This is the accessible text file for GAO report number GAO-10-782
entitled 'Securities And Exchange Commission: Action Needed to Improve
Rating Agency Registration Program and Performance-Related
Disclosures' which was released on September 22, 2010.
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.
Report to Congressional Committees:
United States Government Accountability Office:
GAO:
September 2010:
Securities And Exchange Commission:
Action Needed to Improve Rating Agency Registration Program and
Performance-Related Disclosures:
GAO-10-782:
GAO Highlights:
Highlights of GAO-10-782, a report to congressional committees.
Why GAO Did This Study:
In 2006, Congress passed the Credit Rating Agency Reform Act (Act),
which intended to improve credit ratings by fostering accountability,
transparency, and competition. The Act established Securities and
Exchange Commission (SEC) oversight over Nationally Recognized
Statistical Rating Organizations (NRSRO), which are credit rating
agencies that are registered with SEC. The Act requires GAO to review
the implementation of the Act. This report (1) discusses the Act’s
implementation; (2) evaluates NRSROs’ performance-related disclosures;
(3) evaluates removing NRSRO references from certain SEC rules; (4)
evaluates the impact of the Act on competition; and (5) provides a
framework for evaluating alternative models for compensating NRSROs.
To address the mandate, GAO reviewed SEC rules, examination guidance,
completed examinations, and staff memoranda; analyzed required NRSRO
disclosures and market share data; and interviewed SEC and NRSRO
officials and market participants.
What GAO Found:
SEC’s implementation of the Act involved developing an NRSRO
registration program and an examination program. As currently
implemented and staffed, both programs require further attention.
* The process for reviewing NRSRO applications limits SEC staff’s
ability to fully ensure that applicants meet the Act’s requirements.
While SEC had registered 10 of 11 credit rating agency applicants as
of July 2010, some staff memoranda to the Commission summarizing their
review of applications described concerns that were not addressed
prior to registration. According to staff, the 90-day time frame for
SEC action on an application and the lack of an express authority to
examine the applicants prior to registration prevented the concerns
from being addressed prior to approval. Unlike other registration
application programs that have built in greater authority and
flexibility for their staff to clarify outstanding questions regarding
applications before approval, the NRSRO registration program requires
SEC to act within 90 days of receiving the application. As a result,
staff recommended granting registration with ongoing concerns about
NRSROs meeting the Act’s requirements.
* With its current level of staffing for NRSRO examinations, SEC’s
Office of Compliance Inspections and Examinations (OCIE) would likely
not have been able to meet its routine examination schedule of
examining the three largest NRSROs every 2 years and others every 3
years. OCIE has requested additional resources to fully staff the
NRSRO examination program. While the Dodd-Frank Wall Street Reform and
Consumer Protection Act (Dodd-Frank Act) requires SEC to establish an
Office of Credit Ratings to conduct annual examinations of each NRSRO
and staff the office sufficiently to carry out these examinations, SEC
may face challenges in meeting the required examination timetable and
providing quality supervision over NRSROs unless it develops a plan
that clearly identifies staffing needs, such as requisite skills and
training.
While SEC has increased the amount of performance-related data NRSROs
are required to disclose, the usefulness of the data is limited.
First, SEC requires NRSROs to disclose certain performance statistics,
increasing the amount of performance information available for some
NRSROs. However, because SEC does not specify how NRSROs should
calculate these statistics, NRSROs use varied methodologies, limiting
their comparability. Second, SEC issued two rules requiring NRSROs to
make certain ratings history data publicly available. However, the
data sets do not contain enough information to construct comparable
performance statistics and are not representative of the population of
the credit ratings at each NRSRO. Without better disclosures, the
information being provided will not serve its intended purpose of
increasing transparency.
In July 2008, SEC proposed amendments that would have removed
references to NRSRO ratings from several rules. While SEC removed
references from six rules and two forms, it retained the use of the
ratings or delayed further action on two rules. These rules govern
money market fund investments and the amount of capital that broker-
dealers must hold, and use NRSRO references as risk-limiting measures.
GAO reviewed SEC’s proposals to remove NRSRO references from these two
rules and identified concerns with how SEC examiners would have
evaluated compliance with the proposed alternative credit standards
and whether it had staff with the requisite skills. Going forward, the
Dodd-Frank Act requires SEC to remove NRSRO ratings from its rules.
SEC’s previous experience with proposals to remove credit rating
references highlights the importance of developing a plan to help
ensure that (1) any adopted alternative standards of creditworthiness
for a particular rule facilitate its purpose and (2) that examiners
have the requisite skills to determine that the adopted standards have
been applied. Without such a plan, SEC may develop alternative
standards of creditworthiness that are not effective in supporting the
purpose of a particular rule.
Since the implementation of the Act, the number of NRSROs has
increased from 7 to 10; however, industry concentration as measured by
NRSRO revenues, the number of entities rated, and the dollar volume of
new asset-backed debt rated remains high. Several factors likely have
contributed to the continued high concentration among NRSROs. First,
relatively little time has passed since SEC implemented the NRSRO
registration program and NRSRO rulemaking. Second, the three new
NRSROs have not had much time to build market share. Finally, the
recent financial crisis occurred soon after the Act’s implementation,
substantially slowing certain sectors of the securitization markets.
Moreover, there are barriers to entering the rating industry and to
becoming an NRSRO. For example, establishing a reputation as a
credible provider of credit ratings can take years. The reference to
specific NRSROs in private contracts and investment guidelines also
acts as a barrier.
As part of an April 2009 roundtable held to examine oversight of
credit rating agencies, SEC requested perspectives from users of
ratings and others on whether it should consider additional rules to
better align the raters’ interest with those who rely on those
ratings, and specifically, whether one business model represented a
better way of managing conflicts of interest than another. GAO
identified five unique alternative models for compensating NRSROs that
have been proposed by roundtable participants and others, although
they vary in the amount of detail available. To assist Congress and
others in assessing these proposals, GAO created an evaluative
framework of seven factors that any compensation model should address
to be effective. By applying these factors, users of the framework can
identify the potential benefits of the model consistent with
policymakers’ goals as well as any tradeoff.
Table: Framework for Evaluating Alternative Models for Compensating
NRSROs:
Factor: Independence;
Description: The ability for the compensation model to mitigate
conflicts of interest inherent between the entity paying for the
rating and the NRSRO. Key questions include: What potential conflicts
of interest exist in the alternative compensation model and what
controls, if any, would need to be implemented to mitigate these
conflicts?
Factor: Accountability;
Description: The ability of the compensation model to promote NRSROs’
responsibility for the accuracy and timeliness of their ratings. Key
questions include: How does the compensation model create economic
incentives for NRSROs to produce quality ratings over the bond’s life?
How is NRSRO performance evaluated and by whom?
Factor: Competition;
Description: The extent to which the compensation model creates an
environment in which NRSROs compete for customers by producing higher-
quality ratings at competitive prices. Key questions include: To what
extent does the compensation model encourage competition around the
quality of ratings, ratings fees, and product innovation? To what
extent does it allow for flexibility in the differing sizes,
resources, and specialties of NRSROs?
Factor: Transparency;
Description: The accessibility, usability, and clarity of the
compensation model and the dissemination of information on the model
to market participants. Key questions include: How transparent are the
model’s processes and procedures for determining ratings fees and
compensating NRSROs? How would NRSROs obtain ratings business?
Factor: Feasibility;
Description: The simplicity and ease with which the compensation model
can be implemented in the securities market. Key questions include:
What are the costs to implement the compensation model and who would
fund them? Who would administer the compensation model? What, if any,
infrastructure would be needed to implement it?
Factor: Market acceptance and choice;
Description: The willingness of the securities market to accept the
compensation model, the ratings produced under that model, and any new
market players established by the compensation model. Key questions
include: What role do market participants have in selecting NRSROs to
produce ratings, assessing the quality of ratings, and determining
NRSRO compensation?
Factor: Oversight;
Description: The evaluation of the model to ensure it works as
intended. Key questions include: Does the model provide for an
independent internal control function? What external oversight does
the compensation model provide to ensure it is working as intended?
Source: GAO.
[End of table]
What GAO Recommends:
time frames and authorities it needs to review NRSRO applications,
develop a plan to help ensure the NRSRO examination program is
sufficiently staffed, improve NRSROs’ performance-related disclosure
requirements, and develop a plan to approach the removal of NRSRO
references from its rules. SEC generally agreed with these
recommendations.
View [hyperlink, http://www.gao.gov/products/GAO-10-782] or key
components. For more information, contact Orice Williams Brown at
(202) 512-8678 or williamso@gao.gov.
[End of section]
Contents:
Letter:
Background:
NRSRO Application Review Process Limits SEC Staff's Ability to Ensure
That Applicants Meet the Act's Requirements and NRSRO Examination
Program Faces Staffing Challenges:
SEC Has Increased the Amount of Performance-related Data NRSROs Are
Required to Disclose, but These Data Have Limited Usefulness:
As SEC Works to Remove NRSRO References from SEC Rules, It Will Need
To Ensure It Has the Staff with the Requisite Skills to Evaluate
Compliance with Any Alternative Creditworthiness Standard:
The Number of NRSROs Has Increased since the Act Was Implemented but
Industry Concentration Remains High:
Models Proposing Alternative Means of Compensating NRSROs Intend to
Address Conflicts of Interests in the Issuer-Pays Model:
Conclusions:
Recommendations for Executive Action:
Agency Comments and Our Evaluation:
Appendix I: Objectives, Scope, and Methodology:
Appendix II: Other Registration Processes Provide Greater Flexibility
to the Regulators:
Appendix III: Comments from the Securities and Exchange Commission:
Appendix IV: GAO Contact and Staff Acknowledgments:
Tables:
Table 1: SEC Rules Pertaining to NRSROs, 2007 and 2009:
Table 2: Hypothetical 1-Year Transition Matrix and 1-Year Default
Rates Relative to Beginning-of-Year Ratings for 2009 Cohorts:
Table 3: Methods Used by NRSROs to Calculate SEC-required 1-, 3-, and
10-year Transition Rates:
Table 4: Methods Used by NRSROs to Calculate SEC-required 1-, 3-, and
10-year Default Rates:
Table 5: NRSROs Registered by Asset Class, 2010:
Table 6: HHI for NRSROs Based on Total Revenues, 2006-2009:
Table 7: HHI for NRSROs Based on Number of Issuers Rated, 2006-2009:
Table 8: HHI Based on Dollar Value of Newly Issued U.S.-ABS, January
2004-June 2010:
Figures:
Figure 1: Years the Current NRSROs Have Produced Credit Ratings and
Have Been Recognized as NRSROs:
Figure 2: NRSRO U.S. Annual Market Coverage by ABS, Dollar Volume,
2004-June 2010:
Figure 3: NRSRO U.S. Annual Market Coverage by Traditional ABS, Dollar
Volume, 2004-June 2010:
Figure 4: NRSRO U.S. Annual Market Coverage by Prime RMBS, Dollar
Volume, 2004-June 2010:
Figure 5: NRSRO U.S. Annual Market Coverage by Nonprime RMBS, Dollar
Volume, 2004-June 2010:
Figure 6: NRSRO U.S. Annual Market Coverage by CMBS, Dollar Volume,
2004-June 2010:
Figure 7: NRSRO U.S. Annual Market Coverage by CDO, Dollar Volume,
2004-June 2010:
Abbreviations:
ABS: asset-backed securities:
Act: Credit Rating Agency Reform Act:
CMBS: commercial mortgage-backed securities:
CDO: collateralized debt obligations:
Dodd-Frank Act: Dodd-Frank Wall Street Reform and Consumer Protection
Act:
Enforcement: SEC's Division of Enforcement:
Exchange Act: Securities Exchange Act of 1934:
FINRA: Financial Industry Regulatory Authority:
Fitch: Fitch Ratings:
HHI: Herfindahl-Hirschman Index:
Investment Company Act: Investment Company Act of 1940:
Investment Management: SEC's Division of Investment Management:
DOJ: Department of Justice:
Moody's: Moody's Investors Service:
NRSRO: Nationally Recognized Statistical Rating Organization:
OCIE: Office of Compliance Examinations and Inspections:
ORA: SEC's Office of Risk Analysis:
RMBS: residential mortgage-backed securities:
SEC: Securities and Exchange Commission:
Trading and Markets: SEC's Division of Trading and Markets:
[End of section]
United States Government Accountability Office:
Washington, DC 20548:
September 22, 2010:
The Honorable Christopher J. Dodd:
Chairman:
The Honorable Richard C. Shelby:
Ranking Member:
Committee on Banking, Housing, and Urban Affairs:
United States Senate:
The Honorable Barney Frank:
Chairman:
The Honorable Spencer Bachus:
Ranking Member:
Committee on Financial Services:
House of Representatives:
A credit rating is an assessment of the creditworthiness of an obligor
as an entity or with respect to specific securities or money market
instruments.[Footnote 1] In the past few decades, credit ratings have
assumed increased importance in the financial markets, in large part
due to their use in law and regulation. In 1975, the Securities and
Exchange Commission (SEC) first used the term Nationally Recognized
Statistical Rating Organization (NRSRO) to describe those rating
agencies whose ratings could be relied upon to determine capital
charges for different types of debt securities (securities) broker-
dealers held.[Footnote 2] Since then, SEC has used the NRSRO
designation in a number of regulations, and the term has been widely
embedded in numerous federal and state laws and regulations as well as
in investment guidelines and private contracts.
The highly publicized, alleged failures by the three largest NRSROs to
warn investors in a timely manner about the impending bankruptcies of
Enron and other issuers raised concerns in Congress and among others
about the role and performance of NRSROs in the securities market and
SEC's supervision of the industry.[Footnote 3] In response to a
congressional mandate, SEC prepared a report in 2003 discussing its
findings from examinations of these rating agencies, which revealed
concerns related to potential conflicts of interest caused by the
business model they employ--in which the issuers of securities pay the
rating agencies for their ratings (issuer-pays model).[Footnote 4] The
report also discussed the lack of a formal regulatory program to
oversee NRSROs, and SEC efforts over the years to establish one.
Participants in congressional hearings held at the time noted the high
concentration of market share among a small number of large NRSROs and
criticized SEC's no-action letter process (used to recognize NRSROs)
as a barrier to entry to the market for new credit rating agencies,
and thus a hindrance to competition.[Footnote 5]
To address these concerns, Congress passed the Credit Rating Agency
Reform Act (Act) in 2006[Footnote 6]. The Act amended the Securities
Exchange Act of 1934 (Exchange Act) to improve ratings quality for the
protection of investors by fostering accountability, transparency, and
competition in the credit rating industry.[Footnote 7] The Act added
section 15E to the Exchange Act, which establishes SEC oversight over
those credit rating agencies that register as NRSROs. Section 15E also
provides SEC with examination authority and establishes a registration
program for credit rating agencies seeking NRSRO designation, defines
eligibility requirements, prescribes the minimum information
applicants must provide in their application, and establishes a time
frame and parameters for SEC review and approval of applications.
NRSRO applicants and registered NRSROs are also required to disclose
information, including ratings performance, conflicts of interest, and
the procedures used to determine ratings.
More recently, the performance of the three largest NRSROs in rating
subprime residential mortgage-backed securities (RMBS) and related
securities renewed questions about the accuracy of their credit
ratings generally, the integrity of the ratings process, and investor
reliance on NRSRO ratings for investment decisions.[Footnote 8] In
July 2008, SEC made public its report on its examinations of these
three NRSROs, which identified significant issues with their
documentation and disclosure of critical ratings processes and the
management of conflicts of interest related to these products.
[Footnote 9] Partially in response to these findings, SEC issued
additional rules in 2008 and 2009 intended to enhance NRSRO disclosure
to investors, strengthen the integrity of the ratings process, and
more effectively address the potential for conflicts of interest. SEC
also held a roundtable relating to its oversight of credit rating
agencies in April 2009, where roundtable participants offered
proposals to, among other things, reduce conflicts of interest and
increase NRSRO incentives to produce accurate ratings by establishing
alternative means for compensating NRSROs.
Section 7 of the Act requires us to review, by September 2010, the
implementation and impact of the Act and the rules issued under it on
the quality of credit ratings, financial markets, competition in the
credit rating industry, the process for NRSRO registration, and other
matters. This report responds to that mandate. It also responds to
your interest in identifying and assessing alternative models for
compensating NRSROs and assessing the potential impact of removing
NRSRO references from SEC rules.
Specifically, this report (1) discusses the implementation of the Act,
focusing on SEC rulemaking and SEC's implementation of the NRSRO
registration and examination programs; (2) evaluates the performance-
related NRSRO disclosures that the Act and SEC rules require; (3)
evaluates the potential regulatory impact of removing NRSRO references
from certain SEC rules; (4) evaluates the impact of the Act on
competition among NRSROs; and (5) provides an overview of proposed
alternative models for compensating NRSROs and an evaluative framework
for assessing the models.
To address the first objective, we reviewed the rules SEC adopted to
implement the Act, SEC's July 2008 public report discussing
examination findings on selected NRSROs, SEC's Division of Trading and
Markets (Trading and Markets) internal memoranda to the commissioners
documenting its review of NRSRO applications, and an internal
memorandum on NRSRO monitoring. We also reviewed the Office of
Compliance Examinations and Inspections' (OCIE) guidance for
conducting an NRSRO examination and reviewed completed examinations.
We discussed the application review process with Trading and Markets
staff, and conducted interviews with staff from SEC's Division of
Investment Management (Investment Management) and the Financial
Industry Regulatory Authority (FINRA) about their respective
registration programs and with OCIE staff regarding NRSRO
examinations. For the second objective, we analyzed SEC rules
requiring NRSRO disclosure of performance statistics and ratings
history samples. We analyzed and compared the NRSROs' 2009 disclosures
of performance statistics, focusing on the corporate and structured
finance asset classes, and we reviewed voluntary disclosures of
additional performance statistics by several NRSROs. We also assessed
NRSRO ratings history data disclosed by the seven issuer-pays NRSROs
pursuant to SEC's rule to determine their reliability and usability
for generating comparative performance statistics. We identified a
number of issues that led us to determine that the data were not in a
format that allowed us to generate comparative performance statistics.
For the third objective, we obtained and reviewed SEC's proposed rules
to remove references to NRSRO ratings, focusing on proposed amendments
to rule 2a-7 under the Investment Company Act of 1940 (Investment
Company Act) and Exchange Act rule 15c3-1, as well as the comment
letters submitted to SEC on the proposals.[Footnote 10] We also
reviewed OCIE examinations of money market funds from FY 2003-2009
that reviewed rule 2a-7 compliance. We conducted interviews with OCIE
staff, Trading and Markets staff, Investment Management staff, and
market participants and observers.
For the fourth objective, we reviewed proposed and final SEC rules
intended to promote competition among NRSROs, as well as the comment
letters SEC received in response to those rules. We reviewed SEC's
2008 and 2009 mandated annual reports on NRSROs, including SEC's
studies on competition in the credit rating industry. We used the
Herfindahl-Hirschman Index (HHI)--a measure of industry concentration
that reflects both the number of firms in the industry and each firm's
market share--to track industry concentration over time.[Footnote 11]
We calculated the HHI using three alternative variables: (1) NRSRO
revenues, (2) the number of organizations or entities rated by the
NRSROs that issue debt securities, and (3) for asset-backed securities
(ABS), the dollar amount of new U.S.-issued debt rated. We obtained
the revenue data from the NRSROs' Form NRSRO filings, the registration
form SEC requires credit rating agencies to submit when applying for
NRSRO registration and then annually after registration. We
interviewed staff from Trading and Markets to determine the steps they
took to ensure the data represented the firms' total revenues. We
obtained data on the number of rated organizations or entities that
issue debt securities from the NRSROs. To ensure consistency among the
data, we specified how the NRSROs were to count rated organizations
and entities and classify them. We also examined trends within the
data. We obtained data on the dollar value of rated new U.S.-asset-
backed securitization from an industry newsletter that tracks these
issuances. We obtained information from this firm on the processes and
procedures it used to collect and manage the data. We also used these
data to generate a series of descriptive graphs showing the NRSROs'
market coverage, that is, the percentage of new U.S. asset-backed
issuance that each rated. For all of our data sources, we determined
that the data were reliable enough for our purposes, which were to
show the relative concentration of NRSROs in the industry. We also
reviewed academic studies on competition in the industry and obtained
the views of SEC's Office of Risk Analysis (ORA), the NRSROs, credit
rating agencies that are not registered NRSROs, institutional
investors, issuers, and industry experts on the impact of the Act on
competition.
For the fifth objective, we identified proposals on alternative models
for compensating NRSROs by reviewing white papers submitted to the SEC
roundtable on credit rating agency oversight, academic and white
papers, and interviewed the authors of the proposed models. We
obtained the views of Trading and Markets, ORA, NRSROs, and credit
rating agencies that are not registered as NRSROs, institutional
investors, issuers, and academic and industry experts. To develop the
framework for evaluating the models, we reviewed prior GAO reports and
academic and market research to identify appropriate factors for
inclusion. We then convened a panel of GAO experts (financial markets
specialists, economists, an attorney and a social scientist) to review
the framework and incorporated their comments. Finally, we solicited
comments from NRSROs, proposal authors, industry experts, and trade
associations and incorporated them as appropriate. For all relevant
objectives, we also reviewed the recently passed Dodd-Frank Wall
Street Reform and Consumer Protection Act (Dodd-Frank Act) to
understand additional changes to SEC's oversight of NRSROs.[Footnote
12]
We conducted this performance audit from May 2009 to September 2010 in
accordance with generally accepted government auditing standards.
Those standards require that we plan and perform the audit to obtain
sufficient, appropriate evidence to provide a reasonable basis for our
findings and conclusions based on our audit objectives. We believe
that the evidence obtained provides a reasonable basis for our
findings and conclusions based on our audit objectives. Appendix I
provides a more detailed description of our scope and methodology.
Background:
The ratings produced by the NRSROs generally are letter-based symbols
intended to reflect assessments of credit risk for entities issuing
securities in public markets. Typically, credit rating agencies
designate issuers or securities considered investment-grade, or lower
risk, with higher letter ratings, and issuers or securities considered
speculative-grade, or higher risk, with lower letter ratings. For
example, Standard & Poor's and Fitch Ratings (Fitch) designate
investment-grade, long-term debt with ratings of AAA, AA, A, and BBB,
and speculative-grade, long-term debt with ratings of BB, B, CCC, CC,
and C. The rating scale employed by Moody's Investors Service
(Moody's) uses Aaa, Aa, A and Baa for investment-grade, long-term
debt, and Ba, B, Caa, Ca, and C for speculative-grade, long-term debt.
Rating agencies may employ different rating scales for different
regions, sectors, jurisdictions, or types of securities. For example,
the rating scale that a ratings agency uses to assign short-term
obligations may differ from the scale it uses for long-term
obligations.
NRSRO credit ratings are designed to measure the likelihood of default
for an issue or issuer, although some also measure other variables,
such as the expected value of dollar losses given a default. These
assessments reflect a variety of quantitative and qualitative factors
that vary based on sector. The NRSROs describe ratings as intended
only to reflect credit risk, not other valuation factors such as
liquidity or price risk. To determine an appropriate rating, credit
analysts use publicly available information, and market and economic
data, and may obtain nonpublic information from the issuer and engage
in discussions with its senior management. However, not all NRSROs
rely on nonpublic information in producing credit ratings.[Footnote 13]
Issuers seek credit ratings for a number of reasons, such as to
improve the marketability or pricing of their financial obligations,
or to satisfy investors, lenders, or counterparties. In certain
markets, such as the U.S. long-term corporate debt market, a single-
rated debt issue may be priced below an issue with similar ratings
from two agencies, because the absence of a second rating is
interpreted as an issuer's inability to obtain another equivalent
rating. However, in other markets such as the ABS market, a single
rating may be adequate confirmation of asset quality.
Institutional investors, such as mutual funds, pension funds, and
insurance companies, are among the largest owners of debt securities
in the United States and are substantial users of credit ratings.
Retail participation in the debt markets generally takes place
indirectly through these fiduciaries. Institutional investors may use
credit ratings as one of several important inputs to their own
internal credit assessments and investment analysis, or to identify
pricing discrepancies for their trading operations. Broker-dealers
that make recommendations and sell securities to their clients also
use ratings. These firms often act as dealers in markets that place
significant importance on credit ratings. For example, in the over-the-
counter derivatives markets, broker-dealers tend to use credit ratings
(when available) to determine acceptable counterparties, as well as
collateral levels for outstanding credit exposures. Large broker-
dealers also frequently obtain credit ratings as issuers of long-and
short-term debts.
Academic literature suggests that credit ratings affect financial
markets both by providing information to investors and other market
participants and by their use in regulations.[Footnote 14] Several
studies suggest that bond prices, stock returns, and credit-default
swap spreads vary with credit ratings downgrades. Other studies find
that obtaining a credit rating generally increases a firm's access to
capital markets and that firms with credit ratings have different
capital structures than firms without them as a result. Furthermore,
some studies suggest that firms adjust their capital structure to
achieve a particular credit rating. One explanation for these
relationships is that rating agencies have access to private
information about the issuers and issues they rate, and the ratings
they assign incorporate this information. Thus, ratings are a
mechanism for communicating this otherwise unavailable information to
market participants. Alternatively, at least in market segments with
rating-based regulations, investors' willingness and ability to
purchase bonds with credit ratings above a regulatory threshold could
be greater than their willingness and ability to purchase bonds with
ratings below the threshold. Thus, firms with credit ratings above the
regulatory threshold have lower costs of capital than those with
credit ratings below the threshold.
NRSROs today operate primarily under one of two compensation models:
issuer-pays or subscriber-pays. Under the issuer-pays model, issuers
pay the NRSRO for a rating. These ratings are generally free to the
public, although users may have to purchase access to in-depth reports
explaining the basis for the rating. Under the subscriber-pays model,
users pay a subscription to the NRSRO for access to its ratings.
Trading and Markets administers and executes the agency's programs
relating to NRSRO oversight, which includes administration of the
NRSRO registration program and rulemaking. OCIE administers SEC's
nationwide examination and inspection program. Within OCIE, the NRSRO
examination team within the Office of Market Oversight conducts NRSRO
examinations. The purpose of an NRSRO examination is to promote
compliance with applicable laws and rules, identify any violations of
such laws and rules, and ensure remedial action. Examinations also
serve to inform SEC and SEC staff of NRSROs' compliance with their
regulatory obligations and noteworthy industry developments. If OCIE
discovers potential violations of federal securities laws or rules
during an NRSRO examination, it may refer the case to Enforcement,
which is responsible for further investigating these potential
violations; recommending SEC action when appropriate, either in a
federal court or before an administrative law judge.
NRSRO Application Review Process Limits SEC Staff's Ability to Ensure
That Applicants Meet the Act's Requirements and NRSRO Examination
Program Faces Staffing Challenges:
SEC's implementation of the Act involved developing an NRSRO
application review process and an examination program. As currently
implemented and staffed, both programs require further attention.
First, in June 2007, SEC adopted final rules that established a
voluntary registration program for recognizing credit rating agencies
as NRSROs.[Footnote 15] Over the past 3 years, SEC has registered 10
credit rating agencies as NRSROs, instituted proceedings to determine
whether to deny registration to one applicant, and has begun
examinations. However, as implemented, the registration process
potentially limits the staff's ability to ensure that applicants meet
the Act's requirements and may create a situation in which ratings
from an NRSRO that may not meet the Act's requirements are used by
investors and for regulatory purposes. Second, although SEC has
established an OCIE branch dedicated to the examination of NRSROs and
hired individuals with experience in credit rating analysis and
structured finance to fill these positions, OCIE has not completed
timely examinations of the NRSROs and has expressed concerns about its
ability to meet its planned NRSRO routine examination schedule of
examining the three largest NRSROs every 2 years and the other NRSROs
every 3 years. While SEC requested additional resources that it
anticipated using to fully staff this oversight function, it will
likely need to revisit those requests due to the passage of the Dodd-
Frank Act, which among other things, requires that each NRSRO be
examined every year and that SEC establish an Office of Credit Ratings
to carry out these examinations. Formalizing a plan to assess not only
the number of staff it needs for this office but also the skills
required of this staff would help SEC be strategically positioned to
implement the Dodd-Frank Act requirements. SEC may face challenges in
meeting the required examination timetable and providing quality
oversight over NRSROs unless it develops a plan that ensures SEC has
sufficient staff that have the appropriate qualifications and are
appropriately trained.
SEC Established a Formal Registration and Oversight Program for NRSROs
and Continues Rulemaking under the Act:
SEC adopted final rules for a formal regulatory and oversight program
for NRSROs in June 2007.[Footnote 16] The rules establish a voluntary
registration program for those credit rating agencies that seek to be
recognized as NRSROs and require that NRSROs make and retain certain
records, furnish financial reports to SEC, and establish procedures to
address uses of material nonpublic information. The rules also require
the disclosure of certain performance measures and ratings
methodologies, prohibit certain conflicts of interest and require the
management of other conflicts of interest, and prohibit specified
coercive and unfair practices by NRSROs. SEC amended several of these
rules in February and December 2009 with the goal of further
increasing transparency of NRSRO rating methodologies, strengthening
the disclosures of ratings performance, prohibiting NRSROs from
engaging in certain practices, and enhancing NRSRO record keeping.
[Footnote 17] These amendments were designed in part to address
concerns that SEC staff identified in its July 2008 report about the
integrity of the process by which the three largest NRSROs rated
structured finance products.[Footnote 18] Table 1 summarizes the rules
and amendments to those rules adopted by SEC under the Act.
Table 1: SEC Rules Pertaining to NRSROs, 2007 and 2009:
Rule 17g-1:
Prescribes how an NRSRO must apply to be registered with SEC, keep its
registration up-to-date, and comply with the statutory requirement to
furnish SEC with an annual certification. Specifically, all of these
actions must be accomplished by furnishing SEC with information on a
Form NRSRO. As part of registration, NRSROs must disclose certain
performance statistics and general descriptions of their ratings
processes and methodologies.
Rule 17g-2:
Requires an NRSRO to make and retain certain types of business records
and disclose certain ratings history data.
Rule 17g-3:
Requires an NRSRO to furnish SEC with four, or in some cases five,
financial reports annually. The first report requires the submission
of audited financial statements. The remaining reports are unaudited.
Also requires the NRSRO to provide SEC with an unaudited report of the
number of credit rating actions during the fiscal year in each class
of credit rating for which it is registered.
Rule 17g-4:
Prescribes minimal requirements for the policies and procedures that
registered NRSROs are required to establish, maintain, and enforce in
order to address specific areas in which material, nonpublic
information could be inappropriately disclosed or used.
Rule 17g-5:
Prohibits certain actions that constitute an impermissible conflict of
interest and prescribes minimal requirements to manage other inherent
conflicts of interest. For structured finance products, requires an
NRSRO hired by an arranger to rate a structured finance product to
obtain representation from the arranger that it will provide the
information given to the hired NRSRO to the nonhired NRSROs.
Rule 17g-6:
Prohibits any act or practice by an NRSRO that SEC determines to be
unfair, abusive, or coercive.
Source: 17 C.F.R. §§ 240.17g-1 - 17g-6 (2010).
[End of table]
The Act replaced the SEC staff no-action letter process for
recognizing credit rating agencies as NRSROs with a speedier and more
transparent registration system.[Footnote 19] It created a
registration process with required information for applicants to
submit, a specific time frame for SEC's review of the application, and
specific reasons for which SEC could deny an application. According to
the Senate report accompanying the Act's passage, the new registration
program does not favor any particular business model, thus encouraging
purely statistical models to compete with the qualitative models of
the dominant rating agencies and subscriber-pays models to compete
with issuer-pays models.[Footnote 20] The Senate Report stated that
the new registration program would enhance competition and provide
investors with more choices, higher-quality ratings, and lower costs.
