Barclays 2007 Annual Report Download - page 93

1
Business review
Barclays PLC Annual Report 2007 91
Measurement, reporting and internal ratings
The principal objective of credit risk measurement is to produce the
most accurate possible quantitative assessment of the credit risk to
which the Group is exposed, from the level of individual facilities up
to the total portfolio.
The key building blocks in this quantitative assessment are:
– Probability of default (PD)
– Exposure in the event of default (EAD)
– Severity of loss given default (LGD)
Barclays first began to use internal estimates of PD (internal ratings) in all
its main businesses in the 1990s. Internally derived estimates for PD, EAD
and LGD have been used since then in all our major risk decision making
processes, enabling the application of coherent risk measurement across
all credit exposures, retail and wholesale.
With the advent of the Basel II accord on banking, Barclays has been
given permission to use internal rating models as an input in its regulatory
capital calculations. In preparation, Barclays has spent considerable time
developing and upgrading a number of such models across the Group,
moving towards compliance with the Basel II advanced internal ratings
based approach. As part of this process, all Basel credit risk models are
assessed against the Basel II minimum requirements prior to model
sign-off to ensure that they are fit to be used for regulatory purposes.
Applications of internal ratings
The three components described above – the probability of default,
exposure at default and loss given default – are building blocks used in a
variety of applications that measure credit risk across the entire portfolio.
Two examples are Risk Tendency (RT) and Expected Loss (EL) which
are statistical estimates of the average loss for the loan portfolio for
a 12-month period, taking into account the portfolio’s size and risk
characteristics under either current credit conditions (RT) or average
credit conditions (EL). As such, RT uses a point-in-time PD while EL uses
a through-the-cycle PD but the basic calculation is the same for both:
PD x EAD x LGD
Since through-the-cycle PDs provide a measure of risk that is independent
of the current credit conditions for a particular customer type, they are
more stable than point-in-time ratings. RT and EL provide insight into the
credit quality of the portfolio and assist management in tracking risk
changes as the Group’s stock of credit exposures evolves in size or risk
profile in the course of business.
As our understanding and experience have developed, we have extended
the use and sophistication of internal ratings. The other main business
processes that use internal estimates of PD, LGD and EAD, are as follows:
– Credit Grading – originally introduced in the early 1990s to provide a
common measure of risk across the Group using an eight point rating
scale; wholesale credit grading now employs a 21 point scale (Barclays
Masterscale).
– Credit Approval – a rating scale is used to set differentiated credit
sanctioning levels based upon a PD, so that credit risks are reviewed at
appropriate levels.
– Risk Appetite – measures of expected loss and the potential volatility
of loss are used in the Group’s Risk Appetite framework (see page 86).
– Pricing – within the corporate mass market portfolios we first
developed and used risk adjusted pricing models in the early 1990s
to differentiate risk reward decisions.
– IAS Impairment calculations – many of our collective impairment
estimates incorporate the use of our PD and LGD models.
– Economic capital (EC) allocation – most EC calculations use the same
through-the-cycle PD and EAD inputs as the regulatory capital (RC)
process. The process also uses the same underlying LGD model
outputs as the RC calculation, but does not incorporate the economic
downturn adjustment used in RC calculations.
– Risk management information – Group and the main business units
have for several years received either Key Information Packs or other
risk reports focused on EL and EC information to inform senior
management on issues such as the business performance, Risk
Appetite and consumption of EC.
Calculation of internal ratings
To calculate probability of default (PD), Barclays assesses the credit
quality of borrowers and other counterparties and assigns them an
internal risk rating.
Multiple rating methodologies may be used to inform the rating decision
on individual large credits, such as internal and external models, rating
agency ratings, and for wholesale assets market information such as credit
spreads. For smaller credits, a single source may suffice such as the result
from an internal rating model.
Barclays recognises the need for two different expressions of PD
depending on the purpose for which it is used. For the purposes of
calculating regulatory and economic capital, long-run average through-
the-cycle PDs are required. However, for the purposes of pricing and risk
tendency, PDs should represent the best estimate of probability of default,
typically in the next 12 months, dependent on the current position in the
credit cycle. Hence, point-in-time PDs are also required.
When each PD model is constructed, its output is specified as one of
point-in-time (PIT) or through-the-cycle (TTC) or a hybrid, e.g. a 50:50
blend. Using this distinction between PIT and TTC, the PDs are bucketed
into both PIT Default Grades (DGs) and TTC bands, adopting techniques
that are relevant to the model’s initial output calibration, the industry and
location of the counterparty and an understanding of the current and
long-term credit conditions. Two grades are therefore recorded for each
client, the DG and the TTC band. A customer may therefore be rated DG 6
reflecting sectoral performance and TTC band 8 reflecting long-term
credit conditions.
This same PIT/TTC distinction is applied to agency ratings. Within
Barclays, an agency alphabet rating is also expressed in terms of PIT DG
and TTC band. It is therefore no longer possible to produce a static
mapping of agency letter ratings to either DGs or TTC bands because they
are considered a hybrid of both PIT and TTC. As such, any mappings would
change over time with movements in the credit cycle.
Barclays internal rating system also differentiates between corporate and
retail customers.
For corporate portfolios (primarily Barclays Capital, BCB and the
commercial areas of IRCB), the rating system is constructed to ensure
that each client receives the same rating independent of the part of the
business with which they are dealing. To achieve this, a model hierarchy is
adopted which requires users to adopt a specific approach to rating each
counterparty depending upon the nature of the business and its location.
A range of methods is approved for estimating counterparty PDs. These
include bespoke grading models developed within the Barclays Group
(Internal Models), vendor models such as MKMV Credit Edge and RiskCalc,
and a conversion of external alphabet ratings from either S&P, Moody’s
or Fitch.