This is the html version of the file http://law.wustl.edu/WULQ/77-3/773-619.pdf.
G o o g l e automatically generates html versions of documents as we crawl the web.
To link to or bookmark this page, use the following url: http://www.google.com/u/WashULaw?q=cache:-nRPWwFcb6MJ:law.wustl.edu/WULQ/77-3/773-619.pdf+Partnoy&hl=en


Google is not affiliated with the authors of this page nor responsible for its content.
These search terms have been highlighted:  partnoy 

Page 1
619
Washington University
Law Quarterly
V
OLUME
77
N
UMBER
3
1999
ARTICLES
THE SISKEL AND EBERT OF FINANCIAL
MARKETS?:
TWO THUMBS DOWN FOR THE CREDIT
RATING AGENCIES
FRANK PARTNOY
*
I. I
NTRODUCTION
.......................................................................................... 620
II. A “R
EPUTATIONAL
C
APITAL
” V
IEW OF
C
REDIT
R
ATINGS
....................... 627
A.
Reputational Capital .................................................................... 628
B.
Development of Credit Ratings .................................................... 636
C.
The Modern Credit Rating Agency............................................... 649
III. P
ROBLEMS WITH THE
R
EPUTATIONAL
C
APITAL
V
IEW
............................ 655
A.
Credit Spread Estimation............................................................. 656
B.
Ratings-Driven Transactions ....................................................... 665
1.
Asset-Backed Securities........................................................ 666
2.
Derivative Product Companies ............................................ 670
3.
Financial Guarantees........................................................... 672
4.
Arbitrage Vehicles................................................................ 674
5.
Other Problems and Corrections ......................................... 676
C.
Credit Derivatives ........................................................................ 677
* Associate Professor, University of San Diego School of Law. J.D., 1992, Yale Law School. I am
grateful for comments on an earlier draft of this article by participants in a conference on Reexamining the
Regulation of Capital Markets for Debt Securities, sponsored by the Duke University Global Capital
Markets Center and The Bond Market Foundation, and held in Washington, D.C., on Oct. 18-19, 1999,
and at a faculty colloquium held at the University of San Diego on Jan. 15, 1999. In particular, I want to
thank Jim Cox, Lynne Dallas, Howard Schneider, Steven Schwarcz, Stephen Wallenstein, and Chris
Wonnell.

Page 2
620
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
1.
Credit Swaps ........................................................................ 678
2.
Credit Options...................................................................... 680
3.
Credit Notes ......................................................................... 681
IV. A “R
EGULATORY
L
ICENSE
” V
IEW OF
C
REDIT
R
ATINGS
........................ 683
A.
Regulatory Licenses ..................................................................... 683
B.
1930s Regulation.......................................................................... 688
C.
Post-1973 Regulation: The Creation of NRSROs ........................ 692
V. R
ECOMMENDATIONS
................................................................................ 706
A.
Substituting Credit Spreads for Credit Ratings............................ 706
B.
Market Risk Ratings ..................................................................... 709
C.
Recent Litigation .......................................................................... 711
VI. C
ONCLUSION
........................................................................................... 713
“There are two superpowers in the world today in my
opinion. There’s the United States and there’s Moody’s Bond
Rating Service. The United States can destroy you by
dropping bombs, and Moody’s can destroy you by
downgrading your bonds. And believe me, it’s not clear
sometimes who’s more powerful.”
1
“The most that we can safely assert about the evolutionary
process underlying market equilibrium is that harmful
heuristics, like harmful mutations in nature, will die out.”
2
I. I
NTRODUCTION
Is a AAA
3
rating of an institution’s bonds any different from a five-star
Morningstar
4
rating of a mutual fund, or a four-diamond American Automobile
1. The News Hour with Jim Lehrer: Interview with Thomas L. Friedman (PBS television
broadcast, Feb. 13, 1996) (transcript on file with author).
2. Merton H. Miller, Debt and Taxes, 32 J. F
IN
. 261, 273 (1977).
3. A credit rating of “AAA” generally is regarded as the highest quality rating; a credit rating of
“D” (signifying default) is the lowest. The various rating systems are described in detail infra at Part II.C.
Credit rating agencies provide ratings for credit instruments, typically bonds, although they also rate more
complex investments, including derivatives, whose value is based on or derived from some other
underlying financial instrument or index. See discussion infra at Part III.B.
4. Morningstar Inc., the Chicago-based publisher of mutual fund information, has a “star” rating
system of from one to five stars. Mutual fund managers and analysts whose funds earn the coveted “Five
Star” ratings from Morningstar earn the highest, seven-figure salaries. See Charles Gasparino, Mutual
Funds Show Managers The Money, Seeking Big Returns, W
ALL
S
T
. J., March 7, 1997, at C1. There are
only about 100 Five Star managers out of approximately 8,000 U.S. mutual funds. To gain a Five Star

Page 3
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
621
Association rating of a hotel, or a three-star Michelin rating of a restaurant, or a
“two-thumbs-up” Siskel and Ebert rating of a film, or a single UL
5
symbol on a
blender?
6
And if bond ratings differ from ratings of mutual funds, hotels,
restaurants, films, and appliances, to what extent have legal rules caused those
differences? To what extent should legal rules operate to minimize differences?
This article addresses these questions.
Credit ratings pose an interesting paradox. On one hand, credit ratings
purport to provide investors with valuable information they need to make
informed decisions about purchasing or selling bonds;
7
credit rating agencies
seem to have impressive reputations and most bond issues are rated by multiple
agencies. On the other hand, particularly since the mid-1970s, the informational
value of credit ratings has plummeted;
8
credit rating agencies, faced with the
challenges of globalized, technologically innovative markets and with
competition from providers of more current, detailed, and accurate information,
have become reactive rather than proactive, and some evidence indicates they
have maintained accurate credit ratings (i.e., ratings correlated with actual
default experience) due more to after-the-fact corrections than to predictive
power.
9
rating, a fund must place within the top 10 percent of its peer group for at least three years, based on its
performance and a measure of its risk versus return. See id. at C23. Some top mutual fund managers have
been compared to professional sports stars. See id.
5. I analyze the Underwriters Laboratories (UL) rating system infra at 685-86.
6. Institutions, not individuals, provide most ratings, presumably because institutions are better
able to capture economies of scale in gathering and providing information. Nevertheless, a few individuals
have thrived as raters, including the late Gene Siskel and Roger Ebert, the well-known film critics from
Chicago; Robert Parker, the international wine connoisseur; and numerous celebrity endorsers. Although
an individual, John Moody, invented and first issued credit ratings, and although some individual stock
analysts acquire a kind of “brand equity” due to their prominent, public role in rating securities (e.g.,
Abby Joseph Cohen, a partner of Goldman, Sachs & Co.), the financial market—and this article—are
centered on raters as institutions, not individuals.
7. Numerous studies conducted by both rating agencies and independent groups have demonstrated
that credit ratings have a high correlation with actual default experience. See discussion infra at 647-48.
8. There have been multiple unexpected defaults and sudden credit downgrades in recent years,
involving major issuers such as Orange County, Mercury Finance, and the governments and banks of
numerous countries in Latin American and Southeast Asia. See discussion infra at 658-62.
9. Many recent credit downgrades occurred after the rated entity already had disclosed
substantially increased risk. For example, the recent near-collapse of Long-Term Capital Management, the
Greenwich, Connecticut, hedge fund that reported losses of $4 billion in September 1998, prodded
Standard and Poor’s (“S&P”) and Moody’s to review the credit quality of all major U.S. and European
banks, and to downgrade the senior debt of one of them, Bankers Trust. See CNN Moneyline News Hour
with Lou Dobbs, at 10 (CNN Television Broadcast, Sept. 25, 1998) (available in LEXIS-NEXIS,
Curnews database; transcript on file with author) (addressing credit issues related to Long-Term Capital
Management); see also Frank Partnoy, Playing Roulette with the Global Economy, N.Y. T
IMES
, Sept.
30, 1998, at A17 (describing role of derivatives and hedge funds in the near-collapse of Long-Term
Capital Management). Such evidence supports arguments by economists that rating agencies’ claims that

Page 4
622
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
This paradox—continuing prosperity of credit rating agencies in the face of
declining informational value of ratings—has generated extensive debate both in
the popular press and among financial economists.
10
Since John Moody
introduced his clever and simple rating system for railroad bonds in 1909,
11
credit ratings have played a critical role in the development of modern financial
markets. Oddly, during two key time periods—the 1930s and the mid-1970s
through 1990s—despite the declining informational value of credit ratings, such
ratings became more, not less, important to financial market participants.
12
Particularly since the mid-1970s, credit ratings have assumed elevated status.
Not only do credit rating agencies now rate the vast majority of all new public
ratings correlate with default rates are seriously flawed, both logically and empirically, because (1) ratings
often are simply a response to negative information (i.e., correlation does not mean causation), and (2)
ratings changes tend to influence market prices (i.e., rating downgrades are partially self-fulfilling because
they cause price declines). See discussion infra at Part III.A. The credit crunch of late 1998 is additional
evidence in support of these views. In October 1998, the credit quality and liquidity of most bonds and
swaps plummeted, yet most credit ratings remained constant. See Thomas J. Sowanick, M
ERRILL
L
YNCH
F
IXED
I
NCOME
W
KLY
., Nov. 13, 1998, at 3 (copy on file with author).
10. A sample of articles from The Economist is representative. See, e.g., Credit-Rating Agencies.
AAArgh!, E
CONOMIST
, Apr. 6, 1996, at 80; Credit-Rating Agencies; Beyond the Second Opinion,
E
CONOMIST
, Mar. 30, 1991, at 80; Rating the Rating Agencies, E
CONOMIST
, July 15, 1995, at 53; Room
for Improvement: Ratings Agencies, E
CONOMIST
, July 15, 1995, at 54; Sovereign Debt; The Ratings
Game, E
CONOMIST
, Oct. 30, 1993, at 88; The Use and Abuse of Reputation, E
CONOMIST
, Apr. 6, 1996,
at 18.
Legal commentators typically have assumed ratings are valuable due to their information content.
Perhaps as a result, legal scholars have focused on questions ancillary to the core issue of why rating
agencies continue to prosper, and have shied from analyzing the substantial volume of finance literature
addressing the predictive value of ratings. See discussion infra at 647-48. In contrast, while finance
theorists have focused on core economic issues related to the informational value of ratings, they have not
addressed in detail the effects of regulation or law on ratings. See discussion infra at Part IV.C.
Regulators have expressed mixed views. Richard Roberts, then a Commissioner of the Securities and
Exchange Commission, first discussed the use of credit ratings in federal securities regulations in 1992.
See K. Susan Grafton, The Role of Ratings in the Federal Securities Laws, I
NSIGHTS
, Aug. 1992, at 22;
Roberts Says SEC Needs Oversight Authority of Credit Rating Services, S
EC
. W
K
., Apr. 13, 1992, at 3.
Soon thereafter, the Securities and Exchange Commission requested comment on the use of ratings in a
release proposing a new Rule 3a-7 under the Investment Company Act of 1940. See 57 Fed. Reg. 23,980
(1992). A series of releases and comments followed, but as of November 1999 no decision had been made.
See discussion infra at note 386.
11. I describe the evolution of rating agencies, including Moody’s scale of classification, in greater
detail at Part II.B.
12. Today, credit ratings are of paramount importance to debt issuers. A poor rating can drive up an
issuers borrowing costs or even put it out of business. For example, in February 1991, when S&P and
Moody's downgraded a $1.1 billion debt issue by Chrysler, its interest costs jumped by $38 million a year.
See Credit-Rating Agencies: Beyond the Second Opinion, supra note 10, at 80. Issuers of debt have
complained about the power of the rating agencies, and analysts from at least one large American
institution have called the concentration of power at Moody's and S&P “dangerous.” Id. More recently,
credit downgrades of the Japanese government and Japanese banks (which occurred months after market
prices already had reflected information about bad loans and debts) caused huge increases in borrowing
costs. See, e.g., Gillian Tett, Moody’s Downgrades Sovereign Debt, F
IN
. T
IMES
, Nov. 18, 1998, at 6.

Page 5
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
623
bond issues, but rating changes are important public events, even front-page
news. Both the press and the public now follow ratings very closely, which,
some scholars assert, shows the social value of the ratings.
13
In this article, I argue that the paradox of rating agencies is best explained
by the practice of linking substantive securities regulation to private credit
ratings. I argue that, from the mid-1970s until today, the Securities and
Exchange Commission (SEC) and other regulatory bodies, by promulgating
rules that depend substantively on credit ratings, have given the handful of
approved credit rating agencies—dubbed Nationally Recognized Statistical
Ratings Organizations (NRSROs)—a substantial degree of market power. In
particular, these rules permit NRSROs to grant “regulatory licenses,” a term I
use to describe the valuable property rights associated with the ability of a
private entity, rather than the regulator, to determine the substantive effect of
legal rules. Such regulatory licenses have benefitted rating agencies to varying
degrees since the 1930s.
14
I argue that the recent increase in the scope of these regulatory licenses not
only has caused substantial deadweight loss due to the agencies’ ensuing
oligopoly, but also has encouraged the rating agencies to shift from the business
of providing valuable credit information to the far more lucrative business of
selling regulatory licenses. The new regulatory scheme has had dramatic
effect,
15
not only causing a decline in the informational value of credit ratings,
13. Professor Kenneth Lehn has argued credit ratings must have substantial informational content
because of the hundreds of stories that appear in the financial press about bond rating changes issued by
the major rating agencies. See Letter from Kenneth Lehn, Professor of Business Administration,
University of Pennsylvania, to Jonathan G. Katz, Secretary, SEC 4 (Dec. 5, 1994) (available at SEC
office headquarters, file no. S7-23-94; copy on file with author). Similarly, finance scholars have argued
that higher-rated debt instruments have lower default rates. In particular, W. Braddock Hickman—whose
study later served as the scholarly basis for the development of the junk bond market during the 1980s—
found that Moody’s Aaa-rated debt had a default rate of 10 percent during 1900-43, while Ba-rated debt
had a default rate of over 40 percent during the same period. See W. B
RADDOCK
H
ICKMAN
, N
AT
L
B
UR
.
E
CON
. R
ESEARCH
, C
ORPORATE
B
OND
Q
UALITY AND
I
NVESTOR
E
XPERIENCE
(1958).
In recent years, the credit rating business has generated astonishing revenues and profits. Several of
the rating agencies are either privately held or are divisions of companies that do not separately report the
agencies’ financial results. However, there is evidence that the rating agencies have been able to maintain
very high profit margins. See discussion infra at 654. Yet, at the same time, the number of bankruptcies,
financial fiascos, and defaults that rating agencies have failed to anticipate has skyrocketed. I discuss
several of the recent shortcomings of credit ratings in Part III.
14. See discussion infra at Part IV.C.
15. As Thomas J. McGuire, then Executive Vice President and Director-Corporate Development of
Moody’s Investors Service, perhaps the premier credit rating agency, put it: “[T]he weight of government
regulators on private institutions can be enormous. Eight hundred pound gorillas should be careful where
they sit.” Thomas J. McGuire, Ratings in Regulation: A Petition to the Gorillas, Delivered to the SEC
Fifth Annual International Institute for Securities Market Development, at 17 (April 28, 1995) (copy on
file with author).

