MERRILL LYNCH & CO.
Advanced Notice of Proposed Rulemaking on Implementation of the
New Basel Capital Accord
Response to Federal Deposit Insurance Corporation, the Board of
Governors of the Federal Reserve System, the Office of Thrift Supervision and the Office of the
Comptroller of the Currency
Merrill Lynch & Co.
November 3, 2003
Merrill Lynch
4 World Financial Center North Tower
New York, New York 10080
Robert E. Feldman, Executive Secretary
Attention: Comments
Federal
Deposit Insurance Corporation
550 17th Street, N.W.
Washington, D.C. 20429
Ms. Jennifer J. Johnson, Secretary
Board of Governors of the Federal Reserve System
20th Street and
Constitution Avenue, N.W.
Washington, D.C., 20551
ATTN: Docket No. R-1154
Regulation Comments
Chief Counsel's Office
Office of Thrift
Supervision
1700 G Street, N.W.
Washington, D.C. 20552
ATTN: 2003-27
Office of the Comptroller of the Currency
250 E Street, S.W.
Public Information Room, Mailstop 1-5
Washington, D.C. 20219
ATTN:
Docket No. 03-14
Dear Sir / Madam:
Advanced Notice of Proposed Rulemaking on Implementation of the New
Basel Capital Accord
Merrill Lynch & Co., Inc. and its subsidiaries, including Merrill
Lynch Bank USA and Merrill Lynch Bank & Trust Co. (“ML” or “we”) very
much welcome this opportunity to respond to the Federal Deposit
Insurance Corporation (the “FDIC”), the Board of Governors of the
Federal Reserve System (the “Fed”), the Office of Thrift Supervision
(the “OTS”) and the Office of the Comptroller of the Currency (the “OCC”)
(collectively the “Agencies”) on the interagency proposals set out in
the Advanced Notice of Proposed Rulemaking (“ANPR”) published on August
4, 2003.
The proposals in the ANPR will have both direct and indirect
application to ML's operations. The FDIC regulates the major element of
ML's Banking business. The Securities and Exchange Commission (the
“SEC”) has recently announced its intention to adopt the Basel
Committee's proposals as the basis for its consolidated capital adequacy
supervision of Investment Bank Holding Companies and Consolidated
Supervised Entities. Major ML subsidiaries worldwide, particularly in
Europe, will have to follow regulations that will have their basis in
the Basel Committee's proposals.
We would like to state initially that we have found the ANPR document
to be extremely useful. It has greatly improved our understanding of the
Basel Committee's proposals. In particular, the ANPR has provided
helpful guidance on a variety of key aspects, such as the supervisory
expectations during planned implementation in the U.S.
We have fully participated in all the prior consultations by the
Basel Committee for Banking Supervision whose proposals underlie the
ANPR proposals. We have been, and remain, in active dialogue with the
European Commission on their implementation of the Basel Committee's
proposals in the form of the Risk Based Capital Directive (“RBCD”).
In addition, we participated in the collaborative development of the
joint response to the ANPR by the International Swaps and Derivatives
Association (“ISDA”) and The Bond Market Association (“TBMA”), and also
the response by the Securities Industry Association (“SIA”). We fully
endorse the comments in those responses.
Further, we responded to the Basel Committee's CP3 in July 2003. Many
of the comments in our CP3 response are relevant to this response. To
avoid repetition we append our CP3 response to this letter (see Appendix
2). However, we do seek to highlight some significant "Matters of Note"
(see Appendix 1) specifically in respect of the ANPR, which we set out
below:
1 Counterparty Risk in the Trading Book
1.1 Need for a Wider Review of Treatment of Counterparty Risk
The treatment of Counterparty Risk in the Trading Book has not
received the same level of attention and discussion as Credit Risk in
the Banking Book. The present treatment of OTC derivatives is unchanged.
The proposed treatment (as we understand it) for settling / unsettled
transactions has been hastily constructed. The new treatment of repo-style
transactions ignores loss history and diverges from economically similar
transactions (e.g., OTC derivatives). The treatment of credit
derivatives fails to recognize market developments and the risk
mitigating effects of credit derivatives and imposes inappropriate
capital requirements on the market making activities necessary to
underpin this important market.
We view the separate development of different treatments for repo-style
transactions and OTC derivatives as a weakness in the proposed capital
framework. Repo-style transactions are fundamentally forward contracts.
Therefore, they should have the same treatment as OTC derivatives for
loan equivalent purposes. In particular, we strongly support the ISDA
discussion and proposal that the loan equivalent for both OTC
derivatives and repo-style transactions should be determined by the
expected level of Potential Exposure, with the alpha factor as a
multiplier, rather than by the 99-percentile. The Counterparty Risk of
credit derivatives would be most appropriately covered by this approach
rather than through a segregated treatment.
