November 12, 2003
To: Addressees
Re: Draft Supervisory Guidance on Internal Ratings-Based Systems for
Corporate Credit
Wells Fargo & Company appreciates the opportunity to participate in
the ongoing dialogue on the Basel capital reform proposal. We are a
diversified financial services company, providing banking, insurance,
investments, mortgage and consumer finance from more than 5,600 stores,
as well as through the Internet and other distribution channels across
North America. As such, we have a keen interest in the framing of the
Basel Accord and hope that the comments that we offer in this paper will
be of assistance in providing solutions to the issues that exist in the
current proposal.
Sincerely,
Howard I. Atkins
Executive Vice President and Chief Financial Officer
Addressees:
Docket No. R
Ms. Jennifer J. Johnson, Secretary
Board of Governors of the Federal Reserve System
20th Street and
Constitution Avenue, NW
Washington, DC 20551
Docket No. 03
Communications Division
Third Floor
Office of the Comptroller of the Currency
250 E Street, SW
Washington, DC 20219
Mr. Robert E. Feldman, Executive Secretary
Attention: Comments
Federal Deposit Insurance Corporation
550 17th Street, NW
Washington,
DC 20429
Wells Fargo & Company appreciates the opportunity to participate in
the ongoing dialogue on the Basel capital reform proposal. While we
respect the tremendous amount of time and effort that has gone in to
shaping the proposal, we find that we still have some fundamental
differences of opinion with the path on which the Basel Committee and
the U.S. banking supervisors are proceeding and feel that certain
aspects of the proposal must be changed in order for it to be
acceptable.
We will direct our comments here to the Draft Supervisory Guidance on
Internal Ratings-Based Systems for Corporate Credit dated July 1, 2003.
We have drafted separate comment letters for the related Advanced Notice
of Proposed Rulemaking ("ANPR") and the Draft Supervisory Guidance on
Operational Risk Advanced Measurement Approaches for Regulatory Capital,
although we may allude to some of that commentary in the course of this
dialogue.
There has been a relatively uniform set of concerns communicated to
the Basel Committee in response to Consultative Paper 3 (CP3) on the
topic of prescriptiveness. However, we fear that these criticisms have
been too general in nature to be of much value as an agent of change. In
fact, the Committee maybe receiving mixed signals from the industry in
terms of its requests to have more rigidity built into the Accord on
some issues and less rigidity on others.
The areas where we feel that clarity is required relate primarily to
definitional issues within the Accord - a common definition of default
or future margin income, long-run average versus point in time PD or LGD
estimates, and similar metrics or terms that are necessary to create an
unambiguous foundation upon which the new, more risk-sensitive,
regulatory capital calculations can be computed.
Where clarity is not required, and where the Supervisory Guidance
steps over the line and into the realm of unwarranted prescriptiveness,
comes from its attempts to dictate how banks actually manage risk. The
Supervisory Guidance is too prescriptive and inflexible in its vision of
the risk management processes to which banks must adhere.
This is in stark contrast to the original supposition of Basel II --
that each bank would be allowed to continue the use of its existing risk
management practices, so long as they could be shown to have been
effective over time. The Accord and Supervisory Guidance should only
aspire to establish a more risk-sensitive framework for constructing
minimum bank regulatory capital requirements. They cannot, and should
not, attempt to dictate how banks actually manage risk. For those
institutions, such as Wells Fargo, with proven risk management processes
in place, it would be imprudent, and perhaps dangerous, for them to make
significant changes to their risk management systems in the absence of
quantifiable and validated data that clearly demonstrates that an
alternate system is more robust and accurate, and could be successfully
inculcated into their risk management process. And, even if an alternate
system were deemed to be superior, to attempt such changes on an
accelerated timetable would be risky where credit portfolios of
significant size and scope were at stake.
Do the Basel Committee and the U.S. Banking Supervisors really intend
to force the migration of well-functioning, customized risk management
processes into an untested, complex framework with the potential to
actually confuse, or undermine, the control and understanding that banks
currently have of their credit portfolios?
The primary points that Wells Fargo will emphasize, and where we feel
that we must be successful in helping the Basel Committee and the U.S.
regulatory authorities implement a more appropriate regulatory capital
regime, are in those areas where we believe that the Accord has ventured
beyond its intended scope. We have organized our comments into the same
chapters presented in the Supervisory Guidance.
