Morgan Stanley 2010 Annual Report Download - page 107

The table below presents the Company’s 95%/one-day VaR:
Table 1: 95% Total VaR 95% One-Day VaR for 2010 95% One-Day VaR for 2009
Primary Market Risk Category
Period
End Average High Low
Period
End Average High Low
(dollars in millions)
Interest rate and credit spread .......... $102 $129 $ 147 $100 $142 $128 $149 $106
Equity price ........................ 30 28 52 19 23 21 36 14
Foreign exchange rate ................ 21 24 50 9 26 20 47 7
Commodity price .................... 30 28 36 21 24 24 38 18
Less: Diversification benefit(1) ........ (65) (70) (120) (32) (57) (55) (108) (34)
Total Trading VaR .................. $118 $139 $ 165 $117 $158 $138 $162 $111
Total Non-trading VaR ............... $ 77 $ 82 $137 $ 57 $ 67 $ 63 $ 89 $ 33
Aggregate VaR ..................... $146 $173 $ 217 $143 $187 $163 $205 $119
(1) Diversification benefit equals the difference between Total VaR and the sum of the VaRs for the four risk categories. This benefit arises
because the simulated one-day losses for each of the four primary market risk categories occur on different days; similar diversification
benefits also are taken into account within each category.
The Company’s average Trading VaR for 2010 is relatively unchanged at $139 million compared with $138
million for 2009. Increased equity, foreign exchange and commodity risks were offset by increased
diversification in the trading portfolios.
The Company’s average Non-trading VaR for 2010 increased to $82 million compared with $63 million for
2009. The Company’s average Aggregate VaR for 2010 was $173 million compared with $163 million for 2009.
The increase in both Non-trading and Aggregate VaR was due primarily to the Company’s investment in
Invesco, which was sold in November 2010, as well as increased interest rate sensitivity of deposits in the
declining rate environment.
VaR Statistics under Varying Assumptions.
VaR statistics are not readily comparable across firms because of differences in the breadth of products included
in each firm’s VaR model, in the statistical assumptions made when simulating changes in market factors and in
the methods used to approximate portfolio revaluations under the simulated market conditions. These differences
can result in materially different VaR estimates for similar portfolios. The impact varies depending on the factor
history assumptions, the frequency with which the factor history is updated and the confidence level. As a result,
VaR statistics are more reliable and relevant when used as indicators of trends in risk taking rather than as a basis
for inferring differences in risk taking across firms.
Table 2 presents the VaR statistics that would result if the Company were to adopt alternative parameters for its
calculations, such as the reported confidence level (95% versus 99%) for the VaR statistic or a shorter historical
time series (four-year versus one-year) for market data upon which it bases its simulations. The four-year VaR
measure continues to be sensitive to the high market volatilities experienced in the fourth quarter of 2008, while
the one-year VaR is no longer affected by this phenomenon. Consequently, the four-year VaR is a more
conservative approximation of the Company’s portfolio risk.
101