Goldman Sachs 2015 Annual Report Download - page 111

THE GOLDMAN SACHS GROUP, INC. AND SUBSIDIARIES
Management’s Discussion and Analysis
Market Risk Management Process
We manage our market risk by diversifying exposures,
controlling position sizes and establishing economic hedges
in related securities or derivatives. This process includes:
‰Accurate and timely exposure information incorporating
multiple risk metrics;
‰A dynamic limit setting framework; and
‰Constant communication among revenue-producing
units, risk managers and senior management.
Risk Measures. Market Risk Management produces risk
measures and monitors them against market risk limits set
by our risk committees. These measures reflect an extensive
range of scenarios and the results are aggregated at product,
business and firmwide levels.
We use a variety of risk measures to estimate the size of
potential losses for both moderate and more extreme
market moves over both short-term and long-term time
horizons. Our primary risk measures are VaR, which is
used for shorter-term periods, and stress tests. Our risk
reports detail key risks, drivers and changes for each desk
and business, and are distributed daily to senior
management of both our revenue-producing units and our
independent control and support functions.
Value-at-Risk. VaR is the potential loss in value due to
adverse market movements over a defined time horizon
with a specified confidence level. For assets and liabilities
included in VaR, see “Financial Statement Linkages to
Market Risk Measures.” We typically employ a one-day
time horizon with a 95% confidence level. We use a single
VaR model which captures risks including interest rates,
equity prices, currency rates and commodity prices. As
such, VaR facilitates comparison across portfolios of
different risk characteristics. VaR also captures the
diversification of aggregated risk at the firmwide level.
We are aware of the inherent limitations to VaR and
therefore use a variety of risk measures in our market risk
management process. Inherent limitations to VaR include:
‰VaR does not estimate potential losses over longer time
horizons where moves may be extreme;
‰VaR does not take account of the relative liquidity of
different risk positions; and
‰Previous moves in market risk factors may not produce
accurate predictions of all future market moves.
When calculating VaR, we use historical simulations with
full valuation of approximately 70,000 market factors.
VaR is calculated at a position level based on
simultaneously shocking the relevant market risk factors
for that position. We sample from five years of historical
data to generate the scenarios for our VaR calculation. The
historical data is weighted so that the relative importance of
the data reduces over time. This gives greater importance to
more recent observations and reflects current asset
volatilities, which improves the accuracy of our estimates of
potential loss. As a result, even if our positions included in
VaR were unchanged, our VaR would increase with
increasing market volatility and vice versa.
Given its reliance on historical data, VaR is most effective in
estimating risk exposures in markets in which there are no
sudden fundamental changes or shifts in market conditions.
Our VaR measure does not include:
‰Positions that are best measured and monitored using
sensitivity measures; and
‰The impact of changes in counterparty and our own
credit spreads on derivatives, as well as changes in our
own credit spreads on unsecured borrowings for which
the fair value option was elected.
We perform daily backtesting of our VaR model (i.e.,
comparing daily trading net revenues to the VaR measure
calculated as of the prior business day) at the firmwide level
and for each of our businesses and major regulated
subsidiaries.
Stress Testing. Stress testing is a method of determining
the effect of various hypothetical stress scenarios on the
firm. We use stress testing to examine risks of specific
portfolios as well as the potential impact of significant risk
exposures across the firm. We use a variety of stress testing
techniques to calculate the potential loss from a wide range
of market moves on our portfolios, including sensitivity
analysis, scenario analysis and firmwide stress tests. The
results of our various stress tests are analyzed together for
risk management purposes.
Sensitivity analysis is used to quantify the impact of a
market move in a single risk factor across all positions (e.g.,
equity prices or credit spreads) using a variety of defined
market shocks, ranging from those that could be expected
over a one-day time horizon up to those that could take
many months to occur. We also use sensitivity analysis to
quantify the impact of the default of any single entity,
which captures the risk of large or concentrated exposures.
Goldman Sachs 2015 Form 10-K 99