Ratings for IRB Systems
1) Definition of Default -- We believe that the definition of default
outlined in Paragraphs 29-32 of the Supervisory Guidance should be
simplified to correspond more closely to what is more commonly used by
risk practitioners. That is, loans that fall under the corporate and
specialized lending models should define default to coincide solely with
the incidence of non-accrual or charge-off status (to exclude the 90
days past due and other isolated conditions present in the Accord's
current definition), and loans that fall under the retail model should
define default to coincide with the Uniform Retail Credit Classification
standards published by the FFIEC.
With respect to retail lending, the ANPR presents an updated point of
view from the U.S. banking supervisors that the FFIEC definitions of
loss recognition for retail credit will prevail. However, the ANPR goes
on to state that retail default will also include the occurrence of any
of the following events: 1) full or partial charge-off, 2) a distressed
restructuring or workout involving forbearance and loan modification; or
3) notification that the obligor has sought or been placed in
bankruptcy. We believe that the retail charge-off and bankruptcy
conditions are addressed in the FFIEC guidelines, and, as such, would be
appropriately triggered as defaults by those procedures. However, the
distressed restructuring criterion is outside of the scope of FFIEC and
should be excluded from the Basel definition of default.
Our comments here will address primarily the application of the
default definition to corporate and specialized lending portfolios. We
are concerned that, in the absence of moving the Basel default
definition for wholesale loans to be based solely on the occurrence of
non-accrual or charge-off status, banks will be forced to track two
separate measures of default - one for internal risk assessment and a
second for regulatory capital purposes. This would seem to be a
meaningless, yet costly, exercise, since the ultimate driver of risk is
loss, and these fine lines of default definition will only serve to
shift the mix of PD and LGD in an offsetting fashion, without
significantly affecting ultimate loss.
Non-Accrual status already subsumes the more detailed definitions of
default. Generally, an asset is placed on non-accrual when it is 90 days
past due or when reasonable doubt exists about a loan's collectibility.
And, a declaration of bankruptcy would almost certainly trigger the
condition of reasonable doubt regarding collectibility.
An exception to these general rules occurs when a loan is well
secured and in the process of collection, in which case it will not
necessarily be placed on non-accrual status. However, this exception
only applies in limited situations. To be well secured, the asset must
be secured by lien or pledge of collateral with realizable value
sufficient to fully meet the obligation or guaranteed by a financially
responsible party. An asset is in the process of collection if the
collection through legal or other means is in due course. Generally, an
asset can only remain that status for 30 days unless it can be
demonstrated that the amount and timing of the payment is sufficient and
reasonably certain.
There are already internal controls, internal audits, external audits
and supervisory processes to ensure that non-accrual and charge-off
policies are applied correctly. These policies, which govern whether
banks continue to recognize income on their financial statements, should
be sufficient to satisfy the Basel definition of default. The broader IRB definition of default, which includes bankruptcy, selling at a loss,
distressed restructuring (either wholesale or retail), and 90 days past
due, is likely to arrive at virtually the same overall conclusion
regarding the frequency of defaults, once consideration is given to
materiality and purely technical defaults are excluded.
The U.S. banking supervisors seem overly concerned regarding the
potential for "silent defaults;" that is, instances where the well
secured and in the process of collection exceptions to non-accrual
policies are triggered. Capturing this data is a meaningless exercise
for two reasons. First, these are exceptions precisely because there is
a strong expectation of zero loss. And, second, as we previously stated,
the net result of tagging such events as defaults would be negligible,
since increased PD estimates would be offset by lower LGD estimates.
The same thought process around silent defaults also seems to have
driven the additional criterion to include loan sales at material credit
related discounts as defaulted assets. We oppose this criterion on both
practical and conceptual grounds. Loan sales are a part of the portfolio
management function. Portfolio management strategies differ
significantly across banks, with some institutions being much more
active than others. Even within a single institution, loan sale
strategies will vary across time depending on overall balance sheet
management and liquidity issues. Clearly, including performing loan
sales in the definition of default would introduce comparability
problems. Further, discounts on loan sales can be due to a variety of
factors unrelated to credit such as interest rates, liquidity or
technical supply and demand issues. It would be quite difficult, and
ultimately arbitrary, to disentangle these effects.
