BPCE - 2018 Registration document

FINANCIAL REPORT IFRS Consolidated Financial Statements of BPCE SA group as at December 31, 2018

for the Large Corporates, Banks and Sovereigns loan books, the ● quantitative criterion is based on the level of variation in the rating since initial recognition. The same qualitative criteria as for Individual Customers, Professional Customers and SMEs apply, as do complementary criteria based on the change in sector rating and the level of country risk; for Specialized Financing, the criteria applied vary according to ● the characteristics of the exposures and the related ratings system. Exposures rated by the tool dedicated to large exposures are treated in the same way as Large Corporates; other exposures are treated in the same way as SMEs. For all these loan books, the ratings used to measure the increase in risk correspond to the ratings produced by internal systems when they are available, as well as on external ratings, particularly when an internal rating is not available. The standard provides that the credit risk of a financial instrument has not increased materially since its initial recognition if this risk is considered to be low at the end of the fiscal year. This provision is applied to certain investment-grade debt securities held by Corporate & Investment Banking. For Stage 1 and 2 financial instruments, expected credit losses are also measured on an individual basis, depending on the features of each contract. Collective provisions may be established by the different Group institutions, corresponding to “sector” provisions. Group entities are therefore responsible for assessing the consistency of provisions determined for the Group with the local and sector characteristics of their portfolio and for defining additional sector provisions if necessary. The few portfolios not covered by the methodologies described below (not material at the Group level) may also be measured on a collective basis [Note: details to be provided by entities making impairment and provisions on their own scopes: CFF, Palatine, BRED, CEMP, CEBPL, Crédit Coopératif… if not provided above]. Expected credit losses on financial instruments classified as Stage 1 or Stage 2 are measured as the product of several inputs: cash flows expected over the lifetime of the financial instrument, ● discounted on the valuation date – these flows are determined according to the characteristics of the contract, its effective interest rate and the level of early repayment expected on the contract; Loss Given Default (LGD); ● Probabilities of Default (PD), for the coming year in the case of ● Stage 1 financial instruments and until the contract’s maturity in the case of Stage 2 financial instruments. The Group draws on existing concepts and mechanisms to define these inputs, and in particular on internal models developed to calculate regulatory capital requirements and on projection models used in the stress test system. Certain adjustments are made to comply with the specifics of IFRS 9: IFRS 9 inputs aim to provide an accurate estimate of expected ● credit losses for accounting provision purposes, whereas

prudential inputs are more cautious for regulatory framework purposes. Several of the safety buffers applied to the prudential inputs are therefore restated; IFRS 9 inputs must allow expected credit losses to be estimated ● until the contract’s maturity, whereas prudential inputs are defined to estimate 12-month expected losses. Twelve-month inputs are thus projected over long periods; IFRS 9 inputs must be forward-looking and take into account the ● expected economic environment over the projection period, whereas prudential inputs correspond to the cycle’s average estimates (for PD) or bottom-of-the-cycle estimates (for LGD and the flows expected over the lifetime of the financial instrument). The PD and LGD prudential inputs are therefore also adjusted based on this expected economic environment. Inputs are adjusted to economic conditions by defining three economic scenarios over a three-year period. The variables defined in each of these scenarios allow for the distortion of the PD and LGD parameters and the calculation of an expected credit loss for each economic scenario. Inputs for periods longer than three years are projected based on the mean reversion principle. The models used to distort the PD and LGD inputs rely on those developed as part of the stress test system for the purpose of ensuring consistency. These economic scenarios are associated with probabilities of occurrence, ultimately making it possible to calculate an average probable loss used as the IFRS 9 impairment amount. These scenarios are defined using the same organization and governance as that defined for the budget process, requiring an annual review based on proposals from the Economic Research department and approval by the Executive Management Committee. To ensure consistency with the budget scenario, the central scenario is the budget scenario. Two variants – an optimistic view and a pessimistic view – are also developed around this scenario. The likelihood that the scenarios will occur is reviewed on a quarterly basis by the Group’s Watch List and Provisions Committee. The inputs thus defined allow expected credit losses for all rated exposures to be valued, regardless of whether they belong to a scope approved using an internal method or they are processed using the standard method for the calculation of risk-weighted assets. Conservative default rules are applied to unrated exposures (the stakes are not material for the Group). These rules involve assigning the highest rating on the internal scale in the absence of a rating on approval and the lowest rating on the scale before the at-risk stage in the absence of a rating to date. The mechanism for validating IFRS 9 inputs is fully integrated in the validation mechanism for existing models within the Group. Inputs are subject to review by an independent unit responsible for internal model validation, which is then examined by the Group Model Committee. Furthermore, the recommendations issued by the validation unit are subsequently monitored.

5

475

Registration document 2018