RBS 2012 Annual Report Download - page 164

162
Business review Risk and balance sheet management continued
Credit risk management framework continued
Credit risk measurement*
The Group uses credit risk models to support quantitative risk
assessments element within the credit approval process, ongoing credit
risk management, monitoring and reporting and portfolio analytics. Credit
risk models used by the Group may be divided into three categories, as
follows.
Probability of default/customer credit grade
These models assess whether a customer will be able to repay its
obligations over a one year period.
Wholesale models - As part of the credit assessment process, the Group
assigns each counterparty an internal credit grade based on its
probability of default (PD). The Group uses a number of credit grading
models which consider risk characteristics relevant to the customer.
Credit grading models utilise a combination of quantitative inputs, such as
recent financial performance and qualitative inputs such as management
performance or sector outlook. The Group uses a credit grade in many of
its risk management and measurement frameworks, including credit
sanctioning and managing single-name concentration risk.
Retail models - Each customer account is scored using models based on
the likelihood of default. Scorecards are statistically derived using
customer data; customers are given a score that reflects their probability
of default, and this score is used to support automated credit decision
making.
Exposure at default models
Exposure at default (EAD) models estimate the level of use of a credit
facility at the time of a borrower’s default, recognising that customers may
make more use of their existing credit facilities as they approach default.
For revolving and variable draw-down type products that are not fully
drawn, the EAD is higher than the current utilisation. This estimate of
exposure can be reduced with financial collateral provided by the obligor
or a netting agreement.
Models that measure counterparty credit risk exposure are used for
derivatives and other traded instruments, where the amount of credit risk
exposure may depend on one or more underlying market variables, such
as interest or foreign exchange rates. These models drive the Group’s
internal credit risk management activities.
Loss given default models
Loss given default (LGD) models estimate the amount that cannot be
recovered by the Group in the event of default. When estimating LGD,
the Group takes into account both borrower and facility characteristics, as
well as any security held or credit risk mitigation, such as credit protection
or insurance. The cost of collections and a time discount factor for the
delay in cash recovery are also incorporated.
Changes to wholesale credit risk models
The Group is updating its wholesale credit risk models, incorporating
more recent data and reflecting new regulatory requirements applicable
to wholesale internal ratings based (IRB) modelling. In 2012, the Group
implemented updates to certain models, such as those used in the
sovereign and financial institution asset classes; these updates affected
the risk measures in the Group’s disclosures. Further updates, primarily
of models used for the corporate asset class, are planned for 2013.
Updates to models have generally affected relatively low-risk segments of
the Group’s portfolio. For example, the changes stemming from the
introduction of updated probability of default models largely affected
assets bearing the equivalent of investment-grade ratings.
In anticipation of these changes, the Group modified various risk
frameworks, including its risk appetite framework and latent loss
assessment. In addition, with the agreement of its regulators, the Group
adjusted upwards the risk-weighted assets (RWAs) of some portfolios
prior to the introduction of the new models.
Model changes affect year-on-year comparisons of risk measures in
certain disclosures. Where meaningful, the Group in its commentary has
differentiated between instances where movements in risk measures
reflect the impact of model changes, and those that reflect movements in
the size of underlying credit portfolios or their credit quality. However, it is
not practicable to quantify the impact of model updates on individual
asset quality bands.
Separately, as agreed with the Financial Services Authority (FSA), the
Group has started to apply a slotting approach to calculate RWAs related
to commercial real estate assets; this approach does not use modelled
measures to determine RWAs and capital requirements.
*unaudited