Finally, on a more fundamental level, the loss in a loan's value due
to credit deterioration is migration risk and not default risk.
Migration risk is already included in the framework through the maturity
adjustment portion of the IRB formula. To be consistent with the
derivation of the formula, the default probability that is estimated
should not be artificially inflated for downgrades, and then only for
those that are "realized" through discretionary loan sales. Such
regulation could create perverse incentives for bank credit portfolio
management and actually add to risk in the portfolio.
2) Obligor Ratings -- Paragraph 35 of the Supervisory Guidance states
that separate exposures to the same obligor must have the same obligor
rating grade. This contradicts paragraph 359 of CP3, which states that
there are two exceptions to this rule. Firstly, in the case of country
transfer risk, where a bank may assign different borrower grades
depending on whether the facility is denominated in foreign or local
currency. And, secondly, when the treatment of associated guarantees to
a facility may be reflected in an adjusted borrower grade. There is a
third instance not explicitly mentioned in CP3 (but which we feel is
equally applicable), where certain types of specialized lending may be
symptomatic of instances where fluctuations in collateral value
influence not only the LGD of a facility, but its PD as well.
We think that there should be no such restriction that all exposures
to the same obligor must have the same obligor rating grade. At the same
time, banks should be held accountable for defending instances where
this rule of thumb does not hold true. For similar reasons, we do not
agree with Paragraph 36, which states that, "once an obligor is in
default on any material credit obligation... all of the facilities at
that institution are to be considered in default," and believe that it
should be re-worded to exclude that stipulation.
3) Default Grades - Paragraph 36 also references the regulatory
expectation that obligors in default will be assigned to one obligor
default grade. We see no reason why it should be necessary to create a
risk rating bucket that, by design, has a 100% PD, so long as a bank
would always be able to identify what the actual default rate is for
each of its rating buckets. While it is highly likely that defaulting
borrowers would congregate at the lower end of a rating scale, we do not
think that a unilateral default ruing construct should be prescribed to
banks.
Given the overly broad regulatory definition of default that has been
proposed, placing all defaulted borrowers into one risk rating category
would cause Wells Fargo to actually lose risk rating granularity. In the
design of our risk rating scale, defaulted borrowers are assigned a
variety of obligor ratings, depending on the probability that they will
ultimately repay their obligations. We believe that it is important to
preserve this granularity of "distressed" asset risk rating, both for
loan loss reserving purposes, and due to the fact that it drives our
problem loan and collection processes. Paragraph 36 should be changed to
acknowledge that defaulted borrowers can represent a wide variety of
risks of repayment and, therefore, may be assigned different obligor
ratings.
The proposed mandate for a single default bucket becomes a
potentially bigger issue when added to the fact that we disagree with
the proposed definition of default in the first place. Without some
change in the default definition, banks would be faced with the
unnecessary cost of actually creating parallel risk rating methodologies
- one for internal risk assessment and a second for regulatory capital
purposes, with no value added to the risk management process, and,
indeed, the potential to create confusion among those responsible for
identifying and managing risk in the portfolio.
4) Obligor Rating Granularity -- We are also apprehensive that the
language of the Supervisory Guidance may be interpreted by banking
supervisors in such a way that certain concentration limits may be
imposed on the fraction of a portfolio that can be present in any one
risk rating classification (without regard to the nature of the business
being conducted). Paragraphs 44-46 describe some of the tests that banks
must conduct in order to justify the number of obligor grades used in
its rating system, but it clearly leaves banks open to regulatory
criticism on that issue. This should not be the subject of capital
regulation. Rating systems should be tasked solely with rank-ordering
risk in the portfolio and producing valid estimates of PD and LGD that
can be used in the construction of a risk-based capital requirement. The
Accord and the Supervisory Guidance should be silent on the issue of
granularity.
5) Stress Condition LGD's - Paragraph 51 of the Supervisory Guidance
suggests that conservatism be built into the estimates provided for LGD
by limiting the underlying observation set to particular years that can
be called stress conditions. We believe that LGD should simply be
estimated using a "default-weighted" process that is naturally weighted
toward periods with high defaults. Stressed parameters, such as
recessionary LGD's, should be used separately in stress analyses.
