BPCE - 2018 Risk report / Pillar III

CREDIT RISK Risk measurement and internal ratings

STANDARDIZED APPROACH The “risk measurement and internal ratings” section describes the various approved models used by Groupe BPCE for the different exposures classes. Where the Group does not have an internal model authorized for use in determining capital requirements for a given exposure class, they have to be estimated based on corresponding inputs under the standardized approach. These inputs are based in particular on the credit assessments (ratings) performed by rating agencies recognized by the supervisory authority as meeting ECAI (External Credit Assessment Institutions) requirements, in particular Moody’s, Standard & Poor’s, Fitch Ratings and the Banque de France for GroupeBPCE. In accordance with Article 138 of Regulation No. 575/2013 (the Capital RequirementsRegulationor CRR) on prudential requirements for credit institutionsand investmentfirms, where a counterpartyhas been rated by several rating agencies, the counterparty’s rating is determinedon the basisof the secondhighest rating. When an externalcredit rating directly applicableto a given exposure is requiredand exists for the issuer or for a specificissuanceprogram, the procedures used to determine the weighting are applied in accordance withCRR Article139. For fixed-incomesecurities(bonds),short-termexternalratings of the bond take precedenceover external ratings of the issuer. If there are no external ratings for the bond, the issuer’s long-term external rating is taken into accountfor senior debtonly, except in the specific case of exposureto institutionswhose risk weight is derived from the credit ratingof the sovereign country in which it is established. BACKTESTING All information used to measure the Group’s exposure to all counterparties bearing credit risk is saved. Furthermore, all information on counterparties in default (collections, deterioration, write-offs) for the period is archived. Validity tests are conducted once a year on each of these internally-estimated inputs. All three credit risk inputs are subjectto yearly backtestingin order to verify the performance of the rating system. More specifically, backtestingis aimed at measuringthe overall performanceof models used, primarily to ensure that the model’s discriminatingpower has not declinedsignificantlyrelative to the modelingperiod. The average of estimated and observed values is calculated over several years using the information available for each model. Observed default rates are then comparedwith estimateddefault rates for each rating. Ratings are checked for through-the-cycle applicability. More specifically, for portfolios with low default rates (large corporates, banks, sovereigns and specialized lending), a detailed analysis is carried out using additional indicators such as severity differences, adjustmentsto agency ratings and changes in ratings before default. A more qualitative analysis is also performed. The scope of LGD values is consistent with the values observed, i.e. limited exclusivelyto exposuresat default.Estimatedvalues therefore cannot be directly compared with LGD values measured in the outstandingportfolio. The average of estimated and observed values is calculated over several years using the information available for each model. Actual collections are compared with estimated collections. Downturn LGDs are also verified. Backtestingresults may call for the implementationof action plans if the system is deemed not sufficiently prudent or effective.

Backtesting results and the associated action plans are discussed by the Group Models Committee, then reviewed by the DRCCP Group Standards & Methods Committee (see governance of the internal rating system). On the basis of these exercises, the rating system has been deemed satisfactory overall in terms of effective risk management. The calibrations are conservativewith respect to observed risk: observed default rates are lower than expected default rates over the entire cycle and over the most recent period. Observed losses on assets in default are lower thanexpected losses. REPORTS ON CREDIT RISK MODELS Since the Single SupervisoryMechanism (SSM) was implemented in 2014, the European Central Bank (ECB) has been working to strengthengovernanceof internal model supervisionthrough various investigations. These include the TRIM (Targeted Review of Internal Models), aimed at assessing the regulatory compliance of internal models specifically targeted by the ECB. To that end, TRIM investigations are based on a set of standardized inspection methodologies and techniques, which the teams mandated by the ECB use on-site. BPCE has been subject to TRIM reviews covering several scopes of operation since December 2015, giving rise to reports prepared by the ECB. As a result, several new initiativeswere launched with the aimof furtherimproving the existing system. TRIM - General Topics The “General Topics” TRIM focused on principles not specific to the models (overall governance) and took place over a short period (December 2015 to January 2016). At the end of the review, Groupe BPCE globally updated its roll-out/permanentpartial use (PPU) plan, primarily to better formalize PPU inputs. TRIM - Corporate exposures The “Corporateexposures”TRIM took place on-site from July 2017 to November 2017 and covered Probability of Default (PD) models for corporate exposures, targeting companies publishing consolidated financial statements as well as small enterprises generating revenue of less than € 10 million and publishing parent company balance sheets. This review covered the Banque Populairenetwork, the Caisse d’Epargne network and Natixis. It should be noted that a request for approval was submitted to allow the Small Enterprises(SE) model to be used for the Caisse d’Epargne network. In its final decision letter, the ECB validatedthe use of the new SE modelsfor the entire audited scope and confirmed its approval of the model applied to companies publishinga consolidatedbalance sheet. TRIM - Professional Retail exposures The “Professionalretail exposures”TRIM was carried out on-site from September 2017 to December 2017 and covered Probability of Default (PD) models for retail exposuresin the professionalcustomers segment. This review covered the Banque Populaire network, the Caisse d’Epargne network and Natixis. Note: a request for a material change in the calibrationof PD scales was made. In its final decision letter, the ECB validated the use of PD models for the entire audited scope and confirmed the material change in the calibration of PD scales.

5

89

Risk Report Pillar III 2018