The Act added new section 15E to the Exchange Act, which provides, in
pertinent part, that a credit rating agency electing to register as an
NRSRO must submit the following information to SEC:
* statistics that measure the performance of credit ratings over
short- , mid-, and long-term periods;
* the procedures and methodologies the applicant uses in determining
credit ratings;
* policies or procedures the applicant adopted and implemented to
prevent the misuse of material, nonpublic information;
* the organizational structure of the applicant;
* its code of ethics and, if one does not exist, why not;
* any conflicts of interest relating to the issuance of credit ratings;
* the categories for which the applicant intends to apply for
registration;
* on a confidential basis, a list of the 20 largest issuers and
subscribers that use its credit rating services;
* on a confidential basis, written certification from 10 or more
qualified institutional buyers (QIB) that have used the credit ratings
of the applicant for at least 3 years immediately preceding the data
of the certification;[Footnote 21] and:
* any other information and documents concerning the applicant and any
person associated with such applicant as SEC, by rule, may prescribe
as necessary or appropriate in the public interest or for the
protection of investors.[Footnote 22]
In implementing Section 15E of the Exchange Act, SEC created and
adopted Form NRSRO to collect the required information as well as
audited financial statements and revenue and compensation information.
Form NRSRO requires certification by the applicant that the
information contained is accurate in all significant respects.
Section 15E(a)(2) [15 U.S.C. § 78o-7(a)(2)] sets forth the application
review requirements. SEC must either grant registration or institute
proceedings to determine if registration should be denied within 90
calendar days of a credit rating agency furnishing its application to
SEC. The deadline can be extended if the applicant consents. If SEC
institutes proceedings, it has to provide the applicant a notice of
the grounds for denial under consideration and an opportunity for a
hearing and conclude the proceedings not later than 120 days after the
date on which the application for registration was furnished.[Footnote
23] SEC can extend the conclusion of the proceedings for up to 90
days, if it finds good cause for such extension and publishes its
reasons for so finding, or for longer periods if the applicant
consents.[Footnote 24] The Act requires that SEC shall grant
registration if it finds the requirements of section 15E are
satisfied, unless it makes either of two findings (in which case
registration must be denied): first, that the applicant does not have
adequate financial and managerial resources to consistently produce
credit ratings with integrity and to materially comply with the
procedures and methodologies disclosed in Form NRSRO as well as with
the requirements of the Act regarding the prevention of misuse of
nonpublic information, management of conflicts of interest, prohibited
conduct, and the designation of a compliance officer.[Footnote 25]
Second, that the applicant or a person associated with it committed or
omitted any act such that if the applicant were registered, its
registration would be subject to suspension or revocation under
subsection (d) of Section 15E of the Exchange Act.[Footnote 26]
SEC Staff View Their Ability to Ensure That an NRSRO Applicant Meets
the Act's Requirements as Limited:
The NRSRO registration program has provided greater transparency and
shortened the time between application and SEC recognition. However,
as currently implemented, the registration process potentially limits
staff's ability to ensure information provided on applications is
accurate and lacks criteria needed to recommend that SEC deny an
application. Since the implementation of Section 15E, 11 credit rating
agencies have applied for NRSRO registration.[Footnote 27] To apply, a
credit rating agency must fill out Form NRSRO and submit it and the
required accompanying information to SEC. Trading and Markets staff
review these documents and draft a memorandum to the Commission with
the results of their review and recommendation for registration or
denial of the application. We reviewed 10 memoranda that Trading and
Markets provided the Commission and found that staff recommended that
all be registered. The Commission has instituted proceedings to deny
the application of the eleventh applicant.[Footnote 28] A few of the
10 memoranda we reviewed described concerns that Trading and Markets
had with the applications.
Staff told us these concerns generally were not resolved before
registration for several reasons: (1) staff lacked criteria against
which to measure certain concerns, (2) staff lacked the ability to
examine the credit rating agencies before registration, and (3) staff
came up against the 90-day time frame for the review of applications.
According to Trading and Markets staff, some of these concerns were
not addressed because they were qualitative in nature and would have
required subjective judgments by the staff for which section 15E,
implementing rules, or Form NRSRO do not provide criteria. For
example, staff noted a concern about one applicant's managerial
resources because the designated compliance officer lacked experience
as a credit analyst. However, because section 15E, the rules
implementing section 15E, and Form NRSRO do not prescribe minimum
qualifications for the compliance officer position, staff were unable
to support a finding that the applicant lacked the necessary
managerial resources. As previously noted, a finding that the
applicant did not have adequate financial and managerial resources
would have been grounds for denying registration.
Staff noted that because of the newness of the registration and
oversight programs and the rules, no history of regulatory compliance
could be used as a benchmark or criteria by which SEC could evaluate
whether the applicant had adequate financial or managerial resources.
Staff reviewed financial statements, other required financial
information, and required managerial information, and provided
summaries of the information in their memoranda.
Trading and Markets staff also stated that for some of the concerns
raised, they likely would need to conduct examinations to obtain the
information necessary to assess if the concerns were legitimate.
However, staff stated they cannot examine the applicants' books and
records to investigate qualitative concerns because an applicant does
not become subject to SEC's oversight authority until it becomes a
registered NRSRO. For example, staff noted that one NRSRO appeared to
produce ratings that are significantly more volatile than those of
other applicants. Staff noted that appropriate explanations for the
ratings volatility could exist but without the ability to examine the
applicant, determining the causes of this volatility would be
difficult, as would determining whether or not it demonstrated
inadequate managerial resources. Similarly, staff said that any
assessment of financial sufficiency would have to be determined
through an examination because of the uniqueness of the applicants'
business models. For example, they said a firm that uses only publicly
available information and a quantitative model to produce its ratings
likely needs far fewer financial resources than a firm that uses
qualitative information and must employ analysts to assess this
information.
In addition to qualitative concerns, staff also noted factual concerns
about the veracity of some information provided on Form NRSRO.
However, staff said that the 90-day deadline for action and the lack
of express authority to conduct an examination of the applicant did
not allow for the resolution of these types of concerns during the
application process.[Footnote 29] The deadline can be extended with
the applicant's consent but to date only two applicants have done so.
[Footnote 30]
Staff said that even if SEC had authority to examine an NRSRO
applicant prior to acting on its application, the Act's 90-day
deadline for acting on an application would not provide enough time
for a more thorough review. An application is deemed "furnished" to
SEC, and thus begins the 90-day time frame, when SEC receives a
complete and properly executed Form NRSRO. Staff said the Act does not
provide SEC with any way to extend the review period without the
applicant's consent or an SEC decision to institute proceedings (which
would extend the deadline by 30 days, or an additional 90 days based
upon a finding of good cause or upon the applicant's consent). Staff
said that generally speaking, they would not recommend that SEC
institute proceedings absent sufficient evidence to support the
findings necessary to deny an application. While possible, they said
that providing sufficient material on Form NRSRO to support the
institution of proceedings would be unlikely.[Footnote 31]
Trading and Markets staff said their principal purpose in raising
qualitative and factual concerns in memoranda was to inform the
Commission and notify OCIE so these issues could be monitored through
future examinations. Trading and Markets staff said that conducting
the lengthy examinations that likely would be needed to resolve many
of these qualitative issues in effect could be viewed as a return to
the prior staff no-action letter process in which examinations were
used to make qualitative and subjective assessments of a credit rating
agency seeking NRSRO designation. Staff pointed to the legislative
history of the Act as illustrating that Congress clearly found that
the prior staff no-action letter process created artificial barriers
to entry to NRSRO registration, and that the Act specifically was
designed to replace that process with a more efficient and transparent
registration program.[Footnote 32] For this reason, staff told us they
have interpreted the Act to mean it is not appropriate for SEC to
institute proceedings to deny an application merely to resolve staff
questions about an application, and absent sufficient evidence at the
application stage to support one of the statutory grounds for denial
of registration. As Trading and Markets staff interpret the NRSRO
registration program, they believe that the Commission must find an
applicant has satisfied the requirements of Section 15E if the
applicant meets the definition of a credit rating agency, submits the
required material, and is capable of complying with the applicable
U.S. securities laws.
Because the Act allows SEC to deny an application if the applicants
are found to have inadequate managerial or financial resources, the
investing public may have some expectation that SEC determined that
applicants had the financial and managerial resources to produce
ratings with integrity before registering them. Furthermore, because
each applicant must certify the accuracy of its information and
statements, an expectation could exist that the information provided
to SEC and on which the Commission bases its approval decision was
accurate and complete. This not only includes the qualified
institutional buyer certifications, which are used to determine
whether or not the market recognizes and uses the ratings provided by
the NRSRO, but also the public disclosures, such as the descriptions
of the ratings methodologies. We identified other registration
processes that have built in greater authority and flexibility for the
staff to clarify issues before registering applicants. For example,
both FINRA's registrant application process for broker-dealers and
SEC's registration process for investment advisors require applicants
to provide specific information and utilize deadlines to ensure an
efficient process. According to staff from these programs, they are
able to clarify any outstanding questions they have regarding
information required on the application and to delay registering the
applicant until they are satisfied the applicant has met all of the
necessary requirements.[Footnote 33] However, because the NRSRO
registration program requires SEC to act within 90 days of receiving
the application and SEC has limited ability to extend that deadline,
staff have recommended granting registration to credit rating agencies
as NRSROs with some concerns outstanding about their meeting the Act's
requirements. Furthermore, the uncertainty as to the extent of SEC's
authority to compel the production of certain additional information
that could be used to verify the information provided on Form NRSRO
and the lack of specific criteria against which to assess the
application may lead to SEC granting registration to an applicant that
does not fully meet Section 15E's requirements as an NRSRO.
New Dedicated Teams in SEC Provide Input on Regulatory Initiatives and
Examine NRSROs:
In December 2008, Trading and Markets created an NRSRO monitoring unit
to provide input on regulatory initiatives related to NRSROs. The
members of this unit are responsible for meeting periodically with
NRSROs to establish an ongoing dialogue and discuss topics such as
updates to rating methodologies and practices, financial results, and
compliance and internal audit activity. The unit is also responsible
for preparing internal periodic profile reports of each NRSRO and the
annual report to Congress on NRSROs, and analyzing and preparing
reports on topics of interest or potential concern. According to
documents we reviewed, the unit also is supposed to meet periodically
with NRSROs to discuss issues specifically related to model
development, validation, and governance to gain a better understanding
of the models and the controls around each. As of August 2, 2010, this
unit has three members with qualifications and experience, ranging
from a former rating agency managing director with more than 20 years
experience to an analyst with 6 years of experience as a corporate
credit analyst.
Following its registration as an NRSRO, a credit rating agency
immediately becomes subject to SEC oversight, including compliance
examinations and enforcement. OCIE, which conducts NRSRO examinations,
in May 2009 reorganized the Office of Market Oversight to create a new
NRSRO examination team (a branch chief, a senior specialized examiner,
and three staff examiners) to perform NRSRO examinations.[Footnote 34]
The senior specialized examiner and three staff examiners were new SEC
employees and include two former NRSRO analysts (both former managing
directors of a major rating agency with over 20 and over 10 years
experience respectively) and attorneys with experience in structured
finance products and corporate law. According to OCIE staff, the new
examiners have received standard OCIE training and on-the-job training
from the branch chief and other examiners who completed previous NRSRO
examinations. In addition, OCIE is considering incorporating NRSRO-
specific training into its standard examiner training and has
developed written guidance to assist examiners and foster consistency
in examinations.
Based on our review of OCIE's NRSRO examination guidelines and
interviews with OCIE staff, OCIE plans to conduct routine and special
NRSRO examinations. Routine examinations assess an NRSRO for
compliance with the Act and applicable rules and regulations at
regular intervals. Special, or cause, examinations typically originate
from a tip or need to gather specific information (limited scope) or
follow up on past examination findings and recommendations. OCIE
expects to conduct special or cause examinations as necessary.
According to the guidelines, each routine examination will generally
begin with an initial risk assessment and scope analysis of the NRSRO
to be reviewed. Specific review areas for each examination (such as
conflicts of interest or document retention) are determined during the
risk assessment process and throughout the examination. OCIE explained
that all examinations are based on risk assessment to maximize the
examination team's time and resources.
NRSRO examination staff typically review, among other things, whether
the NRSRO adequately has (1) disclosed a description of its ratings
procedures and methodologies; (2) documented internally its ratings
procedures and methodologies; (3) documented its ratings process in
ratings files, including making and retaining required documentation;
and (4) adhered to its ratings policies and procedures in the creation
of ratings. OCIE examinations do not assess whether the NRSRO produces
accurate ratings, as SEC is prohibited from evaluating the substance
of credit ratings or the procedures and methodologies by which an
NRSRO determines credit ratings. Areas of review may include credit
rating and surveillance process, record retention, prevention of the
misuse of material nonpublic information, conflicts of interest,
prohibited acts and practices, financial operations, marketing,
compliance, internal audit, and unsolicited ratings. For example,
under the conflicts of interest area, the guidance provides a
description of the applicable rule (17g-5) and a description of the
disclosure requirements. It suggests OCIE request all current written
policies and procedures related to conflicts of interest, and then
assess if these policies and procedures were designed reasonably,
captured all the relevant conflicts, and were followed. OCIE has been
working with the NRSROs to standardize the level of detail that the
examiners would expect to see. This process is ongoing and examiners
likely will have to complete a few examinations before determining
exactly what they would need to see to make compliance determinations.
If deficiencies were found, OCIE would send a deficiency letter to the
NRSRO requesting that it make the appropriate corrections and notify
OCIE of how it plans to make those corrections. In cases of potential
violations, OCIE may refer the NRSRO to the Enforcement Division for
further review.
OCIE completed its first series of examinations of the largest NRSROs
in July 2008. The examinations focused on the rating of subprime RMBS
by Standard & Poor's, Moody's, and Fitch and were initiated in
response to the recent mortgage crisis. OCIE conducted these
examinations jointly with Trading and Markets and made its examination
results public.[Footnote 35] Although these examinations occurred
after the enactment of the Act, the period subject to examination
predated the Act.
Among other things, the examinations found that:
* RMBS and collateralized debt obligations (CDO) deals substantially
increased in number and complexity since 2002, and some of the rating
agencies appear to have struggled with that growth;
* significant aspects of the rating process were not always disclosed;
* the management of conflicts of interest raised some concerns; and:
* the surveillance processes the ratings agencies used appeared to
have been less robust than the processes used for initial ratings.
The examinations also resulted in remedial actions that the examined
NRSROs stated they would take to address the findings of these
examinations and additional rulemaking by SEC. Examples of remedial
actions included that the examined NRSROs evaluate, both at that time
and on a periodic basis, whether they have sufficient staff and
resources to manage their volume of business and meet their
obligations under Section 15E of the Exchange Act and the rules
applicable to NRSRO, and that each NRSRO conduct a review of its
current disclosures relating to process and methodologies for rating
RMBS and CDOs to assess whether it is fully disclosing its ratings
methodologies in compliance with Section 15E of the Exchange Act and
the rules applicable to NRSROs. The additional rulemaking included
amendments to SEC's rules identifying a series of conflicts of
interest that NRSROs must disclose and manage and others that were
outright prohibited.
In October 2008, OCIE began routine examinations of four of the
remaining seven NRSROs. As of August 30, 2010, OCIE had completed
three examinations and closed the remaining examination.[Footnote 36]
OCIE staff explained these routine examinations took longer to
complete than anticipated because of the transition from the old to
the new staff, the resignation or departure from the area of some
examination staff, and the need to conduct other examinations and
initiatives. For example, OCIE also has completed three special
(limited scope) examinations and begun another special (cause)
examination since the NRSRO examination program started. Given the
small number of NRSRO examiners, the need to conduct these additional
examinations slowed progress on the routine examination schedule.
NRSRO examiners also have been conducting outreach initiatives for
designated compliance officers of the NRSROs, which are intended to
educate them on SEC's expectations with regard to compliance officers'
skills and backgrounds. As a result, with the current level of
staffing it is unlikely that OCIE would have been able to meet its
planned routine examination schedule of examining the three largest
NRSROs every 2 years and the remaining NRSROs every 3 years depending
on staffing resources, and two examinations have taken over 18 months
to complete. SEC requested additional resources which it anticipated
using to fully staff this oversight function.
Under the recently passed Dodd-Frank Act, SEC must establish an Office
of Credit Ratings and conduct annual examinations of each NRSRO. The
Dodd-Frank Act also outlines eight specific areas these examinations
must review and requires that the Commission produce an annual public
report summarizing the findings of these examinations. According to
OCIE staff, in August 2010, the NRSRO examination team will begin new
examinations of all NRSROs as mandated by the Dodd-Frank Act. OCIE
staff said they are relying on currently designated NRSRO examiners
and examiners from other OCIE examinations programs to staff the teams.
The Dodd-Frank Act requires SEC to staff the office sufficiently to
carry out these requirements. Although SEC may be able to staff the
teams needed to conduct the first round of annual examinations by
using staff from other examination programs, such an approach is not
sustainable in the long term. In creating this new office, developing
a plan to assess not only the number of staff it needs but also the
skills required of this staff would help SEC be strategically
positioned to meet the Act's requirements. Without a plan that helps
ensure SEC has sufficient staff that have the appropriate
qualifications and are appropriately trained, SEC may face challenges
in meeting the required examination timetable and providing quality
oversight over NRSROs.
SEC Has Increased the Amount of Performance-related Data NRSROs Are
Required to Disclose, but These Data Have Limited Usefulness:
Since the implementation of the Act, SEC has made several revisions to
the Form NRSRO that are intended to make more information publicly
available for evaluating and comparing NRSRO performance. The revised
Form NRSRO requires NRSROs to disclose credit rating performance
statistics over 1-, 3-, and 10-year periods. SEC has also required
NRSROs to make ratings history data publicly available. SEC intended
these disclosures to allow users to better evaluate and compare NRSRO
performance. However, because SEC did not specify how NRSROs should
calculate these statistics, the NRSROs used varied methodologies,
limiting their comparability. Further, we found that the ratings
history data sets do not contain enough information to construct
comparable performance statistics and are not representative of the
population of credit ratings at each NRSRO. Without better
disclosures, the information being provided will not serve its
intended purpose of increasing transparency.
SEC-required Performance Statistics Have Increased the Data Available
from NRSROs, but Their Comparability Is Limited:
Pursuant to the requirements of section 15E of the Exchange Act,
NRSROs are required to disclose credit rating performance statistics
over short-, mid-, and long-term periods, as applicable. Form NRSRO
specifies that these statistics must at a minimum show the performance
of credit ratings in each class of class for which an NRSRO is
registered over 1-, 3-, and 10-year periods, including, as applicable,
historical transition and default rates for each rating category and
notch. The statistics must include defaults relative to initial
ratings.[Footnote 37] Transition rates compare ratings at the
beginning of a time period with ratings at the end of the time period,
while default rates show the percentage of ratings with each rating
that defaulted over a given time period. SEC requires NRSRO applicants
to furnish the required transition and default rates as part of their
NRSRO application on Form NRSRO, and once registered, to update the
statistics annually. As part of these disclosures, NRSROs must define
the credit rating categories they use and explain these performance
measures, including the inputs, time horizons, and metrics used to
determine them.
SEC Intended Statistics to Facilitate Comparison of Credit Rating
Performance among NRSROs:
In adopting these requirements, SEC stated that these types of
statistics are important indicators of the performance of a rating
agency in terms of its ability to assess the creditworthiness of
issuers and, consequently, would be useful to users of ratings in
evaluating a rating agency's performance. SEC specified the 1-, 3-,
and 10-year periods so that the performance statistics the NRSROs
generated would be more easily comparable. SEC also stated that
requiring NRSROs to define the ratings categories used and explain
their performance statistics would assist users of ratings in
understanding how the measurements were derived and in comparing the
measurement statistics of different NRSROs.
In deciding which statistics to require, SEC identified default and
transition rates as common benchmarks.[Footnote 38] To compute 1-year
transition rates by rating category, an NRSRO will form cohorts by
grouping all of the entities (issues or issuers) with ratings
outstanding at the beginning of the year by their rating at the
beginning of the year. The NRSRO then calculates the number or
percentage of entities in each cohort that have each possible rating
at the end of the year. Table 2 provides a hypothetical example of a 1-
year transition matrix for cohorts of rated entities for 2009.
Table 2: Hypothetical 1-Year Transition Matrix and 1-Year Default
Rates Relative to Beginning-of-Year Ratings for 2009 Cohorts:
Rating as of Jan 1, 2009: AAA;
Rating as of Dec. 31, 2009: AAA: 93.3%;
Rating as of Dec. 31, 2009: AA: 5.4%;
Rating as of Dec. 31, 2009: A: 1.3%;
Rating as of Dec. 31, 2009: BBB: 0.0%;
Rating as of Dec. 31, 2009: BB: 0.0%;
Rating as of Dec. 31, 2009: B: 0.0%;
Rating as of Dec. 31, 2009: CCC: 0.0%;
Rating as of Dec. 31, 2009: CC: 0.0%;
Rating as of Dec. 31, 2009: D: 0.0%;
Rating as of Dec. 31, 2009: W: 0.0%;
Rating as of Dec. 31, 2009: Cohort size[A]: 149.
Rating as of Jan 1, 2009: AA;
Rating as of Dec. 31, 2009: AAA: 0.4%;
Rating as of Dec. 31, 2009: AA: 92.1%;
Rating as of Dec. 31, 2009: A: 7.0%;
Rating as of Dec. 31, 2009: BBB: 0.4%;
Rating as of Dec. 31, 2009: BB: 0.0%;
Rating as of Dec. 31, 2009: B: 0.0%;
Rating as of Dec. 31, 2009: CCC: 0.0%;
Rating as of Dec. 31, 2009: CC: 0.0%;
Rating as of Dec. 31, 2009: D: 0.0%;
Rating as of Dec. 31, 2009: W: 0.2%;
Rating as of Dec. 31, 2009: Cohort size[A]: 543.
Rating as of Jan 1, 2009: A;
Rating as of Dec. 31, 2009: AAA: 0.0%;
Rating as of Dec. 31, 2009: AA: 1.3%;
Rating as of Dec. 31, 2009: A: 91.3%;
Rating as of Dec. 31, 2009: BBB: 6.9%;
Rating as of Dec. 31, 2009: BB: 0.2%;
Rating as of Dec. 31, 2009: B: 0.0%;
Rating as of Dec. 31, 2009: CCC: 0.0%;
Rating as of Dec. 31, 2009: CC: 0.0%;
Rating as of Dec. 31, 2009: D: 0.0%;
Rating as of Dec. 31, 2009: W: 0.3%;
Rating as of Dec. 31, 2009: Cohort size[A]: 1,181.
Rating as of Jan 1, 2009: BBB;
Rating as of Dec. 31, 2009: AAA: 0.0%;
Rating as of Dec. 31, 2009: AA: 0.2%;
Rating as of Dec. 31, 2009: A: 3.6%;
Rating as of Dec. 31, 2009: BBB: 90.4%;
Rating as of Dec. 31, 2009: BB: 5.0%;
Rating as of Dec. 31, 2009: B: 0.5%;
Rating as of Dec. 31, 2009: CCC: 0.0%;
Rating as of Dec. 31, 2009: CC: 0.0%;
Rating as of Dec. 31, 2009: D: 0.1%;
Rating as of Dec. 31, 2009: W: 0.3%;
Rating as of Dec. 31, 2009: Cohort size[A]: 1,266.
Rating as of Jan 1, 2009: BB;
Rating as of Dec. 31, 2009: AAA: 0.0%;
Rating as of Dec. 31, 2009: AA: 0.0%;
Rating as of Dec. 31, 2009: A: 0.3%;
Rating as of Dec. 31, 2009: BBB: 8.1%;
Rating as of Dec. 31, 2009: BB: 83.6%;
Rating as of Dec. 31, 2009: B: 6.9%;
Rating as of Dec. 31, 2009: CCC: 0.4%;
Rating as of Dec. 31, 2009: CC: 0.0%;
Rating as of Dec. 31, 2009: D: 0.1%;
Rating as of Dec. 31, 2009: W: 0.6%;
Rating as of Dec. 31, 2009: Cohort size[A]: 700.
Rating as of Jan 1, 2009: B;
Rating as of Dec. 31, 2009: AAA: 0.0%;
Rating as of Dec. 31, 2009: AA: 0.0%;
Rating as of Dec. 31, 2009: A: 0.0%;
Rating as of Dec. 31, 2009: BBB: 0.2%;
Rating as of Dec. 31, 2009: BB: 8.7%;
Rating as of Dec. 31, 2009: B: 83.2%;
Rating as of Dec. 31, 2009: CCC: 4.1%;
Rating as of Dec. 31, 2009: CC: 1.5%;
Rating as of Dec. 31, 2009: D: 0.7%;
Rating as of Dec. 31, 2009: W: 1.7%;
Rating as of Dec. 31, 2009: Cohort size[A]: 541.
Rating as of Jan 1, 2009: CCC;
Rating as of Dec. 31, 2009: AAA: 0.0%;
Rating as of Dec. 31, 2009: AA: 0.0%;
Rating as of Dec. 31, 2009: A: 0.0%;
Rating as of Dec. 31, 2009: BBB: 2.1%;
Rating as of Dec. 31, 2009: BB: 0.0%;
Rating as of Dec. 31, 2009: B: 19.1%;
Rating as of Dec. 31, 2009: CCC: 68.1%;
Rating as of Dec. 31, 2009: CC: 2.1%;
Rating as of Dec. 31, 2009: D: 4.3%;
Rating as of Dec. 31, 2009: W: 4.3%;
Rating as of Dec. 31, 2009: Cohort size[A]: 47.
Rating as of Jan 1, 2009: CC;
Rating as of Dec. 31, 2009: AAA: 0.0%;
Rating as of Dec. 31, 2009: AA: 0.0%;
Rating as of Dec. 31, 2009: A: 0.0%;
Rating as of Dec. 31, 2009: BBB: 0.0%;
Rating as of Dec. 31, 2009: BB: 0.0%;
Rating as of Dec. 31, 2009: B: 20.0%;
Rating as of Dec. 31, 2009: CCC: 20.0%;
Rating as of Dec. 31, 2009: CC: 0.0%;
Rating as of Dec. 31, 2009: D: 20.0%;
Rating as of Dec. 31, 2009: W: 40.0%;
Rating as of Dec. 31, 2009: Cohort size[A]: 40.
Source: GAO.
[A] Cohort size refers to the number of entities that had a particular
rating at the beginning of the time period, in this case January 1,
2009.
[End of table]
The rows of the matrix show all possible credit ratings an entity
could have at the beginning of the year, by rating category. The
columns of the matrix show all possible credit ratings an entity could
have at the end of the year, also by rating category. The matrix also
includes columns for defaults (D) and withdrawals (W), since an entity
could default or have its rating withdrawn during the course of the
year. The table cells show the rates at which ratings migrate upward,
downward, or stay the same. For example, 1.3 percent of the entities
rated A at the beginning of the year were rated AA at the end of the
year, 6.9 percent of the entities rated A at the beginning of the year
were rated BBB at the end of the year, and 91.3 percent of the
entities rated A at the beginning of the year were rated A at the end
of the year.
By showing the number or fraction of entities in each cohort with
stable ratings--that is, with the same beginning-of-year and end-of-
year ratings--transition matrixes allow users to compare the stability
of different rating categories (for the same NRSRO). Table 1 shows
that 93.3 percent of entities rated AAA at the beginning of the year
were still rated AAA at the end of the year, but 90.4 percent of
entities rated BBB at the beginning of the year were still rated BBB
at the end of the year. Thus, AAA ratings demonstrated more stability
than BBB ratings. In general, an NRSRO's credit ratings are intended
to order rated entities by their creditworthiness, with high ratings
indicating relatively more creditworthiness than low ratings. One
aspect of creditworthiness is the extent to which it changes over
time, with more creditworthy entities demonstrating greater stability.
Thus, an NRSRO's higher ratings should exhibit more stability than its
lower ratings. Transition rates illustrate how well a particular
rating scale rank orders credit risk on this margin.
As previously stated, default rates show the percentage of entities
with each rating that defaulted over a given time period. For example,
to calculate simple 1-year default rates relative to beginning-of-year
ratings, an NRSRO will form cohorts by grouping all of the entities
with ratings outstanding at the beginning of the year by their rating
at the beginning of the year. The NRSRO then calculates the fraction
of entities in each cohort that default during the year. Table 2 shows
hypothetical 1-year default rates for the 2009 cohorts of rated
entities. For example, no bonds rated AAA defaulted during 2009, while
20 percent of bonds rated CC defaulted during this period.
SEC rules require that NRSROs disclose on their Form NRSRO 1-, 3-, and
10-year default rates relative to initial ratings. For example, to
calculate simple 1-year default rates relative to initial ratings, an
NRSRO could form cohorts by grouping all of the entities assigned
initial ratings during a given time period by their initial rating.
The NRSRO could then calculate the fraction of entities in each cohort
that default within 1 year after they receive their initial rating. As
previously discussed, an NRSRO's credit ratings are intended to put
rated entities in order of their creditworthiness. It follows that
entities with higher credit ratings should default less often than
entities with lower credit ratings. Default rates also illustrate how
accurately a particular rating scale rank orders credit risk on this
margin.
NRSROs' Differing Methodologies for Calculating Performance Statistics
Limit Their Comparability:
We reviewed the 2009 performance statistics published by the 10 NRSROs
as part of their annual update to Form NRSRO, focusing on the
corporate and structured finance asset classes. The required
disclosures increased the amount of information publicly available
about the performance of some NRSROs, particularly those that were
newly registered. However, SEC did not specify how the NRSROs were to
calculate the required performance statistics, and, as a result, the
NRSROs used different methodologies for calculating the transition and
default rates. For the transition rates, they differed by whether they
(1) were for a single cohort or averaged over many cohorts, (2)
constructed cohorts on an annual basis or monthly basis, (3) were
adjusted for entities that have had their ratings withdrawn or
unadjusted, and (4) allowed entities to transition to default or not.
Because of these differences, users cannot use the performance
statistics to compare transition rates across NRSROs, as the rule
intended.
First, some NRSROs provided transition rates for individual cohorts
for the most recent 1-, 3-, and 10-year period.[Footnote 39] These
NRSROs provided statistics summarizing the transition rates for
individual cohorts--for example, 1-year transition rates for the 2009
cohort, 3-year transition rates for the 2007 cohort, and 10-year
transition rates for the 2000 cohort.[Footnote 40] Other NRSROs
provided average transition rates for multiple 1-, 3-, and 10-year
time periods over a range of years. For example, one NRSRO calculated
1-year transition rates for every annual cohort from 1981 to 2009
(obtaining 28 separate 1-year transition rates) and then averaged the
rates in those matrixes to obtain average 1-year transition rates for
each rating category and notch. The NRSRO used the same methodology to
calculate average 3-year and average 10-year transition rates over the
same period.
Single and average cohort approaches provide different information
about an NRSRO's performance. The single cohort approach uses
information from only the most recent 1-, 3-, and 10-year periods and
thus describes the NRSRO's performance at specific points in time. As
such, it is useful for describing the historical experience of a
particular group of ratings under a particular set of circumstances.
Single cohort transition matrixes are thus useful as predictors of the
performance of ratings in future time periods under similar
circumstances, but they are less useful as predictors of the
performance of ratings in future time periods under different economic
and other conditions. On the other hand, the average cohort approach
uses information from multiple time periods and thus describes the
NRSRO's performance during an average 1-, 3-, or 10-year time period.
As such, average cohort transition rates are useful indicators of
expected transition rates in the future, given that future economic
and other conditions are unknown. Both approaches are valid, depending
on the needs of the user, but they do not yield comparable information.