6) Recognition of Risk Mitigation - Paragraph 59 states that "while
guarantees may provide grounds for adjusting PD or LGD, they cannot
result in a lower risk weight than that assigned to a similar direct
obligation of the guarantor. While this application of the "substitution
approach" may be roughly appropriate to certain forms of guarantees in
which the financial condition of the borrower and guarantor are closely
linked (say, a proprietor who provides a personal guarantee against the
performance of his business), there are other forms of guarantees (such
as credit derivatives), where this approach does not adequately
recognize the lower risk of joint default or the benefit of double
recovery associated with guarantees.
Failure to recognize the risk mitigation effect of double default in
credit derivatives would send inappropriate signals to banks about the
use of guarantees and credit derivatives -financial instruments that
have provided enormous value in the active management of portfolio
credit risk.
As one illustration of the proposal's inadequacy, consider the case
where a AA-rated counterparty is used to enact a hedge on an unrelated
AA-rated exposure in the banking book. Using the substitution approach,
there would be no capital benefit. Moreover, the bank would have to add
a capital charge for the counterparty exposure associated with the hedge
provider. In effect, the bank would be required to hold more capital
than if it had not hedged at all.
As a solution to this situation, we would support the use of some
form of the modified ASRF approach suggested in the recent Federal
Reserve paper on guarantees and credit derivatives. Under this approach,
regulators could (at least initially) assign the necessary 3 "types" of
asset value correlation (AVC) in conservative fashion (e.g., obligor and
guarantor AVCs according to the Basel AVC-PD equation for commercial
credits, and a "wrong-way" asset-value-correlation of, say, 50%). This
would produce significant reductions in the regulatory capital charges
for a hedged transaction.
Quantification of IRB Systems
1) Short-Dated Maturities -- Paragraph 173 of the Supervisory
Guidance states that most credit exposures must be assigned a maturity
value of no less than one year. We believe that the resultant capital
treatment for legitimate short-term maturities under one year would be
excessive. This unintuitive outcome could easily be adjusted by
essentially decompounding the annualized PD for the subject facility to
the appropriate fraction of a year that corresponded to the remainder of
the exposure. An obligor's probability of default over, say, the next
quarter, must be lower than his cumulative probability of default over
the next 4 quarters, even assuming no credit quality deterioration.
Consequently, unexpected losses (and, therefore, capital) must be less
for the shortdated facility. Implicit in this conclusion is the
requirement that the bank must have the unquestioned right to cancel the
facility at the end of its current term.
Data Maintenance
1) Data Warehousing Requirements -- We are highly skeptical that the
data maintenance standards outlined in Paragraphs 189 to 196 constitute
"best practice." Based on this language, it appears possible that
Paragraphs 189 to 191 may be interpreted by our national banking
supervisors in such a way that they may impose detailed data warehousing
requirements on Expert Judgment risk rating systems of Advanced IRB
banks. The apparent goal of such requirements would be to validate not
only the accuracy of PD and LGD estimates made from a bank's rating
system at the "back-end" of an account's life cycle (which is
understandable), but also the accuracy of the account's initial risk
ratings through an Evaluation of Developmental Evidence at the
"front-end" of its life cycle.
We are unaware of any form of front-end "validation" of either
judgmental or modeled risk ratings that has been demonstrated to have
any statistical power in use anywhere in the financial services
industry, so we have a basic question about the underlying objective for
the supervisory expectations regarding data maintenance. And, if such
interpretive data maintenance standards are a precursor to the required
development of credit scoring models for large wholesale credits, we
believe that over-reliance on credit scoring models for many types of
wholesale lending could produce disastrous results. We believe that such
an approach would actually increase risk in the banking system.
The Supervisory Guidance should be re-worded such that the data
maintenance standards contained in Paragraphs 189 to 196 are not
interpreted as a universal requirement of the Accord, but rather as a
principle to be follow&: by banks wishing to investigate credit scoring
models as "challengers" to the rating systems that they currently have
in place. There should be no implicit, or explicit, mandate in the
Accord for the development of credit scorinq models for large wholesale
credits. Furthermore, the Accord should focus solely on back-testing the
accuracy of PD and LGD estimates made from a bank's rating system, while
eliminating any requirements to validate the accuracy of initial risk
ratings through an Evaluation of Developmental Evidence.