We are encouraged by the Basel Committee's apparent willingness to
review the wider treatment of Counterparty Risk. This review must
include the possibility of introducing a modeling approach for OTC
derivatives and repo-style transactions as noted above. We hope that the
Agencies support the need for this review. However, the timetable for
this review is uncertain and any delay in fully reviewing the
Counterparty Risk treatment will create additional problems. It is not
readily apparent from the ANPR how the Agencies would seek to
incorporate this future Basel work.
1.2 Treatment of OTC Derivatives
We note above that the treatment of OTC derivatives has not been
updated, and that this may be rectified by the development of a modeling
approach. Our analysis suggests that the proposed treatment of OTC
derivatives results in capital requirements that significantly exceed
the actual credit risk created by these trades. We reiterate the need
for substantial revision in this area, as presently there is a distinct
lack of risk sensitivity for OTC derivatives. In particular, retaining
the EAD representation of OTC derivatives from Basel I approach would
result in a very significant distortion of their relative contribution
to the overall risk capital. In addition, we strongly believe that
appropriate revisions to OTC derivatives EAD treatment should include
repo-style transactions and counterparty risk of credit derivatives.
1.3 Systems and Procedures
Our concern as to the lack of focus on Counterparty Risk (as noted
above) is best highlighted by looking at the proposed impact on systems
and procedures in the wholesale market by using our current
understanding of the proposed treatment of settling DVP trades as an
example. We understand that capital would be required against the credit
exposure arising from the date of execution of a trade to the date of
settlement. The procedures undertaken by broker-dealers have developed
over the years and are finely tuned to the nature of the market. Any
(theoretical) credit risk in this market is so insignificant that we do
not monitor it - to do so would entail development of systems and
procedures whose costs far outweigh any risk management benefits.
Monitoring will only normally occur if a trade failed to settle at
settlement date. Therefore, a firm's credit and regulatory systems
currently do not measure credit risk associated with DVP trades prior to
the scheduled settlement date. To do so will impose significant costs
with a minimal, if any, benefit.
Further, under an IRB methodology, firms would have to internally
rate their counterparties and exposures. Given the negligible credit
risk involved, DVP counterparties are not normally rated1. To require
these DVP-only counterparties to be internally rated would involve
significant cost in terms of credit risk analysis. This cost would be
out of proportion with respect to the perceived negligible credit risk
present and our history of broker-dealer transactions.
Additionally, and for the reasons above, we support the argument that
normally settling DVP trades should be excluded from the calculation.
Rather, the focus should be on the inclusion of failed trades subject to
an appropriate grace period.
1.4 Repo VaR Backtesting
Our analysis suggests that the relative risk capital contribution of
repos may be disproportionate to the cost of backtesting implementation.
We would therefore propose that a risk materiality threshold be
introduced to justify such an effort.
2 Operational Risk
2.1 Pillar 2 Treatment
We have long advocated that a Pillar 2 Approach is more appropriate
for the treatment of Operational Risk. Certain operational risk events,
particularly low-frequency high-impact ones, are unsuited to measurement
and evaluation. In this regard, the AMA Approach, as currently proposed,
will involve a fair degree of subjectivity to derive a reasonable risk
capital requirement. In some cases arbitrary decisions will have to be
made in the following areas:
• the choice of operational risk categories for some types of risks;
• the adjustments to internal loss data to reflect the changes in
business mix; and
• the use of external data to complement internal losses
2.2 Confidence Interval
As the responses to Basel's CP3 demonstrate, there is a widespread
concern that the stated confidence interval, 99.9%, is too onerous given
data availability at this stage of development of the operational risk
discipline. We have already discussed this topic in our own CP3
response, attached as Appendix 2. Technically, the 99.9% threshold
raises two significant issues:
1. Estimating a high quantile with reasonable precision requires very
large data samples. In the case of operational risk, the discrete events
that create the tail of the loss distribution are infrequent, even after
taking into account relevant external data. As a result, the estimated
unexpected loss can be far from the true 99.9% quantile, especially when
using fat-tailed distributions.
2. Mandating a very exacting level of confidence provides a
disincentive to include conservatism in the statistical modeling of the
loss distribution. This is because as fatter tails are built in the
models, the 99.9% quantile quickly produces unreasonably large numbers
compared to the 99% quantile adopted for other risks. This disincentive
goes against the spirit of other parts of the ANPR that require that a
degree of conservatism be included in the analytical framework.
We would recommend that the explicit reference to the 99.9%
confidence level be deleted and that the confidence level be selected
according to the internal economic capital model used by the firm.
Emphasis should be put on the necessary conservatism in the modeling to
arrive at a reasonable number when compared to benchmarks (e.g., % of
total capital, number of times largest losses incurred).
