This is the accessible text file for GAO report number GAO-14-621 
entitled 'Large Bank Holding Companies: Expectations of Government 
Support' which was released on July 31, 2014. 

This text file was formatted by the U.S. Government Accountability 
Office (GAO) to be accessible to users with visual impairments, as 
part of a longer term project to improve GAO products' accessibility. 
Every attempt has been made to maintain the structural and data 
integrity of the original printed product. Accessibility features, 
such as text descriptions of tables, consecutively numbered footnotes 
placed at the end of the file, and the text of agency comment letters, 
are provided but may not exactly duplicate the presentation or format 
of the printed version. The portable document format (PDF) file is an 
exact electronic replica of the printed version. We welcome your 
feedback. Please E-mail your comments regarding the contents or 
accessibility features of this document to Webmaster@gao.gov. 

This is a work of the U.S. government and is not subject to copyright 
protection in the United States. It may be reproduced and distributed 
in its entirety without further permission from GAO. 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. 

United States Government Accountability Office: 
GAO: 

Report to Congressional Requesters: 

July 2014: 

Large Bank Holding Companies: 

Expectations of Government Support: 

GAO-14-621: 

GAO Highlights: 

Highlights of GAO-14-621, a report to congressional requesters. 

Why GAO Did This Study: 

“Too big to fail” is a market notion that the federal government would 
intervene to prevent the failure of a large, complex financial 
institution to avoid destabilizing the financial sector and the 
economy. Expectations of government rescues can distort investor 
incentives to properly price the risks of firms they view as too big 
to fail, potentially giving rise to funding and other advantages for 
these firms. 

GAO was asked to review the benefits that the largest bank holding 
companies (those with more than $500 billion in assets) have received 
from perceived government support. This is the second of two GAO 
reports on government support for bank holding companies. The first 
study focused on actual government support during the 2007-2009 
financial crisis and recent statutory and regulatory changes related 
to government support for these firms. This report examines how 
financial reforms have altered market expectations of government 
rescues and the existence or size of funding advantages the largest 
bank holding companies may have received due to perceived government 
support. 

GAO reviewed relevant statutes and rules and interviewed regulators, 
rating agencies, investment firms, and corporate customers of banks. 
GAO also reviewed relevant studies and interviewed authors of these 
studies. Finally, GAO conducted quantitative analyses to assess 
potential “too-big-to-fail” funding cost advantages. 

In its comments, the Department of the Treasury generally agreed with 
GAO's analysis. GAO incorporated technical comments from the financial 
regulators, as appropriate. 

What GAO Found: 

While views varied among market participants with whom GAO spoke, many 
believed that recent regulatory reforms have reduced but not 
eliminated the likelihood the federal government would prevent the 
failure of one of the largest bank holding companies. Recent reforms 
provide regulators with new authority to resolve a large failing bank 
holding company in an orderly process and require the largest bank 
holding companies to meet stricter capital and other standards, 
increasing costs and reducing risks for these firms. In response to 
reforms, two of three major rating agencies reduced or removed the 
assumed government support they incorporated into some large bank 
holding companies' overall credit ratings. Credit rating agencies and 
large investors cited the new Orderly Liquidation Authority as a key 
factor influencing their views. While several large investors viewed 
the resolution process as credible, others cited potential challenges, 
such as the risk that multiple failures of large firms could 
destabilize markets. Remaining market expectations of government 
support can benefit large bank holding companies if they affect 
investors' and customers' decisions. 

GAO analyzed the relationship between a bank holding company's size 
and its funding costs, taking into account a broad set of other 
factors that can influence funding costs. To inform this analysis and 
to understand the breadth of methodological approaches and results, 
GAO reviewed selected studies that estimated funding cost differences 
between large and small financial institutions that could be 
associated with the perception that some institutions are too big to 
fail. Studies GAO reviewed generally found that the largest financial 
institutions had lower funding costs during the 2007-2009 financial 
crisis but that the difference between the funding costs of the 
largest and smaller institutions has since declined. However, these 
empirical analyses contain a number of limitations that could reduce 
their validity or applicability to U.S. bank holding companies. For 
example, some studies used credit ratings which provide only an 
indirect measure of funding costs. 

GAO's analysis, which addresses some limitations of these studies, 
suggests that large bank holding companies had lower funding costs 
than smaller ones during the financial crisis but provides mixed 
evidence of such advantages in recent years. However, most models 
suggest that such advantages may have declined or reversed. GAO 
developed a series of statistical models that estimate the 
relationship between bank holding companies' bond funding costs and 
their size or systemic importance, controlling for other drivers of 
bond funding costs, such as bank holding company credit risk. Key 
features of GAO's approach include the following: 

* U.S. Bank Holding Companies: The models focused on U.S. bank holding 
companies to better understand the relationship between funding costs 
and size in the context of the U.S. economic and regulatory 
environment. 

* Bond Funding Costs: The models used bond yield spreads—-the 
difference between the yield or rate of return on a bond and the yield 
on a Treasury bond of comparable maturity-—to measure funding costs 
because they are a risk-sensitive measure of what investors charge 
bank holding companies to borrow. 

* Extensive Controls: The models controlled for credit risk, bond 
liquidity, and other variables to account for factors other than size 
that could affect funding costs. 

* Multiple Models: GAO used 42 models for each year from 2006 through 
2013 to assess the impact of using alternative measures of credit 
risk, bond liquidity, and size and to allow the relationship between 
size and bond funding costs to vary over time with changes in the 
economic and regulatory environment. 

* Credit Risk Levels: GAO compared bond funding costs for bank holding 
companies of different sizes at the average level of credit risk for 
each year, at low and high levels of credit risk for each year, and at 
the average level of credit risk during the financial crisis. 

The figure below shows the differences between model-estimated bond 
funding costs for bank holding companies with $1 trillion in assets 
and bank holding companies with $10 billion in assets, with average 
levels of credit risk in each year. Circles represent statistically 
significant model-estimated differences. 

Figure: Estimates from 42 Models of Average Bond Funding Cost 
Differences between Bank Holding Companies with $1 Trillion and $10 
Billion in Assets, 2006-2013: 

[Refer to PDF for image: illustrated graph] 

Models predicting statistically significant differences above zero: 
2006: 0; 
2007: 0; 
2008: 0; 
2009: 0; 
2010: 2; 
2011: 11; 
2012: 15; 
2013: 18. 

Models predicting statistically insignificant differences above zero: 
2006: 4; 
2007: 7; 
2008: 0; 
2009: 0; 
2010: 12; 
2011: 13; 
2012: 8; 
2013: 12. 

Models predicting statistically significant differences below zero: 
2006: 28; 
2007: 20; 
2008: 39; 
2009: 37; 
2010: 9; 
2011: 14; 
2012: 4; 
2013: 8. 

Models predicting statistically insignificant differences below zero: 
2006: 10; 
2007: 15; 
2008: 3; 
2009: 5; 
2010: 19; 
2011: 4; 
2012: 15; 
2013: 4. 

Estimated average bond funding costs for larger bank holding companies 
(BHC) were higher than for smaller BHCs. 

Estimated average bond funding costs for larger BHCs were lower than 
for smaller BHCs. 

Basis point = .01% (1/100th of a percent). 

Depicted on graph: 
Average difference estimated by one of 42 models (statistically 
significant at 10% level); 
Average difference estimated by one of 42 models (not statistically 
significant at 10% level). 

Source: GAO analysis of data from Bloomberg and the Financial 
Stability Board. GAO-14-621. 

Notes: We estimated econometric models of the relationship between 
bank holding company size and funding costs using data for U.S. bank 
holding companies and their outstanding senior unsecured bonds for the 
first quarter of 2006 through the fourth quarter of 2013. The models 
used bond yield spreads to measure funding costs and controlled for 
credit risk factors such as capital adequacy, asset quality, earnings, 
maturity mismatch, and volatility, as well as bond liquidity and other 
characteristics of bonds and bank holding companies that can affect 
funding costs. We estimated 42 models for each year from 2006 through 
2013 to assess the sensitivity of estimated funding cost differences 
to alternative measures of capital adequacy, volatility, bond 
liquidity, and size or systemic importance. We used the models to 
compare bond funding costs for bank holding companies of different 
sizes but the same levels of credit risk, bond liquidity, and other 
characteristics. This figure compares bond funding costs for bank 
holding companies with $1 trillion and $10 billion in assets, for each 
model and for each year, with average levels of credit risk. Each 
circle and dash shows the comparison for a different model, where 
points below zero suggest bank holding companies with $1 trillion in 
assets have lower bond funding costs than bank holding companies with 
$10 billion in assets, and vice versa. 

[End of figure] 

All 42 models found that larger bank holding companies had lower bond 
funding costs than smaller ones in 2008 and 2009, while more than half 
of the models found that larger bank holding companies had higher bond 
funding costs than smaller ones in 2011 through 2013, given the 
average level of credit risk each year (see figure). However, the models
' comparisons of bond funding costs for bank holding companies of 
different sizes varied depending on the level of credit risk. For 
example, in hypothetical scenarios where levels of credit risk in 
every year from 2010 to 2013 are assumed to be as high as they were 
during the financial crisis, GAO's analysis suggests that large bank 
holding companies might have had lower funding costs than smaller ones 
in recent years. However, reforms in the Dodd-Frank Wall Street Reform 
and Consumer Protection Act, such as enhanced standards for capital 
and liquidity, could enhance the stability of the financial system and 
make such a credit risk scenario less likely. 

This analysis builds on certain aspects of prior studies, but 
important limitations remain and these results should be interpreted 
with caution. GAO's estimates of differences in funding costs reflect 
a combination of several factors, including investors' beliefs about 
the likelihood a bank holding company will fail and the likelihood it 
will be rescued by the government if it fails, and cannot precisely 
identify the influence of each factor. In addition, these estimates 
may reflect factors other than investors' beliefs about the likelihood 
of government support and may also reflect differences in the 
characteristics of bank holding companies that do and do not issue 
bonds. Finally, GAO's estimates, like all past estimates, are not 
indicative of future trends. 

[End of section] 

Contents: 

Letter: 

Scope and Methodology: 

Background: 

Reforms Have Reduced Expectations of Government Support among Key 
Market Observers and Increased Regulation for Large Institutions: 

Evidence Suggests Large Banks Had a Funding Cost Advantage over Small 
Ones during the Financial Crisis That May Have Declined or Reversed 
Since: 

Agency Comments and Our Evaluation: 

Appendix I: Methodology for Analysis of Funding Cost Differences 
between Large and Small Bank Holding Companies: 

Appendix II: Comments from the Department of the Treasury: 

Appendix III: GAO Contact and Staff Acknowledgments: 

Tables: 

Table 1: Overview of Variables Used in Econometric Models: 

Table 2: Empirical Studies of Too-Big-to-Fail Funding Cost Differences: 

Table 3: Numbers of Bank Holding Companies and Bonds, 2006-2013: 

Table 4: Summary Statistics for Bank Holding Companies and Bonds 
Analyzed, 2008 and 2013: 

Table 5: Parameter Estimates for Baseline Regressions, 2008 and 2013: 

Table 6: Baseline Comparisons of Bond Yield Spreads for Bank Holding 
Companies of Different Sizes by Year, 2008 and 2013 (basis points): 

Table 7: Comparisons of Bond Yield Spreads for Bank Holding Companies 
with $1 trillion versus $10 billion in Assets, All 42 Models, 2008 and 
2013 (basis points): 

Figures: 

Figure 1: Estimates from 42 Models of Average Bond Funding Cost 
Differences between Bank Holding Companies with $1 Trillion and $10 
Billion in Assets, 2006-2013: 

Figure 2: Difference in Estimated Bond Funding Costs for Average 
Credit Risk Bank Holding Companies with $1 Trillion versus $10 Billion 
in Assets by Indicator of Size or Systemic Importance, 2013: 

Figure 3: Difference in Estimated Bond Funding Costs for Bank Holding 
Companies with $1 Trillion versus $10 Billion in Assets by Level of 
Credit Risk, 2013: 

Figure 4: Difference in Estimated Bond Funding Costs for Average 
Credit Risk Bank Holding Companies with $1 Trillion versus $10 Billion 
in Assets by Model Specification, by Year, 2006-2013: 

Abbreviations: 

AIG: American International Group, Inc. 

BHC: bank holding company: 

CCAR: Comprehensive Capital Analysis and Review: 

FDIC: Federal Deposit Insurance Corporation: 

FDICIA: FDIC Improvement Act of 1991: 

FSOC: Financial Stability Oversight Council: 

GSIB: global systemically important bank: 

IDI: insured depository institution: 

LTCM: Long-Term Capital Management: 

OCC: Office of the Comptroller of the Currency: 

OFR: Office of Financial Research: 

OLA: Orderly Liquidation Authority: 

S&P: Standard & Poor's: 

SIFI: systemically important financial institution: 

SPOE: Single Point-of-Entry: 

TARP: Troubled Asset Relief Program: 

[End of section] 

United States Government Accountability Office: 
GAO:
441 G St. N.W. 
Washington, DC 20548: 

July 31, 2014: 

The Honorable Sherrod Brown: 
Chairman: 
Subcommittee on Financial Institutions and Consumer Protection: 
Committee on Banking, Housing, and Urban Affairs: 
United States Senate: 

The Honorable David Vitter: 
United States Senate: 

"Too big to fail" refers to a market notion that the federal 
government would intervene to prevent the failure of a large, 
interconnected financial institution to avoid harm to the economy. 
Events during the 2007-2009 financial crisis heightened concerns that 
market participants had come to view several of the largest U.S. 
financial institutions as too big to fail. Most notably, the U.S. 
government intervened to provide tens of billions of dollars of 
capital and other support to a few large troubled financial 
institutions out of concern that allowing these firms to go into 
bankruptcy would have further disrupted troubled credit markets and 
damaged confidence in the U.S. financial system.[Footnote 1] Market 
expectations of government rescues can distort the incentives of 
investors and counterparties to properly price and restrain the risks 
of firms they believe to be too big to fail, potentially giving rise 
to funding cost and other advantages for these firms relative to 
smaller competitors. For example, creditors may be willing to accept 
lower interest rates on debt issued by these firms if they believe the 
possibility of a government rescue reduces the likelihood that they 
could suffer losses. If creditors and other counterparties do not 
fully charge a firm for the risks it is taking, that firm may have 
incentives to take on greater risks in the pursuit of higher returns. 
Excessive risk-taking in response to such incentives can increase the 
likelihood that such a firm could become distressed and disrupt 
financial markets. 

The Dodd-Frank Wall Street Reform and Consumer Protection Act (Dodd-
Frank Act) prohibits rescues of individual failing financial 
institutions and provides new tools and authorities for the Federal 
Deposit Insurance Corporation (FDIC) to resolve a large failing 
financial institution in a manner that imposes losses on the firm's 
creditors and other investors while minimizing adverse impacts to the 
economy.[Footnote 2] Nevertheless, market observers have continued to 
debate whether some of the largest and most complex financial 
institutions may continue to benefit from beliefs among their 
investors and counterparties that the government might intervene to 
prevent their failure. 

You asked us to review any economic benefits that the largest bank 
holding companies (those with more than $500 billion in total 
consolidated assets) have received as a result of actual or perceived 
government support. This is the second of two reports we are issuing 
on this topic.[Footnote 3] In this report, we review (1) what is known 
about how financial reforms have altered market expectations of 
government rescues and the relative advantages or disadvantages of 
being a large bank holding company and (2) the extent to which the 
largest bank holding companies have received funding cost advantages 
as a result of perceptions that the government would not allow them to 
fail. 

Scope and Methodology: 

To address our first objective, we reviewed information from relevant 
statutory provisions and regulations and prior GAO reports to describe 
financial reforms that could impact market expectations of government 
rescues or relative advantages or disadvantages of being a large bank 
holding company.[Footnote 4] We obtained perspectives on the potential 
impacts of these reforms from credit rating agencies, investment 
firms, and corporations that are customers of banks. Where available 
and relevant, we reviewed some public statements, reports, and other 
analyses by these groups. For example, to obtain information about 
credit rating agencies' assessments of the likelihood and level of 
government support for large bank holding companies, we reviewed 
relevant publications by the three largest credit rating agencies: 
Fitch Ratings (Fitch), Moody's Investors Service (Moody's), and 
Standard & Poor's (S&P). We interviewed representatives from each of 
these rating agencies to obtain their perspectives on factors 
contributing to changes in their assessments of government support 
over time. 

We conducted interviews with representatives from 10 investment firms 
and six corporations to learn about (1) factors that influence their 
decisions to invest in or do business with bank holding companies of 
various sizes; (2) how they assess the risks of banks and the extent 
to which they rely on credit rating agencies' assessments of these 
risks; (3) their views on the likelihood that the federal government 
would intervene to prevent the failure of a large bank holding company 
and factors that have influenced these views over time; and (4) how, 
if at all, expectations of government support have impacted their 
decisions to invest in or do business with banks of various sizes. In 
selecting investment firms and large corporations for interviews, we 
selected nonrepresentative samples of firms. As a result, the views we 
present from these firms are not generalizable to the broader 
community of bank investors and customers and do not indicate which 
views are most prevalent. We selected investment firms with experience 
investing in debt or equity securities of banks and bank holding 
companies and selected different types of investment firms to obtain 
perspectives reflecting a range of investing strategies. Specifically, 
we selected three large asset management firms (each with more than $1 
trillion in assets under management); three public pension funds (each 
with more than $50 billion in assets under management); three hedge 
funds; and one large insurance company. We selected U.S. corporations 
from different industry sectors and with a range of banking needs. We 
identified four of these firms and contacted them with the assistance 
of the U.S. Chamber of Commerce, which reached out to its members on 
our behalf, and selected two additional firms to achieve additional 
diversity across industry sectors. The corporations we interviewed 
included four multinational corporations (a chemical company, a 
delivery and logistics company, an energy company, and a technology 
company) and two corporations with all or close to all of their 
operations in the United States (a regional electric utility company 
and a national retail services company). 

To obtain additional information and perspectives on how financial 
reforms or credit ratings could impact the relative advantages or 
disadvantages of being a large bank holding company, we reviewed 
relevant publicly available information in the financial statements of 
bank holding companies and conducted interviews with bank holding 
companies of various sizes, bank industry associations, public 
interest groups, academics, and other experts. For example, we 
reviewed bank holding companies' financial disclosures about how Dodd-
Frank reforms could increase certain fees and how a credit rating 
downgrade could impact the amount of collateral required of them under 
certain financial contracts. We also reviewed our prior work on 
potential impacts of Dodd-Frank Act implementation.[Footnote 5] 

As part of our first objective, we reviewed regulators' efforts to 
assess their progress in addressing too-big-to-fail perceptions and 
market distortions that can result. We reviewed Dodd-Frank Act 
provisions that outline statutory responsibilities for the Financial 
Stability Oversight Council (FSOC) and reviewed relevant sections of 
the FSOC annual report. We interviewed officials from FSOC, the 
Department of the Treasury (Treasury), the Board of Governors of the 
Federal Reserve System (Federal Reserve Board), FDIC, and the Office 
of the Comptroller of the Currency (OCC) about their efforts to 
analyze the impacts of Dodd-Frank reforms on too-big-to-fail 
perceptions and to evaluate whether additional policy actions may be 
needed to address any remaining market distortions. We also reviewed 
relevant congressional testimonies and other public statements by 
agency officials. 

To assess the extent to which the largest bank holding companies have 
received funding cost advantages as a result of perceptions that the 
government would not allow them to fail, we conducted an econometric 
analysis of the relationship between a bank holding company's size and 
its funding costs. To inform our econometric approach and understand 
the breadth of results and methodological approaches, we reviewed 
studies that estimated the funding cost difference between large and 
small financial institutions that could be associated with the 
perception that some institutions are too big to fail. We evaluated 
studies that met the following criteria: (1) used a comparative 
empirical approach that attempted to account for differences across 
financial institutions that could influence funding costs, (2) 
included U.S. bank holding companies, and (3) included analysis of 
data from 2002 or later. We chose these criteria to identify the most 
relevant and rigorous studies related to our research objective. To 
identify studies that met these criteria, we sought input from 
individuals, agencies, and groups that we interviewed, identified 
studies cited in an initial set of studies we had already identified, 
and conducted a systematic search of research databases (including 
Google Scholar and SSRN). 

Our criteria excluded studies that used option-pricing approaches--
that is, techniques that use tools for pricing stock options to 
estimate the value associated with possible government interventions 
to assist distressed banks--because these studies assume a too-big-to-
fail funding cost advantage exists and only estimate its magnitude. We 
also excluded two studies that otherwise met our criteria, but did not 
attempt to control for important differences between financial 
institutions. We were aware of potential conflicts of interest 
associated with a number of studies in our review. For example, one 
study was conducted by researchers at a large bank holding company and 
two others were sponsored by a trade group representing large 
commercial banks. We considered the potential impact these conflicts 
of interest might have on their methods and results. We ultimately 
included 16 studies in our review that we determined were sufficiently 
reliable for the purposes of this report. 

In reviewing these studies, we assessed what they identified as the 
level of funding cost differences and how that level has changed over 
time and we identified the strengths and limitations of the studies' 
approaches. Because of limitations of the methodologies of these 
studies, their results, while suggestive of general trends, are not 
definitive and thus should be interpreted with caution. We interviewed 
authors of selected studies, federal financial regulators, and other 
experts to obtain perspectives on the strengths and limitations of 
relevant quantitative approaches that have been used. 

Taking into consideration the strengths and limitations of different 
methodologies, we developed our own econometric approach to evaluate 
the extent to which the largest bank holding companies may have 
received funding cost advantages as a result of perceptions that the 
government would not allow them to fail. In addition, we selected 
three experts with relevant expertise to review our econometric 
approach and assess its strengths and limitations. These experts 
reviewed our approach before we implemented it and provided comments 
on our methodology. In many instances, we made changes or additions to 
our models to address their comments, and in other instances, we 
disclosed additional limitations of the models. Before selecting these 
experts, we reviewed potential sources of conflicts of interest, and 
we determined that the experts we selected did not have any material 
conflicts of interest for the purpose of reviewing our work. 

We used a multivariate regression model to estimate the relationship 
between bank holding companies' funding costs and their size while 
controlling for factors other than size that may also influence 
funding costs. Our general regression model is the following: 

[Formula illustration] 

In this model, denotes the bank holding company, denotes the quarter, 
is the bank holding company's cost of funding in a quarter, is a 
measure of the bank holding company's size at the beginning of the 
quarter, is a list of proxies for the bank holding company's credit 
risk--the risk that the bank holding company will not repay the funds 
it borrowed as agreed, is a list of other variables that may influence 
funding costs, is an idiosyncratic error term, and are parameters to 
be estimated. The parameter captures the direct relationship between a 
bank holding company's funding cost and its size. The parameter 
captures the indirect relationship between a bank holding company's 
funding cost and its size that exists if the size of a bank holding 
company affects the relationship between its funding cost and credit 
risk. If investors view larger bank holding companies as less risky 
than smaller bank holding companies due to beliefs that the government 
is more likely to rescue larger bank holding companies in distress, 
then either is less than zero, is less than zero, or both. [Footnote 
6] However, the parameters and may also reflect factors other than 
these beliefs. 

We used a measure of funding costs based on bonds issued by bank 
holding companies. Bank holding companies use a variety of funding 
types from different sources, including various types of deposits, 
bonds, and equity. We used bond yield spreads--the difference between 
the yield on a bond and the yield on a Treasury bond of comparable 
maturity--to measure a bank holding company's cost of bond funding. 
Treasury securities are widely viewed as a risk-free asset, so the 
yield spread measures the price that investors charge a bank holding 
company to borrow to compensate them for credit risk and other 
factors. We focused on bond yield spreads because they are a measure 
of funding costs that is available for bank holding companies of a 
range of sizes, including bank holding companies with less than $10 
billion in assets. Furthermore, bonds are traded in secondary markets, 
so changes in bond yield spreads can be publicly observed in a timely 
manner. Finally, bond yield spreads are a direct measure of funding 
costs, unlike alternatives such as credit ratings. 

We used Bloomberg to identify U.S. bank holding companies with more 
than $500 million in assets that were operating in 1 or more years 
from 2006 through 2013, and to identify all plain vanilla, fixed-rate, 
senior unsecured bonds issued by these bank holding companies, 
excluding bonds with an explicit government guarantee.[Footnote 7] We 
collected data on bond yield spreads, bank holding company size, 
variables associated with bank holding company credit risk, and bond 
characteristics from Bloomberg. We used these data to assemble a 
dataset with one observation for each bond in each quarter from the 
first quarter of 2006 through the fourth quarter of 2013. We 
constructed alternative measures to control for size, bond liquidity, 
and credit risk due to uncertainty about how to appropriately capture 
these important factors influencing bond yields and because the 
regression results may be sensitive to alternative specifications (see 
table 1). The numbers of bank holding companies and bonds we analyzed 
and summary statistics for our indicators of size, credit risk, and 
other factors are in appendix I. 

Table 1: Overview of Variables Used in Econometric Models: 

Funding costs: 
Quarterly average bond yield spreads based on prices from executed 
trades, executable quotes, and composites derived from executable and 
indicative quotes[B]; 
Quarterly average bond yield spreads based on executed trades only; 
Size: 
Total assets; 
Total assets and total assets squared; 
An indicator for whether or not a bank holding company had $50 billion 
or more in assets[C]; 
An indicator for whether or not a bank holding company was designated 
a globally systemically important bank in November 2013; 
Credit risk[A]: 
Capital adequacy: Equity capital and subordinated debt as percentages 
of assets and risk-weighted assets[D]; 
Asset quality: Nonperforming assets as a percentage of tangible common 
equity plus loan loss reserves[E]; 
Earnings: Net income as a percentage of assets; 
Maturity mismatch: Volatile liabilities minus liquid assets as a 
percentage of total liabilities[F]; 
Volatility: Standard deviation of equity prices over the quarter, 
option-implied volatility for the quarter, the standard deviation of 
equity returns over the quarter, the standard deviation of excess 
equity returns over the quarter, and the standard deviation of 
earnings[G]; 
Other bond and bank holding company characteristics: 
Bond liquidity: Issue size, total volume traded during a quarter, and 
average bid-ask spread over a quarter[H]; 
Coupon rate: The current interest rate on the bond; 
Time to maturity: The number of quarters until the bond matures; 
Operating expenses: Noninterest expense as a percentage of total 
assets. 

Source: GAO. GAO-14-621. 

Note: Unless otherwise noted, all balance sheet and income statement 
variables--such as assets, liabilities, and net income--are measured 
as of the beginning of the quarter. 

[A] Credit risk increases with higher values of our asset quality, 
maturity mismatch, and volatility measures and with lower values of 
our capital adequacy and earnings measures. 

[B] Bond yield spreads are the difference between the yield on a bond 
and the yield on a comparable Treasury bond. Executed trades are 
completed transactions for which the terms have been agreed upon and 
assets have been exchanged. Executable quotes (or firm quotes) are 
prices quoted by dealers at which they would be willing to trade. 
Indicative quotes are prices quoted by dealers that give an indication 
of the price at which a bond might trade but at which they are not 
obligated to trade. Composite quotes or prices are derived from 
quantitative algorithms that aggregate information on prices from 
executed trades, executable quotes, and indicative quotes to give an 
indication of the market price of a bond. 

[C] The Dodd-Frank Act requires bank holding companies with $50 
billion or more in total assets to be subjected to enhanced regulatory 
standards and supervision. Pub. L. No. 111-203, § 165, 124 Stat. 1376, 
1423 (2010) (codified at 12 U.S.C. § 5365). 

[D] Regulators generally require that banks maintain certain ratios of 
capital as a share of assets to ensure that they have sufficient 
capital to absorb losses. Under the Basel approaches, banks may weight 
certain assets based on their risks, and use these risk-weighted 
assets to calculate their capital adequacy ratios. 

[E] Nonperforming assets include assets in nonaccrual status, other 
real estate owned, restructured loans, and restructured loans in 
compliance. Tangible common equity is total capital minus perpetual 
preferred stock minus goodwill and other intangibles. 

[F] Volatile liabilities include federal funds purchased and 
repurchase agreements, trading liabilities, other borrowed funds, 
foreign deposits, and jumbo deposits less derivatives with negative 
fair value. Liquid assets include cash, securities, federal funds sold 
and reverse repurchase agreements, and trading assets less pledged 
securities. 

[G] Option-implied volatility is a measure of a security's expected 
volatility as reflected by the market price of traded options on that 
security. The theoretical price of an option is a function of the 
underlying price, strike price, historical volatility of the 
underlying, the risk-free rate, and the time to expiration. Implied 
volatility is calculated by using the market price of the option and 
solving for volatility. Excess return is a security's return minus the 
return from a no-risk security during the same time period. 

[H] The bid-ask spread is the difference between the best buying price 
and the best selling price. 

