12 CFR § 324.205 - VaR-based measure.
(a) General requirement. An FDIC-supervised institution must use one or more internal models to calculate daily a VaR-based measure of the general market risk of all covered positions. The daily VaR-based measure also may reflect the FDIC-supervised institution's specific risk for one or more portfolios of debt and equity positions, if the internal models meet the requirements of § 324.207(b)(1). The daily VaR-based measure must also reflect the FDIC-supervised institution's specific risk for any portfolio of correlation trading positions that is modeled under § 324.209. An FDIC-supervised institution may elect to include term repo-style transactions in its VaR-based measure, provided that the FDIC-supervised institution includes all such term repo-style transactions consistently over time.
(1) The FDIC-supervised institution's internal models for calculating its VaR-based measure must use risk factors sufficient to measure the market risk inherent in all covered positions. The market risk categories must include, as appropriate, interest rate risk, credit spread risk, equity price risk, foreign exchange risk, and commodity price risk. For material positions in the major currencies and markets, modeling techniques must incorporate enough segments of the yield curve - in no case less than six - to capture differences in volatility and less than perfect correlation of rates along the yield curve.
(2) The VaR-based measure may incorporate empirical correlations within and across risk categories, provided the FDIC-supervised institution validates and demonstrates the reasonableness of its process for measuring correlations. If the VaR-based measure does not incorporate empirical correlations across risk categories, the FDIC-supervised institution must add the separate measures from its internal models used to calculate the VaR-based measure for the appropriate market risk categories (interest rate risk, credit spread risk, equity price risk, foreign exchange rate risk, and/or commodity price risk) to determine its aggregate VaR-based measure.
(3) The VaR-based measure must include the risks arising from the nonlinear price characteristics of options positions or positions with embedded optionality and the sensitivity of the fair value of the positions to changes in the volatility of the underlying rates, prices, or other material risk factors. An FDIC-supervised institution with a large or complex options portfolio must measure the volatility of options positions or positions with embedded optionality by different maturities and/or strike prices, where material.
(4) The FDIC-supervised institution must be able to justify to the satisfaction of the FDIC the omission of any risk factors from the calculation of its VaR-based measure that the FDIC-supervised institution uses in its pricing models.
(5) The FDIC-supervised institution must demonstrate to the satisfaction of the FDIC the appropriateness of any proxies used to capture the risks of the FDIC-supervised institution's actual positions for which such proxies are used.
(b) Quantitative requirements for VaR-based measure.
(1) The VaR-based measure must be calculated on a daily basis using a one-tail, 99.0 percent confidence level, and a holding period equivalent to a 10-business-day movement in underlying risk factors, such as rates, spreads, and prices. To calculate VaR-based measures using a 10-business-day holding period, the FDIC-supervised institution may calculate 10-business-day measures directly or may convert VaR-based measures using holding periods other than 10 business days to the equivalent of a 10-business-day holding period. An FDIC-supervised institution that converts its VaR-based measure in such a manner must be able to justify the reasonableness of its approach to the satisfaction of the FDIC.
(2) The VaR-based measure must be based on a historical observation period of at least one year. Data used to determine the VaR-based measure must be relevant to the FDIC-supervised institution's actual exposures and of sufficient quality to support the calculation of risk-based capital requirements. The FDIC-supervised institution must update data sets at least monthly or more frequently as changes in market conditions or portfolio composition warrant. For an FDIC-supervised institution that uses a weighting scheme or other method for the historical observation period, the FDIC-supervised institution must either:
(i) Use an effective observation period of at least one year in which the average time lag of the observations is at least six months; or
(ii) Demonstrate to the FDIC that its weighting scheme is more effective than a weighting scheme with an average time lag of at least six months representing the volatility of the FDIC-supervised institution's trading portfolio over a full business cycle. An FDIC-supervised institution using this option must update its data more frequently than monthly and in a manner appropriate for the type of weighting scheme.
(c) An FDIC-supervised institution must divide its portfolio into a number of significant subportfolios approved by the FDIC for subportfolio backtesting purposes. These subportfolios must be sufficient to allow the FDIC-supervised institution and the FDIC to assess the adequacy of the VaR model at the risk factor level; the FDIC will evaluate the appropriateness of these subportfolios relative to the value and composition of the FDIC-supervised institution's covered positions. The FDIC-supervised institution must retain and make available to the FDIC the following information for each subportfolio for each business day over the previous two years (500 business days), with no more than a 60-day lag:
(1) A daily VaR-based measure for the subportfolio calibrated to a one-tail, 99.0 percent confidence level;
(2) The daily profit or loss for the subportfolio (that is, the net change in price of the positions held in the portfolio at the end of the previous business day); and
(3) The p-value of the profit or loss on each day (that is, the probability of observing a profit that is less than, or a loss that is greater than, the amount reported for purposes of paragraph (c)(2) of this section based on the model used to calculate the VaR-based measure described in paragraph (c)(1) of this section).