Examination of Witnesses (Questions 1000
TUESDAY 13 NOVEMBER 2007
Q1000 Ms Keeble: When it is peeled
off now, it is quite clear that there was gross mis-selling, or
what would be deemed to be mis-sellingpeople could not
repay, they did not have the income, all kinds of thingsso
how reliable can your statistical tools have been? I could understand
that if you applied them to middle income mortgages they might
be very reliable, but sub-prime is different, is it not?
Mr Bell: It is, and that is why
our rules for sub-prime and the way in which we stress them is
very different. We do not stress them at all the way we stress
prime mortgages. Clearly there are lessons that we are going to
have to learn about the US sub-prime. Clearly we are looking at
what went wrong and how we could possibly learn from this and
how we can change our criteria, change our tools of analysis.
Are there any items we should be looking at that we did not look
at before? Such crises are always an opportunity for us to learn.
Q1001 Ms Keeble: I want to ask you
in a minute about the lessons you have learned, but you have also
said repeatedly that your advice functions are different from
your ratings functions and that you have got different structures,
different agencies. You say that, but everybody else who is coming
here (and obviously we all talk to people informally too) says
there are conflicts of interest: that you advise and that you
rate on the same products. How is everybody else so mistaken about
what you are doing?
Mr Bell: I cannot answer that.
What I can say is that we have a potential conflict of interest
which is known and managed. We simply do not provide advice. It
is very difficult for me to say anything more than that. We do
not have advisory functions, we do not have a consultancy function
for structured finance, our analysts are hired and our company
is hired to rate transactions. We do not have any advisory contracts.
Q1002 Ms Keeble: Okay. You have sat
here, Mr Bell, and you have talked about the cushion that is needed,
this and that and the other, to get a triple-A rating. That comes
perilously close to saying, if you do this so we can tick these
boxes, you can get a triple-A rating. You have sat there and said
it, and we will see it when the transcript comes out. How do you
actually make sure that other people do not listen to you, say
"Oh, well, to get the triple-A, I have to do X, Y and Z and
then I will get it, and that is the model and that is all I need
to do"? How do you safeguard against that?
Mr Bell: All our ratings are done
by a committee; all of our rated transactions and structured finance
are analysed. Because our criteria are transparent and available
to the market and available to investors so that they can understand
how we rate, we are very conscious that there will be a tendency
by investment bankers and market participants to game our ratings
and, therefore, the rating process is never a mechanical one.
We always look at each transaction and try to understand and try
to see whether or not anybody has tried to game our criteria.
Q1003 Ms Keeble: Can you say whether,
as a result of this, you have actually tightened up things? Do
you actually sit on doing the final assessment or do you have
a completely arms-length group of people who get all the data,
a bit like the MPC, I suppose, and look at it all and then think,
Mr Bell: Different from whom?
Because we do not have an advisory function.
Q1004 Ms Keeble: You have sat here
and described how we could go about. You have described
briefly some of the things that are needed to get a triple-A rating.
Do you actually take decisions on the ratings that people get
or do you have some people who will assess a product, do all the
reports and then a separate group of people who take the decisions
on what ratings they should get?
Mr Bell: The latter. Maybe it
is easier if I just explain the way in which we operate. Our criteria
are public so that market participants know how we apply the rules.
They have provisions so that we can have different criteria if
we feel someone is gaming them, but basically our criteria is
public. The client will approach us; an analyst will be assigned
to a particular transactionthat is the primary analyst.
Sometimes on big transactions, complex transactions, there may
be two analysts. They will gather the information; they will ask
the questions that they believe they should have obtain answers
to in order to achieve the rating. They will then, once they have
done this, go to a committee and they will present the conclusions
of their work to the committee. The committee will vote on the
rating. That is how we operate.
Q1005 Ms Keeble: Do the assessors
get information from separate sources or just from the client?
Mr Bell: They will get assessment,
they will get information from all the sources they deem relevant,
so they will get it from the client, they may get it from the
press, they will get it from other market participants if need
Q1006 Ms Keeble: What happens if
the client says, "If I tweak it here or there, will I get
a triple-A"? What do you do then?
Mr Bell: We basically answer the
issue of our criteria. So if a client says, "I thought I
had followed your criteria. I thought I would get a triple-A.
I have not got a triple-A. What is wrong? Why did I not get a
triple-A?" We will give them an answer and say the criteria
had not been followed.
Q1007 Ms Keeble: But before hand,
I mean, when you are doing the assessment?
