Centre for Economic Policy Research

partly the result of barriers to entry. The cost of acquiring the necessary 
reputation to function as an effective rating agency is a natural barrier to new
entrants, but its height may have been artificially increased by the procedures by
which reputation is now officially recognized for regulatory purposes.
4.5
Do rating agencies add value?
The fact that ratings are widely used and that most issuers are prepared to pay to
receive them cannot be regarded as conclusive evidence that agencies add value.
The privileged position created by market practice and regulatory requirements
46 Conflicts of Interest in the Financial Services Industry

---

could generate a demand for the services of rating agencies that is independent of
the underlying value of their assessments (Partnoy, 1999; Wakeman, 1984; White,
2001). Analysis of the contribution of the agencies has therefore relied on other
means of judging the quality and influence of their assessments.
One approach is to simply look at the accuracy of ratings. Since ratings are
intended as a guide to the likelihood of the rated entity servicing its obligations
on time and in full, correlations between ratings and subsequent default history
provide an initial measure of their accuracy. Many studies have shown that,
notwithstanding prominent counterexamples such as Enron, there is a reasonably
close correlation between ratings and default probabilities (see studies reviewed in
Bank for International Settlements, 2000). The agencies themselves also provide
evidence of the relationship between their ratings and subsequent default history
(Brand and Bahar, 1999; Keenan, 1999).
For several reasons, however, this cannot be regarded as demonstrating the
value of the information supplied by rating agencies. In the first place, such 
correlations do not show how much rating agencies’ judgements add to what can
be gleaned from other indicators of credit quality, such as interest spreads, or
other sources of credit analysis, including analysts’ and auditors’ reports. Partnoy
(1999) surmises that the added value might be quite low, in view of the fact that
the staff working in rating agencies are spread very thinly. For example, at
Moody’s an analyst rates an average of 35 issues.
Second, while credit agencies’ assessments may reflect the relative risk at a
point in time, the correspondence of ratings to default probabilities seems to be
subject to substantial change over time (Cantor and Packer, 1994). If a given grade
is supposed to measure probability of default, the correlation between rating
grades and subsequent defaults should be more stable.
A more powerful test of the extent to which the market uses ratings is to look
at how market prices respond to changes in ratings. If a rating provided 
information over and above that which was already reflected in market prices, it
is to be expected that market prices would react in a systematic manner to changes
in ratings. Most academic studies confirm this expectation, although the effects
for downgrades are stronger than for upgrades.
40
Even this cannot be regarded as conclusive. It is possible that observed rela-
tionships reflect an element of reverse causality. In other words, credit 
downgrades may not necessarily reflect an underlying weakening in credit 
quality; they may themselves be a determinant of credit quality via their influence
on financing costs. This possibility appears to have become more realistic in
recent years, as ratings have been increasingly used in bond covenants, and 
ratings downgrades are more often used as triggers for contract revisions. Such
triggers seem to have played a significant role in the end-game of a number of
recent corporate failures, such as Enron and Worldcom. 
Another reason some observers have offered for questioning the finding that
rating actions are associated with market reactions is that rating adjustments often
come out contemporaneously with other news (Partnoy, 1999). In an effort to get
around this problem, Kliger and Sarig (2000) analyse the rating refinement made
by Moody’s in 1982. At that time, Moody’s effectively tripled the number of 
rating categories by adding a + or – suffix to existing ratings. Kliger and Sarig’s 
contention is that the addition of a suffix provided previously unknown 
additional information about Moody’s judgement, with no change in the 
situation of the entity being rated. If this information was regarded as having
value, a systematic change in the relevant market prices for bonds should be
observable. Their results show that there was a statistically significant effect, 
suggesting that the market values the judgements of rating agencies.
41
Rating Agencies: Conflicts of Interest in Credit Assessment and Consulting 47

---

Although there are few studies that attempt to determine whether rating 
agencies, analysts or other monitors contribute more information to the market,
they do conclude that rating agencies independently provide additional 
information (Ederington and Goh, 1998). In any event, it is difficult to provide a
conclusive assessment of the exact contribution of rating agencies because the 
picture is muddied by the regulatory situation. It is always possible for the 
agencies’ critics to argue that the observed impact of rating agencies’ judgements
on the market is due to a reverse causation, in other words a downgrade leads to
a weakening of a firm’s prospects, rather than a weakening of prospects being 
identified by a rating agency and revealed to the market in a rating downgrade.
The growing practice of tying interest rates on certain categories of debt to credit
ratings is likely to amplify this effect.
One thing is hard to dispute, however. Agencies’ ratings are a convenient 
shorthand for synthesizing information about credit quality (Wakeman, 1984;
Bank for International Settlements, 2000). They are looked at by the market, and
widely used in financial covenants. Borrowers expend considerable effort to secure
a rating, so presumably they believe there are advantages in having a good one.