Market risk measurement methodology

The Group uses a number of approaches to measure market risk in its trading and treasury portfolios. These approaches include:

(i) VaR

VaR is a technique that produces estimates of the potential negative change in the market value of a portfolio over a specified time horizon at given confidence levels. For internal risk management purposes, the Group's VaR assumes a time horizon of one trading day and a confidence level of 95%. The Group also calculates VaR at a confidence interval of 99% and a time horizon of ten trading days for the purposes of calculating trading book market risk capital.

The Group uses historical simulation models in computing VaR. This approach, in common with many other VaR models, assumes that risk factor changes observed in the past are a good estimate of those likely to occur in the future and is, therefore, limited by the relevance of the historical data used. The Group's method, however, does not make any assumption about the nature or type of underlying loss distribution. The Group typically uses the previous 500 trading days of market data.

The Group calculates both general market risk (i.e. the risk due to movement in general market benchmarks) and idiosyncratic market risk (i.e. the risk due to movements in the value of securities by reference to specific issuers) using its VaR models.

The Group's VaR should be interpreted in light of the limitations of the methodology used. These limitations include:

  • Historical data may not provide the best estimate of the joint distribution of risk factor changes in the future and may fail to capture the risk of possible extreme adverse market movements which have not occurred in the historical window used in the calculations.
  • VaR using a one-day time horizon does not fully capture the market risk of positions that cannot be liquidated or hedged within one day.
  • VaR using a 95% confidence level does not reflect the extent of potential losses beyond that percentile.

The Group largely computes the VaR of trading portfolios at the close of business and positions may change substantially during the course of the trading day. Further controls are in place to limit the Group's intra-day exposure, such as the calculation of the VaR for selected portfolios. These limitations and the nature of the VaR measure mean that the Group cannot guarantee that losses will not exceed the VaR amounts indicated. The Group undertakes stress testing to identify the potential for losses in excess of the VaR.

The VaR for the Group's trading portfolios segregated by type of market risk exposure, including idiosyncratic risk, is presented in the table below.

2007 2006
Excluding ABN AMRO Average
£m		 Period end
£m		 Maximum
£m		 Minimum
£m		 Average
£m		 Period end
£m		 Maximum
£m		 Minimum
£m
Interest rate 11.7 9.6 17.6 7.6 8.7 10.2 15.0 5.7
Credit spread 17.7 37.9 44.0 12.6 13.2 14.1 15.7 10.4
Currency 2.6 2.6 6.9 1.1 2.2 2.5 3.5 1.0
Equity 2.4 1.9 6.8 1.4 1.1 1.6 4.4 0.5
Commodity 0.2 0.1 1.6 – 0.2 – 1.1 –
Diversification (12.4) (12.8)
Total trading VaR 20.3 39.7 45.5 13.2 14.2 15.6 18.9 10.4
2007
Including ABN AMRO Average
£m		 Period end
£m		 Maximum
£m		 Minimum
£m		  
Interest rate 12.5 15.0 21.8 7.6  
Credit spread 18.8 41.9 45.2 12.6  
Currency 2.6 3.0 6.9 1.1  
Equity 5.4 14.0 22.0 1.4  
Commodity 0.2 0.5 1.6 –  
Diversification (28.7)  
Total trading VaR 21.6 45.7 50.1 13.2  

Backtesting

The Group undertakes a programme of daily backtesting, which compares the actual profit or loss realised in trading activity to the VaR estimation. The results of the backtesting process are one of the methods by which the Group monitors the ongoing suitability of its VaR model. Backtesting exceptions are those instances when a realised loss exceeds the predicted VaR. At the 99% confidence level, no more than one backtesting exception is expected every 100 trading days. The Group experienced three backtesting exceptions at the consolidated Group level during 2007.

The Group's trading activities are carried out principally by Global Banking & Markets. The chart below depicts the number of days on which Global Banking & Markets' trading income fell within stated ranges.

Depicts the number of days on which Global Banking & Markets' trading income fell within stated ranges.

(ii) Stress testing

Stress testing measures the impact of abnormal changes in market rates and prices on the fair value of the Group's trading portfolios. GEMC approves the high-level market risk stress test limit for the Group.

The Group calculates a range of market risk stress tests each day. The objective of stress testing is to identify the loss that the Group's current portfolio of trading book exposures would generate in plausible but adverse market events. The Group calculates historical stress tests and hypothetical stress tests.

Historical stress tests calculate the loss that would be generated if the market movements that occurred during a historical market event were to be repeated. Hypothetical stress tests calculate the loss that would be generated if a specific set of adverse market movements were to occur.

In addition to the Group-level consolidated market risk stress tests, stress testing is also undertaken at key trading strategy level. Additional stress tests are undertaken for those strategies where the associated market risks are not adequately captured by VaR.

Stress test exposures are discussed with senior management and are reported to GRC, GEMC and the Board. Breaches in the Group's market risk stress testing limit are reported to GRC, GEMC and the Board.

(iii) Position risk and sensitivity analyses

In addition to the VaR and stress testing measures discussed above, the Group calculates a wide range of sensitivity and position risk measures, for example interest rate ladders or option revaluation matrices. These measures provide valuable additional controls, often at individual desk or strategy level.