2) Electronic Storage of Guarantor Rating Histories - A specific,
unnecessary data maintenance cost that we perceive in the proposed
Accord is the requirement in paragraph 392 of CP3 that banks maintain
rating histories on recognized guarantors. While we agree with the
standards that are laid out in the Accord for the recognition of
guarantees, we feel that a lender's supporting documentation for 1) the
recognition of a guarantee, 2) analysis of the strength of a guarantor,
and 3) the PD estimate attached to the guarantor should only need to
appear in the physical credit files. It would be unnecessarily costly,
confusing, and without any value, to reproduce this data electronically
when Expert Judgment risk rating systems are employed. In particular,
instances of partial guarantees or multiple guarantees make the
systematic storage of such data problematic.
There may be instances, such as the use of credit derivatives in
hedging the risk in large commercial loan facilities, where it becomes
necessary to explicitly track both the PD of the borrower and the PD of
the guarantor in order to properly model their joint probability of
default. However, such forms of risk mitigation are much less prevalent
in Wells Fargo's commercial loan portfolio of predominantly middle
market and small business customers, and, therefore, present fewer
systematic issues.
The Accord should be re-worded such that the systematic maintenance
of rating histories on recognized guarantors is an option, not a
requirement.
Control and Oversight Mechanisms
1) Independence of Rating Assignments - Paragraph 217 of the
Supervisory Guidance states that "ratings assignments and periodic
rating reviews must be completed or approved by a party that does not
directly stand to benefit from the extension of credit."
Wells Fargo's view on the credit approval process is 180-degrees
opposed to this perspective. We employ an Expert Judgment rating process
for wholesale credits, where lending officers are responsible and held
accountable for assigning and maintaining accurate and timely risk
ratings. They are the principal owners of the risk ratings. The
requirement of ownership is probably the most important aspect of our
credit culture.
We believe that lending officers need to know their customers,
monitor their customers' financial condition and collateral, and surface
deteriorating situations and problem loans early. The requirement and
expectation that lending officers own the risk rating, and are
responsible for the risk rating, forces them to meet management's
expectations with respect to the credit process; that is, properly
underwriting, analyzing, grading, and monitoring their credits. We
further believe that, because of their frequent contact with their
customers, lending officers are the best prepared to surface
deteriorating situations in a timely manner. This fundamental pillar of
our credit culture is augmented by a strong, independent loan review
function (called Risk Asset Review) that evaluates each office's lending
practices and has final authority on the risk ratings assigned.
Now, we are being told that we must change our risk rating system to
conform to some theoretical, yet unproven, new system, even though we
have had one of the best credit quality trends in the industry. We are
not an internationally active bank. Our credit extensions are neither
exotic nor complex. 95% of our commercial loans outstanding are to
middle market and small business customers. Judgments by talented and
experienced lending officers and credit supervisors are particularly
important for these types of customers, as compared to the analysis of
loans to large corporations. A fundamental tenet of our credit culture
is to "know our borrowers."
To shift the responsibility for assigning wholesale lending risk
ratings to an independent rating function seems totally implausible to
us. This would be like asking an auditor to audit a system that has been
developed and implemented by auditors. We strenuously object to
regulatory efforts to disrupt successful risk management cultures that
have been developed through years of training and practice, with proven
results. The Accord should be re-worded to exclude this directive.
Paragraph 220 states that "independence of the rating process can be
achieved through a range of practices that will be carefully reviewed by
supervisors."
Based on this language in the proposal and discussions that we have
held with our national banking supervisors, we believe that our banking
supervisors will require someone besides the lending officer to sign-off
on every rating decision made by the lending officer. This idea is not
only impractical from a cost/benefit standpoint, but it would also, over
time, have a significant adverse impact on our credit culture; that is,
the notion that lending officers own the risk ratings. The Supervisory
Guidance should be reworded so that it does not permit such an
interpretation of its design.
2) Transparency of Risk Ratings -- We are also apprehensive that the
language of the Supervisory Guidance may be interpreted by banking
supervisors in such a way that a certain specificity of risk rating
definitions is prescribed to banks. Although Basel lI allows for an
Expert Judgment system, Paragraphs 221-222 of the Supervisory Guidance
require banks to identify and track specific criteria for each factor
that is considered in a rating decision. This will be required to
achieve "transparency" of a rating system. However, by dictating such a
requirement, the national banking supervisors will, in effect, have
eliminated Expert Judgment systems as a risk rating practice and imposed
Constrained Judgment systems in their place.