[End of table] 

We developed a variety of econometric models that use alternative 
measures of bond liquidity, bank holding company credit risk, and the 
size or systemic importance of a bank holding company. We estimated 
the parameters for each of our models separately for each year from 
2006 through 2013 to allow the relationship between bank holding 
company size and bond funding costs to vary over time.[Footnote 8] Our 
baseline models used average yield spreads on senior unsecured bonds 
based on actual trades, executable quotes, and composites derived from 
executable and indicative quotes to measure bond funding costs; total 
assets to measure size; equity capital and subordinated debt as 
percentages of total assets to measure capital adequacy; and issue 
size and total volume to measure bond liquidity. We estimated the 
baseline model for each year and for each of our five measures of 
volatility, as well as for each year without a measure of volatility. 
We also estimated models that added average bid-ask spread to our 
baseline indicators of bond liquidity, models that used average yield 
spreads based only on actual trades, models that used equity capital 
and subordinated debt as percentages of risk-weighted assets as our 
indicators of capital adequacy, models that used global systemically 
important bank (GSIB) designation as an indicator of size, models that 
used the $50 billion asset threshold as an indicator of size, and 
models that used both total assets and the square of total assets as 
indicators of size.[Footnote 9] Altogether, we used 42 separate models 
for each year from 2006 through 2013. For all models, we included 
indicators for each quarter to control for the influence on yield 
spreads of economic conditions, the regulatory environment, and other 
factors that vary over time but not across bank holding companies. The 
details of the models we estimated and the results for our baseline 
models for select years are in appendix I. 

We used our models to compare bond funding costs for bank holding 
companies of different sizes, all else being equal.[Footnote 10] 
Because our models account for the possibility that investors' beliefs 
about government rescues depend on the credit risk level of the bank 
holding company, we made comparisons for bank holding companies with 
the average level of credit risk that prevailed each year. In 
addition, we assessed the impact of credit risk on our comparisons by 
making comparisons at credit risk levels higher and lower than the 
average for each year and also while holding the level of credit risk 
constant over time at the average level for 2008--the year when the 
financial crisis peaked and credit risk for bank holding companies was 
high. By holding credit risk constant, we can assess the extent to 
which changes in average credit risk over time may have influenced 
changes in funding costs relative to other factors. 

Our approach is limited by several factors. Investors' beliefs about 
the likelihood of government support are composed of several different 
elements, including the likelihood that a bank holding company will 
fail, the likelihood that it will be rescued by the government if it 
fails, and the size of the losses that the government may impose on 
investors if it rescues the bank holding company, but our methodology--
like the methodologies used by other researchers--does not allow us to 
precisely identify the influence of each of these components. Although 
we have taken into account many factors that may influence bond yield 
spreads and that differ for bank holding companies of different sizes, 
our estimates of differences in bond yield spreads for bank holding 
companies of different sizes may reflect factors other than investors' 
beliefs about the likelihood of government support because our control 
variables are imperfect or may be incomplete. In addition, our 
estimates of differences in bond yield spreads for bank holding 
companies of different sizes may reflect differences in the 
characteristics of bank holding companies that choose to issue bonds. 
The section of this report that addresses our second objective 
contains a fuller discussion of the limitations associated with our 
empirical work. 

For parts of our work that involved the analysis of computer-processed 
data, such as market data used in our analysis of funding cost 
differences, we assessed the reliability of these data by reviewing 
relevant documentation and conducting interviews with data providers 
to review steps they took to collect and ensure the reliability of the 
data. In addition, we electronically tested data fields for missing 
values, outliers, and obvious errors. We determined that these data 
were sufficiently reliable for our purposes. 

We conducted this performance audit from January 2013 to July 2014 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. 

Background: 

While the 2007-2009 financial crisis highlighted concerns about the 
market distortions that can result from too-big-to-fail perceptions, 
concern about such distortions pre-dated the crisis. A key factor 
giving rise to the too-big-to-fail dilemma has been the emergence of 
financial institutions of such size, interconnectedness, and market 
importance that their failure could threaten to severely disrupt the 
financial system and damage the economy. Although the federal 
government's policy responses to failing financial institutions in 
recent decades have not formed a clear pattern in terms of the 
availability or structure of government support, these responses may 
have influenced market views on the likelihood of government support. 
Several observers trace too-big-to-fail concerns back to 1984 when 
FDIC provided support to Continental Illinois National Bank, then the 
sixth largest U.S. bank in terms of total assets, to prevent its 
failure and losses to its depositors and creditors.[Footnote 11] The 
Federal Reserve Board's response to the near failure of a large U.S. 
hedge fund, Long-Term Capital Management (LTCM), in 1998 was another 
significant event that may have contributed to too-big-to-fail 
perceptions.[Footnote 12] While LTCM was not itself a large bank, the 
Federal Reserve Board's intervention in helping to facilitate private-
sector assistance to LTCM may have signaled the willingness of federal 
government authorities to intervene to avoid potential systemic 
consequences from a large, interconnected financial firm's failure. 
Other factors may have contributed to some ambiguity surrounding the 
likely recipients and circumstances of government support in the years 
leading up to the 2007-2009 financial crisis. For example, failures 
and near-failures of large financial firms had been infrequent and 
occurred under varying circumstances, making it difficult to discern a 
clear pattern of government support. 

During the 2007-2009 crisis, the federal government took actions to 
stabilize the financial system by creating new emergency programs with 
broad-based eligibility and providing firm-specific assistance to 
prevent the failures of large financial institutions.[Footnote 13] 
Notably, however, U.S. government authorities' initial responses to 
impending failures of large financial institutions did not send a 
clear signal about the availability of government support. In March 
2008, the Federal Reserve Board authorized emergency assistance to 
prevent the failure of one large investment bank (Bear Stearns 
Companies, Inc.), but 6 months later, Federal Reserve Board officials 
determined that they could not assist another large failing investment 
bank, Lehman Brothers Holdings, Inc. (Lehman Brothers).[Footnote 14] 
Following Lehman Brothers' bankruptcy announcement on September 15, 
2008, which triggered an intensification of the financial crisis, U.S. 
government authorities took actions that signaled a stronger near-term 
commitment to prevent the failure of systemically important financial 
institutions. On the day after Lehman Brothers' bankruptcy 
announcement, the Federal Reserve Board authorized up to $85 billion 
of credit assistance for American International Group, Inc. (AIG) to 
prevent its failure.[Footnote 15] In addition, on September 29, 2008, 
the Secretary of the Treasury invoked the systemic risk exception for 
the first time since the enactment of the FDIC Improvement Act of 
1991(FDICIA) to authorize FDIC to provide assistance to avert the 
failure of Wachovia Corporation--then the fourth-largest banking 
organization in terms of assets in the United States--by facilitating 
Citigroup Inc.'s acquisition of its banking operations.[Footnote 16] 

At the height of the crisis in late 2008, the United States and other 
G7 countries announced an agreement to implement a comprehensive 
action plan to provide liquidity to financial markets and prevent the 
failure of any systemically important institution, among other 
objectives.[Footnote 17] That month, U.S. government agencies launched 
two of the largest U.S. emergency initiatives: the Troubled Asset 
Relief Program, through which Treasury provided capital and other 
assistance to eligible financial institutions, and the Temporary 
Liquidity Guarantee Program, through which FDIC guaranteed certain 
uninsured deposits and newly issued unsecured debt of eligible 
financial institutions.[Footnote 18] In late 2008 and early 2009, 
Treasury, the Federal Reserve Board, and FDIC authorized packages of 
firm-specific assistance for two of the largest U.S. bank holding 
companies, Citigroup Inc. and Bank of America Corporation. Foreign 
governments launched parallel initiatives to provide broad-based 
liquidity support and also took steps to prevent the failures of large 
financial institutions. Examples of large foreign financial 
institutions that received firm-specific assistance from their 
governments include Royal Bank of Scotland Group PLC (United Kingdom) 
and UBS (Switzerland). 

Since the onset of the financial crisis, the largest banks have grown 
bigger in many major advanced economies, even as the financial sector 
has shrunk, and U.S. and foreign policymakers have acknowledged that 
crisis policy interventions raised moral hazard concerns.[Footnote 19] 
As discussed earlier, market perceptions that some firms are too big 
to fail can distort market participants' incentives to properly price 
and restrain risk-taking by these firms. U.S. regulators have 
coordinated with foreign counterparts through the G20 and the 
Financial Stability Board to develop a policy framework for addressing 
the risks posed by large, complex financial institutions.[Footnote 20] 
In November 2010, G20 leaders endorsed the Financial Stability Board's 
framework for addressing too-big-to-fail concerns. The framework aims 
to reduce the probability and impact of the failure of systemically 
important firms.[Footnote 21] Key elements of this framework include 
developing effective resolution regimes and strengthening capital 
standards for systemically important financial institutions. FDIC, the 
Federal Reserve Bank of New York, and Treasury helped to develop 
standards the Financial Stability Board issued for effective 
resolution regimes in October 2011.[Footnote 22] In addition, U.S. 
banking regulators have worked with their foreign counterparts to 
develop a strengthened capital regime that will require global 
systemically important banks to have additional loss absorbing 
capacity.[Footnote 23] U.S. federal financial regulators are 
implementing these and other elements of the Financial Stability 
Board's framework for addressing too big to fail as part of the 
process of implementing relevant Dodd-Frank Act provisions. 

Dodd-Frank Act Provisions: 

U.S. federal financial regulators have made progress in implementing 
Dodd-Frank Act provisions and related reforms to restrict future 
government support and reduce the likelihood and impacts of the 
failure of a systemically important financial institution 
(SIFI).[Footnote 24] These reforms can be grouped into four general 
categories: (1) restrictions on regulators' emergency authorities to 
provide assistance to financial institutions; (2) new tools and 
authorities for regulators to resolve a failing SIFI outside of 
bankruptcy if its failure would have serious adverse effects on the 
U.S. financial system; (3) enhanced regulatory standards for SIFIs 
related to capital, liquidity, and risk management; and (4) other 
reforms intended to reduce the potential disruptions to the financial 
system that could result from a SIFI's failure. 

Restrictions on Emergency Authorities. The Dodd-Frank Act revised 
Federal Reserve Board and FDIC emergency authorities so that emergency 
assistance can no longer be provided to assist a single and specific 
firm but rather can only be made available through a program with 
broad-based eligibility--that is, a program that provides funding 
support to institutions that meet program requirements and that choose 
to participate.[Footnote 25] 

New Tools and Authorities for Resolving SIFIs. The Dodd-Frank Act 
includes two key reforms intended to facilitate the orderly resolution 
of a large failing firm without a taxpayer-funded rescue: (1) 
requirements for SIFIs to formulate and submit to regulators 
resolution plans (or "living wills") that detail how the companies 
could be resolved in bankruptcy in the event of material financial 
distress or failure; and (2) the Orderly Liquidation Authority (OLA), 
through which FDIC can resolve large financial firms, and which 
provides an alternative to bankruptcy if resolution under the 
bankruptcy code would have serious adverse effects on financial 
stability.[Footnote 26] 

* Living wills. Title I of the Dodd-Frank Act requires bank holding 
companies with $50 billion or more in consolidated assets and nonbank 
financial companies designated by FSOC to formulate and submit to 
FDIC, the Federal Reserve Board, and FSOC resolution plans (or "living 
wills") that detail how the companies could be resolved in bankruptcy 
in the event of material financial distress or failure.[Footnote 27] 
The Federal Reserve Board and FDIC have finalized rules relating to 
resolution plans, and the large bank holding companies that were the 
first firms required to prepare and submit such plans submitted these 
to regulators as expected in July 2012.[Footnote 28] Regulators 
reviewed these initial plans and developed guidance on what 
information should be included in 2013 resolution plan 
submissions.[Footnote 29] If FDIC and the Federal Reserve Board 
jointly determine that a resolution plan is not credible or would not 
facilitate an orderly resolution of the company under the bankruptcy 
code, after giving the company an opportunity to remedy the plan's 
deficiencies, the agencies may jointly decide to impose more stringent 
regulatory requirements on the company.[Footnote 30] 

* Orderly Liquidation Authority. OLA gives FDIC the authority, subject 
to certain constraints, to resolve large financial firms, including 
nonbanks, outside of the bankruptcy process.[Footnote 31] This 
authority allows for FDIC to be appointed receiver for a financial 
firm if the Secretary of the Treasury determines, among other things, 
that the firm's failure and its resolution under applicable federal or 
state law, including bankruptcy, would have serious adverse effects on 
U.S. financial stability and no viable private-sector alternative is 
available to prevent the default of the financial company.[Footnote 
32] While the Dodd-Frank Act does not specify how FDIC must exercise 
this authority, FDIC is developing an approach to resolving a firm 
under OLA that it refers to as the Single Point-of-Entry (SPOE) 
approach. 

Under the SPOE approach, FDIC would be appointed receiver of the top-
tier U.S. parent holding company of a financial group determined to be 
in default or in danger of default following the completion of the 
appointment process set forth under the Dodd-Frank Act. Immediately 
after placing the parent holding company into receivership, FDIC would 
transfer assets (primarily the equity and investments in subsidiaries) 
from the receivership estate to a bridge financial holding company. By 
allowing FDIC to take control of the firm at the holding company 
level, this approach is intended to allow subsidiaries (domestic and 
foreign) carrying out critical services to remain open and operating. 
In a SPOE resolution, at the parent holding company level, 
shareholders would be wiped out, and unsecured debt holders would have 
their claims written down to reflect any losses that shareholders 
cannot cover. Under the Dodd-Frank Act, officers and directors 
responsible for the failure cannot be retained.[Footnote 33] The new 
resolution authority under the Dodd-Frank Act provides a back-up 
source for liquidity support, the Orderly Liquidation Fund, which 
could provide liquidity support to the bridge financial company if 
customary sources of liquidity are unavailable.[Footnote 34] The law 
requires FDIC to recover any losses arising from a resolution by 
collecting assessments from bank holding companies with $50 billion or 
more in consolidated assets, nonbank financial holding companies 
designated for supervision by the Federal Reserve Board, and other 
financial companies with $50 billion or more in consolidated assets. 
[Footnote 35] 

Enhanced Regulatory Standards. The Dodd-Frank Act also introduced a 
number of regulatory changes designed to reduce the risks that the 
largest financial institutions pose to the financial system. The act 
requires the Federal Reserve Board to create enhanced capital and 
prudential standards for SIFIs.[Footnote 36] According to Federal 
Reserve Board officials, in implementing these reforms, the Federal 
Reserve Board aims to design prudential standards that will both 
reduce the likelihood of a covered institution's failure and create 
incentives for these institutions to reduce their systemic footprint. 
The act's provisions related to enhanced prudential standards for 
these covered firms include, but are not limited to, the following: 

* Risk-based capital requirements and leverage limits. The act 
required the Federal Reserve Board to establish capital and leverage 
standards, which, as finalized, include a requirement for covered 
firms to develop capital plans to help ensure that they maintain 
capital ratios above specified standards, under both normal and 
adverse conditions.[Footnote 37] In addition, the Federal Reserve 
Board has announced its intention to apply capital surcharges to some 
or all firms based on the risks these firms pose to the financial 
system.[Footnote 38] 

* Liquidity requirements. The act required the Federal Reserve Board 
to establish liquidity standards, which as finalized include 
requirements for covered firms to hold liquid assets that can be used 
to cover their cash outflows over short periods and in stressed 
conditions.[Footnote 39] In addition, the Federal Reserve Board, FDIC, 
and OCC have issued a proposed rule that would implement a minimum 
liquidity requirement that is consistent with the Basel III liquidity 
coverage ratio and would apply to internationally active U.S. banking 
organizations and U.S. depository institutions with $250 billion or 
more in total consolidated assets.[Footnote 40] 

* Risk management requirements. Publicly traded covered firms must 
establish a risk committee and be subject to enhanced risk management 
standards.[Footnote 41] 

* Stress testing requirements. The Federal Reserve Board is required 
to conduct an annual evaluation of whether covered firms have 
sufficient capital to absorb losses that could arise from adverse 
economic conditions.[Footnote 42] 

The Federal Reserve Board has been implementing the enhanced standards 
required by the Dodd-Frank Act in conjunction with its implementation 
of Basel III, a set of risk-based capital, leverage, and liquidity 
standards developed by the Basel Committee on Banking Supervision. 
[Footnote 43] The Basel capital reforms include a risk-based capital 
surcharge that will apply to financial institutions that have been 
designated as GSIBs. Further, the U.S. banking agencies have already 
adopted a leverage capital surcharge that will apply to the eight U.S. 
banking organizations that are GSIBs.[Footnote 44] 

Other Reforms. The act includes other reforms that could help reduce 
the likelihood or impacts of a SIFI's failure. 

* Authorities related to SIFI size and complexity. The Dodd-Frank Act 
grants regulators new authorities to take certain actions if they 
determine that a SIFI poses risks of serious adverse effects on the 
stability of the financial system. These include the authority for the 
Federal Reserve Board to require a SIFI to meet even stricter 
regulatory standards, the authority for the Federal Reserve Board to 
limit (with the approval of FSOC) the ability of a SIFI to merge with 
another company if it determines that the merger would pose a grave 
threat to U.S. financial stability, and, as noted above, the joint 
authority for the Federal Reserve Board and FDIC to require a firm to 
take steps to become more resolvable in bankruptcy.[Footnote 45] 

* Volcker rule. Section 619 of the Dodd-Frank Act (also known as the 
Volcker rule) generally prohibits proprietary trading by insured 
depository institutions and their affiliates and places restrictions 
on sponsorship or investment in hedge and private equity 
funds.[Footnote 46] The Volcker rule's restrictions may have greater 
impacts on larger bank holding companies that have been more involved 
in the types of activities the rule restricts. To the extent that 
Volcker rule implementation prevents these large institutions from 
engaging in certain risky activities, it could serve to reduce the 
likelihood of their failure. 

* Swaps clearing and margin requirements. Title VII of the Dodd-Frank 
Act establishes a new regulatory framework for swaps to reduce risk, 
increase transparency, and promote market integrity in swaps 
markets.[Footnote 47] As we previously reported, requirements for 
swaps to be cleared through clearinghouses can reduce the 
vulnerability of the financial system to the failure of one or a few 
of the major swap dealers by transferring credit risk from the swap 
counterparties to the clearinghouse.[Footnote 48] At the same time, 
experts have pointed out that clearinghouses concentrate credit risk 
and thus represent a potential source of systemic risk.[Footnote 49] A 
benefit of the central clearing requirement is that clearinghouses 
require members to post margin for their trades and the Dodd-Frank Act 
also includes provisions that require regulators to develop margin 
requirements for uncleared swaps.[Footnote 50] These new requirements 
could help reduce systemic risk by preventing the build-up of large, 
undercollateralized exposures. 

Although federal financial regulators have finalized a number of rules 
related to these reforms, implementation of some key reforms has not 
yet been completed. For example, FDIC has largely completed the core 
rulemakings necessary to carry out its systemic resolution 
responsibilities, and is continuing to develop its SPOE approach. FDIC 
requested public comments on its SPOE resolution strategy in December 
2013, and the comment period closed in March 2014.[Footnote 51] In 
addition, regulators have not finalized certain rules that would 
subject SIFIs to enhanced prudential standards. For example, 
regulators have not finalized rules on single-counterparty credit 
limits.[Footnote 52] 

Reforms Have Reduced Expectations of Government Support among Key 
Market Observers and Increased Regulation for Large Institutions: 

While views among investment firms we interviewed and credit rating 
agencies varied, many believe the Dodd-Frank Act has reduced but not 
eliminated the possibility of a government rescue of one of the 
largest bank holding companies. Two of the three largest credit rating 
agencies cited FDIC's resolution process as a key factor in their 
decisions to reduce or eliminate "uplift"--an increase in the credit 
rating--they had assigned to the credit ratings of eight of the 
largest bank holding companies due to their assumptions of government 
support for these firms. Several representatives from large investment 
firms with whom we spoke told us that FDIC's resolution process makes 
significant progress in reducing expectations of government support, 
but several agreed that uncertainty around its implementation or the 
circumstances of its use remains. As such, some market perceptions 
that the government might not allow the largest bank holding companies 
to fail remain and can give rise to advantages for these firms if 
these perceptions affect decisions by investors, counterparties, and 
customers of these firms. For example, credit rating agencies' 
assignment of higher credit ratings due to assumed government support 
can create benefits for these firms, but because investors may rely on 
credit ratings to varying degrees, the impact of such benefits may 
vary accordingly. In addition, Dodd-Frank Act provisions and related 
rules subject the largest firms to higher fees and stricter regulation 
that may reduce their risk of failure and increase costs on them 
relative to smaller competitors. Officials from FSOC and some of its 
member agencies have stated that financial reforms have not completely 
removed too-big-to-fail perceptions, but have made significant 
progress toward doing so. They anticipate that remaining expectations 
of government support will decline as Dodd-Frank implementation 
progresses. 

Some Credit Rating Agencies and Large Investors Believe Dodd-Frank Has 
Reduced the Likelihood of Government Rescues: 

While views among credit rating agencies and investment firms varied, 
many believe the Dodd-Frank Act has reduced but not eliminated the 
possibility of a government rescue of one of the largest bank holding 
companies. During the financial crisis, credit rating agencies 
assigned or increased "uplift"--or an increase in the credit rating--
for several large bank holding companies' credit ratings to reflect 
their view that the increased possibility of government support for 
these firms reduced the risk that the firms' creditors would suffer 
losses.[Footnote 53] We reviewed changes in credit rating agencies' 
assumptions about government support over time and interviewed credit 
rating agency representatives. Because large investors do not 
necessarily rely on credit ratings or rating agencies' assessments of 
government support, we obtained perspectives from representatives of 
large asset management firms, pension funds, hedge funds, and other 
investment firms that purchase debt and equity issued by bank holding 
companies.[Footnote 54] 

Citing progress in Dodd-Frank implementation and other changes, two 
credit rating agencies have reduced or eliminated their rating uplift 
for the largest bank holding companies since the end of the 2007-2009 
financial crisis, and a third rating agency has noted that regulatory 
developments may lead it to reduce or eliminate the uplift it assigns 
to these bank holding companies' ratings. The three largest credit 
rating agencies--Fitch, Moody's, and S&P--have incorporated government 
support into their ratings over time, with the amount of increase in 
credit rating being driven by their assessment of the firm's 
standalone credit rating--the credit rating that assumes no government 
support--and the impact on the firm's creditworthiness they attribute 
to the likelihood of government support.[Footnote 55] These three 
rating agencies have incorporated government support into their 
ratings using different approaches and under Fitch's approach, 
assumptions of extraordinary government support have not led to an 
increase in rating for U.S. bank holding companies in all cases. Fitch 
incorporates government support into its ratings by assigning a 
"support rating floor," a minimum credit rating that reflects its view 
about the likelihood an entity will receive company-specific 
government support in case of need, though the entity only receives 
"uplift" in its rating if its standalone credit rating is below the 
"support rating floor." Following the July 2010 enactment of the Dodd-
Frank Act, Fitch and Moody's removed their assumptions of government 
support for several large regional bank holding companies and their 
subsidiary banks and reduced (but did not eliminate) their assumptions 
of support for eight of the largest U.S. bank holding companies. 
[Footnote 56] Fitch and Moody's reports cited FDIC's new resolution 
authority and a reduced willingness by the U.S. government to assist a 
failing bank holding company as key factors influencing these changes 
in assumed government support. As of June 2014, S&P had not changed 
its level of assumed government support since the financial 
crisis.[Footnote 57] However, in June 2013, S&P noted that regulatory 
developments may lead it to reassess its assumptions of government 
support for the eight bank holding companies.[Footnote 58] The three 
credit rating agencies each noted that their remaining assumptions of 
government support reflected continued uncertainty about the ability 
of the U.S. government to effectively resolve one of the largest bank 
holding companies in OLA. In September 2013, Fitch indicated that it 
would conduct a global review of its support ratings and in March 
2014, Fitch reported that it expects to remove its support rating 
floor for several of the largest U.S. bank holding companies within 
the next one or two years.[Footnote 59] In November 2013, Moody's 
removed all uplift from assumed government support from its credit 
ratings for the remaining eight large bank holding companies.[Footnote 
60] Moody's cited regulators' substantial progress in establishing the 
SPOE receivership framework as a main consideration in their decision 
to remove the uplift. They noted that the SPOE framework would allow 
FDIC to impose losses on the creditors of a U.S. bank holding company 
to recapitalize and preserve the operations of the bank's systemically 
important subsidiaries in a stress scenario. As a result, they believe 
that the holding company creditors of systemically important U.S. 
banks are unlikely to receive government support. 

Representatives of large investment firms with whom we spoke said that 
they rely primarily on their own assessments of government support 
when investing in financial institutions, and they identified OLA and 
other reforms as factors influencing their views. While 
representatives of several firms said that Dodd-Frank reforms have 
significantly reduced or eliminated expectations of government 
rescues, others said they continue to expect that the government would 
rescue one of the largest bank holding companies under certain 
scenarios if policymakers judged the potential costs to the economy 
from such a failure to be too great. 

Investors generally cited progress on OLA and enhanced regulatory 
standards for the largest bank holding companies as among the most 
important factors influencing their views on the likelihood of 
government support, and many considered living wills and other reforms 
to be less significant factors. 

FDIC's resolution process. Investors with whom we spoke said that 
FDIC's progress in developing its resolution process to implement OLA 
as an alternative to bankruptcy has caused them to significantly 
reduce their expectations of government support, but uncertainty 
around its implementation and circumstances of its use remains. 
Although several investors believed that FDIC's resolution process is 
credible for managing a single large failure, two expressed doubts 
about whether it could be used to resolve multiple failing firms in a 
systemic crisis. They noted that if the economic costs of a large 
firm's failure were judged to be too high, the federal government 
might not want to risk using OLA if regulators believed it would 
destabilize markets. Two investors noted that in the event that 
concerns about destabilizing markets led the federal government to 
provide emergency assistance to a failing firm in lieu of using OLA, 
policymakers might face political pressure to structure the assistance 
in a manner that imposed losses on creditors.[Footnote 61] Other 
factors being equal, an investor's belief that there is a possibility 
of incurring losses even if the government prevents a firm's failure 
would reduce that investor's willingness to provide funds to that firm 
on more favorable terms because of a too-big-to-fail perception. 
Because OLA is untested, some uncertainty may exist about its 
viability as an alternative to bankruptcy and government rescues until 
it is used. 

Some investors identified areas where further progress is needed to 
enhance the credibility of OLA. First, some market observers have 
pointed to opportunities to further minimize the adverse market 
impacts that could result from resolving a firm under OLA. For 
example, although OLA provides for a 1-day stay on qualified financial 
contracts to allow for the selection of contracts to transfer to the 
bridge company, derivatives contracts written under the laws of other 
countries could allow counterparties to close out those contracts 
immediately, possibly posing liquidity issues for the firm and leading 
it to sell assets at depressed prices into the market. Some regulatory 
officials have said that cross-border agreements that create 
conformity in the treatment of derivatives contracts in resolution 
processes would enhance OLA's effectiveness and practicality as a 
resolution tool. In addition, some investors noted that progress on 
the Federal Reserve's planned proposal for a minimum long-term debt 
requirement could create greater certainty that the largest bank 
holding companies would have enough equity and debt to absorb losses 
and recapitalize their operating subsidiaries under OLA.[Footnote 62] 

Enhanced regulatory standards. Many investment firm representatives 
credited enhanced regulatory standards for the largest bank holding 
companies with improving the safety and soundness of these firms and 
reducing the likelihood that they would experience distress that could 
result in failure or government support. One representative from a 
large investment firm said that the best defense against banks needing 
government support is to make sure they are well-capitalized. 
Similarly, another investment firm representative said that higher 
capital ratios and strengthened balance sheets have given confidence 
to the markets that the institutions are more sound, in turn reducing 
the likelihood that they would fail and potentially receive government 
assistance. A representative from one large asset management firm said 
that enhanced capital and liquidity standards are a positive from a 
debt holder's perspective because increased capital provides a bigger 
buffer to absorb losses and increased liquidity makes a run on the 
firm less likely. 

Living wills. Several investors said the living wills may have 
positive effects, but some investors have expressed doubts about the 
effectiveness of the plans, with one investor citing a lack of public 
transparency. In a public comment letter to FDIC, The Credit 
Roundtable, a financial industry association, noted that additional 
living will disclosures would improve the market's ability to gauge 
the level of risk under a SPOE scenario. Additionally, while the 
purpose of living wills is to make SIFIs resolvable in bankruptcy, 
several large investors said they assume that a failing SIFI would be 
resolved through OLA. 

Remaining Market Expectations of Government Support Can Have Benefits 
for Large Firms: 

Remaining market assumptions about government support can give rise to 
advantages for the largest bank holding companies in three broad 
categories to the extent these assumptions affect decisions by 
investors, counterparties, and customers of these firms. Those 
categories are funding costs, financial contracts that reference 
ratings, and ability to attract customers. 

Funding Costs: 

Market beliefs about government support could benefit a firm by 
lowering its funding costs. However, the extent to which this occurs 
depends in part on the extent to which providers of funds--such as 
depositors, bond investors, and stockholders--rely on credit ratings 
that assume government support or incorporate their own expectations 
of government support into their decisions to provide funds. For 
example, an investor that relies on credit ratings may view a firm 
with a rating that incorporates implied government support as having 
lower risk--other factors being equal--and may be more inclined to 
invest in the firm and accept a lower interest rate or return on the 
firm's obligations. These effects can be more pronounced during a 
financial crisis, particularly if market strains cause credit rating 
agencies to reduce ratings more for firms they believe the government 
would not rescue and if providers of funds seek to reduce their risk 
exposures to firms they believe are not too big to fail. 