Second, the NRSROs differ in whether they construct cohorts on an
annual or monthly cohort basis. The frequency with which the cohorts
are formed affects the accuracy of average transition rates. The
higher the sampling frequency--the shorter the time period between
cohort formation dates--the more observations are available for
calculating the averages and the more accurate the transition rates
are for predictive purposes. Most of the NRSROs used annual cohorts
(that is, they formed a new cohort on December 30 or 31 or January 1
of each year), but one NRSRO used monthly cohorts (that is, it formed
a new cohort on the first day of each month). Using monthly cohorts
means that 12 times as many observations contribute to average
transition rates.
Third, we found some NRSROs adjusted their transition rates to reflect
those entities with ratings that were withdrawn during the time
period, while others did not. NRSROs withdraw ratings for a number of
reasons. In many cases, the issue matures and the rating is no longer
needed. In other cases, the NRSRO discontinues a rating for lack of
information. Transition rates that include entities with withdrawn
ratings in the cohorts are called "unadjusted" rates, while those that
exclude entities with withdrawn ratings are called "withdrawal-
adjusted" rates. The treatment of withdrawn ratings in calculating
transition rates can have a significant impact on the magnitude of the
rates. For example, suppose a cohort with an initial BBB rating has
100 rated entities at the beginning of the time period, and suppose
that 25 are withdrawn during the period and the remainder are still
rated BBB at the end of the period. The unadjusted transition rate
from BBB to BBB would be 75 percent (75/100) and the withdrawal-
adjusted transition rate would be 100 percent (75/75). For one NRSRO,
we could not determine from the disclosures how it treated withdrawals
in its transition rates.
Fourth, one NRSRO did not include transitions to default in its
transition rates. As a result, its performance statistics do not
include information on the number of ratings that moved from a
particular rating category to default during the 1-, 3-, or 10-year
periods. This information is important to the investor, because
ratings that move from a higher rating category (such as AAA, AA, or
A) directly to default within the time period may signal poor ratings
performance by the NRSROs. Table 3 summarizes the variation in the
calculation of transition rates that we found in reviewing the NRSROs'
2009 performance statistics.
Table 3: Methods Used by NRSROs to Calculate SEC-required 1-, 3-, and
10-year Transition Rates:
NRSRO methods for calculating transition rates on corporate issuer
ratings[A]:
Single-cohort transition rates[B]: Withdrawal-adjusted, includes
transitions to default[C];
Number of NRSROs using these methods: 2.
Single-cohort transition rates[B]: Not withdrawal-adjusted, includes
transitions to default;
Number of NRSROs using these methods: 2.
Average transition rates for multiple cohorts[D]: Withdrawal-adjusted,
includes transitions to default;
Number of NRSROs using these methods: 2.
Average transition rates for multiple cohorts[D]: Not withdrawal-
adjusted;
Number of NRSROs using these methods: [Empty].
Average transition rates for multiple cohorts[D]: Includes transitions
to default;
Number of NRSROs using these methods: 1.
Average transition rates for multiple cohorts[D]: Does not include
transitions to default;
Number of NRSROs using these methods: 1.
Transition rate methodology unclear:
Number of NRSROs using these methods: 1.
NRSRO methods for calculating transition rates on corporate issuer
ratings[A]: Total[E];
Number of NRSROs using these methods: 9.
NRSRO methods for calculating transition rates on corporate issuer
ratings[A]:
Single-cohort transition rates: Withdrawal-adjusted, includes
transitions to default;
Number of NRSROs using these methods: 2.
Single-cohort transition rates: Not withdrawal-adjusted, includes
transitions to default;
Number of NRSROs using these methods: 3.
Average transition rates for multiple cohorts: Withdrawal-adjusted,
includes transitions to default;
Number of NRSROs using these methods: 2.
Average transition rates for multiple cohorts: Not withdrawal-adjusted;
Number of NRSROs using these methods: [Empty].
Average transition rates for multiple cohorts: Includes transitions to
default;
Number of NRSROs using these methods: 1.
Average transition rates for multiple cohorts: Does not include
transitions to default;
Number of NRSROs using these methods: 1.
Transition rate methodology unclear:
Number of NRSROs using these methods: 1.
NRSRO methods for calculating transition rates on corporate issuer
ratings[A]: Total;
Number of NRSROs using these methods: 10.
Source: GAO analysis of transition rates reported on 2009 Form NRSRO
filings.
[A] Two NRSROs did not provide transition rates, but provided data so
users could calculate these rates. Transition matrixes may also vary
across NRSROs on margins not reported in the table.
[B] Single-cohort transition rates describe the performance of
individual cohorts.
[C] Withdrawal-adjusted transition rates are those for which withdrawn
ratings are excluded from cohorts.
[D] Average transition rates describe the average performance of
multiple cohorts over a given time period.
[E] Total does not add up to 10 because one NRSRO is not registered in
the corporate asset class.
[End of table]
NRSROs also used different methodologies for calculating default
rates. In general, default rates differed by whether they were (1)
relative to ratings at the beginning of a given time period or
relative to initial ratings, (2) adjusted for entities that had their
ratings withdrawn or unadjusted, (3) adjusted for how long entities
survived without defaulting or unadjusted, (4) calculated using annual
or monthly cohorts, and (5) calculated for a single cohort or averaged
over many cohorts. Because of these differences, users cannot compare
default rates across NRSROs, as the rule intended.
First, most NRSROs reported default rates relative to ratings at the
beginning of a specific time period, while two reported default rates
relative to initial rating.[Footnote 41] Calculating default rates
relative to ratings at the beginning of a given time period is similar
to calculating transition rates relative to ratings at the beginning
of a given time period. NRSROs form cohorts by grouping entities that
had the same outstanding rating on a specific date, and then
calculating the number or fraction of entities in each group that
defaulted over a given time period. To calculate default rates
relative to initial rating, NRSROs form cohorts by grouping entities
that were assigned the same initial rating, regardless of when the
initial ratings were assigned. One NRSRO calculated the default rates
relative to initial ratings over the entire period for which it had
historical ratings data, from 1983-2009. Issuers that were assigned
initial ratings of AAA at any point during that 26-year period were
grouped, as were issuers that were assigned initial ratings of AA, and
so forth. The NRSRO then determined whether there had been any
defaults at any time for any of those issuers over the 26 years and
calculated the default rate. Another NRSRO used a different
methodology and provided default rates relative to the initial ratings
only for those ratings that were outstanding at the beginning of the
most recent 1-, 3-, and 10-year periods. For example, for the 1-year
period, this NRSRO determined the ratings that had been outstanding as
of December 31, 2008, grouped them according to their initial ratings,
determined whether any defaulted in 2009, and calculated the default
rates.
Default rates for entities based on their initial rating omit
important information about the performance of NRSRO ratings over
time. In some asset classes (specifically corporate, financial
institution, and insurance company), performance statistics are based
on issuer ratings, not the ratings on specific securities of those
issuers.[Footnote 42] For some issuers, their ratings history spans
decades or longer. The initial rating given to those firms is not
relevant information at the time the issuer defaulted, because the
performance of the issuer likely changed throughout the years. For
example, an issuer initially could have been rated AAA 30 years ago
but deteriorated in the last few years. An investor likely would be
more interested in the last ratings the NRSRO provided for the issuer
to determine whether it accurately predicted the default. In contrast,
structured finance products typically do not have maturities that last
for decades, so calculating defaults relative to initial ratings may
provide more useful information in that sector.[Footnote 43]
Second, as with transition rates, some NRSROs calculated their default
rates adjusting for withdrawn ratings, while others did not. In
addition to affecting the relative magnitude of default rates, the
treatment of withdrawn ratings also provides different information
about default risk. Unadjusted default rates describe the historical
frequency of defaults for a cohort during a given time period and
treat entities with withdrawn ratings as if they had remained in the
data sample for the entire period. They can be used to predict the
expected probability of default for entities over a time period that
is at most as long as the period used to calculate the default rate.
However, unadjusted default rates likely underestimate actual default
rates for a cohort because NRSROs are less likely to observe defaults
among entities with withdrawn ratings, either because they choose not
to track those entities or because they have less access to
information about them. Furthermore, unadjusted default rates are only
useful for predicting default rates for entities that have withdrawal
patterns similar to those of the cohort used to calculate the
unadjusted default rates. The greater the differences in the
withdrawal experience of two groups of rated entities, the less useful
the unadjusted default rates for one group are for predicting defaults
in the other group. For one NRSRO, we could not determine from the
disclosures how it treated withdrawals in its default rates.
On the other hand, withdrawal-adjusted default rates describe the
historical frequency of defaults for entities during a given time
period conditional on those entities having a rating outstanding for
the entire period. These statistics treat entities with withdrawn
ratings as if they faced the same likelihood of default as entities
with ratings that were not withdrawn. Withdrawal-adjusted default
rates can be used to predict the expected probability of default for
entities over the same length of time as the period used to calculate
the default rate. The usefulness of the prediction does not depend on
similarities in withdrawal patterns for the entities. However,
withdrawal-adjusted default rates assume that withdrawals are random
and not correlated with the likelihood that an entity defaults.
Withdrawal-adjusted default rates can be biased downward or upward if
entities with withdrawn ratings are more or less likely to default,
respectively, than entities with ratings that were not withdrawn.
Third, some NRSROs reported simple default rates while others reported
default rates conditioned on how long the entities went without
defaulting or withdrawing. NRSROs calculate simple default rates by
dividing the number of defaults that occurred over a specific time
period by the number of rated entities in the cohort at the beginning
of the time period. For example, some NRSROs reported the simple 3-
year default rate for the 2007 cohort. To calculate this, they formed
cohorts for 2007, took the number of defaults that occurred over the 3-
year period in each cohort, and divided them by the number of rated
entities included in each cohort. NRSROs calculate conditional default
rates by adjusting for the fact that entities can default at different
points during the chosen time period. This method is called
conditional because default rates are conditioned upon those issuers
that "survived" for a particular amount of time.[Footnote 44] Simple
defaults rates are equal to conditional defaults rates if neither are
adjusted for withdrawals. However, simple withdrawal-adjusted default
rates are larger than conditional withdrawal-adjusted default rates.
The former assume that all withdrawals occurred at the beginning of
the period and thus never had an opportunity to default. The latter
reflect the fact that withdrawals occur at different times during the
period, so the number of ratings that could default are larger at the
beginning of the period than at the end.
Fourth, as with transition rates, the frequency with which the cohorts
are formed affects the accuracy of average transition rates. Again,
most of the NRSROs used annual cohorts, but two NRSROs used monthly
cohorts.
Fifth, some NRSROs reported default rates for the most recent 1-, 3-,
and 10-year periods for individual cohorts, while others reported
average 1-, 3-, and 10-year default rates for multiple cohorts. For
example, some NRSROs reported simple 3-year default rates for the 2007
cohort. However, one NRSRO reported average simple 3-year default
rates for 1990-2009. It did so by first calculating simple 3-year
default rates for every cohort from 1990 through 2009 and then
averaging those default rates.
Table 4 summarizes the variation in the calculation of the default
rates that we found in our review of the 2009 performance statistics.
Table 4: Methods Used by NRSROs to Calculate SEC-required 1-, 3-, and
10-year Default Rates:
NRSRO methods for calculating default rates on corporate issuer
ratings[A]:
Single-cohort default rates relative to beginning-of-year ratings[B]:
Simple, withdrawal-adjusted[C];
Number of NRSROs using these methods: 2.
Single-cohort default rates relative to beginning-of-year ratings[B]:
Simple, unadjusted;
Number of NRSROs using these methods: 1.
Single-cohort default rates relative to beginning-of-year ratings[B]:
Conditional-on-survival, withdrawal-adjusted[D];
Number of NRSROs using these methods: 2.
Average default rates for multiple cohorts relative to beginning-of-
year ratings[E]: Simple, withdrawal-adjusted;
Number of NRSROs using these methods: 2.
Default rates relative to initial rating: Conditional-on-survival,
withdrawal-adjusted;
Number of NRSROs using these methods: 1.
Default rate methodology unclear:
Number of NRSROs using these methods: 1.
NRSRO methods for calculating default rates on corporate issuer
ratings[A]: Total[F];
Number of NRSROs using these methods: 9.
NRSRO methods for calculating default rates on structured finance
ratings:
Single-cohort default rates relative to beginning-of-year ratings:
Simple, withdrawal-adjusted;
Number of NRSROs using these methods: 1.
Single-cohort default rates relative to beginning-of-year ratings:
Simple, not withdrawal-adjusted;
Number of NRSROs using these methods: 2.
Single-cohort default rates relative to beginning-of-year ratings:
Conditional-on-survival, withdrawal-adjusted;
Number of NRSROs using these methods: 1.
Average default rates for multiple cohorts relative to beginning-of-
year ratings: Simple, withdrawal adjusted;
Number of NRSROs using these methods: 2.
Default rates relative to initial rating: Simple, not withdrawal-
adjusted;
Number of NRSROs using these methods: 1.
Default rates relative to initial rating: Conditional-on-survival,
withdrawal-adjusted;
Number of NRSROs using these methods: 1.
Default rate methodology unclear:
Number of NRSROs using these methods: 1.
NRSRO methods for calculating default rates on corporate issuer
ratings[A]: Total[F];
Number of NRSROs using these methods: 9.
Source: GAO analysis of default rates reported on 2009 Form NRSRO
filings:
[A] Default rates may also vary across NRSROs on margins not
summarized in the table.
[B] Single-cohort default rates describe the performance of individual
cohorts.
[C] Withdrawal-adjusted default rates are those for with withdrawn
ratings are excluded from cohorts.
[D] Conditional default rates are those that are adjusted for the fact
that entities can default at different points during the chosen time
period.
[E] Average default rates describe the average performance of multiple
cohorts over a given time period.
[F] Totals do not add up to 10 because one NRSRO is not registered in
the corporate asset class and one NRSRO did not report any default
rates for issuers of ABS.
[End of table]
Besides not specifying how the NRSROs should calculate their
transition and default rates, SEC did not specify how the NRSROs
should present their performance statistics. For example, four NRSROs
reported their transition and default rates as percentages, but did
not report the absolute number of ratings in each cohort. As a result,
these disclosures have limited utility for comparison purposes. All
else being equal, transition and default rates will be more precise,
and thus more meaningful, the greater the number of observations used
to calculate them. Furthermore, defaults are relatively rare events
that may not be observed at all in samples that are too small. Knowing
the absolute numbers of ratings in each cohort is thus important for
comparative purposes to give users an idea of precisely how transition
and default rates are calculated and the total numbers of entities
involved in each rate. That is, if an NRSRO's default rate is 20
percent, it is useful for users to know if one out of five rated
entities defaulted, or if 10,000 out of 50,000 rated entities
defaulted. As another example, if an NRSRO's default rate increases
from 10 to 12 percent in different years, it is useful for users to
know whether that 2 percentage point difference resulted from the
default of 20 or 2,000 additional rated entities.
At least one SEC-designated asset classes may be too broadly defined
to be meaningful. Two NRSROs may rate the same asset class, but
differences in ratings performance may reflect the differences between
the sets of rated issues and issuers, and not necessarily provide
insights into the relative merits of the ratings methodology used. For
example, several NRSROs specialize in rating certain asset classes.
Realpoint only rates commercial mortgage-backed securities (CMBS)
while Japan Credit Rating Agency and Ratings and Investment, the two
Japanese NRSROs, focus on Japanese issues. Thus, comparing the
transition and default rates of these NRSROs with those of other
NRSROs that may rate more types of structured finance products or
focus on other geographic regions, may not be meaningful. In
structured finance, the NRSROs that rate ABS generally present
performance statistics for this asset class by sectors in their
voluntary disclosures; that is, CMBS, RMBS, and ABS backed by auto,
student, or credit card loans. These ABS sectors have risk
characteristics that vary significantly. Thus, presenting performance
statistics for the ABS asset class as whole, instead of by sectors,
may not be useful. SEC has not yet re-evaluated the appropriateness of
the currently designated asset classes to determine if they are
appropriate for presenting performance statistics.
Trading and Markets staff said SEC issued rules requiring the NRSRO to
publish performance statistics under tight time frames.[Footnote 45]
Because this is a new area for SEC, staff said they wanted to focus on
drafting a rule that would be appropriate. They said that once SEC,
NRSROs, and the market obtain some experience with these disclosures,
SEC could respond to any issues or comments with further rulemakings.
Because the NRSROs do not have specific guidance for calculating and
presenting their currently required performance statistics, they used
different methodologies. As a result, users of these statistics cannot
compare ratings performance across NRSROs. Further, asset classes that
are defined too broadly limit the usefulness of the disclosures.
Under the Dodd-Frank Act, SEC must adopt rules requiring the NRSROs to
publicly disclose information on the initial credit ratings determined
by each NRSRO for each type of obligor, security, and money market
instruments, and any subsequent changes to such credit ratings.
[Footnote 46] The purpose of these rules is to allow users of credit
ratings to evaluate the accuracy of ratings and compare the
performance of ratings by different NRSROs. At a minimum, these
disclosures must be comparable among NRSROs, clear and informative for
investors having a wide range of sophistication who use or might use
credit ratings, and include information over a range of years and for
a variety of credit ratings, including those credit ratings that the
NRSROs withdraw. In developing these new disclosure requirements, it
will be important for SEC to provide clear and specific guidance to
the NRSROs. Otherwise, the resulting disclosures may lack
comparability.
Although Using Consistent Methodologies to Generate Performance
Statistics Is Helpful, Other Differences Can Make Comparisons among
NRSROs Difficult:
Even if NRSROs use the same methodologies to generate and present
performance statistics, there are differences in NRSROs' measures of
creditworthiness, ratings scales, ratings methodologies, and other
processes. It is important that users of NRSRO performance statistics
be aware of this contextual information when comparing NRSRO
performance.
* NRSROs vary in how they measure creditworthiness. For example, some
NRSROs' credit ratings measure the likelihood of default, while others
also measure other characteristics, such as the anticipated severity
of dollar losses given a default.
* NRSROs also vary in how they define the elements of their ratings
scales. As previously discussed, NRSRO rating scales rank rated
entities according to their relative creditworthiness, with higher
ratings indicating higher creditworthiness. However, the
creditworthiness associated with each rating category can vary across
NRSROs. Even within a rating scale, the assignment of ratings in the
same rating category to issuers and issues may not fully reflect small
differences in the degrees of risk. Moreover, the degree of risk
within a particular rating scale can change over time.[Footnote 47]
* NRSROs can differ in how they determine when to withdraw a rating.
As previously discussed, withdrawals typically occur when an issue
matures, but NRSROs also exercise judgment on whether or not to
withdraw ratings in other cases, such as when they believe they do not
have sufficient information to provide a rating. They also can vary in
the extent to which they track withdrawn ratings. We obtained
information from four NRSROs on their treatment of withdrawn ratings.
Three NRSROs continue to track the issue or issuer after a rating is
withdrawn to determine whether it eventually defaulted. These NRSROs
then update their performance statistics to account for these
defaults. One NRSRO did not. NRSROs that track post-withdrawal
defaults will show a higher default rate than those that do not, all
other things being equal.
* NRSROs can differ in how they define default. Therefore, some
agencies may have higher default rates than others as a result of a
broader set of criteria for determining that a default has occurred.
* Differences in NRSROs' rating methodologies can affect the relative
stability of ratings. For example, in their public disclosures, two
NRSROs stated that they rate through the business cycle, meaning that
their ratings are intended to measure how an issuer will weather a
variety of macroeconomic conditions, relative to other issuers. These
rating agencies, upon receiving new information about the issuer, may
not immediately revise the rating. As a result, ratings that reflect a
through-the-cycle approach are less likely to fluctuate over time.
However, another NRSRO told us that it updates ratings more frequently
to reflect current market information and conditions. NRSROs that use
this approach likely will have transition rates that show more
volatility than the transition rates of NRSROs that rate through the
business cycle.
Users of NRSRO performance statistics can obtain some of this
contextual information from other disclosures NRSROs are required to
make under Form NRSRO.[Footnote 48] As previously discussed, NRSRO
applicants and registered NRSROs are required to disclose general
descriptions of their policies and procedures for determining ratings.
For example, these disclosures discuss each NRSRO's approach to
measuring creditworthiness, identifying defaults, and determining when
to withdraw a rating. As part of their required disclosures of
performance statistics, NRSROs also must describe the rating
categories for their ratings scales. However, these descriptions
define only the rank ordering of the elements of the rating scale, and
do not give any indication of the actual degree of risk associated
with a rating category.
When it proposed rules to require performance disclosures from the
NRSROs, SEC requested comments on whether the performance statistics
should use standardized inputs, time horizons, and metrics to allow
for greater comparability.[Footnote 49] Some commenters opposed the
use of standardized measures, stating that such measures would be
impractical because credit rating agencies use different methodologies
to determine credit ratings and different definitions of default and
that the use of such measures could interfere with the methodologies
for determining credit ratings. However, a few commenters supported
the use of standardized measures because it would make it easier to
compare NRSROs. In light of the different views expressed, SEC stated
it would continue considering this issue to determine the feasibility
and potential benefits and limitations of devising measurements that
would allow reliable comparisons of performance between NRSROs. SEC's
ability to achieve comparability on some of these margins may be
limited, however, given Section 15E prohibits SEC from regulating
credit ratings and the procedures and methodologies used to determine
them.
Required Disclosures of Ratings History Data Have Limited Usefulness
for Generating Comparative Performance Statistics:
In February 2009, SEC adopted a rule requiring issuer-pays rating
agencies to disclose a random 10 percent sample of outstanding ratings
in each class of ratings in which they have more than 500 issuer-paid
ratings.[Footnote 50] In December 2009, SEC issued a second rule
requiring all NRSROs to disclose 100 percent of the histories of their
ratings actions for credit ratings initiated on or after June 26,
2007.[Footnote 51] SEC intended the two rules to be complementary and
allow users to generate a variety of performance measures and
comparative studies. However, we found that the data disclosed under
the 10 percent sample disclosure requirement do not contain enough
information to construct comparable performance statistics and are not
representative of the population of credit ratings at each NRSRO and
that the data disclosed under the 100 percent disclosure requirement
likely present similar issues.
Factors Limiting Utility of 10 percent Samples Include Lack of
Information on Ratings Types and Variables, and Sampling Methodologies:
According to SEC, the 10 percent sample requirement is intended to
foster accountability and comparability--and hence, competition--among
NRSROs. SEC stated in the final rule that market observers should be
able to develop performance statistics based on the data to compare
the rating performance of different NRSROs, which will foster NRSRO
competition. The rule specified that the ratings histories NRSROs must
provide for each security that is part of their sample should include
(1) all ratings actions (initial rating, upgrades, downgrades,
placements on watch for upgrade or downgrade, and withdrawals); (2)
the date of such actions; (3) the name of the security or issuer
rated; and (4) if applicable, the CUSIP number of a rated security or
CIK number of a rated issuer.[Footnote 52] New ratings actions must be
disclosed no later than 6 months after they are taken.
We reviewed the 10 percent samples from the seven issuer-paid NRSROs.
We could not use these samples to generate reliable performance
statistics for the NRSROs, as the rule intended, for the following
reasons: (1) the data fields the NRSROs included in their disclosures
were not always sufficient to identify complete ratings histories for
the rated entities comprising each sample, (2) the data fields did not
always give us enough information to identify specific types of
ratings for making comparisons, (3) the data fields did not always
give us enough information to identify the beginning of the ratings
histories in all of the samples, (4) SEC rules do not require the
NRSROs to publish a codebook or any explanation of the variables used
in the samples, (5) not all NRSROs are disclosing defaults in the
ratings histories provided as part of their 10 percent samples, and
(6) SEC guidance to the NRSROs for generating the random samples does
not ensure that the methods used will create a sample that is
representative of the population of credit ratings produced by each
NRSRO. As a result, users cannot generate valid comparative
performance statistics that can be compared across NRSROs.
First, SEC did not specify the data fields the NRSROs were to disclose
in the rule, and the data fields provided by the NRSROs were not
always sufficient to identify a complete rating history for ratings in
each of the seven samples. If users cannot identify the rating history
for each rating in the sample, they cannot develop performance
measures that track how an issue or issuer's credit rating evolves.
Second, the data fields did not always give users enough information
to identify specific types of ratings for making comparisons. In one
sample, we could not distinguish between issue ratings and issuer
ratings. Distinguishing issuer rating histories from issue rating
histories is important because, as we previously discussed,
performance statistics for some asset classes, such as corporate
issuers, financial institutions, and insurance companies, typically
are calculated using issuer ratings, while performance statistics for
issuers of ABS typically are calculated using issue ratings.
Distinguishing between issuer and issue ratings is important for
evaluating comparable entities across agencies. Comparing the
performance of one agency's issuer ratings with another agency's issue
ratings would not be meaningful. For ABS, one NRSRO told us that its
sample did not have enough information to identify the individual
tranches that constitute a deal.[Footnote 53] Without this ability,
users cannot construct meaningful performance measures for ABS.
Third, the data fields did not always give us enough information to
consistently identify the beginning of the ratings histories in all of
the samples. One NRSRO did not include a variable describing rating
actions, so we could not identify the initial rating in the rating
histories. As a result, we could not calculate transitions or defaults
relative to initial rating for this sample. We also could not
calculate measurements, such as path-to-default or time-to-default,
which depend upon comparing a starting point to the state of a rating
at the time of default. The rating histories in three NRSROs' samples
did not always begin with an initial rating action. Those histories
could not be used to calculate the performance statistics discussed
above for the three NRSROs.
Fourth, SEC rules do not require NRSROs to publish a codebook or any
explanation of the variables used in the samples, and none voluntarily
publish one to accompany its sample data. For several NRSROs, we had
to contact them to obtain an explanation of variables used in the
samples. Without the ability to easily determine what data the
variables represent, users could not begin to construct performance
statistics.
Fifth, not all NRSROs have been disclosing defaults in the ratings
histories provided as part of their 10 percent samples. As previously
discussed, SEC requires that the ratings histories disclosed as part
of the sample include all ratings actions taken. However, not all
NRSROs consider the designation of "default" as a rating action. For
example, one NRSRO does not consider default as part of its rating
scale, so it does not disclose any defaults for any of the ratings
that are part of its sample. Without this information, users cannot
calculate any default statistics or other statistics that incorporate
default rates for this sample.
Sixth, SEC guidance to the NRSROs for generating the random samples
does not help ensure that the methods used will create a sample that
is representative of the population of credit ratings at each NRSRO.
The rule requires NRSROs to generate a random 10 percent sample of the
outstanding ratings in each asset class for which the NRSRO is
registered, but does not specify what kind of ratings to draw the
sample from. Depending on how NRSROs construct their universe of
outstanding credit ratings from which they draw the 10 percent
samples, the samples may not represent similar ratings. For example,
two NRSROs said they draw the 10 percent sample from the total number
of entities that are rated, while two other NRSROs said they draw the
sample from the total number of ratings. A corporate issuer may have a
long-term debt rating and a short-term debt rating. An NRSRO drawing
the sample from the total number of entities rated would count the
corporate issuer as one rated entity. However, an NRSRO drawing the
sample from the total number of ratings would count the ratings on
both the issuer and the issue. Further, where samples include both
issuer and issue ratings for asset classes such as corporate,
financial institutions, or insurance companies, the user may have to
first separate out the issuer ratings in order to calculate
performance statistics. However, the fraction of issuers represented
in the samples varies across asset classes and NRSROs and users do not
know what these fractions are. Where NRSROs construct their universe
of ratings based on the total number of ratings issued, and provide
multiple kinds of ratings, the fraction of the sample that represents
rated issuers is likely to be relatively small. Because NRSROs are not
required to draw the sample from the rating types that are typically
analyzed in each asset class, users may not have enough observations
to generate statistically valid performance measures or develop
comparable measures.[Footnote 54]
Furthermore, NRSROs are not required to redraw the 10 percent samples
periodically. The rule requires that NRSROs re-examine their samples
periodically to make sure that they remain 10 percent of outstanding
ratings. We obtained information from five NRSROs on their methods for
maintaining compliance with the rule. Two NRSROs told us they create a
larger-than-required sample so that over time it is unlikely to dip
below 10 percent of outstanding ratings. The other three NRSROs said
they review the samples periodically to identify those securities that
have matured or been withdrawn and replace them with randomly selected
outstanding issues. These NRSROs do not redraw the samples. However,
in both methodologies, the distribution of some ratings types will
become less representative over time. For example, some ratings mature
and are withdrawn at a faster rate than others, but these may be
replaced with ratings that mature more slowly. If the NRSROs do not
periodically redraw their samples, over time, statistics generated
from the samples will become less representative of the population of
credit ratings.
Most importantly, an NRSRO's 10 percent sample is representative only
of the NRSRO's currently outstanding ratings, a subset of all the
ratings the NRSRO has produced. That is, they do not represent ratings
that have been withdrawn in prior time periods. As previously
discussed, the methodologies used by some NRSROs for constructing
default and transitions rates over time factor in ratings that have
been withdrawn so that the statistics represent the population of
ratings that were in effect during the period studied. Moreover,
because withdrawn ratings may be systematically different from
outstanding ratings, the 10 percent samples may not be representative
of the underlying populations of all ratings the NRSROs have issued.
Thus, historical performance statistics calculated using an NRSRO's
sample may contain biases that are not present in the universe of
ratings that the NRSROs use to compose cohorts in their own studies.
Furthermore, the extent of these biases may differ across NRSROs.
Because the samples do not contain information on withdrawn ratings,
they also do not contain information on the post-withdrawal
performance of such ratings. As previously discussed, some NRSROs
track issuers for evidence of default after their ratings are
withdrawn. They then count these defaults as part of their own
studies. One NRSRO said that defaults on withdrawn ratings account for
about 20 percent of all the defaults it reports in its performance
statistics for corporate issuers, financial institutions, and
insurance companies. Unless NRSROs include withdrawn ratings as part
of their samples, users cannot calculate performance statistics that
are representative of the underlying population.
Exclusion of Many Issuers Limits Utility of 100 Percent Requirement
for Comparative Purposes:
As previously mentioned, SEC adopted a second rule in December 2009
requiring all NRSROs to disclose 100 percent of their ratings actions
histories for credit ratings initiated on or after June 26, 2007. In
the case of issuer-paid credit ratings, each new ratings action must
be disclosed no later than 12 months after it is taken. For ratings
actions that are not issuer-paid, each new ratings action must be
disclosed no later than 24 months after it is taken. SEC stated in the
final rule that the 100 percent requirement will help individual users
of credit ratings design their own performance metrics. SEC also noted
in the final rule that the 10 percent requirement and 100 percent
requirement will provide different types of data sets with which to
analyze and compare the performance of NRSROs' credit ratings.
For example, SEC stated in the final rule that because the 10 percent
sample requirement applies to all outstanding and future credit
ratings in the rule's scope, initially it will provide information
that is much more retrospective and include histories for ratings that
have been outstanding for much longer periods. However, SEC stated
that because the 100 percent requirement is broader in scope, the
disclosure eventually will provide for a more granular comparison of
ratings performance. SEC stated that, unlike the 10 percent
requirement, it will permit users of credit ratings and others to take
a specific debt instrument and compare the ratings history of each
NRSRO that rated it.[Footnote 55] SEC also noted that while the 10
percent sample requirement is limited to issuer-paid credit ratings,
the 100 percent requirement covers all credit ratings, thereby
allowing comparisons of a broader set of NRSROs.
The 100 percent requirement will make a larger amount of data
available to users over time than the 10 percent requirement; however,
several factors may limit the usefulness of the data for generating
meaningful and comparable performance statistics. First, according to
Trading and Markets staff, the rule does not require that these data
include the ratings of any issuer that was rated before June 26, 2007.
