|
April 2001
Introduction
The Basel Committee on Banking Supervision's Quantitative Impact
Study questionnaire is attached. The objective of the impact study is
to assess whether the Committee has met its goals with regard to the New
Basel Capital Accord. The impact study will gather the data necessary
to allow the Committee to gauge the impact of the current proposals
across a wide range of banks in the G10 and beyond, given the differing
risk profiles of banks and the extent to which credit risk mitigation is
used.
Banks are asked to provide the data to their national supervisor
by 1 June, in an electronic format. The Committee recognises
that the deadline for the questionnaire is short. However, this is
essential if the New Accord is to be agreed this year as planned and it
is hoped that banks' close co-operation with their supervisors in
filling in the questionnaire will expedite the exercise. Supervisors
will maintain an ongoing dialogue with participating banks throughout
this process. They will also be asked to review banks' responses and
discuss the data with banks on an ongoing basis to ensure that any
difficulties are resolved prior to the 1 June deadline. This should
help the Committee to have a thorough understanding of any limitations
of data provided.
It is also recognised that banks will not have exact data to answer
some of the questions or might not be able to calculate the
requirements. Estimates, therefore, are acceptable as long as they are
well founded and agreed with national supervisors.
The Committee appreciates that the exercise is lengthy and very
detailed and recognises that it will be a burden on participating
institutions. However, we believe that it is in the interests of the
industry that the Committee's proposals have a strong quantitative
basis. The requested information is critical to an accurate assessment
of the overall impact of the revised Accord and to the calibration of
its components (so reducing the need for any buffer against
uncertainty). Your co-operation and participation is, therefore, of
considerable significance and your supervisor will be happy to discuss
the exercise with you should this be necessary or to resolve any queries
that you might have.
All data supplied will be treated as confidential. The Committee
does wish, however, to provide some feedback for the industry on the
overall results. This will be done on a basis that avoids any
disclosure of individual bank balance sheet data.
Questionnaire
The Committee is asking that the capital requirements for credit risk
be calculated using both the standardised and IRB foundation approaches
for a convenient date either in 2000 or the first quarter of 2001. The
questionnaire is divided into three sections:
- Chapter 1 (Parts I through VIII) requests general information on
the nature of the bank (size, current capital, credit rating etc.). In
addition, it asks for the capital requirements for credit risk under the
standardised and IRB foundation approaches to be calculated for a bank's
whole portfolio - at least 80% of group-wide exposures should be
included - as well as the impact of credit risk mitigation. All banks
are asked to calculate capital requirements for credit risk under the
standardised and IRB approaches if they can. This section also asks for
operational risk data and consists of questions pertaining to the Basic
Indicator and the Standardised approaches to operational risk.
- The second section requests details of the standardised approach
for credit risk. Those banks only able to provide information on the
standardised approach should complete this section.
- The third section requests details of the IRB approach This
should be completed by banks which expect to request permission to apply
the IRB approach under the New Accord and which have information on
loans by grades, and can estimate the probability of default associated
with such grades.
The questionnaire is provided in Microsoft Excel format. As the
objective of this study rests on being able to analyse cross-sections of
standard data, the worksheet protection feature has been enabled. You
are asked to provide data precisely in the form in which it is asked.
If banks have further comments they wish to make they should provide
these on additional sheets or should discuss the matter further with
their national supervisor.
Capital calculations
It is essential that, when performing these calculations, the bank
compare the proposed and current capital requirements using the same
portfolio. Banks which can calculate capital requirements for a wider
selection of exposures under the standardised than the IRB approach
should, as a comparator, calculate the standardised approach only for
the narrower book used for the IRB. This is to ensure that the
percentage changes in capital required by the different methods
(standardised requirements versus current requirements, IRB foundation
requirements versus current requirements etc.) are meaningful and
comparable. Similarly, for banks calculating the requirements under the
advanced approach for only part of the portfolio, the IRB and
standardised approach should also be calculated only for this part of
the portfolio. It would be helpful if all banks with own loss given
default (LGD) information could try to calculate the advanced approach
on at least part of their book, even if maturity information is not av
ailable and therefore that dimension cannot be included.
Scope
Large international banks are being asked to ensure that at least 80%
(by value) of their group-wide exposures are included in the data for
capital calculations. This is necessary to ensure that the data
supplied is broadly representative of the banks' portfolios and to give
an accurate impression of the impact. At the level of operating
entities some banks concentrate particular types of exposure in one
entity, making that entity unrepresentative of the whole group. It is
understood that it will not be possible to include the exposures in
every entity within a group, but the main operating entities should be
included. Any intra-group transactions should be excluded when
calculating the group-wide figures. The Committee appreciates that
it may also be necessary to estimate some elements of the data necessary
to calculate the capital requirements for the IRB approach. This is
satisfactory as long as the estimates are representative of the wider
picture in the group.