3 Implementation and need for Transitional Arrangements
3.1 Parallel Running of Basel I Calculations
As with many others in industry, we remain concerned that the need
for the effective parallel running of Basel I calculations for at least
three years (year prior to implementation and two years after) imposes
unnecessary costs on firms. We consider that the further Quantitative
Impact Study, expected prior to implementation, together with the
results of the parallel implementation period and the functioning of the
Pillar 2 regime should provide the Agencies with sufficient knowledge of
the macro and firm-specific impact of the ANPR proposals.
3.2 Leverage Ratio
We are concerned that the Agencies continue to view the leverage
ratio tripwires contained in existing Prompt Corrective Action
regulations as important components of the regulatory capital framework.
We consider the leverage tripwire to be unnecessary particularly in
light of the greater risk sensitivity of the A-IRB approach and the
Pillar 2 role of ensuring that all risks are covered. We believe that an
asset-to-capital leverage ratio is overly simplistic and does not
reflect the risk profile of assets, hedging strategies, or off-balance
sheet exposure and we do not rely internally on overall leverage ratios
to assess risk-based capital adequacy.
In particular, the Basel Committee has previously stated, “increased
capital should not be viewed as the only option for addressing increased
risks confronting the bank. Other means ... must also be considered.2”
The Agencies have acknowledged that in "some" cases the leverage ratio
would continue to be the most constraining capital requirement. We think
that the leverage ratio may be the most constraining requirement in many
cases, which seems to defeat the whole purpose of the Pillar 1 approach.
Accordingly, in light of the supervisory review provided for in Pillar
2, and as the Agencies indicated they would do, the appropriate response
would be to impose additional capital requirements for individual
institutions on a case-by-case basis.
3.3 Impact on Capital Classifications
The ANPR proposes floors that limit the amount risk-weighted assets
can decline in each of the first two years of stand-alone usage of the
advanced approaches. We have long been proponents of compensating caps
or ceilings to minimize the impact where capital increases (see our
response to CP3 at Appendix 2).
The powers and activities of U.S. banks are in part determined by
their capital classification (“well capitalized”, “adequately
capitalized”, or “undercapitalized”). Should a significant increase in
riskweighted assets result from adoption of the advanced approaches, a
bank's capital classification could change, resulting in a loss of
powers and required regulatory enforcement action. Therefore, we would
suggest that the final rule should provide a bank with flexibility to
generate sufficient capital over an appropriate transitional period
(e.g., two to three years). This would allow a bank to maintain the same
capital classification under the advanced approaches as it had prior to
adoption of the new capital standards without limiting its activities or
becoming subject to regulatory enforcement actions.
3.4 Application to Large Groups
The ANPR states that the Agencies believe all bank and thrift
institutions which are members of a consolidated group that is a core
bank, or an opt-in bank, should calculate and report their riskbased
capital requirements under the advanced approaches. We believe this
approach will greatly increase the costs of compliance and provide
minimal, if any, benefits. Further, as product lines and activities
cross legal entity lines, supervision in many instances has evolved away
from evaluation of individual institutions.
We note that the ANPR acknowledges the challenges facing entities in
providing meaningful and readily available disclosure for individual
institutions in a consolidated group. Therefore, we recommend that
second and lower tier entities comply with the advanced approaches only
if an entity by itself qualifies as a core bank. Otherwise, adoption of
the advanced approaches should be at the option of the institution.
4 Treatment of Goodwill
Recent changes in U.S. GAAP have changed goodwill from an amortizing
asset into a permanent one subject to an impairment test. This change
recognizes that goodwill has inherent economic value and is not by
definition a wasting asset. While we realize such assets are illiquid,
we nevertheless believe that it is unnecessarily penal for them to be
subject to a 100% deduction from Tier 1 capital, especially given that
this severity of treatment could prevent acquisitions which strengthen
the overall financial system. We would encourage the Agencies to further
consider the appropriate capital treatment of goodwill.
5 Concerns about Data Adequacy
In our response to the Basel Committee's CP3, attached as Appendix 2,
we noted that lack of data might give rise to implementation issues. We
were concerned that lack of data might act as a barrier to adoption of
the advanced approaches. In light of this, we support the language in
the ANPR Executive Summary that permits use of external data for
business lines where sufficient internal data does not exist to support
PD / LGD assumptions. This development is welcome, though we would note
that the data issue is not specific to Credit Risk because it affects
Operational Risk as well.
Lack of data may be due to positive factors such as minimal loss
history, either as a result of underlying market fundamentals (such as
for DVP trades discussed earlier) and/or due to robust controls and risk
appetite adopted by an individual firm. New products and markets, a
healthy sign of an innovative environment, similarly would lack data
history.
Firms and markets affected by the circumstances described above would
be penalized for their lack of data by rendering the advanced approaches
as unobtainable. This would be damaging to the aim of creating a risk
sensitive capital framework, as well as to the affected firms and
markets themselves.
