JP Morgan Chase 2013 Annual Report Download - page 138

Management’s discussion and analysis
144 JPMorgan Chase & Co./2013 Annual Report
Value-at-risk
JPMorgan Chase utilizes VaR, a statistical risk measure, to
estimate the potential loss from adverse market moves in a
normal market environment consistent with the day-to-day
risk decisions made by the lines of business.
The Firm has one overarching VaR model framework, Risk
Management VaR, used for risk management purposes
across the Firm, which utilizes historical simulation based
on data for the previous 12 months. The framework’s
approach assumes that historical changes in market values
are representative of the distribution of potential outcomes
in the immediate future. The Firm believes the use of Risk
Management VaR provides a stable measure of VaR that
closely aligns to the day-to-day risk management decisions
made by the lines of business and provides necessary/
appropriate information to respond to risk events on a daily
basis.
Risk Management VaR is calculated assuming a one-day
holding period and an expected tail-loss methodology which
approximates a 95% confidence level. This means that,
assuming current changes in market values are consistent
with the historical changes used in the simulation, the Firm
would expect to incur VaR “band breaks,” defined as losses
greater than that predicted by VaR estimates, not more
than five times every 100 trading days. The number of VaR
band breaks observed can differ from the statistically
expected number of band breaks if the current level of
market volatility is materially different from the level of
market volatility during the twelve months of historical data
used in the VaR calculation.
Underlying the overall VaR model framework are individual
VaR models that simulate historical market returns for
individual products and/or risk factors. To capture material
market risks as part of the Firm’s risk management
framework, comprehensive VaR model calculations are
performed daily for businesses whose activities give rise to
market risk. These VaR models are granular and incorporate
numerous risk factors and inputs to simulate daily changes
in market values over the historical period; inputs are
selected based on the risk profile of each portfolio as
sensitivities and historical time series used to generate daily
market values may be different across product types or risk
management systems. The VaR model results across all
portfolios are aggregated at the Firm level.
Data sources used in VaR models may be the same as those
used for financial statement valuations. However, in cases
where market prices are not observable, or where proxies
are used in VaR historical time series, the sources may
differ. In addition, the daily market data used in VaR models
may be different than the independent third-party data
collected for VCG price testing in their monthly valuation
process (see pages 196–200 of this Annual Report for
further information on the Firm’s valuation process.) VaR
model calculations require more timely (i.e., daily) data and
a consistent source for valuation and therefore it is not
practical to use the data collected in the VCG monthly
valuation process.
VaR provides a consistent framework to measure risk
profiles and levels of diversification across product types
and is used for aggregating risks across businesses and
monitoring limits. These VaR results are reported to senior
management, the Board of Directors and regulators.
Since VaR is based on historical data, it is an imperfect
measure of market risk exposure and potential losses, and
it is not used to estimate the impact of stressed market
conditions or to manage any impact from potential stress
events. In addition, based on their reliance on available
historical data, limited time horizons, and other factors, VaR
measures are inherently limited in their ability to measure
certain risks and to predict losses, particularly those
associated with market illiquidity and sudden or severe
shifts in market conditions. As VaR cannot be used to
determine future losses in the Firm’s market risk positions,
the Firm considers other metrics, such as economic-value
stress testing and other techniques, as described further
below, to capture and manage its market risk positions
under stressed scenarios.
For certain products, specific risk parameters are not
captured in VaR due to the lack of inherent liquidity and
availability of appropriate historical data. The Firm uses
proxies to estimate the VaR for these and other products
when daily time series are not available. It is likely that
using an actual price-based time series for these products,
if available, would affect the VaR results presented. The
Firm uses alternative methods to capture and measure
those risk parameters that are not otherwise captured in
VaR, including economic-value stress testing, nonstatistical
measures and risk identification for large exposures as
described further below.
The Firm’s VaR model calculations are continuously
evaluated and enhanced in response to changes in the
composition of the Firm’s portfolios, changes in market
conditions, improvements in the Firm’s modeling techniques
and other factors. Such changes will also affect historical
comparisons of VaR results. Model changes go through a
review and approval process by the Model Review Group
prior to implementation into the operating environment.
For further information, see Model risk on page 153 of this
Annual Report.
Separately, the Firm calculates a daily aggregated VaR in
accordance with regulatory rules (“Regulatory VaR”), which
is used to derive the Firm’s regulatory VaR-based capital
requirements under the Basel 2.5 Market Risk Rule (“Basel
2.5”). This Regulatory VaR model framework currently
assumes a ten business-day holding period and an expected
tail loss methodology which approximates a 99%
confidence level. Regulatory VaR is applied to “covered”
positions as defined by Basel 2.5, which may be different
than the positions included in the Firm’s Risk Management
VaR. For example, credit derivative hedges of accrual loans