Page 6
624
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
but also creating incentives for the agencies to provide inaccurate ratings and for
market participants to pay for regulatory entitlements stemming from the
agencies’ ratings, instead of paying for the informational content of the ratings.
The result is a bewildering array of dysfunctional financial behavior as well as
substantial financial market distortion and inefficiency. One prominent market
participant aptly labeled these phenomena a “chronic sickness” in the financial
markets.
16
The solution I offer for this problem is simple, perhaps bitter, medicine. The
SEC and other regulatory agencies should remove each of these regulatory
licenses by excising the portions of their rules that depend substantively on
credit ratings. Regulators also should resist the temptation to include such
regulatory licenses in new regulation. For example, regulators and market
participants (especially rating agencies) recently have argued that a rating
system similar to the bond rating system should be adopted for mutual funds.
17
The findings in this article raise serious questions about the wisdom of
institutionalizing any such rating system for mutual funds.
In place of ratings-dependent regulation, I recommend a replacement: simply
substitute credit spreads,
18
the market risk measure of bonds, for credit ratings.
Credit spreads are more accurate than credit ratings, and by definition credit
spreads reflect the market price of credit, which should reflect at minimum the
information contained in credit ratings.
This article and its recommendations are a contribution to the recent shift in
16. Mr. McGuire of Moody’s expressed this view:
I should tell you up-front that the prognosis is not good. My comments today are intended, very
deliberately, to be a serious warning to you of a chronic sickness that is affecting the rating agency
business. The cause of that sickness is the use by regulators of ratings as instruments of financial market
regulation.
McGuire, supra note 15, at 11. It is odd that a senior official of a leading rating agency would be leading
the charge against regulations that, as I argue here, preserve the agencies’ oligopoly power. In fact,
officials from Moody’s seem not to be worried about competition from the other members of the oligopoly,
and may be confident in the belief that Moody’s would be better off under a competitive regime. Moody’s
also seems to fear additional regulation. Whether Moody’s could acquire monopoly power absent ratings-
dependent regulation is a question beyond the scope of this article, but I should note that arguments of
some legal scholars about learning-related externalities could be applied to the market for bond ratings.
See, e.g., Marcel Kahan & Michael Klausner, Standardization and Innovation in Corporate
Contracting, 83 V
A
. L. R
EV
. 713 (1997); Marcel Kahan & Michael Klausner, Path Dependence and
Comparative Corporate Governance: Path Dependence in Corporate Contracting, Increasing
Returns, Herd Behavior and Cognitive Biases, 74 W
ASH
. U. L.Q. 347 (1996).
17. See Mutual Funds: X-Rated, E
CONOMIST
, August 22, 1998, at 58 (describing proposals by
regulators outside the U.S. to require that all mutual funds be rated).
18. I discuss credit spreads in detail infra at Part V.A.

Page 7
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
625
legal scholarship on the financial markets,
19
in tandem with recent shifts by
economists, finance theorists, and market participants, away from stocks and
“securities”
20
to other credit and derivative instruments. Following the passage
of the securities laws in 1933 and 1934, there was very little criticism of stock
(or bond) market regulation for nearly 30 years. In 1964, George Stigler began
the debate about the efficiency of federal securities regulation with a study
demonstrating there was very little difference between the return performance of
unregistered and registered stocks.
21
Since Stigler’s study, much of the debate among legal scholars in this area
has addressed stocks and information.
22
Equity securities have occupied the core
of corporate and securities law: stocks carry ownership rights, which are the
subject of fundamental questions about the separation of ownership and control
of corporations and the market for corporate control; stocks are thought to carry
greater risk and therefore greater expected returns, and have proved more
susceptible to manias, panics, and crashes, than other instruments; new stock
issues, in particular initial public offerings, have generated concerns, regulation,
and commentary related to fraud and mispricing; stocks and stock options are
the primary focus of insider trading enforcement efforts and historically have
19. I use the term “financial markets” to describe the markets (both on organized trading exchanges
and through private over-the-counter transactions) for transactions in financial instruments, such as stocks,
bonds, foreign exchange, commodities, and more sophisticated instruments. I label these markets
“financial” rather than “capital” because although they necessarily involve the former concept, they need
not involve the latter; in particular, numerous markets in financial derivatives cannot be said to involve
any type of claim on capital.
20. The term “security” is defined to include certain credit instruments, namely “bonds,” see 15
U.S.C. § 77b(1) (1994), and other investment contracts, see, e.g., SEC v. W.J. Howey Co., 328 U.S. 293
(1946), but it does not include trillions of dollars of other financial instruments, including derivatives. See
Frank Partnoy, Financial Derivatives and the Costs of Regulatory Arbitrage, 22 J. C
ORP
. L. 211, 249-
52 (1997).
21. See George J. Stigler, Public Regulation of the Securities Markets, 37 J. B
US
. 117 (1964).
Stigler studied a sample of new issues of common stock covering two periods, 1923-28 and 1949-55, and
concluded that “investors in common stocks in the 1950’s did little better than [investors] in the 1920’s,
indeed clearly no better if they held the securities only one or two years.” Id. at 121. A more thorough
review of Stigler's work by Gregg Jarrell found that although the pre-1933 market for new equity issues
was, for the most part, efficient, the default risk and risk premia of bonds has fallen from pre- to post-SEC
regulation. See Gregg A. Jarrell, The Economic Effects of Federal Regulation of the Market for New
Security Issues, 24 J.L. & E
CON
. 613, 666-67 (1981).
22. See, e.g., John C. Coffee, Jr., Market Failure and the Economic Case for a Mandatory
Disclosure System, 70 V
A
. L. R
EV
. 717 (1984); Frank H. Easterbrook & Daniel R. Fischel, Mandatory
Disclosure and the Protection of Investors, 70 V
A
. L. R
EV
. 669 (1984); Jeffrey N. Gordon & Lewis A.
Kornhauser, Efficient Markets, Costly Information, and Securities Research, 60 N.Y.U. L. R
EV
. 761
(1985); Jarrell, supra note 21; Marcel Kahan, Securities Law and the Social Costs of “Inaccurate”
Stock Prices, 41 D
UKE
L.J. 977 (1992); Paul G. Mahoney, Mandatory Disclosure as a Solution to
Agency Problems, 62 U. C
HI
. L. R
EV
. 1047 (1995); Thomas A. Smith, Institutions and Entrepreneurs
in American Corporate Finance, 85 C
AL
. L. R
EV
. 1 (1997).

Page 8
626
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
been susceptible to manipulation by traders; and perhaps most importantly, in
the 1970s and 1980s, economists and finance theorists focused on stocks, and
scholarship from that era has served as the backbone for much related, yet often
belated, legal scholarship.
In contrast, financial economists have shifted much attention to the credit
and derivative markets.
23
The growth in credit markets and instruments has been
both rapid and elemental.
24
Since the 1970s, credit markets have been the
driving force in finance, and have generated huge trading volumes, massive
profits, and a potpourri of fascinating legal and regulatory issues.
25
In the past
decade or so, the bond trading floor, far more than the New York Stock
Exchange (or any other equity exchange), has been the center and focus of
financial intermediation.
26
In 1996, 17,614 bond deals were sold in the U.S.
23. Legal scholars’ analysis of the bond market has been less thorough than that of the stock market,
in part because of difficulties associated with collecting bond return data, and in part because bonds have
been regarded as “uniformly low-risk securities compared with common stock.” See Jarrell, supra note
21, at 652.
24. This growth is especially evident outside the U.S. The vast majority of financial assets in
developing countries are debt, e.g., bonds and loans. For example, in Latin America, more than 90 percent
of financial assets are held by commercial banks in the form of fixed income instruments. The vast
majority of developing country financial assets are bonds and loans. One reason for the dominance of debt
instruments in such countries is that most privately-owned corporations in these countries are either
closely-held or not widely held by public shareholders. See discussion infra at 666-68.
25. A few legal commentators have addressed some of these issues in credit markets. Recently,
Marcel Kahan and Michael Klausner have argued that certain provisions in debt contracts are inefficient
due to market failures generated by the path-dependent nature of such provisions over time. See Kahan &
Klausner, Standardization and Innovation in Corporate Contracting, supra note 16; Kahan &
Klausner, Path Dependence and Comparative Corporate Governance, supra note 16. The broad topic
of securitization has received extensive treatment, see, e.g., S
TEVEN
L. S
CHWARCZ
, S
TRUCTURED
F
INANCE
: A G
UIDE TO THE
P
RINCIPLES OF
A
SSET
S
ECURITIZATION
(2d ed. 1993); Steven L. Schwarcz,
The Global Alchemy of Asset Securitization, 1 S
TAN
. J.L. B
US
. & F
IN
. 133, as has the even broader topic
of financial derivatives regulation, see, e.g., Henry T.C. Hu, Hedging Expectations: “Derivative
Reality” and the Law and Finance of the Corporate Objective, 73 T
EX
. L. R
EV
. 985, 996-1000 (1995);
Kimberly D. Krawiec, Derivatives, Corporate hedging, and Shareholder Wealth: Modigliani-Miller
Forty Years Later, 1998 U. I
LL
. L. R
EV
. 1039 (1998). Roberta Romano, A Thumbnail Sketch of
Derivative Securities and their Regulation, 55 M
D
. L. R
EV
. 1 (1996); Lynn A. Stout, Betting the Bank:
How Derivatives Trading Under Conditioins of Uncertainty Can Increase Risks and Erode Returns in
Financial Markets, 21 J. C
ORP
L. 53 (1995). In each of these areas, legal scholars have contributed
significantly to the debate and have been crucial in the formulation of important policy.
26. Although individual investors continue to pour money into equities through mutual fund
investments, investors also have increased holdings of bonds, as evidenced by the extraordinary
movements of capital from bank deposits to money market funds. Most financial advisors recommend that
investors keep a significant portion of their funds in bonds or cash. As of 1997, the total market
capitalizations of the U.S. public stock markets and the U.S. public bond markets were approximately
equal (about $12-13 trillion each), and were far less than the total volume of U.S. bank debt, or the total
notional value of over-the-counter (OTC) derivative contracts (more than $55 trillion). See, e.g., F
RANK
P
ARTNOY
, F.I.A.S.C.O.: B
LOOD IN THE
W
ATER ON
W
ALL
S
TREET
15 (1997). Outside the U.S., stock
markets occupy a much smaller fraction of total capital market size and activity.

Page 9
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
627
public markets.
27
Financial market participants increasingly have shifted from
traditional stock-related activity into new markets, capturing extraordinary
profits and creating even more remarkable financial dislocation.
28
Finally, in this article, I hope to generalize several notions of how financial
markets should be regulated, if at all. In particular, I hope to encourage debate
about mandatory disclosure schemes for all financial instruments from a new
perspective: how should we regulate the provision of information about credit?
On a more fundamental level, I also hope to suggest a model for assessing the
benefits and costs of the privatization of regulatory functions, i.e., functions
where the regulator remains involved in an oversight or gatekeeper role but
delegates important rights and obligations to private entities.
29
In Part II of this Article, I examine the history of credit ratings, with a focus
on the particular view of credit ratings I call the “reputational capital” view. In
Part III, I raise some problems with the reputational capital view. In Part IV, I
propose a new theory of credit markets and information, based on the notion of
a “regulatory license.”
30
In Part V, I offer some recommendations for changing
the regulation of credit rating agencies. Part VI concludes.
II. A “R
EPUTATIONAL
C
APITAL
” V
IEW OF
C
REDIT
R
ATINGS
I begin by describing one view, which I believe is the prevailing one, of the
role of credit ratings in financial markets, based on the agencies’ ability to
generate and retain reputational capital. First, I discuss the role of reputational
27. See Charles Gasparino, Moody’s Bond-Rating Lead is Shrinking, W
ALL
. S
T
. J., Jan. 14, 1997,
at C1. In 1996, Moody’s rated the largest number of bonds, 344 more than S&P. See id.
28. See Partnoy, supra note 20, at 212. Of course, large profits from equity-related activities are not
uncommon; in particular, investment banking profits from sales of equity derivatives and large block
equity trades have been both increasing and astonishingly large. See, e.g., Goldman Sachs Earns A Quick
$15 Million In Sale Of BP Shares, W
ALL
S
T
. J., May 16, 1997, at A4.
29. Such regulatory systems also present difficult questions about the constitutionality of delegations
to private regulators. An analysis of this question, first presented to me by Professor Larry Alexander, is
beyond the scope of this article. It is worth noting, however, that although parties in recent litigation
involving the credit rating agencies have not brought such challenges, there is case law supporting an
argument that the scope of the delegation of regulatory authority to the agencies would not pass
constitutional muster. See, e.g., Larkin v. Grendel’s Den, Inc., 459 U.S. 116 (1982) (state statute giving
churches and schools the right to veto applications for nearby liquor licenses unconstitutional under the
Establishment Clause); Washington v. Roberge, 278 U.S. 116 (1928) (zoning ordinance requiring
approval of neighbors for building project unconstitutional); Eubank v. Richmond, 226 U.S. 137 (1912)
(ordinance establishing building lines upon request of residents unconstitutional); cf. Thomas Kusac Co.
v. Chicago, 242 U.S. 526 (1917) (upholding zoning ordinance requiring consent of neighboring property
owners).
30. The core of the economic model, which applies generally outside of the credit rating context, is
described in Part IV.A.

Page 10
628
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
capital in markets generally, and in credit rating markets specifically. Second, I
set forth certain early historical developments related to the role of credit ratings
and the development of credit rating agencies, and explain how they fit (or do
not fit) this view. Third, I describe the recent explosion in size and importance
of credit rating agencies.
A. Reputational Capital
Economists since Adam Smith have noted the value of reputational capital in
sustaining a self-policing society.
31
Individuals
32
acquire reputations over time
based on their behavior.
33
If an individual’s reputation improves, and other
members of society begin to hold that individual in higher esteem, that
individual acquires a stock of reputational capital, a reserve of good will, on
which other parties rely in transacting with that individual. Reputational capital
leads parties to include “trust” as a factor in their decision-making; trust enables
parties to reduce the costs of reaching agreement. Reputational capital is
especially valuable when a small number of actors interact repeatedly.
34
In such
situations, cooperation among individuals can prevail even without a
government authority, as players learn information about other players’
31. Smith described his views in a 1763 lecture, “Of the Influence of Commerce on Manners”:
Of all the nations in Europe, the Dutch, the most commercial, are the most faithful to their word. The
English are more so than the Scotch, but much inferior to the Dutch. . . . This is not at all to be imputed to
national character, as some pretend; there is no natural reason why an Englishman or a Scotchman
should not be as punctual in performing agreements as a Dutchman. It is far more reducible to self-
interest, . . . [which] is as deeply implanted in an Englishman as a Dutchman. A dealer is afraid of losing
his character, and is scrupulous in observing every engagement. When a person makes perhaps twenty
contracts a day, he cannot gain so much by endeavouring to impose on his neighbours, as the very
appearance of a cheat would make him lose. When people seldom deal with one another, we find that
they are somewhat disposed to cheat, because they can gain more by a smart trick than they can lose by
the injury which it does their character.
A
DAM
S
MITH
, L
ECTURES ON
J
USTICE
, P
OLICE
, R
EVENUE
,
AND
A
RMS
253-54 (Edwin Cannan ed., 1896).
Much of the modern study of reputational economics has involved game theory. For a description of the
game theory view of the effects of reputation and repeated games, see D
OUGLAS
G. B
AIRD ET AL
, G
AME
T
HEORY AND THE
L
AW
, 159-87 (1994); see also David M. Kreps et al., Rational Cooperation in the
Finitely Repeated Prisoners’ Dilemma, 27 J. E
CON
. T
HEORY
245 (1982); David M. Kreps & Robert
Wilson, Reputation and Imperfect Information, 27 J. E
CON
. T
HEORY
253 (1982).
32. I include in my definition of “individuals” not only human beings, but other legal persons such
as corporations, partnerships, trusts, and other entities.
33. For a more complete economic analysis of the acquisition of reputational capital by debt issuers,
see Douglas W. Diamond, Reputational Acquisition in Debt Markets, 97 J. P
OL
. E
CON
. 828 (1989).
34. F.A. Hayek has argued that such cooperation is a human instinct developed during the millions
of years homo sapiens spent in small roving bands adapting to favor cooperation based on the “interaction
of fellows known to and trusted by one another.” F.A. H
AYEK
, T
HE
F
ATAL
C
ONCEIT
: T
HE
E
RRORS OF
S
OCIALISM
11 (W.W. Bartley III ed., 1988).