6 Pillar 3 Disclosures
The Agencies raise questions seeking responses as to the
appropriateness of the proposed Pillar 3 Disclosure requirements. In
response we wish to note that we support the Principles for Strong
Disclosure Practices set out in the Shipley Report (“Working Group on
Public Disclosure” report, dated January 11, 2001, to the Fed, the OCC
and the SEC). These Principles are founded on the basis that disclosures
should be closely aligned to the practices and policies of the
individual firm. Thereby disclosures will always be pertinent and
relevant. We believe this concept is equally relevant for the Pillar 3
disclosures. In this respect, the disclosures proposed in the ANPR
appear to be generally consistent with the Shipley Principles, though
this will only be properly determined upon full implementation given the
practical slant imbedded in the Principles.
7 Ongoing Dialogue
As we have noted above we remain in active and ongoing dialogue with
many Supervisors and Legislators around the world, including the Fed,
FDIC, SEC and OTS. Similarly, significant aspects of the proposals are
subject to revision, for example Securitization. Therefore, we do not
wish our comments set out in this response to be viewed as final or
definitive. We would hope that the Agencies would continue to accept
relevant and informative commentary on aspects of the ANPR beyond the
end of the consultation period.
We hope that you find our comments helpful. We are very happy to
clarify and discuss any of the matters in this response. Please feel
free to contact us, Steve Teather (+44 (0)207 995 4848 or steve teather@ml.com),
or Nicole Degnan (212 449 5042 or ncole degnan@ml.com).
Yours sincerely,
John J. Fosina
First Vice President
Corporate Controller
Managing Director
Christopher B. Hayward
Managing Director
Corporate Risk Management
Merrill Lynch & Co., Inc.
Merrill Lynch Global Bank Group
cc: Michael A. Macchiaroli, Esq.
Associate Director
Division of Market Regulation
Securities and Exchange Commission
450
Fifth Street, N.W.
Washington, D.C. 20549
Appendix 1
Detailed Comments on ANPR Proposals
Appendix 1: Detailed Comments on ANPR Proposals
We have not sought to answer all the questions posed by the Agencies
in the ANPR. Provided below are answers to those questions that are
relevant and appropriate to ML, together with other detailed comments.
A. Credit Risk
ANPR Question: The Agencies seek comment on appropriate thresholds
for determining whether a portfolio, business line, or geographic
exposure would be material.
We would suggest thresholds be functions of loss materiality by
product, portfolio or business line in combination with thresholds for
individual exposure levels. These measures are consistent with our risk
management approach philosophy today. We may adopt a less resource
intensive approach toward counterparties with exposures below a certain
threshold and where such exposure is highly collateralized, short term
in nature or for products that by virtue of a high degree of regulation
or standardization of market practice, are perceived to be less risky.
We would suggest a total threshold for materiality of 1% of total Risk
Weighted Assets.
Loss Given Default: We would appreciate clarification from the
Agencies that they are not in concept averse to firms using very low LGDs for exposure classes where it can be demonstrated that losses are
rare if non-existent, for example, in the retail margin lending
business. In many instances the LGDs could tend to zero.
ANPR Question: The Agencies invite comment on the merits of the SSC
approach in the United States.
We strongly support offering the SSC approach as an option. This
affords institutions the flexibility to conduct their own internal cost
benefit analyses and take the more simplified approach of mapping
internal risk rating grades to one of the five supervisory grades as
appropriate, without compromising the overall Advanced IRB approach.
ANPR Question: The Agencies are seeking comment on the proposed
definitions of the retail AIRB exposure category and sub-category. Do
the proposed categories provide a reasonable balance between the need
for differential treatment to achieve risk-sensitivity and the desire to
avoid excessive complexity in the retail A-IRB framework? What are the
views on the proposed approach to inclusion of small business exposures
in the other retail category?
We support the inclusion of SMEs in the other retail category,
however we would strongly suggest that the threshold of $1 million in
exposure is too low for both SMEs and individuals. We strongly believe
that the threshold should be raised to $5 million to be more consistent
with the practicalities of our business mix and client base. A material
percentage of our individual and SME exposures are between $1 and $5
million. Given these exposures are well collateralized and exhibit
common attributes, we would prefer to categorize them on a pooled basis,
rather than having to rate each borrower and exposure independently. We
acknowledge that $5 million may not be appropriate for all banks, and so
the Agencies may consider that thresholds should be determined on a
case-by-case basis.
Double default:
The proposal as it now stands provides very little incentive to hedge
exposures and gives little benefit for activity that results in
substantial credit risk mitigation.
We strongly support the ISDA position response to the ANPR on this
topic.
Credit Derivatives:
The formula for maturity mismatch is punitive - hedging of default
risk within the one-year horizon is not affected by maturity mismatch
when protection is for more than a year. The formula should affect only
the hedging benefit with respect to the loss of market value due to
spread widening (credit migration).
