Basel
ii in the United States of America
From the
Basel ii
Compliance Professionals Association (BCPA),
the largest association of Basel ii Professionals in the
world
Final Rule, USA: Risk-Based Capital Standards:
Advanced Capital Adequacy Framework — Basel II
Conceptual Overview
1. The
IRB approach for credit risk
The
conceptual foundation of this final rule’s approach to
credit risk capital requirements is similar to the market
risk rule’s approach to market risk capital requirements,
in the sense that each is
VaR-oriented.
Nevertheless, there are important differences between the
IRB approach and the market risk rule.
The
current
market risk rule specifies a nominal confidence
level of 99.0 percent and a ten-business-day horizon, but
otherwise provides banks with
substantial modeling
flexibility
in determining their market risk loss
distribution and capital requirements.
In
contrast, the IRB approach for assessing credit risk
capital requirements is based on a 99.9 percent nominal
confidence level, a one-year horizon, and a supervisory
model of credit losses embodying particular
assumptions
about the underlying drivers of portfolio credit risk,
including loss correlations among different asset types.
The IRB
approach is broadly similar to the credit VaR approaches
used by a number of banks as the basis for their internal
assessment of the economic capital necessary to cover
credit risk.
It is
common for a bank’s internal credit risk models to
consider a one-year loss horizon and to focus on a high
loss threshold confidence level.
As with
the internal credit VaR models used by banks, the output
of the risk-based capital formulas in the IRB approach is
an estimate of the amount of credit losses above ECL over
a one-year horizon that would only be exceeded a small
percentage of the time.
The
agencies believe that a one-year horizon is appropriate
because it balances the difficulty of easily or rapidly
exiting non-trading positions against the possibility that
in many cases a bank can cover credit losses by raising
additional capital should the underlying credit problems
manifest themselves gradually.
The
nominal confidence level of the IRB riskbased capital
formulas (99.9 percent) means that if all the assumptions
in the IRB supervisory model for credit risk were correct
for a bank, there would be less than a 0.1 percent
probability that credit losses at the bank in any year
would exceed the IRB riskbased
capital
requirement.
As noted
above, the supervisory model of credit risk underlying the
IRB approach embodies specific assumptions about the
economic drivers of portfolio credit risk at banks.
As with
any modeling approach, these assumptions represent
simplifications of very complex real-world phenomena and,
at best, are only an approximation of the actual credit
risks at any bank.
If these
assumptions (described in greater detail below) are
incorrect or otherwise do not characterize a given bank
precisely, the actual confidence level implied by the IRB
risk-based capital formulas may exceed or fall short of a
true 99.9 percent confidence level.
In
combination with other supervisory assumptions and
parameters underlying the IRB approach, the approach’s
99.9 percent nominal confidence level reflects a
judgmental pooling of available information, including
supervisory experience.
The
framework underlying this final rule reflects a desire on
the part of the agencies to achieve
(i)
risk-based capital requirements that are reflective of
relative risk across different assets and that are broadly
consistent with maintaining at least an investment grade
rating (for example, at least BBB) on the liabilities
funding those assets, even in periods of economic
adversity; and
(ii) for
the U.S. banking system as a whole, aggregate minimum
regulatory capital requirements that are not a material
reduction from the aggregate minimum regulatory capital
requirements under the general risk-based capital rules.
A number of important explicit general
assumptions and specific parameters are built into the IRB approach to make the
framework applicable to a range of banks and to obtain
tractable information for calculating risk-based capital
requirements.
Chief
among the assumptions embodied in the IRB
approach are:
(i) assumptions that a bank’s credit
portfolio is infinitely granular;
(ii)
assumptions that loan defaults at a bank are driven by a
single, systematic risk factor;
(iii)
assumptions that systematic and non-systematic risk
factors are log-normal random variables; and
(iv)
assumptions regarding correlations among credit losses on various types of
assets.
The specific risk-based capital formulas
in this final rule require the bank to estimate certain risk parameters for its
wholesale and retail exposures, which the bank may do using a variety of techniques.
These risk parameters are PD, LGD, exposure at default (EAD),
and, for wholesale exposures, effective remaining maturity
(M).
The proposed rule included an additional risk
parameter, ELGD.
As
discussed in section III.B.3. of the preamble,
the agencies
have eliminated the ELGD risk parameter from the final
rule.
The
risk-based capital formulas into which the estimated risk
parameters are inserted are simpler than the economic
capital methodologies typically employed by banks, which
often require complex computer simulations.
In
particular, an important property of the IRB risk-based capital
formulas is portfolio invariance.
That is, the risk based capital requirement for a particular
exposure generally does not depend on the
other exposures held by the bank.
Like the
general risk-based capital rules, the total credit risk capital requirement for a
bank’s wholesale and retail exposures is the sum of the credit risk capital requirements on
individual wholesale exposures and segments of retail exposures.
The IRB risk-based capital formulas
contain supervisory
asset value correlation (AVC)
factors, which have a significant
impact on the capital requirements generated by the formulas.
The AVC assigned to a given
portfolio of exposures is an estimate of the degree to which any unanticipated changes
in the financial conditions of the underlying obligors of the exposures are correlated
(that is, would likely move up and down together).
High
correlation of exposures in a period of economic downturn
conditions is an area of supervisory concern.
For a
portfolio of exposures having the same risk parameters, a larger AVC implies less
diversification within the portfolio, greater overall systematic risk, and, hence, a higher
risk-based capital requirement.
For example, a 15 percent AVC for a portfolio of residential
mortgage exposures would result in a lower
risk-based capital requirement than a 20
percent AVC and a higher risk-based capital requirement than a 10 percent AVC.