Officials from two NRSROs told us their samples thus exclude issuer
ratings on many major American corporations. We searched the data
disclosed by a third NRSRO pursuant to this rule, and could not find
issuer ratings for several issuers that this NRSRO currently rates,
such as the Allstate Corporation, Ford Motor Company, and General
Electric Company. As performance statistics for several asset classes,
including corporate issuers, are based on issuer, not issue ratings,
performance statistics calculated using data that do not include the
ratings of issuers rated prior to June 26, 2007, would not reflect the
overall rating performance of NRSROs and may not be representative of
the universe of issuer ratings. For example, one NRSRO told us that
new issuers, especially new nonfinanical companies, generally are
rated speculative. It said that its ratings history data would include
ratings on only these issuers, and not the older, more established
issuers. Where the data do not contain the types of ratings typically
analyzed for a particular asset class, they will have limited
usefulness for generating performance statistics.
Second, as with the 10 percent samples, data on withdrawn ratings and
any subsequent defaults on withdrawn ratings are not required to be
disclosed. To the extent that withdrawn ratings are not included in
the data, users will not be able to generate withdrawal-adjusted
statistics and the data will underrepresent defaulted issuers and
issues. Finally, SEC required that NRSROs disclose the same ratings
history information for the 100 percent disclosure requirement as for
the 10 percent sample disclosure requirement (ratings action, date of
such actions, the name of the issuer or issue, and the CUSIP or CIK
number), but again, did not specify the data fields NRSROs were to
include in their disclosures. As we discussed, the data fields
provided by the NRSROs in their 10 percent samples were not always
sufficient to ensure that the rating histories had enough information
to allow the user to construct complete ratings histories or identify
specific types of ratings for making comparisons. Without additional
SEC guidance to NRSROs on how to format and describe the data, the 100
percent data sets likely will present challenges similar to those for
10 percent sample for users seeking to construct ratings histories and
develop comparable performance statistics.
NRSROs' Different Methodologies for Counting Total Outstanding Ratings
Limit the Usefulness of These Disclosures:
SEC requires that each NRSRO publicly disclose on its initial
application and annual certification of Form NRSRO the approximate
number of total outstanding ratings by each of the five major asset
classes. In requiring public disclosure of this information, SEC said
that users of credit ratings will find this information useful in
understanding an NRSRO. For example, SEC said it would provide
information to users of credit ratings as to how broad an NRSRO's
coverage is within a particular class of credit ratings.
However, SEC did not specify how the NRSROs were to count their
outstanding ratings. As a result, the NRSROs used diverse
methodologies to count up their outstanding ratings. For example, in
the corporate issuers, financial institutions, insurance company, and
government securities asset classes, some NRSROs counted the number of
issuers rated, others counted the number of ratings on issues (which
could be multiple) and others counted the number of rated issuers and
issues.[Footnote 56] As another example, in the structured finance
asset class some NRSROs counted the number of issuers whose deals they
rated, some counted the number of deals, others counted the number of
tranches underlying each deal, and others counted the number of
ratings on each tranche (which could also be multiple). The NRSROs did
not disclose how they determined their total outstanding ratings, so
users have no way of knowing that these differences exist. Because of
the inconsistencies in how the NRSROs count their total outstanding
ratings, users cannot rely on the disclosures to assess how broad an
NRSRO's coverage is within a particular class of credit ratings.
As SEC Works to Remove NRSRO References from SEC Rules, It Will Need
To Ensure It Has the Staff with the Requisite Skills to Evaluate
Compliance with Any Alternative Creditworthiness Standard:
In July 2008, SEC proposed amendments to multiple rules and forms that
would have removed the references to NRSRO ratings from those rules.
While SEC removed references from six rules and two forms, it retained
the use of the ratings or delayed further action on two rules. These
rules govern money market fund investments and the amount of capital
broker-dealers must hold and use NRSRO references as risk-limiting
measures. We found that OCIE examiners had concerns with these
proposals. For example, in the securities rule regulating money-market
fund investments, SEC proposed to remove NRSRO references, which the
rule used to define the minimum credit quality of the securities a
money market fund could hold, and relying instead solely on the
existing requirement that fund boards independently assess the credit
quality of portfolio securities and determine that each presents
minimal credit risks to the fund. OCIE examiners expressed concerns
that the proposed rule might allow money market funds to invest in
riskier securities than the current rule allows. SEC opted not to
remove references at that time. The recently adopted Dodd-Frank Act
requires SEC and other federal agencies to remove references to NRSRO
ratings from their regulations and substitute an alternative standard
of creditworthiness. Given the Dodd-Frank Act requirements, SEC's
previous experience with proposals to remove credit rating references
highlights the importance of developing a plan to help ensure that (1)
any adopted alternative standards of creditworthiness for a particular
rule facilitate its purpose (e.g., that money market funds invest only
in high-quality securities or that broker-dealers hold sufficient
capital against their investments), and (2) examiners have the
requisite skills to apply the adopted standards. Without such a plan,
SEC may develop alternative standards of creditworthiness that are not
effective in supporting the purpose of a particular rule.
SEC Has Removed References from Multiple SEC Rules and Forms but
Retained Their Use or Delayed Action on Two Rules:
In the past 2 years, SEC has proposed or made changes to regulations
that rely on the use of NRSRO ratings.[Footnote 57] Federal securities
and banking regulations rely on NRSRO ratings for a variety of
purposes. For example, NRSRO ratings are components of the definition
of a mortgage-related security and establish criteria for eligibility
for certain types of securities registration. According to a recent
Joint Forum survey, U.S. federal banking and securities statutes,
legislation, regulations, and guidance contain 81 references to NRSRO
ratings, 45 of which are in SEC regulations or guidance and 36 in the
statutes, regulations or guidance of various banking regulators.
[Footnote 58]
In particular, SEC has proposed removing references to NRSRO ratings
from Investment Company Act rule 2a-7, which contains provisions that
limit the types of securities a money market fund can hold, and from
Exchange Act rule 15c3-1 (the "net capital rule"), which includes
provisions to designate the capital that broker-dealers must hold
against their assets.[Footnote 59] Rule 2a-7 governs the operations of
money market funds. Unlike most other investment companies, money
market funds seek to maintain a stable share price, typically $1.00
per share.[Footnote 60] The Investment Company Act and applicable
rules generally require investment companies to calculate current net
asset value per share by valuing portfolio instruments at market value
or, if market quotations are not readily available, at fair value as
determined in good faith by, or under the direction of, the board of
directors. Rule 2a-7 exempts money market funds from these provisions,
but contains conditions on the investments of the fund such as
maturity, quality, liquidity, and diversification, which are designed
to minimize the deviation between a fund's stabilized share price and
the market value of its portfolio. If the deviation becomes
significant, the fund may be required to take certain steps to address
the deviation, including selling and redeeming its shares at less than
$1.00 (breaking the buck).[Footnote 61] Among these risk-limiting
conditions, rule 2a-7 limits a money market fund's portfolio
investments to eligible securities. Under rule 2a-7, eligible
securities are those securities that have received a credit rating
from the "requisite NRSROs"[Footnote 62] in one of the two highest,
short-term rating categories or are comparable unrated securities.
[Footnote 63] Rule 2a-7 further restricts money market funds to
holding securities that the fund's board of directors (or those on
whom they rely) determines present minimal credit risks. This second
requirement specifically requires that the determination "be based on
factors pertaining to credit quality in addition to any rating
assigned to such securities by an NRSRO."
The net capital rule requires broker-dealers to maintain, at all
times, a minimum amount of net capital and uses NRSRO ratings as a
third-party assessment of credit risk in prescribing the level of
capital required to be held. The rule was adopted to create uniform
capital requirements for and help ensure the liquidity of all
registered broker-dealers. In computing net capital, broker-dealers
must deduct from their net worth certain percentages of the market
value of their proprietary securities positions.[Footnote 64] The
deductions are known as haircuts and serve to provide a margin of
safety against losses broker-dealers might incur as a result of market
fluctuations in the prices of, or lack of liquidity in, their
proprietary positions. SEC allows broker-dealers to apply reduced
haircuts for certain types of securities they hold that at least two
NRSROs rate as investment-grade because these securities typically are
more liquid and less volatile in price than securities not so highly
rated. That is, the more highly rated the security, the more it counts
toward the total amount of capital the broker-dealers are required to
hold. In addition to NRSRO ratings, the net capital rule uses measures
such as position concentration, maturity, and type of security to
determine appropriate haircuts.
SEC proposed removing references to ratings in rule 2a-7 and the net
capital rule in July 2008.[Footnote 65] Among other reasons, SEC
proposed these amendments to address the risk that the references to
and use of NRSRO ratings in SEC rules could be interpreted by
investors as an endorsement of the quality of the rating and might
encourage investors to place undue reliance on them. For rule 2a-7,
SEC proposed eliminating the requirement that portfolio securities
have a certain NRSRO rating (or be a comparable unrated security),
while retaining the requirement that portfolio securities be limited
to those that the fund's board of directors determines present minimal
credit risks. The proposal also would have specifically required the
board's determination to be based on factors pertaining to credit
quality and the issuer's ability to meet its short-term financial
obligations.[Footnote 66]
The proposal would have eliminated the requirement that money market
funds restrict themselves to investing in securities highly rated by
NRSROs (or comparable unrated securities), and instead relied on the
existing requirement that the fund's board of directors determine that
the securities present minimal credit risks. Under the proposal, fund
boards could have continued to use quality determinations prepared by
outside sources, including NRSRO ratings, if they concluded these
ratings were credible for making credit risk determinations. In the
rule proposal, SEC stated it expected that boards of directors (or
their designees) would understand the basis for the rating and make an
independent judgment of credit risk. In February 2010, SEC adopted
amendments to rule 2a-7, which continues to use NRSRO ratings in
defining eligible securities. The amendments require money market fund
boards to designate, at least once each calendar year, at least four
NRSROs, the credit ratings of which the boards deem to be sufficiently
reliable for use by their funds to comply with rule 2a-7's eligible
security requirements.[Footnote 67]
As proposed in July 2008, the revisions to the net capital rule would
substitute two new standards for the current NRSRO ratings-based
categories. For determining haircuts on commercial paper, SEC proposed
to replace the top tiers of ratings-based categories with a
requirement that the instrument be subject to a minimal amount of
credit risk and have sufficient liquidity so that it could be sold at
or near its carrying value almost immediately.[Footnote 68] For
determining haircuts on nonconvertible debt securities, SEC proposed a
requirement that the instrument be subject to "no greater than
moderate" credit risk and have sufficient liquidity so that it could
be sold at or near its carrying value in a reasonably short time.
[Footnote 69] According to SEC, the proposed standards are meant to
serve the same purpose as the prior standards. Thus, securities with
"no greater than moderate" credit risk would encompass all so-called
investment-grade securities. SEC believes broker-dealers have the
financial sophistication and the resources necessary to make the basic
determination of whether or not a security meets the requirements in
the proposed amendments and distinguish between securities subject to
minimal credit risk and those subject to moderate credit risk. Under
the proposal, broker-dealers would have to be able to explain how the
securities they used for net capital purposes met the standards in the
proposed amendments. However, SEC stated it would be appropriate, as
one means of complying with the proposed amendments, for broker-
dealers to refer to NRSRO ratings for the purposes of determining
haircuts under the rule. SEC decided to delay any action on this
proposal and as of June 2010, continued to solicit comments.[Footnote
70]
In October 2009, SEC adopted amendments to six rules and two forms
that removed the references to NRSRO ratings made in these rules.
[Footnote 71] Four of these rules and the two forms originally were
adopted in 1998 as part of SEC's new framework for regulation of
exchanges and alternative trading systems and utilized "investment-
grade" and "non-investment-grade" to distinguish between classes of
securities. The adopted amendments and changes to forms eliminated the
distinction between classes of securities and the use of "investment
grade" and "non-investment-grade." The remaining two rules utilize the
terms "highest rating category from an NRSRO" and "investment-grade
rating from at least one NRSRO" to define securities exempted from
specific requirements or define a class of securities eligible for
purchase by funds when the security's principal underwriter had
certain relationships with the fund or its investment adviser. In both
cases, the adopted amendments remove the references to ratings and
either remove the exemption or redefine the class of eligible
securities.[Footnote 72]
Developing a Plan to Address the Implications of the Adopted
Alternative Standards May Help SEC Ensure It Has the Skills and
Resources Necessary to Evaluate Compliance with the Standards:
The proposed changes to rules 2a-7 and 15c3-1 would have eliminated
the bright-line creditworthiness standard OCIE examiners used to
determine that money market funds invested in high quality securities
or the appropriateness of the haircut a broker-dealer took for net
capital purposes on a security. We reviewed OCIE's 2a-7 examination
module and 65 OCIE money market fund examinations (for FY 2003-2009)
identified as having 2a-7 deficiencies to understand how OCIE
examiners assess compliance with the rule's requirements for
determining an "eligible security" and minimal credit risks and how
the removal of NRSRO references would affect SEC's ability to oversee
a fund's exposure to credit risks.
As stated above, rule 2a-7 limits money market funds investments to
those securities that are rated in one of the two highest short-term
categories by an NRSRO or comparable unrated securities and that the
fund's board determines present minimal credit risks for the fund.
OCIE examiners examine money market funds for compliance with this
provision by reviewing the NRSRO ratings at the time of purchase for
securities held. Examiners typically identify if securities held by a
money market fund are eligible securities by requesting a list of all
portfolio holdings, including the current NRSRO rating for each
holding, and verify the NRSRO ratings by reviewing the published
ratings on Bloomberg or on the NRSRO Web site.
OCIE examiners then typically review a fund's compliance and
procedures manuals to help ensure that the board has established
minimal credit risk guidelines and receives periodic credit risk
updates and reports from the adviser verifying that all the securities
comply. Examiners further request and review a small sample of credit
analysis packages that demonstrate that the securities are eligible
and a sample of the materials presented to the fund's security
evaluation committee as evidence of ongoing reviews that a security
continues to present minimal credit risks. According to OCIE
examiners, policies, procedures, and practices for conducting minimal
credit risk analysis vary widely.
Of the 65 examinations of money market funds OCIE completed in FY 2003-
2009 that we reviewed, 36 examinations identified 42 deficiencies in
the funds' compliance with the requirement for a minimal credit risk
determination.[Footnote 73] They generally could be categorized as
deficiencies in fund board oversight, policies and procedures, or
credit file documentation. According to OCIE examiners, citing funds
for a deficiency in documenting its analysis of minimal credit risk in
an examination is not unusual. For example, in one examination
deficiency letter OCIE found no current written documentation in the
credit files substantiating that the fund adequately determined that
each security purchased presented minimal credit risks and requested
that the fund bring its files up-to-date. According to Enforcement
staff, SEC has not brought any enforcement actions against a money
market fund for violations of this requirement.
OCIE examiners expressed concerns with the proposed rule because they
believed it might allow money market funds to invest in riskier
securities than the current rule allows. Under the proposed rule, a
money market fund could invest in any security it finds to present a
minimal credit risk. OCIE examiners stated they would have likely
continued to evaluate for compliance with the minimal credit risk
determination requirement as they do under the current rule. As such,
OCIE only examines a fund's policies and procedures to assess if they
effectively address credit risk and to assess whether a fund follows
its policies and procedures in making credit risk determinations. It
does not evaluate the standards used to determine whether a security
is deemed to represent a minimal credit risk, dictate the types of
analyses that must be included in a minimal credit risk determination,
or make any of its own determinations as to whether the security
represents a minimum credit risk to that fund[Footnote 74]. According
to Investment Management staff, the minimum credit risk requirement
was not designed for these purposes as the rule recognizes that funds
can have different investment objectives and positions, and as such,
the same security could present different risks to different funds.
One fund might consider a particular security an appropriate
investment, while another would not. OCIE examiners stated that the
proposed rule eliminated the floor, in terms of creditworthiness, that
NRSRO references provided and it was unclear how, if at all, the
standard for eligible securities under the proposed rule would ensure
that money market funds continued to invest only in securities of the
highest credit quality.
Further, if OCIE examiners were given the authority to evaluate funds'
credit risk determinations, OCIE staff told us that additional
resources and skill sets would be needed to conduct such examinations
and they questioned OCIE's ability to evaluate the credit risk
determinations.[Footnote 75] To date, examiners have not needed to
have these skills because examiners, as dictated by the rules and
interpretations, relied on NRSRO ratings. OCIE examiners told us that
as proposed, they likely would approach compliance examinations by
continuing to focus on ensuring that funds had reasonable policies and
procedures in place for determining what constituted an eligible
security and documentation demonstrating that those policies and
procedures were followed and an analysis of credit risk completed.
The proposal to remove NRSRO references from the net capital rule also
would eliminate the credit-risk criteria OCIE examiners currently use,
among other factors, to determine whether a broker-dealer was taking
appropriate haircuts. Under the current rule, broker-dealers use a
variety of factors, including whether the security is rated investment
grade, to determine the haircut they must take on debt securities when
determining their net worth for regulatory capital purposes.[Footnote
76] OCIE examiners generally confirm the net capital calculation by
reviewing and confirming a firm's inventory, selecting a sample of
securities with which to verify the existence of a ready market, and
verifying that the haircuts were accurate and considered in the net
capital computation. Under the proposed rule, broker-dealers would be
responsible for determining the level of risk a security presented and
the amount of the subsequent haircut, which could be different for
each broker-dealer, depending on the methods used.
Going forward, the Dodd-Frank Act requires SEC to remove NRSRO ratings
from its rules. SEC's previous experience with proposals to remove
credit rating references highlights the importance of developing a
plan to help ensure that (1) any adopted alternative standards of
creditworthiness for a particular rule facilitate its purpose (e.g.,
that money market funds invest only in high-quality securities or that
broker-dealers hold sufficient capital against their investments), and
(2) examiners have the requisite skills to determine that the adopted
standards have been applied. Without such a plan, SEC may develop
alternative standards of creditworthiness that are not effective in
supporting the purpose of a particular rule.
The Number of NRSROs Has Increased since the Act Was Implemented but
Industry Concentration Remains High:
Since the implementation of the Act, the number of NRSROs has
increased from 7 to 10. However, the market remains highly
concentrated. Continued concentration is likely a result of multiple
factors. First, relatively little time has passed since SEC
implemented the NRSRO registration program and NRSRO rulemaking.
Second, credit rating agencies face barriers in entering the credit
rating industry and registering as an NRSRO. Academic research
suggests that increasing competition among NRSROs improves information
availability but the impact on ratings quality is unclear.
The Number of NRSROs Has Increased from 7 to 10, but the Industry
Remains Concentrated:
Since the implementation of the Act, the number of NRSROs has
increased from 7 to 10; however, the market remains highly
concentrated. As previously discussed, 7 credit rating agencies had
received SEC staff no-action letters recognizing them as NRSROs prior
to the Act. When the NRSRO registration program became effective,
these firms applied to register as NRSROs and received SEC
approval.[Footnote 77] All of these operate primarily under an issuer-
pays business model. SEC also granted NRSRO registration to 3
additional credit rating agencies that operate primarily under a
subscriber-pays business model.[Footnote 78]
Figure 1 indicates when the 10 NRSROs began producing credit ratings
and the year that SEC first recognized them as NRSROs, either through
the no-action letter process or the NRSRO registration program.
Figure 1: Years the Current NRSROs Have Produced Credit Ratings and
Have Been Recognized as NRSROs:
[Refer to PDF for image: timeline]
A.M. Best Company:
Years credit rating agency has produced credit ratings: 1898-2010;
Years SEC has recognized credit rating agency as an NRSRO: 2005-2010.
Moody’s Investors Service[A]:
Years credit rating agency has produced credit ratings: 1909-2010;
Years SEC has recognized credit rating agency as an NRSRO: 1975-2010.
Fitch Ratings[B]:
Years credit rating agency has produced credit ratings: 1924-2010;
Years SEC has recognized credit rating agency as an NRSRO: 1975-2010.
Standard & Poor’s Ratings Services[C]:
Years credit rating agency has produced credit ratings: 1922-2010;
Years SEC has recognized credit rating agency as an NRSRO: 1975-2010.
DBRS:
Years credit rating agency has produced credit ratings: 1976-2010;
Years SEC has recognized credit rating agency as an NRSRO: 2003-2010.
LACE Financial:
Years credit rating agency has produced credit ratings: 1984-2010;
Years SEC has recognized credit rating agency as an NRSRO: 2008-2010.
Japan Credit Rating Agency:
Years credit rating agency has produced credit ratings: 1985-2010;
Years SEC has recognized credit rating agency as an NRSRO: 2007-2010.
Rating and Investment Information:
Years credit rating agency has produced credit ratings: 1985-2010;
Years SEC has recognized credit rating agency as an NRSRO: 2007-2010.
Egan-Jones Rating Company:
Years credit rating agency has produced credit ratings: 1993-2010;
Years SEC has recognized credit rating agency as an NRSRO: 2007-2010.
Realpoint:
Years credit rating agency has produced credit ratings: 2001-2010;
Years SEC has recognized credit rating agency as an NRSRO: 2008-2010.
Source: NRSROs, SEC.
[A] Moody's Investors Service was founded as John Moody & Company,
which began producing credit ratings in 1909.
[B] Fitch Ratings was founded as the Fitch Publishing Company, which
began producing credit ratings in 1924.
[C] Poor's Publishing and Standard Statistics began producing credit
ratings in 1922 and 1923, respectively. Standard Statistics merged
with Poor's Publishing forming Standard & Poor's Corporation.
[End of figure]
None of the 10 NRSROs, including the 3 newly registered subscriber-
pays NRSROs, is a new entrant into the credit rating industry.
Further, all 10 NRSROs have been producing ratings for a number of
years. A.M. Best, Fitch, Moody's, and Standard & Poor's have been
producing ratings the longest--for more than 80 years. Several of
these NRSROs have undergone mergers with or acquisitions of other
rating agencies or NRSROs over the years. For example, Poor's
Publishing and Standard Statistics merged in 1941 to form Standard &
Poor's, and Moody's was acquired by Dun and Bradstreet in 1962. Fitch
merged with IBCA Ltd in 1997, and in April 2000, acquired Duff &
Phelps Credit Rating Company and Thomson BankWatch.[Footnote 79] More
recently, in May 2010 Realpoint was acquired by Morningstar,
Inc.[Footnote 80]
Credit rating agencies can apply to register as NRSROs in five
distinct asset classes: financial institutions, insurance companies,
corporate, ABS, and government securities. Table 5 describes the asset
classes in which each NRSRO is registered.
Table 5: NRSROs Registered by Asset Class, 2010:
A.M. Best:
Financial institutions: [Check];
Insurance companies: [Check];
Corporate issuers: [Check];
ABS: [Check];
Government municipal and sovereign securities: [Empty].
DBRS:
Financial institutions: [Check];
Insurance companies: [Check];
Corporate issuers: [Check];
ABS: [Check];
Government municipal and sovereign securities: [Check].
Egan-Jones Ratings:
Financial institutions: [Check];
Insurance companies: [Check];
Corporate issuers: [Check];
ABS: [Check];
Government municipal and sovereign securities: [Check].
Fitch Ratings:
Financial institutions: [Check];
Insurance companies: [Check];
Corporate issuers: [Check];
ABS: [Check];
Government municipal and sovereign securities: [Check].
Japan Credit Rating Agency:
Financial institutions: [Check];
Insurance companies: [Check];
Corporate issuers: [Check];
ABS: [Check];
Government municipal and sovereign securities: [Check].
LACE:
Financial institutions: [Check];
Insurance companies: [Check];
Corporate issuers: [Check];
ABS: [Empty];
Government municipal and sovereign securities: [Check].
Moody's Investors Service:
Financial institutions: [Check];
Insurance companies: [Check];
Corporate issuers: [Check];
ABS: [Check];
Government municipal and sovereign securities: [Check].
Ratings and Investment, Inc.:
Financial institutions: [Check];
Insurance companies: [Check];
Corporate issuers: [Check];
ABS: [Empty];
Government municipal and sovereign securities: [Check].
Realpoint:
Financial institutions: [Empty];
Insurance companies: [Empty];
Corporate issuers: [Empty];
ABS: [Check];
Government municipal and sovereign securities: [Empty].
Standard & Poor's:
Financial institutions: [Check];
Insurance companies: [Check];
Corporate issuers: [Check];
ABS: [Check];
Government municipal and sovereign securities: [Check].
Source: 2009 SEC Form NRSRO filings.
[End of table]
Some NRSROs, such as Moody's, Standard & Poor's, and Fitch, cover a
wide range of securities that span all five asset classes. Others have
specialized in a particular asset class, sector, or geographic region.
For example, although Realpoint is designated in ABS, it specializes
in one type of ABS, specifically CMBS. LACE is designated in four
asset classes, but specializes in rating financial institutions.
Similarly, A.M. Best is designated in four asset classes, but
specializes in rating insurance companies and related securities.
[Footnote 81] Japan Credit Rating Agency and Ratings and Investment,
Inc., are Japanese rating agencies that mainly rate Japanese issuers.
Although the number of NRSROs has increased, the credit rating
industry remains highly concentrated. To assess the impact of the Act
on competition among NRSROs, we calculated the HHI, a key statistical
measure used to assess market concentration and the potential for
firms to exercise their ability to influence market prices.[Footnote
82] The HHI reflects the number of firms in the industry and each
firm's market share. It is calculated by summing the squares of the
market shares of each firm competing in the market. The HHI also
reflects the distribution of market shares of the top firms and the
composition of the market outside the top firms. The HHI is measured
on a scale of 0 to 10,000, with larger values indicating more
concentration. According to DOJ, markets in which the value of the HHI
is between 1,500 and 2,500 points are considered to be moderately
concentrated, and those in which the value of the HHI is in excess of
2,500 points are considered to be highly concentrated, although other
factors also play a role.
We calculated the HHI by summing the squares of the market shares of
all the firms competing in the industry. Doing so requires defining
what constitutes the industry and specifying our measure of market
share. We defined the relevant industry as the set of credit rating
agencies that have NRSRO status, and we used a variety of market share
definitions to ensure that any trends in industry concentration we
observed were robust to alternative specifications of NRSROs' market
shares. A firm's market share typically is measured in terms of
dollars, as either its sales or revenue as a fraction of total sales
or revenue for all firms in the industry, or in terms of quantities,
such as its output as a fraction of total output produced by all firms
in the industry.
We first calculated the HHI using market shares based on total
revenues earned by the NRSROs.[Footnote 83] NRSROs generally earn
revenues from a number of activities related to the production of
credit ratings. Issuer-pays NRSROs earn the bulk of their revenues
from the fees paid by issuers to have their issues rated. However,
issuer-pays NRSROs offer other services as well, including
subscription services to users of credit ratings. Subscriber-pays
NRSROs earn their revenues from subscription fees and other services.
Some of the types of services offered by the subscriber-pays NRSROs
are data and valuation and proxy services for financial institutions.
[Footnote 84]
NRSRO applicants and registered NRSROs must provide data on the total
revenues earned in the prior calendar year to SEC on Form NRSRO.
[Footnote 85] We used these data to calculate the HHI from 2006 to
2009.[Footnote 86] Table 6 provides the results of these calculations.
Table 6: HHI for NRSROs Based on Total Revenues, 2006-2009:
All asset classes;
2006: 3,617;
2007: 3,511;
2008: 3,333;
2009: 3,324.
Annual Percentage Change;
2006: [Empty];
2007: -2.93%;
2008: -5.08%;
2009: -0.27%.
Source: GAO analysis of NRSRO revenues provided on Form NRSRO.
[End of table]
The table shows that while the HHI declined between 2006 and 2009, the
industry remains highly concentrated according to DOJ standards. This
decline is likely influenced by the entrance of the three new NRSROs
in late 2007.
An NRSRO's total revenue does not necessarily reflect its total
output; that is, the number of ratings it produces. For example, both
issuer-pays and subscriber-pays NRSROs could provide ratings on the
same group of entities, but receive vastly different revenues. Because
market shares based on numbers of ratings can differ from those based
on total revenue, so can the HHI, possibly revealing a different trend
in industry concentration. To assess industry concentration using an
output-based measure of market share, we attempted to calculate the
HHI using market shares based on the number of each NRSRO's
outstanding ratings.[Footnote 87] However, as previously discussed, we
found inconsistencies in the methods that the NRSROs use to count
their outstanding ratings. As such, the data were not valid for
purposes of calculating the HHI.
As an alternative assessment of industry concentration using an output-
based measure of market share, we calculated the HHI using market
shares based on the number of issuers each NRSRO rates (see table
7).[Footnote 88] We obtained data from nine of the NRSROs on the
number of issuers they rated in each asset class in 2006-2009 and used
it to calculate the HHI for these 4 years. We were unable to obtain
data on the number of rated issuers from one NRSRO because it said it
did not track rated organizations by whether or not they issue debt
securities. However, this NRSRO did provide us with the total number
of organizations it rated.
Table 7: HHI for NRSROs Based on Number of Issuers Rated, 2006-2009:
Asset Class: Corporate issuers;
2006: 3,069;
2007: 2,625;
2008: 2,596;
2009: 2,483.
Asset Class: Financial institutions;
2006: 2,773;
2007: 2,555;
2008: 2,550;
2009: 2,452.
Asset Class: Insurance companies;
2006: 3,353;
2007: 3,066;
2008: 2,826;
2009: 2,749.
Asset Class: Issuers of government securities;
2006: 3,822;
2007: 3,820;
2008: 3,846;
2009: 3,889.
Asset Class: Issuers of asset-backed securities;
2006: 3,602;
2007: 3,561;
2008: 3,553;
2009: 3,493.
Source: GAO analysis of NRSRO data.
Note: Calculations are based on data from nine of the ten currently
registered NRSROs.
[End of table]
The table shows that the industry is concentrated in every asset class
in every year according to DOJ standards, although the industry has
become less concentrated in corporate issuers, financial institutions,
insurance companies, and issuers of ABS asset classes with the HHIs
decreasing by 5 percent, 4 percent, 10 percent, and 2 percent,
respectively, between 2007 and 2009. The industry has become more
concentrated in the issuers of government securities asset class, with
the HHI increasing by about 2 percent between 2007 and 2009. These
results, however, assume that none of the organizations rated by the
10th NRSRO issues debt securities.
To assess the sensitivity of these results to the missing data from
the NRSRO that did not track which of its rated organizations issue
debt securities, we recalculated the HHIs assuming that all of the
organizations this NRSRO reported rating issue debt securities. We did
so because it is likely that some of the organizations this NRSRO
rates do issue debt securities, but we cannot determine how many.
Calculating the HHIs based on the alternative assumption that all of
the organizations this NRSRO rates issue debt securities gives us a
range within which the true value of the HHI is likely to fall.
The main difference between our alternative and baseline estimates is
in the HHI for the financial institutions asset class. The alternative
assumption produces estimates of the HHI for the financial
institutions asset class for 2008 and 2009 that are 133 percent and
136 percent, respectively, larger than our baseline estimates. The two
estimates differ because the NRSRO that did not provide us with data
rates at least 10 times as many organizations in the financial
institutions asset class as any other NRSRO. Assuming that all of
these organizations issue debt securities produces high HHI estimates
because it implies that the excluded NRSRO has a relatively high
market share and thus that the industry is relatively highly
concentrated.[Footnote 89]
Finally, to assess trends in concentration in the market for rating
structured finance securities, we calculated the HHI for January 2004-
June 2010 using market shares based on the dollar value of issuance of
U.S.-issued ABS rated by an NRSRO.[Footnote 90] Issuance-based HHI
declined by about 18 percent over this time period (1 percent during
2004-2007 and 17 percent since 2007), indicating that this market has
become less concentrated (see table 8). We note that the market for
ABS declined considerably since 2007. According to data from Asset-
Backed Alert, the number of ABS deals declined from over 3,000 in 2006
to about 370 in 2009. For 2010, 223 deals were reported as of the end
of June.
Table 8: HHI Based on Dollar Value of Newly Issued U.S.-ABS, January
2004-June 2010:
Asset Class: All U.S. asset-backed securities;
2004: 3,444;
2005: 3,375;
2006: 3,469;
2007: 3,398;
2008: 3,396;
2009: 2,973;
2010[A]: 2,809.
