Академический Документы
Профессиональный Документы
Культура Документы
8 May 2007
www.fitchratings.com
Banks
A Proven Track Record nomic capital management framework of these or-
Models that have been consistently employed to as- ganisations. The growing popularity of these models
sess risk over a period of time, and where manage- is only expected to accelerate with the establishment
ment have a history of decision making and manag- of a new regulatory framework for assessing bank
ing their business using the output provided by the risk-based capital, better known as Basel II. Basel II
models will be assigned higher weights. Credit risk adapts many of the concepts and techniques used by
models that are in place but not integrated with the large financial institutions in their own internal credit
institution’s risk management and decision-making risk models for supervisory purposes.
framework are, in general, ineffective.
The agency believes that the anticipated convergence
Appropriate Control of the Model between economic capital and regulatory capital
As models can be powerful tools within a bank’s frameworks has been a positive impetus to further
business, model controls are important and are a key refine and more broadly spread advanced risk meas-
ingredient of Fitch’s assessment. Models should be urement and management practices such as credit
embedded within the business. Change control and portfolio risk modelling.
documentation standards should ensure model integ-
rity. The assessment of credit risk models and their output
represents an important part of Fitch’s overall view
Quality of Model of an institution’s risk appetite and profile within the
Models should be chosen that provide a strong fit agency’s rating analysis. A number of institutions
between their capabilities, assumptions and the prop- use standard vendor models available in the market
erties of the portfolio. Some contemporary credit risk to measure and manage credit risk. There is an in-
models employ a single-factor model framework creasing trend by many institutions to (1) attempt to
and/or assume an infinitely fine-grained portfolio. separate the conceptual model from its implementa-
They are appealing because of their simplicity. Such tion, (2) internally develop credit risk models built to
models may be appropriate for highly diverse portfo- meet their organisation’s unique requirements, and
lios, as found in retail banking or SME divisions, but (3) access greater diversification benefit through the
can only rarely be applied to corporate portfolios. joint analysis of multiple asset classes.
The tractability of credit risk models is lost when
This report identifies the principal drivers of credit
less finely-grained portfolios or portfolios which
risk models in the measurement of credit risk and the
intrinsically depend on a large number of disparate
influence of these drivers on the economic capital
market factors are to be modelled. Thus, it is essen-
charge, as well as the impact of these drivers if ap-
tial that a suitably sophisticated model be employed.
plicable in determining the capital charges under the
Models should be updated and shortcomings ad- new regulatory capital framework (Basel II – internal
dressed as the industry continues its enhancements. ratings-based approach). The report uses a sample
Appendix 1 provides a brief overview of common portfolio of loans as an illustration to compare the
credit risk model approaches. influence of these drivers in determining capital
charges under economic and regulatory capital
Validation frameworks. In addition, the report also outlines the
While validation processes and methodologies are agency’s opinion on the use of these credit risk mod-
the prime responsibility of the institutions them- els in the measurement of credit risk and the associ-
selves, Fitch will evaluate and assess the process and ated capital charge and how it will factor this into its
techniques deployed and attempt to identify best rating process.
practice amongst institutions.
Credit Risk Models and Risk Management
Ongoing Model Review Fitch believes that credit risk models form a very
Models should be subject to regular periodic audits, important part of risk management and are a vital
continuous testing and benchmarking. Related input into the pricing of products and services. The
change control procedures are also important to en- understanding of risk on an aggregate basis and an
sure that only reasonable changes are made without assessment of how these risks associated with single
sacrificing model integrity. borrowers interact with each other is very difficult to
achieve in the absence of some kind of risk model,
Introduction especially where risks are complex. It is also the
Fitch recognises the advances made by banks in risk agency’s opinion that while measuring risks is a ne-
management and in particular the increasing use of cessity, to achieve full economic benefit from the
credit portfolio risk models for the measurement and models, their output should be an influencing factor
management of credit risk as part of the overall eco- in decision making within an institution.
