32 Using Credit Risk Models for Regulatory Capital
A number of issues not discussed in this article would have to
be addressed before an internal-models-based (IM) approach
to regulatory capital for credit risk could be implemented.
These issues include:
• Loan loss reserves and expected loss: A capital charge
based on unexpected losses raises important issues
concerning the role and definition of loan loss reserves.
Recall that unexpected losses equal losses at the target
percentile minus expected losses. Therefore, if loan loss
reserves fall short of expected losses, the total resources
available to absorb losses—reserves plus capital—will
not be sufficient to provide protection at the desired
soundness standard. Unfortunately, there is no necessary
correspondence between the accounting definition of
loan loss reserves and the concept of expected losses
from a credit risk measurement model. Thus, over the
longer run, basing a regulatory capital charge on
unexpected losses may require a rethinking of the
treatment of loan loss reserves.
• Eligible institutions: The set of institutions subject to an
IM capital requirement will most likely be defined by the
minimum standards that are developed. Initially, only a
small set of banks would likely have models that were
sufficiently well-developed; many banks currently
employ default mode models and few, if any, fully
capture the correlation between risk drivers such as the
potential correlation between defaults and recovery
rates. Over time, however, the set of institutions with
comprehensive credit risk models is likely to grow as
modeling expertise disseminates through the industry,
as data sources become more readily available, and as the
competitive incentives for institutions to manage their
credit risk exposures in a more active way intensify.
• Scope of application: An important issue is whether an
IM capital requirement could be designed to cover all of
a bank’s credit exposures, or only those in selected
portfolios (for instance, large commercial loans). The
models discussed in this article are applied primarily to
commercial lending portfolios, while other portfolios—
such as retail lending—are either covered by models
whose structures are very different or, occasionally, not
covered at all. In this situation, it might make sense to
allow banks to apply an IM capital requirement only to
those portfolios covered by comprehensive credit risk
models of the type described here and to use a non-
models-based regulatory capital requirement for other
portfolios. However, “cherry picking,” or selective
adoption, is a clear concern if banks are allowed to use
internal models to determine capital charges for some,
but not all, of their exposures. That is, a bank may have
an incentive to model only those portions of its portfolio
in which capital charges are reduced.
• Scaling factor: The IM capital requirement for market
risk incorporates a multiplicative scaling factor that is
intended to translate value-at-risk estimates into an
appropriate minimum capital requirement, reflecting
considerations both about the accuracy of a bank’s
value-at-risk model and about prudent capital coverage.
There could be a similar role for a scaling factor in an IM
credit risk capital regime. For instance, given
shortcomings in data availability, uncertainty
surrounding the calibration of credit risk model
parameters (so-called model uncertainty) is a significant
concern in using these models for regulatory capital
purposes. More generally, supervisors and banks lack
long-term experience with credit risk models, a fact that
creates uncertainty about how the models will perform
over future credit cycles and during times of financial
market distress. These concerns could be addressed—
albeit roughly—by scaling up the raw loss figures
reported by the banks. In this instance, a scaling factor
might be incorporated when an IM approach is initially
implemented, and then revisited as both supervisors and
banks gain experience with the IM regime.
• Frequency of capital calculations: Prudential standards
would have to specify how frequently banks would be
required to run their credit risk models and report the
results to supervisors. Unlike value-at-risk models,
which are run on a daily basis to assess the market risk in
banks’ trading activities, credit risk models are run less
frequently. Monthly runs of the model—where a “run”
of the model means a new estimate of the PDF of future
losses incorporating changes in portfolio composition,
credit ratings, market prices, and parameter updates,
where warranted—seem a reasonable minimum
standard in the near term, though over the longer run,
banks would probably be expected to develop the
capability to generate fresh model estimates on an even
more frequent basis (perhaps weekly or biweekly).
Given frequent model results, capital could be based
on the average of monthly or weekly estimates during the
quarter. Using an average should mitigate banks’
incentives to window dress, as might be the case if the
capital charge were based on model outputs as of a single
point in time, such as quarter-end. In addition, averaging
should smooth short-run volatility in the model
Appendix: Practical Implementation Issues