Asset Class: U.S. commercial mortgage-backed securities;
2004: 3,224;
2005: 3,222;
2006: 3,359;
2007: 3,212;
2008: 3,751;
2009: 2,916;
2010[A]: 2,804.
Asset Class: U.S. traditional asset-backed securities;
2004: 3,374;
2005: 3,338;
2006: 3,314;
2007: 3,280;
2008: 3,305;
2009: 3,262;
2010[A]: 3,046.
Asset Class: U.S. prime residential mortgage-backed securities;
2004: 3,677;
2005: 3,672;
2006: 3,542;
2007: 3,376;
2008: 3,148;
2009: 3,222;
2010[A]: 4,145.
Asset Class: U.S. nonprime residential mortgage-backed securities;
2004: 3,390;
2005: 3,177;
2006: 3,344;
2007: 3,515;
2008: 3,531;
2009: 10,000[B];
2010[A]: 6,009.
Asset Class: U.S. Collateralized Debt Obligations;
2004: 3,772;
2005: 3,944;
2006: 4,173;
2007: 4,253;
2008: 4,846;
2009: 3,795;
2010[A]: 5,561.
Source: GAO analysis of Asset -Backed Alert data.
[A] The HHIs for 2010 are based on data through June 30, 2010.
[B] Only one deal was issued in 2009 and was rated by a single NRSRO.
[End of table]
Since the U.S. ABS asset class includes distinct products, we also
examined five sectors in the ABS asset class to determine if trends in
market concentration varied across these sectors (table 8). The five
sectors are traditional ABS (that is, securities backed by student
loans, auto loans and credit card loans, but not by mortgages), prime
RMBS, nonprime RMBS, CMBS, and CDO.[Footnote 91] We calculated the HHI
for the market for rating securities in each of these sectors using
market shares based on the dollar value of issuance rated by an NRSRO.
The 17 percent reduction in the HHI for the market for rating
securities in the ABS asset class as a whole since 2007 was driven
primarily by the reduction in concentration in the market for rating
traditional ABS and CMBS. For these markets, the issuance-based HHI
declined by about 7 percent and 13 percent, respectively, since 2007.
On the other hand, the markets for rating prime RMBS, and CDOs have
become more concentrated since 2007. While the market for rating
nonprime RMBS also has become more concentrated since 2007, the number
of issuances offered since then has declined so rapidly that trends in
the HHI are difficult to interpret.[Footnote 92]
The HHI indicates that the market for rating ABS remains highly
concentrated by DOJ standards, even in those sectors in which
concentration has declined since 2007. To assess which NRSROs are
dominating this market, we examined the NRSROs' annual market
coverage. Annual market coverage is an indication of the quantity of
ratings an NRSRO produces relative to the quantity of issues or
issuers that are available to be rated. Because more than one NRSRO
can rate an issue or issuer, the sum of each NRSRO's annual market
coverage can add to more or less than 100 percent. We measure an
NRSRO's annual market coverage as the dollar volume an NRSRO rates as
a fraction of the total volume issued. We did not assess the causal
factors behind any trends we observed.
Trends in annual market coverage from January 2004 through June 2010
among the six NRSROs that rated U.S. ABS issuance generally shifted
beginning in 2007 (see figure 2).[Footnote 93] These six were the only
NRSROs to rate new U.S. ABS issuance during this period. From 2004-
2007, Standard & Poor's, Moody's, and Fitch provided the most annual
market coverage in this asset class, rating an average of about 94
percent, 89 percent, and 49 percent, respectively, of issuance.
Starting in 2008, Standard & Poor's and Moody's annual market coverage
began to decline. For the first half of 2010, Standard & Poor's and
Moody's were rating only about 73 percent and 62 percent,
respectively, of issuance. Fitch's annual market coverage peaked in
2009 at about 59 percent of issuance, and then declined to about 31
percent for the first half of 2010. On the other hand, DBRS increased
its coverage from about 4 percent in 2007 and 2008 to about 33 percent
in June 2010.[Footnote 94]
Figure 2: NRSRO U.S. Annual Market Coverage by ABS, Dollar Volume,
2004-June 2010:
[Refer to PDF for image: multiple line graph]
Percentage of total issuance:
Year: 2004;
Total number of deals: 493;
S&P: 94%;
Moody's: 88.1%;
Fitch: 50.1%;
DBRS: 3.4%;
Realpoint: 0%;
A.M. Best: 0%.
Year: 2005;
Total number of deals: 597;
S&P: 93.7%;
Moody's: 89.6%;
Fitch: 48.4%;
DBRS: 6.8%;
Realpoint: 0%;
A.M. Best: 0%.
Year: 2006;
Total number of deals: 819;
S&P: 93.5%;
Moody's: 91.9%;
Fitch: 46.7%;
DBRS: 4.3%;
Realpoint: 0%;
A.M. Best: 0%.
Year: 2007;
Total number of deals: 648;
S&P: 94.5%;
Moody's: 86.2%;
Fitch: 51.6%;
DBRS: 4.4%;
Realpoint: 0%;
A.M. Best: 0%.
Year: 2008;
Total number of deals: 198;
S&P: 84.6%;
Moody's: 79.7%;
Fitch: 45.3%;
DBRS: 4.5%;
Realpoint: 0%;
A.M. Best: 0.2%.
Year: 2009;
Total number of deals: 178;
S&P: 67.8%;
Moody's: 58.4%;
Fitch: 56.2%;
DBRS: 12.5%;
Realpoint: 0.2%;
A.M. Best: 0%.
Year: 2010;
Total number of deals: 119;
S&P: 72.6%;
Moody's: 61.9%;
Fitch: 31.3%;
DBRS: 33.3%;
Realpoint: 0.5%;
A.M. Best: 0%.
Source: GAO analysis of ABA data.
[End of figure]
We also examined NRSROs' coverage of the five sub-sectors in the ABS
asset class and found that trends in annual market coverage varied
across sectors. Trends in coverage of traditional ABS are similar to
those for the ABS asset class as a whole, the main difference being
that Fitch's coverage did not peak in 2009 before declining (see
figure 3).
Figure 3: NRSRO U.S. Annual Market Coverage by Traditional ABS, Dollar
Volume, 2004-June 2010:
[Refer to PDF for image: multiple line graph]
Percentage of total issuance:
Year: 2004;
Total number of deals: 493;
S&P: 97.4%;
Moody's: 93.5%;
Fitch: 74.2%;
DBRS: 0.1%;
A.M. Best: 0%.
Year: 2005;
Total number of deals: 597;
S&P: 98.3%;
Moody's: 96.7%;
Fitch: 71.9%;
DBRS: 2.3%;
A.M. Best: 0%.
Year: 2006;
Total number of deals: 819;
S&P: 95.7%;
Moody's: 93%;
Fitch: 75.6%;
DBRS: 2.1%;
A.M. Best: 0%.
Year: 2007;
Total number of deals: 648;
S&P: 96%;
Moody's: 95%;
Fitch: 74.4%;
DBRS: 3.8%;
A.M. Best: 0.1%.
Year: 2008;
Total number of deals: 198;
S&P: 93.4%;
Moody's: 97.4%;
Fitch: 71.6%;
DBRS: 3.1%;
A.M. Best: 0.3%.
Year: 2009;
Total number of deals: 178;
S&P: 82.6%;
Moody's: 81.9%;
Fitch: 60.6%;
DBRS: 4.7%;
A.M. Best: 0%.
Year: 2010;
Total number of deals: 119;
S&P: 76.6%;
Moody's: 81.5%;
Fitch: 40%;
DBRS: 20.2%;
A.M. Best: 0%.
Source: GAO analysis of ABA data.
[End of figure]
In prime RMBS, Moody's market coverage began to decline in 2007 from
about 88 percent to about 6 percent in June 2010. Standard & Poor's
annual market coverage of prime RMBS began to decline in 2008, falling
from about 94 percent in 2007 to about 42 percent in 2009, but it has
since increased to about 63 percent (see figure 4). Fitch's market
share peaked in 2009 at about 62 percent, but has since declined to
about 9 percent.
Figure 4: NRSRO U.S. Annual Market Coverage by Prime RMBS, Dollar
Volume, 2004-June 2010:
[Refer to PDF for image: multiple line graph]
Percentage of total issuance:
Year: 2004;
Total number of deals: 574;
S&P: 89.7%;
Moody's: 81.9%;
Fitch: 36.6%;
DBRS: 1%.
Year: 2005;
Total number of deals: 752;
S&P: 90%;
Moody's: 84.5%;
Fitch: 30.5%;
DBRS: 5%.
Year: 2006;
Total number of deals: 713;
S&P: 89.4%;
Moody's: 87.5%;
Fitch: 42.8%;
DBRS: 2.5%.
Year: 2007;
Total number of deals: 584;
S&P: 94%;
Moody's: 74.2%;
Fitch: 57.9%;
DBRS: 2.9%.
Year: 2008;
Total number of deals: 83;
S&P: 84.1%;
Moody's: 40.5%;
Fitch: 45%;
DBRS: 17.8%.
Year: 2009;
Total number of deals: 126;
S&P: 42%;
Moody's: 5.5%;
Fitch: 61.5%;
DBRS: 41.3%.
Year: 2010;
Total number of deals: 60.
S&P: 68.3%;
Moody's: 6.4%;
Fitch: 8.5%;
DBRS: 73.4%.
Source: GAO analysis of ABA data.
[End of figure]
Trends in coverage of the nonprime RMBS market are the most dramatic,
with both Standard & Poor's and Moody's coverage plummeting from more
than 90 percent to zero for 2009 and 2010, and Fitch's coverage
falling from about 57 percent for 2005 to about 1 percent for 2008
(see figure 5). On the other hand, DBRS coverage of nonprime RMBS
increased from about 14 percent for 2005, and 12 percent in 2006-2007
to about 73 percent for 2008. Furthermore, DBRS was the lead NRSRO to
rate the most nonprime RMBS deals issued in 2009 and the first half of
2010, rating three more than Moody's.
Figure 5: NRSRO U.S. Annual Market Coverage by Nonprime RMBS, Dollar
Volume, 2004-June 2010:
[Refer to PDF for image: multiple line graph]
Percentage of total issuance:
Year: 2004;
Total number of deals: 608;
S&P: 99.1%;
Moody's: 93.7%;
Fitch: 50.4%;
DBRS: 6.8%.
Year: 2005;
Total number of deals: 621;
S&P: 97.2%;
Moody's: 95%;
Fitch: 56.5%;
DBRS: 13.6%.
Year: 2006;
Total number of deals: 654;
S&P: 99.2%;
Moody's: 99.1%;
Fitch: 45.4%;
DBRS: 12%.
Year: 2007;
Total number of deals: 344;
S&P: 98.4%;
Moody's: 94.3%;
Fitch: 31.9%;
DBRS: 12.4%.
Year: 2008;
Total number of deals: 13;
S&P: 94.5%;
Moody's: 47.7%;
Fitch: 1.2%;
DBRS: 73.4%.
Year: 2009;
Total number of deals: 1;
S&P: 0%;
Moody's: 0%;
Fitch: 0%;
DBRS: 100%.
Year: 2010;
Total number of deals: 3;
S&P: 0%;
Moody's: 38%;
Fitch: 0%;
DBRS: 100%.
Source: GAO analysis of ABA data.
[End of figure]
Declines in Standard & Poor's, Moody's, and Fitch's coverage of the
CMBS market through 2009 were almost as dramatic as those in the
nonprime RMBS market, but they have not been matched by
correspondingly dramatic increases in DBRS's coverage of the nonprime
RMBS market (see figure 6). Rather, Realpoint's CMBS coverage has
increased from virtually zero through 2008 to 19 percent in 2010.
[Footnote 95] In addition, Standard & Poor's, Moody's, and Fitch's
coverage of the CMBS market have all rebounded somewhat in 2010, with
Fitch's annual market coverage in 2010 about equal to its annual
market coverage in 2005 (about 56 percent).
Figure 6: NRSRO U.S. Annual Market Coverage by CMBS, Dollar Volume,
2004-June 2010:
[Refer to PDF for image: multiple line graph]
Percentage of total issuance:
Year: 2004;
Total number of deals: 117;
S&P: 77.7%;
Moody's: 72%;
Fitch: 46.8%;
DBRS: 8%;
Realpoint: 0%.
Year: 2005;
Total number of deals: 125;
S&P: 82%;
Moody's: 75.2%;
Fitch: 56%;
DBRS: 6.6%;
Realpoint: 0%.
Year: 2006;
Total number of deals: 130;
S&P: 79.1%;
Moody's: 71.9%;
Fitch: 57.1%;
DBRS: 1%;
Realpoint: 0%.
Year: 2007;
Total number of deals: 103;
S&P: 85.8%;
Moody's: 72.9%;
Fitch: 67.5%;
DBRS: 5.6%;
Realpoint: 0%.
Year: 2008;
Total number of deals: 24;
S&P: 69.4%;
Moody's: 63.4%;
Fitch: 26.8%;
DBRS: 0%;
Realpoint: 0%.
Year: 2009;
Total number of deals: 51;
S&P: 16.3%;
Moody's: 11.8%;
Fitch: 12.7%;
DBRS: 0.5%;
Realpoint: 3.2%.
Year: 2010;
Total number of deals: 8;
S&P: 39.3%;
Moody's: 52.3%;
Fitch: 56.4%;
DBRS: 0%;
Realpoint: 19%.
Source: GAO analysis of ABA data.
[End of figure]
Finally, in the CDO market, Moody's coverage fell from about 97
percent in 2007 to about 39 percent in 2009 (see figure 7).
Figure 7: NRSRO U.S. Annual Market Coverage by CDO, Dollar Volume,
2004-June 2010:
[Refer to PDF for image: multiple line graph]
Percentage of total issuance:
Year: 2004;
Total number of deals: 270;
S&P: 94.7%;
Moody's: 86.3%;
Fitch: 35.3%;
DBRS: 0.1%.
Year: 2005;
Total number of deals: 428;
S&P: 97.5%;
Moody's: 89.8%;
Fitch: 28.2%;
DBRS: 0.3%.
Year: 2006;
Total number of deals: 767;
S&P: 97.2%;
Moody's: 98.4%;
Fitch: 20.9%;
DBRS: 0.1%.
Year: 2007;
Total number of deals: 529;
S&P: 97.4%;
Moody's: 97.3%;
Fitch: 18.3%;
DBRS: 0%.
Year: 2008;
Total number of deals: 98;
S&P: 71.2%;
Moody's: 63.4%;
Fitch: 2%;
DBRS: 0.3%.
Year: 2009;
Total number of deals: 12;
S&P: 58.9%;
Moody's: 39%;
Fitch: 22.4%;
DBRS: 0%.
Year: 2010;
Total number of deals: 33;
S&P: 49.8%;
Moody's: 100%;
Fitch: 0%;
DBRS: 0%.
Source: GAO analysis of ABA data.
[End of figure]
Moody's CDO coverage was about the same as Standard & Poor's in 2007,
but has declined steadily since then to about 40 percent in 2009. It
has since increased to 100 percent. Fitch's CDO coverage declined from
about 35 percent in 2004 to zero in 2010 (with a brief increase in
2009 covering about 22 percent of the market). DBRS's CDO coverage has
remained negligible throughout the period.
In December 2009, SEC adopted rule amendments that are intended, in
part, to increase competition in the rating of ABS.[Footnote 96]
Beginning in June 2010, the amended rule requires an NRSRO hired by
arrangers to rate a structured finance product to disclose on its
password-protected Website each structured finance product it has been
hired to rate, along with the type of structured finance product, the
name of the issuer, the date the rating process began, and the Web
site at which the issuer will disclose the information it has provided
to the NRSRO for the rating. The amended rule requires the arranger to
provide representations to the hired NRSRO that it will make available
the information it has provided to the hired NRSRO for determining an
initial rating or for monitoring a rating on the issuer's password-
protected Web site. The issuer must also provide representations to
the hired NRSRO that it will allow other NRSROs access to the
information so that the other NRSROs can produce unsolicited ratings
on the same structured finance product. SEC proposed these rule
amendments after its examinations of the three largest NRSROs
identified issues in the management of conflicts of interest
particular to structured finance ABS. In particular, SEC found that
analysts appeared to be aware of NRSROs' business interest in securing
the rating of the deal and that rating agencies did not appear to take
steps to prevent considerations of market share and other business
interests from influencing ratings or rating criteria. In the proposed
rule amendments, SEC stated it believed that the issuer-pays conflict
is particularly acute in the case of structured finance products
because certain arrangers of structured finance products bring repeat
business to NRSROs. As such, SEC believes that some arrangers have the
potential to influence NRSROs on structured finance products more than
on corporate securities.
In the amended rule, SEC stated that one of its goals is to facilitate
the issuance of credit ratings for structured finance products by
nonhired NRSROs at the same time as the hired NRSRO and provide
investors with more views on the creditworthiness of the structured
finance product. SEC stated this practice may serve to increase
unsolicited ratings for structured finance products, mitigate ratings
shopping, and affect competition among NRSROs by having more ratings
in the market. Furthermore, SEC stated that market participants could
use unsolicited ratings to evaluate the ratings issued by the hired
rating agency. Specifically, SEC intends that by opening up the
ratings process to more NRSROs, hired NRSROs will find it easier to
resist any pressure by the arranger to obtain better-than-warranted
ratings, because of the likelihood that any steps taken to
inappropriately favor the arranger could be exposed to the market
through credit ratings issued by other NRSROs.
Although not enough time has passed to assess the impact of the
amended rule on competition, we found that one NRSRO has withdrawn its
NRSRO registration in this asset class and the other NRSROs had
varying views on its potential effectiveness. One NRSRO said that the
high cost of establishing and maintaining the data systems could
negatively impact both NRSROs and issuers. Further, this NRSRO said
the amended rule could deter issuers from taking innovative structured
finance products to the market because they would have to disclose
proprietary information. Another NRSRO said the cost of implementing
the rule could burden nonhired NRSROs, who may need to post only very
limited information under the rule, and thus compromise the
effectiveness of the rule. In May 2010, one NRSRO announced it was
withdrawing its NRSRO registration in this asset class.[Footnote 97]
This NRSRO said it made this decision in part because it believed that
the funds raising activities through structured finance products in
the geographic region it serves might be negatively affected. A fourth
NRSRO agreed that publishing unsolicited ratings could enhance the
transparency of the ratings process, but said that rating new
structured finance issuances was too costly without fees from an
issuer. For example, this NRSRO said that its costs were approximately
$2,500 to verify the data underlying the security in an ABS that it
rates. Additionally, this NRSRO said that legal costs for analyzing
the securities in a deal ranged from $25,000 to $50,000. However,
another NRSRO was not concerned with the costs associated with
implementing the rule and believed the new rule could be effective.
Multiple Factors Likely Account for Continued Concentration among
NRSROs:
The continued concentration among NRSROs since the implementation of
the Act likely resulted from several factors. First, little time has
passed since the Act took effect. Second, the three new NRSROs
registered under the new program have not had much time to build
market share. Furthermore, SEC rules implementing the new NRSRO
registration program and requiring disclosures of ratings performance,
ratings methodologies, and conflicts of interest have been in place
since June 2007, and have been amended twice. Finally, the credit
crunch and the ensuing financial crisis occurred soon after the
implementation of the Act, substantially slowing certain sectors of
the credit market.[Footnote 98]
Generally speaking, barriers to becoming an NRSRO create challenges
for newer and smaller credit rating agencies. Two types of barriers to
entry likely contributed to the continued concentration of the
industry: entering the credit rating industry and registering as an
NRSRO.
We have identified three barriers to entering the credit rating
industry. First, credit rating agencies may have relatively high fixed
costs.[Footnote 99] Credit rating agencies that are established can
produce ratings in volume leading to economies of scale. The
combination of high fixed costs and economies of scale favor larger
established rating agencies and pose barriers to smaller firms
entering the market. And, the markets for some asset classes may be
more difficult for rating agencies to enter than others. For example,
a credit rating agency told us it is difficult for a credit rating
agency to get into the credit rating market for the structured finance
class because of costs associated with acquiring the expertise to rate
this type of product. Further, rating methodologies for structured
finance products are complex, requiring expertise to develop and apply
the models needed to rate this type of product.
Second, establishing a reputation takes time.[Footnote 100] The better
a rating agency's reputation for producing ratings, the more business
it will be able to attract as compared with a credit rating agency
without a reputation.[Footnote 101] However, given the nature of the
ratings, reputation can take years to establish. In most cases,
ratings are intended to predict the likelihood of default over the
life of the bond, but some securities take from 10 to 30 years to
mature. When a new rating agency begins to rate securities, evaluating
the quality of those ratings at the time of purchase is difficult.
Instead, users need to see how the ratings perform over the life of
the bond to determine how accurate and timely they are. Several NRSROs
have commented that their reputation is critical to their success. One
NRSRO said that a reputation is difficult to earn and easy to lose.
Third, network effects pose a challenge to entry in the ratings
industry and favor the larger, more established credit rating
agencies.[Footnote 102] The more securities a specific NRSRO rates,
the more value to assigning ratings to that same NRSRO, because
comparing securities rated by that NRSRO would be easier for investors
and other market participants. Network effects can make gaining market
share difficult for new entrants if investors and other market
participants already are using an existing NRSRO's ratings. NRSRO
references have been widely embedded in numerous federal and state
laws and regulations.[Footnote 103] Further, many investment
guidelines and private contracts reference specific NRSROs, which
makes marketing to investors, other users, or issuers more difficult
for newer NRSROs. Several institutional investors and investment
advisors with whom we spoke told us they use the big three NRSROs'
ratings either because of investor guidelines, regulatory guidelines,
or depth of ratings coverage. Although there is no limit on the number
of NRSROs that can rate a particular issue, asset managers may face
budgetary constraints that limit their ability to subscribe to NRSROs
beyond those their investment agreements require them to
consider.[Footnote 104] For example, one asset manager told us that
they have a limited budget for subscriptions to ratings and that
purchasing subscriptions for each analyst in the credit research
department is costly. They told us they have subscription services
with four NRSROs but have a limited working relationship with the
others. Three issuers with whom we spoke told us that their choice of
NRSRO is driven by investor expectations, which directs them to the
big three firms.
A credit rating agency also encounters barriers to entry when
registering as an NRSRO. Despite the efficiency and transparency of
the new NRSRO registration program, compliance with the Act and SEC
rules may result in higher costs for smaller NRSROs and may inhibit
credit rating agencies from registering as NRSROs. For example, one
small NRSRO estimated it spent $500,000 annually to maintain the NRSRO
designation. Two rating agencies with which we spoke said they would
not register because of the regulatory burden associated with being in
compliance with the Act. Furthermore, one NRSRO told us it might de-
register as an NRSRO should regulation became too costly.
Besides barriers to entry, differences between NRSROs' compensation
models and a degree of specialization may also contribute to market
concentration. For example, subscriber-pays NRSROs may be limited in
their ability to rate newly issued securities. One NRSRO explained
that it uses the subscriber-pays model to provide ongoing surveillance
of rated securities in the secondary market, which it produces using
publicly available information. However, it said it could not use the
subscriber-pays model to rate the initial offering of securities
because, under this model, it would not have access to the data
provided by the issuer to complete the analysis and produce an initial
ratings. To the extent the subscriber-pays NRSROs are not rating new
issues, market coverage of these securities will be concentrated among
issuer-pays NRSROs. As another example, the difference between the
issuer-pays and the subscriber-pays compensation models could also
impact market concentration when it is measured in terms of total
revenues. As previously discussed, both issuers-pays and subscriber-
pays NRSROs could provide ratings on the same group of entities, but
receive vastly different revenues. Issuer-pays NRSROs charge the
issuers fees for every rating produced, while subscriber-pays NRSROs
are charging users a subscription fee for access to their ratings.
Thus, if issuer-pays and subscriber-pays rate the same entities, total
revenue will likely be concentrated among NRSROs using the
compensation model that generates the greatest revenues per rated
entity. Finally, to the extent certain NRSROs specialize in a
particular asset class, sector, or geographic, the overall credit
rating industry will likely be highly concentrated among those NRSROs,
which rate across asset classes, sectors, and geographic regions.
However, a specialized NRSRO could have a significant presence in its
market.
Academic Research Suggests Increasing Competition in the Credit Rating
Industry Improves Information Availability, but the Impact on Rating
Quality Is Unclear:
The impact of increasing competition on the quality of credit ratings
is not yet well understood. Academic researchers generally measure the
quality of credit ratings according to how much information they
convey about the risk of default or of loss in the event of default.
Their findings suggest that the entry of new credit rating agencies
can improve overall information available to investors and other
market participants. However, the effect of entry on the quality of
ratings produced by any one rating agency is not clear. Moreover,
there have been few studies investigating the effect of new entrants
and competition in the credit rating industry. These studies are
unpublished and, thus, their findings should be viewed as preliminary
in nature.
We reviewed three studies that examined the impact of competition on
ratings quality.[Footnote 105] Based on an analysis of insurance
company ratings, one study suggests that entry of a new credit rating
agency improves the amount of information available to investors and
other market participants. This study suggests that new entrants have
stricter criteria than incumbents for assigning the same rating,
assuming they both use the same rating scale. It is more difficult for
an issuer to get the highest rating from the new rating agency than
from the incumbent rating agency. An issuer can choose to be rated by
the incumbent rating agency, by the new rating agency, or both.
Furthermore, the two ratings agencies' criteria are different. As a
result, an issuer can communicate more information about its riskiness
to the market by its choice of ratings agency and the combination of
ratings it gets than it could communicate when there was only one
rating agency.
A different study of CDO ratings also suggests that the number of
rating agencies from which an issuer requests ratings is informative.
Specifically, tranches rated by more than one rating agency were less
likely to be downgraded than those rated by a single rating agency.
This result is consistent with the hypothesis that issuers of less-
risky CDOs were more likely to request two or more ratings.
Based on ratings of corporate issuers, insurance companies, and
financial institutions, a third study analyzes the impact of
competitive pressure from a new credit rating agency on the quality of
an incumbent rating agency's ratings. The study uses three alternative
indicators of quality. The first indicator is the correlation between
a bond's rating and its yield, with lower correlations indicating that
ratings are less informative about bond repayment and thus are of
lower quality. The second indicator is the magnitude of the effect of
a downgrade on an issuer's stock price, with larger magnitudes
indicating that the downgrade is worse news.[Footnote 106] The last
indicator is the rating it assigns to an issuer or a bond, with higher
ratings presumed to be more favorable to the issuer and thus of lower
quality. This study suggests that the incumbent credit rating agencies
produce lower-quality ratings in market segments in which smaller,
newer credit rating agencies have higher market share.
Together, the three studies' findings have implications for the amount
of information available to credit market participants and for the
quality of credit ratings and may offer some preliminary observations
about the impact of new entrants on rating quality and competition.
The first and second studies both suggest that entry of new credit
rating agencies will allow issuers and other rated entities to
communicate more information to the market, both by the numbers of
ratings they request and by the combination of ratings they receive
from different rating agencies. The findings of the second and third
studies together seem to suggest that entry of new credit rating
agencies will lead incumbent rating agencies to produce lower-quality
ratings, either relative to the new entrant's ratings or relative to
their own ratings in markets in which they face less competitive
pressure. However, it is difficult to predict what the effect will be
on the incumbent rating agencies' ratings when a new entrant enters
the market.
Models Proposing Alternative Means of Compensating NRSROs Intend to
Address Conflicts of Interests in the Issuer-Pays Model:
As part of an April 2009 roundtable held to examine oversight of
credit rating agencies, SEC requested perspectives from users of
ratings and others on whether it should consider additional rules to
better align the raters' interest with those who rely on those
ratings, and specifically, whether one business model represented a
better way of managing conflicts of interest than another. In
response, some roundtable participants proposed alternative models for
compensating NRSROs, and market observers have proposed others in
congressional hearings and academic literature. We identified five
unique models that have been proposed, although they are in various
stages of development. To assist Congress and others in assessing
these proposals, we created an evaluative framework of seven factors
that any compensation model should address to be effective. By
applying these factors, users of the framework can identify the
potential benefits of the model consistent with policymakers' goals as
well as any tradeoffs.
Proposed Alternative Compensation Models:
In recent years, academic researchers and industry experts have begun
to develop a number of alternative compensation models for credit
rating agencies in response to concerns about conflict of interest,
ratings integrity, and competition. In a July 2008 report discussing
the examinations of the three most active NRSROs and their performance
of in rating subprime RMBS and related CDOs, SEC staff identified
issues in the management of conflicts of interest resulting from the
issuer-pays model the firms used. NRSROs using this model have an
interest in generating business from the firms that seek the rating,
which could conflict with providing quality ratings. In response to
the examination findings, SEC introduced new and amended rules
intended to improve the management of conflicts of interest in the
issuer-pays model.[Footnote 107] In April 2009, as part of the
roundtable held to examine oversight of credit rating agencies, SEC
requested perspectives from users of ratings and others on whether SEC
should consider additional rules to better align the NRSROs' interest
with those who rely on those ratings, and specifically, whether one
form of model represented a better way of managing conflicts of
interest than another.
In response, some roundtable participants proposed alternative models
for compensating NRSROs. These models generally intend to address the
conflict of interest in the issuer-pays model, better align the
NRSROs' interest with users of ratings, or improve the incentives
NRSROs have to produce reliable and high-quality ratings. Other models
with similar goals have been presented in Congressional hearings and
in academic literature. Below, we provide a summary of the key
provisions of five distinct alternative models for compensating NRSROs
(alternative compensation models). Given their theoretical nature,
they vary greatly in the amount of detail currently available. None of
these models has been implemented to date.
Five alternative compensation models have been proposed: random
selection model, investor-owned credit rating agency model, stand-
alone model, designation model, and user-pay model.[Footnote 108]
Random Selection Model:
Under the random selection model, a ratings clearinghouse randomly
would select a credit rating agency to rate a new issuance.[Footnote
109] All issuers or sponsors that wanted to obtain ratings for their
issuances would be required to request ratings from the clearinghouse,
which would use a random number generator, such as a computerized
algorithm, to assign a credit rating agency. The clearinghouse would
notify the credit rating agency of the opportunity to rate the
issuance and provide basic information pertaining to the type of
issuance, but not the issuer's name. Not until the credit rating
agency agreed to complete the rating would the clearinghouse disclose
to the credit rating agency the identity of the issuer and the details
of the issuance. If the selected credit rating agency agreed to rate
the issuance, the issuer would pay a fee to the ratings clearinghouse.
The clearinghouse then would distribute the fees to the credit rating
agency upon the completion of the initial and maintenance ratings. The
letter rating would be free of charge to the public. In addition to
this function another primary role of the clearinghouse would be to
design the criteria by which new entrants could qualify as a credit
rating agency.
According to the proposed model, the ratings clearinghouse could be a
nonprofit organization, a governmental agency such as SEC, or a
private-public partnership. Funding for this ratings clearinghouse
would be paid for by the issuer, on top of that required to rate the
security, to cover clearinghouse costs. The ratings clearinghouse also
would be responsible for setting the ratings fees for the credit
rating agency depending on the type of security issued.
The proposal incorporates a peer comparison review to create an
incentive for credit rating agencies to produce quality ratings. As
part of this review, the ratings clearinghouse would evaluate the
performance of all credit rating agencies on the basis of two
empirical tests. As one potential test, the proposal suggests an
analysis of the magnitude of debt instruments that default or lose
substantial value for investment-grade debt instruments. If the
default percentage for a given credit rating agencies differed from
its peers by a set parameter, then it would be subject to sanctions,
which would range from losing a percentage of business to losing a
percentage of rating fees. The second potential test would evaluate
annual yields, as set by the market, to be compared to identically
rated debt securities from different asset classes for each credit
rating agency. Securities in different asset classes that are rated
similarly should have the same yield. If a threshold differential
exists between the yields of identically-rated securities for a credit
rating agency, then it would be subject to sanctions.