Additionally, within the substitution approach, risk mitigation
achieved by buying credit derivative protection from a counterparty that
is rated lower than the reference obligation, but with CSA protection
providing credit enhancement (e.g., is collateralized) is completely
missed.
B. Operational Risk
Independent Testing and Verification Function:
It is unclear whether internal (audit) staff can fulfill this
function.
Quarterly reporting interval:
The regulators prescribe at least a quarterly reporting interval for
risk exposures, loss experience, business environment and internal
control assessments. Given that we do not expect some of these items to
change with such frequency, we suggest that the regulators allow each
institution to report quarterly what is meaningful, decide the reporting
intervals for all other items, and discuss their rationale for choosing
these intervals with the regulators.
Exception reporting:
We would appreciate clarification as to the meaning of this section,
which is unclear at present.
Demonstrate appropriate internal loss event data, relevant external
loss event data, assessments of business environment and internal
controls, and results from scenario analyses:
We suggest that the Agencies provide additional guidance to clarify
the relative emphasis to be put on the use of these elements in the
management of operational risk (e.g., risk assessments, resource
planning and risk mitigation initiatives) or for modeling purposes. If
the latter should predominate then it should be explicitly stated that
firms have the flexibility to use only certain elements (or combinations
of elements) over others. For example, businesses that have enough
internal loss data points to build a conservative loss distribution
should be allowed to use only this internal dataset.
Event end date:
Given that the date of the loss and the discovery date of the loss
are both suggested, it seems to be unnecessarily burdensome to track yet
a third date for every data point. The regulators should highlight
whether or not this is a mandatory item. If not, to avoid similar
confusion by other firms, it might be best to delete it from the
“suggested” list.
System to identify and assess business environment and internal
control factors:
The agencies should clarify what the key elements of such a system
should be. In particular can an institution have different assessment
systems in different businesses?
Supervisory standards for removing the Expected Loss Offset in the
capital computation:
The guidance should specify what these standards are so that firms
expecting to qualify for the offset in their data can meet these
standards.
Independent verification of the analytical framework:
We seek clarification that internal audit or another independent,
competent area within the firm can perform this independent
verification.
Expected Loss offsets and thresholds:
It is unclear why firms with higher thresholds should be penalized
with a smaller expected loss offset so long as they demonstrate diligent
efforts to capture the tail or "unexpected loss" exposures, particularly
if they can show that the threshold is sufficient.
Impact of risk mitigants:
It is not clear why a 20% limit on the impact of risk mitigants was
chosen. The recognition of risk mitigation in the operational risk
calculations should be left to the institution subject to strict
criteria and principles. If a ceiling were maintained then we would ask
for clearer guidance on acceptable ways to apply this ceiling in the
calculations (e.g., historical vs. prospective basis, loss data offset
vs. offset to final capital results).
Future risk mitigation products:
We suggest that the Agencies should provide guidance as to how future
risk products introduced by either the insurance industry or capital
markets can get an expedient joint review and approvals from all four
agencies.
Public reports on operational risk measurement and management
results:
Given the sensitive nature of operational risk event information, we
would caution against any suggestion that the information should be made
available beyond internal reports and regulatory discussions.
AMA computation interval:
We suggest that the computations for determining operational risk
capital is updated less frequently than those for either market or
credit risk capital. Since a firm's operational risk profile is largely
dependent on its senior management, governance processes, and
risk-minded culture, we would anticipate that such factors would change
much more slowly than the market fluctuations, market positions, and
credit exposures that drive the market and credit risk capital
computations. If the regulators wish to have regulatory capital
requirements timed with the management cycle for determining internal
economic capital requirements, then we suggest that this be done
annually. Regulators, of course, will retain the option of intra-year
updates whenever it appears that a firm may have abruptly altered its
profile (e.g., through a strategic restructuring or significant
insurance purchase).
Appendix 2
ML Response to the Basel Committee's CP3
Third Consultative Document on The New Basel Capital Accord
Response to the Basel Committee on Banking Supervision
Merrill Lynch & Co.
July 2003
30 July 2003
Basel Committee Secretariat
Basel Committee on Banking Supervision
Bank for International
Settlements
CH-4002
Basel
Switzerland
Dear Sirs:
Response to the Basel Committee on Banking Supervision on Proposals
for the New Capital Accord
Merrill Lynch & Co., Inc. (“ML” or “we”) very much welcomes this
opportunity to respond to the Basel Committee on Banking Supervision
(“the Basel Committee”) on its proposals set out in the third
consultative paper on the New Basel Capital Accord (“the Accord” or
“CP3”) published on 29 April 2003.
