Source: https://www.bankofengland.co.uk/-/media/boe/files/prudential-regulation/policy-statement/2026/january/ps126app1.pdf · captured 2026-09-29 · checked 2026-09-29 · c75b8196

Page 272 thumbnail
14341 This document is effective from 1 January 2027 to accompany PS1/26.
14342Please see: www.bankofengland.co.uk/prudential-regulation/publication/2026/january/implementation-of-basel-3-1-final-rules-policy-statement.
14343 PRA2026/1
14345 (h) the risk measurement model shall use a separate risk factor at least for each commodity in
14346 which the institution holds significant positions, unless the institution has a small aggregate
14347 commodity position compared to all its trading activities, in which case it may use a
14348 separate risk factor for each broad commodity type; for material exposures to commodity
14349 markets, the model shall capture the risk of less than perfectly correlated movements
14350 between commodities that are similar, but not identical, the exposure to changes in
14351 forward prices arising from maturity mismatches, and the convenience yield between
14352 derivative and cash positions;
14354 (i) the proxies used shall show a good track record for the actual position held, shall be
14355 appropriately conservative, and shall be used only where the available data are
14356 insufficient, such as during the period of stress referred to in point (c) of Article 325bc(2);
14358 (j) for material exposures to volatility risks in instruments with optionality, the risk
14359 measurement model shall capture the dependency of implied volatilities across strike
14360 prices and options' maturities; and
14362 (k) an institution shall periodically and at least annually demonstrate that the modelling of
14363 positions in CIUs in their risk measurement model leads to own funds requirements that
14364 are at least as conservative as if a look-through approach was applied to those positions.
14366 2. An institution may use empirical correlations within broad categories of risk factors and, for the
14367 purpose of calculating the unconstrained expected shortfall measure UES t as referred to in
14368 Article 325bb(1), across broad categories of risk factors only where the institution's approach
14369 for measuring those correlations is sound, consistent with the applicable liquidity horizons, and
14370 implemented with integrity.
14371 3. An institution shall ensure that:
14372 (a) for the purpose of calculating the partial expected shortfall calculations referred to in Article
14373 325bc, the data inputs used in their risk measurement model meet the requirements in
14374 paragraphs 4 to 10;
14376 (b) where the data inputs used for a risk factor in the risk measurement model do not meet the
14377 requirements in paragraphs 4 to 10, institution deems the risk factor shall be deemed as
14378 non-modellable and shall calculate the own funds requirements for market risk in
14379 accordance with Article 325bk for that risk factor; and
14381 (c) it considers the coefficients of a multifactor model as non-modellable risk factors in
14382 accordance with Article 325be unless the coefficients of that multifactor model are
14383 determined empirically based on historical data.
14385 Data inputs derived from a combination of modellable risk factors
14386 4. An institution shall ensure that:
14387 (a) it derives data input used in an institution’s risk measurement model from only modellable
14388 risk factors. An institution may use interpolation from a combination of modellable risk
14389 factors to determine a data input; provided that if so specified in the IMA permission, an
14390 institution may use extrapolation to determine a data input if the extrapolation is only a
14391 reasonable distance from the closest modellable risk factor;
14393 (b) where an institution uses interpolation or extrapolation to generate a data input for the
14394 institution’s risk measurement model, it must determine the theoretical changes in portfolio
14395 value for the P&L attribution requirements in accordance with Article 325bg using that
14396 same interpolation or extrapolation; and
14398 Page 272 of 492