Page 11
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
629
strategies (i.e., their reputations).
35
Reputational capital and ratings are closely related. Rating services such as
Michelin or Consumer Reports survive and prosper based on their ability to
acquire and retain reputational capital. Raters who invest more in their
investigative and decision-making processes (and who therefore generate more
accurate and valuable ratings) acquire greater reputational capital. Individuals
and institutions look to a rater’s accumulated reputational capital in deciding
whether to rely on the assessment of the rater or, instead, to undertake
independent investigation. Absent other factors, the consumer of a product will
purchase a rating if the expected benefit of the rating minus the actual cost of
the rating is both positive and greater than the expected benefit of an
independent investigation minus the actual cost of such an investigation.
36
Reputational capital and credit ratings are even more closely related, for two
reasons. First, the concept of “credit”,
37
broadly defined as the promise to pay in
the future, includes notions of trust and credibility.
38
In a market economy,
financial markets allocate the supply of and demand for credit, and thereby
determine the price of various types of credit risk. Lenders make lending
35. See, e.g., R
OBERT
A
XELROD
, T
HE
E
VOLUTION OF
C
OOPERATION
145-46 (1984). The
reputation of a player typically is established by others observing the interactions of that player with other
players. Axelrod describes reputation as a means of escaping the Prisoner’s Dilemma, a game in which
two prisoners fail to cooperate even though cooperation would be advantageous to each, or as a means of
achieving deterrence, particularly for government. For example, Axelrod cites Britain’s decision to take
back the Falkland Islands in response to the Argentine invasion as enhancing Britain’s reputation for being
provocable, and the U.S.’s decision to commit major combat forces to Vietnam as enhancing the U.S.’s
reputation for threatening a major war in response to a power grab by the former Soviet Union. See id. at
150-54.
36. Accordingly, raters should exist in those markets where the cost of investigation to the rater is
less than the sum of costs of investigations to consumers, i.e., in markets where there are economies of
scale in rating. The difference between investigation costs to the rater and to the consumer can be thought
of as “surplus” which may be divided between the rater and consumers, depending on the competitive
dynamics in the market for ratings. In a competitive market, the net marginal benefit to the rater from
undertaking additional investigation will equal the net marginal benefit to the consumer from undertaking
additional investigation.
37. The concept of credit predates the existence of financial markets and is more fundamental than
the relatively recent conceptions of stocks, bonds, and other financial instruments. Credit exists whenever
there is intertemporal trade and is a necessary component of any market, financial or otherwise. Human
beings have been extending credit to each other for thousands of years, and basic credit arrangements exist
even where there are no financial markets to allocate credit. See Catherine Mansell-Carstens, Popular
Financial Culture in Mexico: The Case of the Tanda, in C
HANGING
S
TRUCTURE OF
M
EXICO
:
P
OLITICAL
, S
OCIAL
,
AND
E
CONOMIC
P
ROSPECTS
77 (Laura Randall ed., 1996) (describing the “tanda” or
“rosco,” an informal mechanism used by members of many small Mexican villages to allocate credit).
38. The following are two definitions of “credit” given by Standard & Poor’s: “1. trust; credibility.
2. Confidence in the ability and the intention of a person or institution to pay, based on its solvency and
probity, thereby entitling it to be trusted in borrowing or buying.” S
TANDARD
& P
OOR
S
D
EBT
R
ATINGS
C
RITERIA
: I
NDUSTRIAL
O
VERVIEW
iii (1986) [hereinafter D
EBT
R
ATINGS
C
RITERIA
].

Page 12
630
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
decisions based on the perceived riskiness of borrowers. Both lenders and
borrowers consider trust and credibility: in equilibrium, the supply of
39
and
demand for
40
funds at a particular risk level results in an equilibrium cost of
capital for the borrower. The quantity and quality of information market
participants require to make such decisions varies based on the nature of the
relationship with the counterparty, the availability of the information, and other,
often intangible, factors. In practice, borrowers and lenders use shorthand
approximations for credit. A person may be a “good” or “bad” credit risk. A
bank may establish size and time limits for extension of credit, based on
categories of creditworthiness. Mortgage lenders may use formulas to determine
eligibility or to cap the amount of a loan. In each case, the credit risk a lender
perceives depends in large part on the reputation of the borrower.
Credit ratings are closely related to reputational capital for a second,
derivative, reason: the success and function of a credit rating agency also
depends on trust and credibility.
41
Each credit rating agency depends for its
livelihood on its reputation for objectivity and accuracy.
42
Consider the
following description of the credit rating business by S&P: “Credibility is
fragile. S&P operates with no governmental mandate, subpoena powers, or any
other official authority. It simply has a right, as part of the media, to express its
opinions in the form of letter symbols.”
43
If this view of the credit rating agencies is correct,
44
then credit ratings are
simply opinions, not unlike a restaurant star rating from Michelin. Credit ratings
respond to investors’ demand for information about risks associated with fixed
income investments. By specializing in the gathering, analysis, examination, and
39. The supply of funds depends on the compensation for lending and the riskiness of the loan. The
riskiness of the loan depends on future contingencies that typically cannot be specified based on past
information, a predicament the rating agencies recognize. See id. (“In the analytical experience, we are
constantly reminded that the past is less and less prologue to the future.”) (emphasis in original).
40. The demand for funds depends on the cost of borrowing. For example, if A wanted to borrow
$100 from B for a year, the two might agree on an interest rate of, for example, ten percent, payable in one
year. This interest rate would have two components: (1) a risk-free rate, which is composed of a real
interest rate plus an expected inflation component, and (2) a credit spread. With a short-term risk free rate
of around five percent, the rate B would charge A likely would range from six to fifteen percent,
depending on the term and security of the loan. The resulting credit spread of one to ten percent would be
compensation for the chance that A might default on the loan. See discussion infra at 655-56.
41. The President of Moody’s, John Bohn Jr., recently proclaimed, “We’re in the integrity business:
People pay us to be objective, to be independent and to forcefully tell it like it is.” Richard House, Ratings
Trouble, I
NSTITUTIONAL
I
NV
., Oct. 1995, at 245. Similarly, an S&P publication states, “Ratings are of
value only so long as they are credible.” D
EBT
R
ATINGS
C
RITERIA
, supra note 38, at 3.
42. See Credit-Rating Agencies. AAArgh!, supra note 10, at 80.
43. See D
EBT
R
ATINGS
C
RITERIA
, supra note 38, at 3.
44. As demonstrated in Part IV, I believe S&P’s description of its business is incorrect.

Page 13
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
631
dissemination of such information, credit rating agencies eliminate the
duplicative and wasteful (i.e., inefficient) efforts of individuals engaging in such
activities.
45
According to this view, credit ratings are a competitive, reputation-
driven business, and agencies will survive only to the extent they are accurate
and reliable in assessing the credit risks of borrowers.
46
Financial economists have viewed bond credit ratings variously as screening
mechanisms for information unavailable publicly,
47
as attempts to distinguish
issuers of inferior quality and thereby avoid “average quality pricing,”
48
and as
proxies for other economic variables and statistics, such as state and municipal
borrowers, total net direct debt, per capita debt, unemployment rate, and median
home value, that closely relate to issuer quality.
49
In the classical view of the market for financial information, there is both a
supply of and a demand for information. Investors demand information about
the likelihood of receiving timely payments on financial instruments. Securities
law scholars have concluded that a market, efficient or not, should be in
informational equilibrium. In other words, investors should not only lack
incentives to change their portfolios, they should also lack incentives to change
their information acquisition strategies.
50
Financial economists maintain that the value of a bond rating lies in its
certification of the debt issue’s credit quality.
51
For example, one study found
45. For the classic statement of how individuals may overinvest in the production of information, at
substantial social cost, see Jack Hirschleifer, The Private and Social Value of Information and the
Reward to Inventive Activity, A
M
. E
CON
. R
EV
., June 1971, at 561.
46. According to Thomas McGuire, formerly an executive vice president at Moody's, “[w]hat's
driving us is primarily the issue of preserving our track record. That's our bread and butter.” House, supra
note 41, at 246.
47. See, e.g., Joseph E. Stiglitz, The Theory of “Screening”, Education, and the Distribution of
Income, A
M
. E
CON
. R
EV
., June 1975, at 283.
48. See, e.g., George A. Akerlof, The Market for “Lemons”: Qualitative Uncertainty and the
Market Mechanism, 84 Q.J. E
CON
. 488 (1970).
49. See, e.g., Pu Liu & Anjan V. Thakor, Interest Yields, Credit Ratings, and Economic
Characteristics of State Bonds: An Empirical Analysis, J. M
ONEY
, C
REDIT
& B
ANKING
, Aug. 1984, at
344, 345-48.
50. See Gordon & Kornhauser, supra note 22, at 786. Professor John Coffee has explained this
equilibrium as follows:
Analysts should invest in verifying and obtaining material information about corporate securities until the
marginal cost of this information to them equals their marginal return. Ordinarily, this private
equilibrium should also result in allocative efficiency: social resources would be devoted to information
verification until the social costs rose to meet the social benefits.
See Coffee, supra note 22 (arguing in part that rating agencies are few in number and have conflicts of
interest that may prevent them from acting competitively).
51. See, e.g., L. Paul Hsueh & David S. Kidwell, Bond Ratings: Are Two Better Than One?, F
IN
.
M
GMT
., Spring 1988, at 46, 47.

Page 14
632
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
that the purchase of a second credit rating provided additional information and
therefore reduced borrowing costs, even if the two ratings were the same.
52
Another study found that rating agencies rated corporate tax-exempt debt
backed by a standby letter of credit based primarily on the irrevocable
commitment of the bank issuing the letter of credit.
53
This finding suggests that
banks are superior to rating agencies in their certification function, although the
bank’s reputation for quality may not be transmitted directly to the market.
54
In
other words, obtaining a standby letter of credit from a bank reduces the interest
cost to the borrower, but it does so primarily because of the reaction of the
rating agencies.
55
Rating agencies may exist because of information asymmetry between debt
issuers and investors. Information asymmetry exists in markets where sellers
have superior information to buyers about product quality, yet cannot costlessly
convey this information to buyers.
56
If buyers are economically rational, prices
in a market with information asymmetry will reflect the average quality of a
product, and sellers with superior products will bear the cost of the information
asymmetry. Consequently, sellers in such a market will have an incentive to
disclose the superior nature of their product so that they can receive the highest
price.
57
How can sellers in capital markets do this? One way is for sellers of capital
to use credit ratings to signal to the market the credit quality of their debt issues.
Under such a theory, credit ratings provide a signal about the credit quality of
an issuer to potential investors in a market. Investors infer the actual credit
quality of a bond from the credit rating, and price the bond accordingly. The use
of a third party for signaling reduces the moral hazard problem of transferring
such information directly.
58
52. Interestingly, this study also found that the reduction in borrowing cost was greater when the
second rating was different from the first. See id. The study concluded that the second rating provided
additional information to the market and reduced borrowing costs because the value of the additional
information exceeded the cost of obtaining a second rating. Even though borrowers cannot predict what
the second rating will be and normally apply for both ratings simultaneously, the expected savings from
obtaining two credit ratings, whether identical or split, should outweigh the added rating fee expense. See
id. at 52 n.8.
53. See Roger Stover, Third-Party Certification in New Issues of Corporate Tax-Exempt Bonds:
Standby Letter of Credit and Bond Rating Information, F
IN
. M
GMT
., Spring 1996, at 62-63.
54. See id. at 64.
55. See id. at 65. This study found that the total effect of employing a standby letter of credit was a
17-basis-point-reduction in the new issue interest cost. See id.
56. See Hsueh & Kidwell, supra note 51, at 46.
57. See id. at 47.
58. Moral hazard refers to the risk that sellers may gain some advantage by exaggerating their credit

Page 15
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
633
Information-gathering agencies may acquire and process information for the
purpose of certifying asset quality. Three criteria must be satisfied for
certification to be credible to outside investors. First, the certifying agent must
have reputational capital at stake in the certification activity. In other words, the
certifying agent would suffer a loss of future relationships because of reduced
trustworthiness if it suggested a fair market value in excess of the offering price.
Second, the loss in reputational capital must exceed the gain possible from false
certification. Third, the agent’s services must be costly and the cost must be
related to the asymmetric information associated with the issuing firm.
59
According to the reputational capital view of credit rating agencies, these
three criteria are satisfied.
60
First, credit rating agencies have reputational
capital at stake in issuing ratings. Second, rating agencies would lose more in
reputational capital from giving false ratings than they would gain in increased
fees. Third, ratings are costly, and, if they contain valuable information, are
needed to overcome information asymmetry between issuers and investors.
Of course, certifying agencies may include not only credit rating agencies,
but other third-party providers of credit information, including bond analysts
and industry periodicals. Third-party information providers not involved in
funding bond issues as financial intermediaries are not financially accountable
for the quality of information produced, and therefore are more likely to produce
accurate information than are, for example, bond analysts employed by
underwriters. On the other hand, even independent third-party information
providers may face a “multi-period incentive contract,” where future revenues
depend on the quality of current information.
61
According to this theory, the
existence of the information provider depends on the accuracy of the
information it provides, and providers attempt to persuade others to accept the
information it signals about issuers.
Law journal commentary on credit ratings seems to accept many of the
above arguments, which collectively present the credit rating business as
competitive and reputation-driven.
62
According to this view, S&P and Moody’s
quality, and that therefore buyers may not receive accurate information about credit from sellers. See id.
59. See Stover, supra note 53, at 63.
60. I discuss reasons why, according to the regulatory license view, these three criteria may not be
satisfied by modern credit rating agencies infra at 703.
61. See Hsueh & Kidwell, supra note 51, at 47 (discussing implications of this “multi-period
incentive contract”).
62. Consider the following examples:
In many markets, intermediaries play a certification role without any regulatory intervention. Standard
and Poor’s (S&P) and Moody’s, for example, certify the credit risk of company debt.