We have fully participated in the Committee's prior consultations and
other related work since deliberations began back in 1999. We recognize
and strongly welcome the efforts of the Committee to seek to address the
many and varied concerns held by industry. This is reflected in the
current proposals, which represent a considerable evolution from earlier
versions. However, we do have some significant "Matters of Note" which
we set out in the accompanying pages to this letter. In particular we
have the following significant concerns with the proposals:
Treatment of Counterparty Risk in the Trading Book
• We believe that this is an area that still requires considerable
thought and discussion in order to deliver risk sensitive and
proportionate proposals. To this effect we fully endorse the
Counterparty Risk comments in the International Swaps and Derivatives
Association (“ISDA”)/ The Bond Markets Association (“TBMA”) joint
response (summarized in the main body of our response).
• We welcome the promise by the Basel Committee to work with industry
to update the treatment of OTC Derivatives. This review must also
include Securities Financing Transactions and should commence without
delay.
Operational Risk
• We continue to believe that Operational Risk is more suited to
treatment under Pillar 2, especially for low-frequency high-impact events
where measurement and evaluation is subjective.
• We continue to have strong reservations about the proposals for the
Standardized Approach. These appear to be based on the view that from an
Operational Risk perspective the Trading Book is riskier than the
Banking Book. We find this presumption troubling and can find no
justification for it in the proposals.
We have fully participated in, and accordingly generally endorse the
responses of trade associations of which we are a member firm. In
addition to ISDA and TBMA mentioned above, these are the London
Investment Banking Association, the British Bankers Association and the
Securities Industry Association.
We hope that you consider our comments helpful. We are very happy to
clarify and discuss any matters in this response. Please feel free to
call or e-mail Steve Teather (+44-20-7867-4848 or steve teather(a)ml.com).
Yours sincerely,
John Fosina
Corporate Controller
Merrill Lynch & Co., Inc.
David Brooks Gendron
First Vice President, Chief Financial Officer
Merrill Lynch Europe, Middle East and Africa
Matters of Note
We appreciate that the Committee wishes to complete the Accord by the
end of 2003. We therefore restrict our comments to those matters of
particular concern and importance to ML. We believe that many industry
members share these concerns. ML is not a so-called "Basel Bank" and not
a direct constituent of the group for which the Accord is prepared.
However, the ML Group does contain subsidiaries that will be subject to
the Accord and so are impacted by the Committee's proposals. We hope our
comments will be considered.
Credit Risk
The Committee has naturally concentrated on the Banking Book during
the development of the Accord. Consequently, Trading Book matters have
received less attention; though we welcome the increased attention that
has been paid recently. However, a number of matters remain to be
addressed.
ISDA / TBMA Response - Counterparty Risk Comments
We do not wish to repeat here the well-articulated comments in the
ISDA / TBMA response which we fully endorse. ISDA / TBMA raise matters
that are particularly important to ensure that the Accord does not have
unintended consequences or disproportionate impact on the Trading Book.
The ISDA / TBMA comments are summarized below for convenience:
• Capital Treatment of Credit Derivatives
• Treatment of restructuring risk in credit default swaps: welcomes
the recognition of restructuring where it is in the control of the
guarantor and seeks to provide some recognition where such control is
not demonstrated to exist.
• Credit default swap add-ons: queries the application of the add-on
to sellers of credit risk and notes that the add-on for qualifying items
is too large.
• Substitution / double default risk: concern that there is no
recognition for the smaller probability of both counterparty and
guarantor defaulting under the substitution methodology.
• Specific risk off-sets: arbitrary percentage off-set prescribed is
not risk sensitive and suggests that credit risk positions should be
represented as Floating Rate Notes to allow appropriate offset.
• Operational requirements applied to CDSs: seeks to ensure that the
proposals will allow the use of Master Netting Agreements.
• Counterparty Risk
• Use of VaR for repo-style transactions: concerns expressed over a
potentially onerous and penal back-testing regime.
• Treatment of potential exposure: current add-on methodology is too
crude and does not meet the same risk sensitivity standards of the rest
of the proposed Accord. A review and update has been promised but this
must be timely.
• Maturity
• Maturity adjustment below one year: seeks to ensure that IRB
methodology is appropriate for shorter maturities.
• Effective maturity adjustment for repo and derivatives: suggests a
standard maturity of one year for OTC derivative trades and six months
for repo transactions.
• Treatment of maturity mismatches: queries why a standardized linear
scaling factor approach is used.
It is essential these matters that relate largely to Trading Book
exposures be satisfactorily addressed.
QIS 3
The Committee's QIS3 results showed that the Trading Book charge is
expected to increase significantly, even under the most advanced
methodology3. This accords with our own observations. We are concerned
that the Committee dismissed the Trading Book impact as immaterial.
However, this is simply a function of the relative size of the Trading
Book in the QIS3 sample. This sample did not include firms with
predominant Trading Books, such as ML. The importance of the Trading
Book business means that the impact cannot be ignored.