Several factors influence the extent to which investors rely on 
ratings. For example, an investor's reliance on credit ratings can 
depend on the extent to which the investor conducts its own credit 
analysis. While representatives of large investment firms with whom we 
spoke said they rely primarily on their own assessments of credit risk 
and do not rely on credit ratings, smaller investors lacking the 
resources to do their own credit analysis may rely more on credit 
ratings and rating agencies' assessments of the impact of possible 
government support on a firm's risk profile. In addition, while an 
investment firm's assessment of government support can be relevant to 
funds that it actively manages, it may not incorporate this factor 
into the investment decisions of funds that it manages using passive 
investment strategies.[Footnote 63] Finally, some representatives of 
large investment firms said that while they do not rely on credit 
ratings for investment decisions, they pay attention to them when 
managing funds for clients whose investments must meet minimum credit 
rating requirements and for clients who may use credit ratings to 
assess their performance. 

Representatives of large investment firms with whom we spoke generally 
said their views on the likelihood of government support do not affect 
their investment decisions. Some representatives of investment firms 
said that while they believe some probability of government rescues 
remains, there is too much uncertainty surrounding future government 
support to factor it into their current investment decisions. Several 
bond investors said it is difficult to distinguish any pricing impacts 
from market expectations of government support from the variety of 
other factors related to firm size that can impact debt pricing and 
investors' investment decisions. For example, compared to smaller 
institutions, large bank holding companies issue bonds more frequently 
and in larger amounts, which increases the liquidity of their bonds. 
Investors may accept lower interest rates on more liquid bonds because 
more liquid bonds can be sold more easily without reducing the price. 
In the section addressing the second objective of this report, we 
analyze the existence and size of potential funding cost advantages 
for the largest bank holding companies using quantitative approaches 
that control for factors outside of government support that can 
influence funding cost differences. 

Financial Contracts That Reference Ratings: 

Higher credit ratings from assumed government support can also benefit 
firms through private contracts that reference credit ratings. For 
example, derivative contracts often tie collateral requirements to a 
firm's credit rating. Representatives of some large bank holding 
companies said that reduced credit ratings would require them to post 
more collateral. Additional collateral requirements would demand 
additional funds that could otherwise be used in other investments. 
The largest bank holding companies disclose information in their 
financial statements about how a credit rating downgrade could cause 
them to post more collateral. While estimates of these collateral 
impacts have varied over time and across firms, several of these firms 
have estimated that a downgrade in their credit rating could require 
them to post between $1 billion and $4 billion of additional 
collateral, depending on the size of the downgrade. Another way that 
private contracts can reference credit ratings is by setting minimum 
credit rating requirements. Examples of such requirements include 
investment funds that cannot purchase securities that are below 
minimum ratings requirements and counterparties that will not accept a 
letter of credit from a bank with a low credit rating. 

Ability to Attract Customers: 

Corporate customers with whom we spoke expressed varying views on the 
degree to which expectations of government support influence their 
banking decisions. Two corporate customers with whom we spoke said 
that they believe the government would intervene to prevent the 
failure of the largest bank holding companies, but that potential 
government support is only one of several factors they consider in 
choosing a bank and is not necessarily a decisive factor. Several 
corporate treasurers identified size-related factors that are 
unrelated to government support that make them more inclined to use 
the largest banks for their banking needs. For example, treasurers of 
global firms noted that the largest U.S. banks have the geographic 
presence and ability to provide funding on the scale they need to 
support their operations around the world. One corporate customer 
noted that although the company's credit facility includes both 
regional banks and some of the largest banks, they tend to use the 
services of large banks more because of their capacity for handling 
large transactions and the variety of their business lines. 

However, while two treasurers said that they tend to select the 
largest U.S. banks primarily for reasons that are unrelated to 
government support, their beliefs about which banks would be rescued 
by the government can impact how they manage their risk exposures to 
banks of different sizes. For example, a treasurer for a large 
domestic corporation said that the possibility of government rescues 
can be a factor when evaluating counterparty risk and the safety of 
deposits. She noted that in normal economic conditions, the likelihood 
of government support for banks is not a significant factor, but when 
markets become strained, her company may reduce its deposits and other 
exposures to regional banks they believe the government would allow to 
fail. Outside of these treasurers, a treasurer from a large global 
company said that potential government support may impact his 
company's banking decisions indirectly through credit ratings. He 
noted that the company uses credit ratings as a factor in assessing a 
bank's creditworthiness and adjusting its exposures to banks. For 
example, if a bank's credit rating falls, the company may reduce its 
intraday exposure to that bank by shifting deposits and other 
exposures away from that bank. A few corporate customers told us they 
do not consider the possibility of government support for large banks 
when they decide how to allocate their banking business. 

Stricter Capital and Other Regulatory Standards May Reduce Risk of 
Failure and Increase Costs for Largest Firms: 

The Dodd-Frank Act imposes new and higher fees on large bank holding 
companies and requires the Federal Reserve Board to subject large bank 
holding companies to enhanced regulatory standards for capital, 
liquidity, and risk management. These enhanced standards may help to 
reduce the likelihood and potential market impacts of the failure of a 
large bank holding company. Taken together, higher fees, stricter 
regulatory standards, and other reforms could increase costs for the 
largest bank holding companies relative to smaller competitors. 
[Footnote 64] New or revised fees and assessments impose higher direct 
costs on bank holding companies with more than $50 billion in total 
assets. 

* Deposit insurance assessments. The Dodd-Frank Act required FDIC to 
change the definition of an insured depository institution's 
assessment base, which can affect the amount of deposit insurance 
assessment the institution pays into the deposit insurance 
fund.[Footnote 65] According to FDIC, this change shifted some of the 
overall assessment burden from smaller banks to larger institutions 
that rely less on deposits but did not affect the overall amount of 
assessment revenue collected. The base was changed from total domestic 
deposits to average consolidated total assets minus average tangible 
equity. The largest bank holding companies generally saw the largest 
percentage increases in their deposit insurance assessments because 
they rely less on domestic deposits for their funding than smaller 
institutions. One of the largest bank holding companies reported that 
the change to the assessment calculation resulted in a $600 million 
increase in its deposit insurance assessments in 2011. In the quarter 
after the rule became effective, those banks with less than $10 
billion in assets saw a 33 percent drop in their assessments (from 
about $1 billion to about $700 million), while those banks with over 
$10 billion in assets saw a 17 percent rise in their assessments (from 
about $2.4 billion to about $2.8 billion). 

* Fees on SIFIs. In addition, the Dodd-Frank Act directs the Federal 
Reserve Board to collect fees from bank SIFIs equal to the expenses 
the Federal Reserve Board estimates are necessary or appropriate to 
carry out its supervision and regulation of those companies.[Footnote 
66] In addition, the Dodd-Frank Act directs Treasury to collect 
assessments from bank and nonbank SIFIs to fund the operations of the 
Office of Financial Research. These assessments totaled $137 million 
in 2012 and $35 million in 2013. 

The Dodd-Frank Act requires the Federal Reserve Board to subject large 
bank holding companies to heightened standards for capital, liquidity, 
and stress testing, as well as other provisions, all of which could 
reduce the risk of their failure and the costs that their distress 
could impose on the financial system. Following Dodd-Frank enactment, 
bank SIFIs significantly increased their capital and liquidity in 
advance of finalization of new rules for capital, leverage, and 
liquidity standards. As of December 31, 2013, the six largest U.S. 
GSIBs had an average tier 1 common equity capital ratio of 12.1 
percent, compared to the 4.5 percent minimum required under Basel III 
and an average of 8.4 percent among these firms as of December 31, 
2009.[Footnote 67] In addition, pursuant to the Dodd-Frank Act, the 
Federal Reserve Board conducts stress testing and evaluates the 
capital planning process of large bank holding companies to help 
ensure these firms are resilient to periods of economic or financial 
stress. In the most recent round of capital planning reviews, the 
Federal Reserve Board rejected the capital plan of one U.S. GSIB and 
required another to resubmit its capital plan after errors were 
discovered.[Footnote 68] Pending approval of their revised capital 
plans, the Federal Reserve Board did not allow proposed actions by 
these firms, such as dividend increases, that would have reduced their 
capital. In April 2014, U.S. bank regulators adopted a new rule that 
strengthens the leverage ratio standards for the largest, most 
interconnected U.S. banking organizations.[Footnote 69] 

Beyond the new rules and regulatory reviews to ensure capital 
adequacy, the Federal Reserve Board has indicated that eight of the 
largest U.S. bank holding companies will be subject to a capital 
surcharge--an increase in their risk-based capital requirement--based 
on their size, complexity, and interconnectedness.[Footnote 70] 
Federal Reserve Board officials have stated that the capital surcharge 
is intended to force the largest bank holding companies to internalize 
the costs they could impose on the financial system from their 
systemic footprint. Federal Reserve Board and Treasury officials said 
that this capital surcharge could also help to offset any funding cost 
advantages that remain from market perceptions that the government 
would not allow the largest bank holding companies to fail. 

Higher capital and liquidity requirements for banks can increase their 
funding and other costs. For example, higher capital requirements can 
require banks to increase the portion of their funding that comes from 
equity capital rather than debt, which can increase funding costs. 
[Footnote 71] In prior work, we have summarized the results of studies 
by the Bank for International Settlements and others on the benefits 
and costs of increasing capital requirements for banks, but these 
studies generally estimated cost impacts to the economy rather than 
the incidence of increased costs for the institutions themselves. 
While banks can respond to additional costs in a variety of ways, 
including passing on some costs to borrowers by charging higher 
interest rates on loans, a Federal Reserve Board official noted that 
costs associated with the GSIB capital surcharges--which will not 
apply to most banks and will not apply evenly among the GSIBs--may be 
more difficult for the largest bank holding companies to pass on to 
customers. 

Dodd-Frank also imposes additional compliance costs on large bank 
holding companies required to comply with other Dodd-Frank provisions. 
For example, as previously discussed, Dodd-Frank requires SIFIs to 
periodically submit resolution plans to the Federal Reserve and FDIC, 
as well as to conduct company-run stress tests semiannually. 
Regulators and industry officials have stated that SIFIs have devoted 
significant staffing resources to developing these resolution plans. 
According to industry representatives, stress testing requires newly 
covered firms to incur significant compliance costs associated with 
building information systems, contracting with outside vendors, 
recruiting personnel, and developing stress testing models that are 
unique to their organization. Furthermore, changes to the market 
infrastructure for swaps--such as clearing and exchange-trading 
requirements--and real-time reporting requirements for designated 
major swap dealers or major swap participants will require firms to 
purchase or upgrade information systems. Industry representatives and 
regulators said that while some compliance costs of the derivatives 
reforms could be recurring, a large part of these costs will come from 
one-time upfront investments to update processes and technology. 
Additionally, by generally prohibiting banks from engaging in 
proprietary trading and limiting their ability to sponsor or invest in 
hedge and private equity funds, the Volcker rule restrictions could 
eliminate past sources of trading and fee income for some banks. 

As we have noted in prior work, measuring the costs of financial 
regulation is challenging because of the multitude of intervening 
variables, the complexity of the financial system, and data 
limitations.[Footnote 72] For example, the extent to which regulated 
institutions pass on a portion of their increased costs to their 
customers may be impacted by competitive forces or other factors. 
Other sources of uncertainty, such as the potential for regulatory 
arbitrage, add to the challenges of estimating the act's potential 
costs. For example, increased regulation could cause certain financial 
activities in the United States to move to foreign jurisdictions with 
less stringent regulations. U.S. regulators have acknowledged the 
importance of harmonizing international regulatory standards and noted 
that it can be advantageous for the United States to be the leader in 
implementing new regulatory safeguards. 

Agencies See Progress in Addressing Too-Big-to-Fail Perceptions and 
Expect Further Progress as Reforms Are More Fully Implemented: 

Officials from FSOC and its member agencies have stated that financial 
reforms have not completely removed too-big-to-fail perceptions but 
have made significant progress toward doing so. In a December 2013 
speech, Treasury Secretary Jack Lew said there is growing recognition 
of the Dodd-Frank reforms and that market analysts are factoring them 
into their assumptions. However, he noted that there is still more 
work to be done. 

Under the Dodd-Frank Act, FSOC is, among other things, charged with 
promoting market discipline by eliminating expectations on the part of 
shareholders, creditors, and counterparties of large bank holding 
companies that the U.S. government will shield them from losses in the 
event of failure. FSOC and its member agencies monitor progress in 
addressing expectations of government support primarily through 
monitoring progress in implementing relevant Dodd-Frank reforms. 
FSOC's 2014 annual report includes a discussion of progress made on 
OLA, enhanced prudential standards, and other relevant reforms. 
According to Treasury officials, several key areas require continued 
progress: 

* International regulatory reform. In its 2013 annual report, FSOC 
writes that international coordination of financial regulation is 
essential to mitigate threats to financial stability. FDIC officials 
said they continue to work with foreign regulators to address issues 
related to creating a viable process for effecting the orderly 
resolution of a failing financial institution with significant cross-
border activities. For example, FDIC is working with foreign 
counterparts on changes needed to ensure that derivatives contracts 
under other countries' laws include a stay similar to that which 
applies to U.S. contracts under Dodd-Frank to prevent termination of 
these contracts by counterparties of a firm pulled into resolution. 
Federal Reserve Board staff said U.S. regulators are considering steps 
that may be needed to help ensure that foreign regulators do not take 
disruptive actions with respect to foreign operations of a U.S. firm 
pulled into resolution. They noted that global U.S. SIFIs may need to 
create intragroup loss absorbency arrangements that provide clarity 
and assurance to foreign regulators about how loss absorbency from the 
U.S. holding company will be made available to support foreign 
operations during a resolution. 

* The Federal Reserve's long-term debt requirement. The Federal 
Reserve Board has identified the implementation of a long-term debt 
requirement as a regulatory priority that it and other agencies are 
actively considering. In testimony before the Senate Banking 
Committee, Federal Reserve Board Governor Daniel Tarullo said that 
successful resolution without taxpayer assistance would be most 
effectively accomplished if a firm has sufficient long-term, unsecured 
debt to absorb additional losses and to recapitalize the business 
transferred to a bridge operating company. 

* General education of market participants on reforms. Treasury 
officials identified the education of market participants as a key 
area for progress. Public outreach and education often take the form 
of speeches from agency officials and meetings with industry 
stakeholders. Regulators also solicit feedback on proposed rulemakings 
and regulations during public comment periods. For example, on 
December 18, 2013, FDIC published a public notice on the framework for 
a SPOE approach for resolution of failed financial institutions under 
OLA and solicited comments from the public through February 18, 2014, 
before subsequently extending the comment period through March 20, 
2014.[Footnote 73] 

Treasury officials also monitor market trends and outside research to 
inform their assessment of progress in addressing too-big-to-fail 
perceptions. Treasury staff have looked at trends in bond prices, 
credit-default-swap prices, and other market data for bank holding 
companies of different sizes for evidence that investors have reduced 
their expectations of government support. Treasury staff also monitor 
relevant outside research, including a growing body of research by 
academics and others that has used quantitative approaches to analyze 
the existence and size of potential funding cost advantages that the 
largest bank holding companies could receive because of market 
expectations of government support. The next section of this report 
includes a summary of selected studies in this literature and 
discusses the strengths and limitations of the methods they use. FSOC 
and Treasury staff have reviewed these studies and noted that while 
the studies have limitations, their findings are consistent with a 
reduction in expectations of government support following the Dodd-
Frank Act. 

Evidence Suggests Large Banks Had a Funding Cost Advantage over Small 
Ones during the Financial Crisis That May Have Declined or Reversed 
Since: 

Our analysis and the results of studies we reviewed provide evidence 
that the largest bank holding companies had lower funding costs than 
smaller bank holding companies during the 2007-2009 financial crisis 
but that differences may have declined or reversed in more recent 
years. To inform our econometric approach, we reviewed studies that 
estimated funding cost differences between large and small financial 
institutions that could be associated with the perception that some 
institutions are too big to fail. Studies we reviewed generally found 
that the largest financial institutions had lower funding costs during 
the 2007-2009 financial crisis but that the difference between the 
funding costs of the largest and smaller financial institutions has 
since declined. In some cases these findings could be interpreted as 
evidence of advantages driven by too-big-to-fail perceptions; however, 
these empirical analyses are imperfect and contain a number of 
limitations that could reduce their validity or applicability to U.S. 
bank holding companies. Our analysis, which addresses certain 
limitations of these studies, also provides evidence that large or 
systemically important bank holding companies had lower funding costs 
than smaller bank holding companies during the 2007-2009 financial 
crisis, which may have been associated with expectations of government 
assistance. In addition, our analysis provides some evidence that 
funding cost differences may have declined or reversed in recent years 
and that large bank holding companies may have had higher funding 
costs since the crisis. However, we also analyzed what funding cost 
differences might have been since the crisis in hypothetical scenarios 
where levels of credit risk in every year from 2010 to 2013 are 
assumed to be as high as they were during the financial crisis. This 
analysis suggests that large bank holding companies might have had 
lower funding costs than smaller bank holding companies since the 
crisis if levels of credit risk had remained high, indicating that 
changes in funding cost differences over time may be due in part to 
improvements in bank holding companies' financial conditions. Although 
our analysis improves on certain aspects of prior studies, important 
limitations remain and our results should be interpreted with caution. 

Studies Generally Found the Largest Banks Had Lower Funding Costs 
during the Financial Crisis but Results Have Limitations: 

Studies we reviewed generally found that the largest financial 
institutions had lower funding costs than smaller ones during the 2007-
2009 financial crisis but that the difference between the funding 
costs of the largest and smaller financial institutions has since 
declined. In some cases these findings could be interpreted as 
evidence of advantages driven by too-big-to-fail perceptions; however, 
these empirical analyses are imperfect and contain a number of 
limitations that could reduce their validity or applicability to U.S. 
bank holding companies. 

We reviewed studies that estimated the funding cost difference between 
large and small financial institutions that could be associated with 
the perception that some institutions are too big to fail. We 
evaluated studies that met the following criteria: (1) used a 
comparative empirical approach that attempted to account for 
differences across financial institutions that could influence funding 
costs, (2) included U.S. bank holding companies, and (3) included 
analysis of data from 2002 or later. See our scope and methodology 
section for more information on our criteria and approach. 

The 16 studies we reviewed made a wide variety of methodological 
decisions and came to a range of conclusions. We present the variety 
of methodological decisions along two key dimensions--which source of 
funding is analyzed (e.g., deposits, bonds) and time period of 
analysis--in table 2 below. The source of funding that is analyzed is 
an important methodological decision because investors may have 
differing expectations regarding the likelihood that various sources 
of funding might receive government support, and these expectations 
could differ by the size of the financial institution. Results could 
differ across studies because of differences in creditor priority 
(subordinated debt versus senior debt) or maturity (bonds that mature 
several years in the future versus deposits that can be demanded at 
any time). We also include information in table 2 on the reported 
affiliations of the study authors. 

Table 2: Empirical Studies of Too-Big-to-Fail Funding Cost Differences: 

Study authors (Year): Acharya, Anginer & Warburton (2013); 
Measure of funding cost: Bonds; 
Time period: 1990-2011; 
Affiliation category: Academic. 

Study authors (Year): Araten & Turner (2012); 
Measure of funding cost: Combination of funding sources; 
Time period: 2002-2011; 
Affiliation category: Private[A]. 

Study authors (Year): Balasubramnian & Cyree (2014); 
Measure of funding cost: Bonds[B]; 
Time period: 2009-2011; 
Affiliation category: Academic. 

Study authors (Year): Balasubramnian & Cyree (2013); 
Measure of funding cost: Bonds[C]; 
Time period: 2009-2011; 
Affiliation category: Academic. 

Study authors (Year): Barth & Schnabel (2013); 
Measure of funding cost: Credit default swaps; 
Time period: 2005-2011; 
Affiliation category: Academic. 

Study authors (Year): Bertay, Demirgüç-Kunt & Huizinga (2013); 
Measure of funding cost: Combination of funding sources; 
Time period: 1990-2011; 
Affiliation category: Academic and public. 

Study authors (Year): International Monetary Fund (2014); 
Measure of funding cost: Credit ratings, credit default swaps; 
Time period: 2005-2014; 
Affiliation category: Public. 

Study authors (Year): Jacewitz & Pogach (2012); 
Measure of funding cost: Deposits; 
Time period: 2005-2010; 
Affiliation category: Public. 

Study authors (Year): Keppo & Yang (2013); 
Measure of funding cost: Bonds, deposits; 
Time period: 1990-2011; 
Affiliation category: Academic and public. 

Study authors (Year): Kumar & Lester (2014a); 
Measure of funding cost: Deposits; 
Time period: 2006-2012; 
Affiliation category: Private[D]. 

Study authors (Year): Kumar & Lester (2014b); 
Measure of funding cost: Bonds; 
Time period: 2009-2013; 
Affiliation category: Private[D]. 

Study authors (Year): Li et al. (2011); 
Measure of funding cost: Credit default swaps; 
Time period: 2001-2010; 
Affiliation category: Private. 

Study authors (Year): Santos (2014); 
Measure of funding cost: Bonds; 
Time period: 1985-2009; 
Affiliation category: Public. 

Study authors (Year): Tsesmelidakis & Merton (2012); 
Measure of funding cost: Credit default swaps; 
Time period: 2002-2010; 
Affiliation category: Academic and private. 

Study authors (Year): Ueda & Di Mauro (2012); 
Measure of funding cost: Credit ratings; 
Time period: 2007, 2009; 
Affiliation category: Academic and public. 

Study authors (Year): Völz & Wedow (2009); 
Measure of funding cost: Credit default swaps; 
Time period: 2002-2007; 
Affiliation category: Public. 

Source: GAO. GAO-14-621. 

Note: We read and evaluated studies that met the following criteria: 
(1) used a regression or similar comparative empirical methodology to 
estimate any funding cost difference, (2) included U.S. bank holding 
companies, and (3) included analysis of data from 2002 or later. Our 
criteria excluded studies that used option-pricing approaches--this 
methodology assumes a too-big-to-fail funding cost advantage exists 
and only estimates its magnitude. We identified author affiliations 
based on information included in the papers, which may not reflect all 
relevant affiliations. Academic affiliation includes authors at 
universities or research institutes. Private affiliation includes 
authors at financial institutions and other private firms. Public 
affiliation includes authors at government and regulatory agencies 
(including the regional Federal Reserve banks which are private 
corporations), and international financial institutions such as the 
World Bank. We also excluded two studies that otherwise met our 
criteria but did not attempt to control for important differences 
between financial institutions. We were aware of potential conflicts 
of interest associated with a number of studies in our review. As with 
other studies we reviewed, these studies must have attempted to 
address factors that might account for differences in funding costs in 
order to be included, and we also considered the potential impact 
these conflicts of interest might have on their methods and results. 
See our scope and methodology for more details on our approach. 

[A] This study was conducted by researchers at JP Morgan Chase. 

[B] This study focused on subordinated bonds. 

[C] This study focused on senior bonds. 

[D] This study was conducted by researchers at Oliver Wyman and 
sponsored by the Clearinghouse Association, a trade association of the 
world's largest commercial banks. 

[End of table] 

Studies we reviewed generally found that the largest financial 
institutions had lower funding costs than smaller ones during the 2007-
2009 financial crisis, but that the difference between the funding 
costs of the largest and smaller financial institutions has since 
declined. For example, one study estimated that large U.S. financial 
institutions had roughly 100 basis points lower bond funding costs 
than smaller ones in 2009, but this difference had declined to around 
40 basis points by 2011.[Footnote 74] Similarly, a study of U.S. bank 
credit default swaps found that large U.S. bank holding companies had 
roughly 100 basis points lower funding costs in 2009, but this 
difference had declined to around 15 basis points in 2013.[Footnote 
75] In some cases these differences could be interpreted as evidence 
of funding cost advantages driven by too-big-to-fail perceptions. In 
other cases, limitations in the studies make it difficult to eliminate 
other explanations of why funding cost differences might exist--such 
as greater liquidity or diversification that could be associated with 
size or spurious results driven by imperfect measures of funding 
costs. Time period of analysis was another important difference across 
studies we reviewed. Few studies in our review included data beyond 
2011. Therefore, most results may not reflect recent changes in the 
regulatory environment and market expectations discussed earlier in 
the report. 

Studies also varied in their approach to identifying financial 
institutions that might be perceived as too big to fail, using a 
variety of size and other thresholds. For example, some studies 
measured too-big-to-fail status by a bank's assets; however, the 
threshold between too-big-to-fail and other banks varied from $50 
billion to $500 billion. Several papers estimated too-big-to-fail 
status by size relative to industry, such as the largest 20 banks or 
top 10 percent by assets. These different approaches indicate that 
there is no consensus within the literature on which financial 
institutions may be considered too big to fail for the purposes of 
comparing funding costs. 

The studies we reviewed can be grouped into categories based on their 
approaches. While all studies included in our review used standard 
approaches and attempted to address factors that might account for 
differences in funding costs, these empirical analyses remain 
imperfect. 

* Regression. Most studies we reviewed adopted a regression 
methodology in which some measure of funding costs was explained by a 
variety of control variables, such as risk, liquidity, and maturity, 
to attempt to account for differences across financial institutions. 
These models are standard empirical tools and are flexible in terms of 
the information about financial institutions and markets that they can 
incorporate. In some instances these models rely on a small number of 
indicators that may only imperfectly measure underlying default risks. 
As a result, some analyses may not correctly estimate the size of any 
too-big-to-fail advantages because they omit important factors that 
influence funding costs. In other studies that account for a more 
thorough set of factors that influence funding costs, results may be 
sensitive to alternative measurements of these factors.[Footnote 76] 
For example, default risk is an underlying driver of funding costs, 
and studies may produce different results by using a bank's earnings 
volatility as an indicator for default risk as opposed to other 
indicators such as the quality of a bank's assets. In addition, 
liquidity is another important factor to account for when attempting 
to explain funding cost differences--investors charge banks more for 
less liquid sources of funding--and some studies do not adequately 
control for the liquidity of the funding source. Challenges similar to 
those involved in accurately capturing default risk arise in finding 
appropriate indicators for a bond's liquidity. 

* Equity-based. Three papers we reviewed measured the difference 
between observed credit default swap spreads (which approximate bond 
funding costs) and hypothetical credit default swap spreads (which are 
estimated based on information implied by equity prices).[Footnote 77] 
This approach estimates hypothetical spreads with a standard 
theoretical model used in finance that uses the risk of a firm's 
equity to estimate the risk of a firm's debt.[Footnote 78] In doing 
so, the approach assumes that hypothetical spreads derived from equity 
prices are not influenced by any expectations of government support, 
but that observed credit default swap spreads are influenced by such 
expectations. By comparing the two spreads the approach can estimate 
the magnitude of expectations of government support. While this 
approach has some advantages, it relies on critical assumptions about 
how a limited number of factors influence the risk of default. 
[Footnote 79] As a result, these analyses may also omit important 
factors that influence funding costs, such as earnings.[Footnote 80] 

* Ratings-based. Two papers used Fitch credit ratings to estimate the 
funding cost difference that could be associated with potential 
government support.[Footnote 81] Models based on credit ratings offer 
a convenient way to incorporate all the factors the rating agency 
considers relevant to default risk and take advantage of the rating 
agency's explicit separation of the impact of expected government 
support through, for example, the assignment of a standalone credit 
rating (assuming no government support) and a higher credit rating 
assuming government support.[Footnote 82] However, this approach 
assumes that all information about market expectations of default risk 
and government support are incorporated into credit ratings, which is 
a potentially weak assumption. Credit ratings had a limited impact on 
the views of large investors we interviewed, as previously discussed. 
Moreover, funding costs vary for firms within a particular rating. As 
a result, these studies may estimate funding costs with considerable 
error. Finally, results of these studies are sensitive to the credit 
rating agency used--for example, results based on Moody's ratings 
could be quite different than other rating agencies because Moody's 
removed expectations of government support for U.S. bank holding 
companies in 2013. 

In addition to the approach-specific limitations, a number of general 
limitations related to implementation of the various approaches exist 
across studies we reviewed that could reduce their validity or 
applicability to U.S. bank holding companies. For example, studies 
varied in the countries that were included in the analysis--some 
studies focused on the United States, while others included a broad 
cross-section of more than 20 countries. Studies that pooled a large 
number of countries in their analysis have results that may not be 
applicable to U.S. bank holding companies. For example, studies that 
included Switzerland and Iceland in their analyses may not apply to 
the United States because banking sectors in those countries are much 
larger relative to the economy. As noted above, because few studies 
included data past 2011, results may not reflect recent changes in the 
regulatory environment and market sentiment; for example, the Federal 
Reserve's rule for enhanced prudential standards for large bank 
holding companies and FDIC's proposed strategy for orderly liquidation. 

As a result of the limitations associated with these methodological 
choices, estimates of the size of the funding cost difference 
associated with a too-big-to-fail advantage based on this literature--
while suggestive of general trends--are not definitive and should be 
interpreted with caution. 