Page 16
634
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
have remained the dominant credit rating agencies for nearly a century because
they have performed well in assessing credit risk, and thus have developed and
sustained their reputations for quality. In short, the top rating agencies have
acquired and maintained reputational capital. This view depends on the
assumption that rating agencies have a comparative advantage in gathering and
analyzing information, primarily because of economies of scale.
According to this view, if S&P and Moody’s had not continued to generate
quality information and their ratings had therefore become less accurate or
reliable, they would have suffered a loss in reputation. Over time the agencies
would have lost revenue, and perhaps been forced out of the rating industry.
Stephen Choi, Market Lessons for Gatekeepers, 92 N
W
. U. L. R
EV
. 916, 934 (1998) (emphasis added).
The very value of an agency’s ratings, like an accountant’s opinions, lies in their independent, reliable
evaluation of a company’s financial data.
Gregory Husisian, What Standard of Care Should Govern the World’s Shortest Editorials?: Analysis of
Bond Rating Agency Liability, 75 C
ORNELL
L. R
EV
. 411, 426 (1990).
Indeed, the only reason that rating agencies are able to charge fees at all is because the public has enough
confidence in the integrity of these ratings to find them of value in evaluating the riskiness of investments.
Jonathan R. Macey, Wall Street Versus Main Street: How Ignorance, Hyperbole, and Fear Lead to
Regulation, 65 U. C
HI
. L. R
EV
. 1487 (1998) (reviewing my book, F.I.A.S.C.O.: Blood in the Water on
Wall Street, and concluding “[t]he author does not understand the important role reputation plays on Wall
Street.”)
Finally, credit rating agencies enhance the capital markets infrastructure by distilling a great deal of
information into a single credit rating for a security. That rating reflects the informed judgment of the
agency regarding the issuer's ability to meet the terms of the obligation. Such information is frequently
critical to potential investors and could not be acquired otherwise, except at substantial cost.
Susan M. Phillips & Alan N. Rechtschaffen, International Banking Activities: The Role of the Federal
Reserve Bank in Domestic Capital Markets, 21 F
ORDHAM
I
NT
'
L
L.J. 1754, 1762-63 (1998).
Rating agencies help solve this problem by processing the flow of information and distilling it into a
rating useful to the investor at a much lower cost than could potentially be incurred by individual
investors. In a world without rating agencies, investors’ costs at the margin in conducting research would
outweigh the benefits and issuers might have to pay higher interest rates to signal their ability in the
market and thus encourage investors to invest in their securities. The production of securities information
with rating agencies increases the efficiency of the capital markets because rating agencies have expertise,
economics of scale, and can communicate distilled information quickly and effectively in the capital
markets.
Amy K. Rhodes, The Role of the SEC in the Regulation of the Ratings Agencies: Well-Placed Reliance
or Free-Market Interference?, 20 S
ETON
H
ALL
L
EGIS
. J. 293, 295-96 (1996) (citations omitted).
Information intermediaries, such as securities analysts or credit rating agencies, facilitate such
conventions by decoding ambiguous signals.
George G. Triantis & Ronald J. Daniels, The Role of Debt in Interactive Corporate Governance, 83
C
AL
. L. R
EV
. 1073, 1110 (1995).
For the more general argument that third-party certification agencies prosper based on their ability to
sustain a good reputation for signaling value to purchasers, see Ronald J. Gilson, Value Creation by
Business Lawyers: Legal Skills and Asset Pricing, 94 Y
ALE
L.J. 239, 288-93 (1984); see also Reinier
H. Kraakman & Ronald J. Gilson, The Mechanisms of Market Efficiency, 70 V
A
. L. R
EV
. 549, 613-21
(1984)

Page 17
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
635
Assuming few barriers to entry in the credit rating business,
63
new entrants
would have displaced any agency suffering a loss of reputational capital. Such
market forces would have acted continuously on credit rating agencies,
especially given the technological innovation in financial markets in recent
decades,
64
so that the playing field would have shifted constantly if agencies had
not out-innovated and out-improved the competition. Therefore, the argument
goes, Moody’s and S&P must have survived because of the quality of their
ratings. Again, according to this view, the top rating agencies have acquired and
maintained reputational capital.
In sum, according the “reputational capital” view, credit ratings do not differ
from the numerous other ratings in financial markets, including Morningstar
ratings of mutual funds; Institutional Investor ratings of securities analysts,
dealers, and banks; individual bank ratings of securities;
65
and numerous other
ratings of securities and funds.
66
Credit rating agencies exist in a competitive
market of information providers and live or die based on their reputational
capital. Credit ratings reduce information costs and therefore reduce issuers’
cost of capital. In other words, the purpose of rating agencies is viewed as
providing information about the amount of credit that can be extended to a party
without undue risk. Supporting the investors are mountains of data, supplied
primarily by credit rating agencies, that show a statistical correlation between
ratings and default.
67
Public reports often support this historical view, and show
63. Although this assumption may seem reasonable as to the accumulation of research and
information, substantial evidence supports the proposition that the value of a credit rating derives from
information other than basic, publicly available information such as firm size, asset valuation, average
interest coverage, and other debt-related statistical measures. See, e.g., Louis H. Ederington et al., The
Informational Content of Bond Ratings, 10 J. F
IN
. R
ES
. 211, 218-25 (1987); see also discussion infra at
639. To the extent the agencies gather information from non-obvious or non-public sources, there may be
costly barriers to entry. In addition, a government-created oligopoly obviously imposes barriers to entry.
64. Merton Miller, who won the Nobel Price in Economics in 1990 for his work in corporate
finance, has both predicted and explicated the major technological advances in finance. See, e.g., Merton
H. Miller, Financial Innovation: The Last Twenty Years and the Next, 21 J. F
IN
. & Q
UANTITATIVE
A
NALYSIS
459 (1986) (prediction); M
ERTON
H. M
ILLER
, M
ERTON
M
ILLER ON
D
ERIVATIVES
1-14
(1997) (explication).
65. For example, stocks are grouped into various categories, including growth and value. Services
such as Valueline, and even major financial periodicals and newsletters, rate stocks based on performance.
Stock analysts at financial intermediaries issue recommendations of buy, sell, hold, or, euphemistically,
“accumulate.” These recommendations can have enormous effect on the behavior of buyers even though
there is evidence of systematic upward bias in ratings. Managers of stock issuers maintain close
relationships with these analysts. See Husisian, supra note 62, at 426 n.25.
66. Some banks have developed more sophisticated benchmarks for comparing investments in stocks
and mutual funds. See discussion infra at 707-08.
67. See, e.g., S
TANDARD
& P
OOR
S
R
ATINGS
P
ERFORMANCE
1997: S
TABILITY AND
T
RANSITION
(1998) [hereinafter R
ATINGS
P
ERFORMANCE
].

Page 18
636
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
that investors agree that the agencies’ principal function is not to influence
investment decisions, but to provide an independent view of the quality of
securities.
68
Ratings are mnemonics for credit quality; simply put, AAA-rated
companies don’t default.
B. Development of Credit Ratings
The next two sections analyze the development of credit ratings through the
lens of the reputational capital view. The precursors to twentieth-century credit
rating agencies were mercantile credit agencies, and their history seems to fit the
above view of ratings and reputational capital. During the seventeenth and
eighteenth centuries, colonial importers customarily extended up to a year of
credit to their retail customers, shopkeepers, and general stores.
69
Payments
were often late, and it was difficult for sellers to gather credible information
about the reputation of buyers: letters of reference were faked or forged, detailed
financial data were not available, and the process was tediously slow.
70
As
markets and trade evolved during the nineteenth century, it became clear that
there were economies of scale associated with gathering and disseminating
credit information in a systematic, organized way.
One of the victims of the crisis of 1837
71
was Lewis Tappan who operated a
substantial silk business with his brother.
72
Fortunately for the Tappans, they
kept detailed credit information about current and prospective customers, which
68. Although many investors perform additional credit analyses, some rely on rating agencies to
avoid the expense associated with engaging in independent credit review. See House, Rating the Raters,
I
NSTITUTIONAL
I
NV
., Oct. 1995, Int’l Edition, at 53.
69. See J. Wilson Newman, Dun & Bradstreet: For the Promotion and Protection of Trade, in
R
EPUTATION
: S
TUDIES IN THE
V
OLUNTARY
E
LICITATION OF
G
OOD
C
ONDUCT
85, 86 (Daniel Klein ed.,
1997).
70. See id. at 86-89.
71. These are reasons why credit ratings might become increasingly popular and valuable following
a major financial crisis, such as the crisis of 1837, despite the agencies’ failure to anticipate the crisis. In a
time of crisis, the value of information about ability or willingness to pay debts increases, as the
probability increases that any individual creditor will not be able or willing to pay its debts. The crisis of
1837 fit these expectations:
With the passing of the clouds, merchants throughout the country began to realize that one of the chief
contributory causes of the crash and the depression which followed was inherent in the conditions which
governed credit granting in this country. . . . Thus the crisis of 1837 brought the merchants face to face
with the necessity of clearer and more thorough scrutiny of credit risks. The eagerness with which
information was sought by sellers of goods on time finally resulted in the establishment of the mercantile
agency.
T
HEODORE
N. B
ECKMAN
, C
REDITS AND
C
OLLECTIONS IN
T
HEORY AND
P
RACTICE
135 (1930).
72. See G
ILBERT
H
AROLD
, B
OND
R
ATINGS AS AN
I
NVESTMENT
G
UIDE
: A
N
A
PPRAISAL OF
T
HEIR
E
FFECTIVENESS
6 (1938); Rhodes, supra note 62, at 300.

Page 19
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
637
included many large commercial enterprises.
73
When the silk business collapsed,
that customer information proved valuable to other merchants, and in 1841,
Tappan formed The Mercantile Agency, the first mercantile credit agency.
74
As Tappan and other mercantile credit raters thrived during the late 1800s,
other raters began to copy Tappan’s idea, particularly in rating investments in
stocks and bonds. By 1890, Poor’s Publishing Company (S&P’s predecessor)
was publishing Poor’s Manual, an analysis of various types of investments,
including bonds.
75
By the early 1900s, various American firms were classifying
bonds, primarily railroad bonds, into groups according to quality, and there was
at least one bond rating system in Europe.
76
John Moody, a Wall Street analyst during the early 1900s, observed the
success of mercantile credit ratings and became interested in applying this
simple rating methodology to bonds.
77
In his words, “[s]omebody, sooner or
later, will bring out an industrial statistical manual, and when it comes, it will be
a gold mine.”
78
By 1907, several analysts had issued reports on the railroad
73. See H
AROLD
, supra note 72, at 7. Tappan also began extensive correspondence with business
people and lawyers, who provided opinions and data regarding the standing and reputation of owners of
businesses throughout the U.S. See Newman, supra note 69, at 89. Some sellers, especially those with
poor credit, opposed the formation of rating agencies. See id.
74. See, e.g., H
AROLD
,
supra note 72, at 7-8; Richard Cantor & Frank Packer, The Credit Rating
Industry, F
EDERAL
R
ESERVE
B
OARD
N.Y. Q. R
EV
., Summer-Fall 1994, at 1. In 1859, Robert Graham
Dun became the sole proprietor of the agency, renamed R.G. Dun and Company, that issued the Dun
rating book containing about 20,000 names. See H
AROLD
, supra note 72, at 7-8. At the same time, John
M. Bradstreet, a Cincinnati lawyer, founded a competing firm in 1849, and in 1857 began publishing the
world’s first commercial rating book. Newman, supra note 69, at 90; see also H
AROLD
,
supra note 72, at
8. In 1933, the two companies merged to form Dun and Bradstreet, Inc. See The Formation of Dun and
Bradstreet, Inc., D
UN
& B
RADSTREET
W
KLY
. R
EV
., March 4, 1933, at 3. Various other agencies also
were formed throughout the 1800s. See Cantor & Packer, supra, at 1-2.
75. See, e.g., In re Bartol, 38 A. 527 (Pa. 1897) (approving of reference to Poor’s Manual of
1890).
76. See H
AROLD
,
supra note 72, at 11. According to John Moody, founder of the credit rating
system:
While no one in this country had attempted such a thing as investment ratings by means of symbols, yet
even in those days bonds were classified into groups according to quality and salability, especially by
large investment institutions, such as insurance companies. Moreover there had existed for a considerable
time, I think, a bond rating system in Vienna and also, I believe, in Berlin. These foreign systems had
been developed by symbols and the Austrian Manual of Statistics, which carried these symbols, was
quite well known in Europe, although not at all in this country.
Id. (quoting from John Moody, in a letter to Harold dated August 21, 1934). For a detailed description of
the role of rating agencies outside the U.S., and especially in Europe, see Carsten Thomas Ebenroth &
Thomas J. Dillon, Jr., The International Rating Game: An Analysis of the Liability of Rating Agencies
in Europe, England, and the United States, 24 J. L
AW
& P
OL
Y
I
NT
L
B
US
. 783 (1993).
77. In 1934, Moody wrote, “I cannot claim much credit for creating the idea, and certainly I think
the general use of commercial and credit ratings had something to do with bringing the idea of possible
bond ratings to my mind.” H
AROLD
,
supra note 72, at 11.
78. See L. Macdonald Wakeman, The Real Function of Bond Rating Agencies, in T
HE
M
ODERN

Page 20
638
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
industry, with elaborate statistics and detailed operating and financial data.
79
Moody believed that if he could synthesize the complex data in these reports
into a single rating symbol for each bond, he could make his fortune selling such
ratings to the public. Other bankers opposed Moody’s plan, saying Moody
could earn superior returns by using such “inside information”
to trade, instead
of publicly flaunting it.
80
Moody ignored these protestations,
81
persevered,
82
and
in 1909 published the first rating scheme for bonds, in a book entitled Analysis
of Railroad Investments.
83
Moody’s Investors Service was incorporated in 1914
and created a formal rating department in 1922.
84
By 1924 Moody’s ratings
T
HEORY OF
C
ORPORATE
F
INANCE
391, 392 (Michael C. Jensen & Clifford W. Smith, Jr. eds., 1984)
(quoting John Moody).
79. Moody relied heavily on reports published in 1906 and 1907 concerning the railroad industry,
including “The Earning Power of Railroads” by Floyd Mundy and “American Railways as Investment”
by Carl Snyder, each of which contained elaborate statistics on the railroad industry, including detailed
operating and financial data. See H
AROLD
,
supra note 72, at 11-12.
80. Moody and other bankers seem to have believed that Wall Street analysts could use non-public
information concerning the railroad industry to earn superior profits from bond trading. See id. at 11-12.
When Moody was considering his rating idea, one “old Wall Street buccaneer” advised him:
You young pipe dreamer, why throw away your ten years’ experience of learning the rules of the game?
Why give the public all the facts regarding the corporations for the price of a book? You will be showing
them how to play safe and get rich, while you will make nothing yourself. Anyway, if you begin to flaunt
too many facts, there won’t be much inside knowledge left to work on; you will be spoiling our game.
Use your information yourself; don’t be a philanthropist. There’s no money in it!”
J
OHN
M
OODY
, T
HE
L
ONG
R
OAD
H
OME
91 (1933). Moreover, Moody’s rationale for marketing bond
ratings was that he did not have sufficient capital to benefit from using inside information directly to earn
profits from trading. According to Moody, although a person with capital could take advantage of inside
information in the short-run, a person without capital (such as Moody)
could earn greater long-run returns
by selling the information “wholesale in a book.” Id.
81. Other bankers, who presumably had access to capital and were happy to earn short-run profits
from inside information, were opposed to Moody’s ratings idea. They protested ratings, and later
threatened Moody and his agency; one bank reportedly offered options on stock to Moody’s in exchange
for a more favorable rating, and it was rumored that one agency had a bank employee on its staff
(although agency managers claimed they did not know this). See H
AROLD
,
supra note 72, at 16-17.
Moody seems to have believed that bankers eventually supported his ratings, although it is not clear when
this support was manifested. See <http://www.moodys.com/moodys/mdyindex.htm> (visited Nov. 15,
1999) (quoting John Moody as saying, “[i]n the Spring of 1909 I brought out the first edition of a new
type of Railroad Manual which attempted to analyze railroad reports and rate their bond issues. While it
raised a storm of opposition, not to mention ridicule from some quarters, it took hold with dealers and
investment houses). In contrast, when Moody presented the idea of bond ratings to Thomas F. Woodlock,
then editor and co-publisher of the Wall Street Journal, Woodlock enthusiastically supported Moody. See
H
AROLD
,
supra note 72, at 10.
82. Moody’s first efforts do not seem to have been successful; he became insolvent in 1907. See id.
at 11-12.
83. See, e.g., House, supra note 41, at 53; McGuire, supra note 16, at 2; Rating the Rating
Agencies, supra note 10, at 53; Cantor & Packer, supra note 74, at 1-2.
84. “Moody’s first ratings were assigned to the bonds of more than 250 railroads.” Kenneth Pinkes,
The Function of Ratings in Capital Markets: Rating Scale, Rating Approach and Credit Topics: A
Moody’s Symposium, Dec. 5, 1997, at 2 (Special Report from Moody’s, transcript of speech in Vienna,
Austria on Oct. 23, 1997; copy on file with author). Later, in 1918, Moody’s began to rate bonds issued

