Article 294

1. As part of the initial and on-going validation of its CCR exposure model and its risk measures, an institution shall ensure that the following requirements are met:

(a) the institution shall carry out back-testing using historical data on movements in market risk factors prior to the permission by the competent authorities in accordance with Article 283(1). That back-testing shall consider a number of distinct prediction time horizons out to at least one year, over a range of various initialisation dates and covering a wide range of market conditions;

(b) the institution using the approach set out in Article 285(1)(b) shall regularly validate its model to test whether realised current exposures are consistent with prediction over all margin periods within one year. If some of the trades in the netting set have a maturity of less than one year, and the netting set has higher risk factor sensitivities without these trades, the validation shall take this into account;

(c) it shall back-test the performance of its CCR exposure model and the model's relevant risk measures as well as the market risk factor predictions. For collateralised trades, the prediction time horizons considered shall include those reflecting typical margin periods of risk applied in collateralised or margined trading;

(d) if the model validation indicates that Effective EPE is underestimated, the institution shall take the action necessary to address the inaccuracy of the model;

(e) it shall test the pricing models used to calculate CCR exposure for a given scenario of future shocks to market risk factors as part of the initial and on-going model validation process. Pricing models for options shall account for the nonlinearity of option value with respect to market risk factors;

(f) the CCR exposure model shall capture the transaction-specific information necessary to be able to aggregate exposures at the level of the netting set. An institution shall verify that transactions are assigned to the appropriate netting set within the model;

(g) the CCR exposure model shall include transaction-specific information to capture the effects of margining. It shall take into account both the current amount of margin and margin that would be passed between counterparties in the future. Such a model shall account for the nature of margin agreements that are unilateral or bilateral, the frequency of margin calls, the margin period of risk, the minimum threshold of un-margined exposure the institution is willing to accept, and the minimum transfer amount. Such a model shall either estimate the mark-to-market change in the value of collateral posted or apply the rules set out in Chapter 4;

(h) the model validation process shall include static, historical back-testing on representative counterparty portfolios. An institution shall conduct such back-testing on a number of representative counterparty portfolios that are actual or hypothetical at regular intervals. Those representative portfolios shall be chosen on the basis of their sensitivity to the material risk factors and combinations of risk factors to which the institution is exposed;

(i) an institution shall conduct back-testing that is designed to test the key assumptions of the CCR exposure model and the relevant risk measures, including the modelled relationship between tenors of the same risk factor, and the modelled relationships between risk factors;

(j) the performance of CCR exposure models and its risk measures shall be subject to appropriate back-testing practice. The back testing programme shall be capable of identifying poor performance in an EPE model's risk measures;

(k) an institution shall validate its CCR exposure models and all risk measures out to time horizons commensurate with the maturity of trades for which exposure is calculated using IMM in accordance to the Article 283;

(l) an institution shall regularly test the pricing models used to calculate counterparty exposure against appropriate independent benchmarks as part of the on-going model validation process;

(m) the on-going validation of an institution's CCR exposure model and the relevant risk measures shall include an assessment of the adequacy of the recent performance;

(n) the frequency with which the parameters of an CCR exposure model are updated shall be assessed by an institution as part of the initial and on-going validation process;

(o) the initial and on-going validation of CCR exposure models shall assess whether or not the counterparty level and netting set exposure calculations of exposure are appropriate.

2. A measure that is more conservative than the metric used to calculate regulatory exposure value for every counterparty may be used in place of alpha multiplied by Effective EPE with the prior permission of the competent authorities. The degree of relative conservatism will be assessed upon initial approval by the competent authorities and at the regular supervisory reviews of the EPE models. An institution shall validate the conservatism regularly. The on-going assessment of model performance shall cover all counterparties for which the models are used.

3. If back-testing indicates that a model is not sufficiently accurate, the competent authorities shall revoke its permission for the model, or impose appropriate measures to ensure that the model is improved promptly.