GAO Analysis Suggests Large Banks Had Lower Funding Costs during the 
Financial Crisis but This Advantage May Have Declined or Reversed 
Since: 

We conducted our own analysis to assess the extent to which the 
largest bank holding companies have had lower funding costs as a 
result of perceptions that the government would not allow them to 
fail. Overall, our analysis provides some evidence that large or 
systemic bank holding companies had lower funding costs than smaller 
ones during the 2007-2009 financial crisis that may have been 
associated with expectations of government assistance. Our analysis 
provides only limited evidence that large bank holding companies had 
lower funding costs since the crisis and instead provides some 
evidence that the opposite may have been true at the levels of credit 
risk that prevailed in those years. However, in hypothetical scenarios 
where levels of credit risk in every year from 2010 to 2013 are 
assumed to be as high as they were during the financial crisis, our 
analysis suggests that large bank holding companies might have had 
lower funding costs than smaller bank holding companies. Although our 
analysis improves on certain aspects of prior studies, important 
limitations remain and our results should be interpreted with caution. 

To conduct our analysis, we developed a series of econometric models--
models that use statistical techniques to estimate the relationships 
between quantitative economic and financial variables--based on our 
assessment of relevant studies and expert views. These models estimate 
the relationship between bank holding companies' bond funding costs 
and their size, while also controlling for other drivers of bond 
funding costs, including credit risk and bond liquidity. Key features 
of our econometric approach include the following: 

* U.S. bank holding companies. To better understand the relationship 
between bank holding company funding costs and size in the context of 
the U.S. economic and regulatory environment, we only analyzed U.S. 
bank holding companies. In contrast, some of the literature we 
reviewed analyzed nonbank financial companies and foreign companies. 

* 2006-2013 time period. To better understand the relationship between 
bank holding company funding costs and size in the context of the 
current economic and regulatory environment, we analyzed the period 
from 2006 through 2013, which includes the recent financial crisis as 
well as years before the crisis and following the enactment of the 
Dodd-Frank Act. In contrast, some of the literature we reviewed did 
not analyze data in the years after the financial crisis. 

* Bond funding costs. We used bond yield spreads as our measure of 
bank holding company funding costs because they are a direct measure 
of what investors charge bank holding companies to borrow money and 
because they are sensitive to credit risk and hence expected 
government support. This indicator of funding costs has distinct 
advantages over certain other indicators used in studies we reviewed, 
including credit ratings, which do not directly measure funding costs, 
and total interest expense, which mixes the costs of funding from 
multiple sources. 

* Alternative measures of size. Size or systemic importance can be 
measured in multiple ways, as reflected in our review of the 
literature. Based on that review and the comments we received from 
external reviewers, we used four different measures of size or 
systemic importance: total assets, total assets and the square of 
total assets, whether or not a bank holding company was designated a 
GSIB by the Financial Stability Board in November 2013, and whether or 
not a bank holding company had assets of $50 billion or more. 

* Extensive controls for bond liquidity, credit risk, and other key 
factors. To account for the many factors that could influence funding 
costs, we controlled for credit risk, bond liquidity, and other key 
factors in our models. We included a number of variables that are 
associated with the risk of default, including measures of capital 
adequacy, asset quality, earnings, and volatility. We also included a 
number of variables that can be used to measure bond liquidity. 
Finally, we included variables that measure other key characteristics 
of bonds, such as time to maturity, and key characteristics of bank 
holding companies, such as operating expenses. Our models include a 
broader set of controls for credit risk and bond liquidity than some 
studies we reviewed and, as we discuss later, we directly assess the 
sensitivity of our results to using alternative controls on our 
estimates of funding costs. 

* Multiple model specifications. In order to assess the sensitivity of 
our results to using alternative measures of size, bond liquidity, and 
credit risk discussed above, we estimated multiple different model 
specifications. We developed models using four alternative measures of 
size, two alternative sets of measures of capital adequacy, six 
alternative measures of volatility, and three alternative measures of 
bond liquidity to assess the impact of using alternative measures on 
our results. In contrast, some of the studies we reviewed estimated a 
more limited number of model specifications. 

* Annual estimates of models. To allow for changes in investors' 
beliefs about the likelihood of government rescues between the years 
of the financial crisis--when emergency government programs designed 
to assist financial institutions were available--and the years 
following the crisis, our models allow the relationship between bank 
holding company funding costs and size to vary over time. In contrast, 
some of the studies we reviewed assumed that the relationship between 
bank holding company funding costs and size was constant over time. 

* Link between size and credit risk. To account for the possibility 
that investors' beliefs about government rescues affect their 
responsiveness to credit risk, our models allow the relationships 
between bank holding company funding costs and credit risk to depend 
on size. 

Altogether, we estimated 42 different models for each year from 2006 
through 2013 and then used these models to compare bond yield spreads--
our measure of bond funding costs--for bank holding companies of 
different sizes but with the same level of credit risk.[Footnote 83] 
Figure 1 shows our models' comparisons of the difference between bond 
funding costs for bank holding companies with $1 trillion in assets 
and average credit risk and bond funding costs for similar bank 
holding companies with $10 billion in assets, for each model and for 
each year.[Footnote 84] Each circle and dash in figure 1 shows the 
comparison of bond funding costs for a different model. Circles show 
model-estimated differences that were statistically significant at the 
10 percent level, while dashes represent differences that were not 
statistically significant at that level.[Footnote 85] Circles and 
dashes below zero correspond to models suggesting that bank holding 
companies with $1 trillion in assets have lower bond funding costs 
than bank holding companies with $10 billion in assets, and vice versa. 

Figure 1: Estimates from 42 Models of Average Bond Funding Cost 
Differences between Bank Holding Companies with $1 Trillion and $10 
Billion in Assets, 2006-2013: 

[Refer to PDF for image: illustration] 

Models predicting statistically significant differences above zero: 
2006: 0; 
2007: 0; 
2008: 0; 
2009: 0; 
2010: 2; 
2011: 11; 
2012: 15; 
2013: 18. 

Models predicting statistically insignificant differences above zero: 
2006: 4; 
2007: 7; 
2008: 0; 
2009: 0; 
2010: 12; 
2011: 13; 
2012: 8; 
2013: 12. 

Models predicting statistically significant differences below zero: 
2006: 28; 
2007: 20; 
2008: 39; 
2009: 37; 
2010: 9; 
2011: 14; 
2012: 4; 
2013: 8. 

Models predicting statistically insignificant differences below zero: 
2006: 10; 
2007: 15; 
2008: 3; 
2009: 5; 
2010: 19; 
2011: 4; 
2012: 15; 
2013: 4. 

Estimated average bond funding costs for larger bank holding companies 
(BHC) were higher than for smaller BHCs. 

Estimated average bond funding costs for larger BHCs were lower than 
for smaller BHCs. 

Basis point = .01% (1/100th of a percent). 

Depicted on graph: 
Average difference estimated by one of 42 models (statistically 
significant at 10% level); 
Average difference estimated by one of 42 models (not statistically 
significant at 10% level). 

Source: GAO analysis of data from Bloomberg and the Financial 
Stability Board. GAO-14-621. 

Notes: We estimated econometric models of the relationship between 
bank holding company size and funding costs using data for U.S. bank 
holding companies and their outstanding senior unsecured bonds for the 
first quarter of 2006 through the fourth quarter of 2013. The models 
used bond yield spreads to measure funding costs and controlled for 
credit risk factors such as capital adequacy, asset quality, earnings, 
maturity mismatch, and volatility, as well as bond liquidity and other 
characteristics of bonds and bank holding companies that can affect 
funding costs. We estimated 42 models for each year from 2006 through 
2013 to assess the sensitivity of estimated funding cost differences 
to alternative measures of capital adequacy, volatility, bond 
liquidity, and size or systemic importance. We used the models to 
compare bond funding costs for bank holding companies of different 
sizes but the same levels of credit risk, bond liquidity, and other 
characteristics. This figure compares bond funding costs for bank 
holding companies with $1 trillion and $10 billion in assets, for each 
model and for each year, with average levels of credit risk. Each 
circle and dash shows the comparison for a different model, where 
circles and dashes below zero suggest bank holding companies with $1 
trillion in assets have lower bond funding costs than bank holding 
companies with $10 billion in assets, and vice versa. 

[End of figure] 

Our analysis provides evidence that the largest bank holding companies 
had lower funding costs during the 2007-2009 financial crisis but that 
differences may have declined or reversed in recent years. However, we 
found that the outcomes of our econometric models varied with the 
various controls we used to capture size, credit risk, and bond 
liquidity. This variation indicates that uncertainty related to how to 
model funding costs has an important impact on estimated funding cost 
differences between large and small bank holding companies. As figure 
1 shows, most models found that larger bank holding companies had 
lower bond funding costs than smaller bank holding companies during 
the 2007-2009 financial crisis, but the magnitude of the difference 
varied widely across models, as indicated by the range of results for 
each year. For example, for 2008, our models suggest that bond funding 
costs for bank holding companies with $1 trillion in assets and 
average credit risk were from 17 to 630 basis points lower than bond 
funding costs for similar bank holding companies with $10 billion in 
assets. For 2009, our models suggest that bond funding costs for bank 
holding companies with $1 trillion in assets and average credit risk 
were from 30 to 628 basis points lower than bond funding costs for 
similar bank holding companies with $10 billion in assets. 

Our models' comparisons of bond funding costs for different-sized bank 
holding companies for 2010 through 2013 also vary widely. For bank 
holding companies with average credit risk, more than half of our 
models suggest that larger bank holding companies had higher bond 
funding costs than smaller bank holding companies from 2011 through 
2013, but many models suggest that larger bank holding companies still 
had lower bond funding costs than smaller ones during this period. For 
example, for 2013, our models suggest that bond funding costs for 
average credit risk bank holding companies with $1 trillion in assets 
ranged from 196 basis points lower to 63 basis points higher than bond 
funding costs for similar bank holding companies with $10 billion in 
assets (see figure 1). For 2013, thirty of our models suggest that the 
larger banks had higher funding costs, and 12 of our models suggest 
that the larger banks had lower funding costs. 

Our models' comparisons were particularly sensitive to the measure of 
size or systemic importance we used. For example, for 2013, models 
that used total assets as the indicator of size or systemic importance 
suggest that bond funding costs for bank holding companies with $1 
trillion in assets and average credit risk ranged from 23 to 59 basis 
points higher than bond funding costs for similar bank holding 
companies with $10 billion in assets (see fig. 2). Models that used 
the GSIB designation suggest that bond funding costs for bank holding 
companies with $1 trillion ranged from 11 basis points lower to 36 
basis points higher than bond funding costs for bank holding companies 
with $10 billion in assets. Models that used total assets and the 
square of total assets--which allows for a nonlinear relationship 
between size and yield spreads--suggest that bond funding costs for 
bank holding companies with $1 trillion in assets ranged from 52 basis 
points lower to 4 basis points higher than bond funding costs for bank 
holding companies with $10 billion in assets. Finally, models that 
used whether or not a bank holding company had $50 billion in assets 
suggest that bond funding costs for bank holding companies with $1 
trillion ranged from 161 to 196 basis points lower than bond funding 
costs for bank holding companies with $10 billion in assets. 

Figure 2: Difference in Estimated Bond Funding Costs for Average 
Credit Risk Bank Holding Companies with $1 Trillion versus $10 Billion 
in Assets by Indicator of Size or Systemic Importance, 2013: 

[Refer to PDF for image: illustration] 

Estimated average bond funding costs for larger bank holding companies 
(BHC) were higher than for smaller BHCs. 

Estimated average bond funding costs for larger BHCs were lower than 
for smaller BHCs. 

Basis point = .01% (1/100th of a percent). 

Total assets: 
Total statistically significant number above/below zero: 5/0; 
Total statistically insignificant number above/below zero: 1/0. 

Global systemically important bank designation: 
Total statistically significant number above/below zero: 2/0; 
Total statistically insignificant number above/below zero: 3/1. 

Total assets and total assets squared: 
Total statistically significant number above/below zero: 0/3; 
Total statistically insignificant number above/below zero: 1/2. 

Systemically important financial institution (total assets $50 billion
or more): 
Total statistically significant number above/below zero: 0/5; 
Total statistically insignificant number above/below zero: 0/1. 

Depicted on graph: 
Average difference estimated by one of 42 models (statistically 
significant at 10% level); 
Average difference estimated by one of 42 models (not statistically 
significant at 10% level). 

Source: GAO analysis of data from Bloomberg and the Financial 
Stability Board. GAO-14-621. 

Notes: We estimated econometric models of the relationship between 
bank holding company size and funding costs using data for U.S. bank 
holding companies and their outstanding senior unsecured bonds for the 
first quarter of 2006 through the fourth quarter of 2013. The models 
used bond yield spreads to measure funding costs and controlled for 
credit risk factors such as capital adequacy, asset quality, earnings, 
maturity mismatch, and volatility, as well as bond liquidity and other 
characteristics of bonds and bank holding companies that can affect 
funding costs. We estimated multiple models for each year from 2006 
through 2013 to assess the sensitivity of estimated funding cost 
differences to alternative measures of capital adequacy, volatility, 
bond liquidity, and size or systemic importance. We used the models to 
compare bond funding costs for bank holding companies of different 
sizes but the same levels of credit risk, bond liquidity, and other 
characteristics. This figure compares bond funding costs for bank 
holding companies with $1 trillion and $10 billion in assets with 
average levels of credit risk for 2013 for models that differ in how 
they measure size. Each circle and dash shows the comparison for a 
different model, where circles and dashes below zero suggest bank 
holding companies with $1 trillion in assets have lower bond funding 
costs than bank holding companies with $10 billion in assets, and vice 
versa. 

[End of figure] 

We also found that our models' comparisons of bond funding costs for 
bank holding companies of different sizes varied depending on bank 
holding companies' level of credit risk. For low credit risk bank 
holding companies, most models suggest that bond funding costs for 
larger bank holding companies were lower than bond funding costs for 
smaller bank holding companies during the 2007-2009 financial crisis 
and that bond funding costs for larger bank holding companies were 
higher than bond funding costs for smaller bank holding companies from 
2010 through 2013. Depending on the year, 25 to 42 of our 42 models 
suggest that bond funding costs for larger, low risk bank holding 
companies were lower during the financial crisis, while 31 to 41 of 
our 42 models suggest that bond funding costs for larger low risk bank 
holding companies were higher following the crisis. However, for high 
credit risk bank holding companies, most models suggest that bond 
funding costs for larger bank holding companies were lower than bond 
funding costs for smaller bank holding companies in every year (28 to 
41 of our 42 models, depending on the year). 

Given that most models suggest that large bank holding companies' bond 
funding costs are typically lower than small bank holding companies' 
bond funding costs at high levels of credit risk, this suggests that 
size-related funding cost differences that favor large bank holding 
companies are more likely to emerge when the likelihood that a bank 
holding company will fail increases. As we discuss later, investors' 
overall beliefs about the likelihood of government support may have 
several components, including beliefs about whether or not a bank 
holding company will fail--which is related to its credit risk--and 
beliefs about whether or not a bank holding company will be supported 
by the government if it fails. For example, investors may believe that 
larger bank holding companies are more likely to be supported than 
smaller bank holding companies in the event of failure, but investors 
may also believe that all bank holding companies are relatively safe 
and unlikely to fail. In this case, investors' overall expectations of 
government support for all bank holding companies are likely to be 
low, and differences in funding costs due to varying expectations of 
government support for bank holding companies of different sizes are 
likely to be small or nonexistent. In contrast, if investors believe 
that all bank holding companies are risky and prone to fail, then 
investors' overall expectations of government support for larger bank 
holding companies are likely to be higher than for smaller bank 
holding companies, and differences in funding costs are likely to be 
greater. 

To assess how investors' beliefs that the government will support 
failing bank holding companies have changed over time, we compared 
bond funding costs for bank holding companies of various sizes while 
holding the level of credit risk constant over time at the average for 
2008--a relatively high level of credit risk that prevailed during the 
financial crisis.[Footnote 86] In these hypothetical scenarios, most 
models suggest that bond funding costs for larger bank holding 
companies would have been lower than bond funding costs for smaller 
bank holding companies in most years. For example, most models for 
2013 predict that bond funding costs for larger bank holding companies 
would be higher than for smaller bank holding companies at the average 
level of credit risk in that year, but would be lower at financial 
crisis levels of credit risk (see fig. 3). These results suggest that 
changes over time in funding cost differences we estimated (depicted 
in fig. 1) have been driven at least in part by improvements in the 
financial condition of bank holding companies.[Footnote 87] At the 
same time, more models predict lower bond funding costs for larger 
bank holding companies in 2008 than in 2013 when we assume that 
financial crisis levels of credit risk prevailed in both years, which 
suggests that investors' expectations of government support have 
changed over time.[Footnote 88] However, it is important to note that 
the relationships between variables estimated by our models could be 
sensitive to the average level of credit risk among bank holding 
companies, making these estimates of the potential impact of the level 
of credit risk from 2008 in the current environment even more 
uncertain.Moreover, Dodd-Frank Act reforms discussed earlier in this 
report, such as enhanced regulatory standards for capital and 
liquidity, could enhance the stability of the U.S. financial system 
and make such a credit risk scenario less likely. However, the extent 
to which such benefits will materialize depends on many factors that 
remain difficult to predict. 

Figure 3: Difference in Estimated Bond Funding Costs for Bank Holding 
Companies with $1 Trillion versus $10 Billion in Assets by Level of 
Credit Risk, 2013: 

[Refer to PDF for image: illustration] 

Models predicting statistically significant differences above zero: 18; 
Models predicting statistically insignificant differences above zero: 
12. 

Models predicting statistically significant differences below zero: 0; 
Models predicting statistically insignificant differences below zero: 
12. 

Estimated average bond funding costs for larger bank holding companies 
(BHC) were higher than for smaller BHCs. 

Estimated average bond funding costs for larger BHCs were lower than 
for smaller BHCs. 

Basis point = .01% (1/100th of a percent). 

Depicted on graph: 
Average difference estimated by one of 42 models (statistically 
significant at 10% level); 
Average difference estimated by one of 42 models (not statistically 
significant at 10% level). 

Source: GAO analysis of data from Bloomberg and the Financial 
Stability Board. GAO-14-621. 

Notes: We estimated econometric models of the relationship between 
bank holding company size and funding costs using data for U.S. bank 
holding companies and their outstanding senior unsecured bonds for the 
first quarter of 2006 through the fourth quarter of 2013. The models 
used bond yield spreads to measure funding costs and controlled for 
credit risk factors such as capital adequacy, asset quality, earnings, 
maturity mismatch, and volatility, as well as bond liquidity and other 
characteristics of bonds and bank holding companies that can affect 
funding costs. We estimated 42 models for each year from 2006 through 
2013 to assess the sensitivity of estimated funding cost differences 
to alternative measures of capital adequacy, volatility, bond 
liquidity, and size or systemic importance. We used the models to 
compare bond funding costs for bank holding companies of different 
sizes but the same levels of credit risk, bond liquidity, and other 
characteristics. This figure compares bond funding costs for bank 
holding companies with $1 trillion and $10 billion in assets, for each 
model for 2013, with the average level of credit risk in 2013 and the 
average level of credit risk in 2008 during the financial crisis. Each 
circle and dash shows the comparison for a different model, where 
circles and dashes below zero suggest bank holding companies with $1 
trillion in assets have lower bond funding costs than bank holding 
companies with $10 billion in assets, and vice versa. 

[End of figure] 

Estimates of Size-Related Funding Cost Differences Should Be 
Interpreted with Caution: 

Our estimates of the relationship between the size of a bank holding 
company and the yield spreads on its bonds are limited by several 
factors and should be interpreted with caution. These factors present 
challenges to using our results and the results of other studies as 
the basis for public policy responses to concerns about the risks 
posed by large financial institutions. 

Investors' beliefs about the likelihood of government support are 
composed of several different elements, including the likelihood that 
a bank holding company will fail, the likelihood that it will be 
rescued by the government if it fails, and the size of the losses that 
the government may impose on investors if it rescues the bank holding 
company. Like the methodologies used in the literature we reviewed, 
our methodology does not allow us to precisely identify the influence 
of each of these factors. As a result, changes over time in our 
estimates of the relationship between bond funding costs and size may 
reflect changes in one or more of these components, but we cannot 
identify which with certainty. For example, if bond funding costs for 
a bank holding company with $1 trillion are less than those for a bank 
holding company with $10 billion but the difference decreases over 
time, this trend may indicate that investors believe that the larger 
bank holding company is relatively less likely to fail, which could be 
the case if the level of credit risk is falling over time, either due 
to market pressure, regulatory requirements, or other reasons. This 
trend could also indicate that investors believe that the government 
has become less likely to rescue the larger bank holding company if it 
fails or more likely to impose losses on investors in a rescue. 

In addition, our estimates of differences in bond funding costs for 
bank holding companies of different sizes may reflect factors other 
than investors' beliefs about the likelihood of government support. We 
have taken into account many of the factors that may help explain 
differences in yield spreads for bank holding companies of different 
size, such as credit risk and bond liquidity. However, we may not have 
taken into account all possible factors. If a factor that we have not 
taken into account is associated with size, then our results may 
reflect the relationship between bond funding costs and this omitted 
factor instead of, or in addition to, the relationship between bond 
funding costs and bank holding company size. 

Our estimates of differences in bond funding costs for bank holding 
companies of different sizes may also reflect differences in the 
characteristics of bank holding companies that choose to issue bonds. 
Bank holding companies that issue bonds may differ from those that do 
not in ways that may or may not be observable. If such differences 
exist and are unobservable, then our models' comparisons are likely to 
be consistently either too high or too low depending on the 
relationship between size and the relevant unobservable 
characteristic. However, if bank holding companies that issue bonds 
differ from those that do not in ways that are observable in our 
model, then our models' comparisons of bond funding cost differences 
for bank holding companies of different sizes are unlikely to be 
consistently either too high or too low. We found some evidence that 
this may be the case. Investors with whom we spoke told us that larger 
bank holding companies are generally more likely to issue bonds than 
smaller ones because they can issue a large enough quantity of debt to 
satisfy investors' demand for liquidity and to allow investors to make 
a large enough investment to cover their transaction costs. Thus, 
size, which is observable, may be an important difference between bank 
holding companies that issue bonds and those that do not. Importantly, 
bank holding company size matters in this case because it is 
associated with bond issue size, which we control for, not because it 
is associated with investors' beliefs about government rescues. 

In general, our estimates of the impact of size on bond funding costs 
may reflect a relationship between size, credit risk, or other 
explanatory variables and the part of bond funding costs that is not 
explained by our model (endogeneity). This could occur if any of our 
control variables are influenced by bond funding costs. In this case, 
our estimates of the magnitude of the association between size and 
bond funding costs will be imperfect and our ability to infer a causal 
relationship between size and bond funding costs will be limited. 

Historical estimates of differences in bond funding costs for bank 
holding companies of different sizes are not indicative of future 
differences. As we have discussed, our estimates of the historical 
relationship between bank holding company size and bond funding costs 
vary from year to year. Thus, it is likely that the relationship 
between bond funding costs and bank holding company size may change in 
the future. As we have noted, the Dodd-Frank Act imposes new and 
higher fees on large bank holding companies and requires the Federal 
Reserve Board to subject large bank holding companies to enhanced 
regulatory standards for capital, liquidity, and risk management. 
These enhanced standards may help to reduce the likelihood that a 
large bank holding company will fail, which may in turn alter 
investors' beliefs about the likelihood of government support and thus 
affect the size of any differences in yield spreads on bonds issued by 
large and small bank holding companies. Improvements in economic 
conditions, such as faster economic growth and lower unemployment, may 
have a similar effect. Finally, changes in the structure of financial 
markets, such as an increase in the share of credit provided by 
nonbank financial companies that reduces the systemic importance of 
large bank holding companies, could also lead investors to change 
their beliefs about government rescues in future episodes of 
individual or system-wide distress. 

Finally, our estimates of the differences in bond funding costs for 
bank holding companies of different sizes do not necessarily reflect 
the harm that the failure of a large bank holding company could do to 
the economy. Bond funding costs reflect the risk that a bank holding 
company might fail and not be able to fully repay its investors. 
However, parties other than investors may be harmed if a bank holding 
company fails. For example, the customers of a failed bank holding 
company may be harmed if they have less access to credit or to 
specialized services provided by the bank holding company, which could 
be the case if the bank holding company has a large enough share of 
the market. 

Agency Comments and Our Evaluation: 

We made copies of the draft report available to FDIC, the Federal 
Reserve Board, FSOC, OCC and Treasury for their review and comment. We 
also provided excerpts of the draft report for technical comment to 
Fitch, Moody's, Standard and Poor's, and the International Monetary 
Fund. All of these agencies and third parties, except for FSOC, 
provided technical comments, which we have incorporated, as 
appropriate. 

In its written comments, which are reprinted in appendix II, Treasury 
generally agreed with the results of our analysis and commented that 
our draft report represents a meaningful contribution to the 
literature. Treasury further commented that the Dodd-Frank Act makes 
clear that shareholders, creditors and executives--not taxpayers--will 
be responsible if a large company fails and that our results reflect 
increased market recognition that the Dodd-Frank Act ended "too big to 
fail" as a matter of law. While our results do suggest bond funding 
cost differences between large and smaller bank holding companies may 
have declined or reversed since the 2007-2009 financial crisis, we 
also report that a higher credit risk environment could be associated 
with lower bond funding costs for large bank holding companies than 
for small ones. Furthermore, as we have noted, many market 
participants we spoke with believe that recent regulatory reforms have 
reduced but not eliminated the perception of "too big to fail" and 
both they and Treasury officials indicated that additional steps were 
required to address "too big to fail." As discussed in the final 
section of our report on page 56, changes over time in our estimates 
of the relationship between bond funding costs and size may reflect 
changes in one or more components of investors' beliefs about 
government support--such as their views on the likelihood that a bank 
holding company will fail and the likelihood it will be rescued if it 
fails--but we cannot precisely identify the influence of each of these 
components with certainty. A decline or reversal of funding cost 
advantages for large bank holding companies could indicate that 
investors believe that the government has become less likely to rescue 
a large bank holding company if it fails or more likely to impose 
losses on investors in a rescue. This trend could also indicate that 
investors believe that large bank holding companies are less likely to 
fail. 

On separate dates in July 2014, Treasury, the Federal Reserve Board, 
OCC and FDIC provided via email technical comments related to the 
draft report's analysis of funding cost differences between large and 
small bank holding companies. We summarize their most significant 
comments and our responses below. 

* Treasury provided comments on our presentation of the impact of a 
higher credit risk environment on our analysis of bond funding costs 
and the statistical robustness of these results. In response to these 
comments, we revised text on the Highlights and in the report body to 
clarify that the results of this analysis reflect hypothetical 
scenarios and to provide greater attention to the potential impacts of 
regulatory reforms. With respect to the statistical robustness of 
these results, we note that the draft report contained clear 
information about the statistical significance of our results. 
Importantly, we note that whether one considers the estimates from all 
42 models for 2013 or only the 10 models with statistically 
significant results, higher credit risk substantially increases (1) 
the number of models that suggest bond funding costs would have been 
lower for the largest bank holding companies than for smaller bank 
holding companies and (2) the size of funding cost differences in 
2013. In addition, we amended the draft to clarify that our results 
for the hypothetical scenario for 2013 differ from our results for 
2008, in which all 42 models predicted lower funding costs for larger 
bank holding companies. 

* Treasury and the Federal Reserve Board provided comments related to 
the draft report's presentation of statistical significance in figures 
1, 2, and 3. In response to these comments, we made formatting changes 
to the figures to more clearly differentiate estimates that are 
statistically significant from those that are not. In addition, we 
note that table 7 on pages 79-81 of the report contains some of the 
data used to create figures 1, 2 and 3 and differentiates between 
estimates that are statistically significant at the 1 percent, 5 
percent, and 10 percent levels. Finally, while statistically 
insignificant estimates may be viewed as weaker evidence than 
statistically significant estimates and may influence how our results 
are interpreted, we note that statistical significance is not the only 
relevant characteristic of an econometric estimate and that by 
presenting the full range of results one can better assess their 
magnitude and economic significance. 

* Treasury and the Federal Reserve Board also commented that in 
comparing bond funding costs for large and small bank holding 
companies, a bank holding company with $10 billion in assets is too 
small to make a meaningful comparison to a bank holding company with 
$1 trillion in assets. They commented that a bank holding company with 
at least $50 billion in assets would provide a more relevant 
comparison for this analysis. While we agree that bank holding 
companies with $50 billion in assets may be more similar to $1 
trillion bank holding companies than bank holding companies with $10 
billion in assets, we used a smaller size for small bank holding 
companies because bank holding companies with $50 billion or more in 
assets may be viewed by investors as "large" and systemically 
important, in part because $50 billion in assets is the size threshold 
for Dodd-Frank Act requirements related to enhanced regulatory 
standards. While we agree that bank holding companies of different 
sizes have different characteristics, we compared estimated funding 
costs for bank holding companies assuming their credit risk and other 
characteristics are identical. Finally, increasing the size of the 
small bank holding company in our comparisons would not have a 
substantive impact on the sign or statistical significance of our 
estimated differences in funding costs, nor would it change the trends 
in estimated differences in funding costs over time. 

* The Federal Reserve Board and OCC commented that few of the 
estimated coefficients on the variables measuring size and size 
interacted with credit risk reported in table 5 were individually 
statistically significant, suggesting that there is little statistical 
evidence of a relationship between bond funding costs and size. To 
address this concern, we conducted hypothesis tests that the 
coefficients on the size and size-credit risk interaction terms are 
jointly equal to zero. The results of these hypothesis tests suggest 
that the coefficients on the size and size-credit risk interaction 
terms are jointly significant at the 5 percent level, suggesting that 
there is statistical evidence of a relationship between bond funding 
costs and size. We report the results of our joint hypothesis tests in 
table 5 on pages 74 and 76 of the report. In addition, we note that 
the draft report only contains coefficient estimates for the 6 
baseline models of the 42 total models for 2008 and 2013 and that 
those 6 models are presented as examples and do not fully reflect the 
impact of size in all the specifications in those years or in other 
years. However, we believe the regression-level detail on the 6 
baseline models included in the report is sufficient to assist readers 
looking to understand our methodology and conclusions. 