Page 21
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
639
covered nearly 100 percent of the U.S. bond market.
85
Within a few years after Moody published his first ratings, three other credit
rating organizations entered the market. Poor’s Publishing Company was the
second rating agency; it began rating stocks and bonds in 1916.
86
Standard
Statistics Company, Inc. followed in 1922.
87
Fitch Publishing Company began
publishing ratings in 1924.
88
Such early, consecutive, entries into the credit rating market are evidence
that there were no substantial barriers to entering the credit rating industry
during the 1920s. The cost of accumulating the relevant statistics and data to
generate a rating were not prohibitively high. Indeed, any “barriers to exit”
should not have been high either, and an agency that did not generate accurate
and credible ratings probably would not have expected to survive long.
89
It appears that, at least during the 1920s, rating agencies functioned in a
manner consistent with the reputational capital view described above. For
example, investors and analysts often would call or write the rating agencies to
point out inaccuracies or to disagree with the agency’s rating.
90
In this way,
rating agencies served as an information intermediary. Their ratings became a
pricing mechanism for information. Valuable information was incorporated in
the rating; worthless or incredible information was not. In general, bond
by U.S. cities and other municipalities. See id.
85. See A Century of Market Leadership (visited Oct. 31, 1999) <http://www.moodys.com/
moodys/mdyindex.htm>; Pinkes, supra note 84, at 2. Ironically, Dun and Bradstreet—with its roots in the
first mercantile credit rating agency—purchased Moody’s Investors Service in 1962. See Cantor &
Packer, The Credit Rating Industry, supra note 74, at 2.
86. See H
AROLD
,
supra note 72, at 13.
87. Poor’s merged with Standard Statistics in 1941 to form S&P. See Cantor & Packer, supra note
74, at 2. S&P was an independent, publicly-owned corporation until 1966, when McGraw-Hill Inc., a
major publishing company, acquired all of S&P’s common stock. See id. Today, S&P can, and does,
claim it has assigned ratings to corporate bonds since 1923, and to municipal bonds since 1940. See D
EBT
R
ATINGS
C
RITERIA
, supra note 38, at 5.
88. See H
AROLD
,
supra note 72, at 13. Fitch merged with IBCA, Ltd. in 1997 and is now operating
as Fitch IBCA, Inc. See <http://www.fitchibca.com/about_us/philosophy/> (visited Nov. 15, 1999).
89. Nevertheless, the only significant new entry into the credit rating market since the 1920s has
been Duff and Phelps Credit Rating Co., which began providing bond ratings for a wide range of
companies in 1982. Although Duff and Phelps had researched public utility companies since 1932, it did
not provide formal ratings until fifty years later. The only other entrant into the U.S. market has been
McCarthy, Crisanti, and Maffei, which was founded in 1975, and was acquired by Duff and Phelps in
1991. See id. The Fitch-IBCA merger followed IBCA’s failed attempt to obtain NRSRO status. Today,
Moody’s, S&P, Fitch IBCA, and Duff are the four major U.S. credit rating agencies. See Rhodes, supra
note 62, at 301. There are several Canadian and Japanese raters of credit, and Thomson BankWatch in the
United States rates the obligations of financial institutions only; similarly, A.M. Best rates only the ability
of U.S. insurance companies to pay claims. See Cantor & Packer, supra note 74, at 2. I assess the reasons
why barriers to entering these markets might have increased more recently in Part III.
90. See H
AROLD
,
supra note 72, at 14.

Page 22
640
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
investors supported and welcomed ratings during this period.
In contrast, bond issuers opposed the ratings, much as they had opposed the
earlier commercial credit ratings. Some issuers regarded the role of the credit
rating agency as an intrusion into the corporation’s business. Nevertheless, such
“intrusions,” and even the opposition from issuers, also increased the amount
and value of information flowing to the rating agencies.
91
Issuers had no choice
but to provide the agencies with valuable information, including non-public
information.
92
In turn, the rating agencies published accurate and reliable ratings
and thereby increased their stock of reputational capital.
Rating agencies continued to accumulate reputational capital during the
1920s because they were able to gather and synthesize valuable information.
During this time, ratings were financed entirely by subscription fees paid by
investors,
93
and the rating agencies competed to acquire their respective
reputations for independence, integrity, and reliability.
94
In a market with low
barriers to entry, a rating agency issued inaccurate ratings at its peril. Every
time an agency assigned a rating, that agency’s name, integrity, and credibility
were subject to inspection and critique by the entire investment community.
Reputational considerations would have been especially acute in such an
environment.
There is evidence that rating agencies continued to accumulate reputational
capital during the 1930s. Rating agencies and ratings became much more
important to both investors and issuers during this period. In the years following
the stock market crash of 1929, demand for credit ratings increased, as investors
became concerned about high bond default rates and credit risk.
95
By the end of
the 1920s, the U.S. bond market included approximately 6,000 bond issues with
91. For example, a corporation receiving a low rating might protest and challenge the rating, saying
something like, “[s]end your man around, and we’ll show him a few things that will cause him to raise
your rating.” Id. at 16. Such additional information might be of value to the rating agency as well as to
investors and therefore might assist the agency in accumulating reputational capital.
92. Just as the rating agencies’ direct access to issuers generated opportunities for a rater to acquire
non-public information, which would have been of great value to the rater, the rating agency employees’
direct access to issuer employees often put agency employees in a position to receive favors, tangible or
intangible, from issuers in return for assigning a higher rating than they otherwise would have assigned.
There is some evidence of precisely such scenarios: “Some corporations indeed have gone so far as to
assure the rating agency that if a given rating were raised the corporation or its sponsors ‘could do
something’ for the rating agency.” Id. at 58-59. These criticisms survive today. See Credit-Rating
Agencies: Beyond the Second Opinion, supra note 10, at 80 (criticizing agency analysts for becoming
too friendly with the companies they rate).
93. See McGuire, supra note 15, at 4.
94. See id. at 5.
95. See id. at 2.

Page 23
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
641
a total face amount of more than $26 billion; the vast majority of those bond
issues were rated by the rating agencies.
96
Court decisions during this period are consistent with the view that rating
agencies had been able to accumulate and retain reputational capital by the
1920s.
97
A few cases explicitly relied on ratings as evidence of the propriety of
bond purchases in assessing whether fiduciaries had satisfied their duties. For
example, the court in In re Bartol, addressing a challenge of a trustee’s
purchase of an electric railway bond, referred to Poor’s Manual of 1890 in
holding that “[i]t must be conceded under the evidence, that the trustees used all
the care that a person of ordinary care and prudence would use in determining
upon an investment of his personal funds.”
98
Likewise, in In re Detre’s Estate,
the court relied on a Moody’s rating in finding that a trust properly purchased
certain bonds: “In Moody’s Manual for 1914, these . . . bonds are rated:
Security, very high; Salability, good; net rating, A.”
99
And in In re Winburn’s
Will, the court relied on the ratings given by Moody’s, holding that “[t]here is a
distinction between seasoned securities of this character here involved and
investments in speculative securities.”
100
The credit ratings systems and scales were well established by 1929. Ratings
were divided into categories, based on the credit quality of the rated financial
instrument. Generally, a bond rating was intended to indicate the likelihood of
default or delayed payment for that bond. As with many ratings, the practice
was to assign the letter A or the number 1 to the highest grade, with A1
signifying a high, if not the highest, grade. Relative rankings, in descending
96. See id.
97. See Wakeman, supra note 78, at 393-94. In contrast, recent suits against rating agencies, most
of which have been settled, reflect the failure of the credit rating agencies to anticipate defaults or other
credit problems. These suits include class action litigation related to the Washington Public Power Supply
System default in 1983, and the Executive Life bankruptcy in 1991. See, e.g., Francis A. Bottini, Jr. An
Examination of the Current Status of Rating Agencies and Proposals for Limited Oversight of Such
Agencies, 30 S
AN
D
IEGO
L. R
EV
. 579, 584-95 (1993); Cantor & Packer, supra note 74, at 4. On June
11, 1996, Orange County, California, filed a complaint in the U.S. Bankruptcy Court for the Central
District of California, alleging breach of contract, professional negligence, and aiding and abetting breach
of fiduciary duty by S&P. County of Orange v. McGraw-Hill Cos., No. SA 94-22272 JR (June 11,
1996). On March 17, 1997, the district court upheld in part the bankruptcy court’s denial of the motion to
dismiss. See District Court Case No. SACV 96-0765-GLT (filed Mar. 18, 1997). McGraw-Hill settled
the case in June 1999 by agreeing to pay Orange County $140,000, a fraction of the $860.7 million paid
by other defendants in related litigation brought by the county. See Orange County Litigation Ends with
Settlement of McGraw Hill Suit, P
ROF
. L
IABILITY
L
ITIG
. R
EP
., Aug. 1999, at 11.
98. 38 A. 527 (Pa. 1897).
99. 117 A. 54 (Pa. 1922).
100. 249 N.Y. Supp. 758, 762 (1931).

Page 24
642
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
order, would be B or 2, then C or 3, and so on.
101
The agencies’ scales were similar, with each agency employing both ordinal
(e.g., A,B,C) and cardinal (e.g., Aaa, Aa, A) ratings. Each agency used three
subcategories for each broad rating category (e.g., three levels of “As,” three
levels of “Bs”). By 1930, it was possible to match each agency’s rating symbols
one-for-one with each of the other agency’s symbols.
102
Moreover, although the agencies did not agree on every rating, ratings were
loosely correlated and there was a certain amount of rating “inflation” evident in
each of the agency’s scales. For example, the vast majority of ratings were in
the A category. Very few bonds were rated C or lower.
103
Following the stock market crash of 1929, numerous ratings were lowered,
101. The following table summarizes the various ratings as of 1929:
Symbols of the Principal Rating Agencies
Fitch
Moody
Poor
Standard
Majority
Interpretation
AAA
Aaa
A**
A1+
Highest
AA
Aa
A*
A1
High
A
A
A
A
Sound
BBB
Baa
B**
B1+
Good
BB
Ba
B*
B1
Fair
B
B
B
B
Somewhat speculative
CCC
Caa
C**
C1+
Speculative
CC
Ca
C*
C1
Highly speculative
C
C
C
C
Extremely speculative
DDD
D**
D1+
Low or weak
DD
D*
D1
Small or very weak
D
D
D
Practically valueless
See H
AROLD
, supra note 72, at 75.
102. The single exception to this one-for-one matching was Moody’s, which did not use the D
category of ratings.
103. A representative sample chosen for one study was as follows:
Distribution of Issues by Ratings, July 15, 1929
Rating
Fitch
Moody
Poor
Standard
A+
147
97
68
78
A
64
63
89
93
A-
80
99
110
104
B+
40
59
61
40
B
17
25
22
26
B-
4
2
7
16
C+
3
4
C
C-
D+
1
Unrated
8
18
6
1
See Harold,
supra note 72, at 90.

Page 25
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
643
as the rating agencies, along with most individuals and institutions, failed to
anticipate the rapid decline in the prices of hundreds of bond issues.

Page 26
644
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
For example, the Chicago, Rock Island & Pacific 4s-1988, was rated Aaa, or
the equivalent highest rating, by all four agencies in 1929, was only rated Aa, or
the equivalent rating, for another five years thereafter, and by 1934 was in
default.
104
Notwithstanding the large number of ratings changes (mostly downgrades)
in the early 1930s and the considerable lag between the time market prices
incorporated negative information about bonds and the time credit ratings
incorporated such information, ratings continued to be a respected and
important institution in the bond market.
105
The preface to Moody’s 1931
Manual of Industrials stated that “[t]he fundamental thought back of the system
of rating investment securities . . . has been to furnish an authoritative key to
the relative security and stability, from an investment standpoint, of all types of
bonds and stocks.”
106
Moody’s and other agencies apparently were able to
retain this “authority” during the 1930s, despite the obvious decline in the
accuracy of their ratings.
Although credit rating agencies continued to employ the so-called “best and
brightest” during the 1930s, there is reason to doubt the agencies’ ability to
generate valuable information during this period.
107
Although rating agencies
claimed their information was from “unique sources,” much of it obviously was
from publicly available investment news.
108
Moreover, most bond issues during
the 1930s were not rated until after they were distributed, a sign that credit
ratings were viewed as valuable only in the secondary market, not in the
primary market for new issues, where the agencies’ information arguably should
have been of much greater value.
109
104. See id. at 46.
105. Harold provides interesting anecdotal evidence of the popularity of credit ratings during this
period, as of 1934: “Hardly a bond is purchased anywhere in the United States but that the purchaser asks,
‘How is it rated?’” id. at v; “Nearly every commercial bank, investment bank, insurance company,
investment trust, and investing trustee from the Atlantic to the Pacific, from Canada to Mexico, consults
them.” id. at 3; “It is difficult, in fact almost impossible, to discover a bank that does not use bond
ratings.” id. at 20; “One banker, nicknamed ‘Triple-A James’ because of his insistence on buying only
AAA bonds, is typical of thousands in the banking profession.” id. at 20; “practically all” brokerage
offices displayed one or more of the ratings manuals in their reading rooms and lobbies, id. at 21.
106. Id. at 48 (quoting from M
OODY
S
, M
ANUAL OF
I
NDUSTRIALS
vii (1931) (emphasis added).
107. The agencies remained relatively small during this period, especially compared to banks and
other financial institutions, and to the companies the agencies rated. Claims that agency employees
possessed specialized expertise not available to other bond analysts are dubious. One example: “There is
some evidence that those individuals who are engaged in the rating process are above ‘normal’ in
intelligence. Many are college men.” Id. at 57.
108. See id. at 58.
109. See H
AROLD
, supra note 72 at 21.

Page 27
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
645
Institutions relied on credit ratings to varying degrees during this period.
Large New York banks used the ratings merely as a check on their own
findings, but smaller “country” banks continued to depend on ratings almost
exclusively as an authoritative guide in placing values on the credit risk
component of securities they owned or were considering purchasing.
110
Large
insurance companies placed less weight on ratings and relied more on their own
analysts, who they believed based their opinions on more recent, higher-quality
information. At large industrial companies, ratings were at least consulted,
primarily because they were regarded as having some “recognized publicity
value.”
111
Certain institutions regarded credit ratings as so important and accurate that
they relied on ratings in formulating internal guidelines. Trust companies, in
particular, placed great reliance on ratings, and some trustees were restricted by
the terms of their fund to investments in securities of a stipulated rating or
higher.
112
Some trusts limited the range of permissible investments, based on
assigned credit ratings.
113
One important development paralleled the increase in importance of rating
agencies during the 1930s. Both academic studies and anecdotal
110. See id. at 20 (citing Letter from J.S. Love, Superintendent of Banking, State of Mississippi
(Aug. 21, 1934)).
111. Id. at 22.
112. For a detailed description of how regulation has incorporated credit ratings, see Grafton, supra
note 10, at 22. One example from 1934: Dillman A. Rash, of the bond department of The Louisville Trust
Company, stated that as to a particular security, “[t]he AAA rating accorded (a certain security) . . . was
the only way we were able to ‘sell’ it to our Trust Investment Committee.” H
AROLD
,
supra note 72, at 23
(quoting Letter from Dillman A. Rash, Bond Department, The Louisville Trust Company (Aug. 31,
1934)).
113. For example, a 1920s agreement and declaration of trust between the United States Shares
Corporation and the Chase National Bank of the City of New York specified, among other things, that:
The following conditions shall govern all reinvestments: . . .
(2) No stock shall be acquired for substitution if it is then rated lower than B in “Moody’s Manual of
Investments and Security Rating Service.”
(3) Not more than ten (10%) per cent of the total investment in a unit shall by reason of such substitution
be at the date thereof in stock rated lower than Ba in said Moody’s Manual.
(4) Not more than fifty (50%) per cent of the total investment in a unit shall by reason of such
substitution be at the date thereof in stock rated lower than Baa in said Moody’s Manual.
(5) Not less than twenty (20%) per cent of the total investment in a unit shall by reason of such
substitution be at the date thereof in stock rated A or higher in said Moody’s Manual.
If publication of said Moody’s Manual shall be discontinued, its successor, or if none, a similar
reference and rating service of at least equal standing shall be used in ascertaining the ratings hereinbefore
mentioned.
H
AROLD
,
supra note 72, at 24 (citing L
ELAND
R
EX
R
OBINSON
, I
NVESTMENT
T
RUST
O
RGANIZATION AND
M
ANAGEMENT
557-58 (New York, 1929)).