Such interpretive rulings would not represent principles of sound
risk management that are unilaterally applicable. They would be
prescriptions, pure and simple. We note, with interest, that the U.S.
banking regulators, themselves, have not followed these prescriptions
when articulating the risk rating scale on which their Shared National
Credit examinations will be conducted in the future. We are adamantly
opposed to this form of capital regulation. The Supervisory Guidance
should be re-worded to exclude the possibility of such supervisory
interpretations of risk rating definitions.
There is nothing inherently wrong with any of the systems proposed by
the Basel Committee and our national banking supervisors. Each may be an
appropriate rating system given certain circumstances, such as the
nature of a particular bank's business or the state of a company's
credit culture. However, Wells Fargo has operated successfully for many
years with its current Expert Judgment rating system. This is confirmed
not just by our financial results, but also by the fact that independent
third parties (principally, our national banking supervisors and rating
agencies) have consistently concluded that Wells Fargo has a sound
credit risk management process, a rating system that rank orders risk,
and a rating system that is accurate.
Wells Fargo would be doing a disservice to its shareholders and
debtholders if it did not defend the risk management practices that have
operated so successfully for the Bank over the years. We strongly
encourage the Basel Committee and the U.S. banking supervisors to
reconsider some of the points that we have made above, to remove some of
the regulatory prescriptiveness relative to the operational detail of
bank risk management policies and practices, and to allow banks like
Wells Fargo to preserve well-functioning credit cultures that they have
developed.
Comment Two
November 12, 2003
To: Addressees
Re: Supervisory Guidance on Operational Risk Advanced Measurement
Approaches for Regulatory
Wells Fargo & Company appreciates the opportunity to participate in
the ongoing dialogue on the Basel capital reform proposal. We are a
diversified financial services company, providing banking, insurance,
investments, mortgage and consumer finance from more than 5,600 stores,
as well as through the Internet and other distribution channels across
North America. As such, we have a keen interest in the framing of the
Basel Accord and hope that the comments that we offer in this paper will
be of assistance in providing solutions to the issues that exist in the
current proposal.
Sincerely,
Howard I. Atkins
Executive Vice President and Chief Financial Officer
Addressees:
Docket No. R
Ms. Jennifer J. Johnson, Secretary
Board of Governors of the Federal Reserve System
20th Street and
Constitution Avenue, NW
Washington, DC 20551
Docket No. 03
Communications Division
Third Floor
Office of the Comptroller of the Currency
250 E Street, SW
Washington, DC 20219
Mr. Robert E. Feldman, Executive Secretary
Attention: Comments
Federal Deposit Insurance Corporation
550 17th Street, NW
Washington,
DC 20429
Wells Fargo & Company appreciates the opportunity to participate in
the ongoing dialogue on the Basel capital reform proposal. While we
respect the tremendous amount of time and effort that has gone in to
shaping the proposal, we find that we still have some fundamental
differences of opinion with the path on which the Basel Committee and
the U.S. banking supervisors are proceeding and feel that certain
aspects of the proposal must be changed in order for it to be
acceptable.
We will direct our comments here to the Draft Supervisory Guidance on
Operational Risk Advanced Measurement Approaches for Regulatory Capital
dated July 2, 2003. We have drafted separate comment letters for the
related Advanced Notice of Proposed Rulemaking ("ANPR") and the Draft
Supervisory Guidance on Internal Ratings-Based Systems for Corporate
Credit, although we may allude to some of that commentary in the course
of this dialogue.
Wells Fargo has a basic difference of opinion with the Basel
Committee with respect to the capital treatment of Operational Risk,
insofar as we don't believe that capital should be required for
Operational Risk at all. To understand this perspective, one must first
bifurcate operational losses into two segments -- 1) high frequency/low
severity losses that can be statistically assessed, expensed, and priced
for, and 2) low frequency/high severity losses that cannot be reliably
modeled.