According to the architect of the model, this model would eliminate
the conflict of interest when an issuer pays a credit rating agency
for a rating by making the compensation neutral, eliminating the
linkage between the credit rating agency and the issuer (the conflict
of interest). The elimination of the conflict of interest would remove
a barrier to entry and would allow for new competition. Furthermore,
he believes that the peer comparison review coupled with economic
sanctions for poor performance would motivate the credit rating
agencies to continually adjust their rating models and produce quality
ratings.
Investor-Owned Credit Rating Agency Model:
Under the investor-owned credit rating agency (IOCRA) model,
sophisticated investors--termed highly sophisticated institutional
purchasers (HSIP)--would create and operate an NRSRO that would
produce ratings. Issuers would be required to obtain two ratings; one
from the IOCRA and the second from their choice of NRSRO. More
specifically, an NRSRO would be prohibited from publicly releasing a
rating that was paid for by the issuer or sponsor, unless the NRSRO
received written notification that the issuer had made arrangements
and paid an IOCRA to publicly release its rating. The IOCRA would
publish its rating simultaneously when the NRSRO published its rating.
Institutional investors would have to qualify as an HSIP before
forming an IOCRA or joining an existing one. To qualify as an HSIP, an
institutional investor would have to demonstrate that it was large and
sophisticated, managed billions of dollars in assets, and could be
relied upon to represent the buy-side interest in accurately rating
debt market instruments. The HSIPs would hold a majority voting and
operational control over the IOCRA. The proposal contemplates that the
IOCRA could be a for-profit or a not-for-profit entity. There would
not be a regulatory limit on the number of IOCRAs that could be formed.
Under the proposal, market forces would set IOCRA fees, which likely
would be comparable to fees currently charged by the dominant NRSROs.
The letter rating and the underlying research would be free to the
public.
Proponents of this model believe it would improve the rating process
by changing the incentive structure of the NRSROs' business. They said
the IOCRA would affect competition and ratings quality by introducing
new competition to the industry, and the investors' interest would be
counter-balanced against the interest of the issuers.
Stand-Alone Model:
Under the stand-alone model, NRSROs only would be permitted to produce
credit ratings. The NRSROs would be able to interact with and advise
organizations being rated, but could not charge fees for providing
advice.[Footnote 110] Instead of receiving issuer fees, the NRSROs
would be compensated through transaction fees imposed on original
issuance and on secondary market transactions. Part of the fee would
be paid by the issuer or secondary-market seller, and the other
portion of the fee by the investor purchasing the security in either
the primary or secondary market. The NRSRO would be compensated over
the life of the security based on these transaction fees. The letter
rating would be free to the public.
Proponents of this model believe that by creating a funding source
that is beyond the influence of both issuers and investors, the focus
of the NRSRO will be on producing the most accurate and timely credit
analysis rather than on satisfying the desires of any other vested
interest.
Designation Model:
Under the designation model, all NRSROs would have the option of
rating a new issuance, and security holders would direct, or
designate, fees to the NRSROs of their choice, based on the proportion
of securities that they owned. When an issuer decided to bring a
security to market, it would be required to provide all interested
NRSROs with the information to rate the issuance. The issuer would pay
the rating fees to a third-party administrator, which would manage the
designation process.[Footnote 111] When the security was issued, the
security holders would designate which of the NRSROs that rated the
security should receive fees, based on their perception of research
underlying the ratings. The security holders could designate one or
several NRSROs. The third-party administrator would be responsible for
disbursing the fees to the NRSROs in accordance with the security
holders' designations. After the initial rating, the issuer would
continue to pay maintenance rating fees to the third-party
administrator, which bond holders also would allocate through the
designation process every quarter over the life of the security. When
the debt was repaid (or repurchased by the issuer), a final rating fee
would be paid in conjunction with the retirement of the security. The
letter rating would be free to the public, while the research
underlying it would be distributed to securityholders and (at the
discretion of the relevant NRSROs) to potential securityholders.
The proposed model suggests that the issuer's transfer agent could
perform the responsibilities of the third-party administrator. The
transfer agent currently is responsible for maintaining ownership
records of the security holders.
The authors of this model believe this model would eliminate the
conflict of interest between the issuers paying for the rating and the
NRSRO and would increase competition by encouraging NRSROs to prepare
unsolicited ratings, because each NRSRO would be assured of receiving
compensation for its rating, provided some group of investors or other
users of ratings found them useful enough to allocate to the provider
a portion of the fees they designated or paid.
User-Pay Model:
Under the user-pay model, issuers would not pay for ratings. Rather,
to address the free-rider problem, the model specifies that all users
of ratings would be required to enter into a contract with the NRSRO
and pay for the rating services of an NRSRO. The proposal defines
"user" as any entity that included a rated security, loan, or contract
as an element of its assets or liabilities as recorded in an audited
financial statement. Users of ratings would include holders of long or
short positions in a fixed-income instrument, as well as parties that
refer to a credit rating in contractual commitments (that is, as
parties to a lease) or that are parties to derivative products that
rely on rated securities or entities. A user would be required to pay
for ratings services supplied during each period in which it booked
the related asset or liability.
The model relies on third-party auditors to ensure that NRSROs receive
payment from users of ratings for their services. Any entity that
required audited financial statements in which the rated instrument or
covenant was included among the assets or liabilities would be
required to demonstrate to the auditors that the holder had paid for
the rating services. No audit opinion would be issued until the
auditor was satisfied that the rating agencies had been properly
compensated. The model would require the close cooperation of the
auditing community and the Public Company Auditing Oversight Board.
The architects of this model believe that, while more cumbersome, the
model attempts to capture "free riders"--those users of ratings that
do not compensate NRSROs for the use of their intellectual property
and require them to pay for the ratings.
Framework to Evaluate Alternative Compensation Models:
In this report, we are not evaluating the proposed alternative
compensation models. Instead, we are providing a framework that
Congress and others can use to evaluate or craft alternative
compensation models for NRSROs. The framework contains seven factors,
all of which are essential for a compensation model to be fully
effective. Furthermore, we have provided key questions under each
factor that can be applied to an alternative compensation model to
identify its relative strengths and weaknesses, potential trade offs
(in terms of policy goals), or areas in which further elaboration or
clarification would be warranted. Similarly, the framework could be
used to further develop proposals or identify aspects of current
regulations to make them more effective and appropriate for addressing
the limitations of the current credit rating system.
1. Independence. The ability for the compensation model to mitigate
conflicts of interest inherent between the entity paying for the
rating and the NRSRO.
* What potential conflicts of interest exist in the alternative
compensation model?
* What controls, if any, would need to be implemented to mitigate
these conflicts? How does the compensation model seek to limit
conflicts of interest between the entity paying for the ratings and
the NRSRO? Between users of ratings and the NRSRO? Between issuers and
the NRSRO?
As previously discussed, conflicts of interest arise between the
entity paying for the rating and the NRSRO. The alternative
compensation models we have identified continue to rely on issuer fees
to fund ratings and to help ensure that ratings remain free to the
public. However, several intend to mitigate the potential for the
issuer to influence NRSROs in different ways--either by increasing the
investor's role in the rating process or assigning NRSROs using a
rotational process or randomly. In assessing these as well as other
potential compensation models, it is important to consider whether the
models introduce any new conflicts of interest and evaluate the steps
the models propose to mitigate them.
2. Accountability. The ability of the compensation model to promote
NRSRO responsibility for the accuracy and timeliness of their ratings.
* How does the compensation model create economic incentives for
NRSROs to produce quality ratings over the life of a bond?
* How is NRSRO performance evaluated and by whom? For example, does
the compensation model rely on market forces or third parties to
evaluate performance? For models that rely on third parties, how are
"quality" credit ratings defined and what criteria would be used to
assess ratings performance?
* When an NRSRO demonstrates poor performance, what are the economic
consequences under the compensation model and who determines these
consequences? For example, how is an NRSRO's compensation or
opportunity for future ratings business linked to ratings performance?
The quality of an NRSRO's ratings largely is determined by the ratings
methodologies it employs, but the compensation model also can affect
the ratings process. An effective compensation model will provide
economic incentives for the rating agency to produce not only a
quality rating at issuance, but also appropriate surveillance of the
security over its life. It also should link NRSRO compensation to the
performance of the rated entity. As such, when evaluating various
compensation models it is important to consider how NRSRO performance
is evaluated and by whom. NRSRO performance could be evaluated by
market participants or an independent arbiter, with the consequences
of performance dictated by the compensation model. For example, models
can rely on market discipline to evaluate NRSRO ratings and determine
which ones merit future business. Models also could rely on third
parties to evaluate NRSRO performance which might require that the
third party develop performance measures. However, as we previously
discussed, there are differences in the NRSROs' measures of
creditworthiness, ratings scales, ratings methodologies, and other
processes that can make comparison of NRSRO performance difficult when
using these measures.
3. Competition. The extent to which the compensation model creates an
environment in which NRSROs compete for customers by producing higher-
quality ratings at competitive prices.
* On which dimensions does the compensation model encourage NRSROs to
compete? To what extent does the compensation model encourage
competition around the quality of ratings? Ratings fees? Product
innovation?
* To what extent would the compensation model encourage new entrants
and reduce barriers to entry in the industry?
* To what extent does the model allow for flexibility in the differing
sizes, resources, and specialties of NRSROs?
* To what extent do market forces determine ratings fees?
When evaluating an alternative compensation model, considering its
potential impact on the competition in the ratings industry is
important. Most importantly, the model should not in itself present a
barrier to entry or increase existing barriers. It should not promote
convergence of one class of products or methodologies by NRSROs, but
should foster diversity in ratings methodologies and products. For
example, the compensation model should be flexible to allow for
relatively smaller NRSROs or NRSROs with specialties to adapt to any
new requirements and not inadvertently hinder them from competing with
the larger NRSROs or expanding their product lines to meet market
demand.
An effective compensation model also will promote competition around
the quality of ratings. In that sense, this factor is closely related
to the accountability factor, in that the compensation model should
not economically reward NRSROs that consistently produce poor-quality
ratings. Some compensation models could increase competition by
reducing barriers to entry for smaller or newer NRSROs; for example,
by offering or guaranteeing them more opportunity to produce ratings
and increase their coverage of the market. However, it is unclear
whether a model that increases the number of NRSROs would result in
more competition among them to produce quality ratings over time. In
assessing these models, considering their potential impact on NRSROs'
incentives to compete around ratings quality, product innovation, and
overall efficiency is important.
Similarly, an effective compensation model will promote competition
around ratings fees. Some NRSROs are highly specialized, serving
particular markets or asset classes. As such, the ratings fees charged
by each NRSRO reflects its own cost structure. Compensation models
should not incorporate a uniform approach to setting ratings fees.
Such an approach would promote inefficiencies in the market and
dissuade some NRSROs from continuing to offer services if they
believed they were economically disadvantaged. Those NRSROs with
comparatively lower-cost structures for producing ratings might
benefit from such an approach, but overall it would not encourage
NRSROs to produce ratings cost effectively.
4. Transparency. The accessibility, usability, and clarity of the
compensation model and the dissemination of information on the model
to market participants.[Footnote 112]
* How clear are the mechanics of the compensation model to market
participants? How transparent are the following procedures and
processes:
- how the NRSROs obtain ratings business;
- how ratings fees are determined;
- how NRSROs are compensated; and:
- how the compensation model links ratings performance to NRSRO
compensation.
An effective compensation model should be transparent to market
participants to help them understand it and to increase market
acceptance. For example, the model should be transparent about how
NRSROs obtain ratings business, such as (1) whether issuers will
select the NRSROs; (2) whether ratings business will be assigned,
randomly awarded, or mandated; or (3) whether NRSROs will have the
option to provide ratings on any new business. If the model relies on
third parties or systems for this function, it should be explicit
about the criteria or procedures employed. Similarly, if ratings fees
are not determined by market forces, the model should clearly explain
the process and criteria for determining fees.[Footnote 113] Issuers,
NRSROs, oversight bodies, and users of ratings also should understand
the proposed compensation mechanism, and it should clearly link
ratings performance to NRSRO compensation. Any criteria for evaluating
ratings performance and the process for determining these criteria
should be disclosed. Lack of transparency in any of these areas could
hinder support and trust in the model.
5. Feasibility. The simplicity and ease with which the compensation
model can be implemented in the securities market.
* Is the model easily implemented? If not, how difficult will
implementing the model be?
* Could the compensation model be instituted through existing
regulatory or statutory authority or are additional authorities needed?
* What are the costs to implement the compensation model and who would
fund them?
* Which body would administer the compensation model, and is this an
established body? If not, how would it be created?
* What, if any, infrastructure would be needed to implement the
compensation model? What information technology would be required?
Which body would be responsible for developing and maintaining it?
* What impact would the alternative compensation model have on
bringing new issuances to market and trading on the secondary market?
* How many NRSROs would be required for the compensation model to
function as intended? How would the exit of an NRSRO from the ratings
industry affect the model's feasibility? What impact would the
alternative compensation model have on the financial viability of an
NRSRO?
When assessing a compensation model, considering the model's
feasibility for successful implementation is essential. We note that
the market itself has not undertaken the implementation of any
compensation model other than the current issuer-and subscriber-pays
models. As such, SEC or Congress likely would have to direct market
participants to implement any alternative model.[Footnote 114] Models
that are technically simplistic in nature will be more feasible to
implement than complex ones. Further, some alternative models we
identified involve potentially significant costs. For example, some
models would require the development of information technology systems
that would be used across the market by potentially thousands of
participants. Assessing not only the costs of implementing these
models, but also determining who would be responsible for overseeing
and paying for their development upfront is a key question. Costly
models could deter market participants from implementing and
participating in them.
Assessing the impact of any potential model on the efficiency of the
securitization market is an important part of the evaluation process.
An alternative compensation model should be flexible and adaptable to
real-time demands of the securities markets, and not hinder the timing
of initial issuance and trading on the secondary market. The model
also should be viable regardless of the number of NRSROs that enter or
exit the market.
For compensation models that require a third-party administrator, the
process for selecting or creating this administrator could have a
significant impact on the success of its implementation. If market
participants question the independence or capability of the
administrator to run the model effectively and efficiently, they may
be less likely to accept the model.
For a compensation model to work as intended, it should not have to
rely on a certain number of NRSROs to attain its goal. Such models
could be undermined if only one or two NRSROs participated in the
rating of specific types of securities or if an NRSRO exited the
industry. While some proposed alternative compensation models intend
to encourage competition among the NRSROs, the models themselves
should not hinder the financial viability of an NRSRO. For example,
the potential impact on the smaller NRSROs of any participation costs
should be considered. The model also should not introduce undue
uncertainty into the industry, so that NRSROs could not conduct
appropriate business planning (that is, attract and retain qualified
staff).
6. Market Acceptance and Choice. The willingness of the securities
market to accept the compensation model, the ratings produced under
that model, and any new market players established by the compensation
model.
* What role do market participants have in selecting NRSROs to produce
ratings, assessing the quality of ratings, and determining NRSRO
compensation? More specifically, what are the roles of issuers and
investors in these processes? Where do these roles differ between
models and what are the trade offs?
* Are all market participants likely to accept the ratings produced
under the compensation model? If not, what are the potential
consequences for the securitization market?
* What impact, if any, would the model have on each market participant
using the ratings?
* Would market participation need to be mandated, and if so, for which
participants?
The likelihood that market participants will accept an alternative
compensation model is another important consideration in its
evaluation. In achieving its goals, such as increasing independence in
the ratings process or competition among NRSROs, a particular model
may limit or promote the participation and choices of some market
participants over others and could affect the market's acceptance of
and participation in the model. For example, as we have pointed out,
the proposed models we have identified require that the issuer pays
for the rating; however, in most of the models the issuer is no longer
able to select which NRSRO rates the security. Limiting the issuer's
choice may address conflicts of interest and increase the independence
of the ratings process, but also could deter issuers from soliciting
NRSROs to rate their debt. Market acceptance by institutional
investors is also instrumental to the success of the model. For
example, many private investment guidelines require the use of
specific NRSROs. If these specified NRSROs are not producing ratings
under an adopted model, then these investors may be limited in the
securities that they can consider purchasing. Such tradeoffs would
need to be carefully evaluated to ensure the model's viability and
minimize its impact on the securities market.
Market participants should accept the ratings produced by the
compensation model. If market participants, particularly issuers and
end users of ratings, do not have confidence in the ratings produced
under the model, its viability could be significantly undermined. This
is a particular concern with models that would mandate the use of a
particular NRSRO or otherwise limit the market's influence in the
supply and demand for ratings. Market participants, including
regulators outside of the United States, also could affect the
acceptance of the model. A rating that is not accepted could create
inconsistencies between domestic and foreign securities markets for
investors that rely on ratings.
7. Oversight. The evaluation of the model to help ensure it works as
intended.
* Does the model provide for an independent internal control function?
* What external oversight (from a regulator or third-party auditor)
does the compensation model provide to ensure it is working as
intended?
* If third-party auditors provide external oversight, how are they
selected, what are their reporting responsibilities, and to whom do
they report?
* Who will compensate the regulator or third-party auditor for
auditing the compensation model? How will the compensation for
regulator/auditor be determined?
* To what extent will a third-party auditor allow flexibility in
oversight to accommodate NRSROs of different sizes?
An effective alternative compensation model also will provide for
independent internal controls and robust, external oversight to ensure
its integrity. This is especially important when a model calls for
third-party administration or the use of information technology
systems in its implementation. For example, any centralized system
that collects fees from issuers and compensates the NRSRO(s) should be
audited, as should any procedures used to award ratings business to an
NRSRO, evaluate NRSRO performance, or apply economic penalties. Such
oversight will help ensure the model functions as intended, thus
increasing transparency and market acceptance. Funding for this
oversight should be specified.
The Dodd-Frank Act contains a mandate for SEC to conduct a study of
the feasibility of establishing a system in which a public or private
utility or a self-regulatory organization assigns NRSROs to determine
the credit ratings of structured finance products. The study must
include an assessment of the potential mechanisms for determining fees
for NRSROs, appropriate methods for paying fees to the NRSROs, and the
range of metrics that could be used to determine the accuracy of
credit ratings. SEC also must evaluate alternative means for
compensating NRSROs that would create incentives for accurate credit
ratings.[Footnote 115] Our framework could be used to evaluate current
proposals for compensating NRSROs, develop new proposals, and identify
tradeoffs among them.[Footnote 116]
Conclusions:
The Credit Rating Agency Reform Act sought to improve ratings quality
for the protection of investors, including establishing SEC oversight
over credit rating agencies that register as NRSROs. However, SEC
faced a number of challenges in implementing the law, and the recent
financial crisis resulted in additional changes to SEC's oversight of
NRSROs under the Dodd-Frank Act. As SEC starts to implement its new
requirements, there are a number of challenges from its existing
responsibilities that must also be addressed.
SEC's implementation of the Act involved developing an NRSRO
registration program and an examination program. As currently
implemented and staffed, both programs require further attention.
* As intended by the Act, the new registration program for NRSROs
reduced the time SEC staff took to act on an application and improved
the transparency of SEC's process for awarding NRSRO designations to
interested credit rating agencies. However, due to the time
constraints, a lack of criteria, and lack of express preregistration
examination authority, Trading and Markets staff said the registration
process generally does not allow staff to conduct reviews that would
allow staff to confirm that the information provided on Form NRSRO was
accurate and the applicant met all of the Act's requirements. However,
SEC has yet to explicitly identify the legislative changes needed to
address this limitation and work with Congress to ensure it has the
authority needed to effectively carry out its oversight
responsibilities. This raises a concern because SEC may approve
applicants that do not meet the Act's requirements.
* Although SEC has established an OCIE branch dedicated to the
examination of NRSROs and hired individuals with experience in credit
rating analysis and structured finance to fill these positions, OCIE
has not completed timely examinations of the NRSROs and expressed
concerns about its ability to meet its planned NRSRO routine
examination schedule of examining the three largest NRSROs every 2
years and the other NRSROs every 3 years. While SEC requested
additional resources that it anticipated using to fully staff this
oversight function, it will likely need to revisit those requests due
to the passage of the Dodd-Frank Act, which requires SEC to establish
an Office of Credit Ratings and examine each NRSRO every year. As SEC
begins its planning of this new office, it is essential that SEC
assess not only the number of staff it needs but also the skills
required of this staff. Approaching this effort strategically may
facilitate the recruitment and training of new hires. Without a plan
that details the amount of staff needed with the requisite
qualifications and training, SEC may face challenges in meeting the
required examination timetable and providing quality oversight of the
NRSROs.
SEC rules requiring NRSROs to publish short-, medium-, and long-term
performance statistics have increased the amount of information
publicly available about the performance of some NRSROs, particularly
those newly registered. Overall, the disclosure of these statistics
has not had the intended effect of increasing transparency for users.
Specifically,
* SEC has not provided specific guidance for the NRSROs for
calculating and presenting the required performance statistics.
Therefore, NRSROs have used different methodologies for calculating
the required performance statistics, which renders them ineffective
for comparative purposes.
* SEC has yet to evaluate the appropriateness of the required
performance statistics for SEC's currently designated asset classes to
determine if the requirements need to be modified. Asset classes that
are defined too broadly limit the usefulness of the disclosures.
The Dodd-Frank Act directs SEC to adopt additional rules requiring
NRSROs to publicly disclose information on the initial credit ratings
determined by each NRSRO for each type of obligor, security, and money
market instruments, and any subsequent changes to such credit ratings.
In developing these new disclosure requirements, it will also be
important for SEC to provide clear and specific guidance to NRSROs.
Otherwise, the resulting disclosures may lack comparability. Recent
SEC rules to make ratings history data publicly available are intended
to generate performance measures and studies to evaluate and compare
NRSRO performance. However, it is unlikely that the ratings histories
NRSROs publish pursuant to the 10 percent and 100 percent requirements
can be used for these purposes. Specifically,
* SEC did not specify the data fields the NRSROs were to disclose as
part of their 10 percent sample disclosures, and the data fields
provided by the NRSROs did not always provide sufficient information
to allow users to identify a complete rating history for each rating
in the sample, including the beginning of rating histories, or
specific types of ratings for making comparisons. As a result, users
cannot develop performance measures that track how an issue or
issuer's credit rating evolves, evaluate comparable entities across
NRSROs, or calculate measures that compare a starting point to the
state of a rating at the time of default.
* Users cannot easily determine what data the variables represent
because NRSROs were not required to provide an explanation of the
variables used in the samples as part of their disclosures. Without
such explanations, users may find it difficult to begin to construct
performance statistics.
* Because SEC did not require it, not all NRSROs disclosed defaults as
part of their ratings histories in the samples. As such, users cannot
calculate default statistics for those NRSROs.
* Because SEC guidance to NRSROs for generating the 10 percent random
samples does not specify that the NRSROs draw the sample from those
ratings that are typically analyzed in each asset class and does not
require the NRSROs to periodically redraw the samples or include
ratings that have been withdrawn in prior time periods, the samples
are not representative of the population of credit ratings at each
NRSRO. As a result, users cannot generate performance measures that
represent the population of credit rating over time and that can be
compared across NRSROs.
* Because the 100 percent disclosure requirement does not require that
the NRSROs disclose the ratings of any issuer rated before June 26,
2007, or include data on withdrawn ratings, performance statistics
calculated using the 100 percent data set would not reflect the
overall rating performance of NRSROs and may not be representative of
the universe of issuer ratings.
* As SEC also did not provide guidance to NRSROs on how to format and
describe the data disclosed under the 100 percent requirement, users
will likely experience challenges when seeking to construct ratings
histories and develop comparable performance statistics.
Finally, SEC's rule requiring NRSROs to disclose total outstanding
ratings is intended for users to assess how broad an NRSRO's coverage
is within a particular class of credit ratings. However, because SEC
did not specify how NRSROs were to count their outstanding ratings,
NRSROs used diverse methodologies to count up their outstanding
ratings. As a result, users of data cannot use them for their intended
purpose.
The Dodd-Frank Act requires SEC and other federal agencies to remove
references to NRSRO ratings from their regulations, and substitute an
alternative standard of creditworthiness. SEC has recently removed or
proposed to remove references to NRSRO ratings from several rules. In
comparison, the Dodd-Frank Act requirement is a broader undertaking
that requires a strategic approach. SEC's previous experience
highlights the importance of developing a plan to ensure that (1) any
adopted alternative standards of creditworthiness for a particular
rule facilitate its purpose and (2) examiners have the requisite
skills to determine that the adopted standards have been applied.
Without such a plan, SEC may develop alternative standards of
creditworthiness that are not effective in supporting the purpose of a
particular rule.
Among its stated goals, the Act intended to improve competition among
NRSROs by creating a more efficient and transparent NRSRO designation
process for SEC to administer. The Act made receiving an NRSRO
designation easier and resulted in three new, subscriber-pays NRSROs.
However, industry concentration remains high. Given the limited time
that has passed since SEC implemented the registration program, the
impact of the credit crisis on the securities market, and uncertainty
about changes in the regulatory environment, significant changes in
industry concentration in the short term are not likely. However,
whether any changes in industry concentration will materialize in the
long term is still unclear. The credit rating industry continues to
manifest its traditional barriers to entry, such as high fixed costs,
that make gaining customers and achieving significant coverage of the
securities market a challenge for new entrants. In some asset classes,
such as structured finance, these barriers are especially high.
Recognizing this, SEC amended the Exchange Act rule 17g-5 to increase
the number of structured finance ratings by requiring issuers that
contract with an NRSRO for a rating to agree to make the underlying
data free to other interested NRSROs in order to encourage nonhired
NRSROs to produce unsolicited ratings. As this rule became effective
in June 2010, not enough time has passed to assess its impact.
However, we note that the impact of competition on the quality of
credit ratings is not well understood. The limited academic research
conducted in this area suggests that the entry of new credit rating
agencies will improve overall information available to investors and
other market participants, but that the effect of entry on the quality
of ratings produced by any one rating agency is not well-established.
To assist Congress and others in evaluating proposed models for
compensating NRSROs, we created an evaluative framework of seven
factors that any compensation model should address to be effective.
The Dodd-Frank Act contains a mandate for SEC to conduct a study of
the feasibility of establishing a system in which a public or private
utility or a self-regulatory organization assigns NRSROs to determine
the credit ratings of structured finance products. SEC also must
evaluate alternative means for compensating NRSROs that would create
incentives for accurate credit ratings. Our framework could be used to
evaluate current proposals for compensating NRSROs, develop new
proposals, and identify trade offs among them.
Recommendations for Executive Action:
To address the concern that the current process for registering credit
rating agencies may result in SEC approving applicants that do not
meet the Act's requirements, the Chairman of the Securities and
Exchange Commission should:
* Identify the additional authorities and time frames necessary to
ensure that staff can verify the accuracy of the information provided
on Form NRSRO and that the applicant meets all of the Act's
requirements; the Chairman should also work with Congress to ensure
that SEC has the authority needed to effectively carry out its
oversight responsibilities.
To ensure that SEC has sufficient staff with the skills necessary to
address the requirement in the Dodd-Frank Act that SEC establish an
Office of Credit Ratings and examine each NRSRO every year, the
Chairman of the Securities and Exchange Commission should:
* Develop and implement a plan for the establishment of this office
that includes the identification of the number of staff and the skills
required of these staff to meet the required examination timetable and
provide quality oversight of the NRSROs, including plans for the
recruitment of any new hires and appropriate training.
To address the inconsistencies in the NRSROs' methodologies for
calculating required performance statistics and total outstanding
ratings for initial and updated Form NRSRO filings, address
limitations in the required 10 percent and 100 percent rating history
disclosures, and increase the comparability and usefulness of these
disclosures, the Chairman of the Securities and Exchange Commission
should take the following eight actions:
* for the disclosures of required performance statistics,
- provide specific guidance for NRSROs for calculating and presenting
these performance statistics, considering the impact of different
methodologies on the information content of the performance statistics
and the purpose for which SEC intends the statistics to be used; and:
- evaluate the appropriateness of SEC's currently designated asset
classes for presenting performance statistics, and where SEC
determines that the asset classes are not appropriate, modify the
requirements accordingly;
* for the disclosures of required 10 percent and 100 percent ratings
histories,
- ensure that the data elements required as part of the datasets allow
users to construct complete ratings histories, identifying the
beginning of ratings histories, and distinguish between different
types of ratings;
- consider requiring NRSROs to publish a codebook to explain the
variables included in the datasets;
- clarify that NRSROs should include defaults in the ratings histories
disclosed;
- review its guidance to NRSROs for generating the 10 percent samples
and modify it as needed to ensure that the samples are 10 percent of
the type of ratings typically analyzed in each asset class, that
withdrawn ratings are not removed from these samples, and that the
samples are periodically redrawn; and:
-review its guidance to NRSROs for generating the 100 percent rating
history disclosures and modify it as needed to ensure that these
histories include those ratings that are typically analyzed in each
asset class; and that withdrawn ratings are not removed from these
disclosures;
* for the disclosures of total outstanding ratings required on Form
NRSRO,
- provide specific guidance to NRSROs to calculate their total
outstanding ratings.
To address the Dodd-Frank Act's requirement for SEC to remove any
references to or requirements of reliance on credit ratings in its
rules and substitute alternative standards of credit worthiness that
it deems appropriate, the Chairman of the Securities and Exchange
Commission should:
* develop and implement a plan for approaching the removal of NRSRO
references from SEC rules to help ensure that (1) any adopted
alternative standards of creditworthiness for a particular rule
facilitate its purpose and (2) examiners have the requisite skills to
determine that the adopted standards have been applied.
Agency Comments and Our Evaluation:
We provided a draft of the report to the SEC Chairman for her review
and comment. SEC provided written comments that are summarized below
and reprinted in Appendix III. We also received technical comments
from SEC that were incorporated, where appropriate.
In its written comments, SEC agreed with our findings. With respect to
our recommendation that addresses concerns about the current
registration process, SEC stated that SEC staff intend to develop
proposals to provide Congress with technical assistance for how the
relevant portions of the Securities Exchange Act of 1934 could be
amended to address the issues we identified with the registration
process for NRSROs. Regarding our recommendation to develop and
implement a plan for the establishment of the Office of Credit
Ratings, SEC stated that it anticipates drawing upon our findings as
it staffs this new office. SEC stated that it is in the process of
hiring staff to meet its new responsibilities under the Dodd-Frank Act
relating to NRSROs and has allocated between twenty-five and thirty-
five positions to this new office.
With respect to the recommendations related to improving the
comparability of the NRSROs required disclosures of performance
statistics, SEC noted the Dodd-Frank Act mandates that it adopt rules
that require NRSRO disclosures of performance-related data to be
comparable among NRSROs in order to allow users of credit ratings to
compare the performance of credit ratings across NRSROs. SEC stated
that our review of the existing disclosure requirements will be
helpful to SEC staff in developing recommendations for the Commission
in response to the congressional mandate.
With respect to the recommendations related to addressing limitations
in the required 10-percent and 100-percent rating history disclosures,
SEC noted that on August 27, 2010, it published on its Web site the
list of XBRL tags that NRSROs must use for these ratings history
disclosure requirements. SEC believes this step may address some of
the our concerns regarding its guidance as to the data fields NRSROs
must use for these disclosures. We encourage SEC to evaluate the
extent to which these tags address the limitations we identified, and,
to the extent that they do not, to take further action. As we
discussed, these limitations render current ratings history
disclosures largely unusable.
With respect to our recommendation regarding SEC's approach to address
the Dodd-Frank mandate that it remove NRSRO references from its rules
and substitute alternative standards of credit worthiness that it
deems appropriate, SEC agreed that any proposed alternative standards
of creditworthiness for a particular rule should facilitate its
purpose and that replacing NRSRO ratings with alternative standards
will require that SEC ensure that examiners have the requisite skills
to determine that the adopted standards have been applied. SEC stated
that it has already hired senior examiners with specialized expertise
and skills and that it continues to increase its expertise in these
areas through its recruiting and training programs.