There are a number of commercial credit risk models 3. Comprehensiveness. It is important to the
in addition to those developed internally. To effec- assessment process to clearly identify the risks
tively employ the results in risk practice and other that are being considered and those that have
management activities, it is essential for banks to been ignored by the credit risk model. While
fully understand how these credit models function. there are many potential inclusions and exclu-
Further, it is important for the bank to be able to sions, two of the most commonly excluded are
provide sufficient detail for Fitch to be able to under- concentration risk and migration risk. Consider
stand, assess and contrast the particular credit risk as an example, concentration risk: Name con-
model being used. centration in loan portfolios arises from either a
dearth of borrowers or when loan amounts are
Assessing Quality of Credit Risk Models asymmetrically distributed. The credit portfolio
Two important components need to be addressed as risk model adopted for the Basel II Internal Rat-
part of the assessment process, and certainly before ings-Based (IRB) approach does not account for
credit risk models can be used to determine name concentration: it assumes that the bank’s
risk-based capital requirements. portfolio is perfectly fine-grained, meaning that
idiosyncratic risk has been fully diversified
1. Model input quality. The need to establish con- away, so that economic capital depends only on
sistent and accurate exposure measurement for systematic risk. As a result, each loan consti-
the chosen credit rating standard is very impor- tutes only a small fraction of the total portfolio
tant. Much work has been conducted regarding exposure. True bank loan portfolios are seldom
the difficulties involved in creating and – if ever – perfectly fine-grained. The asymp-
maintaining accurate internal ratings systems totic assumption, while approximately valid for
and although these present challenging issues, large portfolios, may be invalid for portfolios of
they can be addressed by internal and external smaller or more specialised institutions. Where
qualitative monitoring procedures. material name concentrations of exposure exist,
the residual of undiversified idiosyncratic risk in
2. Model specification and validation. The con- the portfolio can be significant. Fitch will expect
struction of the loss distribution function is pos- financial institutions to have a process in place
sibly the most challenging aspect of credit risk to identify, measure, manage and mitigate con-
modelling. Changes in credit value are due to centration risk.
several factors, such as correlations between
portfolio assets, movements of credit spreads Example
and changes in individual loans’ credit status.
As a result, a variety of modelling assumptions Application of Credit Risk Model
and parameter values are required for the con- Methodology to Loan Portfolios
struction of a credit risk model’s forecasted dis-
A simple Excel-based single-factor model was con-
tribution. The lack of sufficient historical per-
structed to examine how these primary drivers influ-
formance data, however, limits the testing of the
ence model output. The model was also flexed to test
validity of these model components. A relatively
sensitivity of model output to changes in values of
large number of observations are required to ac-
primary drivers, and to compare it to the regulatory
curately evaluate any distribution forecast. This
model.
issue is of less concern for evaluating distribu-
tion forecasts generated by market Value at Risk It is important to note that construction of a
(VaR) models since such models generate daily full-blown credit risk model is not the purpose of this
forecasts of market portfolio changes which can paper. The model employed here (referred to as the
then be evaluated relative to the actual portfolio “Basic Model”) therefore, makes many assumptions
changes. Whilst some 250 observations (one about the structure of credit portfolios. It is thus rela-
year of daily trading data) are currently used in tively simple, aiming to illustrate the principles on
the regulatory evaluation of market VaR models, which these models operate rather than focussing on
it is difficult to gather a large number of ob- the details.
served credit losses with which to evaluate
credit risk models since these have much longer The data used are hypothetical corporate loan portfo-
forecast horizons – typically one year. As a re- lios (which approximate real life bank loan portfo-
sult, one year generates a single observation of lios) and were characterised by the parameters (used
credit losses, so 250 years would be required to as input to the single-factor model) specified below.