We would argue that high severity losses should be outside the scope
of a formulaic approach to minimum regulatory capital standards, because
they are unpredictable and so remote as to be outside the statistical
bounds of what should be captured in capital at risk formulae. We also
feel that the more predictable forms of operational losses are 1)
simply a cost of doing business to a bank and, therefore, routinely
factored into the way that banking products are priced, 2) quite
stable over time, because of their predictability and absence of
correlation across businesses, and 3) likely to be entirely offset by
the Future Margin Income generated by a bank in aggregate, across
all of its operational businesses. Many of these losses are either
expensed or accrued for by banks. Because the Advanced Measurement
Approach (AMA) would allow for the recognition of imperfect correlation
of risks and the impact of Future Margin Income, we feel that the
aggregate outcome of such modeling of predictable operational losses
would typically result in a zero capital requirement, and thus, such
risks should simply be exempted from the minimum regulatory capital
requirements in the first place, as they are today.
In the event that the Basel Committee and the U.S. regulatory
authorities deem it prudent to persist in their direction to assign
regulatory capital to operational risk, the primary point that Wells
Fargo will emphasize lies with the AMA itself. Although the draft
supervisory guidance on the AMA approach is, in many ways, more
enlightened that the Accord's credit risk capital formulations (in terms
of its recognition of Future Margin Income and portfolio diversification
benefits), it should be acknowledged that there is no accepted
methodology for quantifying Operational Risk.
As such, we believe that the AMA approach should not be the only
option made available to U.S. banks. All institutions subject to the
Accord should be allowed to develop any risk measurement methodology
(Basic Indicator, Standard, AMA, or other alternative) that is
acceptable to their national banking supervisors, and to disclose their
methodology and their key controls for managing operational risk in
their public filings.
The proposed AMA framework for Operational Risk leaves banks with the
task of developing a complex and costly methodology for operational loss
estimation. This choice begs the question of whether there may be an
alternative approach that demonstrates that no capital is required for
operational risk, given a particular bank's facts and circumstances, or
whether there are alternative approaches to determining operational risk
capital that are consistent with the way sound businesses actually
operate, without being overly complex or costly to administer.
For example, we note that the well-known concept of operating
leverage, or business risk, seems to be totally overlooked in the Basel
Committee's operational risk capital deliberations. We feel that
constructing a business-based approach to operational risk capital
should be viewed as an acceptable alternative to the AMA track. We would
encourage further discussions between the regulatory agencies and their
regulated institutions along the lines of quantifying the main elements,
definitions, and procedures of this type of framework.
Aside from emphasizing this important, general concept, there are
several specific comments that we have on the details of the Supervisory
Guidance.
Definition of Operational Risk
Paragraph 10 of the Supervisory Guidance defines operational risk to
include "...exposure to litigation from all aspects of an institution's
activities." This would appear to include settlements of baseless
lawsuits as operational risk losses. In many cases, these settlements
are made to control costs or to maintain customer relations and more
appropriately represent strategic risk rather than operational risk. We
believe that this language should be modified, so that banks would have
some flexibility to exclude certain lawsuit settlements from the scope
of operational risk capital.
We also believe that because many operational losses do not get
posted to the general ledger as discrete items (e.g., trading losses),
the U.S. Banking Supervisors should acknowledge that a reconcilement of
operational loss data to the general ledger is not expected or required.
Corporate Governance of Operational Risk
Paragraph 15 of the Supervisory Guidance appears to mandate that an
independent, firm-wide operational risk management function exist, which
is separate from line of business management oversight. We believe that
such a directive is premature, given that a consistent, well-meaning
definition of what operational risk comprises does not yet exist. Under
these conditions, how can there be a central committee to oversee
something that is not defined?
For similar reasons, we are concerned that banking supervisors may
interpret Paragraphs 16 and 17 to develop unrealistic expectations for
the board of directors' involvement in the oversight of operational risk
management. Paragraph 16 states that "the Board is responsible for
overseeing the establishment of the operational risk framework, but may
delegate the responsibility for implementing the framework to
management." It is difficult to require board of directors oversight for
something that is not well defined.
Rather than trying to devise a "one size fits all" central oversight
function for operational risk management, we think that the Supervisory
Guidance should be re-worded to simply require that there be a thorough
governance process for overseeing what may be multiple types of
operational risks, with the details left to the discretion of individual
banks.