Page 28
646
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
evidence indicate that changes in ratings appeared to have a non-trivial effect on
the changes in the price of the rated issue.
114
For example, Gilbert Harold
studied prices of a group of 363 bonds from mid-July 1929 to mid-July 1931 to
determine whether, during this period of extreme financial dislocation, the
actions of the rating agencies had an effect on the market.
115
Harold observed
market action
116
both preceding and following changes in ratings.
117
In a
separate study, Harold found that “there is a very definite tendency for the
market value of specifically recommended bonds to rise within ten days after
publication of the ‘buy’ advice, and, conversely, . . . there is a definite tendency
for the market value of ‘sell’ bonds to decline within the same immediate
period.”
118
Anecdotal evidence from the period supports Harold’s findings. According
to The Wall Street Journal, “[w]hen an issue appreciates substantially, and
nears par, as some railroad obligations have in recent months in reflection of
improved earning power, a higher rating is given at a time when such earnings
may have been well discounted.”
119
Another expert, writing in 1936, agreed that
“[s]ince conditions are constantly changing, individual issues within a group will
be improving, or the reverse. In time, this change will be reflected in ratings . . .
but after the change has taken place.”
120
Reliance on the rating system during the 1930s was sufficiently widespread
that higher-rated bonds enjoyed a more liquid market. Dealers purchased blocks
of bonds when they anticipated an agency would be raising an issuer’s rating,
because the bonds would then trade at higher prices.
121
114. According to one study in the early 1930s, “market prices are frequently affected by changes in
rating.” See Gustav Osterhus, Flaw-Tester for Bond Lists, A
M
. B
ANKERS
A
SSN
. J., Aug. 1931, at 67.
115. Harold used Fitch ratings, which were the easiest data to obtain at the time. In addition to
Harold’s 363-bond statistical study, he conducted personal interviews of all the major rating agency
executives, dozens of state banking commissioners, officials from the major national and Federal Reserve
regional banks, insurance company executives, and many other experts, including James M. Landis, then
Chairman of the Securities and Exchange Commission. See H
AROLD
,
supra note 72, at 261-65.
116. By market action, I mean an increase in value for a raised rating, or a decrease in value for a
lowered rating.
117. See id. at 185.
118. Gilbert Harold, Accuracy in Reading the Investment Spectrum, A
M
. B
ANKERS
A
SSN
. J., July
1934, at 32.
119. Bankers Oppose Eligibility Rule for Investments, W
ALL
S
T
. J., Mar. 13, 1936, at 1.
120. H
AROLD
, supra note 72, at 186 (quoting George D. Bushnell, Investments Join Loans, R
AND
M
C
N
ALLY
B
ANKERS
M
ONTHLY
, June 1936, at 368) (emphasis in original).
121. See id. at 191 (quoting Letter from James H. Oliphant & Co. to Gilbert Harold (Feb. 27,
1936)).

Page 29
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
647
uring the 1930s, a few rating agency executives confessed to a belief that bond
rating changes caused bond price changes, not the reverse, although there is not
substantial evidence that investors could profit from a strategy of buying bonds
immediately after a rating upgrade, or selling after a downgrade.
122
Rating
agency executives do not seem to have attempted to take advantage of this
phenomenon, either.
123
The above evidence about the relationship between bond ratings and the
bond market in the 1930s may be consistent with the reputational capital view.
The fact that rating changes influenced market prices is consistent with the view
that rating agencies reduced both investors’ cost of information and issuers’ cost
of capital.
124
To the extent prices change after a rating agency announces a
rating change, such a price change may be due to additional information the
agency is imparting to the market; however, such evidence is necessary, but not
sufficient, to establish that rating agencies are generating valuable information
and therefore accumulating reputational capital. On the other hand, to the extent
prices change before a rating agency announces a rating change, the later rating
agency announcement cannot have caused the price change; such evidence
would show rating agencies are merely parroting publicly-available information,
and therefore are not accumulating reputational capital.
125
As quickly as credit rating agencies were able to accumulate reputational
capital during their meteoric rise of the early 1930s, they just as quickly
squandered such capital during the following years. As a result, credit rating
agencies did not remain important or influential for long. Following their heyday
in the 1920s and 1930s, the agencies experienced austerity and contraction
during the 1940s and 1950s.
126
During this period, bond prices were not
volatile, the economy was healthy, and few corporations defaulted. As a
consequence, both the demand for and the supply of relevant credit information
dwindled. The rating agencies were struggling when John Moody died in
1958.
127
According to the reputational capital view, the decline of the rating
agencies would have been a response to their inability to generate accurate and
valuable information after the early 1930s.
122. See id.
123. See id. at 191-92.
124. See Rhodes, supra note 62, at 294-95.
125. See discussion infra at 657-60.
126. See, e.g., R
ATINGS
P
ERFORMANCE
, supra note 67
(1998).
127. See House, supra note 68, at 245.

Page 30
648
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
During the Vietnam War, bond price volatility increased somewhat, as did
issuance of commercial paper, and borrowers faced a severe credit
contraction.
128
Demand for credit information increased during this period, but
the agencies remained relatively small and not obviously important as a source
of information to issuers or investors. At the time, the rating agencies employed
only a few analysts each and generated revenues primarily from the sale of
published research reports.
129
The market did not place great value on those
research reports, presumably, according to the reputational capital view,
because rating agencies had lost a large portion of their reputational capital.
Moreover, as the commercial paper market expanded rapidly during the 1960s,
investors were not very precise in assessing credit quality.
130
In the fallout of the
1970 Penn Central default on $82 million of commercial paper, investors began
demanding more sophisticated levels of research.
131
The rating agencies, still
relatively small and without substantial reputational capital, were not in a
position to satisfy this demand.
One study of 207 corporate bond rating changes from 1950 to 1972 found
that rating agencies’ changes generated information of little or no value; instead,
such changes merely reflected information already incorporated into stock
market prices of the companies whose ratings were changed approximately one-
and-one-half years previously.
132
In other words, the lag between the change in
stock market price due to new information and the corresponding change in
bond rating was more than a year.
133
Concern about the failure of the rating
agencies to generate accurate and reliable information, especially during a time
of crisis, led to public arguments for regulation of the credit rating industry.
134
128. See id.
129. See id.
130. See Cantor & Packer, supra note 74, at 4.
131. See House, supra note 68, at 245.
132. See George E. Pinches & J. Clay Singleton, The Adjustment of Stock Prices to Bond Rating
Changes, 33 J. F
IN
. 29, 39 (1978). There were numerous studies of the effects of credit rating changes on
market prices in the Journal of Finance during this period, in part because the performance of the rating
agencies had been so abysmal. See, e.g., George E. Piches & Kent A. Mingo, A Multivariate Analysis of
Industrial Bond Ratings, 28 J. F
IN
. 1 (1973); Frank K. Reilly & Michael D. Joehnk, The Association
Between Market-Dominated Risk Measures for Bonds and Bond Ratings, 31 J. F
IN
. 1387 (1976).
133. Not surprisingly, investors could not profit by acting upon the announcement of a change in
ratings, because the market already had discounted such information. See Pinches & Singleton, supra note
132, at 38.
134. See, e.g., Wakeman, supra note 78, at 392 (setting forth arguments made three years earlier in
L. Macdonald Wakeman, The Real Function of Bond Rating Agencies, 1 C
HASE
F
IN
. Q. 19 (1981)).

Page 31
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
649
One commentator, L. Macdonald Wakeman, found that although the rating
agencies had acquired excellent reputations since the early 1900s for
“accurately evaluating and reporting the risks of new bond issues,” by the
1970s, bond ratings did not actively determine, but instead simply mirrored, the
market’s assessment of a bond’s risk.
135
Wakeman’s study also found that by
the 1970s bond ratings generated little information not already reflected in the
market price of the bonds.
136
C. The Modern Credit Rating Agency
From the mid-1970s to today, credit rating agencies have exploded in size.
The modern credit rating agency
137
is more influential and more profitable than
at any time this century, despite the fact that the rating system hasn’t changed in
any substantial way since the 1930s.
138
The rating scales also are similar to
those used during the 1930s.
139
135. See id. at 393 (“Hence, a rating change does not affect, but merely reflects, but market’s altered
estimation of a bond’s value.”) (italics in original). Wakeman explained the paradox of the importance of
rating agencies based on the agencies’ ability to attest to the quality of an issue and monitor a bond’s risk
so that management did not engage in behavior to benefit shareholders at the bondholders’ expense. See
id. However, this agency cost rationale does not explain why bondholders could not write covenants to
protect themselves, or why investors or other groups could not also provide such a monitoring function, or
why, if the agencies’ true purpose was monitoring management to protect bondholders, this purpose was
not highlighted by the agencies or by investors or even by management as an important or relevant role.
136. For example, Wakeman found no special effect on prices, even when bonds were upgraded to or
downgraded below the “investment-grade” level (Moody’s Baa or S&P’s BBB). See id. Wakeman also
discovered that a much larger number of Moody’s rating changes occurred in May and June, shortly after
most corporate annual reports were published, a discovery that belied the agency’s claims that rating
changes were based on something other than publicly available information. See id.
137. By “modern credit rating agency” I mean to include the four major credit rating agencies: S&P,
Moody’s, Duff and Phelps, and Fitch IBCA. Of these, S&P and Moody’s are by far the largest and share
the vast majority of the market. S&P and Moody’s are now wholly-owned subsidiaries of much-larger
information and publication corporation parents: Moody’s is a subsidiary of Dun & Bradstreet Corp; S&P
is a subsidiary of McGraw-Hill Co. See House, supra note 41, at 245. Ironically, their status as
subsidiaries means not only that S&P and Moody’s are not required to disclose detailed information about
earnings, revenues, and costs, but also that they do not require credit ratings. See Credit-Rating Agencies:
Beyond the Second Opinion, supra note 10, at 80.
138. S&P’s policies, which are representative, state as follows:
In determining a rating, both quantitative and qualitative analyses are employed. The judgment is
qualitative in nature and the role of the quantitative analysis is to help make the best possible overall
qualitative judgment because, ultimately, a rating is an opinion. . . . An S&P rating is not a
recommendation to purchase, sell or hold a security inasmuch as it does not comment as to market price,
market supply or investor preference and suitability.
D
EBT
R
ATINGS
C
RITERIA
, supra note 38, at iii.
139. Today’s agency rating systems closely correspond both to each other and to the rating scales of
the 1930s. The agencies have added plus and minus symbols to their ratings, to distinguish more finely
among issues, but the general framework for long-term senior debt ratings has changed very little during

Page 32
650
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
The number of credit rating agency employees has multiplied more than ten-
fold during the past decade. In 1980, there were thirty professionals working in
the S&P Industrials group; by 1986, there were only forty.
140
By 1995, S&P
had 800 analysts and a total staff of 1,200; Moody’s has expanded at a similar
rate and in 1995 employed 560 analysts and a total staff of 1,700.
141
the past 60 years. The following table of ratings is quite similar to the table of ratings from 1930. See
supra notes 101-02 and accompanying text.
Table 1
Moody’s
S&P/Others
Meaning
Aaa
AAA
Highest quality
Aa1
AA+
High quality
Aa2
AA
Aa3
AA-
A1
A+
Strong payment capacity
A2
A
A3
A-
Baa1
BBB+
Adequate payment
capacity
Baa2
BBB
Baa3
BBB-
INVESTMENT GRADE
Ba1
BB+
Likely to repay; ongoing
uncertainty
SPECULATIVE GRADE, NON-
INVESTMENT GRADE
Ba2
BB
Ba3
BB-
B1
B+
High risk obligations
B2
B
B3
B-
CCC+
Vulnerable to default, or
in default
Caa
CCC
CCC-
Ca
C
In bankruptcy, or default
Moody’s
has no D
rating
D
Short-term debt, including commercial paper, is rated according to a different scale. See discussion infra
at 698-700.
140. See D
EBT
R
ATINGS
C
RITERIA
, supra note 38, at v. In 1986, those 40 professionals monitored
and rated the debt of approximately 1,250 U.S.-based entities, an average of more than 30 companies per
professional. See id. at 5.
141. See House, supra note 41, at 245.

Page 33
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
651
The number of rated issuers has increased by the same order of magnitude.
In 1975, 600 new bond issues were rated, increasing the number of outstanding
rated corporate bonds to 5,500.
142
Today, Moody’s rates 20,000 public and
private issuers in the U.S., and about 1,200 non-U.S. issuers, both corporations
and sovereign states; S&P rates slightly fewer in each category.
143
Moody’s
rates $5 trillion worth of securities; S&P rates $2 trillion.
144
Moody’s and S&P
thus dominate the world business of rating government and corporate debt.
145
In the mid-1980s, credit rating agencies limited their coverage to
predominantly U.S. corporations and the debt of fifteen sovereign states.
146
As
of 1981, for example, S&P rated only thirteen countries, all AAA, and by 1993,
S&P rated the debt of forty-three countries, including many so-called emerging
markets countries, which received low ratings.
147
By 1995, half of the fifty-two
countries rated by Moody’s and S&P were in the emerging markets category.
148
Although there is substantial consensus among credit rating agencies as to
the credit quality of top-investment-grade-issues, the agencies agree about less
than one-third of sub-investment-grade issues.
149
Credit ratings are not as
important or influential for lower-quality issues and investors in such issues rely
142. See Pinches & Singleton, supra note 132, at 31.
143. See House, supra note 41, at 245-46.
144. See id. at 246.
145. See Credit-Rating Agencies: Beyond the Second Opinion, supra note 10, at 80. A 1994 survey
of Chief Financial Officers indicated that issuers preferred S&P to Moody’s by more than two to one, and
noted that Moody’s was the slowest agency to upgrade issues and the fastest to downgrade. See Rating the
Rating Agencies, I
NSTITUTIONAL
I
NV
., April 1995, at 32. By contrast, in Europe, credit ratings are not yet
universal, as they are in the U.S., and the world’s leading credit raters are headquartered in the U.S. See
Ebenroth & Dillon, Jr., supra note 77, at 817. In the U.S., Moody’s and S&P face competition from Fitch
IBCA and Duff & Phelps. See Credit-Rating Agencies: Beyond the Second Opinion, supra note 10, at
80.
146. See House, supra note 68, at 53. In the past, countries did not seek a rating unless they were
sure it would be AAA. See Sovereign Debt: The Ratings Game, supra note 10, at 88.
147. See Sovereign Debt: The Ratings Game, supra note 10, at 88.
148. See House, supra note 41, at 248. Investors seem willing to accept lower credit quality in return
for higher yields, even though there have been 33 sovereign defaults since 1970. See id. Rating agency
rules provide that the sovereign credit rating sets the upper rating limit (known as the sovereign ceiling)
for all debt issuance in a country and thus is an essential precondition for opening up a country's capital
markets. See id. Neither Moody's nor S&P typically gives a corporate issuer a higher credit rating than the
country in which it is incorporated, the rationale being that the government could order the company to
default. See Sovereign Debt: The Ratings Game, supra note 10, at 88.
149. See House, supra note 41, at 248. Moody’s and S&P treat countries differently, too. Moody's
tends to favor Asia over Latin America. For example, in 1993, Moody’s rated China an A rating while
S&P gave China a high BBB rating; Moody’s rated the Philippines higher than Argentina while S&P
rated the two countries the same; and S&P rated Mexico higher than Moody’s did. See Sovereign Debt:
The Ratings Game, supra note 10, at 88. Despite these differences, S&P and Moody’s share 90 percent of
the sovereign ratings market. See id.

