Supplementary memorandum by Fitch Ratings
This memorandum responds to the Committee's
requests at the evidence session on 13 November 2007.
1. Requested Information with respect to Fees:
(a) Please disclose all fees received by Fitch Ratings, and
any sister company within the same group, from Northern Rock plc
("NR") during the period 2003 to 2007. (b) Were
these fees proportionate to fees paid by institutions comparable
to NR for similar services? [Q977]
1.1 Fitch Ratings received £3,857,011
(excluding VAT) with respect to ratings and £46,627 for one
of our subscription products. The ratings fees relate to the rating
of NR and its debt issuances, as well as discrete RMBS and covered
bond transactions. With respect to the subscription fees, there
is no significant difference between the price paid by NR in each
of the relevant years, and the average price paid by other UK
banks and non bank financial institutions which subscribed for
the same service in those years. With respect to the ratings work,
the fees paid by NR in each of the relevant years were in line
with the average fees paid by similar financial institutions,
and often lower than the average. Please note that the amount
of fees with respect to ratings work equates to 0.24% of the total
revenues of Fitch, Inc. during this time period.
1.2 No sister company of Fitch Ratings received
any fees from NR.
2. Requested Information with respect to Current
Ratings: "It would be very helpful if we could have an analysis,
Chairman, by rating category, by type of issuer, the volumes and
the number of issuers". [Q1053]
2.1 Please see attached spreadsheet.
3. Requested Information with respect to Bank
of England Suggestions: Provide your views on the following statements
and proposals made by the Bank of England to improve the information
content of ratings:
(a) "Agencies could publish the expected
loss distributions of structured products, to illustrate the tail
risks around them. Agencies have made significant efforts over
the past few years to increase the transparency of their rating
methodology for structured finance products, through publication
of research reports describing their modeling methodology and
their assumptions on correlations and recovery rates. But published
distributions could provide a visual reminder of the fatter tails
embedded in the loss distribution in structured products".
(b) "Agencies could provide a summary
of the information provided by originators of structured products.
Information on the extent of originators' and arrangers' retained
economic interest in a product's performance could also be included.
Such a summary may satisfy investors that incentives are well
aligned or encourage investors to perform more thorough risk assessments".
(c) "Agencies could produce explicit
probability ranges for their scores on probability of default.
Probability ranges would provide a measure of the uncertainty
surrounding their ratings. Although such figures are already available
retrospectively as transition matrices, an explicit probability
range would allow investors to monitor agencies' performance when
rating different asset classes".
(d) "Agencies could adopt the same scoring
definitions. Currently, some use probability of default, some
loss given default and others a combination. Converging on a single
measure would reduce the risk of misinterpretation by investors".
(e) "Finally, rating agencies could
score instruments on dimensions other than credit risk. Possible
additional categories include market liquidity, rating stability
over time or certainty with which a rating is made. Clear scores
on these dimensions could encourage more sophisticated investment
mandates and easier monitoring of non-credit risks in a portfolio.
It would clearly take some time and money for agencies to develop
the necessary expertise on these other risks, but some agencies
have already proposed these additions".
3.1 Fitch Ratings initiated a dialogue with
the supervisory community, including central banks around Europe,
in the Spring and Summer of 2007. At individual meetings with
central bankers during this period, many of the topics covered
in the suggestions set forth above were raised and discussed.
3.2 With respect to (a) above, we are currently
considering whether we could publish expected loss distributions.
As discussed with the Bank of England in July of this year, this
approach may lend itself more easily to some asset classes than
to others (for example, it would be manifestly easier for those
asset classes where the analytical process already creates a stochastic
distribution of losses). Additionally, there would be challenges
in ensuring that such distributions could be readily communicated
to users in a concise and unambiguous format.
3.3 With respect to (b) above, we support
any initiative to enhance transparency and require more public
disclosure from originators for the benefit of investors. We believe
that the structured finance market is too opaque and that transparency
can best be achieved by originators, servicers, issuers and arrangers
making publicly available all information, both at issuance and
throughout the life of a structured finance transaction, that
they make available to rating agencies (in the context of their
rating analysis). We further believe that the internet provides
an excellent venue to host this information so that it is freely
and publicly available to all interested parties. The public availability
of this information would enable investors to make fully informed
investment decisions without having to rely solely on credit ratings
to determine credit risk and would allow investors to assess independently
issues such as market and liquidity risk that credit ratings do
not address. We do not think, however, that it should be the responsibility
of the rating agencies to disclose the originator's data. Our
Code of Conduct prohibits us from disclosing confidential information,
consistent with the IOSCO Code of Conduct Fundamentals for Credit
Rating Agencies. Moreover, with respect to disclosing any interest
retained by the originator or arranger, rating agencies will not
necessarily know who holds what proportion of the issued bonds.