Validity of Extreme Loss Modeling
Certain operational loss events are relatively small and frequent.
Such events can be successfully modeled through the use of statistical
techniques applied to historical data sets. Because such losses are
relatively predictable, they can effectively be priced into the product,
in much the same manner as expected credit losses are priced into credit
products.
We support the Basel Committee's decision to allow Future Margin
Income (FMI) to offset the expected component of operational losses
(EL). However, we believe that the language used in Paragraph 62 of
the Supervisory Guidance should be strengthened to make it clear that
FMI is an acceptable offset to EL, given the existence of suitable
documentation on the two metrics.
On the other hand, we are doubtful that similar statistical
techniques can be applied to historical data to reliably model extreme
operational loss events. Truly catastrophic loss events cannot be
predicted, and no amount of capital will protect an institution in such
an instance. We believe that some form of qualitative (scenario
analysis) modeling is more appropriate in assessing those types of loss
events that are less predictable. Accordingly, we think that more
development is necessary to finalize exactly what types of loss events
ought, realistically, to be captured under AMA approaches to Operational
Risk capital formulations.
Wells Fargo very much supports the use of information databases and
statistical analysis, but only as a means of understanding/managing its
operating expenses, not as a requirement for establishing capital
levels.
Risk Mitigation
To the extent that extreme operational loss event modeling is deemed
realistic, we see no reason why the recognition of insurance mitigation
should be limited to 20% of the total operational risk capital charge,
as suggested by Paragraph 66 of the Supervisory Guidance. To do so might
lead to imprudent risk management incentives in the use of insurance
programs. We recommend that the capital adjustment for insurance be
based on the full amount of insurance protection provided by insurance
policies, given that the policies meet the qualitative standards
outlined in the Supervisory Guidance.
Correlation of Operational Losses
At Wells Fargo, we believe that we have consciously crafted a
distinct competitive advantage by virtue of the diversity of our
underlying businesses. Between mortgage banking, commercial banking,
insurance, retail deposit taking, and asset management services (to name
a few of our over 80 businesses), along with the significant economies
of scale that we have in each of these businesses, we feel that Wells
Fargo has created a portfolio of risks (both credit and non-credit)
whose worst-case loss potential is substantially less than the sum of
its parts. We are encouraged to see that the capture of the capital
benefits created by business diversification is permissible under the
AMA modeling of operational risk. We believe that this logic should
extend to the modeling of capital for credit risk as well, where the
impact of portfolio diversification is more substantive and more
empirically justifiable.
However, we are concerned by the language of Paragraph 64 of the
Supervisory Guidance, which states that "Under a bottom-up approach,
explicit assumptions regarding cross-event dependence are required to
estimate operational risk exposure at the firm-wide level. Management
must demonstrate that these assumptions are appropriate and reflect the
institution's current environment". The requirement for institutions to
demonstrate that explicit and embedded dependence (correlation)
assumptions are appropriate needs to be clarified. It is important that
reasonability be incorporated into this standard. Insufficient data will
be available to statistically prove correlations across business lines
and event types. Therefore, correlations most likely will be determined
from qualitative reasoning based on the underlying nature of the risks.
We suggest that the language in this section recognize the fact that
qualitative judgment will be necessary and that flexible approaches need
to be allowed, provided that institutions have a well - reasoned basis
for their assumptions. It is important that overly conservative criteria
not be applied regarding correlation assumptions so that banks using
more risk-sensitive "bottoms-up" approaches to the quantification of
operational risk capital are not penalized.
Use Of External Data
We do not believe that the direct incorporation of external loss data
should be a required component of a bank's operational loss modeling.
While it is instructive for banks to be aware of external loss events,
applying that information across all institutions in a formulaic manner
seems problematic to us. The quality and consistency of external data
would prove difficult to verify, especially given the lack of common
data collection standards within the industry. Furthermore, each bank
will have its own inherent and specific causes of risk depending on the
diversification of its lines of business and appetite for risk. Without
a relatively detailed awareness of the internal control conditions that
led to those losses at other institutions, it is difficult, at best, to
do much more than guess the impact of a seemingly similar event on a
given bank. Accordingly, external data should only be one of several,
optional considerations when performing scenario analysis, and not
necessarily the most important.