Page 34
652
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
on other sources of information in making their investment decisions.
150
The recent dominance and growth of a small number of rating agencies is
not necessarily consistent with the reputational capital view. At minimum, it
would have been a daunting task for Moody’s and S&P to maintain both market
share and high margins during a period of intense competition. Rating agency
analysts have faced competition from other analysts at sophisticated financial
services firms. Each rating agency analyst is responsible for tracking the credit
quality of up to thirty-five companies, but is paid significantly less than equity
analysts on Wall Street.
151
Yet the agencies have acquired considerable market
power during a time when they have not generated much additional valuable
information.
152
For the reputational capital view to make sense, Moody’s and
S&P must have continued to generate accurate and valuable information about
tens of thousands of companies, outperforming rivals in a highly competitive
market, throughout this period. Moody’s and S&P must have increased their
respective stocks of reputational capital very rapidly during the past decade
when their respective businesses have been booming.
Moreover, the process agencies use today to generate ratings does not
provide any obvious advantages over those used by competing information
providers and analysts. If agency processes are not unique, one would expect
the agencies to do their best to protect these processes from public view. In fact,
both Moody’s and S&P make rating determinations in secret. The agencies
never describe their terms or analysis precisely or say, for example, that a
particular rating has a particular probability of default, and they stress that the
ratings are qualitative and judgmental.
153
This secretive, qualitative process is
not the type of process one would expect if the agencies had survived based on
their ability to withstand investor scrutiny and accumulate reputational capital.
On the other hand, one might expect such processes in a non-competitive
market; if the rating process had been public or quantitative (rather than
qualitative), other market entrants easily could have duplicated the rating
agencies’ technology and methodology.
150. See House, supra note 41, at 248.
151. See Credit-Rating Agencies: Beyond the Second Opinion, supra note 10, at 80. Curiously,
analysts at rating agencies are judged not only based on predictions of performance, but also based on
predictions of how other rating agencies will change their ratings. See, e.g., Five Stars at the Rating
Agencies, I
NSTITUTIONAL
I
NV
., Aug. 1993, at 80, 80-81.
152. Moody’s and S&P insist they have not acquired market power, and at least one legal
commentator seems to believe them. See Rhodes, supra note 62, at 295-96 (praising rating agencies for
their role in generating information in a competitive marketplace).
153. See House, supra note 41, at 245.

Page 35
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
653
Consider, for example, S&P’s procedures.
154
When the agency is asked to
rate a new bond, representatives of the issuer meet with the agency’s analysts
and disclose facts they believe are relevant to the rating. After three weeks, the
analyst submits a report to a rating committee of up to ten professionals. The
committee meets in secret, and then votes.
155
If the lead analyst’s
recommendation is overruled and he or she protests, the matter can be referred
to an internal appeals court. The issuer can appeal if it is not satisfied with the
rating.
156
It is difficult to imagine how such a lengthy process could generate
timely, valuable information.
157
In addition, both S&P and Moody’s have high levels of staff turnover.
158
Both agencies have modest salary levels and limited upward mobility.
159
It is
questionable whether any agency could have sustained a dominant reputation
for sixty years, given such a process and organizational structure.
160
Perhaps the most important change in the credit rating agencies’ approach
since the mid-1970s has been their means of generating revenue. Today, issuers,
not investors, pay fees to the rating agencies. Ninety-five percent of the
agencies’ annual revenue is from issuer fees, typically two to three basis
points
161
of a bond’s face amount.
162
Moody’s and S&P have aggressively
154. Moody’s evaluation process is similar, and the two agencies reach the exact same results in two-
thirds of all cases. See House, Rating the Raters, supra note 68, at 53. For example, in 1990, the two
agencies assigned the same rating in 64 percent of approximately 1,400 cases. See id. Moody’s and S&P
differ in one important respect. Although S&P commits not to publish a rating the issuer doesn’t want,
Moody’s alone insists on its right to issue even unsolicited ratings. Moody’s claims such ratings contribute
to the agency’s credibility with investors. Competing agencies disagree: Fitch Investors Service has stated
that Moody’s uses the threat of an unsolicited rating to scare reluctant customers into requesting a rating,
and a managing director from Duff & Phelps Credit Rating Co. has stated that “[u]nsolicited ratings are
tantamount to blackmail.” See House, supra note 41, at 245.
155. See House, supra note 68.
156. The issuer typically learns of the rating before it is issued publicly. Only five percent of issuers
appeal their ratings and about 25 percent of issuers who appeal are successful, according to S&P. See
House, supra note 41, at 246.
157. It is possible, of course, that the agencies obtain non-public information. See discussion supra at
640.
158. See House, supra note 41, at 245.
159. Investment banks successfully recruit top rating agency employees. See id.
160. Moreover, in the late 1980s, Moody’s and S&P considered supplementing ratings with “event
risk” numbers in response to the concern that the sudden impact of a buyout or takeover could turn a AAA
bond into a non-investment grade bond overnight. Such considerations raise questions about how accurate
even S&P and Moody’s believe their own ratings are. See Close Your Eyes and Say AAA, F
ORBES
, Jan.
9, 1989, at 16.
161. A “basis point” equals one one-hundredth of a percent (i.e., 0.01%). Rating agencies began
imposing charges of up to $15,000 on each new issue they rated, beginning in the 1970s. See House,
supra note 68, at 53. In 1989, S&P charged 2.5 basis points per issue for a long-term rating; Moody’s
charged $10,000 to $25,000, depending on the issue’s size. See Wendy Cooper, Europe Wrestles with the
Triple-A Question, I
NSTITUTIONAL
I
NV
., Aug. 1989, at 51. More recently, fees of $90,000 are common

Page 36
654
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
expanded and now receive most of their revenue from the corporations they
rate.
163
Duff & Phelps
164
and Fitch IBCA
165
also charge fees to issuers.
Although investors are not paying directly for rating agencies to rate the
securities they buy, issuers who pay for ratings pass on the costs of those
ratings to investors by paying a lower return on debt issues. One reason the
payment to rating agencies may have shifted from investors to issuers is that the
information rating agencies generate has the characteristics of a public good.
For example, an agency publishing a rating to one or more individuals, for a fee,
will find it difficult to exclude other non-paying individuals from access to that
rating. Consequently, the agency should be able to collect higher fees. The
agency also can solve the free-rider problem associated with the provision of a
public good by having the issuer pay for the rating on behalf of all investors in
the issue.
Economically rational issuers will not pay more for a rating than the
expected benefit of the rating. Therefore, the issuer must expect that the
rating—and the informational content associated with the rating—will lower the
issuer’s cost of capital by at least the cost of the rating. Put another way, issuers
must expect that they are able to save at least two to three basis points on an
issue by having an agency rate it. Issuers also may consider the expected costs
of receiving a negative unsolicited rating.
It is doubtful that issuers (and therefore investors) receive two to three basis
points worth of informational value from a rating. For an issue of modest size,
say $100 million, the new information from the rating would have to be worth at
least $20,000. For the multi-trillion dollars of issues rated in aggregate, the
information generated by the agencies—if it is in fact information that issuers
(and therefore investors) are paying for—would have to be worth approximately
one billion dollars.
for complex deals. See House, supra note 68, at 53.
162. See, e.g., House, supra note 68, at 53; McGuire, supra note 15, at 10 ("Rating agencies are
normal profit making institutions which pay people bonuses on the basis of earnings performance. The
great bulk of those earnings comes from issuer fees.”).
163. See House, supra note 41, at 245.
164. Duff & Phelps, a Chicago financial advisory company, is even more aggressive than rivals S&P
and Moody’s; in 1991, it switched from paying salespeople a salary to paying straight commission and has
shown a willingness to rate unusual or new products. See Matthew Schifrin, Hidden Asset, F
ORBES
, Mar.
29, 1993, at 114.
165. In the late 1980s, Fitch Investors Service, a tiny 47-employee rating firm, was on the brink of
bankruptcy. Russell Fraser then purchased the firm for $11 million, including $1 million of his own
money. Fitch became more aggressive and doubled the fees the firm charges. By 1992, rating fees paid by
issuers accounted for 95 percent of Fitch's revenue. See Toddi Gutner, Ratings Shootout, F
ORBES
, Feb.
17, 1992, at 89.

Page 37
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
655
How could credit rating agencies generate information of such enormous
value in a competitive market, given their limited resources? Credit rating
agencies do not independently verify information supplied to them by issuers.
166
All rating agencies get the same data.
167
The rating agencies claim it is how they
interpret such information that draws distinctions among them in how they rate
debt.
168
These claims are dubious, given the above arguments.
Despite the fact that there are no obvious interpretive techniques unique to
rating agencies, it is undeniably true that rating agency profit margins are high.
Although the two largest agencies are subsidiaries of public companies and
therefore do not publish separate operating results, McGraw Hill’s financial
services division has had an operating margin of twenty-nine percent, and
Moody’s numbers are thought to be similar.
169
Such margins should not have
been sustainable over time in a competitive market.
The above discussion demonstrates two points inconsistent with the
reputational capital view. First, the view of credit rating agencies as prospering
based on their ability to accumulate and retain reputational capital does not
explain all of the dramatic changes in the value of ratings over time; it is not
plausible that four rating agencies have risen (or fallen) together in lock-step
based on their collective ability (or inability) to generate valuable credit
information. Second, during two critical periods—the 1930s and the mid-1970s
through 1990s—credit ratings increased in importance, each instance
paradoxically followed by a series of bond defaults demonstrating the rating
agencies’ serious mistakes in rating bonds. The next Part explores the empirical
and analytical support for these points.
III. P
ROBLEMS WITH THE
R
EPUTATIONAL
C
APITAL
V
IEW
During the discussion in Part II, I introduced a few of the conceptual and
theoretical difficulties associated with the reputational capital view of credit
ratings. In this Part, I describe in greater detail how certain types of recent
166. See Rhodes, supra note 62, at 317.
167. See, e.g., James R. Kraus, Making the Grade Its Own Way, A
M
. B
ANKER
, May 17, 1994, at 7.
S&P and Moody’s insist that their credit ratings are not swayed by the judgments of others. See Credit-
Rating Agencies: Beyond the Second Opinion, supra note 10.
168. See Kraus, supra note 167, at 7.
169. See Rhodes, supra note 62, at 317. For the first nine months of 1999, the McGraw-Hill
Companies, Inc. reported $945 million in revenue and $282 million in operating profit for “Financial
Services” (i.e., S&P), an operating margin of almost thirty percent. See the McGraw-Hill Companies, Inc.,
Form 10-Q, Nov. 13, 1999 (available in <http://quicken.elogic.com/>, using ticker symbol MHP (visited
Nov. 15, 1999).

Page 38
656
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
market participant behavior are inconsistent with the reputational capital view.
In particular, I analyze three problems: (1) inaccuracies in credit spread
estimation, (2) increases in ratings-driven transactions, and (3) the growth of
credit derivatives. These three developments present problems for the
reputational capital view because they are not consistent with the notion that
credit rating agencies have survived based on their ability to accumulate and
retain reputational capital. Instead, each development highlights serious flaws in
the rating process and raises questions about the informational content of
ratings: inaccuracies in credit spread estimation show that credit ratings do not
accurately capture credit risk over time; increases in ratings-driven transactions
show that market participants are engaging in transactions to obtain more
favorable ratings based on factors other than improved credit quality; and the
growth of credit derivatives shows how financial market innovation has
generated regulatory arbitrage
170
opportunities which both undercut and exploit
credit ratings.
These problems reveal some of the shortcomings of the reputational capital
view. They also will serve as background and support for a new theory of the
role of credit rating agencies in modern financial markets.
A. Credit Spread Estimation
There are serious problems associated with the estimation of credit
spreads.
171
Credit risk typically is described using the credit spread: the
difference between the yield
172
on a particular bond and the yield on a risk-free
bond with comparable cash-flow characteristics and maturity. Thus, the credit
spread is a reflection of the market’s estimation of the risks associated with a
particular bond compared to its risk-free counterpart. Credit risks must be
distinguished from market risks. Although most market risks can be hedged with
offsetting financial contracts, many credit risks simply cannot be hedged.
173
170. “Regulatory arbitrage” refers to transactions designed to eliminate or reduce regulatory or legal
costs. See Partnoy, supra note 20, at 227.
171. Credit spreads are not the only measure of credit risk. Another measure of bond-specific risk,
known as Option Adjusted Spread (OAS), also takes into account any optionality embedded in particular
bonds, including, for example, a call or convertibility feature. The OAS also is a measure of the riskiness
of a bond compared to its risk-free counterpart. See, e.g., Gregory R. Duffee, On Measuring Credit Risks
of Derivative Instruments, 20 J. B
ANKING
F
IN
. 805, 806 (1996).
172. By yield I mean the yield-to-maturity of the bond. See, e.g., Frank J. Fabozzi, Bond Yield
Measures and Price Volatility Properties, in T
HE
H
ANDBOOK OF
F
IXED
I
NCOME
S
ECURITIES
76-80 (3d
ed. 1991).
173. However, Professor John Hull has argued that a counterparty to a financial contract can virtually

Page 39
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
657
First, it is important to recognize how the rating agencies’ role relates to
credit spreads. Suppose Company A has issued bonds with five years remaining
until maturity. Those bonds pay a semi-annual coupon of 8 percent, return
principal at maturity, and are trading at a price of 100. The U.S. Treasury also
has issued bonds with five years remaining until maturity. Those bonds pay a
semi-annual coupon of 6 percent, return principal at maturity, and are trading at
a price of 100. The credit spread of Company A’s bonds is simply the difference
between the yield to maturity on Company A’s bonds (8 percent) minus the yield
to maturity on the U.S. Treasury bonds (6 percent), i.e., 2 percent. In market
parlance, the credit spread is 200 basis points.
The credit spread is the market’s estimate of the riskiness of the bond
compared to its risk-free counterpart, based on both the probability of default
and the expected recovery in the event of default. From a theoretical perspective,
there are four factors relevant to the decision to lend money, i.e., to extend
credit: (1) the lender’s cost of money, (2) the lender’s anticipated
174
return on the
loaned money, (3) the probability that the borrower will default, and (4) the
lender’s anticipated recovery in the event of default.
Suppose A wants to borrow $100 from B for one year. B has the following
information:
(1) B’s cost of money for one year is 5%,
(2) the anticipated return on the loaned money for one year is 10%,
(3) the probability that A will default is 10%, and
(4) B’s anticipated recovery if A defaults is $50.
Should B make the loan? B’s decision should be based on an analysis of the
expected profit (revenue minus cost) from the loan. B’s expected revenue from
the loan, calculated at the end of one year, is the product of the probability of no
default times the payment in full of principal and interest on the loan (90% x
$110 = $99) plus the product of the probability of default times the recovery in
eliminate credit risk by marking-to-market the financial contract. See J
OHN
H
ULL
, O
PTIONS
, F
UTURES
,
AND
O
THER
D
ERIVATIVE
S
ECURITIES
291 (1993). Marking-to-market refers to the daily settlement of a
margin account based on the value of future contracts. See id. at 4-9.
174. I use the term “anticipated” rather than “expected” to distinguish the general notion of
expectation from the mathematical notion of expectation. For example, if you believe you will receive a
particular payment, because that amount is the payment specified in the relevant contract, then I say you
“anticipate” such a payment. On the other hand, if you cannot anticipate a payment with certainty, then I
say you “expect” a payment equal to the sum of the probabilities of each possible payment times the
amount of that payment, i.e., the mathematical expectation.

