BPCE_REGISTRATION_DOCUMENT_2017

FINANCIAL REPORT IFRS Consolidated Financial Statements of Groupe BPCE as at December 31, 2017

The table below provides the mainunobservableinputs and thevalue rangesfor these instruments:

Unobservable data ranges among relevant Level 3 products

Main types of products comprising Level 3 within the instrument class

Class of instrument

Valuation techniques used

Main unobservable data

Technique for estimating defaults given the correlation effect and recoverymodeling Extrapolation from pricesbased on recoveryassumptions Discounted expected cash flows based onearlyredemption assumptionsonthe underlying portfolio Interestrate options valuation models Model representing severalyield curvefactors

Correlation curve specific to the portfoliounderlying the CDO

CDOs,Indextranche

5%-95% (a)

Private Finance InitiativeCDS (other than CDS onsecuritization assets)

Creditderivatives

Recovery rate

60%-100%

Securitization swaps

Early redemption rate

2%-17%

StickyCMS/Volatility Bonds

Mean reversion inputs

1%-5%

CallableSpreadOptions and Corridor Callable Spread Options

Meanreversionspread

0%-30%

SpreadLock: (2.288 bp),+29.94bp TECvolatility:50bp/70bp TEC-CMS correlation 70%/95%

Bivariate standard model to measurethe time value of Spread Lockoptionsand replication for CMS andTECForwards

SpreadLockSwapand SpreadLock Option

SpreadLockcurve,TECForward volatility and TEC-CMS correlation Interestrate vol.for currencies absentfrom Totem or long maturities

Interestrate vol.: 4.69% to 101.36%

Interestrate derivatives

Volatility cap/floor

Black& Scholes model

Europeanbarriercall option, Asian call option, Vanilladigital call option, European call option

Skew Model, Localvolatility model, Black& Scholesmodel,

Interestrate vol.for current pairs absentfrom Totem or long

Currencyderivatives

maturities ATM vol.:0.84% to 22.25%

Syntheticmodelingof underlying generalbasket(with repoto estimate) and actuarial valuation for TRSor usinga standard equity/interest rate hybridmodel for the TRSauto call

5

Generalcollateralrepo: (0.84%)/+0.5%

Repos and general collateral TRS

TRSandreposindexed to a basket of generalequities

Repo curveof generalbaskets

EUR/CHFcorrelation:36.7%; 40.9% Long-term volatility: 9%-16% USD/CHFcorrelation: (69.10%); (78.80%)

Stripsof long-termoptions,Stripsof quanto options, Strips ofdigital options

Black & Scholes model

Currency/currency correlation USD/CHF & EUR/CHF long-term

Helvetix derivatives

OptionsspreadandDigital optionspread

Gaussian copula

volatility Long-term volatility: 9%-15%

Theapproach used is a hybrid modelthat combines the local volatility-typemulti-underlying equitymodelwith a one-factor Heath-Jarrow-Morton (HJM1F) interestrate model

Fundcorrelation – Interest rates:

Payoffs asTargetVolatilitystrategyand CPPI on MutualFunds

Fund-based derivatives

Fund data

(40%) to 25%

AUD/JPYand USD/JPY correlation:15%-50% Long-term volatility: 8%-15%

Correlationbetween currency and interestrates and long-term volatility levels Correlation inputs (equity-forex, equity-interest rates, interest rates-forex)

Hybridinterest rate/currency derivatives

Long-termPRDC/PRDKO/TARN structures

Hybridcurrency/interest rate optionsvaluation model

EQ/FX= 20%, 50% EQ/IR= 30%, 50% FX/IR=20%, 30%

Long-dated callable range accrual notes (15Y) onseveralassetclasses (equity+forex+interest rates)

Hybridequity/interest rate/forexderivatives

Hybridmodels coupled with equity, forexandinterest rate diffusion

Interestrate/Creditcorrelation: (13%),3% Creditvol:Structureby term ([2Y, 200%],[5Y, 60%],[10Y, 50%]

Hybridmodels coupled with interestrate diffusion and credit diffusion Volatility options valuation model incorporating correlationbetween assets

Hybridinterest rate/credit derivatives

Long-dated interest rateand credit callable range accrual notes (15Y) (default event)

Correlation inputs(interest rate-credit and volatility-credit)

Stock/stock correlation: 18.4 to 92.13

Equity derivatives

Long maturity multi-underlying payoffs

Correlation inputs

All transactions including this type of dataare fullyback-to-back; this inputjustifyingthe Level 3classification is entirely hedged. (a)

265

Registration document 2017