Risk categories

Market risk is the risk of loss resulting from changes in market variables of two broad types: general market risk factors and idiosyncratic components. General market risk factors include interest rates, exchange rates, equity market indices, commodity prices and general credit spreads. Idiosyncratic components are specific to individual companies and affect the values of their securities and other obligations in tradable form, as well as derivatives referenced to those companies.

Most of UBS's market risk comes from the Investment Bank's trading activities. Group Treasury, part of the Corporate Center, assumes foreign exchange and interest rate risk in connection with its balance sheet, profit and loss, and capital management responsibilities. The wealth and asset management operations of UBS take limited market risk in support of client business.

Value at Risk

Value at Risk (VaR) is a statistical measure of market risk, representing a loss greater in absolute value than market risk losses realized over a set time period at an established probability. This assumes no change in the firm's trading positions. The tables on the next page show this statistic calibrated to a 10-day horizon and a 99% probability, using five years of historical data. For UBS and the Investment Bank the tables also show VaR for a 1-day horizon and a 99% probability, using five years of historical data. For a variety of reasons, the actual realized market risk loss experience may differ from that implied by the VaR measures of the firm. For example, the historical period used in creating the VaR measure had fluctuations in market rates and prices that may differ from those in the future; the firm's intra-period trading may mute or accentuate the losses; and the revenue consequences of a market move may differ from those implicitly assumed by the VaR model. All VaR measures are subject to these limitations to some extent and must be interpreted accordingly. UBS continues to review the performance of its VaR implementation and will continue to enhance its VaR model in order to more accurately capture the relationships between the market risks associated with certain positions, as well as the revenue impact of large market movements for some trading positions.

The Investment Bank's regulatory VaR ended the quarter at CHF 485 million, down slightly from CHF 519 million at the prior period end. Average Investment Bank regulatory VaR in the period was CHF 438 million, again only slightly down from CHF 461 million in third quarter 2008. The Investment Bank's internal management average VaR increased from CHF 303 million in the third quarter to CHF 341 million in the fourth quarter. UBS continues to actively reduce its risk exposures. However, VaR is a statistical risk measure which relies on a number of inputs and was impacted by updates to the historical time series in the period. These updates reflected the significant increase in the levels of volatility in many markets and risk factors in the fourth quarter.

Interest rate regulatory VaR, which includes exposures to movements in general credit spreads as well as exposure to the level and shape of yield curves, continued to be the key driver of Investment Bank regulatory VaR and internal management VaR in fourth quarter 2008. Interest rate regulatory VaR is dominated by hedges used to mitigate credit valuation adjustment (CVA) - the estimated sensitivity to credit spreads of protection required to hedge credit risk from counterparties in UBS's over-the-counter derivatives portfolio. CVA must currently be excluded from regulatory VaR (refer to the "Value at Risk developments - treatment of CVA" sidebar on page 26 of UBS's third quarter 2008 financial report for more information). CVA is included in internal management VaR which is dominated by the basis risk between CVA and cash positions, and related credit default swap hedges.

Period-end and average equities regulatory VaR remained relatively stable in fourth quarter 2008 compared with the prior period, as a reduction in risk exposure was more than offset by an increase in volatility which was reflected in a time series update towards the end of the quarter.

Regulatory VaR for UBS as a whole followed a similar pattern to Investment Bank regulatory VaR. Refer to the "Market risk" section of UBS's 2008 annual report, to be published on 19 March 2009, for more information on the scope of VaR.

Backtesting

"Backtesting" compares one-day regulatory VaR calculated on positions at the close of each business day with the revenues arising on those positions on the following business day. These "backtesting revenues" exclude non-trading revenues, such as fees and commissions, and estimated revenues from intraday trading. A "backtesting exception" occurs when backtesting revenues are negative and the absolute value of those revenues is greater than the previous day's VaR.

UBS experienced 25 backtesting exceptions in fourth quarter 2008, up from three in the previous period. UBS's VaR model is based on historical data and thus implicitly assumes that market moves will follow a similar pattern to those that have occurred in the past. As UBS's VaR model uses a look-back period of five years it does not respond quickly to periods of heightened volatility as experienced in the fourth quarter. An extreme lack of liquidity in the period also resulted in a breakdown in the relationships between a number of trading portfolios and their corresponding hedges (commonly known as basis risk). These factors, in addition to extreme market movements in a number of risk factors, were the primary contributors to the backtesting exceptions experienced. These results highlight the limitations of VaR as an absolute measure of risk and reinforce the need for multiple views of risk exposure. As an essential complement to VaR, UBS runs macro stress scenarios bringing together various combinations of market moves to reflect the most common types of potential stress events, and more targeted stress tests for concentrated exposures and vulnerable portfolios. UBS will continue improving its VaR model to better capture all relevant risks in its trading portfolio.

In the first histogram on page 26, daily backtesting revenues are shown for the whole of 2008. In the second histogram, the daily backtesting revenues are compared with the corresponding VaR over the same 12-month period for days when the backtesting revenues are negative. A positive result in this histogram represents a loss less than VaR, while a negative result represents a loss greater than VaR and therefore a backtesting exception.