* OCC suggested that selection bias and omitted variables bias could 
reduce the validity of our econometric results. We agree that these 
biases are potential limitations of the model and are among the 
reasons the results should be interpreted with caution. We discuss the 
potential impact of these concerns on pages 56-57. 

* OCC and FDIC commented on the endogeneity of some independent 
variables and the impact this could have on our results. We agree that 
endogeneity is a potential limitation of the model and is among the 
reasons the results should be interpreted with caution. In response to 
this comment, we added a paragraph discussing the potential impact of 
endogeneity on our results on page 57 of the report. 

We are sending copies of this report to FDIC, the Federal Reserve 
Board, FSOC, OCC, Treasury, interested congressional committees, 
members, and others. In addition, the report is available at no charge 
on the GAO website at [hyperlink, http://www.gao.gov]. 

If you or your staffs have any questions regarding this report, please 
contact me at (202) 512-4802 or EvansL@gao.gov. Contact points for our 
Offices of Congressional Relations and Public Affairs may be found on 
the last page of this report. GAO staff who made major contributions 
to this report are listed in appendix III. 

Signed by: 

Lawrance L. Evans, Jr., PhD: 
Director, Financial Markets and Community Investment: 

[End of section] 

Appendix I: Methodology for Analysis of Funding Cost Differences 
between Large and Small Bank Holding Companies: 

To assess the extent to which the largest bank holding companies have 
received funding cost advantages as a result of perceptions that the 
government would not allow them to fail, we conducted an econometric 
analysis of the relationship between a bank holding company's size or 
systemic importance and its funding costs. Bank holding companies 
obtain funds from investors--such as depositors, creditors, or 
shareholders--which they use to finance assets, such as various types 
of credit. The prices that bank holding companies pay to obtain these 
funds are influenced by several factors, including credit risk--the 
likelihood that bank holding companies will repay the funds they 
borrowed as agreed--and other factors. Funding cost advantages may 
arise if investors believe that the government is more likely to 
support larger bank holding companies in distress than smaller bank 
holding companies in distress. This belief may lead investors to view 
larger bank holding companies as having less credit risk than smaller 
bank holding companies and thus charge larger bank holding companies a 
lower price to borrow than smaller bank holding companies. 

We used a multivariate regression model to estimate the relationship 
between bank holding companies' funding costs and their size while 
controlling for factors other than size that may also influence 
funding costs. Our general regression model is the following: 

[Formula illustration] 

In this model, denotes the bank holding company, denotes the quarter, 
is the bank holding company's cost of funding in a quarter, is a 
measure of the bank holding company's size at the beginning of the 
quarter, is a list of proxies for the bank holding company's credit 
risk, is a list of other variables that may influence funding costs, 
is an idiosyncratic error term, and are parameters to be estimated. 
The parameter captures the direct relationship between a bank holding 
company's funding cost and its size. The parameter captures the 
indirect relationship between a bank holding company's funding cost 
and its size that exists if the size of a bank holding company affects 
the relationship between its funding cost and credit risk. If greater 
values of the size measure are associated with larger bank holding 
companies, if greater values of the credit risk proxies are associated 
with greater credit risk, and if investors view larger bank holding 
companies as less risky than smaller bank holding companies due to 
beliefs that the government is more likely to rescue larger bank 
holding companies in distress, then either is less than zero, is less 
than zero, or both. However, the parameters and may also reflect 
factors other than these beliefs. 

We used a measure of funding costs based on bonds issued by bank 
holding companies that is available for bank holding companies with a 
wide variety of sizes. Bank holding companies use a variety of funding 
types from different sources, including various types of deposits, 
bonds, and equity. We used bond yield spreads--the difference between 
the rate of return on a bond and the rate of return on a Treasury bond 
of comparable maturity to measure a bank holding company's cost of 
bond funding. Treasury securities are widely viewed as a risk-free 
asset, so the yield spread measures the price that investors charge a 
bank holding company to borrow to compensate them for credit risk and 
other factors. We focused on bond funding costs for several reasons. 
First, bonds are traded in secondary markets, so timely information 
about changes in their yield spreads, which reflect investors' 
perceptions of the credit risk of the bond's issuing bank holding 
company, is easily observable.[Footnote 89] In contrast, some 
uninsured deposit products are not traded in secondary markets, so 
changes in the prices of those deposits, which may reflect depositors' 
perceptions of the riskiness of the bank holding company, may be less 
easy to observe. Second, bond yield spreads are a direct measure of 
bank holding companies' funding costs. In contrast, credit ratings are 
indirect measures of bank holding companies' funding costs because 
funding costs can vary for firms with the same rating. Similarly, 
total interest expense as reported on a bank holding company's balance 
sheet is an imperfect measure of funding costs because total interest 
expense may aggregate the prices of liabilities with many important 
differences, including term and creditor priority. Third, bonds 
generally rank higher in a bank holding company's capital structure 
than equity, so bondholders are less likely to suffer losses and more 
likely to be repaid if a bank holding company becomes distressed. 
Bondholders are thus more likely to benefit if a distressed bank 
holding company is rescued by the government. In contrast, equity 
holders generally rank lowest in a bank holding company's capital 
structure and are the first to suffer losses if a bank holding company 
becomes distressed. Shareholders are thus the least likely to benefit 
if a distressed bank holding company is rescued by the government. It 
follows that the cost of bond funding is more likely to reflect 
investors' beliefs about the likelihood of government support than the 
cost of equity funding. Fourth, bank holding companies with a wide 
variety of sizes issue bonds, including some with less than $10 
billion in assets. In contrast, credit default swaps--the prices of 
which likely reflect perceptions of a bank holding company's credit 
risk--are available for only a small number of large bank holding 
companies.[Footnote 90] 

We used Bloomberg to identify U.S. bank holding companies with more 
than $500 million in assets that were active in one or more years from 
2006 to 2013, and to identify all plain vanilla, fixed-rate, senior 
unsecured bonds issued by these bank holding companies, excluding 
bonds with a government guarantee.[Footnote 91] See table 3 for the 
numbers of bank holding companies and bonds we analyzed.[Footnote 92] 

Table 3: Numbers of Bank Holding Companies and Bonds, 2006-2013: 

Year: 2006; 
Number of bank holding companies: 1178; 
Number of bank holding companies with senior unsecured bonds 
outstanding: 22; 
Number of senior unsecured bonds outstanding: 166. 

Year: 2007; 
Number of bank holding companies: 1209; 
Number of bank holding companies with senior unsecured bonds 
outstanding: 23; 
Number of senior unsecured bonds outstanding: 202. 

Year: 2008; 
Number of bank holding companies: 1192; 
Number of bank holding companies with senior unsecured bonds 
outstanding: 22; 
Number of senior unsecured bonds outstanding: 203. 

Year: 2009; 
Number of bank holding companies: 1178; 
Number of bank holding companies with senior unsecured bonds 
outstanding: 26; 
Number of senior unsecured bonds outstanding: 333. 

Year: 2010; 
Number of bank holding companies: 1133; 
Number of bank holding companies with senior unsecured bonds 
outstanding: 29; 
Number of senior unsecured bonds outstanding: 368. 

Year: 2011; 
Number of bank holding companies: 1038; 
Number of bank holding companies with senior unsecured bonds 
outstanding: 29; 
Number of senior unsecured bonds outstanding: 374. 

Year: 2012; 
Number of bank holding companies: 1046; 
Number of bank holding companies with senior unsecured bonds 
outstanding: 31; 
Number of senior unsecured bonds outstanding: 447. 

Year: 2013; 
Number of bank holding companies: 1033; 
Number of bank holding companies with senior unsecured bonds 
outstanding: 30; 
Number of senior unsecured bonds outstanding: 510. 

Source: GAO analysis of Bloomberg data. GAO-14-621. 

Notes: We used Bloomberg to identify U.S. bank holding companies that 
reported positive average assets as of the end of at least one year 
from 2006 to 2013, where average assets consist of the quarterly 
average for total assets, less goodwill, other disallowed intangible 
assets, disallowed deferred tax assets, and any other assets that are 
deducted in determining Tier 1 capital in accordance with the capital 
standards issued by the reporting bank's primary federal supervisory 
authority and are reported on form FR Y-9C. Only bank holding 
companies with $500 million in assets or more are required to file 
form FR Y-9C. We also used Bloomberg to identify all plain vanilla, 
fixed-rate, senior unsecured bonds issued by these bank holding 
companies, where "plain vanilla" bonds are bonds without any 
derivative or equity features. We excluded bonds guaranteed by either 
the Federal Deposit Insurance Corporation (FDIC) or the U.S. 
government. 

[End of table] 

We collected data on bond yield spreads, bank holding company size, 
other variables associated with bank holding company credit risk, and 
bond characteristics from Bloomberg and used these data to assemble a 
dataset with one observation for each bond in each quarter from the 
first quarter of 2006 to the fourth quarter of 2013. 

Yield spreads. For each bond, we collected daily data on its yield 
spread based on prices from executed trades, executable quotes, and 
composites derived from executable and indicative quotes.[Footnote 93] 
For each bond and for each quarter from 2006 to 2013, we calculated 
the average yield spread for each bond based on (1) executed trades, 
executable quotes, and composites derived from executable and 
indicative quotes; and (2) actual trades only.[Footnote 94] 

Size. We constructed four alternative indicators of a bank holding 
company's size or systemic importance: (1) total assets as of the 
beginning of the quarter; (2) an indicator for whether or not a bank 
holding company had $50 billion or more in assets, i.e., is a 
systemically important financial institution (SIFI), as of the 
beginning of the quarter, which captures those bank holding companies 
that are subject to enhanced prudential standards under the Dodd-Frank 
Wall Street Reform and Consumer Protection Act (Dodd-Frank Act); (3) 
an indicator for whether or not the Financial Stability Board 
designated a bank holding company a global systemically important bank 
(GSIB) in November 2013 based on its size, interconnectedness, lack of 
readily available substitutes or financial institution infrastructure, 
global (cross-jurisdictional) activity, and complexity; and (4) total 
assets and total assets squared.[Footnote 95] 

Credit risk. For each bank holding company, we constructed several 
indicators of credit risk, including indicators of capital adequacy, 
asset quality, earnings, maturity mismatch, and volatility.[Footnote 
96] 

* Capital adequacy. Capital absorbs losses, promotes public 
confidence, helps restrict excessive asset growth, and provides 
protection to creditors. We constructed two alternative sets of 
indicators of capital adequacy: (1) equity capital and subordinated 
debt as percentages of assets as of the beginning of the quarter, and 
(2) equity capital and subordinated debt as percentages of risk-
weighted assets as of the beginning of the quarter. 

* Asset quality. Asset quality reflects the quantity of existing and 
potential credit risk associated with a bank holding company's loan 
and investment portfolios and other assets, as well as off-balance 
sheet transactions. It also reflects the ability of management to 
identify and manage credit risk. We used the Texas ratio--
nonperforming assets as a percentage of tangible common equity plus 
loan loss reserves--as of the beginning of the quarter as an indicator 
of asset quality, where nonperforming assets include assets in 
nonaccrual status, other real estate owned, restructured loans, and 
restructured loans in compliance and tangible common equity is total 
capital minus perpetual preferred stock minus goodwill and other 
intangibles. The Texas ratio can indicate a bank holding company's 
likelihood of failure by comparing its troubled loans to its capital. 
The higher the ratio, the more likely the institution is to fail 
because more of its capital could be eroded by realized losses on 
these troubled loans. 

* Earnings. Earnings are the initial safeguard against the risks of 
engaging in the banking business and represent the first line of 
defense against capital depletion that can result from declining asset 
values. We used net income as a percentage of assets as of the 
beginning of the quarter as an indicator of earnings. 

* Maturity mismatch. Maturity mismatch reflects a bank holding 
company's ability to obtain funds at a reasonable price within a 
reasonable time period to meet obligations as they come due. We used 
the difference between volatile liabilities--liabilities that may 
quickly or unexpectedly come due-and liquid assets--assets that can 
easily be converted to cash to cover liabilities--as a percentage of 
total liabilities as of the beginning of each quarter as an indicator 
of maturity mismatch. Volatile liabilities include federal funds 
purchased and repurchase agreements, trading liabilities, other 
borrowed funds, foreign deposits, and jumbo deposits less derivatives 
with negative fair value. Liquid assets include cash, securities, 
federal funds sold and reverse repurchase agreements, and trading 
assets less pledged securities.[Footnote 97] 

* Volatility. Volatility reflects how much the value of a bank holding 
company fluctuates over a given amount of time and the possibility 
that the value will fall below a given threshold for default. We 
constructed five alternative indicators of volatility: (1) the 
standard deviation of equity prices over the quarter, (2) option 
implied volatility for the quarter, (3) the standard deviation of 
equity returns over the quarter, (4) the standard deviation of excess 
equity returns over the quarter, and (5) the standard deviation of 
earnings over the past 16 quarters as alternative indicators of 
volatility. 

Other factors. We also collected data on several other factors that 
may influence bond yield spreads, including bonds' coupon rates, times 
to maturity, and liquidity, and bank holding companies' operating 
expenses. We used three alternative indicators of a bond's liquidity: 
(1) issue size, (2) total volume traded during a quarter, and (3) 
average bid-ask spread over a quarter.[Footnote 98] Finally, we used a 
bank holding company's noninterest expenses as a percentage of total 
assets as an indicator of its operating expenses.[Footnote 99] Table 4 
shows summary statistics for bank holding companies with senior 
unsecured bonds outstanding in 2008 and 2013 and for senior unsecured 
bonds outstanding in 2008 and 2013. 

Table 4: Summary Statistics for Bank Holding Companies and Bonds 
Analyzed, 2008 and 2013: 

Bank holding companies with senior unsecured bonds outstanding, 2008: 

Total assets (dollars in billions); 
Mean: $403.974; 
Standard deviation: $640.816; 
Minimum: $7.685; 
Maximum: $2134.499. 

Operating expense/assets (%); 
Mean: 1.067; 
Standard deviation: 0.801; 
Minimum: 0.576; 
Maximum: 4.451. 

Equity capital/assets (%); 
Mean: 10.602; 
Standard deviation: 5.226; 
Minimum: 5.916; 
Maximum: 31.332. 

Equity capital/risk-weighted assets (%); 
Mean: 14.340; 
Standard deviation: 9.326; 
Minimum: 7.688; 
Maximum: 52.366. 

Subordinated debt/assets (%); 
Mean: 3.308; 
Standard deviation: 1.353; 
Minimum: 0; 
Maximum: 6.010. 

Subordinated debt/risk-weighted assets (%); 
Mean: 4.157; 
Standard deviation: 1.412; 
Minimum: 0; 
Maximum: 6.364. 

Earnings/assets (%); 
Mean: 0.110; 
Standard deviation: 0.440; 
Minimum: -1.032; 
Maximum: 1.349. 

Maturity mismatch (%); 
Mean: 17.869; 
Standard deviation: 18.000; 
Minimum: -53.133; 
Maximum: 34.162. 

Texas ratio (%); 
Mean: 16.287; 
Standard deviation: 10.946; 
Minimum: 0.011; 
Maximum: 40.211. 

Equity price volatility (%); 
Mean: 88.156; 
Standard deviation: 31.924; 
Minimum: 48.106; 
Maximum: 180.773. 

Option implied volatility (%); 
Mean: 66.246; 
Standard deviation: 14.459; 
Minimum: 43.573; 
Maximum: 97.816. 

Equity return volatility (%); 
Mean: 5.079; 
Standard deviation: 0.899; 
Minimum: 3.508; 
Maximum: 6.962. 

Excess equity return volatility (%); 
Mean: 3.723; 
Standard deviation: 0.969; 
Minimum: 2.231; 
Maximum: 5.817. 

Earnings volatility (%); 
Mean: 6.108; 
Standard deviation: 4.353; 
Minimum: 0.460; 
Maximum: 23.510. 

GSIB designation (indicator); 
Mean: 0.273; 
Standard deviation: [Empty]; 
Minimum: 0; 
Maximum: 1. 

Assets $50 billion or more (indicator); 
Mean: 0.818; 
Standard deviation: [Empty]; 
Minimum: 0; 
Maximum: 1. 

Senior unsecured bonds outstanding, 2008: 

Average bond yield spread calculated using prices from executed 
trades, executable quotes, and composites of executable and indicative 
quotes (bps); 
Mean: 456.407; 
Standard deviation: 305.194; 
Minimum: 15.066; 
Maximum: 1,479.019. 

Average bond yield spread calculated using prices from executed trades 
(bps); 
Mean: 474.809; 
Standard deviation: 342.013; 
Minimum: 15.066; 
Maximum: 1,479.019. 

Bond issue size (dollars in millions); 
Mean: $1,004.746; 
Standard deviation: $1,028.839; 
Minimum: $0.143; 
Maximum: $5,500.000. 

Total trade volume (dollars in millions); 
Mean: $160,456.800; 
Standard deviation: $225,838.100; 
Minimum: 0; 
Maximum: $1,663,766.000. 

Average bid-ask spread, all pricing sources (%); 
Mean: 1.535; 
Standard deviation: 1.621; 
Minimum: 0.0156; 
Maximum: 14.224. 

Time to maturity (quarters); 
Mean: 17.177; 
Standard deviation: 19.706; 
Minimum: 0; 
Maximum: 119.000. 

Coupon (%); 
Mean: 5.461; 
Standard deviation: 1.219; 
Minimum: 2.000; 
Maximum: 12.500. 

Bank holding companies with senior unsecured bonds outstanding, 2013: 

Total assets (dollars in billions); 
Mean: $413.488; 
Standard deviation: $672.505; 
Minimum: $4.563; 
Maximum: $2,412.823. 

Operating expense/assets (%); 
Mean: 0.986; 
Standard deviation: 0.598; 
Minimum: 0.613; 
Maximum: 3.899. 

Equity capital/assets (%); 
Mean: 10.993; 
2.375; 
Minimum: 7.207; 
Maximum: 19.144. 

Equity capital/risk-weighted assets (%); 
Mean: 16.206; 
Standard deviation: 4.358; 
Minimum: 10.676; 
Maximum: 31.576. 

Subordinated debt/assets (%); 
Mean: 1.602; 
Standard deviation: 0.996; 
Minimum: 0; 
Maximum: 4.137. 

Subordinated debt/risk-weighted assets (%); 
Mean: 2.402; 
Standard deviation: 1.205; 
Minimum: 0; 
Maximum: 5.171. 

Earnings/assets (%); 
Mean: 0.280; 
Standard deviation: 0.150; 
Minimum: 0.040; 
Maximum: 0.766. 

Maturity mismatch (%); 
Mean: -1.044; 
Standard deviation: 20.512; 
Minimum: -53.133; 
Maximum: 47.001. 

Texas ratio (%); 
Mean: 17.523; 
Standard deviation: 12.061; 
Minimum: 0.812; 
Maximum: 38.690. 

Equity price volatility (%); 
Mean: 20.629; 
Standard deviation: 3.280; 
Minimum: 12.952; 
Maximum: 28.572. 

Option implied volatility (%); 
Mean: 23.996; 
Standard deviation: 3.656; 
Minimum: 15.788; 
Maximum: 30.250. 

Equity return volatility (%); 
Mean: 1.311; 
Standard deviation: 0.202; 
Minimum: 0.828; 
Maximum: 1.758. 

Excess equity return volatility (%); 
Mean: 1.003; 
Standard deviation: 0.200; 
Minimum: 0.630; 
Maximum: 1.478. 

Earnings volatility (%); 
Mean: 3.936; 
Standard deviation: 4.771; 
Minimum: 0.355; 
Maximum: 23.510. 

GSIB designation (indicator); 
Mean: 0.267; 
Standard deviation: [Empty]; 
Minimum: 0; 
Maximum: 1. 

Assets $50 billion or more (indicator); 
Mean: 0.700; 
Standard deviation: [Empty]; 
Minimum: 0; 
Maximum: 1. 

Senior unsecured bonds outstanding, 2013: 

Average bond yield spread calculated using prices from executed 
trades, executable quotes, and composites of executable and indicative 
quotes (bps); 
Mean: 140.684; 
Standard deviation: 69.252; 
Minimum: 15.066; 
Maximum: 500.750. 

Average bond yield spread calculated using prices from executed trades 
(bps); 
Mean: 132.952; 
Standard deviation: 71.563; 
Minimum: 15.066; 
Maximum: 796.406. 

Bond issue size (dollars in millions); 
Mean: $1,042.622; 
Standard deviation: $1,107.804; 
Minimum: $0.350; 
Maximum: $5,500.000. 

Total trade volume (dollars in millions); 
Mean: $161,367.900; 
Standard deviation: $293,879.200; 
Minimum: $0; 
Maximum: $33,334,868.000. 

Average bid-ask spread, all pricing sources (%); 
Mean: 0.537; 
Standard deviation: 0.474; 
Minimum: 0.003; 
Maximum: 2.819. 

Time to maturity (quarters); 
Mean: 24.548; 
Standard deviation: 24.729; 
Minimum: 0; 
Maximum: 126.000. 

Coupon (%); 
Mean: 4.320; 
Standard deviation: 1.712; 
Minimum: 0.800; 
Maximum: 10.000. 

Source: GAO analysis of Bloomberg and Financial Stability Board data. 
GAO-14-621. 

[End of table] 

We estimated a variety of models for each year from 2006 to 2013 to 
address uncertainty about how to appropriately control for bond 
liquidity, bank holding company credit risk, and the size or systemic 
importance of a bank holding company and also to allow the 
relationship between bank holding company size and bond funding costs 
to vary over time. In our baseline specifications, we measured bond 
funding costs using average yield spreads on senior unsecured bonds 
based on executed trades, executable quotes, and composites derived 
from executable and indicative quotes; size using total assets; 
capital adequacy using equity capital and subordinated debt as 
percentages of total assets; and bond liquidity using issue size and 
total volume. We estimated the parameters for each year and for each 
of our five volatility variables, as well as for each year without 
controlling for volatility. We also estimated the parameters of the 
following variations of the baseline specifications for senior 
unsecured bonds: 

* Bid-ask specification. We measured bond liquidity using issue size, 
total volume, and average bid-ask spread. 

* Trade Reporting and Compliance Engine specification. We measured 
bond funding costs using average yield spreads based only on actual 
trades reported in the Trade Reporting and Compliance Engine database. 

* Risk-weighted assets specification. We measured capital adequacy 
using equity capital and subordinated debt as percentages of risk-
weighted assets. 

* GSIB specification. We measured size using the indicator for whether 
or not a bank holding company was designated a GSIB in November 2013. 

* Systemically important financial institution (SIFI) specification. 
We measured size using the indicator for whether or not a bank holding 
company had $50 billion or more in assets. 

* Total assets and total assets squared specification. We measured 
size using both total assets and the square of total assets. 

Altogether, we estimated the parameters of 42 separate specifications 
for each year from 2006 to 2013. For all specifications, we included 
indicators for each quarter to control for the influence on yield 
spreads of economic conditions, the regulatory environment, and other 
factors that vary over time but not across bank holding companies. We 
also adjusted the standard errors of our parameter estimates to allow 
for the possibility that they are not identically distributed and to 
allow for arbitrary correlation between observations on bonds issued 
by the same bank holding company. Table 5 shows the estimates of the 
parameters for our baseline regressions for 2008 and 2013. 

Table 5: Parameter Estimates for Baseline Regressions, 2008 and 2013: 

Dependent variable = average bond yield spreads calculated using 
prices from executed trades, executable quotes, and composites of 
executable and indicative quotes (basis points): 

2008; 

Time to maturity; 
Volatility variable: 
Equity price volatility: -1.269*; (0.614); 
Option implied volatility: -1.187*; (0.641); 
Equity return volatility: -1.255*; (0.606); 
Excess equity return volatility: -1.255*; (0.611); 
Earnings volatility: -1.232**; (0.570); 
No volatility variable: -1.267**; (0.590). 

Coupon; 
Volatility variable: 
Equity price volatility: 30.982; (20.589); 
Option implied volatility: 29.042; (21.853); 
Equity return volatility: 30.547; (20.257); 
Excess equity return volatility: 30.517; (20.379); 
Earnings volatility: 29.479; (19.355); 
No volatility variable: 29.876; (20.006). 

Operating expenses; 
Volatility variable: 
Equity price volatility: -92.810**; (37.956); 
Option implied volatility: -202.517***; (60.420); 
Equity return volatility: -89.526**; (36.812); 
Excess equity return volatility: -70.820*; (35.180); 
Earnings volatility: -138.984**; (55.786); 
No volatility variable: -111.772***; (34.104). 

Issue size; 
Volatility variable: 
Equity price volatility: 0.000; (0.023); 
Option implied volatility: 0.004; (0.024); 
Equity return volatility: 0.000; (0.023); 
Excess equity return volatility: 0.003; (0.023); 
Earnings volatility: 0.002; (0.022); 
No volatility variable: -0.000; (0.022). 

Total trade volume; 
Volatility variable: 
Equity price volatility: -0.000; (0.000); 
Option implied volatility: -0.000; (0.000); 
Equity return volatility: -0.000; (0.000); 
Excess equity return volatility: -0.000; (0.000); 
Earnings volatility: -0.000; (0.000); 
No volatility variable: -0.000; (0.000). 

Total assets; 
Volatility variable: 
Equity price volatility: 0.124; (0.144); 
Option implied volatility: 0.064; (0.139); 
Equity return volatility: 0.124; (0.129); 
Excess equity return volatility: 0.130; (0.104); 
Earnings volatility: 0.055; (0.133); 
No volatility variable: 0.063; (0.129). 

Equity capital; 
Volatility variable: 
Equity price volatility: 19.526**; (6.944); 
Option implied volatility: 23.180**; (10.197); 
Equity return volatility: 18.946**; (7.655); 
Excess equity return volatility: 17.337**; (6.725); 
Earnings volatility: 18.710**; (8.084); 
No volatility variable: 19.686***; (6.863). 

Equity capital x total assets; 
Volatility variable: 
Equity price volatility: 0.013; (0.023); 
Option implied volatility: -0.022; (0.026); 
Equity return volatility: 0.012; (0.024); 
Excess equity return volatility: 0.017; (0.018); 
Earnings volatility: 0.021; (0.020); 
No volatility variable: 0.020; (0.021). 

Subordinated debt; 
Volatility variable: 
Equity price volatility: 7.918; (31.859); 
Option implied volatility: 14.153; (29.805); 
Equity return volatility: 6.180; (31.162); 
Excess equity return volatility: 0.322; (26.788); 
Earnings volatility: 10.856; (29.995); 
No volatility variable: 11.004; (32.695). 

Subordinated debt x total assets; 
Volatility variable: 
Equity price volatility: -0.099; (0.065); 
Option implied volatility: -0.025; (0.046); 
Equity return volatility: -0.090; (0.065); 
Excess equity return volatility: -0.092*; (0.053); 
Earnings volatility: -0.108; (0.064); 
No volatility variable: -0.103; (0.065). 

Earnings; 
Volatility variable: 
Equity price volatility: -167.831*; (96.274); 
Option implied volatility: -108.627; (97.037); 
Equity return volatility: -143.709; (99.636); 
Excess equity return volatility: -146.072; (84.756); 
Earnings volatility: -177.705; (108.083); 
No volatility variable: -163.593*; (93.830). 

Earnings x total assets; 
Volatility variable: 
Equity price volatility: 0.068; (0.138); 
Option implied volatility: 0.169; (0.138); 
Equity return volatility: 0.045; (0.130); 
Excess equity return volatility: 0.060; (0.106); 
Earnings volatility: -0.000; (0.130); 
No volatility variable: 0.013; (0.125). 

Maturity mismatch; 
Volatility variable: 
Equity price volatility: -5.968**; (2.466); 
Option implied volatility: -8.608***; (2.105); 
Equity return volatility: -5.385**; (2.422); 
Excess equity return volatility: -4.915**; (2.181); 
Earnings volatility: -7.370***; (2.349); 
No volatility variable: -6.829**; (2.535). 

Maturity mismatch x total assets; 
Volatility variable: 
Equity price volatility: 0.003; (0.002); 
Option implied volatility: 0.007***; (0.002); 
Equity return volatility: 0.003; (0.002); 
Excess equity return volatility: 0.002; (0.002); 
Earnings volatility: -0.001; (0.003); 
No volatility variable: 0.003; (0.002). 

Texas ratio; 
Volatility variable: 
Equity price volatility: 6.512*; (3.526); 
Option implied volatility: 0.258; (1.979); 
Equity return volatility: 6.320*; (3.453); 
Excess equity return volatility: 4.985; (3.813); 
Earnings volatility: 7.060*; (3.514); 
No volatility variable: 7.531*; (3.615). 

Texas ratio x total assets; 
Volatility variable: 
Equity price volatility: -0.002; (0.002); 
Option implied volatility: -0.000; (0.001); 
Equity return volatility: -0.002; (0.002); 
Excess equity return volatility: -0.002; (0.002); 
Earnings volatility: -0.002; (0.002); 
No volatility variable: -0.003; (0.002). 