We are sending copies of this report to interested congressional
committees and the Chairman of SEC. The report will also be available
at no charge on the GAO Web site at [hyperlink, http://www.gao.gov].
If you or your staffs have any questions about this report, please
contact me at (202) 512-8678 or williamso@gao.gov. Contact points for
our Offices of Congressional Relations and Public Affairs can be found
on the last page of this report. GAO staff who made major
contributions to this report are listed in appendix IV.
Signed by:
Orice Williams Brown:
Director, Financial Markets and Community Investment:
[End of section]
Appendix I: Objectives, Scope, and Methodology:
To discuss the implementation of the Credit Rating Agency Reform Act
of 2006 (Act), focusing on Securities and Exchange Commission (SEC)
rulemaking and SEC's implementation of the registration and
examination programs for Nationally Recognized Statistical Rating
Organizations (NRSRO), we reviewed the rules SEC has adopted to
implement the Act, including the NRSRO registration program and its
oversight of NRSROs. We reviewed the SEC's Inspector General's August
2009 audit report for findings and recommendations pertaining to SEC's
NRSRO registration program as well as SEC's July 2008 public report
discussing its findings on examinations of selected NRSROs. We also
reviewed copies of the Division of Trading and Markets' (Trading and
Markets) internal memoranda to SEC documenting its review of NRSRO
applications and internal memoranda outlining additional NRSRO
monitoring. We obtained and reviewed the Office of Compliance
Examinations and Inspections (OCIE) guidance for conducting an NRSRO
examination and reviewed completed NRSRO examinations. To understand
additional changes to SEC's oversight of the NRSROs, we reviewed the
recently passed Dodd-Frank Wall Street Reform and Consumer Protection
Act (Dodd-Frank Act). We conducted interviews with Trading and Markets
regarding the NRSRO application review process, and SEC's Division of
Investment Management (Investment Management) and the Financial
Industry Regulatory Authority (FINRA) to obtain information on SEC's
process for registering investment advisors and FINRA's process for
registering new broker-dealer members, respectively. We also conducted
interviews with OCIE staff regarding the NRSRO examinations that were
initiated after the Act's passage and obtained information from OCIE
on its staffing of the NRSRO examination team.
To evaluate the performance-related NRSRO disclosures that the Act and
SEC rules require, we analyzed SEC rules requiring NRSRO disclosure of
performance statistics and ratings history samples. First, we analyzed
and compared the 10 NRSROs' 2009 disclosures of performance
statistics, focusing on the corporate and structured finance asset
classes, and we reviewed voluntary disclosures of additional
performance statistics by 5 of the NRSROs. We also reviewed the Dodd-
Frank Act to understand its directive to SEC to ensure the
comparability of NRSRO performance disclosures. Second, we assessed
NRSRO ratings history data disclosed by the 7 issuer-pays NRSROs
pursuant to SEC's 10 percent disclosure requirement. From each NRSRO's
Web site, we obtained its 10 percent random samples of outstanding
ratings. We downloaded these data in February and March 2010.[Footnote
117] To assess the reliability and usability of the samples for
generating comparative performance measures, we reviewed the samples
and obtained information from each of the NRSRO's on the methods used
to draw and maintain the samples. We identified a number of issues
that led us to determine that the data were not in a format that
allowed us to generate comparative performance statistics. For
example, we found that the data fields required by the rule were not
always sufficient to identify a complete rating history for ratings in
each of the seven samples and did not give us enough information to
identify specific types of ratings for making comparisons. We also
could not consistently identify the beginning of the ratings histories
in all of the samples. Furthermore, we found that the guidance
provided by SEC to NRSROs for generating the random samples does not
ensure that the method used will result in samples that are
representative of the population of credit ratings at each NRSRO.
Since it was impossible to compare performance statistics across
NRSROs in any reliable manner, we focused on identifying the issues we
encountered when reviewing the available data and attempting such
comparisons. We also analyzed SEC's 100 percent disclosure requirement
by reviewing the rule and obtaining information from Trading and
Markets and two NRSROs on its implementation. As this rule did not
become effective until June 2010, we did not review the data disclosed
by NRSROs pursuant to the rule, with the exception of one NRSRO. We
conducted a limited review of the data disclosed by this NRSRO to
determine the extent to which the data included ratings of issuers
rated prior to June 26, 2007. As part of this objective, we also
assessed the NRSROs' disclosures of total outstanding ratings required
on Form NRSRO. We reviewed these disclosures and SEC's instructions
pertaining to these disclosures and obtained information from the
NRSROs on their methods for determining total outstanding ratings.
To evaluate the potential regulatory impact of removing NRSRO
references from certain SEC rules, we reviewed SEC's proposed rules to
remove references to NRSRO ratings, focusing on proposed amendments to
rule 2a-7 under the Investment Company Act of 1940 (Investment Company
Act) and Securities and rule 15c3-1under the Securities Exchange Act
of 1934 (Exchange Act) and the comment letters submitted to SEC on
these proposals. We reviewed multiple studies pertaining to the use of
ratings in regulations. To understand the extent to which NRSRO
ratings are used in U.S. federal securities, banking, and insurance
laws, rules, and regulations we obtained a copy of a Joint Forum
report documenting their use. To understand how the ratings were used
in the rules and the regulatory impact the removal of the ratings
might have, we reviewed OCIE's examination guidance for reviewing for
compliance with rules 2a-7 and 15c3-1. We also reviewed the 65
examinations of money market funds OCIE completed in FY 2003-2009,
which reviewed funds for compliance with rule 2a-7. We did so to
understand how examiners ensured that funds complied with the rule's
two-part test for determining if a security was eligible for purchase
and how the removal of NRSRO ratings might affect oversight of this
determination. We reviewed the Dodd-Frank Act to understand its
directive to SEC and other federal agencies to remove NRSRO references
from their rules regulations. We also interviewed staff from
Investment Management to better understand the purpose of rule 2a-7,
and how the removal of references might affect oversight. We conducted
interviews with OCIE staff, Trading and Markets staff, and market
participants and observers about the pros and cons of utilizing NRSRO
references in, and the potential impact of removing them from,
regulations.
To evaluate the impact of the Act on competition among NRSROs, we
reviewed proposed and final SEC rules intended to promote competition
among rating agencies, as well as the comment letters SEC received in
response to those rules. We reviewed SEC's 2008 and 2009 mandated
annual reports on NRSROs, including SEC's studies on competition in
the credit rating industry. We used the Herfindahl-Hirschman Index
(HHI) to track industry concentration over time.[Footnote 118] The HHI
is a measure of industry concentration that reflects both the number
of firms in the industry and each firm's market share. It is
calculated by summing the squares of the market shares of each firm
competing in the market. The HHI reflects both the distribution of
market shares of the top firms and the composition of the market
outside the top firms. The HHI is measured on a scale of 0 to 10,000,
with larger values indicating more concentration. According to the
Department of Justice and Federal Trade Commission, markets in which
the value of the HHI is between 1,500 and 2,500 points are considered
to be moderately concentrated, and those in which the value of the HHI
is in excess of 2,500 points are considered to be highly concentrated,
although other factors also play a role.
Calculating the HHI requires defining what constitutes the industry
and specifying our measure of market share. We define the relevant
industry as the set of credit rating agencies that have NRSRO status,
and we use a variety of market share definitions to ensure that any
trends in industry concentration we observe are robust to alternative
specifications of NRSROs' market shares. A firm's market share
typically is measured in terms of dollars, as either its sales or
revenue as a fraction of total sales or revenue for all firms in the
industry, or in terms of quantities, as its output as a fraction of
total output produced by all firms in the industry. We measured an
NRSRO's overall market share as its revenue as a share of total
revenue for the industry. We measured an NRSRO's market share overall
in each asset class as the number of debt-security issuing
organizations or entities it rates as a percent of the sum of the
numbers of organizations that each NRSRO rates, in which the number of
organizations rated is our proxy for output. Finally, for the ABS
asset class, we measured an NRSRO's market share as the dollar value
of new U.S. issuance it rated as a percentage of the sum of the dollar
value of issuance each NRSRO rated. We then calculated the HHI using
these definitions of market share.
We obtained the revenue data for 2006-2009 from the NRSROs' initial
and annual Form NRSRO filings. We obtained the number of organizations
or entities that issue debt securities and are rated by the NRSROs per
asset class for 2006-2009 from 9 of the 10 NRSROs.[Footnote 119] We
obtained data on the dollar amount of new U.S.-issued asset-backed
debt rated by NRSROs for 2004--June 2010 from an industry newsletter
that tracks asset-backed securitization. Because these data came from
private firms, we were able to conduct only limited assessments of the
data's reliability. We were not, for example, able to conduct our own
reliability tests on the data. For the revenue data obtained from the
NRSROs' responses to Form NRSRO, we interviewed the staff from Trading
and Markets to determine the steps they took to ensure the data
represented the firms' total revenues. As such, we determined that the
data were sufficiently reliable for our purpose, which was to show the
relative concentration of the NRSROs in the industry. For the data we
obtained from the NRSROs on the number of rated organizations or
entities that issue debt securities, we specified the methods by which
the NRSROs were to count and classify the ratings. We did this to
ensure consistency in the data across the NRSROs. We also examined the
data, both within and NRSRO and among NRSROs, to determine whether
there were any illogical trends that would indicate the date had been
prepared incorrectly. As such, we determined that the data were
sufficiently reliable for our purpose, which was to show the relative
concentration of the NRSROs in the industry. For the data we obtained
from the private firm tracking the dollar value of rated new U.S.
asset-backed securitization, we obtained information from the firm on
the processes and procedures used to collect and manage the data. We
determined that the data were sufficiently reliable for our purpose,
which was to show the relative concentration of the NRSROs in the
industry. We also used these data to generate a series of descriptive
graphs showing the NRSROs' market coverage, that is, the percentage of
new U.S. asset-backed issuance that each rated. We also believe that
the data are sufficiently reliable for this purpose.
We also reviewed academic studies on competition in the industry. We
identified three studies that analyze data to assess the impact of the
number of rating agencies on some aspect of the credit rating
industry. We identified these studies by searching databases of both
unpublished working papers and papers published in refereed academic
journals and by searching the bibliographies of studies we found in
the databases.[Footnote 120] The three studies we identified are all
unpublished working papers. We reviewed these studies and determined
they did not raise any serious methodological concerns. However, the
inclusion of these studies is purely for research purposes and does
not imply that we deem them to be definitive. Finally, we obtained the
views of SEC's Office of Risk Analysis, the NRSROs, credit rating
agencies that are not registered NRSROs, institutional investors,
issuers, and industry experts on the impact of the Act on competition.
To provide an overview of proposed alternative models for compensating
NRSROs and an evaluative framework for assessing the models, we
identified proposals on alternative models for compensating NRSROs by
reviewing white papers submitted to the SEC roundtable on credit
rating agency oversight, academic and other white papers, and
interviewed the authors of the proposed models. We obtained the views
of Trading and Markets, SEC's Office of Risk Assessment, and NRSROs.
We also spoke with credit rating agencies that are not registered
NRSROs, institutional investors, issuers, and academic and industry
experts. To develop the framework for evaluating the models, we
reviewed prior GAO reports and obtained the views of market
participants and observers to identify appropriate factors for
inclusion. We then convened a panel of GAO experts (financial markets
specialists, economists, an attorney and a social scientist) to review
the framework and incorporated their comments. Finally, we solicited
comments from NRSROs, proposal authors, industry experts, and trade
associations and incorporated them as appropriate.
We conducted this performance audit from May 2009 to September 2010 in
accordance with generally accepted government auditing standards.
Those standards require that we plan and perform the audit to obtain
sufficient, appropriate evidence to provide a reasonable basis for our
findings and conclusions based on our audit objectives. We believe
that the evidence obtained provides a reasonable basis for our
findings and conclusions based on our audit objectives.
[End of section]
Appendix II: Other Registration Processes Provide Greater Flexibility
to the Regulators:
Other registration programs for securities market participants allow
regulators to exercise greater oversight over applicants than does the
registration program for Nationally Recognized Statistical Rating
Organizations (NRSROs) while maintaining an efficient and transparent
registration process. More specifically, in the Securities and
Exchange Commission's (SEC) Division of Investment Management's
(Investment Management) registration programs for investment advisors
and the Financial Industry Regulatory Authority's (FINRA) Registrant
Application process for broker-dealers, staff can request additional
information from the applicant in specific circumstances, extend the
review process, and reject an application for a broader set of reasons.
SEC's registration process for investment advisers is authorized by
section 203 of the Investment Advisers Act of 1940 (Investment
Advisers Act), which identifies generally the type of information
required in an application for registration as an investment adviser
and prescribes a time frame, 45 days, by which SEC must act on an
application.[Footnote 121] Investment advisers apply by completing and
submitting Form ADV, which requires information such as the type of
company that is applying, what businesses it will be involved in, its
assets under management, its employees and clients, compensation and
ownership arrangements, any financial industry affiliates, and whether
the adviser or any person associated with the adviser is subject to
certain disciplinary events.[Footnote 122] SEC must grant registration
to applicants if it finds that the requirements of section 203 are
satisfied. After instituting a proceeding to determine whether
registration should be denied, SEC must deny a registration if it does
not make such a finding or if it finds that if the applicant were so
registered, its registration would be subject to suspension or
revocation.
Although Investment Management oversees the rules and policies
regarding the registration of investment advisers, the Office of
Compliance Inspections and Examinations (OCIE) is responsible for
reviewing the application materials and evaluating whether the
application is complete. If additional information is necessary to
consider the application or clarify inconsistencies in the information
provided, OCIE contacts the applicant with questions and requests
additional information. If additional information is required,
applicants submit an amended Form ADV, triggering a new 45-day review
period. Investment Management officials characterized the process as
efficient, and stated that OCIE completes application reviews in an
average of 30 days, with variations due to factors such as volume of
applications and complexity. Staff stated that incomplete applications
can be placed in a postponed status. When postponed, OCIE contacts the
applicant through an official letter, which typically states that SEC
received the application and describes the parts of the filing that
need to be corrected or completed. While staff await a response from
the applicant, the running of the 45-day review period is
automatically suspended pending receipt by SEC of the additional
information necessary to complete the application. Staff said that
sometimes an applicant may never provide the requested information or
correct the deficiency in the application. In these situations, OCIE
considers the application incomplete. Once a complete application is
received, a review of disciplinary information is completed to
determine if there is a reason to recommend to SEC to deny, condition,
or limit the registration. If there is none, the application is
approved.
FINRA's registrant process also imposes a time frame on FINRA staff
for reviewing broker-dealer applications for registration. Under FINRA
Rule 1014, FINRA staff must make a decision no more than 180 days from
the filing date unless requested by the applicant and to which FINRA
otherwise agreed. Broker-dealers applying for registration must submit
an application providing information demonstrating they meet the 14
standards specified by the FINRA rule. These include that they have
adequate financial and managerial resources, supervisory systems, and
compliance procedures designed to detect and prevent violations of
securities laws and related rules; recordkeeping systems that enable
compliance with regulatory recordkeeping requirements; and staff
sufficient in qualifications and number to prepare and preserve
required records. FINRA's Department of Member Regulation reviews and
either approves, approves with restrictions or denies broker-dealer
member applications, and staff have the authority, if there are any
questions, to make additional requests for information. The department
also conducts a membership interview to further clarify the
application material, after which additional information is requested
and thereafter, a final decision is made. According to FINRA
officials, staff frequently have additional questions. For example,
staff may have factual questions about capital, employee
registrations, leases, or the location of the broker-dealer facility.
While FINRA officials said that at no point should a decision on an
application take more than 180 days unless agreed to by the applicant
and agreed to by FINRA, they also stated that if staff did not
receive, within prescribed timeframes, the information they requested
to satisfy any questions they would, absent good cause, reject or
lapse the application. They also can reject an application at the time
it is initially filed if that application has a material deficiency
(i.e. was not substantially complete); for example, if staff were
unable to tell what the applicant's business was going to do, how the
business would be supervised, or who was intending to fund the broker-
dealer. If applicants disagree with the department's decision to deny
or restrict an application, they can appeal internally to FINRA and
then to SEC, and finally in Circuit court.[Footnote 123] According to
FINRA officials, the department reviews more than 200 new member
applications in a year and decisions often take less than 180 days.
They estimated that about 80 percent are completed within 180 days.
FINRA officials stated that the process allows FINRA to restrict
certain activities and deny unqualified applicants.
Both FINRA's registrant application process for broker-dealers and
SEC's registration process for investment advisors require applicants
to provide specific information and utilize deadlines to ensure an
efficient process. However, unlike the NRSRO registration program,
staff of these programs are provided the authorities necessary to
clarify any outstanding questions they have regarding an applicant and
to delay approving that applicant until the staff are satisfied that
the applicant has met all of the necessary requirements.
[End of section]
Appendix III: Comments from the Securities and Exchange Commission:
United States:
Securities And Exchange Commission:
Washington, D.C. 20549:
September 10, 2010:
Ms. Orice Williams Brown:
Director:
Financial Markets and Community Investment:
Government Accountability Office:
441 G Street, NW:
Washington, DC 20548:
Dear Ms. Brown:
Thank you for the opportunity to comment on the Government
Accountability Office (GAO) draft report entitled Securities and
Exchange Commission: Action Needed to Improve Rating Agency
Registration Program and Performance-Related Disclosures (GA0-10-782).
At the outset, we want to acknowledge the effort that went into the
preparation of this report and the thoughtful analysis and suggestions
it provides.
The report details the GAO staff's concerns in connection with the
registration process for nationally recognized statistical rating
organizations ("NRSROs"), including in particular its concerns
regarding the constraints placed on the Commission's authority to
review registration applications by the strict statutory deadlines and
the inability to conduct pre-registration examinations of applicants.
We appreciate the report's analysis of these issues and concur with
the conclusion that legislative changes would be necessary to allow
the Commission staff to conduct pre-registration examinations. The
Commission staff intends to develop proposals to provide Congress with
technical assistance for how the relevant portions of the Securities
Exchange Act of 1934 (the "Exchange Act") could be amended to address
these issues based on the GAO's analysis.
We anticipate that the other analyses, findings, and recommendations
in the report will be of help to the Commission staff as it develops
recommendations for the Commission to implement the requirements in
the Dodd-Frank Wall Street Reform and Consumer Protection Act of 2010
(the "Dodd-Frank Act") with respect to the Commission's oversight of
NRSROs. As you know, the Dodd-Frank Act contains a number of
provisions impacting the Commission's regulatory program for NRSROs,
several of which directly involve issues discussed in the report.
For example, the Dodd-Frank Act requires the Commission to establish
an Office of Credit Ratings staffed sufficiently to carry out fully
the Commission's oversight responsibilities with respect to NRSROs,
including the new requirement to perform annual examinations of each
NRSRO. The Commission currently is in the process of hiring personnel
to meet its new responsibilities under the Dodd-Frank Act relating to
NRSROs, and has allocated between twenty-five and thirty-five
positions to this Office. We appreciate the GAO staff highlighting
the considerations that should inform this staffing process and
anticipate drawing upon the report's findings and recommendations in
this area. We also appreciate the GAO's recognition of our efforts to
maintain our current examination cycle goals with limited resources.
[Footnote 1]
In addition, the Dodd-Frank Act mandates that the Commission adopt
rules that require NRSRO disclosures of performance-related data to be
comparable among NRSROs in order to allow users of credit ratings to
compare the performance of credit ratings across NRSROs. The GAO
staff's comprehensive review of the Commission's existing performance
data disclosure requirements and their recommendations as to how to
increase the comparability and usefulness of these disclosures will be
an asset to the Commission staff in carrying out its own review of the
existing disclosure requirements and developing recommendations for
the Commission in response to the congressional mandate.
We note that on August 27, 2010, the Commission published on its
Internet Web site the List of XBRL Tags for NRSROs to be used for the
ratings history disclosure requirements of paragraph (d) of Rule 17g-2
under the Exchange Act. We believe this step may address some of the
GAO's concerns regarding the Commission's guidance as to the data
fields NRSROs must use for these disclosures.
We also note that the Dodd-Frank Act requires the Commission to
undertake a study on the feasibility and desirability of, among other
things, standardizing credit rating terminology across asset classes,
so that named ratings correspond to a standard range of default
probabilities and expected losses independent of asset class and
issuing entity. The work of the GAO staff on how NRSROs calculate
performance statistics will be helpful in performing this study.
The Dodd-Frank Act also requires the Commission, as well as every
other Federal agency, to review any regulation that requires the use
of an assessment of the creditworthiness of a security or money market
instrument and to remove any reference to or requirement of reliance
on credit ratings, substituting such standards of creditworthiness as
it determines is appropriate. As the report notes, the Commission
first proposed removing references to credit ratings in its
regulations in July 2008 and, in October 2009, voted to remove such
references from a number of Commission rules and forms. We agree with
the report's conclusions regarding the importance of ensuring that any
proposed alternative standards of creditworthiness for a particular
rule facilitate the purpose of that rule.
The report further notes that replacing NRSRO ratings with another
standard of creditworthiness will require the Commission to ensure
that examiners have the requisite skills to determine that the adopted
standards have been applied. We agree and note that the SEC has
already hired senior examiners with specialized expertise and skills,
and continues to increase its expertise in these areas through its
recruiting and training programs.
Finally, the Dodd-Frank Act requires the Commission to carry out a
study of and prepare a report on (1) the credit rating process for
structured finance products and the conflicts of interest associated
with the issuer-pay and the subscriber-pay models, (2) the feasibility
of establishing a system in which a public or private utility or a
self-regulatory organization assigns NRSROs to determine the credit
ratings of structured finance products, (3) the range of metrics that
could be used to determine the accuracy of credit ratings, and (4)
alternative means for compensating NRSROs that would create incentives
for accurate credit ratings. We anticipate that the seven-factor
framework for evaluating alternative compensation models set forth in
the report will be a valuable resource to the Commission staff in
carrying out this study and in making recommendations for any
rulemaking determined to be necessary or appropriate in the public
interest or for the protection of investors.
On behalf of the Commission staff, we thank the GAO staff for its work
on the report as well as for the opportunity to review and comment on
the draft before the report is issued in its final form.
Sincerely,
Signed by:
Robert W. Cook:
Director:
Division of Trading Markets:
Signed by:
Carlo di Florio:
Director:
Office of Compliance:
Inspections & Examinations:
Footnote:
[1] The report states that the SEC has been unable to meet its planned
routine examination cycle of examining the three largest NRSROs every
two years and the remaining NRSROs every three years. We note that
NRSROs did not become subject to SEC examination until 2007. We
conducted examinations of the three largest NRSROs in fiscal year 2008
and initiated subsequent examinations of the same three NRSROs in
fiscal year 2010. In addition, we have completed or initiated
examinations of all remaining registered NRSROs.
[End of section]
Appendix IV: GAO Contact and Staff Acknowledgments:
GAO Contact:
Orice Williams Brown, (202) 512-8678 or williamso@gao.gov:
Staff Acknowledgments:
In addition to the individuals named above, Karen Tremba, Assistant
Director; Lucas Alvarez; William Chatlos; Rachel DeMarcus; Courtney
LaFountain; Elizabeth Jimenez; Stefanie Jonkman; Matthew Keeler; Omyra
Ramsingh; and Barbara Roesmann made key contributions to the report.
[End of section]
Footnotes:
[1] Section 3(a)(60) of the Exchange Act (codified at 15 U.S.C. §
78c(a)(60)).
[2] 17 C.F.R. § 240.15c3-1. Rule 15c3-1, also known as the net capital
rule, generally defines net capital as a broker-dealer's net worth
(assets minus liabilities), plus certain subordinated liabilities,
less certain assets that are not readily convertible into cash, and
less a percentage of certain other liquid assets (for example,
securities). In computing their net capital broker-dealers are
required to deduct from their net worth certain percentages of the
market value of their proprietary securities positions, known as a
haircut. NRSRO ratings are used, along with other factors, to
determine the haircut for each security.
[3] The three largest NRSROs are Standard & Poor's, Moody's Investors
Service, and Fitch Ratings. These firms were criticized for rating
Enron as investment grade until only 4 days before the company filed
for bankruptcy on December 21, 2001.
[4] See SEC, Report on the Role and Function of Credit Rating Agencies
in the Operation of the Securities Markets, As Required by Section 702
(b) of the Sarbanes-Oxley Act of 2002. (Washington, D.C.: Jan. 24,
2003). The practice of issuers paying for their ratings creates the
potential for a conflict of interest. Arguably, the dependence of
rating agencies on revenues from the companies they rate could induce
them to rate issuers more liberally, and temper their diligence in
probing for negative information. This potential conflict could be
exacerbated by the rating agencies' practice of charging fees based on
the size of the issuance, as large issuers could be given inordinate
influence with the rating agencies.
[5] Under the no-action letter process, credit rating agencies
requested recognition from SEC as an NRSRO by requesting "no action"
relief. If SEC staff determined that the rating agency could be
considered an NRSRO, it issued a "no-action" letter stating it would
not recommend enforcement action to the Commission if ratings were
used by registrants for regulatory compliance purposes. If SEC staff
concluded the rating agency should not be considered an NRSRO, the
Commission would issue a letter denying a request for no-action relief.
[6] Pub. L. No. 109-291, 120 Stat. 1327 (Sept. 29, 2006) (amending the
Securities Exchange Act of 1934 and codified at various sections of
title 15 of the U.S. Code).
[7] Act of June 6, 1934, ch. 404, Title I (codified, as amended, at 15
U.S.C. §§78a et seq.).
[8] RMBS are debt obligations that represent claims to the cash flows
from pool of residential property loans. Beginning in 2007,
delinquency and foreclosure rates for subprime mortgage loans in the
United States dramatically increased, creating turmoil in the markets
for RMBS backed by such loans and other securities products related to
those securities. As the performance of these securities began to
deteriorate, the three rating agencies most active in rating these
instruments downgraded a significant number of their ratings.
[9] See SEC (Office of Compliance Inspections and Examinations,
Division of Trading and Markets and Office of Economic Analysis),
Summary Report of Issues Identified in the Commission Staff's
Examinations of Select Credit Rating Agencies (Washington, D.C.: Jul.
8, 2008).
[10] Act of August 22, 1940, ch. 686, title II, 54 Stat. 789
(codified, as amended, at 15 U.S.C. §§ 80a-1 et seq.). Investment
Company Act rule 2a-7 governs the operation of money market funds,
which rely on the rule to use valuation and pricing methods different
from those that other investment companies are permitted to use, to
help maintain a stable share price. The rule contains conditions that
restrict money market funds' portfolio investments to securities that
have either received certain minimum credit ratings from NRSROs or are
comparable unrated securities.
[11] The HHI is one of the concentration measures that government
agencies, including the Department of Justice (DOJ) and the Federal
Trade Commission (FTC), use when assessing concentration to enforce
U.S. antitrust laws.
[12] Pub. L. No. 111-203, 124 Stat. 1376 (July 21, 2010). This act
includes a number of provisions intended to improve the regulation of
credit rating agencies. For example, it increases internal control
requirements at NRSROs and directs SEC to issue rules requiring the
NRSRO to submit to it an annual internal controls report. The Dodd-
Frank Act also directs SEC to issue rules requiring greater
transparency of NRSRO rating procedures and methodologies and requires
SEC to establish an Office of Credit Ratings to conduct annual
examinations of the NRSROs.
[13] For example, one NRSRO stated in its 2010 Form NRSRO disclosure
that since its ratings are typically procured by institutional
investors and not issuers, it does not have ready access to management
of the issuer being analyzed. As a result, it does not rely heavily on
information obtained during discussion with issuer management.
[14] We identified and reviewed 24 studies dated 2000 or later that
analyzed the impact of credit ratings on some aspect of the financial
markets. We reviewed these studies and determined that they did not
raise any serious methodological concerns. However, the inclusion of
these studies is purely for research purposes and does not imply that
we deem them to be definitive.
[15] 72 Fed. Reg. 33564, 33619-36
[16] Oversight of Credit Rating Agencies Registered as Nationally
Recognized Statistical Rating Organizations, 72 Fed. Reg. 33564, 33619-
36 (June 18, 2007)(Final Rule)(codified, as amended, at 17 C.F.R. §§
240.17g-1 - 240.17g-6 and 17 C.F.R. § 249b.300) (2010).
[17] Amendments to Rules for Nationally Recognized Statistical Rating
Organizations, 74 Fed. Reg. 6456, 6482-84 (Feb. 2, 2009)(Amending 17
C.F.R. §§ 240.17g-2, 240.17g-3, 240.17g-5 and Form NRSRO); Amendments
to Rules for Nationally Recognized Statistical Rating Organizations,
74 Fed and Reg. 63833, 63863-65 (Dec. 4. 2009)(Amending 17 C.F.R. §§
240.17g-2, 240.17g-5 and 243.100). In December 2009, SEC proposed
rules that would, among other things, require NRSRO compliance
officers to furnish an annual report to SEC, disclose additional
information about sources of revenues on Form NRSRO, and make publicly
available a consolidated report containing information about revenues
of the NRSRO attributable to persons paying the NRSRO for the issuance
or maintenance of a credit rating. Proposed Rules for Nationally
Recognized Statistical Rating Organizations, 74 Fed. Reg. 63866, 63901-
04 (Dec. 4, 2009)(Proposed amendments to 17 C.F.R. §§ 240.17g-3,
249b.300 and Form NRSRO and proposed new § 240.17g-7).
[18] The concerns outlined in the July 8, 2008, report were identified
during SEC examinations of the three largest NRSROs. We discuss these
examinations in more detail later in this section.
[19] Some credit rating agencies stated they tried to obtain NRSRO
status for over a decade under the previous no-action letter process.
In addition, in the prior no-action letter process, SEC would conduct
examinations prior to providing a credit rating agency with a no-
action letter.
[20] S. Rep. No. 109-326, at 7-8 (2006).
[21] Section 15E(a)(1)(D) exempted from this requirement those credit
rating agencies that had been designated as NRSROs by staff prior to
August 2, 2006.
[22] Pub. L. No. 109-291, § 4, 120 Stat. at 1329-38 (codified at 15
U.S.C. §78o-7).
[23] 15 U.S.C. 78o-7(a)(2)(B).
[24] To date two credit rating agencies have agreed to an extension of
the application period. In its response to the SEC Inspectors General
report on the NRSRO registration program, Trading and Markets staff
stated that for at least one application its interactions with the
applicant during the application process made clear that the applicant
would not have consented to such an extension. See SEC, Office of
Audits, The SEC's Role Regarding and Oversight of Nationally
Recognized Statistical Rating Organizations (NRSROs). Report No. 458
(August 27, 2009).
[25] 15 U.S.C. § 78o-7(a)(2)(C).
[26] Section 15E(d) of the Exchange Act [15 U.S.C. § 78o-7(d)] sets
out the circumstances under which the Commission can censure, place
limitations on the activities, functions, or operations of; suspend;
or revoke the registration of an NRSRO. For example, these include
circumstances in which a person associated with the NRSRO has been
convicted of certain civil or criminal offenses or has been the
subject of a Commission order barring or suspending the right of the
person to be associated with an NRSRO. The Act allows SEC to deny a
NRSRO applicant's registration if it determines that the applicant or
a person associated with it has committed or omitted any act, or is
subject to an order or finding enumerated under Section 15E(d).
[27] The 10 NRSROs that SEC approved for registration as NRSROs as
part of the new registration program are A.M. Best Company, Inc.; DBRS
Ltd.; Fitch, Inc.; Japan Credit Rating Agency, Ltd.; Moody's Investors
Service; Rating and Investment Information, Inc.; Standard & Poor's
Ratings Services; Egan-Jones Rating Company; LACE Financial Corp.; and
Realpoint LLC.
[28] The Commission has recently instituted proceedings to determine
if an applicant should be denied registration as an NRSRO. We did not
review the memorandum provided to the Commission for this applicant.