The AVCs that appear in the IRB risk-based
capital formulas for wholesale exposures decline with increasing PD; that
is, the IRB risk-based capital formulas generally imply that a group of low-PD
wholesale exposures are more correlated than a group of
high-PD wholesale exposures.
Thus,
under the rule, a low-PD wholesale exposure would have a higher relative
risk-based capital requirement than that implied by its PD were the AVC in the IRB risk-based
capital formulas for wholesale exposures fixed rather than a decreasing function of
PD.
The AVCs included in the IRB risk-based capital formulas for both wholesale and
retail exposures reflect a combination of supervisory judgment and empirical
evidence.
However,
the
historical data available for estimating correlations among retail
exposures, particularly for non-mortgage retail exposures, was more limited than was the
case with wholesale exposures.
As a result, supervisory judgment played a greater role. Moreover,
the flat 15 percent AVC for residential mortgage exposures is based
largely on supervisory experience with and analysis of traditional long-term,
fixed-rate mortgages.
Several commenters stated that the
proposed AVCs for wholesale exposures were too high in general, and a few claimed
that, in particular, the AVCs for multi-family residential real estate exposures should
be lower.
Other commenters suggested that the AVCs of wholesale exposures should be a
function of obligor size rather than PD.
Similarly, several commenters maintained
that the proposed AVCs for retail exposures were too high. Some of these commenters
suggested that the AVCs for qualifying revolving exposures (QREs), such as credit
cards, should be in the range of 1 to 2 percent, not 4 percent as proposed.
Similarly, some of those commenters opposed the proposed flat 15 percent AVC for
residential mortgage exposures; one commenter suggested that the agencies should
consider employing lower AVCs for home equity loans and lines of credit (HELOCs) to take
into account their shorter maturity relative to
traditional mortgage exposures.
However, most commenters recognized that
the proposed AVCs were consistent with those in the New Accord and
recommended that the agencies use the AVCs contained in the New Accord to avoid
international competitive inequity and unnecessary burden.
Several commenters suggested that
the agencies should reconsider the AVCs going forward, working with the BCBS.
The agencies agree with the prevailing
view of the commenters that using the AVCs in the New Accord alleviates a
potential source of international inconsistency and implementation burden.
The final rule
therefore maintains the proposed AVCs.
As the agencies gain more experience with the
advanced approaches, they may revisit the AVCs for wholesale exposures and retail
exposures, along with other calibration issues identified during the parallel run and
transitional floor periods (as described below) and make
changes to the rule as necessary.
The
agencies would address this issue working with the BCBS and other supervisory and
regulatory authorities, as appropriate.
Another important conceptual element of
the IRB approach concerns the treatment of ECL.
The IRB approach assumes
that reserves should cover ECL while capital should cover credit losses
exceeding ECL (that is, unexpected credit losses).
Accordingly, the final rule, consistent
with the proposal and the New Accord, removes ECL from the risk-weighted assets
calculation but requires a bank to compare its ECL to its eligible credit reserves (as defined
below).
If a bank’s ECL exceeds its eligible credit reserves, the bank must deduct the excess
ECL amount 50 percent from tier 1 capital and 50 percent from tier 2 capital.
If a
bank’s eligible credit reserves exceed its ECL, the bank may include the excess eligible
credit reserves amount in tier 2 capital, up to 0.6 percent of the bank’s credit risk-weighted
assets.
This treatment is intended to maintain a capital incentive to reserve
prudently and ensure that ECL over a one-year horizon is
covered either by reserves or capital.
This
treatment also recognizes that prudent reserving that considers probable
losses over the life of a loan may result in a bank holding reserves in excess of ECL
measured with a one-year horizon.
The BCBS calibrated the 0.6 percent limit on
inclusion of excess reserves in tier 2 capital to be approximately as restrictive as the existing cap on the
inclusion of allowance for loan and lease losses (ALLL) under the 1988 Accord,
based on data obtained in the BCBS’s Third Quantitative Impact Study (QIS-3).
In developing the New Accord, the BCBS
sought broadly to maintain the current overall level of minimum risk-based
capital requirements within the banking system.
Using data from QIS-3, the BCBS conducted
an analysis of the risk-based capital requirements that
would be generated under the New Accord.
Based on
this analysis, the BCBS concluded that a “scaling factor”
(multiplier) should apply to credit risk-weighted assets.
The BCBS, in the New Accord,
indicated that the best estimate of the scaling factor was 1.06.
In May 2006, the BCBS
decided to maintain the 1.06 scaling factor based on the results of a fourth
quantitative impact study (QIS-4) conducted in some jurisdictions, including the United
States, and a fifth quantitative impact study (QIS-5), not conducted in the United States.
The BCBS noted that national supervisory authorities will continue to monitor
capital requirements during implementation of the New Accord, and that the BCBS, in turn,
will monitor national experiences with the
framework.
The agencies generally agree with the BCBS
regarding calibration of the New Accord.
Therefore, consistent with the New Accord and the proposed
rule, the final rule contains a scaling factor of 1.06 for
credit-risk-weighted assets.
As the
agencies gain more experience with the advanced
approaches, the agencies will revisit the scaling
factor along with other calibration issues
identified during the parallel run and transitional floor periods (described below) and will
make changes to the rule as necessary, working with the BCBS and other supervisory and regulatory
authorities, as appropriate.
Return
to Table of Contents
Return to
Index
Read more
about our
Certified Basel
ii Professional (CBiiPro)
program
Read more
about our Certified Pillar 2 Expert
(CP2E)
program
Read more about our
Certified Pillar 3 Expert
(CP3E)
program
Read
more about our Certified
Stress Testing Expert (CSTE)
program
 | |
