ING Direct 2013 Annual Report Download - page 271

Risk management continued ING Bank
Credit risk models
Within ING Bank internal Basel models are used to determine the PD, EAD and LGD for regulatory and economic capital. Bank wide, ING
Bank has implemented more than 100 models, including various sub models that may be applicable for a specific portfolio.
There are three types of modelling which form the foundation of these PD, EAD and LGD models used throughout the bank.
• Statistical models are created where a large set of default or detailed loss data is available. They are characterised by a sufficient
number of data points that facilitate meaningful statistical estimation of the model parameters. The model parameters are estimated
with statistical techniques based on the data set available.
• Expert models are based on the knowledge and experience of experts from both risk management and front office staff and
literature from rating agencies, supervisors and academics. These models are especially appropriate for ‘low default portfolios’, where
limited historical defaults exist; thereby reducing the reliability of a statistical model.
• Hybrid models contain characteristics of both expert and statistical models.
Next to the model choice, the definition of default is an important starting point for model building. ING Bank uses a framework that
integrates elements of the regulatory definition of ‘Default’ and the loan loss provisioning indicators under IAS 39. The rationale is that
several indicators are very close to the indications of an obligor’s ‘unlikeliness to pay’ under Basel II and similar regulations. Integration of
both frameworks makes it possible to use the regulatory risk components PD, LGD and EAD in the collective provisioning process under
IAS 39, further enhancing ING Bank’s compliance with the Basel II ‘use test’. Key differences between the parameters used for loan loss
provisioning and regulatory capital calculations are that Regulatory Capital parameters are typically through the cycle while Loan Loss
parameters tend to be more point in time. Additionally, the LGD for Regulatory Capital calculations is based on a down-turn LGD.
Pre-Settlement measurement models.
For regulatory capital the Pre-Settlement (PS) exposure is calculated using a Marked-to-Market (MtM) plus regulatory-based add-on tables.
For internal capital purposes ING Bank uses two methodologies to calculate its PS exposures. Ideally, all parts of ING Bank would apply
identical methodologies at all times. However, it is recognised that the ability to implement risk measurement methodologies is highly
dependent on systems capabilities, and in certain cases the benefits of implementing a methodology may not be justified by the costs.
Therefore more than one methodology is presently in use at ING Bank.
• MtM plus model based add-on approach: In this approach, the PS risk is calculated as the sum between the MtM of the trade and
the model-based Add-on. The MtM fluctuates through the life of the contract. The model-based Add-on is product-specific, and takes
into account remaining time to maturity, profiling per time-buckets etc. Add-ons are updated with a frequency that takes into account
the major market changes. This methodology is used for pre-deal exposure assessment of all ING Bank financial markets products and
for post-deal risk calculations for financial markets portfolios for which computational efforts and costs associated with implementation
of Scenario Simulation approach are not justifiable;
• Scenario Simulation approach (Monte Carlo approach): Scenario Simulation approach is the most complex of the methods for PS
risk calculations. This approach is the only approach that fully takes into account the daily market conditions, including correlations
between the risk factors and portfolio benefits. This approach is also referred to as Monte Carlo (MC) approach and is currently used
for the largest volume of derivative products such as FX and interest rate derivatives. ING Bank is in the process of implementing this
approach for more products. The monitoring of the PS exposures and the limit setting for the products within the scope of the MC
approach are based on the exposures resulting from the MC approach, the pre-deal check exposure assessment is based on the MtM
plus model add-on approach.
In addition to the two approaches ING recognises that certain trading products that are outside of this scope may be deemed insufficiently
accurate. For example, highly structured or exotic derivative transactions may differ significantly from the generic transactions used to
calculate the add-ons. For the assessment of risk exposures of such complex products a bespoke calculation is made.
The figure below provides a high level summary of the application of model outcomes (PD, EAD and LGD).
Loan Pricing
Loan Loss Provisioning
Regulatory Capital/RWA
Credit Scoring/Acceptance
Basel II Risk Components
PD EAD LGD Maturity
Economic Capital
Performance Measurement
269ING Group Annual Report 2013
1 Who we are 2 Report of the Executive Board 3 Corporate governance 4 Consolidated annual accounts 5 Parent company annual accounts 6 Other information 7 Additional information