Equity price volatility; 
Volatility variable: 
Equity price volatility: 0.586; (0.396); 
Option implied volatility: [Empty]; 
Equity return volatility: [Empty]; 
Excess equity return volatility: [Empty]; 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Equity price volatility x total assets; 
Volatility variable: 
Equity price volatility: -0.000; (0.000); 
Option implied volatility: [Empty]; 
Equity return volatility: [Empty]; 
Excess equity return volatility: [Empty]; 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Option implied volatility; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: 1.696; (1.463); 
Equity return volatility: [Empty]; 
Excess equity return volatility: [Empty]; 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Option implied volatility x total assets; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: 0.000; (0.001); 
Equity return volatility: [Empty]; 
Excess equity return volatility: [Empty]; 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Equity return volatility; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: [Empty]; 
Excess equity return volatility: 34.178*; (17.628); 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Equity return volatility x total assets; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: [Empty]; 
Equity return volatility: -0.005; (0.010); 
Excess equity return volatility: [Empty]; 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Excess equity return volatility; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: [Empty]; 
Equity return volatility: [Empty]; 
Excess equity return volatility: 48.818**; (18.522); 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Excess equity return volatility x total assets; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: [Empty]; 
Equity return volatility: [Empty]; 
Excess equity return volatility: -0.017*; (0.009); 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Earnings volatility; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: [Empty]; 
Equity return volatility: [Empty]; 
Excess equity return volatility: [Empty]; 
Earnings volatility: 7.787; (10.992); 
No volatility variable: [Empty]. 

Earnings volatility x total assets; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: [Empty]; 
Equity return volatility: [Empty]; 
Excess equity return volatility: [Empty]; 
Earnings volatility: 0.014; (0.011); 
No volatility variable: [Empty]. 

Number of observations; 
Volatility variable: 
Equity price volatility: 478; 
Option implied volatility: 428; 
Equity return volatility: 478; 
Excess equity return volatility: 478; 
Earnings volatility: 478; 
No volatility variable: 478. 

Adjusted R-squared; 
Volatility variable: 
Equity price volatility: 0.628; 
Option implied volatility: 0.606; 
Equity return volatility: 0.631; 
Excess equity return volatility: 0.638; 
Earnings volatility: 0.624; 
No volatility variable: 0.622. 

Akaike Information Criteria; 
Volatility variable: 
Equity price volatility: 5917; 
Option implied volatility: 5254; 
Equity return volatility: 5913; 
Excess equity return volatility: 5904; 
Earnings volatility: 5922; 
No volatility variable: 5927. 

Schwartz Information Criteria; 
Volatility variable: 
Equity price volatility: 6000; 
Option implied volatility: 5323; 
Equity return volatility: 5997; 
Excess equity return volatility: 5988; 
Earnings volatility: 6005; 
No volatility variable: 6010. 

Number of bank holding companies; 
Volatility variable: 
Equity price volatility: 21; 
Option implied volatility: 18; 
Equity return volatility: 21; 
Excess equity return volatility: 21; 
Earnings volatility: 21; 
No volatility variable: 21. 

Hypothesis test that the coefficients on total assets and the total 
assets-credit risk interaction terms are jointly equal to zero: 

F-statistic; 
Volatility variable: 
Equity price volatility: 3.579; 
Option implied volatility: 3.792; 
Equity return volatility: 2.972; 
Excess equity return volatility: 4.486; 
Earnings volatility: 2.983; 
No volatility variable: 4.207. 

Probability of observing F-statistic if hypothesis is true; 
Volatility variable: 
Equity price volatility: 0.012; 
Option implied volatility: 0.012; 
Equity return volatility: 0.026; 
Excess equity return volatility: 0.004; 
Earnings volatility: 0.026; 
No volatility variable: 0.007. 

2013: 

Time to maturity; 
Volatility variable: 
Equity price volatility: 0.788***; (0.093); 
Option implied volatility: 0.775***; (0.087); 
Equity return volatility: 0.794***; (0.095); 
Excess equity return volatility: 0.798***; (0.095); 
Earnings volatility: 0.857***; (0.120); 
No volatility variable: 0.878***; (0.128). 

Coupon; 
Volatility variable: 
Equity price volatility: 9.497***; (2.079); 
Option implied volatility: 8.455***; (1.742); 
Equity return volatility: 9.453***; (2.047); 
Excess equity return volatility: 9.197***; (1.926); 
Earnings volatility: 8.396***; (2.459); 
No volatility variable: 10.085***; (2.511). 

Operating expenses; 
Volatility variable: 
Equity price volatility: 5.749; (8.154); 
Option implied volatility: 0.972; (6.588); 
Equity return volatility: 6.741; (8.190); 
Excess equity return volatility: 7.736; (8.433); 
Earnings volatility: -10.274**; (4.296); 
No volatility variable: -13.639**; (5.668). 

Issue size; 
Volatility variable: 
Equity price volatility: -0.013***; (0.004); 
Option implied volatility: -0.010**; (0.004); 
Equity return volatility: -0.013***; (0.004); 
Excess equity return volatility: -0.013***; (0.004); 
Earnings volatility: -0.013***; (0.004); 
No volatility variable: -0.016***; (0.004). 

Total trade volume; 
Volatility variable: 
Equity price volatility: 0.000; (0.000); 
Option implied volatility: 0.000; (0.000); 
Equity return volatility: 0.000; (0.000); 
Excess equity return volatility: 0.000; (0.000); 
Earnings volatility: 0.000; (0.000); 
No volatility variable: 0.000; (0.000). 

Total assets; 
Volatility variable: 
Equity price volatility: 0.013; (0.032); 
Option implied volatility: 0.017; (0.022); 
Equity return volatility: 0.010; (0.034); 
Excess equity return volatility: 0.013; (0.033); 
Earnings volatility: 0.269***; (0.077); 
No volatility variable: -0.037; (0.048). 

Equity capital; 
Volatility variable: 
Equity price volatility: 6.763*; (3.856); 
Option implied volatility: 7.444**; (3.428); 
Equity return volatility: 6.072; (3.923); 
Excess equity return volatility: 3.375; (3.965); 
Earnings volatility: 5.487; (3.716); 
No volatility variable: 0.470; (4.559). 

Equity capital x total assets; 
Volatility variable: 
Equity price volatility: -0.005; (0.005); 
Option implied volatility: -0.009*; (0.005); 
Equity return volatility: -0.004; (0.005); 
Excess equity return volatility: -0.002; (0.005); 
Earnings volatility: -0.011; (0.007); 
No volatility variable: 0.012**; (0.005). 

Subordinated debt; 
Volatility variable: 
Equity price volatility: -29.137**; (13.149); 
Option implied volatility: -29.108***; (9.968); 
Equity return volatility: -29.921**; (13.257); 
Excess equity return volatility: -31.370**; (13.438); 
Earnings volatility: -19.769; (16.353); 
No volatility variable: -25.283*; (14.168). 

Subordinated debt x total assets; 
Volatility variable: 
Equity price volatility: 0.052**; (0.019); 
Option implied volatility: 0.057***; (0.017); 
Equity return volatility: 0.054**; (0.020); 
Excess equity return volatility: 0.057**; (0.021); 
Earnings volatility: 0.065*; (0.033); 
No volatility variable: 0.030; (0.023). 

Earnings; 
Volatility variable: 
Equity price volatility: -69.186; (49.502); 
Option implied volatility: -44.545; (44.445); 
Equity return volatility: -71.544; (49.088); 
Excess equity return volatility: -74.556; (47.102); 
Earnings volatility: -32.870*; (17.759); 
No volatility variable: -48.299**; (21.803). 

Earnings x total assets; 
Volatility variable: 
Equity price volatility: 0.027; (0.041); 
Option implied volatility: 0.044; (0.028); 
Equity return volatility: 0.026; (0.042); 
Excess equity return volatility: 0.018; (0.038); 
Earnings volatility: -0.070***; (0.022); 
No volatility variable: -0.048; (0.034). 

Maturity mismatch; 
Volatility variable: 
Equity price volatility: 0.522; (0.695); 
Option implied volatility: 1.136*; (0.593); 
Equity return volatility: 0.559; (0.714); 
Excess equity return volatility: 0.555; (0.706); 
Earnings volatility: 1.062; (0.779); 
No volatility variable: 2.244***; (0.531). 

Maturity mismatch x total assets; 
Volatility variable: 
Equity price volatility: -0.001*; (0.001); 
Option implied volatility: -0.001***; (0.000); 
Equity return volatility: -0.001*; (0.001); 
Excess equity return volatility: -0.001; (0.001); 
Earnings volatility: -0.002*; (0.001); 
No volatility variable: -0.003***; (0.001). 

Texas ratio; 
Volatility variable: 
Equity price volatility: 3.207***; (1.114); 
Option implied volatility: 2.961***; (0.933); 
Equity return volatility: 3.270***; (1.108); 
Excess equity return volatility: 2.826**; (1.055); 
Earnings volatility: 4.106***; (1.463); 
No volatility variable: 3.447**; (1.464). 

Texas ratio x total assets; 
Volatility variable: 
Equity price volatility: -0.003***; (0.001); 
Option implied volatility: -0.003***; (0.001); 
Equity return volatility: -0.003***; (0.001); 
Excess equity return volatility: -0.004***; (0.001); 
Earnings volatility: -0.008***; (0.002); 
No volatility variable: -0.006***; (0.001). 

Equity price volatility; 
Volatility variable: 
Equity price volatility: 5.959***; (1.434); 
Option implied volatility: [Empty]; 
Equity return volatility: [Empty]; 
Excess equity return volatility: [Empty]; 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Equity price volatility x total assets; 
Volatility variable: 
Equity price volatility: 0.000; (0.001); 
Option implied volatility: [Empty]; 
Equity return volatility: [Empty]; 
Excess equity return volatility: [Empty]; 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Option implied volatility; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: 9.397***; (1.818); 
Equity return volatility: [Empty]; 
Excess equity return volatility: [Empty]; 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Option implied volatility x total assets; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: 0.000; (0.001); 
Equity return volatility: [Empty]; 
Excess equity return volatility: [Empty]; 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Equity return volatility; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: [Empty]; 
Equity return volatility: 98.615***; (25.234); 
Excess equity return volatility: [Empty]; 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Equity return volatility x total assets; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: [Empty]; 
Equity return volatility: -0.001; (0.014); 
Excess equity return volatility: [Empty]; 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Excess equity return volatility; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: [Empty]; 
Equity return volatility: [Empty]; 
Excess equity return volatility: 124.829***; (32.956); 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Excess equity return volatility x total assets; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: [Empty]; 
Equity return volatility: [Empty]; 
Excess equity return volatility: -0.020; (0.017); 
Earnings volatility: [Empty]; 
No volatility variable: [Empty]. 

Earnings volatility; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: [Empty]; 
Equity return volatility: [Empty]; 
Excess equity return volatility: [Empty]; 
Earnings volatility: 12.388***; (3.559); 
No volatility variable: [Empty]. 

Earnings volatility x total assets; 
Volatility variable: 
Equity price volatility: [Empty]; 
Option implied volatility: [Empty]; 
Equity return volatility: [Empty]; 
Excess equity return volatility: [Empty]; 
Earnings volatility: -0.029***; (0.007); 
No volatility variable: [Empty]. 

Number of observations; 
Volatility variable: 
Equity price volatility: 1446; 
Option implied volatility: 1446; 
Equity return volatility: 1446; 
Excess equity return volatility: 1446; 
Earnings volatility: 1535; 
No volatility variable: 1535. 

Adjusted R-squared; 
Volatility variable: 
Equity price volatility: 0.521; 
Option implied volatility: 0.563; 
Equity return volatility: 0.519; 
Excess equity return volatility: 0.519; 
Earnings volatility: 0.521; 
No volatility variable: 0.469. 

Akaike Information Criteria; 
Volatility variable: 
Equity price volatility: 15083; 
Option implied volatility: 14951; 
Equity return volatility: 15090; 
Excess equity return volatility: 15090; 
Earnings volatility: 16221; 
No volatility variable: 16378. 

Schwartz Information Criteria; 
Volatility variable: 
Equity price volatility: 15199; 
Option implied volatility: 15067; 
Equity return volatility: 15206; 
Excess equity return volatility: 15206; 
Earnings volatility: 16338; 
No volatility variable: 16484. 

Number of bank holding companies; 
Volatility variable: 
Equity price volatility: 30; 
Option implied volatility: 30; 
Equity return volatility: 30; 
Excess equity return volatility: 30; 
Earnings volatility: 31; 
No volatility variable: 31. 

Hypothesis test that the coefficients on total assets and the total 
assets-credit risk interaction terms are jointly equal to zero: 

F-statistic; 
Volatility variable: 
Equity price volatility: 3.027; 
Option implied volatility: 3.907; 
Equity return volatility: 3.009; 
Excess equity return volatility: 4.477; 
Earnings volatility: 7.309; 
No volatility variable: 5.258. 

Probability of observing F-statistic if hypothesis is true; 
Volatility variable: 
Equity price volatility: 0.016; 
Option implied volatility: 0.004; 
Equity return volatility: 0.017; 
Excess equity return volatility: 0.002; 
Earnings volatility: less than 0.001; 
No volatility variable: 0.001. 

Source: GAO analysis of Bloomberg and Financial Stability Board data. 
GAO-14-621. 

Notes: We estimated econometric models of the relationship between 
bank holding company size and funding costs using data for U.S. bank 
holding companies and their outstanding senior unsecured bonds for the 
first quarter of 2006 through the fourth quarter of 2013. The models 
used bond yield spreads to measure funding costs and controlled for 
credit risk factors such as capital adequacy, asset quality, earnings, 
maturity mismatch, and volatility, as well as bond liquidity and other 
characteristics of bonds and bank holding companies that can affect 
funding costs. We estimated multiple models for each year from 2006 
through 2013 to assess the sensitivity of our results to alternative 
measures of capital adequacy, volatility, bond liquidity, and size or 
systemic importance. This table shows estimates of the parameters of 
models for 2008 and 2013 that used average bond yield spreads 
calculated using prices from executed trades, executable quotes, and 
composites of executable and indicative quotes to measure bond funding 
costs; total assets to measure bank holding company size; issue size 
and total trade volume to measure bond liquidity; equity capital and 
subordinated debt as percents of assets to measure capital adequacy; 
and each of equity price volatility, option implied volatility, equity 
return volatility, excess equity return volatility, and earnings 
volatility to measure volatility, as well as a model with no 
volatility measure. Other explanatory variables include coupon, time 
to maturity, Texas ratio, maturity mismatch, and operating expenses. 
Standard errors clustered by bank holding company are in parentheses. 
*=statistically significant at the 10 percent level. **=statistically 
significant at the 5 percent level. ***=statistically significant at 
the 1 percent level. 

[End of table] 

Our econometric models allow the size of a bank holding company to 
influence its bond yield spreads both directly and indirectly through 
the relationships between yield spreads and each indicator of credit 
risk, i.e., through the interaction term. To summarize the overall 
relationship between yield spreads and size reflected in each 
specification, we predicted yield spreads on senior unsecured bonds 
for bank holding companies of different sizes and credit risk levels 
using our estimates of the parameters for each specification for each 
year. Specifically, for each year, we predicted bond yield spreads for 
bank holding companies with $10 billion, $50 billion, $100 billion, 
$250 billion, $500 billion, and $1 trillion in assets assuming that 
all bank holding companies had the average level of credit risk each 
year, a low level of credit risk each year, a high level of credit 
risk each year, and the average level of credit risk for 
2008.[Footnote 100] Table 6 shows all of the differences between 
predicted bond yield spreads for bank holding companies with $50 
billion, $100 billion, $250 billion, $500 billion, and $1 trillion in 
assets and predicted bond yield spreads for bank holding companies 
with $10 billion in assets for our baseline regressions for 2008 and 
2013 assuming bank holding companies have the average level of credit 
risk each year. 

Table 6: Baseline Comparisons of Bond Yield Spreads for Bank Holding 
Companies of Different Sizes by Year, 2008 and 2013 (basis points): 

2008, credit risk equal to average for 2008: 

Comparison: $50 billion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: -3.860***; (1.175); 
Option implied volatility: -1.867**; (0.770); 
Equity return volatility: -4.124***; (1.186); 
Excess equity return volatility: -3.590***; (1.181); 
Earnings volatility: -4.353***; (1.235); 
No volatility variable: -4.011***; (1.256). 

Comparison: $100 billion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: -8.686***; (2.644); 
Option implied volatility: -4.201**; (1.734); 
Equity return volatility: -9.278***; (2.669); 
Excess equity return volatility: -8.078***; (2.658); 
Earnings volatility: -9.794***; (2.779); 
No volatility variable: -9.026***; (2.827). 

Comparison: $250 billion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: -23.161***; (7.050); 
Option implied volatility: -11.204**; (4.623); 
Equity return volatility: -24.742***; (7.118); 
Excess equity return volatility: -21.541***; (7.088); 
Earnings volatility: -26.119***; (7.410); 
No volatility variable: -24.069***; (7.537). 

Comparison: $500 billion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: -47.288***; (14.394); 
Option implied volatility: -22.874**; (9.439); 
Equity return volatility: -50.515***; (14.532); 
Excess equity return volatility: -43.980***; (14.471); 
Earnings volatility: -53.326***; (15.128); 
No volatility variable: -49.140***; (15.389). 

Comparison: $1 trillion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: -95.541***; (29.081); 
Option implied volatility: -46.215**; (19.070); 
Equity return volatility: -102.061***; (29.360); 
Excess equity return volatility: -88.858***; (29.237); 
Earnings volatility: -107.739***; (30.565); 
No volatility variable: -99.284***; (31.092). 

2013, credit risk equal to average for 2013: 

Comparison: $50 billion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: 1.152**; (0.446); 
Option implied volatility: 1.011***; (0.365); 
Equity return volatility: 1.239**; (0.463); 
Excess equity return volatility: 1.388***; (0.498); 
Earnings volatility: 0.930; (0.861); 
No volatility variable: 2.365***; (0.727). 

Comparison: $100 billion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: 2.591**; (1.004); 
Option implied volatility: 2.276***; (0.820); 
Equity return volatility: 2.787**; (1.042); 
Excess equity return volatility: 3.123***; (1.120); 
Earnings volatility: 2.093; (1.936); 
No volatility variable: 5.322***; (1.635). 

Comparison: $250 billion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: 6.910**; (2.676); 
Option implied volatility: 6.068***; (2.187); 
Equity return volatility: 7.432**; (2.778); 
Excess equity return volatility: 8.329***; (2.985); 
Earnings volatility: 5.582; (5.164); 
No volatility variable: 14.19***; (4.359). 

Comparison: $500 billion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: 14.107**; (5.464); 
Option implied volatility: 12.389***; (4.465); 
Equity return volatility: 15.174**; (5.673); 
Excess equity return volatility: 17.006***; (6.095); 
Earnings volatility: 11.397; (10.542); 
No volatility variable: 28.977***; (8.900). 

Comparison: $1 trillion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: 28.502**; (11.039); 
Option implied volatility: 25.031***; (9.022); 
Equity return volatility: 30.658**; (11.461); 
Excess equity return volatility: 34.358***; (12.315); 
Earnings volatility: 23.027; (21.30); 
No volatility variable: 58.546***; (17.982). 

2013, credit risk equal to average for 2008: 

Comparison: $50 billion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: 1.976; (3.253); 
Option implied volatility: 2.541; (2.414); 
Equity return volatility: 1.465; (2.899); 
Excess equity return volatility: -0.639; (2.598); 
Earnings volatility: -0.087; (1.704); 
No volatility variable: -1.971; (1.750). 

Comparison: $100 billion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: 4.447; (7.319); 
Option implied volatility: 5.717; (5.431); 
Equity return volatility: 3.296; (6.523); 
Excess equity return volatility: -1.437; (5.846); 
Earnings volatility: -0.195; (3.835); 
No volatility variable: -4.435; (3.937). 

Comparison: $250 billion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: 11.859; (19.517); 
Option implied volatility: 15.245; (14.483); 
Equity return volatility: 8.789; (17.395); 
Excess equity return volatility: -3.833; (15.589); 
Earnings volatility: -0.519; (10.226); 
No volatility variable: -11.828; (10.499). 

Comparison: $500 billion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: 24.212; (39.848); 
Option implied volatility: 31.126; (29.570); 
Equity return volatility: 17.943; (35.514); 
Excess equity return volatility: -7.826; (31.828); 
Earnings volatility: -1.060; (20.879); 
No volatility variable: -24.148; (21.436). 

Comparison: $1 trillion vs. $10 billion in assets; 
Comparison from baseline model with volatility variable: 
Equity price volatility: 48.918; (80.509); 
Option implied volatility: 62.888; (59.743); 
Equity return volatility: 36.253; (71.754); 
Excess equity return volatility: -15.812; (64.305); 
Earnings volatility: -2.142; (42.183); 
No volatility variable: -48.789; (43.309). 

Source: GAO analysis of Bloomberg and Financial Stability Board data. 
GAO-14-621. 

Notes: We estimated econometric models of the relationship between 
bank holding company size and funding costs using data for U.S. bank 
holding companies and their outstanding senior unsecured bonds for the 
first quarter of 2006 through the fourth quarter of 2013. The models 
used bond yield spreads to measure funding costs and controlled for 
credit risk factors such as capital adequacy, asset quality, earnings, 
maturity mismatch, and volatility, as well as bond liquidity and other 
characteristics of bonds and bank holding companies that can affect 
funding costs. We estimated multiple models for each year from 2006 
through 2013 to assess the sensitivity of estimated funding cost 
differences to alternative measures of capital adequacy, volatility, 
bond liquidity, and size or systemic importance and then used the 
models to compare bond funding costs for bank holding companies of 
different sizes but the same levels of credit risk, bond liquidity, 
and other characteristics. This table shows bond funding cost 
comparisons from models for 2008 and 2013 that used average bond yield 
spreads calculated using prices from executed trades, executable 
quotes, and composites of executable and indicative quotes to measure 
bond funding costs; total assets to measure bank holding company size; 
issue size and total trade volume to measure bond liquidity; equity 
capital and subordinated debt as percents of assets to measure capital 
adequacy; and each of equity price volatility, option implied 
volatility, equity return volatility, excess equity return volatility, 
and earnings volatility to measure volatility, as well as a model with 
no volatility measure. Other explanatory variables include coupon, 
time to maturity, Texas ratio, maturity mismatch, and operating 
expenses. *=statistically significant at the 10 percent level. 
**=statistically significant at the 5 percent level. ***=statistically 
significant at the 1 percent level. 

[End of table] 

Table 7 shows the difference in predicted bond yield spreads for bank 
holding companies with $1 trillion and $10 billion in assets for all 
specifications for 2008 and 2013 assuming bank holding companies have 
the average level of credit risk each year. Table 7 also shows the 
difference in predicted bond yield spreads for bank holding companies 
with $1 trillion and $10 billion in assets for all specifications for 
2013 assuming bank holding companies have the average level of credit 
risk in 2008. 

Table 7: Comparisons of Bond Yield Spreads for Bank Holding Companies 
with $1 trillion versus $10 billion in Assets, All 42 Models, 2008 and 
2013 (basis points): 

2008, credit risk equal to average for 2008: 

Baseline model[A]; 
Comparison from model with volatility variable: 
Equity price volatility: -95.54***; (29.08); 
Option implied volatility: -46.22**; (19.07); 
Equity return volatility: -102.06***; (29.36); 
Excess equity return volatility: -88.86***; (29.24); 
Earnings volatility: -107.74***; (30.57); 
No volatility variable: -99.28***; (31.09). 

Add bid-ask spreads to bond liquidity measures; 
Comparison from model with volatility variable: 
Equity price volatility: -98.52***; (32.98); 
Option implied volatility: -21.30; (15.94); 
Equity return volatility: -104.08***; (31.57); 
Excess equity return volatility: -88.60**; (32.09); 
Earnings volatility: -106.98**; (37.49); 
No volatility variable: -104.99**; (37.40). 

Measure size using GSIB designation; 
Comparison from model with volatility variable: 
Equity price volatility: -148.78**; (60.01); 
Option implied volatility: -115.16***; (35.69); 
Equity return volatility: -157.00**; (60.78); 
Excess equity return volatility: -143.10**; (58.73); 
Earnings volatility: -145.12**; (55.21); 
No volatility variable: -148.98**; (57.39). 

Measure size using total assets and total assets squared; 
Comparison from model with volatility variable: 
Equity price volatility: -382.78***; (119.13); 
Option implied volatility: -629.74***; (82.69); 
Equity return volatility: -422.24***; (101.56); 
Excess equity return volatility: -270.29**; (113.61); 
Earnings volatility: -228.54*; (127.27); 
No volatility variable: -291.43**; (120.81). 

Measure capital adequacy using equity capital and subordinated debt as 
percents of risk-weighted assets; 
Comparison from model with volatility variable: 
Equity price volatility: -86.43**; (30.87); 
Option implied volatility: -44.90; (32.54); 
Equity return volatility: -96.96***; (31.12); 
Excess equity return volatility: -77.63**; (29.26); 
Earnings volatility: -96.79***; (32.30); 
No volatility variable: -92.96**; (34.10). 

Measure size using an indicator of whether a bank holding company has 
assets of $50 billion or more; 
Comparison from model with volatility variable: 
Equity price volatility: -432.86***; (90.04); 
Option implied volatility: -397.28**; (146.06); 
Equity return volatility: -420.42***; (67.65); 
Excess equity return volatility: -357.57***; (30.33); 
Earnings volatility: -515.38***; (133.82); 
No volatility variable: -357.49***; (73.88). 

Measure bond yield spreads using prices from executed trades only; 
Comparison from model with volatility variable: 
Equity price volatility: -79.73**; (31.77); 
Option implied volatility: -16.75; (20.62); 
Equity return volatility: -89.17**; (32.85); 
Excess equity return volatility: -73.39**; (32.98); 
Earnings volatility: -93.21**; (34.18); 
No volatility variable: -85.16**; (35.49). 

2013, credit risk equal to average for 2013: 

Baseline model[A]; 
Comparison from model with volatility variable: 
Equity price volatility: 28.50**; (11.04); 
Option implied volatility: 25.03***; (9.02); 
Equity return volatility: 30.66**; (11.46); 
Excess equity return volatility: 34.36***; (12.31); 
Earnings volatility: 23.03; (21.30); 
No volatility variable: 58.55***; (17.98). 

Add bid-ask spreads to bond liquidity measures; 
Comparison from model with volatility variable: 
Equity price volatility: 32.80***; (10.95); 
Option implied volatility: 29.63***; (8.431); 
Equity return volatility: 34.86***; (11.35); 
Excess equity return volatility: 38.49***; (12.41); 
Earnings volatility: 25.87; (21.28); 
No volatility variable: 62.60***; (17.88)]. 

Measure size using GSIB designation; 
Comparison from model with volatility variable: 
Equity price volatility: 3.07; (11.43); 
Option implied volatility: -11.18; (11.32); 
Equity return volatility: 7.49; (11.10); 
Excess equity return volatility: 28.13***; (9.52); 
Earnings volatility: 36.07**; (14.40); 
No volatility variable: 28.97; (20.42). 

Measure size using total assets and total assets squared; 
Comparison from model with volatility variable: 
Equity price volatility: -51.80**; (23.50); 
Option implied volatility: -41.43**; (18.95); 
Equity return volatility: -52.43**; (23.06); 
Excess equity return volatility: -36.89; (22.97); 
Earnings volatility: -19.01; (52.29); 
No volatility variable: 3.73; (43.74). 

Measure capital adequacy using equity capital and subordinated debt as 
percents of risk-weighted assets; 
Comparison from model with volatility variable: 
Equity price volatility: 23.17**; (10.53); 
Option implied volatility: 20.33**; (7.790); 
Equity return volatility: 24.33**; (11.00); 
Excess equity return volatility: 28.05**; (11.91); 
Earnings volatility: 0.264; (15.62); 
No volatility variable: 45.88**; (17.03). 

Measure size using an indicator of whether a bank holding company has 
assets of $50 billion or more; 
Comparison from model with volatility variable: 
Equity price volatility: -183.08**; (81.39); 
Option implied volatility: -195.76***; (59.77); 
Equity return volatility: -190.95**; (80.21); 
Excess equity return volatility: -160.67*; (85.47); 
Earnings volatility: -188.38***; (66.17); 
No volatility variable: -163.51; (100.2). 

Measure bond yield spreads using prices from executed trades only; 
Comparison from model with volatility variable: 
Equity price volatility: 17.32; (14.24); 
Option implied volatility: 13.21; (12.66); 
Equity return volatility: 20.06; (14.78); 
Excess equity return volatility: 24.87; (15.86); 
Earnings volatility: 13.39; (25.28); 
No volatility variable: 49.02***; (17.37). 

2013, credit risk equal to average for 2008: 

Baseline model[A]; 
Comparison from model with volatility variable: 
Equity price volatility: 48.92; (80.51); 
Option implied volatility: 62.89; (59.74); 
Equity return volatility: 36.25; (71.75); 
Excess equity return volatility: -15.81; (64.31); 
Earnings volatility: -2.14; (42.18); 
No volatility variable: -48.79; (43.31). 

