HSBC 2011 Annual Report Download - page 201

199
Overview Operating & Financial Review Corporate Governance Financial Statements Shareholder Information
We employ a range of tools to monitor and limit market risk exposures. These include sensitivity analysis, value at
risk (‘VAR’) and stress testing.
Sensitivity analysis
(Unaudited)
We use sensitivity measures to monitor the market risk positions within each risk type, for example, the present value
of a basis point movement in interest rates for interest rate risk. Sensitivity limits are set for portfolios, products and
risk types, with the depth of the market being one of the principal factors in determining the level of limits set.
Value at risk
(Audited)
VAR is a technique that estimates the potential losses that could occur on risk positions as a result of movements in
market rates and prices over a specified time horizon and to a given level of confidence.
The VAR models we use are based predominantly on historical simulation. These models derive plausible future
scenarios from past series of recorded market rates and prices, taking account of inter-relationships between different
markets and rates such as interest rates and foreign exchange rates. The models also incorporate the effect of option
features on the underlying exposures.
Our historical simulation models assess potential market movements with reference to data from the past two years
and calculate VAR to a 99% confidence level and for a one-day holding period.
We routinely validate the accuracy of our VAR models by back-testing the actual daily profit and loss results,
adjusted to remove non-modelled items such as fees and commissions, against the corresponding VAR numbers.
Statistically, we would expect to see losses in excess of VAR only 1% of the time over a one-year period. The actual
number of excesses over this period can therefore be used to gauge how well the models are performing.
Although a valuable guide to risk, VAR should always be viewed in the context of its limitations. For example:
• the use of historical data as a proxy for estimating future events may not encompass all potential events,
particularly those which are extreme in nature;
• the use of a one-day holding period assumes that all positions can be liquidated or the risks offset in one day.
This may not fully reflect the market risk arising at times of severe illiquidity, when a one-day holding period
may be insufficient to liquidate or hedge all positions fully;
• the use of a 99% confidence level, by definition, does not take into account losses that might occur beyond this
level of confidence;
• VAR is calculated on the basis of exposures outstanding at the close of business and therefore does not
necessarily reflect intra-day exposures; and
• VAR is unlikely to reflect loss potential on exposures that only arise under significant market moves.
Stress testing
(Audited)
In recognition of the limitations of VAR, we augment it with stress testing to evaluate the potential impact on
portfolio values of more extreme, although plausible, events or movements in a set of financial variables.
The process is governed by the Stress Testing Review Group forum which, in conjunction with regional risk
managers, determines the scenarios to be applied at portfolio and consolidated levels, as follows:
• sensitivity scenarios consider the impact of any single risk factor or set of factors that are unlikely to be captured
within the VAR models, such as the break of a currency peg;
• technical scenarios consider the largest move in each risk factor without consideration of any underlying market
correlation;
• hypothetical scenarios consider potential macro economic events, for example, a global flu pandemic; and
• historical scenarios incorporate historical observations of market movements during previous periods of stress
which would not be captured within VAR.