ING Direct 2008 Annual Report Download - page 199

Backtesting
Backtesting is a technique for the ongoing monitoring of the plausibility of the VaR model in use. Although VaR models estimate
potential future results, estimates are based on historical market data. In a backtest, the actual daily result is compared with the 1-day
VaR. In addition to using actual results for backtesting, ING also uses hypothetical results, which measure results excluding the effect of
intraday trading, fees and commissions. When the actual or hypothetical loss exceeds the VaR an ‘occurrence’ has taken place. Based on
ING’s one-sided confidence level of 99% an occurrence is expected once in every 100 business days. In 2008, there was no occurrence
(2007: none) where a daily trading loss exceeded the daily consolidated VaR of ING Wholesale Banking. ING reports the results of this
backtesting to the Dutch Central Bank on a quarterly basis.
Stress testing
Stress tests are used for the monitoring of market risks under extreme market conditions. Since VaR in general does not produce an
estimate of the potential losses that can occur as a result of extreme market movements, ING uses structured stress tests for monitoring
the market risk under these extreme conditions. Stress scenarios are based on historical and hypothetical extreme events. The result of
the stress testing is an event risk number, which is an estimate of the profit and loss account effect caused by a potential event and its
world-wide impact for ING Wholesale Banking. The event risk number for the ING Wholesale Banking trading activity is generated on
a weekly basis. Like VaR, event risk is limited by ALCO Bank. The event-risk policy (and its technical implementation) is specific to ING as
there is no event risk calculation method that is generally accepted by other banks and regulators (like the Value-at-Risk model). ING’s
event risk policy basically consists of defined stress parameters per country and per market (fixed income, equity, foreign exchange, credit
and related derivative markets). The scenarios and stress parameters are back-tested against extreme market movements that actually
occurred in the markets. The market developments in 2008 will be taken into account in the definition of scenarios and stress parameters
during 2009. If and when necessary, ING evaluates specific stress scenarios, as an addition to its structured stress tests. These specific
scenarios relate to current concerns, like political instability in certain regions, terrorist attacks or extreme movements in energy prices.
Other trading controls
VaR and event risk limits are the most important limits to control the trading portfolios. Furthermore, ING uses a variety of other limits
to supplement VaR and event risk. Position and sensitivity limits are used to prevent large concentrations in specific issuers, sectors or
countries. In addition to this, other risk limits are set with respect to the activities in exotic derivatives trading. The market risk of these
products is controlled by product specific limits and constraints.
Development of trading market risks
The following chart shows the development of the overnight VaR under a 99% confidence interval and a 1-day horizon. The overnight
VaR is presented for the ING Wholesale Banking trading portfolio which was risk managed by CMRM Trading during 2007 and 2008.
Several banking books are governed by the trading risk process and are therefore excluded from the non-trading risk table and included
in the trading risk graph and table below.
Consolidated trading VaR ING Wholesale Banking 2007-2008
in EUR millions
Q1 2007
0
10
20
30
40
50
60
70
80
Q2 2007 Q3 2007 Q4 2007 Q1 2008 Q2 2008 Q3 2008 Q4 2008
VaR Limit
Date
Value at Risk
During 2007 and 2008 the overnight VaR for the ING Wholesale Banking trading portfolio ranged from EUR 29 to EUR 72 million.
197
ING Group Annual Report 2008