Treatment of Settling Transactions
The Committee's proposed treatment of settling transactions has only
recently become clear. Paragraph 292, which ostensibly relates to
maturity adjustment, has been interpreted as meaning that the 4-day
settlement grace period will no longer apply. If correct we find this
very troubling.<
The grace period recognizes the administrative nature of resolving
settlement errors (i.e. settlement risk). Genuinely failed trades
uncleared within 4 days appropriately attract a Credit Risk charge.
Settlement Risk is covered under the Accord's proposals for Operational
Risk and removal of the grace period would represent a double capital
hit.
We recognize that conceptually credit risk is present in trades such
as securities sales and purchases settling in less than 4 days. However,
the cost of maintaining systems and processes to enable counterparty
risk management of this high volume settlement activity would be
significant. Such costs would far outweigh any risk management benefits,
particularly given there is little or no historical data evidencing
credit losses associated with this activity.
We question whether the Committee has assessed the likely market
impact of this proposal. The volume and value of unsettled trades in the
market would require a very significant additional amount of capital to
be set aside. We do not believe that this extra capital is justified for
the minimal credit risk present. The Committee must also be mindful that
these proposals will be applied to smaller institutions, and we would
caution that the impact on the retail broking industry must be assessed.
It is likely that most firms would have to make significant changes to
their regulatory systems to capture these unsettled trades.
For clarity we ask the Committee set out its intentions in this area
and particularly to define the scope of "settling transactions".
We note that this matter is also specifically addressed in the ISDA /
TBMA response which we support.
Application of IRB in an Innovative Environment
The Committee is right to bear in mind the impact on innovation of
its proposals. We are concerned as to how new products will be catered
for in the IRB methodology. Innovation is fundamental to our business
and it is essential that new products can be accommodated within IRB, as
otherwise they might not be economically viable. New products by
definition do not have a credit and default history and so may be
excluded from the IRB by reason of this.
We would appreciate specific clarification within the Accord that new
products are not by their nature excluded from the IRB.
Securitization
The prescribed risk weightings for securitization exposure are
calibrated to reflect the risks of corporate bond and loan portfolios.
Such an approach results in excessive risk weights compared to the
economic risks of securitization tranches, particularly for retail and
mortgage portfolios.
We believe that sophisticated banks should be allowed to use models
to determine risk capital allocation for securitization exposures and
expand the use of the Supervisory Formula Approach (“SFA”). Furthermore,
we believe the SFA should be modified to allow application of different
betas depending on the securitization exposures in question.
It is necessary to calibrate the Rating Based Approach better to
reflect the diversity in securitization. We believe that a set of
different risk weights corresponding to different securitized exposures
(e.g. consumer loans, mortgage loans, commercial real estate), as well
as corporate exposures, should be developed.
Operational risk
Recent work has tended to concentrate on the Advanced Measurement
Approach ("AMA"). We believe that certain operational risk events,
particularly low-frequency high-impact ones, are unsuited to measurement
and evaluation. To this effect we continue to believe that they should
be assessed under Pillar 2. We have significant concerns with aspects of
the less risk-sensitive approaches, particularly the Standardized
Approach, and these are set out below.
Calibration of Betas
The set betas give rise to perverse incentives. The Trading Book
business lines (Trading and Sales, and Corporate Finance) attract a beta
of 18%. There is no incentive for firms with significant Trading Book
business to seek to progress from the Basic Indicator Approach to the
Standardized Approach. We find it odd that the Committee should
crystallize such an incentive. The relatively high standards for the
Standardized Approach means it is essential that no beta is higher than
the alpha.
We do not view Trading Book business as more risky from an
operational risk perspective. Indeed we would view Banking Book
activities as generally more risky, if only because we are less active
in these areas. The Committee's own research4 does not suggest that the
Trading Book business lines are inherently more risky.
We note the Committee has recognized the problem of double counting
for certain high-margin banking book business lines. We note that in
general trading book business is also high-margin though largely due to
market risk rather than credit risk. The double counting that the
Committee recognizes as a problem is compounded by artificially high
betas. Failure to lower these betas would result in a triple impact:
high revenue resulting in high operational risk charge, high market risk
charges and a penal beta.
Taking all these matters together we find there is compelling reason
for the Committee to lower the 18% betas to less that 15%. Our
overriding concern is that the apparent riskier nature of the Trading
Book will become institutionalized and that Trading Book Operational
Risk charge will always be benchmarked against 18% even if AMA is
adopted. This is not justified, and is contrary to the level-playing
field concept.
As an alternative we suggest that the betas could be further
differentiated by way of "core" versus "non-core" business. Core
business would attract a lower beta to recognize the fact that firms
would have substantial controls, experience and well-established
governance practices in place as required by the qualitative standards.