For the base scenario, the results and capital charges 60,000,000
40,000,000
Summary of Measures – Base 30,000,000
Scenario 20,000,000
Total exposure EUR1,000,000,000
10,000,000
Expected loss EUR1,436,692
Unexpected loss @ 99/9% EUR53,355,699 0
Capital charge 5.34% 8 grades 15 grades 22 grades
Source: Fitch Source: Fitch
The Impact of Change in PD The UL is more sensitive than the EL to the number
of PD buckets and increases by 22% when PD buck-
ets are restricted to eight grades. Conversely, when
Different PD Bands: Basic Model EL
PD buckets increase to 22 grades UL decreases by
(EUR) 10%. When assessing lower-quality portfolios or
2,000,000 including migration analysis, the impact of rating
scale granularity must be explored in greater depth.
1,600,000
400,000 50,000,000
0 40,000,000
8 grades 15 grades 22 grades 30,000,000
Source: Fitch
20,000,000
10,000,000
The portfolio was tested at different threshold levels 0
for all PD bands. Higher average PDs increase the PD 20% up from PD at base level PD 20% down
UL considerably. This result is not unexpected: base from base
lower average PDs result in larger thresholds which Source: Fitch
must be breached before default (and hence a loss)
can occur. Since these thresholds will be more diffi- Increasing the average PD of the PD bands increases
cult to breach than those for higher average PDs, the UL by 7%. Decreasing the average PD of the PD
losses are reduced. bands decreases the UL by 17%. This result is sensi-
tive to the interaction between the overall average
If starting with more than 15, increasing the number
PD level (i.e., credit quality) and the confidence
of PD buckets used to allocate input data does not
level.
have a significant effect on the EL, and the values
converge to the base case EL. However, a decreasing
number of PD buckets does result in portfolio EL
increasing by 20%. Note that if the net exposure
weighting method were used here, the portfolio
would be mean invariant, so that the number of
buckets would have no effect on EL.
20,000,000 50,000,000
40,000,000
10,000,000
30,000,000
0
20,000,000
Basel threshold PDs Fitch threshold PDs
Source: Fitch 10,000,000
0
Correlations Basel calculated Correlations
The Impact of Changing LGD Source: Fitch 20% up correlations 20% down
Copyright © 2007 by Fitch, Inc., Fitch Ratings Ltd. and its subsidiaries. One State Street Plaza, NY, NY 10004.
Telephone: 1-800-753-4824, (212) 908-0500. Fax: (212) 480-4435. Reproduction or retransmission in whole or in part is prohibited except by permission. All rights reserved. All of the
information contained herein is based on information obtained from issuers, other obligors, underwriters, and other sources which Fitch believes to be reliable. Fitch does not audit or
verify the truth or accuracy of any such information. As a result, the information in this report is provided “as is” without any representation or warranty of any kind. A Fitch rating is an
opinion as to the creditworthiness of a security. The rating does not address the risk of loss due to risks other than credit risk, unless such risk is specifically mentioned. Fitch is not
engaged in the offer or sale of any security. A report providing a Fitch rating is neither a prospectus nor a substitute for the information assembled, verified and presented to investors by
the issuer and its agents in connection with the sale of the securities. Ratings may be changed, suspended, or withdrawn at anytime for any reason in the sole discretion of Fitch. Fitch does
not provide investment advice of any sort. Ratings are not a recommendation to buy, sell, or hold any security. Ratings do not comment on the adequacy of market price, the suitability of
any security for a particular investor, or the tax-exempt nature or taxability of payments made in respect to any security. Fitch receives fees from issuers, insurers, guarantors, other
obligors, and underwriters for rating securities. Such fees generally vary from US$1,000 to US$750,000 (or the applicable currency equivalent) per issue. In certain cases, Fitch will rate
all or a number of issues issued by a particular issuer, or insured or guaranteed by a particular insurer or guarantor, for a single annual fee. Such fees are expected to vary from US$10,000
to US$1,500,000 (or the applicable currency equivalent). The assignment, publication, or dissemination of a rating by Fitch shall not constitute a consent by Fitch to use its name as an
expert in connection with any registration statement filed under the United States securities laws, the Financial Services and Markets Act of 2000 of Great Britain, or the securities laws of
any particular jurisdiction. Due to the relative efficiency of electronic publishing and distribution, Fitch research may be available to electronic subscribers up to three days earlier than to
print subscribers.