Add bid-ask spreads to bond liquidity measures; 
Comparison from model with volatility variable: 
Equity price volatility: 64.91; (76.62); 
Option implied volatility: 86.12; (54.27); 
Equity return volatility: -51.00; (66.88); 
Excess equity return volatility: 3.65; (62.80); 
Earnings volatility: -0.28; (42.16); 
No volatility variable: -43.14; (44.27). 

Measure size using GSIB designation; 
Comparison from model with volatility variable: 
Equity price volatility: -290.72; (210.512); 
Option implied volatility: -264.12**; (103.19); 
Equity return volatility: -266.26; (197.211); 
Excess equity return volatility: -118.16; (184.807); 
Earnings volatility: -41.328; (55.122); 
No volatility variable: 28.798; (54.296). 

Measure size using total assets and total assets squared; 
Comparison from model with volatility variable: 
Equity price volatility: -455.05; (279.97); 
Option implied volatility: -327.16**; (141.48); 
Equity return volatility: -494.45*; (253.52); 
Excess equity return volatility: -473.93*; (242.11); 
Earnings volatility: -241.81; (195.99); 
No volatility variable: -160.75; (153.44). 

Measure capital adequacy using equity capital and subordinated debt as 
percents of risk-weighted assets; 
Comparison from model with volatility variable: 
Equity price volatility: -54.12; (68.63); 
Option implied volatility: -7.73; (71.92); 
Equity return volatility: -67.52; (58.98); 
Excess equity return volatility: -102.48**; (47.92); 
Earnings volatility: -45.56; (35.37); 
No volatility variable: -105.58***; (37.02). 

Measure size using an indicator of whether a bank holding company has 
assets of $50 billion or more; 
Comparison from model with volatility variable: 
Equity price volatility: -194.56; (144.1); 
Option implied volatility: -606.39*; (307.14); 
Equity return volatility: -373.43*; (195.44); 
Excess equity return volatility: -355.84***; (107.35); 
Earnings volatility: -246.16*; (143.55); 
No volatility variable: -154.46; (144.58). 

Measure bond yield spreads using prices from executed trades only; 
Comparison from model with volatility variable: 
Equity price volatility: 38.47; (87.83); 
Option implied volatility: 35.45; (61.70); 
Equity return volatility: 36.14; (75.85); 
Excess equity return volatility: -1.00; (64.28); 
Earnings volatility: 12.43; (35.46); 
No volatility variable: -50.39; (41.98). 

Source: GAO analysis of Bloomberg and Financial Stability Board data. 
GAO-14-621. 

Notes: We estimated econometric models of the relationship between 
bank holding company size and funding costs using data for U.S. bank 
holding companies and their outstanding senior unsecured bonds for the 
first quarter of 2006 through the fourth quarter of 2013. The models 
used bond yield spreads to measure funding costs and controlled for 
credit risk factors such as capital adequacy, asset quality, earnings, 
maturity mismatch, and volatility, as well as bond liquidity and other 
characteristics of bonds and bank holding companies that can affect 
funding costs. We estimated multiple models for each year from 2006 
through 2013 to assess the sensitivity of estimated funding cost 
differences to alternative measures of capital adequacy, volatility, 
bond liquidity, and size or systemic importance and then used the 
models to compare bond funding costs for bank holding companies of 
different sizes but the same levels of credit risk, bond liquidity, 
and other characteristics. This table shows bond funding cost 
comparisons from models for 2008 and 2013 for bank holding companies 
with the average level of credit risk for those years, and from models 
for 2013 for bank holding companies with the average level of credit 
risk for 2008. *=statistically significant at the 10 percent level. 
**=statistically significant at the 5 percent level. ***=statistically 
significant at the 1 percent level. 

[A] The baseline model used average bond yield spreads calculated 
using prices from executed trades, executable quotes, and composites 
of executable and indicative quotes to measure bond funding costs; 
total assets to measure bank holding company size; issue size and 
total trade volume to measure bond liquidity; equity capital and 
subordinated debt as percents of assets to measure capital adequacy; 
and each of equity price volatility, option implied volatility, equity 
return volatility, excess equity return volatility, and earnings 
volatility to measure volatility, as well as a model with no 
volatility measure. Other explanatory variables include coupon, time 
to maturity, Texas ratio, maturity mismatch, and operating expenses. 

[End of table] 

Figure 4 shows estimated bond funding costs for bank holding companies 
with $1 trillion in assets versus those with $10 billion in assets by 
model and year, assuming that all bank holding companies have the 
average level of credit risk in each year. 

Figure 4: Difference in Estimated Bond Funding Costs for Average 
Credit Risk Bank Holding Companies with $1 Trillion versus $10 Billion 
in Assets by Model Specification, by Year, 2006-2013: 

[Refer to PDF for image: illustrated table] 

Average difference estimated by one of 42 models (statistically 
significant at 10% level); 

Average difference estimated by one of 42 models (not statistically 
significant at 10% level). 

Baseline specification: 

2006: 
Total statistically insignificant number above/below zero: 0/2; 
Total statistically significant number above/below zero: 0/4. 

2007: 
Total statistically insignificant number above/below zero: 0/1; 
Total statistically significant number above/below zero: 0/5. 

2008: 
Total statistically insignificant number above/below zero: 0/0; 
Total statistically significant number above/below zero: 0/6. 

2009: 
Total statistically insignificant number above/below zero: 0/1; 
Total statistically significant number above/below zero: 0/5. 

2010: 
Total statistically insignificant number above/below zero: 2/3; 
Total statistically significant number above/below zero: 1/0. 

2011: 
Total statistically insignificant number above/below zero: 2/0; 
Total statistically significant number above/below zero: 4/0. 

2012: 
Total statistically insignificant number above/below zero: 1/0; 
Total statistically significant number above/below zero: 5/0. 

2013: 
Total statistically insignificant number above/below zero: 1/0; 
Total statistically significant number above/below zero: 5/0. 

Global systemically important bank specification: 

2006: 
Total statistically insignificant number above/below zero: 1/1; 
Total statistically significant number above/below zero: 0/4. 

2007: 
Total statistically insignificant number above/below zero: 5/1; 
Total statistically significant number above/below zero: 0/0. 

2008: 
Total statistically insignificant number above/below zero: 0/0; 
Total statistically significant number above/below zero: 0/6. 

2009: 
Total statistically insignificant number above/below zero: 0/0; 
Total statistically significant number above/below zero: 0/6. 

2010: 
Total statistically insignificant number above/below zero: 0/5; 
Total statistically significant number above/below zero: 0/1. 

2011: 
Total statistically insignificant number above/below zero: 0/2; 
Total statistically significant number above/below zero: 0/4. 

2012: 
Total statistically insignificant number above/below zero: 0/5; 
Total statistically significant number above/below zero: 0/1. 

2013: 
Total statistically insignificant number above/below zero: 3/1; 
Total statistically significant number above/below zero: 2/0. 

Total assets and total assets squared specification: 

2006: 
Total statistically insignificant number above/below zero: 0/2; 
Total statistically significant number above/below zero: 0/4. 

2007: 
Total statistically insignificant number above/below zero: 1/0; 
Total statistically significant number above/below zero: 0/5. 

2008: 
Total statistically insignificant number above/below zero: 0/0; 
Total statistically significant number above/below zero: 0/6. 

2009: 
Total statistically insignificant number above/below zero: 0/0; 
Total statistically significant number above/below zero: 0/6. 

2010: 
Total statistically insignificant number above/below zero: 0/0; 
Total statistically significant number above/below zero: 0/6. 

2011: 
Total statistically insignificant number above/below zero: 1/1; 
Total statistically significant number above/below zero: 0/4. 

2012: 
Total statistically insignificant number above/below zero: 1/3; 
Total statistically significant number above/below zero: 0/2. 

2013: 
Total statistically insignificant number above/below zero: 1/2; 
Total statistically significant number above/below zero: 0/3. 

Systemically important financial institution specification
(total assets $50 billion or more): 

2006: 
Total statistically insignificant number above/below zero: 1/2; 
Total statistically significant number above/below zero: 0/3. 

2007: 
Total statistically insignificant number above/below zero: 0/2; 
Total statistically significant number above/below zero: 0/4. 

2008: 
Total statistically insignificant number above/below zero: 0/0; 
Total statistically significant number above/below zero: 0/6. 

2009: 
Total statistically insignificant number above/below zero: 0/0; 
Total statistically significant number above/below zero: 0/6. 

2010: 
Total statistically insignificant number above/below zero: 3/3; 
Total statistically significant number above/below zero: 0/0. 

2011: 
Total statistically insignificant number above/below zero: 0/0; 
Total statistically significant number above/below zero: 0/6. 

2012: 
Total statistically insignificant number above/below zero: 1/2; 
Total statistically significant number above/below zero: 3/0. 

2013: 
Total statistically insignificant number above/below zero: 0/1; 
Total statistically significant number above/below zero: 0/5. 

Risk-weighted assets specification: 

2006: 
Total statistically insignificant number above/below zero: 1/0; 
Total statistically significant number above/below zero: 0/5. 

2007: 
Total statistically insignificant number above/below zero: 0/2; 
Total statistically significant number above/below zero: 0/4. 

2008: 
Total statistically insignificant number above/below zero: 0/1; 
Total statistically significant number above/below zero: 0/5. 

2009: 
Total statistically insignificant number above/below zero: 0/4; 
Total statistically significant number above/below zero: 0/2. 

2010: 
Total statistically insignificant number above/below zero: 1/5; 
Total statistically significant number above/below zero: 0/0. 

2011: 
Total statistically insignificant number above/below zero: 5/1; 
Total statistically significant number above/below zero: 0/0. 

2012: 
Total statistically insignificant number above/below zero: 3/1; 
Total statistically significant number above/below zero: 2/0. 

2013: 
Total statistically insignificant number above/below zero: 1/0; 
Total statistically significant number above/below zero: 5/0. 

Bid-ask specification: 

2006: 
Total statistically insignificant number above/below zero: 0/2; 
Total statistically significant number above/below zero: 0/4. 

2007: 
Total statistically insignificant number above/below zero: 1/4; 
Total statistically significant number above/below zero: 0/1. 

2008: 
Total statistically insignificant number above/below zero: 0/1; 
Total statistically significant number above/below zero: 0/5. 

2009: 
Total statistically insignificant number above/below zero: 0/0; 
Total statistically significant number above/below zero: 0/6. 

2010: 
Total statistically insignificant number above/below zero: 3/1; 
Total statistically significant number above/below zero: 1/1. 

2011: 
Total statistically insignificant number above/below zero: 1/0; 
Total statistically significant number above/below zero: 5/0. 

2012: 
Total statistically insignificant number above/below zero: 1/0; 
Total statistically significant number above/below zero: 5/0. 

2013: 
Total statistically insignificant number above/below zero: 1/0; 
Total statistically significant number above/below zero: 5/0. 

Trade reporting and compliance engine specification: 

2006: 
Total statistically insignificant number above/below zero: 1/1; 
Total statistically significant number above/below zero: 0/4. 

2007: 
Total statistically insignificant number above/below zero: 0/5; 
Total statistically significant number above/below zero: 0/1. 

2008: 
Total statistically insignificant number above/below zero: 0/1; 
Total statistically significant number above/below zero: 0/5. 

2009: 
Total statistically insignificant number above/below zero: 0/0; 
Total statistically significant number above/below zero: 0/6. 

2010: 
Total statistically insignificant number above/below zero: 3/2; 
Total statistically significant number above/below zero: 0/1. 

2011: 
Total statistically insignificant number above/below zero: 4/0; 
Total statistically significant number above/below zero: 2/0. 

2012: 
Total statistically insignificant number above/below zero: 1/4; 
Total statistically significant number above/below zero: 0/1. 

2013: 
Total statistically insignificant number above/below zero: 5/0; 
Total statistically significant number above/below zero: 1/0. 

Estimated average bond funding costs for larger bank holding companies
(BHC) were higher than for smaller BHCs. 

Estimated average bond funding costs for larger BHCs were lower than 
for smaller BHCs. 

Basis point = .01% (1/100th of a percent). 

Source: GAO analysis of data from Bloomberg and the Financial 
Stability Board. GAO-14-621. 

Notes: We estimated econometric models of the relationship between 
bank holding company size and funding costs using data for U.S. bank 
holding companies and their outstanding senior unsecured bonds for the 
first quarter of 2006 through the fourth quarter of 2013. The models 
used bond yield spreads to measure funding costs and controlled for 
credit risk factors such as capital adequacy, asset quality, earnings, 
maturity mismatch, and volatility, as well as bond liquidity and other 
characteristics of bonds and bank holding companies that can affect 
funding costs. We estimated 42 models for each year from 2006 through 
2013 to assess the sensitivity of estimated differences in funding 
costs to alternative measures of capital adequacy, volatility, bond 
liquidity, and size or systemic importance. We used the models to 
compare bond funding costs for bank holding companies of different 
sizes but the same levels of credit risk, bond liquidity, and other 
characteristics. This figure compares bond funding costs for bank 
holding companies with $1 trillion and $10 billion in assets, for each 
model and for each year, with average levels of credit risk. Each 
circle and dash shows the comparison for a different model, where 
circles and dashes below zero suggest bank holding companies with $1 
trillion in assets have lower bond funding costs than bank holding 
companies with $10 billion in assets, and vice versa. 

[End of figure] 

[End of section] 

Appendix II: Comments from the Department of the Treasury: 

Department of The Treasury: 
Under Secretary: 
Washington, D.C. 

July 21, 2014: 

Lawrance Evans, Jr. 
Director: 
Financial Markets and Community Investment: 
Government Accountability Office: 
441 G Street, NW: 
Washington, DC 20548: 

Dear Mr. Evans: 

I am writing regarding the Government Accountability Office's (GAO) 
draft report GAO-14-621 (the Draft Report), Large Bank Holding 
Companies: Expectations of Government Support. 

Five years ago, the U.S. economy was reeling from the devastating 
financial crisis that helped trigger the worst recession since the 
Great Depression. To prevent our economy from melting down, the 
American taxpayer was forced to provide extraordinary assistance to 
financial companies. While the President and Treasury Secretary acted 
decisively to put out the financial fires, they were also determined 
to ensure that such a crisis never happens again. The Administration's 
efforts produced the most comprehensive overhaul of our financial 
system since the Great Depression—the Dodd-Frank Wall Street Reform 
and Consumer Protection Act. The Act's reforms impose more stringent 
requirements on large firms through heightened capital and liquidity 
standards, put in place living wills and resolution authority, and 
prohibit taxpayer bailouts. The law makes clear that shareholders, 
creditors, and executives—-not taxpayers-—will be responsible if a 
large financial company fails. 

The Draft Report reflects an impressive amount of empirical work using 
more recent data than existing research and represents a meaningful 
contribution to the literature. We appreciate the complexity of the 
task you and your team undertook, as well as the limitations of 
available data for comparing funding costs across institutions of 
different size. 

The Draft Report asks readers to interpret its findings with caution, 
and we recognize the issue merits continued monitoring and further 
study. Nevertheless, the results of GAO's analysis are noteworthy: 
While large bank holding companies had clear funding cost advantages 
during the 2007-2009 financial crisis, these advantages had declined 
or reversed by 2013. Specifically, GAO ran 42 models for 2013, 26 of 
which produced statistically significant results. Of these 26 models, 
18 showed that larger bank holding companies had higher funding costs 
than their smaller counterparts in 2013. In other words, more than two-
thirds of these models found a funding cost disadvantage for large 
institutions. The remaining eight models showed a funding cost 
advantage, but one that had declined since the financial crisis. We 
believe these results reflect increased market recognition of what 
should now be evident—Dodd-Frank ended "too big to fail" as a matter 
of law. 

We thank all those involved for their professionalism and diligence 
throughout the study process. Treasury is continuing its efforts to 
fully implement the Dodd-Frank reforms to reduce the likelihood and 
the impact of the failure of a large, interconnected financial 
company. We also recognize that we must remain vigilant on this issue. 
We look forward to working with you on additional matters related to 
strengthening the financial system. 

Sincerely, 

Signed by: 

Mary J. Miller: 

[End of section] 

Appendix III: GAO Contact and Staff Acknowledgments: 

GAO Contact: 

Lawrance L. Evans, Jr., (202) 512-4802, or EvansL@gao.gov: 

Staff Acknowledgments: 

In addition to the contact named above, Karen Tremba (Assistant 
Director), John Fisher (Analyst in Charge), Bethany Benitez, Michael 
Hoffman, Risto Laboski, Courtney LaFountain, Rob Letzler, Marc Molino, 
Jason Wildhagen, and Jennifer Schwartz made significant contributions 
to this report. Other assistance was provided by Abigail Brown, Rudy 
Chatlos, Stephanie Cheng, and José R. Peña. 

[End of section] 

Bibliography: 

Acharya, Viral V., Deniz Anginer, and A. Joseph Warburton. "The End of 
Market Discipline? Investor Expectations of Implicit State 
Guarantees." Social Science Research Network Working Paper. December 
2013. 

Araten, Michel and Christopher M. Turner. "Understanding the Funding 
Cost Differences between Global Systemically Important Banks (G-SIBs) 
and Non-G-SIBs in the United States." Social Science Research Network 
Working Paper. March 2012. 

Balasubramnian, Bhanu and Ken Cyree. "Has Market Discipline on Banks 
Improved after the Dodd-Frank Act?" Journal of Banking & Finance, vol. 
41 (2014): 155-166. 

Balasubramnian, Bhanu and Ken Cyree. "The End of Too-Big-to-Fail? 
Evidence from Senior Bank Bond Yield Spreads around the Dodd-Frank 
Act." Social Science Research Network Working Paper. June 2012. 

Barth, Andreas and Isabel Schnabel. "Why Banks are Not Too Big To Fail-
-Evidence from the CDS Market." Economic Policy, vol. 28, no. 74. 
April 2013: 335-369. 

Bertay, Ata Can, Asli Demirgüç-Kunt, and Harry Huizinga. "Do We Need 
Big Banks? Evidence on Performance, Strategy and Market Discipline." 

Journal of Financial Intermediation, vol. 22 (2013): 532-558. 

International Monetary Fund. "How Big is the Implicit Subsidy for 
Banks Considered Too Important to Fail?" Global Financial Stability 
Report, Ch. 3 (Washington, D.C.: 2014). 

Jacewitz, Stefan and Jonathan Pogach. "Deposit Rate Advantages at the 
Largest Banks." Social Science Research Network Working Paper. 
September 2012. 

Keppo, Jussi and Jing Yang. The Value of Too-Big-To-Fail Subsidy. 
Unpublished working paper. September 2013. 

Kumar, Aditi and John Lester. Do Deposit Rates Show Evidence of Too 
Big To Fail Effects? An Updated Look at the Empirical Evidence through 
2012 among US Banks (Oliver Wyman, Inc., 2014). 

Kumar, Aditi and John Lester. Do Bond Spreads Show Evidence of Too Big 
To Fail Effects? Evidence from 2009-2013 among US Bank Holding 
Companies (Oliver Wyman, Inc., 2014). 

Li, Zan, Shisheng Qu and Jing Zhang. Quantifying the Value of Implicit 
Government Guarantees for Large Financial Institutions (Moody's 
Analytics, 2011). 

Santos, João. "Evidence from the Bond Market on Banks' 'Too Big To 
Fail' Subsidy." Federal Reserve Bank of New York Economic Policy 
Review, vol. 20, no. 2. March 2014. 

Tsesmelidakis, Zoe and Robert C. Merton. "The Value of Implicit 
Guarantees." Social Science Research Network Working Paper. September 
2012. 

Ueda, Kenichi, and B. Weder di Mauro. "Quantifying Structural Subsidy 
Values for Systemically Important Financial Institutions," Journal of 
Banking & Finance, vol. 37, no. 10. October 2013: 3830-3842. 

Völz, Manja and Michael Wedow. "Does Banks' Size Distort Market 
Prices? Evidence for Too-Big-To-Fail in the CDS Market." Deutsche 
Bundesbank Discussion Paper, Series 2: Banking and Financial Studies, 
no. 6 (2009). 

[End of section] 

Footnotes: 

[1] Bankruptcy is a federal court procedure conducted under rules and 
requirements of the U.S. Bankruptcy Code. The goal of bankruptcy is to 
give individuals and businesses a "fresh start" from burdensome debts 
by eliminating or restructuring debts they cannot repay and helping 
creditors receive some payment in an equitable manner through 
liquidation or reorganization of the debtor. 

[2] Pub. L. No. 111-203, 124 Stat. 1376 (2010). 

[3] In November 2013, we issued the first report, which examined (1) 
actual government support for banks and bank holding companies during 
the financial crisis and (2) recent statutory and regulatory changes 
related to government support for banks and bank holding companies. 
See GAO, Government Support for Bank Holding Companies: Statutory 
Changes to Limit Future Support Are Not Yet Fully Implemented, 
[hyperlink, http://www.gao.gov/products/GAO-14-18] (Washington, D.C.: 
Nov. 14, 2013). At a January 2014 hearing, we provided testimony based 
on this report. See GAO, Government Support for Bank Holding 
Companies: Statutory Changes to Limit Future Support Are Not Yet Fully 
Implemented, [hyperlink, http://www.gao.gov/products/GAO-14-174T] 
(Washington, D.C.: Jan. 8, 2014). 

[4] For example, we reviewed [hyperlink, 
http://www.gao.gov/products/GAO-14-18], which describes Dodd-Frank Act 
provisions intended to place limits on agency authorities to provide 
emergency assistance to financial firms and strengthen regulatory 
oversight of large bank holding companies. We also reviewed prior GAO 
work on potential impacts of the Dodd-Frank Act. See GAO, Financial 
Regulatory Reform: Financial Crisis Losses and Potential Impacts of 
the Dodd-Frank Act, [hyperlink, 
http://www.gao.gov/products/GAO-13-180] (Washington, D.C.: Jan. 16, 
2013). 

[5] See [hyperlink, http://www.gao.gov/products/GAO-13-180]. 

[6] Negative values of and are consistent with these beliefs as long 
as greater values of the size measure are associated with larger bank 
holding companies and greater values of the credit risk proxies are 
associated with greater credit risk. 

[7] Plain vanilla bonds refer to bonds that pay interest at regular 
intervals and return the principal at maturity with no additional 
features such as convertibility to stock or options to redeem the bond 
before maturity. Such bonds are also known as straight bonds. 

[8] We estimated the parameters of our models using the least squares 
estimator, which is a standard statistical and econometric technique 
that calculates the values of the parameters of our regression model 
that minimize the sum of the squared errors--the differences between 
bond funding costs observed in the data and bond funding costs 
predicted by the model. 

[9] The Dodd-Frank Act requires bank holding companies with $50 
billion or more in total assets to be subjected to enhanced regulatory 
standards and supervision. Pub. L. No. 111-203, § 165, 124 Stat. 1376, 
1423 (2010) (codified at 12 U.S.C. § 5365). In addition, the Financial 
Stability Board designated 29 GSIBs, including eight U.S. bank holding 
companies, for the purpose of identifying firms that should be subject 
to resolution planning requirements and heightened regulatory 
supervision. The GSIBs generally comprise the largest and most complex 
internationally active financial firms. The bid-ask spread is the 
difference between the best buying price and the best selling price. 

[10] Our models allow the size of a bank holding company to influence 
its bond funding costs directly and also indirectly through the 
interactions between size and the credit risk variables. As a result, 
no single parameter is sufficient to describe the relationship between 
bond funding costs and size. To summarize the overall relationship 
between bond funding costs and size reflected in each specification, 
we calculated bond funding costs for bank holding companies of 
different sizes and credit risk levels using our estimates of the 
parameters for each specification for each year. See appendix I for 
more details on the calculations. 

[11] For more information about the government rescue of Continental 
Illinois National Bank, see GAO, Financial Markets and Institutions: 
Views on the Federal Rescue of the Continental Illinois National Bank 
and Trust Co, Testimony by the Comptroller General of the U.S. before 
the Subcommittee on Financial Institutions Supervision, Regulation, 
and Insurance of the Committee on Banking, Finance and Urban Affairs 
of the U.S. House of Representatives (Washington, D.C.: Dec. 14, 
1984); and GAO, Financial Crisis Management: Four Financial Crises in 
the 1980s, [hyperlink, http://www.gao.gov/products/GAO/GGD-97-96] 
(Washington, D.C.: May 1, 1997). 

[12] For more information on the near-collapse of Long-Term Capital 
Management and broader issues it raised, see GAO, Long-Term Capital 
Management: Regulators Need to Focus Greater Attention on Systemic 
Risk, [hyperlink, http://www.gao.gov/products/GAO/GGD-00-3] 
(Washington, D.C.: Oct. 29, 1999). 

[13] For more discussion of the federal government's emergency actions 
to assist the financial sector, see [hyperlink, 
http://www.gao.gov/products/GAO-14-18]. 

[14] At the time, Bear Stearns was one of the largest investment 
banks. For further discussion of this assistance, see GAO, Federal 
Reserve System: Opportunities Exist to Strengthen Policies and 
Processes for Managing Emergency Assistance, [hyperlink, 
http://www.gao.gov/products/GAO-11-696] (Washington, D.C.: July 21, 
2011). During the weekend of September 13-15, 2008, government 
officials met with leaders of major financial firms to devise a 
private-sector solution to Lehman Brothers' likelihood of defaulting 
on its obligations, but they were unable to find a solution. While the 
Federal Reserve Board was able to provide liquidity to Lehman Brothers 
against collateral through its emergency credit programs, according to 
the Federal Reserve Board Chairman at that time, neither the Federal 
Reserve Board nor any other agency had the authority to provide the 
capital or unsecured guarantee of its obligations that they believed 
Lehman Brothers needed to avert failure. See Ben S. Bernanke, Board of 
Governors of the Federal Reserve System, "Lessons from the Failure of 
Lehman Brothers" (testimony before the Committee on Financial 
Services, U.S. House of Representatives, Washington, D.C., Apr. 20, 
2010). 

[15] Subsequent to the announcement of the $85 billion credit 
assistance, Treasury and the Federal Reserve provided new forms of 
assistance and restructured the assistance package for AIG over time. 
For more information about the federal government's assistance to AIG, 
see GAO, Troubled Asset Relief Program: Status of Government 
Assistance Provided to AIG, [hyperlink, 
http://www.gao.gov/products/GAO-09-975] (Washington, D.C.: Sept. 21, 
2009). 

[16] A few days after the announcement of the proposed Citigroup 
acquisition, Wachovia announced that it would instead merge with Wells 
Fargo in a transaction that would include all of Wachovia's operations 
and require no FDIC assistance. As a result, the FDIC loss-sharing 
agreement on Wachovia assets was not implemented. FDICIA included a 
systemic risk exception to the requirement that FDIC resolve failed 
banks using the least costly method. Pub. L. No. 102-242, § 141, 105 
Stat. 2236, 2275 (codified at 12 U.S.C. § 1823(c)(4)(G)). Under this 
exception, FDIC could provide assistance to a failing bank if 
compliance with its requirements to resolve the bank using the least 
costly approach would have "serious adverse effects on economic 
conditions or financial stability"--that is, would cause systemic risk-
-and if such assistance would "avoid or mitigate such adverse 
effects." Id. FDIC could act under the exception only under a process 
that included recommendations from the FDIC Board of Directors and 
Federal Reserve Board and approval by the Secretary of the Treasury. 
The Dodd-Frank Act restricts FDIC's authority to provide open bank 
assistance to an individual failing bank outside of receivership and 
replaces it with a new authority, subject to certain restrictions and 
a joint resolution of congressional approval, to create a debt-
guarantee program with broad-based eligibility. Pub. L. No. 111-203, § 
1105(a), 124 Stat. 1376, 2121 (2010) (codified at 12 U.S.C. §§ 5612(a) 
and 5613(b)). 

[17] The G7 is an informal forum of coordination among Canada, France, 
Germany, Italy, Japan, the United Kingdom, and the United States. 

[18] For more about the Troubled Asset Relief Program, see GAO, 
Troubled Asset Relief Program: One Year Later, Actions Are Needed to 
Address Remaining Transparency and Accountability Challenges, 
[hyperlink, http://www.gao.gov/products/GAO-10-16] (Washington, D.C.: 
Oct. 8, 2009). For more about the Temporary Liquidity Guarantee 
Program, see GAO, Federal Deposit Insurance Act: Regulators' Use of 
Systemic Risk Exception Raises Moral Hazard Concerns and Opportunities 
Exist to Clarify the Provision, [hyperlink, 
http://www.gao.gov/products/GAO-10-100] (Washington, D.C.: Apr. 15, 
2010). 

[19] The share of total bank assets held by the five largest global 
banks has increased in many advanced countries and highlights concerns 
about too big to fail. 

[20] The G20, established in 1999, is a forum for international 
cooperation on important issues of the global economic and financial 
agenda. Its members include 19 countries and the European Union. The 
G20 leaders established the Financial Stability Board as the successor 
to the Financial Stability Forum and made it responsible for 
coordinating and promoting the implementation of the G20 reform 
commitments. 

[21] Financial Stability Board, "Reducing the Moral Hazard Posed by 
Systemically Important Financial Institutions-FSB Recommendations and 
Time Lines," October 20, 2010. In September 2013, the Financial 
Stability Board issued a report providing an update on progress in 
implementing this framework. See Financial Stability Board, "Progress 
and Next Steps Towards Ending 'Too-Big-to-Fail' (TBTF) - Report of the 
Financial Stability Board to the G20," September 2, 2013. 