Given the wide variety of results likely, the Committee also needs to
consider the typical profiles of book across the G10 and the impact that
the Accord has on banks with these types of portfolio. This is why the
Committee is asking for information on the quality distributions of the
credit books and also the split between corporate, interbank, sovereign
and retail exposures and the split between commitments and drawn
facilities. All this information will be essential to assess the New
Accord's impact.
Internal Ratings
In terms of the IRB, one of the main determinants of the overall
effect is the split in the corporate, sovereign and interbank exposures
between three broad risk categories of internal loan grades as defined
by their associated probabilities of default:
Therefore, at a minimum the Committee requires information on this
broad quality distribution. The Committee is asking also for a
breakdown by the finer loan gradings and default probabilities used by
the bank, which is an important input into the exercise. It is
recognised however, that it may be possible to do this for only certain
portfolios and that the most extensive allocation might simply be into
the three broad bands above. Of course, even if the detailed quality
distributions are not available at an aggregate level, banks should use
whatever finer distributions they have in calculating capital
requirements under the IRB approaches.
Another area of interest to the Committee is the extent to which the
distribution of exposures across rating bands changes over time, in
particular across the cycle. If possible, some indication of how the
distribution has changed would be very helpful. It will also be
necessary to know whether internal grades, when set, are "point in
time" (i.e. assuming that the financials for the borrower remain as
they are currently) or whether they are "through the cycle"
(i.e. taking into account possible stresses in different economic
conditions). If a bank uses an approach that is rather different from
either of these, they should point this out to their supervisor.
Default
The definition of default used is clearly important here and, if
possible, banks should use the definition set out in paragraph 146 of
the supporting document for the IRB approach (i.e. banks should monitor
all four of the different criteria and recognise default as soon as one
of the four is fulfilled). The Committee recognises that some banks may
not be able to do this at this juncture and therefore we are asking that
banks state the definition or definitions used for different portfolios.
Again, supervisors will be speaking with banks to gain an understanding
of how the quality distributions might have differed if the definition
of default would have been different.
Credit Risk Mitigation
The Committee is also requesting information on the extent to which
exposures are collateralised and the consequent loss given default.
Collateralisation is also an important determinant of the final capital
figure. It is essential to try to take the overall effect of
collateralisation into account in the calculation of capital, which is
required under the standardised and IRB approaches. Likewise,
information on a bank's estimates of loss given default for such
collateral would be helpful.
Even if it is difficult to obtain precise data on the extent to which
exposures are backed by the new forms of collateral/credit risk
mitigation which will be allowable under the New Accord, it would be
helpful to have estimates of this and also of the extent to which other
forms of collateral have been taken. One possible method of generating
these estimates might be for banks to conduct a random sample of their
loan documentation to look at the extent of collateralisation by
collateral type and facility. This could be done for both financial and
physical collateral. Banks should discuss with their supervisors how
this might be undertaken to produce reliable data for the Committee.
However, if a bank is unable either to estimate or provide accurate
information, it should leave these tables uncompleted.
Operational Risk
The Committee has outlined proposals for the development of a capital
charge to cover operational risks. It has set out 3 approaches to
operational risk of increasing sophistication - the Basic Indicator
Approach, the Standardised Approach and the Internal Measurement
Approach. Information is requested to allow an estimation of the impact
of the charge using the Basic Indicator Approach and on indicators to
inform the further development of the Standardised and Internal
Measurement Approaches. This should be submitted along with the rest of
the QIS information by 1 June 2001. A further data request for
operational loss data will be made to obtain information to allow an
accurate calibration of all three techniques for assessing the
operational risk charge; the deadline for submission of this data is
likely to be during August.
Definitions
More generally, there are areas where banks and supervisors will need
to consider together the definitions that are used for business lines
(e.g. retail, project-finance etc). Wherever possible, banks should use
definitions that are given in the consultation paper - part 1 of the
questionnaire lists the most important of these. However, the Committee
recognises that current systems may mean that, in the time available,
banks will have to meet the definitions on a "best efforts"
basis. In general, banks should classify exposures (e.g. categories
such as project finance or retail) according to the risk characteristics
of a type of business - a case of "knowing it when you see
it". For example, for retail, it is important to capture certain
characteristics (large number of low value exposures, with a lower
unexpected loss relative to expected loss etc).
|