Mr Bell: This is what I am talking
about. When we are doing the assessment the client will sometimes
come to us and say, "I do not understand; I thought I had
met the criteria", and we will explain, "No, we do not
believe the criteria has been met", and then they will decide
whether they want to go ahead with the transaction or maybe they
want to change the transaction.
Q1008 Ms Keeble: How about the others?
How does what you have described compare with the other agencies?
Mr Drevon: I think the committee
process is probably somewhat similar. One thing to note which
I think is important is that the reason why we have transparent
methodology being made available to the market is because we think
it is good practice. We were being accused a few years ago of
being black boxes: a client, an arranger comes to us and asks
us to rate a transaction, we give the rating but we do not explain
the rationale for that rating. We have taken many steps over the
last few years, in fact, to become much more transparent and make
available our criteria to the market place. From a policy point
of view, I think this was the right track to take.
Q1009 Mr Simon: Going back to the
extent to which you agree or not and why, you started off by saying
that the ratings that you issue measure different things, they
are not all the same; so S&P measures default probability,
Moody's measures expected loss. Then, the next thing you said
is that it is not true that there is an extraordinarily high degree
of agreement between the things that you rate. So far that would
make sense. If you are all measuring different things it is not
surprising that there is not an extraordinary high degree of agreement,
although it is surprising that there are people in the market
saying that there is an extraordinary high degree of agreement.
Then Mr Taylor told us that the reason there is an extraordinary
high degree of agreement is that the facts are the same, at which
point I am starting to lose it. The facts are the same, so there
is an extraordinary high degree of agreement, but you are all
measuring different things, so it is not surprising that there
is not an extraordinary high degree of agreement. If there is
an extraordinary high degree of agreement, what is the point of
having three of you in a normal market where the product is the
same and the price is roughly the same? What would be the point
of having three providers if you are not in any competition, if
you agree about everything, even though you are measuring different
Mr Madelain: Let me answer the
first point, which is that we are measuring different things.
When you talk about investment rate security, the difference between
measuring expected loss and default probability tends to be very
small. The reason it is very small is because the difference is
made up by the expected recovery, effectively.
Q1010 Mr Simon: What are you adding
to the market by measuring these things differently? What is the
point? Why do you not measure the same thing?
Mr Madelain: Obviously Fitch can
comment on their own practices, but we feel that what is important
for the investor is to actually know the ultimate recovery, pay-out,
effectively, that he can expect from the investment he is making.
Q1011 Mr Simon: So you think that
is the best way. You need to measure the loss, not just the probability
Mr Madelain: That is correct.
Q1012 Mr Simon: And you are Moody's?
Mr Madelain: Yes.
Q1013 Mr Simon: S&P, you think
that is wrong. You think that you only need to measure the probability
Mr Taylor: No, I am Fitch. I do
not think it is wrong actually. A lot of this is nuanced, to be
honest. For an investment grade security the impact of recovery
assessment is very limited, because if you are taking a healthy,
strong investment company and saying, "Let us predict what
it is going to look like as it is about to go down the pan",
there is hardly any purpose for doing that, there is no value
we can add. We actually do this for our ratings at the lower end
of non-investment grade; we build in the assumption of recovery.
So we are actually doing the same thing, but we are saying, as
you start getting down to the much riskier levels of assessment,
we think it therefore adds value to talk about recovery prospects
as the risk becomes greater.
Q1014 Mr Simon: So you are only measuring
very slightly different things?
Mr Taylor: In practice, yes.
Q1015 Mr Simon: The facts are the
same, and you are all talking to the same people and using similar
procedures to establish the facts, so it is not surprising that
you all come up with the same answers all the time, which you
obviously do even though you do not like admitting it. In which
case, going back to the fees that you charge and receive, if I
am a typical participant in this market would I normally be attempting
to have a relationship with all of you: I pay you all and you
all rate me.
Mr Madelain: It depends.
Q1016 Mr Simon: I know that sometimes
you decline to rate. I know that you do not all rate everybody
Mr Madelain: Exactly, yes.
Q1017 Mr Simon: If I want to make
a good impression, would I not want to be rated by all three of
Mr Hancock: There will be a number
of clients who are rated by all three, there will be a number
who are rated by two and there will be some who are just rated
by one. There is different market practice in different countries
and sectors of the market that we operate in.
Q1018 Mr Simon: Roughly how would
that break down? Would two be the most common, would you say?
Mr Hancock: Yes, probably.
Q1019 Mr Simon: Why is that? Speculate
a little from a very informal position as to what I as a punter
gain by being rated by two of you when you almost never disagree
with each other and you are measuring the same things and the
Mr Madelain: The point is we can