[22] Financial Stability Board, "Key Attributes of Effective 
Resolution Regimes for Financial Institutions," October 2011. 

[23] For more information about the U.S. role in international 
financial reform efforts, including the development of international 
financial standards, see GAO, International Financial Reforms: U.S. 
and Other Jurisdictions' Efforts to Develop and Implement Reforms, 
[hyperlink, http://www.gao.gov/products/GAO-14-261] (Washington, D.C.: 
Apr. 3, 2014). 

[24] While the Dodd-Frank Act does not use the term "systemically 
important financial institution," this term is commonly used by 
academics and other experts to refer to bank holding companies with 
$50 billion or more in total consolidated assets and nonbank financial 
companies designated by FSOC for Federal Reserve supervision and 
enhanced prudential standards. 

[25] Pub. L. No. 111-203, §§ 1101 and 1105, 124 Stat. 1376, 2113 and 
2121 (2010) (codified at 12 U.S.C. §§ 343(3)(A) and 5612). For more 
information about changes to these authorities, see [hyperlink, 
http://www.gao.gov/products/GAO-14-18]. 

[26] § 165(d), 124 Stat. at 1426 (codified at 12 U.S.C. § 5365(d)); § 
204, 124 Stat. at 1454-56 (codified at 12 U.S.C. § 5384). During the 
financial crisis, several large financial institutions became 
insolvent and filed for bankruptcy when no private-sector solution was 
found. For example, Lehman Brothers filed for Chapter 11 bankruptcy on 
September 15, 2008. Lehman had $639 billion in total assets and $613 
billion in total liabilities as of May 31, 2008. The bankruptcy 
proceedings highlighted inconsistencies in laws and regulations across 
countries and limitations on the ability of countries to coordinate 
effectively during the reorganization or liquidation of international 
financial institutions. 

[27] § 165(d)(1), 124 Stat. at 1426 (codified at 12 U.S.C. § 
5365(d)(1)). 

[28] Resolution Plans Required, 76 Fed. Reg. 67323 (Nov. 1, 2011). 

[29] In October 2013, the Federal Reserve Board and FDIC released the 
public sections of filed annual resolution plans for 11 firms. 
"Federal Reserve Board and FDIC, Agencies Release Public Sections of 
the Second Submission of Resolution Plans for 11 Institutions," 
accessed on November 7, 2013, [hyperlink, 
http://www.federalreserve.gov/newsevents/press/bcreg/20131003a.htm]. 

[30] § 165(d)(4)-(5), 124 Stat. at 1426-27 (codified at 12 U.S.C. § 
5365(d)(4)-(5)). Further, if, after 2 years following the imposition 
of the more stringent standards, the resolution plan still does not 
meet the statutory standards, FDIC and the Federal Reserve Board may, 
in consultation with FSOC, direct a company to divest certain assets 
or operations. Id. 

[31] § 204, 124 Stat. at 1454-1456 (codified at 12 U.S.C. § 5384). 

[32] § 204(b), 124 Stat. at 1455 (codified at 12 U.S.C. § 5384(b)). 
The factors to be considered by the Secretary of the Treasury are set 
forth in Section 203(b) of the Dodd-Frank Act. § 203(b), 124 Stat. at 
1451 (codified at 12 U.S.C. § 5383(b)). Before the Secretary of the 
Treasury, in consultation with the President, makes a decision to seek 
the appointment of FDIC as receiver of a financial company, at least 
two-thirds of those serving on the Board of Governors of the Federal 
Reserve System and at least two-thirds of those serving on the Board 
of Directors of FDIC must vote to make a written recommendation to the 
Secretary of the Treasury to appoint FDIC as receiver. §203(a)(1)(A), 
124 Stat. at 1450 (codified at 12 U.S.C. § 5383(a)(1)(A)). In the case 
of a broker-dealer, the recommendation must come from the Federal 
Reserve Board and the Securities and Exchange Commission, in 
consultation with FDIC, and in the case of an insurance company, from 
the Federal Reserve Board and the Director of the Federal Insurance 
Office, in consultation with FDIC. § 203(a)(1)(B)-(C), 124 Stat. at 
1450 (codified at 12 U.S.C. § 5383(a)(1)(B)-(C)). 

[33] § 204(a)(2), 124 Stat. at 1454 (codified at 12 U.S.C. § 
5384(a)(2)). 

[34] § 210(n), 124 Stat. at 1506 (codified at 12 U.S.C. § 5390(n)). 

[35] § 210(o)(1), 124 Stat. at 1509 (codified at 12 U.S.C. § 
5390(o)(1)). 

[36] §165(a)(1), 124 Stat. at 1423 (codified at 12 U.S.C. § 
5365(a)(1)). 

[37] §165(b)(1)(A)(i), 124 Stat. at 1424; Enhanced Prudential 
Standards for Bank Holding Companies and Foreign Banking 
Organizations; Final Rule, 79 Fed. Reg. 17240 (Mar. 27, 2014) 
(codified at 12 C.F.R. pt. 252). 

[38] The Federal Reserve Board plans to issue a proposal to impose 
graduated common equity risk-based capital surcharges on U.S. bank 
holding companies designated as global systemically important banks 
(GSIB). The Federal Reserve Board has said this proposal will be based 
on the GSIB capital surcharge framework developed by the Basel 
Committee, under which the size of the surcharge for an individual 
GSIB would vary from 1 percent to 2.5 percent, depending on the firm's 
systemic importance. 

[39] §165(b)(1)(A)(ii), 124 Stat. at 1424 (codified at 12 U.S.C. § 
5365(b)(1)(A)(ii)). Enhanced Prudential Standards for Bank Holding 
Companies and Foreign Banking Organizations; Final Rule, 79 Fed. Reg. 
17240 (Mar. 27, 2014) (codified at 12 C.F.R. pt. 252). 

[40] Liquidity Coverage Ratio: Liquidity Risk Measurement, Standards, 
and Monitoring; Proposed Rule, 78 Fed. Reg. 71818 (Nov. 29, 2013).The 
requirement is designed to promote the short-term resilience of the 
liquidity risk profile of internationally active banking 
organizations, thereby improving the banking sector's ability to 
absorb shocks arising from financial and economic stress. Id. 

[41] §§ 165(b)(1)(A)(iii) and 165(h), 124 Stat. at 1424 and 1429 
(codified at 12 U.S.C. §§ 5365(b)(1)(A)(iii) and 5365(h)). 

[42] §165(i)(1), 124 Stat. at 1430 (codified at 12 U.S.C. § 
5365(i)(1)). Companies subject to enhanced prudential standards as 
well as regulated financial firms with more than $10 billion in 
consolidated assets also must conduct their own semiannual or annual 
stress tests, respectively. § 165(i)(2), 124 Stat. at 1430-31 
(codified at 12 U.S.C. § 5365(i)(2)). 

[43] The Basel Committee has developed international standards for 
bank capital for its member economies since the 1980s. In recent 
years, U.S. federal banking regulators have worked with other members 
of the Basel Committee to strengthen the regulatory capital regime for 
internationally active banks and develop a framework for a risk-based 
capital surcharge for the world's largest, most interconnected banking 
companies. The new regime, known as Basel III, seeks to improve the 
quality of regulatory capital and introduces a new minimum common 
equity requirement. Basel III also raises the numerical minimum 
capital requirements and introduces capital conservation and 
countercyclical buffers to require banking organizations to hold 
capital in excess of regulatory minimums. In addition, Basel III 
establishes for the first time an international leverage standard for 
internationally active banks. Federal banking regulators recently 
finalized capital reforms in the United States that are generally 
consistent with Basel III. The Federal Reserve Board will separately 
implement consistent capital and liquidity standards for nonbank 
financial companies designated for enhanced supervision by FSOC. 

[44] Regulatory Capital Rules: Regulatory Capital, Enhanced 
Supplementary Leverage Ratio Standards for Certain Bank Holding 
Companies and Their Subsidiary Insured Depository Institutions, 79 
Fed. Reg. 24528 (May 1, 2014). 

[45] § 121(a)(1), 124 Stat. at 1410 (codified at 12 U.S.C. § 
5331(a)(1)); § 165(a)(1), 124 Stat. at 1423 (codified at 12 U.S.C. § 
5365(a)(1)); § 165(d)(4)-(5), 124 Stat. at 1426-27 (codified at 12 
U.S.C. § 5365(d)(4)-(5)). 

[46] § 619, 124 Stat. at 1620-31 (codified at 12 U.S.C. § 1851). 
Proprietary trading includes trading activities conducted by banking 
entities for their own account as opposed to those of their clients. 

[47] A swap is a type of derivative that involves an ongoing exchange 
of one or more assets, liabilities, or payments for a specified 
period. Financial and nonfinancial firms use swaps and other over-the-
counter derivatives to hedge risk, or speculate, or for other purposes. 

[48] See [hyperlink, http://www.gao.gov/products/GAO-13-180]. 
Counterparty credit risk is the risk to each party in an over-the-
counter derivatives contract that the other party will not perform the 
contractual obligations. Technically, the clearing house members 
interact with the counterparties. 

[49] FSOC has identified certain clearinghouses as systemically 
important financial market utilities, which are subject to risk 
management and other enhanced supervisory and prudential requirements 
under the Dodd-Frank Act and may be afforded access to collateralized 
emergency liquidity from Federal Reserve Banks in unusual or exigent 
circumstances. 

[50] In May 2011, OCC, the Federal Reserve Board, FDIC, the Farm 
Credit Administration, and the Federal Housing Finance Agency issued a 
notice of proposed rule-making on margin and capital requirements for 
covered swap entities. Margin and Capital Requirements for Covered 
Swap Entities; Proposed Rule, 76 Fed. Reg. 27564 (May 11, 2011). 

[51] Resolution of Systemically Important Financial Institutions: The 
Single Point of Entry Strategy, 78 Fed. Reg. 243 (Dec. 18, 2013); 
Resolution of a Systemically Important Financial Institution: The 
Single Point of Entry Strategy, 79 Fed. Reg. 9899 (Feb. 21, 2014). 

[52] In March 2014, the Basel Committee published a standardized 
approach to measuring counterparty credit risk exposures. This 
standardized approach will take effect starting on January 1, 2017. 

[53] For example, in December 2008, S&P introduced government support 
as a factor that could affect a financial institution's ratings. 
Moody's had assigned uplift in ratings for some bank holding companies 
before the financial crisis and during the crisis increased the amount 
of uplift for these firms and began to assign uplift to several other 
financial institutions. 

[54] As explained earlier in this report's scope and methodology 
section, we selected nonrepresentative samples of investment firms for 
interviews. As a result, the views of investment firm representatives 
we interviewed are not generalizable to the broader universe of 
investors. 

[55] A firm with a lower standalone credit rating may receive a bigger 
increase in its rating from government support than a firm with a 
stronger standalone rating. 

[56] These eight companies were Bank of America Corporation, Bank of 
New York Mellon, Citigroup Inc., Goldman Sachs Group, JPMorgan Chase & 
Co., Morgan Stanley, State Street Corporation, and Wells Fargo & 
Company. While Fitch continued to assign a support rating floor to 
eight bank holding companies, only 3 of these companies had standalone 
credit ratings below the support rating floor and thus received a 
higher credit rating from Fitch as a result of assumed government 
support. These three companies were Bank of America Corporation, 
Citigroup Inc., and Morgan Stanley. Following a March 2014 upgrade to 
Citigroup Inc.'s rating, only Bank of America Corporation and Morgan 
Stanley continued to have a standalone rating below the support rating 
floor. 

[57] In a July 2013 report, S&P noted that regulatory developments may 
lead it to reassess its assumptions of extraordinary government 
support on the holding company ratings of the eight systemically 
important banks, but it was not considering removing such support 
assumptions at the subsidiary level. 

[58] Accordingly, S&P revised its rating outlook on the holding 
company of JPMorgan Chase & Co. to negative from stable, while 
maintaining negative rating outlooks on the holding companies of the 
other seven bank holding companies. S&P noted that it is monitoring 
FDIC's progress in developing a resolution mechanism that aims to 
ensure that market confidence will not erode with the failure of a big 
bank and its resolution through Title II of Dodd-Frank. Furthermore, 
it indicated that it would consider the required level of long-term 
debt at the holding company as another factor in removing government 
support from those ratings. 

[59] In March 2014, Fitch revised its outlook on Bank of America 
Corporation from "stable" to "negative," reflecting its expectation 
that its assumption of weakening sovereign support is likely to result 
in a downgrade. Fitch also upgraded Citigroup Inc.'s standalone 
rating, after which only Bank of America Corporation and Morgan 
Stanley--which Fitch noted was likely to be upgraded--had standalone 
ratings below the support rating floor and thus benefited from an 
uplift. 

[60] While Moody's removed rating uplift at the holding company level, 
it retained uplift in its ratings for the senior and subordinated debt 
of bank subsidiaries of these bank holding companies. Moody's did not 
change its assignment of uplift for bank-level senior debt, explaining 
that it continued to believe that the disorderly failure of one of 
these large bank subsidiaries would create the risk of contagion to 
the broader financial system. In contrast, Moody's reduced but did not 
eliminate uplift for subordinated debt of these banks' subsidiaries. 
Moody's attributed differences in its assessment of uplift for senior 
and subordinated debt to potential differences in the level of 
government support for senior and subordinated creditors. Moody's 
noted that it expects support for bank-level creditors to come 
primarily from the protection provided by holding company creditors, 
who take losses before the bank-level creditors under the SPOE 
framework. However, if imposing losses on holding company creditors 
does not prove sufficient to ensure the viability of the bank 
subsidiary, Moody's believes it is unlikely that bank-level 
subordinated creditors would receive direct government support. In 
contrast, in such a scenario, Moody's believes senior creditors might 
still be supported by the government. 

[61] While the Dodd-Frank Act contains provisions that prohibit 
rescues of individual failing institutions, a few investment firm 
representatives believed that, under certain scenarios, the federal 
government might intervene to provide capital injections to distressed 
bank holding companies, which would require an act of Congress. These 
representatives said they did not know what form a government 
intervention would take, but indicated that elected officials could 
face pressure to intervene as the U.S. Congress did in October 2008 
when it passed legislation authorizing TARP. One representative of a 
large asset management firm said that while the political cost to 
elected officials of rescuing large financial institutions would be 
high, the costs to the economy of not intervening could also be high. 

[62] In his February 6, 2014, testimony to the Senate Banking 
Committee, Federal Reserve Governor Daniel Tarullo said that in the 
event that the equity of a financial firm is wiped out, successful 
resolution without taxpayer assistance would be most effectively 
accomplished if a firm has sufficient long-term, unsecured debt to 
absorb additional losses and to recapitalize the business transferred 
to a bridge operating company. In a public comment to FDIC on OLA's 
single-point-of-entry framework, former FDIC Chair Sheila Bair 
identified the long-term debt requirement as a key component to ensure 
that a firm in OLA has sufficient loss-absorbing capacity. 

[63] For example, an investment firm may employ a passive investment 
strategy by managing the selection and allocation of investments in a 
particular fund with the goal of matching the returns of a benchmark 
index. In contrast, the firm may actively manage other funds by 
choosing investments with the goal of generating returns that 
outperform a benchmark index. 

[64] In this report, we do not attempt to quantify the extent to which 
such higher costs for the largest firms could offset benefits they 
receive as a result of expectations of government support. For 
example, we did not attempt to determine the differential impacts of 
various Dodd-Frank Act provisions on bank holding companies of 
different sizes. Implementation of some Dodd-Frank Act provisions 
specifically targets only SIFIs, while other provisions affect both 
SIFIs and non-SIFIs. Representatives of community banks and other non-
SIFIs have noted that while some Dodd-Frank Act provisions--such as 
the Volcker rule--were intended to target activities at the largest 
bank holding companies, smaller banks can still face burdens 
associated with ensuring they comply with these rules. Federal 
financial regulators have acknowledged the importance of minimizing 
regulatory burdens for financial institutions and particularly for 
smaller banks, whose fixed costs arising from regulatory compliance 
must be spread over a smaller base of revenues. 

[65] Pub. L. No. 111-203, § 331(b), 124 Stat. 1376, 1538 (2010). 

[66] § 318(c), 124 Stat. at 1527. In addition to these fee assessments 
related to Federal Reserve Board supervision, in April 2014 OCC 
proposed increasing the assessments it collects for national banks and 
federal savings associations with more than $40 billion in total 
assets. Assessment of Fees, 79 Fed. Reg. 23297 (Apr. 28, 2014). Under 
the proposal, the assessment increases would range from 0.32 percent 
to 14 percent, depending on the total assets of the institution 
reflected in its June 30, 2014, call report. Id. The proposed rule 
presented estimates of the size of the increase for institutions of 
various sizes. For example, a national bank with assets of $2 trillion 
would see its semiannual assessment increase by an estimated $9.3 
million, or 14 percent. 

[67] National banking regulators classify capital as either Tier 1--
currently the highest-quality form of capital and includes common 
equity--or Tier 2, which is weaker in absorbing losses. Tier 1, or 
core, capital consists primarily of common equity. Tier 2 is 
supplementary capital and includes limited amounts of subordinated 
debt, loan loss reserves, and certain other instruments. 

[68] On March 26, 2014, the Federal Reserve Board announced that it 
approved the capital plans of 25 bank holding companies participating 
in its Comprehensive Capital Analysis and Review (CCAR) program. It 
objected to the capital plans of 5 bank holding companies, including 
Citigroup Inc., one of the U.S. GSIBs. On April 28, 2014, the Federal 
Reserve Board announced that it was requiring Bank of America 
Corporation to resubmit its capital plan and to suspend planned 
increases in dividend distributions. The decision related to the 
disclosure by Bank of America that it incorrectly reported data used 
in the calculation of regulatory capital ratios and submitted as 
inputs for the most recent stress tests conducted by the Federal 
Reserve. The Federal Reserve Board can require a banking organization 
that is part of the annual CCAR program to resubmit its capital plan 
at any time if there has been or likely will be a material change in a 
firm's capital position. 12 C.F.R. § 225.8(d)(4)(i)(C)(2). 

[69] Regulatory Capital Rules: Regulatory Capital, Enhanced 
Supplementary Leverage Ratio Standards for Certain Bank Holding 
Companies and Their Subsidiary Insured Depository Institutions, 79 
Fed. Reg. 24528 (May 1, 2014). The final rule applies to U.S. bank 
holding companies with more than $700 billion in consolidated total 
assets or more than $10 trillion in assets under custody (covered 
BHCs) and their insured depository institution (IDI) subsidiaries. 
Covered BHCs must maintain a leverage buffer greater than 2 percentage 
points above the minimum supplementary leverage ratio requirement of 3 
percent, for a total of more than 5 percent, to avoid restrictions on 
capital distributions and discretionary bonus payments. IDI 
subsidiaries of covered BHCs must maintain at least a 6 percent 
supplementary leverage ratio to be considered "well capitalized" under 
the agencies' prompt corrective action framework. The final rule, 
which has an effective date of January 1, 2018, currently applies to 
eight large U.S. banking organizations that meet the size thresholds 
and their IDI subsidiaries. 

[70] As noted earlier in this report, the Federal Reserve Board has 
said its proposal will be based on the G-SIB capital surcharge 
framework developed by the Basel Committee, under which the size of 
the surcharge for an individual G-SIB would vary from 1 percent to 2.5 
percent, depending on the firm's systemic importance. 

[71] In theory, increasing the required proportion of equity funding 
relative to debt funding should not affect a firm's overall cost of 
funding as it reduces the risk that the firm will fail, thereby 
reducing the returns demanded by both equity and debt holders. 
However, certain government policies make equity financing (such as 
through issuing stock to investors) more expensive for financial 
institutions than debt financing. For example, interest on debt is tax 
deductible, while dividends on equity securities are not. In addition, 
bank deposits benefit from federal guarantees and the interest rates a 
bank pays on its insured deposits may not fall as capital levels and 
the perceived safety of the firm increases. 

[72] See [hyperlink, http://www.gao.gov/products/GAO-13-180]. 

[73] Resolution of Systemically Important Financial Institutions: The 
Single Point of Entry Strategy, 78 Fed. Reg. 243 (Dec. 18, 2013); 
Resolution of a Systemically Important Financial Institution: The 
Single Point of Entry Strategy, 79 Fed. Reg. 9899 (Feb. 21, 2014). 

[74] A basis point is one one-hundredth of a percentage point. Viral 
V. Acharya, Deniz Anginer and A. Joseph Warburton, "The End of Market 
Discipline? Investor Expectations of Implicit State Guarantees," 
Social Science Research Network Working Paper (December 2013). 

[75] International Monetary Fund, "How Big is the Implicit Subsidy for 
Banks Considered Too Important to Fail?" Global Financial Stability 
Report, Ch. 3 (Washington, D.C.: 2014). 

[76] Results of these models will also be sensitive to the typical 
assumption that the impact of variables on funding costs is linear--
that is, changes in credit risk and liquidity have a constant and 
proportional impact on funding costs. 

[77] International Monetary Fund, "How Big is the Implicit Subsidy for 
Banks Considered Too Important to Fail?" Global Financial Stability 
Report, Ch. 3 (Washington, D.C.: 2014), Zoe Tsesmelidakis and Robert 
C. Merton, "The value of implicit guarantees," Social Science Research 
Network working paper, September 1, 2012, and Zan Li, Shisheng Qu and 
Jing Zhang, Quantifying the Value of Implicit Government Guarantees 
for Large Financial Institutions (Moody's Analytics, 2011). 

[78] This approach can be referred to as a "Merton model" based on 
Robert C. Merton, "On the pricing of corporate debt: the risk 
structure of interest rates," Journal of Finance vol. 29, 1974. 
Holders of equity and debt in a bank both face a risk of loss. Equity 
holders face a higher risk because they are the first to take losses 
in a failure (i.e., debt holders have higher priority). However, the 
risk that a debt holder will take a loss is likely to be proportional 
to--but smaller than--the risk facing an equity holder. As a result, 
one can estimate the risk of default on a debt based in part on the 
volatility of the stock price of the bank. 

[79] Another weakness of this model is that the market pricing of 
credit default swaps may not be reliable during a crisis. 

[80] See Sreedhar T. Bharath and Tyler Shumway, "Forecasting default 
with the Merton distance to default model," Review of Financial 
Studies, vol. 21, no. 3, May 2008. 

[81] International Monetary Fund, "How Big is the Implicit Subsidy for 
Banks Considered Too Important to Fail?" Global Financial Stability 
Report, Ch. 3 (Washington, D.C.: 2014), and Kenichi Ueda and Beatrice 
Weder di Mauro, "Quantifying structural subsidy values for 
systemically important financial institutions," Journal of Banking & 
Finance vol. 37, no. 10, October 2013. 

[82] In assessing the creditworthiness of bank holding companies, all 
three major rating agencies discuss the extent to which a rating is 
influenced by potential government support, though the specific 
approach differs by rating agency. By comparing a bank holding 
company's credit rating--with and without expected government support-
-to the average historical funding cost for corporations at those 
ratings, one can estimate the benefit of the amount of government 
support associated with the rating uplift. 

[83] Our models allow the size of a bank holding company to influence 
its bond funding costs directly and also indirectly through the 
interaction between size and credit risk. As a result, no single 
parameter is sufficient to describe the relationship between bond 
funding costs and size. To summarize the overall relationship between 
bond funding costs and size reflected in each specification, we 
calculated bond funding costs for bank holding companies of different 
sizes and credit risk levels using our estimates of the parameters for 
each specification for each year. See appendix I for more details on 
these calculations. 

[84] We also compared funding costs for bank holding companies with 
$50 billion, $100 billion, $250 billion, and $500 billion in assets to 
bank holding companies with $10 billion in assets. See appendix I. 

[85] Many of the estimates that were statistically significant at the 
10 percent level were also statistically significant at the 5 percent 
or 1 percent level. See table 5 in appendix I. 

[86] Although higher, credit risk in 2008 was not outside the range of 
credit risk in 2013 in every dimension. Specifically, the average 
values of the credit risk variables for 2008 were less than the 
maximum values of the credit risk variables for 2013, with the 
exceptions of the variables measuring equity price volatility, option 
implied volatility, equity return volatility, and excess equity return 
volatility. 

[87] As discussed earlier in this report, many investment firm 
representatives with whom we spoke credited enhanced regulatory 
standards with improving the safety and soundness of the largest bank 
holding companies and reducing the likelihood that they would 
experience distress that could result in failure or government support. 

[88] To see this, compare the 2008 estimates in figure 1 to the 
"financial crisis level" estimates for 2013 in figure 3. Both sets of 
estimates are derived assuming that the level of credit risk is equal 
to the average for 2008. 

[89] Secondary markets are markets where investors purchase securities 
or assets from other investors, rather than from issuing companies 
themselves. 

[90] A credit default swap is a credit derivative in which two parties 
enter into an agreement whereby one party pays the other a fixed 
periodic coupon for the specified life of the agreement and the other 
party makes no payments unless a credit event relating to a 
predetermined reference asset occurs. If such an event occurs, the 
party will then make a payment to the first party and the swap will 
terminate. The size of the payment is usually linked to the decline in 
the reference asset's market value following the determination of the 
occurrence of a credit event. 

[91] A plain vanilla bond, also known as a straight bond, is a bond 
that pays interest at regular intervals and at maturity pays back the 
principal that was originally invested. These bonds are debt 
instruments because they are essentially loaning money (creating debt) 
to an entity, which promises to pay the interest on the debt and at 
maturity pay back the original loan. 

[92] Our sample includes only bank holding companies, so it excludes 
companies like Goldman Sachs and Morgan Stanley before they became 
bank holding companies. Similarly, our sample excludes other financial 
companies that were never bank holding companies. 

[93] Executed trades are trades that have been completed. Executable 
quotes are quotes provided by bond dealers at which they are willing 
to trade. Indicative quotes are quotes provided by bond dealers to 
give an indication of the price at which a bond might trade, but at 
which they are not obligated to trade. Composites are aggregates of 
quotes from multiple sources that are suggestive of the price at which 
a bond might trade. 

[94] The correlation between the two measures of bond yield spreads 
was 0.96. 

[95] A bank holding company's total assets may change from one quarter 
to the next. Similarly, a bank holding company may have less than $50 
billion in assets in some quarters and more than $50 billion in assets 
in others. An alternative measure of a bank holding company's size or 
systemic importance is whether or not it participated in the Federal 
Reserve's Comprehensive Capital Analysis and Review (CCAR). However, 
the measure of size based on CCAR participation is highly correlated 
with the measure of size based on whether or not a bank holding 
company had $50 billion or more in assets, so these two measures of 
size are not substantively different. 

[96] Credit risk increases with higher values of our asset quality, 
maturity mismatch, and volatility measures and with lower values of 
our capital adequacy and earnings measures. 

[97] Several components of volatile liabilities, such as federal funds 
purchased, repurchase agreements, and demand deposits, have relatively 
short maturities and may not be rolled over by the creditor when they 
mature. Maturity mismatch measures the bank holding company's capacity 
to cover these liabilities if the creditor chooses to not roll them 
over. 

[98] The bid-ask spread is the difference between the best buying 
price and the best selling price. We expressed the bid-ask spread as a 
percentage of the midpoint between the bid price and the ask price. 

[99] Some studies suggest that there may be economies of scale or 
scope in banking. For example, see Joseph P. Hughes and Loretta J. 
Mester, "Who Said Large Banks Don't Experience Scale Economies? 
Evidence From a Risk-Return-Driven Cost Function," Journal of 
Financial Intermediation 22 (2013), pp. 559-585. We include 
noninterest expense as a percentage of total assets to control for a 
channel through which economies of scale may affect bond funding 
costs. Other channels through which economies of scale or scope may 
affect bond funding costs are profitability, which is captured in our 
earnings variable, and riskiness, which is captured in our volatility 
variable. 

[100] For predictions at average levels of credit risk, we set 
variables equal to the unweighted average for the sample used to 
estimate the regression for each year. For predictions at low levels 
of credit risk, we set variables measuring asset quality, maturity 
mismatch, and volatility equal to their 25th percentiles for each year 
and we set variables measuring capital adequacy and earnings equal to 
their 75th percentiles for each year. For predictions as high levels 
of credit risk, we set variables measuring asset quality, maturity 
mismatch, and volatility equal to their 75th percentiles for each year 
and we set variables measuring capital adequacy and earnings equal to 
their 25th percentiles for each year. 

[End of section] 

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 website [hyperlink, http://www.gao.gov]. Each 
weekday afternoon, GAO posts on its website newly released reports, 
testimony, and correspondence. To have GAO e-mail you a list of newly 
posted products, go to [hyperlink, http://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 
website, [hyperlink, 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. 

Connect with GAO: 

Connect with GAO on facebook, flickr, twitter, and YouTube.
Subscribe to our RSS Feeds or E mail Updates. Listen to our Podcasts.
Visit GAO on the web at [hyperlink, http://www.gao.gov]. 

To Report Fraud, Waste, and Abuse in Federal Programs: 

Contact: 
Website: [hyperlink, http://www.gao.gov/fraudnet/fraudnet.htm]; 
E-mail: fraudnet@gao.gov; 
Automated answering system: (800) 424-5454 or (202) 512-7470. 

Congressional Relations: 

Katherine Siggerud, Managing Director, siggerudk@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. 

[End of document]