Page 40
658
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
the event of default (10% x $50 = $5), for a total of $104, or a return of $4.
175
B’s cost of money, calculated at the end of one year, is $5 (5% x $100), so B
should not make the loan.
176
In a competitive market, the borrower and lender would bargain over the
anticipated return on the loaned money, and, in equilibrium, the lender’s
anticipated return on the loaned money would have to rise to approximately
11.1% before the lender would be willing to make the loan.
177
If the risk-free
rate for the same maturity were 4%, then the credit spread for this particular
loan would be 7.1%, or 710 basis points.
What determines the credit spread? The above example assumed the lender
has perfect information about the four factors to be considered. In fact, this
assumption necessarily is true as to the first two items: in every case in which a
lender is deciding whether to extend credit, the lender will know its cost of
money for the relevant period and the specified return on the loaned money
during that period. This information is what distinguishes credit instruments
from all other financial instruments. Put another way, a credit instrument is a
financial instrument whose return and maturity are specified. In contrast, a
purchaser of stocks would not know the amount of dividends plus capital gain
for her investment.
However, a lender will almost never have perfect information about the last
two items: the probability of default and the expected recovery in the event of
default.
178
It is the uncertainty and imperfect information associated with these
175. More generally, suppose p is the probability that A will repay B a given loan amount, L, in full.
Then (1-p) is the probability that A will not repay B in full. Further suppose x is the future value of the
amount B will collect if A does not repay B in full. (Put another way, x is the market value of the loan after
an event of default.) Finally, suppose r is the anticipated rate of return on a loan. Then mathematically, the
expected return on the loan is p*L*(1+r) + (1-p)*x.
176. This analysis assumes B is risk neutral or risk averse. If B is risk seeking, B might decide to lend
A $100, because B “overweights” the probability of A not defaulting.
177. The precise interest rate that solves the preceding equation is 11 1/9%. Mathematically,
0.9*100*(1+0.111) + (1-0.9)*50 =~105. More generally, a credit loss occurs only if (1) the counterparty
defaults, and (2) the credit exposure on the instrument is positive. See Duffee, supra note 171, at 810. The
credit spread premium depends primarily on the probability of default. See Jerome S. Fons, Using Default
Rates to Model the Term Structure of Credit Spreads, F
IN
. A
NALYSTS
J., Sept./Oct. 1994, at 25.
Although the model introduced by Jerome Fons provides a minimum estimate of credit spreads, its
estimates are lower than actual spreads on corporate bonds because the model only takes into account
credit risk and not other factors including liquidity, taxes, and capital requirements, each of which will
increase credit spreads. See Leland E. Crabbe, Estimating the Credit-Risk Yield Premium for Preferred
Stock, F
IN
. A
NALYSTS
J., Sept./Oct. 1996, at 49.
178. In the above mathematical formula, supra note 175, the variables L and r are observable, but p
and x are not. A lender should be willing to pay an information provider for information about the values
of p and/or x if the cost of that information is less than the expected net benefit from the information.
There is some empirical evidence of the expected recovery, x, in the event of default. According to

Page 41
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
659
two items that generates variability in credit spreads and, therefore, the demand
for credit information. Put another way, the credit spread reflects the market’s
best estimate of these two variables. The major rating agencies claim, and the
reputational capital view argues, that credit ratings have value because they
include additional information not already reflected in credit spreads.
The major rating agencies claim their ratings have informational value
because ratings are highly correlated with actual credit spreads. However, this
claim, even if true, does not establish that credit ratings have independent
informational value. The credit spread is a reflection of all available information
in the market, including the rating. The correlation between ratings and credit
spreads could be a reflection that either the market is reacting to the
informational content of credit ratings or credit ratings are reacting to the
informational content of the market.
A more precise analysis of the informational content of credit ratings must
involve the behavior of credit spreads of bonds rated in the same category over
time. If credit ratings are accurate and reflect information not already reflected
in the market, one would expect credit spreads of bonds given a particular rating
to remain relatively constant over time. Additionally, if credit ratings are
accurate, one would expect bonds with the same credit rating to have the same
credit spread.
In fact, studies by financial economists and anecdotal evidence (including
public statements by these economists) suggest that credit ratings have become
less accurate over time, and that credit spreads of bonds in particular rating
categories have changed dramatically.
179
James Van Horne, a professor of
finance at Stanford, has concluded that “[w]hile the assignment of a rating for a
new issue is current, changes in ratings of existing bond issues tend to lag
behind the events that prompt the change.”
180
Kenneth Lehn, a professor of
one study, the trading values of various types of debt in the event of default, measured one month after the
default, are as follows (as a percent of the par amount of the bond, or 100):
Senior Secured
53.3
Senior Unsecured
44.6
Senior Unsubordinated
36.0
Subordinated
28.7
Junior Subordinated
16.3
Preferred Stock
6.0
See Crabbe, supra note 177, at 46.
179. See discussion supra notes 132-36.
180. See J
AMES
C. V
AN
H
ORNE
, F
INANCIAL
M
ARKET
R
ATES AND
F
LOWS
181 (1990). There is,
however, substantial evidence that equity returns are influenced by changes in credit ratings. See, e.g.,
I
LIA
D. D
ICHEV
& J
OSEPH
D. P
IOTROSKI
, T
HE
L
ONG
-
RUN
S
TOCK
R
ETURNS
F
OLLOWING
B
OND
R
ATINGS

Page 42
660
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
business administration at the University of Pittsburgh (and an advisor to
Moody’s) has concluded that only seventy-five percent of the ratings process is
based on statistical information and equations, and that twenty-five percent is
subjective.
181
Frank Packer’s initial research for the Federal Reserve Bank into
the movement of sovereign bond yields indicates that yields typically decline
several days before the agencies act on a rating, suggesting that the agencies lag
behind the market.
182
Moreover, economists have criticized agency ratings
because although the agencies publish volumes describing how they award
ratings and the basis for their decisions, they do not describe the probability of
default. According to Richard Cantor, an assistant vice president at the Federal
Reserve Bank of New York, the market has never required that the agencies
define their ratings more clearly.
183
The New York Federal Reserve Bank has expressed concern that ratings are
not consistent over time and can be influenced by competitive pressures.
184
Similarly, other types of ratings that depend directly on credit ratings have
exhibited similar problems and inconsistencies. Insurance regulators, in
particular the Securities Valuation Office (SVO) of the National Association of
Insurance Commissioners (NAIC), became concerned in 1995 that reliance on
credit ratings as a trigger for more lenient regulatory treatment might
inadvertently mask critical distinctions relevant for financial solvency
monitoring.
185
Numerous professionals view the ratings as “rearview mirror”
analyses. According to Professor Bruce N. Lehmann of Columbia University
Business School, ratings “are lagging indicators of credit quality. I have never
known a portfolio manager who goes by the ratings.”
186
Credit estimation mistakes are especially acute with respect to sovereign
C
HANGES
(University of Michigan Business School Working Paper, Oct. 1998) (copy on file with author)
(finding that after bond downgrades the corresponding stocks earned abnormally low returns while after
bond upgrades the corresponding stocks earned normal or slightly above-normal returns for up to one year
after the rating change).
181. See Lyn Perlmuth, Is Turnabout Fair Play?, I
NSTITUTIONAL
I
NV
., April 1995, at 34 (quoting
University of Pittsburgh professor of business administration Kenneth Lehn); see also Letter from Kenneth
Lehn, supra note 13, at 4.
182. See Richard Cantor & Frank Packer, Determinants and Impacts of Sovereign Credit Ratings,
F
ED
. R
ESERVE
B
OARD
N.Y. E
CON
. P
OL
Y
R
EV
., Oct. 1996, at 45-46.
183. See House, supra note 68 (quoting Cantor as stating “[y]ou can bet that if this was all
quantitative, others could duplicate the model and new players might be drawn into business.”)
184. See Limits Cited in Judgment of NRSROs, 5 I
NS
. A
CCT
., A
M
. B
ANKER
, April 10, 1995, at 5.
185. See id.
186. See David Zigas, Why the Rating Agencies Get Low Marks on the Street, B
US
. W
K
., Mar. 12,
1990, at 104.

Page 43
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
661
ratings, a new and important line of business for the agencies.
187
Even AAA
ratings are neither accurate, nor consistent across bonds.
188
Nevertheless,
although the most recent and complete study of sovereign credit ratings, by
Cantor and Packer, found that eight publicly-available statistics predicted
eighty-six percent of the variation in ratings, the study also found that ratings
directly affected corporate securities prices.
189
These findings that the ratings
changes affect market prices are in sharp contrast to the findings from the
1970s.
190
Even the most sophisticated market participants are unable to generate
accurate methods of estimating credit risk, despite the importance of doing so.
For example, bank regulators need accurate measures of the credit risk involved
in financial derivatives transactions so that capital requirements can be set
optimally.
191
Yet even the most sophisticated current methods of analyzing
credit risk are seriously flawed.
192
Estimates of future credit exposures
generated by “Monte Carlo” simulations ignore the uncertainties in the models
used to generate such estimates, and therefore can vary by as much as sixty
percent.
193
Each participant has its own procedures, which vary and change
quickly, and the most sophisticated techniques are proprietary.
194
Even
187. For example, Moody’s gave its highest commercial paper rating to the Spanish bank Banco
Espanol de Credito, just before it collapsed; Turkey was running a deficit equivalent to at least 15 percent
of GDP yet was rated investment-grade, and the rating agencies were among the last to notice Turkey’s
problems and downgrade the sovereign credit rating; on the other hand, the rating agencies held firm with
sub-investment grade ratings for Mexico’s sovereign debt during 1993 and 1994, when the market was
anticipating an upgrade and yields were trading near investment-grade levels (the rating agencies later
claimed they had been correct when the Mexican peso and markets crashed on December 20, 1994). See
House, supra note 41.
188. See, e.g., Cooper, supra note 161, at 51 (“Even within the charmed triple-A circle, wide price
discrepancies exist depending on the type of borrower involved. One new-issues maven cites a 45-basis-
point gap between a triple-A sovereign and a triple-A U.S. corporate bond issued recently on the same
day.”).
189. See Cantor & Packer, supra note 182, at 45. For other studies finding that credit rating changes
had an impact on market prices, see Jeremy Goh & Louis Ederington, Is a Bond Rating Downgrade Bad
News, Good News, or No News for Stockholders?, 48 J. F
IN
. 2001 (1993); John Hand et al., The Effect
of Bond Rating Agency Announcements on Bond and Stock Prices, 47 J. F
IN
. 733 (1992).
190. See discussion supra at 647-48.
191. See Duffee, supra note 171, at 806.
192. The primary difficulty with modeling credit risk is that it is non-linear, and therefore requires
complex computer simulations for valuation. There have been serious criticisms, for example, of Credit
Metrics, the proprietary credit risk-management system J.P. Morgan offers as a risk management tool to
quantify credit exposure. See J.P. Morgan to Launch Credit Risk Data Set, D
ERIVATIVES
W
K
., Feb. 3,
1997, at 1.
193. See Duffee, supra note 171, at 806. A “Monte Carlo” simulation is a valuation methodology in
which a computer model simulates different price paths and then generates a valuation based on the
probability of particular paths. See id.
194. See id. at 812.

Page 44
662
WASHINGTON UNIVERSITY LAW QUARTERLY
[
VOL
. 77:619
sophisticated computer models don’t work particularly well in estimating credit
spreads. Finance theorists continue to debate whether their theoretical models
are correct, and S&P’s and Moody’s methods of rating Derivative Product
Companies (DPCs),
195
which engage in the most sophisticated attempts at
estimating credit and credit spreads, are questionable.
196
Recently, finance
theorists have seized on the ways in which credit changes can affect duration, a
mathematical measure of the risk associated with a particular credit
instrument.
197
The consensus of work on the topic is that credit risk shortens the
effective duration of corporate bonds, yet many assessments of credit do not
account for such changes.
198
Finally, studies of so-called “split ratings” also provide evidence of credit
spread differentials which are not consistent with the reputational capital view.
A split rating occurs when two rating agencies rate the same bond differently.
For example, Moody’s might have rated bonds Aaa, while S&P rated the same
bonds AA. Split ratings do not support the conclusion that ratings reflect
valuable and accurate information, because if ratings were valuable and
accurate, one would expect that ratings from different agencies would not
consistently agree. In fact, S&P’s and Moody’s ratings have a correlation of
approximately ninety-seven percent.
199
In addition to the above financial studies, there is abundant anecdotal
evidence of rating inaccuracy. Most recently, there has been a substantial
increase in defaults. According to Moody’s, companies defaulted on $22 billion
of bonds in 1990, the highest figure in twenty years.
200
Only a few months
before Orange County filed for bankruptcy in December 1994, both S&P and
Moody’s gave the county their highest short-term rating for a $600 million
taxable note issue.
201
In early 1997, several sub-prime auto lending institutions,
195. See discussion infra at 668-70.
196. See Charles Scheyd and Reza Bahar, Derivative Product Company Rating Criteria, June 6,
1994 <Error! Bookmark not defined.> (visited Oct. 1, 1998). Scheyd and Bahar set forth S&P’s
approach to rating DPCs, and suggest both that S&P is willing to use any computer model supplied by a
private party that can adequately address the potential risks that a DPC might face, and that the approach
is an unfixed, dynamic process, subject to change based on financial innovation and on changes in law. I
discuss DPCs in greater detail infra at 668-70.
197. See Babell et al., Default Risk.and the Effective Duration of Bonds, F
IN
. A
NALYSTS
J.,
Jan./Feb. 1997, at 35.
198. See id. at 41.
199. See, e.g., Cantor & Packer, The Credit Rating Industry, supra note 74, at 25.
200. See Credit-Rating Agencies: Beyond the Second Opinion, supra note 10, at 80. Of course, the
amount of debt rated by Moody’s also has been increasing.
201. See Barry B. Burr, Credit Raters Miss Many Danger Flags, P
ENSIONS
& I
NV
., Jan. 23, 1995,
at 11.

Page 45
1999]
TWO THUMBS DOWN FOR THE CREDIT RATING AGENCIES
663
including Mercury Finance Co. and Jayhawk Acceptance Corp., announced
accounting irregularities and bankruptcy filings, in part because the companies
were posting earnings based on “little more than an educated guess about how
much cash will flow in over three or four years.”
202
The rating agencies
downgraded the issues only after these problems became public. There are
similar, major inaccuracies in estimations of credit related to the insurance
industry.
203
And most recently, there have been defaults related to the Asian and
Russian financial crises.
204
Ratings increasingly are subject to political pressure. In the late 1980s, when
New York State, running a large deficit and facing serious revenue shortfalls,
sought ratings for its short-term borrowings, analysts expected a downgrade.
However, the rating agencies, under pressure, decided not to change their top
ratings; Moody’s decision came after a one-hour meeting with New York
Governor Mario M. Cuomo.
205
If ratings were accurate, these credit spreads should have remained relatively
constant. This is because a particular credit spread, say 100 basis points,
reflects the market’s expectation with respect to the probability of default and
payment in the event of default, as well as the supply and demand for a
particular issue. This market expectation includes the information associated
with already-published credit ratings. Accordingly, one would expect that if
credit ratings were accurate, bonds with credit spreads of 100 basis points
would have the same ratings. As previously noted, bonds of a particular given
credit rating should have the same credit spread.
Absent some explanation of why the market would ignore credit ratings, the
fact that the credit spread for a given rating changes so much over time cannot
be an indication that the market is not accurately pricing bonds with given credit
ratings, because a bond’s market price by definition has incorporated already-
published credit ratings.
206
One would only expect that the market price would
202. See Mark Maremont & Rick Melcher, Hard Answers for the Crisis in Easy Credit, B
US
. W
K
.,
Feb. 24, 1997, at 118.
203. Buying insurance exposes the buyer to the credit risks of the insurer. This exposure is
substantial. Americans spend approximately five percent of their disposable income on life insurance.
There are more than 150 million Americans covered by life insurance, and the U.S. insurance industry had
$1.4 trillion in assets, as of 1991. See Larry Light, et al., Are You Really Insured?, B
US
. W
K
., Aug. 5,
1995, at 42 (warning purchasers that the insurance industry is “notoriously opaque, with its own
bewildering lingo” and noting that raters failed to warn of the dangers at Executive Life and Mutual
Benefit Life Insurance Co. before such dangers became public).
204. See C
HARLES
<div style="position:absolute;top:48562;left:2