However, the Order Instituting Administrative Proceedings identified
for consideration two grounds for denial: (1) whether the applicant
has sufficient connection with U.S. interstate commerce to register as
an NRSRO and thereby invoke the regulatory and oversight authority of
the SEC; and (2) whether the application should be denied on grounds
that if registered as an NRSRO, the applicant would be subject to
having its registration suspended or revoked under section 15E(d)(1)
of the Exchange Act because, in light of requirements in its home
jurisdiction, the applicant would be unable to comply with provisions
of the U.S. securities laws and rules, including, in particular,
Section 17 of the Exchange Act and Rules 17g-2 and 17g-3. See 75 Fed.
Reg. 20645-46 (Apr. 20, 2010).
[29] In one case, Trading and Markets staff asked an NRSRO applicant
to consent to a 2-day extension of the 90-day review requirement to
allow a Commissioner, who had been unable to vote on the application
earlier, to vote. The applicant initially resisted granting the
Commission the extension. Ultimately, it consented to the 2-day
extension but made clear it would not consent to a longer time period.
Trading and Markets staff said they have requested additional
documents from applicants in other cases but, without express
examination authority, an applicant may deny such requests. SEC does
not have the authority to conduct an examination before approving a
credit rating agency as an NRSRO.
[30] One applicant consented to two extensions, one for 7 days and
another for 14 days.
[31] As indicated by SEC's public order, in the case of the applicant
that SEC has instituted proceedings to determine if registration
should be denied; the information provided that led to the institution
of proceedings was not related to the financial or managerial
resources or other types of "qualitative concerns" raised by staff in
the other memoranda. SEC instituted proceedings based on the fact that
due to its home jurisdiction, the applicant may not be able to comply
with provisions of the U.S. securities laws and rules, in particular
the requirement that credit rating agencies make their books and
records available to examiners without notice.
[32] See, e.g., S. Rep. No. 109-326, at 7-8.
[33] See appendix II for a more detailed discussion of these
registration programs.
[34] The Market/Self-Regulatory Organization (SRO) Oversight groups
within OCIE are responsible for examining SROs to ensure that they and
their members comply with applicable federal securities laws and SRO
rules. The SROs include national stock exchanges, such as the New York
Stock Exchange, and national securities associations, such as FINRA.
Other SROs are registered clearing agencies and the Municipal
Securities Rulemaking Board.
[35] Because these examinations were conducted before the
establishment of the new NRSRO examination team, examiners from the
Office of Market Oversight/Self-Regulatory Organizations performed
them.
[36] OCIE wrapped any outstanding issues and examination work from the
unfinished examination into a new examination initiated to fulfill the
Dodd-Frank Act requirement that every NRSRO be examined annually.
[37] Section 15E(a)(1)(B)(i) requires an application for NRSRO
registration to contain information regarding credit rating
performance measurement statistics over short-, medium-, and long-term
performance measurements. 15 U.S.C. §78o-7(a)(1)(B)(i). In addition,
section 15E(a)(3) directs the SEC to require, by rule, that registered
NRSROs make the information and documents submitted in its application
for registrations publicly available on the NRSRO's Web site. As part
of its June 2007 rules implementing section 15E, SEC required NRSRO
applicants and registered NRSROs to disclose on their Form NRSRO
short- , medium-, and long-term historical transition and downgrade
rates for each class of credit rating for which they are registered.
The Form NRSRO was revised, effective April 2009, to require default
and transition rates (the latter include both upgrades and downgrades)
and specify short-, medium-, and long-term as 1-, 3-, and 10-year time
periods. Amendments to Rules for Nationally Recognized Statistical
Rating Organizations, 74 Fed. Reg. at 6483-84.
[38] Several NRSROs published their own performance measures prior to
SEC's requirements. The measures published vary considerably, but most
of these NRSROs published some form of transition and default rates.
In some cases, their statistics encompass longer time frames or focus
on particular geographic regions or industry sectors. Some NRSROs
publish other types of performance statistics. For example, two NRSROs
also publish Lorenz curves, also sometimes called "power curves" or
"cumulative accuracy profiles." Lorenz curves are visual tools for
assessing the accuracy of the rank ordering of creditworthiness that a
set of ratings provides. They are considered useful for comparing the
relative accuracy of different rating systems or the relative accuracy
of a single rating system measured at different points of time for
different cohorts. However, the NRSROs' ability to publish performance
statistics beyond what is required by SEC depends on data
availability. One NRSRO explained that the largest NRSROs are able to
generate more performance statistics and a more granular level than
the smaller NRSROs because they have many more ratings in their
database that span more years.
[39] Three NRSROs did not calculate the transition rates for each
rating category. Two provided the number of ratings in each rating
category at the beginning of the rating period and the number of
ratings that transitioned during the rating period. Users could
calculate transition rates from this information. The third provided
the number of ratings that transitioned during the rating period, but
did not provide the total number of ratings in each rating category.
Users could not calculate transition rates from these data.
[40] Some NRSROs created cohorts for a year based on ratings as of
January 1 of that year. Other NRSROs created cohorts for a year based
on ratings as of December 31 of the previous year.
[41] One NRSRO provided the number of ratings that transitioned to
default during the rating period, but did not provide the total number
of ratings in each rating category. Users could not calculate default
rates from these data.
[42] Several NRSROs said they base performance statistics for the
corporate, financial institution, and insurance company asset classes
on issuer ratings, because some of the issuers in these asset classes
are responsible for multiple issues, and the issuer's credit rating is
highly correlated with the ratings on its issues. They said if
performance statistics were based on issues for these asset classes,
they could be biased toward the ratings performance of large issuers.
On the other hand, they said that performance statistics for the
structured finance asset class are based on issues. One NRSRO
explained that this is because each issue has its own default
probability.
[43] Some NRSROs also presented lists of the firms that defaulted in
the last 10 years or over the time period for which they had ratings
histories, along with the initial rating assigned to the firm.
[44] For example, one NRSRO reported conditional default rates
relative to ratings at the beginning of a 3-year period for the 2007
cohorts. The NRSRO would group the rated entities by rating category.
For each cohort, the NRSRO then would calculate first-year, second-
year, and third-year survival rates. The first-year survival rate is
the number of entities that did not default in 2007 divided by the
number in the cohort. The number that survived the first year, 2007,
is the number in the cohort minus the number that defaulted in 2007.
Some NRSROs adjust for withdrawals by also subtracting the number of
entities with ratings withdrawn in 2007. The second-year survival rate
is the number of entities that survived the first year and did not
default in 2008 divided by the number that survived the first year.
The number of entities that survived the second year, 2008, is the
number that survived the first year minus the number that defaulted in
2008 (with some NRSROs also subtracting the number of entities with
ratings withdrawn in 2008). The third-year survival rate is the number
of entities that survived the second year and did not default in 2009
divided by the number that survived the second year. The 3-year
conditional default rate for a cohort is one minus the product of the
first-year, second-year, and third-year survival rates. In a
variation, one NRSRO reported conditional default rates relative to
initial ratings, which it calculated with a similar method.
[45] The Act mandated that SEC issue these rules within 9 months of
the date of enactment.
[46] Dodd-Frank Act § 932(q).
[47] Cantor and Packer demonstrated that the observed default rates of
bonds rates BBB or lower (typically the last rating in the investment-
grade category) vary over time for a single NRSRO. Richard Cantor and
Frank Packer, "The Credit Rating Industry," Quarterly Review, Federal
Reserve Bank of New York, 19, no. 2 (1994). Blume, Lim, and Mackinlay
find evidence that one rating agency applied more stringent rating
standards between 1978 and 1995, so that firms with the same
observable characteristics were assigned lower ratings in later years
than they were assigned in earlier years. See Marshall E. Blume, Felix
Lim, and A. Craig Mackinlay, "The Declining Credit Quality of U.S.
Corporate Debt: Myth or Reality?" Journal of Finance, 53, no. 4,
Papers and Proceedings of the Fifty-Eighth Annual Meeting of the
American Finance Association, Chicago, Illinois, January 3-5, 1998
(August 1998), pp. 1389-1413.
[48] We did not evaluate the adequacy of these disclosures.
[49] Oversight of Credit Rating Agencies Registered as Nationally
Registered Statistical Rating Organizations, 72 Fed. Reg. 6378 (Feb.
9. 2007).
[50] See 74 Fed. Reg. at 6482 (codified, as amended, at 17 C.F.R.
240.17g-2(d)(2) (2010)). The 10 percent disclosure requirement became
effective in August 2009. SEC did not apply the 10 percent disclosure
requirement to subscriber-pays NRSROs. SEC noted in the final rule as
subscriber-pays NRSROs make their ratings available only for a fee,
the rule requiring them to make 10 percent of their outstanding
ratings available for free could cause them to lose subscribers.
[51] See, 74 Fed. Reg. at 63863-65 (codified, as amended, at 17 C.F.R.
240.17g-2(d)(3) (2010). The 100 percent disclosure requirement became
effective in June 2010. As part of both rules, SEC also required that
the NRSROs make the ratings history data available on their Web sites
in eXtensible Business Reporting Language (XBRL) format. The XBRL
format is intended to provide a uniform standard format for presenting
the data and allow users to dynamically search and analyze the data.
SEC published the list of XBRL tags that the NRSROs must use to comply
with this requirement on August 27, 2010. The NRSROs have 60 days
after this date to publish the data using this format.
[52] CUSIP stands for the Committee on Uniform Securities and
Identification. A CUSIP number consists of nine characters that
uniquely identify a company or issuer and the type of security. CIK is
the unique number that SEC's computer system assigns to individuals
and corporations that file disclosure documents with SEC. CIK is an
acronym for Central Index Key. All new electronic and paper filers,
foreign and domestic, receive a CIK number.
[53] Some securitizations--such as RMBS--are divided into different
classes, or tranches. A tranche is a piece of a securitization that
has specified risks and returns.
[54] Sample size also may limit the kinds of comparative performance
statistics that can be developed. Transition and default rates are
more useful the larger the sample of data used to construct them. This
is particularly true of default rates because they are rare events and
may not be observed in samples that are too small. Some of the 10
percent samples have relatively small numbers of observations,
particularly the samples of smaller NRSROs. For example, three of the
samples had no observed defaults or impairments.
[55] Some academic studies evaluate the comparative performance of
NRSROs by observing instances where NRSROs offer ratings on the same
entity or security. It is unlikely that in the 10 percent samples two
or more NRSROs randomly will select the same entity or security for
inclusion in their samples, making studies of such "split" ratings
difficult.
[56] For example, some NRSROs provide both financial strength and
issuer credit ratings for issuers. These NRSROs varied by whether they
counted just the rated entity once, and not the separate ratings,
while other NRSROs count both ratings as part of their total
outstanding ratings.
[57] References to Ratings of Nationally Recognized Statistical Rating
Organizations, 73 Fed. Reg. 40088 (July 11, 2008) (Exchange Act
Proposing Release); References to Ratings of Nationally Recognized
Statistical Rating Organizations, 73 Fed. Reg. 40106 (July 11, 2008)
(Securities Act Proposing Release), References to Ratings of
Nationally Recognized Statistical Rating Organizations, 73 Fed. Reg.
40124 (July 11, 2008) (Investment Company Act Proposing Release); and
References to Ratings of Nationally Recognized Statistical Rating
Organizations, 74 Fed. Reg. 52374 (Oct. 9, 2009) (Proposed Rule; re-
opening of comment period; request for additional comments).
[58] The Joint Forum, Stocktaking on the Use of Credit Ratings (Basel,
Switzerland: June 2009).
[59] supra note 58.
[60] To maintain a stable share price, most money funds use the
amortized cost method of valuation or the penny-rounding method of
pricing permitted by rule 2a-7. Under the amortized cost method,
portfolio securities are valued by reference to their acquisition cost
as adjusted for amortization of premium or accretion of discount. 17
C.F.R. § 270.2a-7(a)(1). Share price is determined under the penny-
rounding method by valuing securities at market value, fair value, or
amortized cost and rounding the per-share net asset value to the
nearest cent on a share value of a dollar, as opposed to the nearest
one-tenth of 1 cent. 17 C.F.R. § 270.2a-7(a)(15). See also Valuation
of Debt Instruments and Computation of Current Price Per Share by
Certain Open-End Investment Companies (Money Market Funds), 48 Fed.
Reg. 32555 (July 18, 1983) (Final Rule) ("Release 13380") and
Investment Company Act Rel. No. 12206, 47 Fed. Reg. 5428, 5430 n. 5
(Feb. 5, 1982) (Proposed Rules) ("Release 12206").
[61] From 1971 to 2007, only one money market fund, Community Bankers
U.S. Government Fund, broke the buck. On September 16, 2008, the
Reserve Primary Fund broke the buck. The resulting investor anxiety
caused a near run on money market funds and on September 19, 2008, the
U.S. Department of the Treasury announced a program to insure the
holdings of any publicly offered eligible money market fund that paid
a fee to participate in the program to quell investor fears.
[62] Requisite NRSROs are defined as any two NRSROs that have issued a
rating with respect to a security or class of debt obligations of an
issuer, or if only one NRSRO has issued a rating with respect to such
security or class of debt obligations of an issuer at the time the
fund purchases or rolls over the security, that NRSRO.
[63] 17 C.F.R. § 270.2a-7. Short-term ratings refer to short-term
debt, which has a maturity of 397 days or less.
[64] 17 C.F.R. § 240.15c3-1. The net capital rule generally defines
net capital as a broker-dealer's net worth (assets minus liabilities),
plus certain subordinated liabilities, less certain assets that are
not readily convertible into cash, and less a percentage of certain
other liquid assets (for example, securities). In computing their net
capital, broker-dealers are required to deduct from their net worth
certain percentages of the market value of their proprietary
securities positions, known as a haircut. NRSRO ratings are used,
along with other factors, to determine the haircut for each security.
[65] See Investment Company Act Proposing Release and Exchange Act
Proposing Release, supra note 58.
[66] The proposal also would have changed the definition of "first-
tier security" to a security whose issuer the board has determined has
the "highest capacity to meet its short-term financial obligations."
Any eligible security not deemed first-tier would be deemed second
tier. Under the current rule 2a-7, as amended in February 2010, a
money market fund generally must limit its investments in second-tier
securities to no more than 3 percent of fund assets, with investment
in second-tier securities of any one issuer being limited to the half
of 1 percent of fund assets.
[67] Money Market Fund Reform, 75 Fed. Reg. at 10109-10120 (March 4,
2010)(Final Rule).
[68] Commercial paper is an unsecured short-term obligation with
maturities ranging from 2 to 270 days issued by banks, corporations,
and other borrowers.
[69] Non-convertible debt securities are securities that cannot be
exchanged for shares of stock from the issuing corporation.
[70] See generally, References to Ratings of Nationally Recognized
Statistical Rating Organizations, ( 74 Fed. Reg. 52374 (Oct. 9, 2009)
(Proposed Rule; re-opening of comment period; request for additional
comments)).
[71] See References to Ratings of Nationally Recognized Statistical
Rating Organizations, 74 Fed. Reg. 52358, 52371-73 (Oct. 9,
2009)(Final Rule)(codified, inter alia, at 17 C.F.R. §§ 240.3a1-1,
242.300, 242.301, 270.5b-3, 270.10f-3). These rules include rules
under the Exchange Act and under the Investment Company Act. The rules
under the Exchange Act include 3a1-1, 300, 301(b)(5) and 301(b)(6) of
Regulation ATS, and Forms ATS-R and PILOT. The rules under the
Investment Company Act include 5b-3 and 10f-3. SEC has not taken
further action on its remaining proposals to remove NRSRO references
from its rules and forms.
[72] The Dodd-Frank Act rescinds the exemption for NRSROs under Rule
436(g) of the Securities Act of 1933. Issuers of ABS are required to
disclose the credit ratings that are a condition of the issuance of
the ABS and the identity of the rating agency. Rule 436(g) had
provided that ratings assigned by an NRSRO (but not other credit
rating agencies) would not be deemed part of a registration statement
and the NRSRO would not be subject to liability as an expert for the
rating under the Securities Act. The Securities Act requires that an
expert who is named as having prepared a report in connection with a
registration statement must file a written consent with the
registration statement. Going forward, credit ratings assigned by an
NRSRO that are incorporated into registration statements or
prospectuses will require consent by that NRSRO since they will be
considered expert opinions. The Division of Corporation Finance issued
a no-action letter on July 22, 2010, stating it will not recommend an
enforcement action to the Commission if an issuer of ABS omits the
ratings disclosure required by Regulation AB from a prospectus that is
part of a registration statement relating to an offering of ABS. SEC
noted in the no-action letter that the NRSROs have indicated they are
not willing to provide their consent at this time. SEC issued the no-
action letter to facilitate ABS transactions.
[73] Not all of the 65 examinations had deficiencies related to the
minimum credit risk determination requirement and some had multiple
deficiencies in this area.
[74] Since the implementation of rule 2a-7, Investment Management has
provided guidance to money market funds about what it will and will
not accept as evidence of an adequate minimal credit risk
determination. In a May 8, 1990, letter to the industry, Investment
Management states that the focus of any minimal credit risk analysis
must be on those elements that indicate the capacity of the issuer to
meet its short-term debt obligations. The letter provides examples of
elements that the analysis could include. While funds or their
advisers are not required to have these specific elements in their
credit files, the guidance states that the determination that money
market fund portfolio investments present minimal credit risks must be
based on factors pertaining to credit quality in addition to the
rating assigned to such instruments by an NRSRO.
[75] OCIE staff told GAO that OCIE has approximately 450 staff
dedicated to examinations of investment advisors and funds. It does
not have a unit devoted specifically to conducting money market funds.
Currently there are approximately 11,500 registered advisers and 860
investment company complexes (with thousands of individual funds,
including money market funds). SEC staff estimates that there are less
than 150 fund complexes offering investors approximately 850-900
different money market funds.
[76] The net capital rule uses additional criteria to establish the
appropriate haircut for a security including time to maturity and type
of security (for example, government security, nonconvertible debt,
and preferred stock).
[77] A.M. Best, DBRS, Fitch, Japan Credit Rating Agency, Moody's,
Rating and Investment Information, and Standard & Poor's received
NRSRO designation prior to the Act. The Act nullified the no-action
letters and required them to subsequently register as NRSROs with SEC
when the NRSRO registration program became effective.
[78] Egan-Jones Ratings, Realpoint, and LACE are primarily subscriber-
pays NRSROs.
[79] IBCA, Inc.; Duff & Phelps Credit Rating Company; and Thomson
BankWatch were all recognized as NRSROs under SEC's prior no-action
letter process in 1990, 1982, and 1991, respectively. A fourth credit
rating agency, McCarthy, Crisanti, and Maffei, Inc., also received a
no-action letter recognizing it as an NRSRO in 1983, and was later
acquired by Duff & Phelps in 1991.
[80] According to a press release, Morningstar does not plan to
register as an NRSRO, but Realpoint will continue as an NRSRO.
[81] A.M. Best provides financial strength ratings on insurance
organizations and credit ratings on bonds and other financial
instruments that insurers and reinsurers issue, and recently has
expanded into ratings for financial institutions.
[82] The HHI is one of the market concentration measures that
government agencies, including the DOJ and FTC, use when assessing
concentration to enforce U.S. antitrust laws. DOJ and FTC often
calculate the HHI as the first step in providing insight into
potentially anticompetitive conditions for an industry. However, the
HHI is a function of firms' market shares, and market shares may not
fully reflect the competitive significance of firms in the market.
Thus, DOJ and FTC use the HHI in conjunction with other evidence of
competitive effects when evaluating market concentration.
[83] In this case, an NRSRO's market share is equal to its total
revenue divided by the sum of all NRSROs' total revenues.
[84] Data and valuation service includes in-depth market analysis for
particular market segments. Proxy service includes research,
recommendations, and voting services for domestic and foreign proxy
proposals.
[85] NRSRO applicants and registered NRSROs provide these data to SEC
as part of Form NRSRO. These data are not required to be made public
for each NRSRO.
[86] For the NRSROs that reported their total revenue not ending on
December 31 we estimated their revenue for the 12-month periods ending
December 31 of 2006, 2007, 2008, and 2009.
[87] In this case, an NRSRO's market share is equal to the number of
its outstanding ratings divided by the sum of each NRSRO's outstanding
ratings. Additionally, the HHI could be calculated using the number of
new ratings assigned by NRSROs, instead of outstanding ratings. Since
ratings on securities or issuers can be outstanding for many years, a
rating agency that issued a lot of ratings in the past might look
dominant, even if all the new ratings were being issued by different
companies. Currently, NRSROs are not required to provide data on new
ratings assigned on Form NRSRO.
[88] In this case, an NRSRO's market share is equal to the number of
issuers it rates divided by the sum of the numbers of issuers all
NRSROs rate. More than one NRSRO can produce a credit rating for an
issuer. Thus, the sum of the numbers of organizations rated by all
NRSROs likely will be greater than the total number of issuers with a
credit rating. For example, if three NRSROs rate the same issuer, then
all three of those NRSROs will count that company in their numbers of
rated issuers. We define market share this way so that NRSROs' market
shares sum to 100 percent. As a result, our concept of market share
differs from the concept of market coverage. An NRSRO's market
coverage would be the number of issuers it rates as a share of the
number of issuers with a credit rating. Since more than one NRSRO can
rate an issuer, NRSROs' market coverage can sum to more than 100
percent.
[89] Other differences between our alternative and baseline estimates
are in the HHIs for the corporate issuers and insurance companies
asset classes. The alternative assumption produces an estimate of the
HHI for the corporate issuers asset class in 2009 that is 12 percent
smaller than the baseline estimate. The alternative assumption also
produces estimates of the HHI for the insurance companies asset class
in 2008 and 2009 that are 13 percent and 9 percent, respectively,
smaller than the baseline estimates. The alternative assumption
produces HHIs that indicate that concentration in these asset classes
has declined more rapidly between 2007 and 2009 than the HHIs produced
by the baseline assumption indicate.
[90] We obtained data on U.S.-issued ABS from Asset-Backed Alert. In
this case, an NRSRO's market share is equal to the dollar value of
issuance it rates divided by the sum of the dollar value of issuance
that each NRSRO rates.
[91] In a basic CDO a group of loans or debt securities are pooled and
securities are then issued in different tranches that vary in risk and
return depending on how the underlying cash flows produced by the
pooled assets are allocated.
[92] The HHIs for U.S. sub-prime RMBS for 2009 and 2010 are both based
on a single deal rated by a single NRSRO.
[93] For all ABS, it is important to note that the number of new deals
decreased by 2,715 from 3,083 to 368 between 2006 and 2009,
respectively. We did not assess the reasons ABS issuance declined. AM
Best was recognized as a NRSRO under SEC's former no-action letter
process in 2005. Realpoint was registered as an NRSRO in 2008.
[94] In November 2008, the Federal Reserve Bank of New York created
the Term Asset-Backed Securities Loan Facility (TALF) to increase
credit availability and support economic activity by facilitating
renewed issuance of ABS. ABS issued under TALF had to be rated by two
TALF-eligible NRSROs. For ABS other than CMBS, TALF-eligible NRSROs
included DBRS, Inc, Fitch Ratings, Moody's, and Standard & Poor's. For
CMBS, TALF-eligible NRSROs also included Realpoint LLC. Realpoint has
been issuing surveillance ratings since 2001; it began issuing initial
ratings in December 2009.
[95] The number of CMBS deals decreased significantly over the review
period, from a high of 130 in 2004 to a low of 24 in 2008. From
January 2010 through December 2009, 51 deals were issued, illustrating
the freezing of the CMBS market. Ratings coverage data only reflect
public ratings provided by issuer-pays NRSROs and do not reflect CMBS
deals that are rated privately or ratings paid by investors.
[96] Amendments to Rules for Nationally Recognized Statistical
Organizations, 74 Fed. Reg. 63832, 63864-63865 (Dec. 4, 2009)(amending
17 C.F.R. § 240.17g-5).
[97] On May 14, 2010, Ratings and Investment, Inc. issued a press
release announcing it was withdrawing its NRSRO registration from the
ABS asset class, effective June 28, 2010.
[98] As a response to the financial crisis the Federal Reserve created
the Term Asset-backed Securities Loan Facility to restore the
securitization markets. The Federal Reserve program targeted
securitizations in the asset-backed classes, specifically ABS. See
GAO, Troubled Asset Relief Program Treasury Needs to Strengthen Its
Decision-Making Process on the Term Asset-Backed Securities Loan
Facility, [hyperlink, http://www.gao.gov/products/GAO-10-25]
(Washington, D.C.: Feb. 5, 2010).
[99] Herwig M. Langhor, "The Credit Rating Agencies and Their Credit
Ratings," address given to the Bond Market Association in Paris
February 2006.
[100] For examples of recent academic papers that discuss the role of
reputation in the credit rating industry see Lawrence J. White, "The
Credit Rating Industry: An Industrial Organization Analysis," NYU
Center for Law and University and Business Research Paper (April
2001), and Frank Partnoy, "The Paradox of Credit Ratings," University
of San Diego Law & Economics Research Paper #20. (2001).
[101] Fabian Dittrich, "The Credit Rating Industry: Competition and
Regulation," Social Science Research Network, July 2007.
[102] Market participants use the ratings of a particular NRSRO
because other market participants use it too. When network effect is
present, the value of a product or service increases as more people
use it.
[103] As previously noted, the recently adopted Dodd-Frank Act
required SEC and other federal agencies to remove references to NRSRO
ratings from its regulations and substitute an alternative standard of
creditworthiness.
[104] This asset manager told us that his firm is removing the
references of the three largest NRSROs in contracts and investment
guidelines at renewal. However, we do not know to what extent this may
be occurring.
[105] See appendix I for our literature review methodology.
[106] The researchers hypothesized that downgrades will be considered
worse news when the rating standards are low to begin with, and thus
larger magnitudes will imply lower-quality ratings.
[107] The subscriber-pays model is also subject to conflicts of
interests because there is the potential for investors to influence
the NRSRO to upgrade or downgrade securities the investors are holding
to their advantage. For example, a subscriber may want to hold only
investment grade securities because its investment guidelines makes
this a requirement. An upgrade to investment grade of a security would
allow the subscriber to hold that security.
[108] See appendix1 for a detailed discussion on how we identified
these models.
[109] Several variations of this model have been proposed by others.
[110] Under this proposed model, if the rating agency was part of a
larger company, interaction between the parent company and the rating
agency would be prohibited.
[111] The model as proposed did not specify how ratings fees were
determined, but suggested that issuers could negotiate with the NRSROs
to determine the rating fee, or the NRSROs could establish a fee
schedule for rating different kinds of securities.
[112] Transparency in this context does not refer to the transparency
or disclosure regime of the NRSROs but is specific to the transparency
of the compensation model only.
[113] However, we are not suggesting that the model reveal the actual
rating fees that are charged by an NRSRO.
[114] Section 15E(h) of the Exchange Act provides SEC with the
authority to implement rules for the management of conflicts of
interest relating to the issuance of credit ratings by NRSROs. 15
U.S.C. § 78o-7(h). However, we did not assess whether SEC could
implement any of the proposed alternative compensation models under
this authority.
[115] After submission of the report, SEC is authorized to issue
regulations establishing a system for the assignment of NRSROs to
determine initial credit ratings of structured finance products in a
manner that prevents the arranger from selecting the NRSRO that will
determine the credit rating. SEC is to give thorough consideration to
the provisions of the Senate-passed financial reform bill that would
have required an issuer desiring an initial credit rating for
structured finance products to submit a request to a credit rating
agency self-regulatory organization, which would select an NRSRO from
a qualified pool based on a selection method intended to reduce
conflicts of interest. SEC is to implement this system unless it
determines that an alternative system would better serve the public
and its investors.
[116] This act also mandates GAO to conduct a study on alternative
means for compensating NRSROs, with the intent of creating incentives
for NRSROs to provide accurate credit ratings. The Dodd-Frank Wall
Street Reform and Consumer Protection Act, Pub. L. No. 111-203, §939D,
124 Stat. 1376, 1888 (2010) (to be codified at 15 U.S.C. 78o-9 note).
[117] We downloaded Moody's data on February 12, 2010; Standard &
Poor's data on February 26, 2010; Fitch's data on March 1, 2010; A. M.
Best's, DBRS's, and Japan Credit Rating Agency's data on March 2,
2010; and Rating and Investment Information's data on March 8, 2010.
[118] The HHI is one of the concentration measures that government
agencies, including the Department of Justice (DOJ) and the Federal
Trade Commission (FTC) use when assessing concentration to enforce
U.S. antitrust laws. DOJ and FTC often calculate the HHI as the first
step in providing insight into potentially anticompetitive conditions
for an industry. However, the HHI is a function of firms' market
shares, and market shares may not fully reflect the competitive
significance of firms in the market. Thus, DOJ and FTC use the HHI in
conjunction with other evidence of competitive effective effects when
evaluating market concentration.
[119] We were unable to obtain data on the number of rated issuers
from one NRSRO that indicated that it did not keep track of which of
the organizations it rates that issue debt securities. However, this
NRSRO did provide us with the total number of organizations is rated.
To assess the sensitivity of these results to the missing data from
the NRSRO that did not track which of its rated organizations issue
debt securities, we recalculated the HHIs assuming that all of the
organizations this NRSRO reported rating issue debt securities. We did
so because it is likely that some of the organizations this NRSRO
rates do issue debt securities, but we cannot determine how many.
Calculating the HHIs based on the alternative assumption that all of
the organizations this NRSRO rates issue debt securities gives us a
range within which the true value of the HHI is likely to fall.
[120] We searched the EconLit, the National Bureau of Economic
Research (NBER) Working Paper Series, and the Social Science Research
Network (SSRN) databases. The three studies identified included:
Becker, Bo and Milbourn, Todd T., "Reputation and Competition:
Evidence from the Credit Rating Industry," Harvard Business School
Finance Working Paper No. 09-051, June 21, 2009; Benmelech, Efraim and
Dlugosz, Jennifer, "The Credit Rating Crisis," National Bureau of
Economic Research Working Paper Series No. 15045, June 2009; and
Doherty, Neil A., Kartasheva, Anastasia and Phillips, Richard D.,
"Competition Among Rating Agencies and Information Disclosure,"
February 13, 2009.
[121] Act of August 22, 1940, ch. 686, title II, § 203, 54 Stat. 847,
850 (codified, as amended, at 15 U.S.C. § 80b-3(c)). Section 203(c)(2)
of the Investment Advisers Act provides that within 45 days of the
date of the filing of an application for registration (or within such
longer period to which the applicant consents), the Commission shall
either grant such registration or institute proceedings to determine
whether registration should be denied. 15 U.S.C. § 80b-3(c)(2). These
proceedings must include notice of the grounds for denial under
consideration and opportunity for hearing. Such proceedings must
conclude within 120 days of the date of filing but can be extended by
the Commission for an additional 90 days if it finds good cause and
publishes its reasons or if the applicant consents.
[122] Form ADV has two parts. Part 1 contains information about the
adviser's education, business, and disciplinary history within the
last 10 years, and is filed electronically through FINRA's IARD
system. Part 2 includes information on an adviser's services, fees,
and investment strategies. Currently, SEC does not require advisers to
file Part 2 electronically.
[123] The appeal process does happen but FINRA staff stated it is not
frequent.
[124] FINRA provided an informal estimate of the percentage of
applications it completes within 180 days.
[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 Web site [hyperlink, http://www.gao.gov]. Each
weekday, GAO posts newly released reports, testimony, and
correspondence on its Web site. To have GAO e-mail you a list of newly
posted products every afternoon, 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 Web site,
[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.
To Report Fraud, Waste, and Abuse in Federal Programs:
Contact:
Web site: [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:
Ralph Dawn, Managing Director, dawnr@gao.gov:
(202) 512-4400:
U.S. Government Accountability Office:
441 G Street NW, Room 7125:
Washington, D.C. 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, D.C. 20548: