via emailFleet
FleetBoston Financial
November 3, 2003
Ms. Jennifer J. Johnson
Secretary
Attention: Docket No. R-1154
Board of Governors of the Federal Reserve System
20th Street and Constitution Avenue, NW
Washington, DC 20551
Mr. Robert E. Feldman
Executive Secretary
Attention: Comments
Federal Deposit Insurance Corporation
550 17th Street, NW
Washington, DC 20429
Office of the Comptroller of the Currency
Attention: Docket No. 03-14
250 E Street, SW
Public Information Room
Mail Stop 1-5
Washington, DC 20219
Regulation Office
Chief Counsel’s Office |
Attention: No. 2003-27
Office of Thrift Supervision
1700 G Street, NW
Washington, DC 20552
Re: Risk-Based Capital Guidelines; Implementation of the New Basel
Capital Accord; Proposed Rule and Notice
Ladies and Gentlemen:
FleetBoston Financial Corporation (“FleetBoston”) is the seventh
largest diversified financial holding company in the United States with
total assets of U.S. $197 billion. Headquartered in Boston,
Massachusetts, FleetBoston has consumer and commercial banking
platforms, as well as asset management and capital markets businesses,
serving approximately 20 million customers worldwide.
The U.S. version of the New Basel Capital Accord (“New Accord”),
issued for comment in August 2003 as the Advanced Notice of Proposed
Rulemaking ("ANPR"), would have a significant impact on our institution
and on the banking industry in general. As such, we appreciate the
opportunity to provide the U.S. bank supervisory agencies (“Agencies”)
with our thoughts and concerns regarding the ANPR.
FleetBoston commends the Agencies for their continued work with the
Basel Committee on Banking Supervision (“Committee”) to improve the
proposed rules. The Third Consultative Paper on the New Basel Capital
Accord ("CP3"), on which the ANPR is based, is a significant improvement
over the previous version, especially in the calculation of minimum
regulatory capital in Pillar I. We believe this stems from the Agencies’
openness to industry comments and suggestions and their willingness to
incorporate capital “best practices” into the rules. We especially
appreciate the proactive process by which industry input has been
solicited.
This letter offers FleetBoston's specific comments and concerns with
the rules as presented in the ANPR, as well as the supervisory guidance
released simultaneously in the Internal Ratings-Based Approach for
Corporate Credit (“AIRB SG”) and the Operational Risk Advanced
Measurement Approach for Regulatory Capital (“AMA SG”). Much of the
substance of this letter is unchanged from our response to CP3 dated
July 31, 2003 because the advanced approaches incorporated into the ANPR
are largely unchanged. We have attempted to provide additional color on
many of the topics plus additional detail not covered in the previous
letter. You will see that our responses are divided into two sections:
(a) foundation issues that are of major concern and (b) implementation
issues.
Many statements and related stories have been disseminated following
the October, Madrid conference of the Basel Committee, indicating that
certain key elements of the New Accord have been modified and/or
clarified. We have restricted our comments herein to the existing ANPR,
which does not include any of these potential changes. Once the final
decisions of the Madrid conference are agreed to and disseminated, we
will provide feedback on these modifications.
FOUNDATION ISSUES
A Full Internal Models-Based Approach for Regulatory Capital should
be adopted. FleetBoston commends the Agencies for their attempt to
improve upon the regulatory capital rules originally laid out in the
1988 Basel I accord. The goal of more directly linking the levels of
regulatory capital with underlying economic risks is key to this
improvement and one that we wholeheartedly support. Many large banks,
including FleetBoston, have been managing their business for years using
risk-sensitive internal economic capital models. Drawing on our own
experience, the economic capital process has proven quite successful in
the management of risk, the allocation of capital, servicing our
customers effectively, and improving returns to our shareholders. While
the advanced approaches are an improvement over the existing rules of
Basel I, it is our strong belief that the use of the computation methods
proposed by the ANPR would result in a process that falls short of
achieving a truly economically-based, regulatory capital methodology. We
will elaborate further throughout this letter.
FleetBoston recommends that a bank's regulatory capital should be
determined solely using its internal models, regardless of risk type
(i.e., credit, market, and operational risks). Naturally supervisory
oversight, review, and approval would be required. In the ANPR, we see
specific areas where reliance on internal models is accepted: the
internal models-based approach for equity exposures, the existing
market-risk capital process, and now in the AMA for operational risk. We
question why a full commitment to the use of internal models is not
pursued for credit-risk capital. With credit risk, data is plentiful and
increasingly available, quantification techniques are well developed,
and the market practices to both price and hedge risk are improving
every day. Yet with all this, the ANPR continues to provide a great deal
of prescription that in total arrives at capital results that are too
conservative. In contrast, the determination of operational risk
capital, a discipline very much in its infancy, is required to make use
of a bank-developed AMA. An internal models-based approach incorporates
flexibility and simultaneously ensures that regulatory capital
appropriately reflects economic risks; as such, we believe that capital
allocation for all risks should be addressed through this approach.
Regulatory Capital for Expected Losses is inconsistent with Financial
Theory and Industry Practice. The New Accord, and hence the ANPR, leave
the definition of regulatory capital unchanged from the current rules.
We believe this has forced the Agencies to treat regulatory capital as
covering expected losses in addition to unexpected losses. This practice
is inconsistent with financial theory and industry practice, where
capital provides protection against unexpected losses only. Typically,
economic capital consists of common equity and is used for protection
against unexpected losses only. That is, capital protects against the
volatility of actual losses around the long-term average or expected
level. Unfortunately, having different meanings for the term "capital"
has been a barrier to effective communication about the similarities and
differences between regulatory and economic capital methodologies, often
leading to confusion and a lack understanding of the New Accord.
Additionally, the Agencies have had to incorporate "work arounds" into
the ANPR (e.g., inclusion of Future Margin Income, “FMI”) to ensure that
capital is not required where income has already been provided to absorb
expected losses. This adds unnecessary complexity to the rules.
Expected losses are a recurring “cost of doing business” that are
explicitly incorporated into all transaction pricing, particularly for
credit products. Consequently, losses are charged against income as
incurred or via a provision to build a reserve. To require capital for
expected losses suggests that actual credit charge-offs or operating
losses should be charged directly against capital, not earnings. If this
were so, then banks would be freed from the need to charge these losses
against earnings, as they have already set aside capital.
FleetBoston recommends that the regulatory capital methodology be
revised to cover the potential for unexpected losses only. However, if
the Agencies continue to pursue the approach of covering both
loss-types, then (1) all of the loan-loss reserves should be included as
regulatory capital; and (2) a credit to regulatory capital that reflects
all of an institution’s earnings power should be provided (i.e., the
inclusion of all the Future Margin Income needed to cover expected
losses and not just the limited amount as proposed in the ANPR). In
addition, we view FMI and loan loss reserves as our “first line of
defense” against potential losses. Consequently, we believe that these
two forms of protection should count as Tier 1 capital not Tier 2.
Recently, the Basel Committee on Banking Supervision (“BCBS”)
released a statement detailing an approach for regulatory capital that
will cover unexpected losses only. Included in this proposal is a
redefinition of regulatory capital. We are pleased to see the
supervisory authorities considering this needed change to the New
Accord. As discussed earlier, we will offer our comments on this
proposal in a separate letter to be submitted after completion of these
comments.
The proposed capital floors are inappropriate given the best-practice
standards identified and the disincentives that are created for non-core
institutions. First, with all the effort expended over the past five or
so years by the supervisory community and the banking industry on the
New Accord, we believe it is a mistake to restrict its ability to
function as designed. The first year (90%) and second year (80%)
risk-weighted asset floors, which effectively limit the regulatory
capital reduction available from the ANPR, impair the Accord's
effectiveness and reduce incentives for banks to devote the resources
needed to adopt the advanced approaches. We understand the need for some
amount of conservatism, but not the exclusively unidirectional way in
which it is applied here. We believe that these floors should be
removed, allowing the methodologies in the Accord to function as
intended, with regulatory capital requirements reflecting true, economic
risks. Supervisory oversight can be invoked as needed to have
institutions alter their capital levels if the proposed rules do not
properly reflect the risk profile. If, however, the Agencies decide to
retain these floors, then we urge that caps, similar in nature to the
floors, be instituted.
Credit Risk Capital is too conservative. We believe that the
calculation of capital for credit risk includes several adjustments,
which individually seem reasonable, but collectively are too
conservative. Some examples include: (a) the LGDs on residential
mortgages have a floor at 10%, (b) not providing the full regulatory
capital benefit provided by Future Margin Income ("FMI") (e.g., only 75%
for qualifying revolving retail exposures and nothing for other credit
products), (c) not recognizing the full risk-reduction benefit provided
by guarantees through "joint probability of default," and (d) the strict
matching required for the risk reduction provided by credit default
swaps.
We believe that supervisory review and validation of bank PDs, LGDs,
and EADs — a prerequisite for the use of internal data in the regulatory
capital calculations — provides enough oversight without additional
minimums and maximums, which themselves appear arbitrary. During the
internal model certification process, concerns surrounding assumptions
and data calculations should be raised and addressed. It should not be
assumed, a priori, that the inputs are required to be more conservative
using rules-based procedures. If, during the validation process,
examiners conclude that a bank has not shown proper back-up or
sufficient justification for its inputs, supervisors can then require
additional conservatism in the parameters. The advanced approach is, by
its nature, the best method to assess economic risk. Arbitrarily
constraining the results of such risk assessment would seem to send an
unintended message to bank management. Why, for instance, would
management require mortgage borrowers to have private mortgage insurance
("PMI"), a prudent risk management action, when it receives only partial
regulatory capital relief?
Further, we point to the statement within the AIRB SG (page 45960),
in the context of credit-risk parameter estimation: “Margins of
conservatism need not be added at each step; indeed, that could produce
an excessively conservative result.” The cumulative effect of the
requirements, including stressed PDs and LGDs, would appear to be
inconsistent with the intent of avoiding an excessively conservative
methodology.
These restrictions will continue to invite capital arbitrage, which
undermines the New Accord at its inception. With regulatory capital
continuing to be significantly more conservative (i.e., higher) than the
economic risks would suggest, efforts to circumvent the New Accord will
continue (i.e., opportunities for regulatory capital arbitrage would
continue to be available). This result could be eliminated up-front
through a closer alignment of capital with the economic risks.
For all the reasons above, FleetBoston strongly urges the Agencies to
adopt the Advanced IRB approach without any specific supervisory
minimums for the major inputs. It is our view that this would also
motivate non-core institutions to move towards a more robust
risk-management framework.
A Specific Capital Charge for Operational Risks is inappropriate at
this time. In general, we support the Agencies’ desire to develop a
quantitative methodology for operational risk capital. Our view is that
the current state-of-the-art practices for operational risk measurement
and the collection of the loss events needed to develop and calibrate
the models have not progressed sufficiently to warrant their use in the
assignment of a minimum regulatory capital charge at this time.
We continue to believe that the measurement of operational risk is an
emerging discipline of research that is an important risk management
area for the industry and its regulatory authorities, and that capital
should ultimately be assigned for this risk. While good progress has
been made in the quantification of the risk, much important work remains
before a capital charge can be determined. These tasks include:
• Collection of enough historical loss-event data to be statistically
meaningful. This applies both to an individual bank's collection of its
internal loss events and to the collection across the industry in total.
• Development of accurate and conceptually sound quantitative
techniques needed to transform loss data into capital requirements.
• Resolution of context-dependency issues within the same
institution, reconciling loss-event data across lines of business, and
within the same line of business where the business model, internal
control factors, or other foundation variables have changed
dramatically.
• Incorporation of external loss data into individual bank capital
calculation requirements with as yet unperfected methods of scaling
these data to a bank's activity level, as well as for the differences
that exist in control environments.
While capital is an important component of a bank's risk management
toolkit and can provide meaningful protection against unexpected
operating loss events, the cost of operational risk is typically a “cost
of doing business” to be covered through current period operating
earnings. Banks have made a significant investment in risk-mitigation
costs, such as establishing and maintaining risk-control systems,
internal and external audit oversight, and insurance protection. All of
these form the first line of defense against the effect of operational
errors and are paid for through charges against annual earnings. With
all the focus on capital that the New Accord and the ANPR bring to bear,
we fear that these time-tested risk mitigation and control techniques
will be potentially overlooked, opening the door to significant losses.
In addition to the quantification challenges, a specific capital
charge does, in our view, create an uneven playing field. Products and
services that are primarily operating in nature form a significant
source of revenue and profit for FleetBoston through our ability to
serve our customers' complete financial needs. Our concern is that a
specific capital charge will put us and the global banking industry at a
competitive disadvantage vis-à-vis non-bank competitors who are not
subject to the same rules.
At this time, FleetBoston urges the Committee to remove the explicit
capital charge for operational risk from Pillar I and to include the
assessment of this risk as part of the examiners’ periodic reviews of a
bank’s risk profile (i.e., Pillar II approach in the vernacular of CP3).
Our recommendation is not without precedent. The assessment of
structural interest rate risk (i.e., that which resides in the basic
banking book) has been recognized as an explicit risk long before
operational risk, and has been the explicit cause of a number of bank
failures in the past. The measurement processes are very well developed
and have undergone extensive scrutiny by the industry, supervisors, and
academicians. All would agree that there is some level of economic risk
here, but the Agencies have decided, and we strongly agree, that this
risk is better handled within a supervisory review framework unless a
bank is running an extreme position.
We continue to believe in a supervisory review (Pillar II) approach
for now. If, however, the Agencies decide to continue with a Pillar-I
capital charge for operational risk, the AMA approach appears to be a
“distant second” option, but may provide workable solution. As we stated
earlier, allowing a bank to use its internally-developed data and
models, with regulatory oversight of course, will generate a more
accurate representation of the underlying risks, which will in turn
result in a more accurate level of regulatory capital. Additionally, we
feel that an AMA affords the desired flexibility, including the
reduction in risk afforded by insurance, and lack of prescriptiveness
needed at this stage of development. Forcing banks to use a particular
approach when industry, market, and supervisory "best practices" are
still very much in the development phase would be a mistake. At this
point in the development of operational risk measurement, supervisors
need to encourage the creation of multiple innovative techniques, which
we believe the AMA allows.
We understand the dilemma with which the Agencies are faced,
balancing standardization of rules to foster comparability versus
recognizing differences to allow for flexibility. The AMA provides
ultimate flexibility but is lacking guidance on how an examiner is going
to determine if a bank's process is certified for use as an AMA. This
issue requires additional discussion and specificity.
In any event, FleetBoston urges continued collaboration between the
industry and the bank regulatory authorities in advancing risk
assessment techniques for operational risk.
The capital-reduction benefit of Line-of-Business (or risk type)
diversification should be incorporated into the new framework. Nowhere
in the ANPR is there a capital benefit provided to banks that operate a
diverse mix of businesses. Business-line diversification, or
alternatively risk-type diversification, mitigates both the possibility
and magnitude of unexpected loss, and as a result, banks should be
allowed a capital credit or risk-weighted-asset reduction in recognition
of such diversification. For a well-diversified institution, there is a
decreased probability of a bank experiencing significant losses in each
of its businesses simultaneously. For example, the ANPR would determine
capital for a bank operating only two business lines — let's say
consumer banking (primarily a credit-risk activity) and asset management
(primarily an operational-risk activity) — as the simple addition of
credit-risk capital and operational-risk capital, computed
independently. We feel this approach overstates the institution’s total
risk and as a result, the amount of regulatory capital required. In
addition, ignoring the benefit sends an inappropriate message that
diversification of risk provides no capital benefit, which we believe is
incorrect.
We offer two examples in support of this risk-diversification
benefit. First, banks consider this benefit in their economic capital
models. FleetBoston has developed internal estimates of the capital
reduction that comes from its existing mix of businesses. The results
demonstrate that business-line diversification allows FleetBoston to
reduce its economic capital from the levels that would result from the
simple addition of the stand-alone capital needed for each business. JP
Morgan Chase publicly discloses a capital-reduction benefit from
diversification of about 16%.1 We are not suggesting the appropriate
level of diversification benefit that would be determined by an
individual bank's business mix and risk profile. Instead, we are
attempting to point out that banks recognize its value in their economic
capital models and in the management of capital.
Second, in our informal discussions with external rating agencies,
they acknowledge that diversification is an inherent risk mitigant and
is reflected in their ratings of financial institutions. For example,
monoline credit card companies are viewed as riskier than
well-diversified financial institutions, all else being equal. This can
clearly be seen in the higher levels of capital required by the markets
for concentrated businesses.
FleetBoston strongly suggests that any final rule must include a
reduction in regulatory capital for the benefit of line-of-business
diversification, at the total bank level.
The following example provides a simple approach on how the benefits
might be computed using a sliding scale and the businesses outlined in
the operational-risk proposal of CP3.
In the consultative document, there are eight Level-1 business units.2
To qualify for any decrease, an institution must first have at least two
business units that each contributes at least 10% of the bank’s annual
pre-tax net income. This would set a base reduction in the amount of 6%
of the aggregate regulatory capital requirement. For each additional
business that contributes at least 10% of annual pre-tax net income, an
extra 1% reduction in regulatory capital would be provided. Therefore, a monoline bank receives no diversification benefit, while those with two
to eight businesses can have total regulatory capital reduced by an
amount ranging from 6% to 12%. While conservative correlations of 0.75
were assumed in creating the preceding example, it is meant to suggest a
framework to quantify a benefit we feel is warranted, meaningful, and
quantifiable
The ANPR results in an Uneven Competitive Playing Field. As with the
current regulatory capital rules, the ANPR applies only to banking
firms. This continues to leave non-bank competitors free to pursue their
business activities unencumbered by supervisory capital rules. Two
examples come to mind: commercial real estate (“CRE”) and operating
services (e.g., asset management). We, and other U.S. banks, compete
against captive finance companies, insurance companies, and investment
banks for CRE business. In operating services, the non-bank competition
is largely from investment banks and other independent service
providers.
We feel that this approach will result in assets or businesses
flowing from the banking system to entities without these capital
regulations. This result hurts banks’ earnings potential, making them
more susceptible to economic downturns. We have heard arguments that
banks have advantages (e.g., access to discount window borrowings in
time of liquidity stress) that are paid for through strict regulation.
In spite of this, national banking regulators need to be aware of
unintended consequences. For example, with the potential for so many
activities to be conducted outside the banking system, the ability of
national regulators and central bankers to control systemic risks is
greatly reduced, which could pose threats to the stability of national,
as well as global financial systems.
Another potentially adverse competitive element of the New Accord
relates to interpretations. It appears that supervisors in each country
are allowed wide discretion to interpret the New Accord as they see
fit—witness the decision to adopt only the advanced approaches in the
U.S. This could provide banks in a particular country operating under a
more favorable interpretation of the capital rules with a distinct
competitive advantage. Often this is referred to as the "home / host"
issue and will be a concern for all banks not just internationally
active ones. For instance, a non-indigenous bank may be able to attract
business and customers from local banks if it is operating under a
less-restrictive interpretation of the New Accord provided by its "home"
regulator, which in turn allows for lower pricing on products and
services. Therefore, great care needs to be taken to ensure this does
not happen and that a standard and uniform interpretation of the
proposed rules is used by all global regulatory agencies. While we have
reviewed the High-level principles for the cross-border implementation
of the New Accord published by the Basel Committee in August and
understand that an Accord Implementation Group has been established to
deal with these concerns, we have not yet seen the definitive rules or
explicit guidance required to resolve this issue.
Similarly, it is not clear from the ANPR how uniformly the advanced
approaches will be applied across subsidiaries of core institutions.
This issue must be addressed consistently by all regulators across the
globe.
The Asset Securitization Capital calculation is unnecessarily
complex. First, we welcome the Committee's approach that caps the
regulatory capital requirement in situations where a bank retains
first-loss exposure in a securitization of its own assets (i.e., banks
as an issuer in securitizations). Any set of rules that results in more
capital after a securitization than if the assets remained on a bank's
balance sheet is fundamentally flawed. Under the proposed approach,
banks that have on-balance-sheet, credit-enhancing IO strips first must
deduct these positions from capital and thereafter apply the AIRB cap.
This process could result in capital charges that exceed the capital
requirements of the “non-securitization” alternative mentioned above. We
feel that the maximum capital charge should by capped by the AIRB
capital including the IO deduction.
Securitization of assets is an important liquidity management tool
for banks, since it represents funding with limited issuer event risk
and enables banks to tap a non-traditional funding base, which
ultimately lowers liquidity risk through the diversification of funding
sources. We are strongly opposed to any capital rules that do not
accurately reflect the economic risks and benefits of securitization.
FleetBoston's fundamental belief is that capital for securitization
activities should be based on the results of a bank's own internal
models and subject to regulatory review. Banks participating in these
markets have devoted significant resources developing their own,
economically-based models to understand the risks inherent in the pool
of assets backing the securities issued to investors and retained by the
firm. The bank's evaluation of economic risks is further validated by
the rating agencies as both issued securities and retained interests are
frequently rated. In addition, since these securities are sold in the
capital markets, the issuer is able to receive further independent
confirmation as to the economic risk inherent in these assets. We
believe the calculation of retained economic risk should ultimately
determine capital requirements.
As in other areas of the ANPR, the capital required under the
proposed approach is too conservative (i.e., the capital is too high
based on the underlying economic risks). This conclusion is based upon
the results of a comparison of our own internal economic capital models
with the results of QIS3. Also, the additional complexity of "Kirb",
"L", and "T" terms does not materially improve the resulting capital
requirements. As a simpler, more direct approach, institutions would
first determine Kirb as proposed. Then, the needed capital would be the
lesser of this amount or the total retained first-loss position. Capital
requirements for other tranches held by the bank would utilize the AIRB
approach to calculate their risk-weighted assets or the RBA, as
proposed.
A bank that is an investor in a securitization should be allowed to
use its own internal credit models in the determination of regulatory
capital, which is consistent with our earlier comment on the use of
internal models for all credit risks. While CP3's Internal Ratings-Based
Approach ("IRB") is a welcomed improvement over the current rules and a
step in the right direction, we feel that it still overestimates the
capital needed based on the securities' economic risk, especially in the
case of liquidity facilities. This is another example of the unnecessary
conservatism that is used in the New Accord for minimum regulatory
capital calculations.
Additional Disclosure Requirements are excessive and unnecessary. The ANPR requires that a bank using the advanced approaches make extensive
additional disclosures about its risk profile and risk management
processes. These disclosure requirements are fundamentally flawed and
should be dropped from the proposal for several reasons. First, the
additional disclosures will not achieve the intended effect of
increasing market understanding of financial institutions. The market is
sufficiently well informed already as evidenced by the size of the
market for debt issued by financial institutions. A financial
institution transacts business daily in a variety of capital markets by
raising wholesale funding, issuing debt, and providing clients with
risk-management products. All of these transactions require the market
to constantly assess the financial institution's creditworthiness
including, but not limited to, its capital structure. Market
participants have ample opportunity and resources to evaluate credit
risk with the information already at hand. If the market needs more
information, it will demand it. We do not believe the Agencies have any
unique insight into the additional information that may be required by
the market to make the same credit decisions it makes today without
these disclosures.
Complying with the disclosure process would be quite burdensome and
costly. Much of the information exists in formats designed for internal
use and access. To translate and transmit the data would require
additional staff and systems to ensure that the data is available,
understandable, and current. Our biggest additional burden would be the
effort and resources to ensure that the ANPR disclosures are put into
the proper context of the bank's risk profile and not misinterpreted.
Oftentimes, analysts or investors have limited time in which to
understand changes in a company's financial position. This can force
them into generic assumptions (e.g., all banks are the same) or cursory
reviews that lead to incorrect conclusions. For example, a simple
disclosure of an increase in exposures in poor-quality PD bands may be
picked up, but without consideration of mitigation techniques employed
to offset this apparent increase in risk.
The disclosure requirements may also create additional securities law
liabilities for financial institutions that are subject to U.S.
securities laws. Capital reserves represent an implied view of future
expected losses. Any required disclosure of a bank’s assessment of its
capital position is, by its nature, a “forward-looking statement” which
litigants could use to bring suits under U.S. securities laws with the
benefit of “20/20 hindsight”. While securities laws provide some safe
harbors that may mitigate this risk, it cannot be eliminated. This
incremental litigation liability would not accrue to financial
institutions that are not subject to U.S. securities laws. Ultimately
this would represent a competitive advantage for those institutions. We
also believe that forced disclosure of much of a bank’s risk profile
represents an unfair compromise of confidential and proprietary business
intellectual property. Financial services companies, not regulated by
banking authorities, would also gain significant insight and competitive
advantages as a result of these disclosures. Ultimately, a weakening of
the banking system could result, which clearly, is contrary to the
objectives and intent of the rule. The disclosure requirement may also
result in a constriction of credit availability to less creditworthy
customers. Banks using the AIRB approach would have an added incentive
to avoid extending credit to any borrower that could trigger a further
disclosure of increasing risk positions.
Finally, we urge that the Agencies do not make these disclosure
decisions in a vacuum. Any changes to the current requirements should be
made in conjunction with the appropriate accounting oversight bodies and
securities regulators, such as the FASB and SEC in the U.S. and IASB
internationally. Lack of collaboration with other authorities would
result in unnecessary and potentially all too frequent changes to the
requirements. This would also provide a way to ensure that the proposed
disclosures are relevant by consulting with these authorities prior to
implementation.
IMPLEMENTATION ISSUES
Implementation timing must be reassessed. Given the extended horizon
for finalization of the New Accord and the ANPR, and the foundation
issues that are not yet resolved, we would urge the regulators to
reconsider the U.S. implementation date of January 1, 2007. The
feasibility of this deadline is further strained by the data-history
requirements and data-collection challenges faced by core institutions,
both across specific commercial-credit segments and within the
operating-risk area. Within the commercial-credit discipline in
particular, this deadline will be challenging, given the 5-year
data-history requirements for PD estimation and the 7-year data-history
requirements for LGD and EAD estimation. On a cumulative basis, the
implementation deadline does not provide an adequate timeframe for core
institutions, particularly given the requirement that these institutions
must receive approval from their primary Federal supervisor by year-end
2005. At a minimum, we would urge the regulators to relax the
data-history requirements and recognize the evolving nature of specific
methodologies within core institutions.
The definition of default needs to be refined, to more accurately
reflect credit risk in both the commercial and consumer areas. In the
commercial-credit area, the definition of default should be consistent
with market convention (“credit events” defining default in the
credit-derivative market) and rating-agency practices (which focus on
“failure to pay” and “non-accrual”). In particular, “silent defaults” by
definition are not reflective of credit risk, and as such, their
inclusion would serve to artificially increase PD, decrease LGD, with
little or no incremental value added to the predictive ability of the
rating system. Similarly, leases of ‘essential-use’ equipment often
remain current even while the underlying creditor has defaulted on other
obligations. In fact, asset-based lending exists as a product because a
specific obligor is “…unlikely to pay its credit obligations…in full,
without recourse by the bank to actions such as liquidating collateral…”
(one of the ANPR default-recognition conditions).
Further, as the ANPR states, for consumer exposures, “…default on one
obligation would not require a banking organization to treat all other
obligations of the same obligor as defaulted.” This distinction is
important, since it will avoid creating artificially high PDs and
similarly low LGDs for specific products and segments. Consistent with
this interpretation, for specific commercial segments, defaults should
be defined at the facility rather than at the obligor level.
In the consumer-credit segment, one of the triggers for recognizing
default is “…a distressed restructuring or workout involving forbearance
and loan modification;” our behavioral experience demonstrates that a
“re-aging” (for open-ended accounts such as credit cards) or
restructuring (for closed-end accounts such as mortgages) as defined in
detail within the FFIEC guidelines, are very successful in reducing
default rates. Should the ANPR definition of default remain as stated,
core institutions may loose incentives to restructure debt now deemed to
be in default.
With respect to undrawn exposures, both commercial and consumer,
general banks will hold less capital than core banks. Specifically,
under existing rules, any undrawn consumer-credit commitments that are
immediately cancelable (such as credit cards), attract no regulatory
capital; similarly, any undrawn commercial-credit commitments that
mature in less than one year attract no regulatory capital. Under the
proposed rules, core institutions will have to hold capital against the
full amount of any undrawn lines. At a minimum, we urge the regulators
to upgrade the existing regulatory-capital guidelines, to create a
consistent framework for the treatment of all undrawn lines, between
general and core institutions.
Establishing floors on specific parameters is arbitrary, excessively
conservative, and contrary to the intent of enhancing the risk
sensitivity of regulatory capital. Specifically, the PD floor of 3 basis
points penalizes high-quality portfolios, particularly in the
consumer-credit area. A significant portion of our mortgage and
home-equity portfolio consists of low-LTV/high-FICO product, with an
empirically-robust PD estimate of less than 3 basis points. Similarly,
the 10% LGD floor for residential mortgages, across all risk segments
and without regard to private mortgage insurance (PMI) is unnecessarily
conservative. For example, for a 60% LTV mortgage, the LGD floor implies
a liquidation loss of 50% in the event of a default on the underlying
loan. Finally, the minimum asset-value correlation (AVC) of 15% for
residential mortgages is artificially high and significantly above our
internal estimates.
To more precisely reflect credit risk, these parameters should be
calculated using an internal models-based approach, and where
appropriate, reflect a sensitivity to variables such as loan-to-value
(LTV) ratios and FICO scores. Additionally, these floors are contrary to
the Accord’s stated objective of improving the risk sensitivity of
regulatory capital. For example, the implied decline in regulatory
capital requirements for the credit risk of prime mortgages is
empirically supported by proven predictive variables (LTV, FICO, and
behavioral scores), significant data samples, and experience.
We disagree with the proposed rule that material losses on loan sales
should be treated as default. Such losses should only be recognized as a
default if they are charged against the loan-loss reserve. It is not
clear what constitutes “material” in this context. Further, the market
price of any asset reflects numerous conditions, in addition to
idiosyncratic credit risk. Also reflected are interest-rate risk,
liquidity risk, event risk, sector risk, supply risk, etc. The ANPR
assumes that idiosyncratic credit risk can be isolated in this context,
and this assumption is incorrect. For example, for any fixed-rate
instrument, an increase in market interest rates will create
depreciation in that instrument’s price; similarly, for a floating-rate
instrument, an increase in market spreads will create depreciation in
that instrument’s price. What is not clear is the attribution of these
interest-rate and spread changes across the risk spectrum identified
above. Hence, given that all or any portion of market-sale losses cannot
be consistently and reliably attributed to the idiosyncratic credit risk
of the underlying obligor, requiring that any such losses above some
arbitrary threshold be considered default is neither practical nor
logical.
Further, utilizing the ANPR’s definition of commercial default (i.e.,
(1)”…the banking organization determines that the borrower is unlikely
to pay its obligations to the organization in full…; or (2) the borrower
is more than 90 days past due on principal or interest on any material
obligation to the organization”), would market-sale losses above some
threshold automatically require default recognition across all
obligations of the underlying borrower, regardless of whether that
borrower was current on these other obligations?
The combination of scarce historical data for specific specialized
lending (“SL”) segments, such as commercial real estate (“CRE”), and the
resulting capital requirements under the Supervisory Slotting Criteria
(“SSC”) approach may represent a significant competitive disadvantage
for core institutions. Specifically, the SSC approach will result in
significantly higher regulatory-capital requirements relative to the
existing regulatory-capital framework. Hence, core institutions will be
required to hold substantially more regulatory capital in these segments
relative to general institutions when historical experience does not
support this increase. Further, lending activities for these segments
may migrate to non-regulated institutions, since these institutions will
be able to support more aggressive pricing, driven by lower capital
requirements.
The definition of and capital requirements for residual-value risk
are unclear and potentially excessively conservative. When defining the
treatment of residual-value risk for lease exposures, it is not clear
whether the agencies are referring to actual residual book value, or the
variation between the residual book value and the residual’s fair market
value. Regardless of which definition is intended, we urge the Agencies
to reconsider a seemingly arbitrary assignment of 100% that appears to
have no supporting empirical evidence. This approach gives no
recognition to the durability and market value of the physical assets
nor the ability of the lessor to extract value from the assets — a core
competency of any experienced lessor. The Agencies should consider an
approach that reflects actual historical experience, in terms of the
realization of residual values. The experience at Fleet has been
exceptionally successful, and it is important to ensure that capital
allocation methodologies reflect the true, economic risks.
Given the inclusion of expected losses in the calculation of required
capital, we believe that the entire amount of ALLL should be treated as
regulatory capital. In our view, the loan loss reserve covers both EL
and UL, in part determined by estimated loan losses over the next year
(EL), and the remainder determined by external constituents like rating
agencies, investors, etc. (UL). Because the ALLL in total represents the
first line of defense against losses, we further believe that this
amount should be included in Tier-1 capital.
While we support the Ratings-Based Approach (“RBA”) for
securitizations, the application of this methodology is not clear. For
example, in Table 1 on page 45935, column 2 refers to “Thick tranches
backed by highly granular pools”, but neither the term “thick” nor any
threshold thereof is addressed in the context of the RBA. Further, the
effective number of exposures (N) should be defined as the actual number
of underlying loans rather than the number of separate exposures in the
pool. For example, in a securitization involving two mortgage-backed
securities (“MBSs”), each with 1,000 underlying loans, N should be
defined as 2,000; the ANPR would appear to calculate N as 2 for this
example, which is clearly inconsistent with both the nature of the
economic exposures and market convention.
The capital required for securitization liquidity facilities is
significantly overstated. Liquidity facilities that can be drawn only in
the event of a general market disruption are assigned a 20% credit
conversion factor (CCF); all other facilities are assigned a 100% CCF.
If a liquidity facility can be drawn only in the event of a general
market disruption, the risk created is necessarily liquidity not
credit;
hence, by implication, the CCF should be close to or equal to zero.
Similarly, liquidity facilities that can be drawn without a general
market disruption may nonetheless represent significantly less credit
risk than an outright exposure to the underlying assets; as such, these
facilities should be assigned a CCF that is significantly less than
100%.
Similarly, if it is not possible to calculate the “Kirb capital
amount” using either the top-down or bottom-up approach, a
“Look-Through” approach is defined. Under this methodology, facilities
maturing in one year or less are assigned a 50% CCF, while facilities
with a maturity of longer than one year are assigned a 100% CCF. If the
facilities do represent liquidity, rather than credit support, these
factors overstate the risk exposures and the capital requirement.
We do not support the proposed rule that all ‘non-retail’ revolving
securitizations be considered 100% on balance sheet, as if they had
never been securitized. This approach is inconsistent with the ANPR’s
intent, as articulated on page 45932, which emphasizes that “…both the
designation of positions as securitization exposures and the calculation
of A-IRB capital requirements for securitization exposures would be
guided by the economic substance of a given transaction, rather than by
its legal form.” There is no economic justification for creating such
disincentives for non-retail, revolving securitization exposures, which
legitimately transfer significant credit risk.
Securitizations of committed or non-retail facilities that utilize a
controlled-amortization feature are arbitrarily assigned a CCF of 90
percent (the CCF is 100 percent if these facilities utilize a
non-controlled amortization feature). We strongly disagree with this
treatment. FleetBoston has successfully managed a revolving
commercial-loan securitization program for 5 years, and we believe that
there is no justification for the asymmetric treatment of
securitized-retail and securitized-committed/non-retail revolving
facilities. This treatment arbitrarily penalizes committed and
non-retail revolving facilities, and is contrary to the ANPR’s stated
recognition criteria “…which are intended to ensure that securitization
transactions transfer significant credit risk…”.
The proposed treatment of securitization arbitrarily penalizes
issuers relative to investors. While we favor the RBA for
externally-rated exposures, we are opposed to using a different
implementation standard for issuers and investors. Specifically, for an
identical, rated position that is below Kirb, investors are permitted to
apply the RBA; issuers are required to deduct these exposures from
capital. We are mindful of concerns about the validity of ratings for
non-traded positions, but would argue that the rating standards for
non-offered notes are the same as those for offered notes; thus, their
capital treatment should be the same.
Use of internal bank models for equity exposures is appropriate. In
general, the approach for determining equity exposure capital seems
reasonable. Use of a bank’s own internal economic models, with regulator
review and approval, is our preferred approach for any and all
regulatory capital determinations. The materiality threshold that
determines whether or not the advanced approach is required seems
somewhat arbitrary, and we would like to better understand that
rationale. Exclusion of nationally-legislated programs including CDCs
and CEDEs, seems appropriate. We would add that fund investments should
be treated as a single investment as the “look trough” approach would be
impractical to implement. And finally, the definition of “a return
directly linked to equity” needs clarification.
VAR-based methods for credit-risk capital of repos are appropriate
and their use should be expanded. Use of a VAR approach for determining
regulatory capital requirements for the counterparty credit risk
inherent in repos and securities lending is an industry “best practice,”
and one we fully endorse. We would recommend that this approach be
expanded to include all derivative transactions residing in the trading
book. This would substantially improve upon the less-accurate,
factor-based methodology employed in the current market-risk capital
rules.
Capital benefits from use of credit derivatives seem unnecessarily
conservative. The benefits of risk-mitigation are substantially reduced
if the ANPR is adopted as written. We feel this treatment would not only
be an incorrect portrayal of the economics but would also provide
disincentives to banks to manage their credit risk with these important
tools. First, there is the issue of “under recognition” of the benefits
of “double-default.” As with guarantees, because the risk of loss
requires the occurrence of a double-default event, the risk of loss is
materially smaller than provided for in the ANPR. We ask that this
treatment be changed in the final rules.
Second, the impact of credit derivatives should be consistent with
the A-IRB methodology applied to the underlying loan exposures. A
potential and simple solution would be to run the underlying credit
instrument of the credit derivative (or guarantee, for that matter)
through the ANPR’s corporate risk-weight function. This means that the
risk reduction would be computed using the same PD, LGD, and EAD as the
exposure being protected with the potential for a change in the maturity
of the protection. The sign of the result, of course, would be negative.
The maturity could be computed in the same manner as the exposure
itself, eliminating any concern for amortizing versus bullet. Also, the
maturity would be capped at the maturity of the exposure being hedged.
The credit risk of the counterparty providing the protection would be
determined separately as outlined in the ANPR.
Finally, FleetBoston makes use of credit derivatives (largely, credit
default swaps) to hedge credit risks in its banking book. In our case,
the derivatives are marked-to-market with the results flowing through
earnings into equity (Tier 1 regulatory capital) while the hedged items
remain on an accrual basis. To align the accounting results with the
underlying economics, any gain or loss for credit derivatives used to
hedge the credit portfolio should be removed from Tier 1 capital.
Conclusion
We commend the Agencies for their efforts to develop a more
risk-sensitive regulatory capital framework and trust that they will
continue to take into consideration the many positive suggestions
offered by the industry as they craft the final version of the capital
rules. While the proposal represents a significant improvement over the
existing rules, we believe they do not go far enough. The banking
industry will willingly work with regulators to arrive at a set of
economically robust capital rules that will provide the appropriate
economic incentives to develop a more safe, sound, and competitive
banking system.
FleetBoston is prepared to provide further input to the Agencies’
deliberations on this topic. Please contact Thomas Loeffler
(617-434-7501 or thomas_h_loeffler@fleet.com) or William Schomburg
(617-434-6158 or william_h_schomburg_iii@fleet.com) with further
questions or comments.
Sincerely,
Robert C. Lamb, Jr.
Executive Vice President
and Chief Financial Officer
Copied to: Mr. Eric S. Rosengren
Senior Vice President
Federal Reserve Bank of Boston
600 Atlantic Avenue
Boston, MA 02106
Mr. Douglas W. Roeder
Senior Deputy Comptroller, Large Bank Supervision
Office of the Comptroller of the Currency
250 E Street, S.W.
Washington, DC 20019
Mr. Jack Hall
Examiner in Charge
Office of the Comptroller of the Currency
c/o FleetBoston Financial Corporation
Mail Stop MA DE 10304N
100 Federal Street
Boston, MA 02110
Mr. Timothy MacDonald
Directing Examiner
Federal Reserve Bank of Boston
c/o FleetBoston Financial Corporation
Mail Stop MA DE 10304N
100 Federal Street
Boston, MA 02110
________________________________
JP Morgan Chase 2002 Annual Report, page 41.
Annex 6, page 199, Consultative Document — The New Basel Capital
Accord, April 2003
|