Operational Risk Boundary with Market Risk
There is overlap between Market Risk and Operational Risk similar to
that identified between Credit Risk and Operational Risk. Where a
pricing loss derives from an operational failure it is unclear whether
it should be treated as either Market Risk or Operational Risk. Clarity
is needed to avoid double counting and overstating capital. It would not
be practical to isolate market risk losses deriving from operational
failure within the market-to-market process. The mark-to-market approach
means that losses are treated as a 100% capital charge anyway.
We would appreciate clarification of the treatment of the boundary
between Market Risk and Operational Risk.
Reporting Interval
We suggest that the computations for determining Operational Risk
capital be updated less frequently than those for Market or Credit Risk.
Since a firm's Operational Risk profile is largely dependent on its
senior management, governance processes, and controls, we would
anticipate that such factors would change much more slowly than market
fluctuations, market positions, and credit exposures that drive the
Market and Credit Risk capital computations. We suggest that the
calculation be done annually. Regulators should retain the option of
intra-year updates whenever it appears that a firm may have
significantly altered its risk profile (e.g., through a strategic
restructuring or acquisition).
99.9% Confidence Interval Requirement
We believe that it is premature to require a specific 99.9%
confidence interval. This may turn out to be an unfair or unattainable
standard. At this stage in the evolution of operational risk
methodologies regulators should require firms to justify the confidence
intervals used in their models.
Pillar 2
Stress Testing
We accept the importance and value of stress testing in determining
economic capital. It is an essential component of risk management and
therefore a valid requirement in the Accord.
However, if the Committee is unwilling to recognize a firm's Credit
Risk model for regulatory capital purposes then it is inappropriate to
require stress testing in Pillar 1. As such paragraphs 396 to 399 are
not appropriate to Pillar 1 and should be moved to Pillar 2 to
complement paragraph 684.
General
Transitional Arrangements
We note the Committee's aim to maintain the level of capital across
the banking industry. This means that on average the expected decrease
in Credit Risk capital is matched by the new Operational Risk charge.
The Committee seeks to monitor this by requiring firms to continue to
perform the Basel 1 computation in parallel. Additionally the Committee
seeks to limit any beneficial impact by maintaining a floor for two
years after implementation.
We understand the need for these arrangements, though we should point
out that this approach is unbalanced. There are many firms that will see
their regulatory capital significantly increase. If the imbalance
remains the Committee will find it difficult to keep to its aim of
maintaining the level of capital in the banking system. Limiting the
benefits without similarly limiting the costs will lead to distortions
and increase the level of capital. It is therefore essential that as
part of a prudent approach to implementation some form of capital cap or
ceiling be imposed so as to allow a smooth transition. Failure to do so
will introduce unwarranted competitive distortions.
Scope
We do not believe that consolidation at every node in a group
("sub-consolidation") is appropriate. This will be an expensive
imposition for groups, particularly large groups. We believe the
marginal benefit to supervisors will not outweigh the cost to firms.
Use of Data
Data scarcity is an issue, and is more acute in some areas, e.g. the
Trading Book. As a result we are keen that the Committee allows group
data to be used at the individual legal entity level. Such data will of
necessity span geographical and regulatory boundaries, but given it
would derive from a similar system and control environment we would
consider it more relevant than third party external data.
Data collection and use are key element of the Credit and Operational
Risk methodologies. It is essential that the Committee state clearly its
requirements in respect of data collection and use. Given the ambiguity
of the term "bank" in the Accord5 it is unclear whether data should be
collected and used at the legal entity or consolidated basis.
We would appreciate clarification that firms could use wider group
data at the legal entity level.
Home / Host Issue
As a global organization with entities in multiple jurisdictions we
are subject to the regulatory requirements of many different
supervisors. We therefore wish to add our support to the many industry
comments that seek a sensible and pragmatic solution to the issue of
lead supervision and regulatory approval across multiple jurisdictions.
This is particularly acute for ML, in comparison to many other financial
groups, in that we do not have single dominating bank to which the CP3
proposals will apply.
1 A rating may exist where that counterparty transacts other non-DVP
products with the firm.
2 Par. 590 in the Basel Committee's January 2001 CP2 document.
3 Per "QIS 3 -Overview of Global Results" published by the Basel
Committee in May 2003: % change in capital requirement for the Trading
Book under (1) Standardized Approach = +12%, (2) Foundation IRB = +4%, and
(3) Advanced IRB = +2%.
4 2002 Loss Data Collection Exercise for Operational Risk published
by the Basel Committee on 14 March 2003.
5 For example, in paragraph 640 there are references to the singular
"bank" that appears to be contradicted by the reference in the forth
bullet that "the bank... roll out the AMA across all material legal
entities and business lines ". It is not clear therefore whether partial
AMA use is at the legal entity or consolidated level.