3.4 With respect to (c) above, we have been
recognised, or are currently applying for recognition, as an External
Credit Assessment Institution ("ECAI") in all major
markets where we operate. One part of this recognition process
involves an explicit comparison between the historical default
experience of an ECAI against a target band of cardinal default
rates. This comparison is repeated on a regular basis going forward
in assessing the performance of ECAIs over time. As we intend
to maintain our ECAI status in future, we will be subject to a
de facto review of our cardinal default process by every bank
supervisory authority in whose jurisdiction we are recognised
as an ECAI. Additionally, we are considering how we may better
communicate publicly the analysis we carry out internally of the
movements in historical default and transition rates, relative
to long-term trends.
3.5 With respect to (d) above, we note that
the topic of diverging definitions has been extensively reviewed
as part of the wider debate surrounding comparability of rating
scales over the past several years. In general, the distinctions
in rating definitions among rating agencies are modest. The most
independent review of this topic has been done in the context
of the above-noted ECAI recognition process, which includes a
"mapping" or correspondence table, established by each
bank supervisory authority to plot rating agencies' scales against
default rates prescribed by the Basel II process. This process
also results in a mapping table plotting the ratings of one agency
against the ratings of other agencies. It is important to note
that this process did not result in pressure from the user base
to standardise rating definitions. The topic has also been raised
more recently at hearings in October, convened by the Committee
on the Global Financial System, at the Bank of England's offices,
involving investors, arrangers and central bank representatives.
The consensus was that the current modest variation in rating
definitions reflected a healthy variety in methodological approaches
and at the same time did not provide material obstacles to comprehension
or application of the ratings by users. Movement towards a stricter
homogeneity of output would likely not serve the market well and
would reduce investor choice. In addition, there would be a variety
of practical problemsfor example, how (and by whom) would
such a set of common definitions and related criteria be developed
and maintained. We believe, however, that the market would benefit
from a common definition of default for purposes of the production
of transition and default studies by the rating agencies and any
other interested parties. A common definition of default would
provide investors and all interested parties the ability to more
readily compare these studies against each other and thereby more
accurately assess the relative performance of rating agencies.
3.6 With respect to (e) above, as discussed
with the Bank of England at our July meeting, we are currently
investigating whether we could provide opinions with respect to
risk elements other than credit risk. For example, we have been
working on ways to measure rating volatility, including stability
scores. In fact, Fitch was the first agency to introduce such
scores, after a lengthy development period, in early 2006. At
the time we initiated Stability Scores, these were designed precisely
to highlight the differing performance characteristics of instruments
which carried the same rating on the "AAA" scale. At
the time, it is fair to say the market was less interested in
rating stability, or indeed rating differentiation, given the
generally much higher appetite for risk at that time. As a result,
penetration of the Stability Score concept in the market was limited.
In the meantime, we have worked on revising the Stability Scores
to broaden the scope of products covered. Equally, it is fair
to assume the market is now more sensitised to risk differentiation
than it was in March 2006. As a result, we will be looking at
proposing an enhanced version of the Stability Score process in
More generally, Fitch has done much work researching
the best way in which to reflect additional risks affecting issuers
and transactions. Our discussions with institutional investors,
central banks and other interested parties have indicated a marked
preference for discrete scales for discrete dimensions of risk
(ie, one scale covering default risk, one scale covering loss
severity, etc) rather than combinations of risks on the same scale.
Examples of successful multiple-scale sectors include our bank
and insurance sectors, with a range of Support, Individual and
Financial Strength ratings complementing the mainstream Issuer
Default Ratings. In contrast, other areas where multiple scales
have been introduced have seen more muted investor uptake. It
is important to note that we cannot "force" usage of
any of our scales, and equallyand reluctantly sometimeswe
will withdraw scales for which we perceive low investor demand.
12 The financial institutions used for purposes of
this comparison were Bradford and Bingley, Abbey, HBOS, Alliance
and Leicester and Nationwide. Back