NATIXIS // 2021 Universal Registration Document

CONSOLIDATED FINANCIAL STATEMENTS AT DECEMBER 31, 2021 Consolidated financial statements and notes

The main models used to value and manage currency products are local volatility and stochastic models (like the LSV model for the equity scope), as well as the hybrid models combining an underlying currency model with two one-factor Hull & White models to understand domestic and foreign economies’ fixed-income curves; credit derivatives: products generally have specific characteristics V that justify the choice of model. The main models used for the valuation and management of credit products are the Hull & White credit factor model (HW1F Credit) and the hybrid Bi-Hull & White Rate-Credit model (Bi-HW Rate/Credit). The HW1F Credit model is used to disseminate the credit curve (CDS curve) with a Gaussian factor. The Bi-HW Rate/Credit model enables the yield curve and the credit curve to be disseminated jointly each with a Gaussian factor correlated with each other; commodities products: commodities products generally have their V own characteristics that justify the choice of model. The main models used for the valuation and management of commodities products are the Black & Scholes models, with local volatility and local volatility combined with the Hull & White 1-factor (H&W1F), an extended version for all of these models to a multi-framework underlying asset to manage all futures of the commodity family. The Black & Scholes model is based on a log-normal dynamic of the underlying asset and a deterministic volatility assumption. The local volatility model treats volatility as a function of time and the price of the underlying. Its main property is that it considers the implied volatility of the option (derived from market data) relative to its exercise price. The H&WIF model consists of combining the local volatility model described above with a one-factor Hull & White model, also described above (see fixed-income products) . Inputs relating to all such Level 2 instruments were demonstrated to be observable and documented. From a methodology perspective, observability is based on four inseparable criteria: inputs are derived from external sources (primarily a recognized V contributor, for example); they are updated periodically; V they are representative of recent transactions; V their characteristics are identical to the characteristics of the V transaction. If necessary, a proxy may be used, provided that the relevance of such an arrangement is demonstrated and documented. The fair value of instruments obtained using valuation models is adjusted to take account of liquidity risk (bid-ask), counterparty risk, the risk relating to the cost of funding uncollateralized or imperfectly collateralized derivatives, own credit risk (measurement of liability derivative positions), modeling risk and input risk. The margin generated when these instruments begin trading is immediately recognized in income.

Complex instruments Some more hybrid and/or long-maturity financial instruments are measured using a recognized model on the basis of market inputs derived from observable data such as yield curves, implied volatility layers of options, market consensus data or active OTC markets. The main models for determining the fair value of these instruments are described below by type of product: equity products: complex products are valued using: V market data, V a payoff, i.e. a calculation of positive or negative cash flows V attached to the product at maturity, a model of changes in the underlying asset. V The products traded may be mono-underlying, multi-underlying or hybrid (e.g. fixed income/equity) products. The main models used for equity products are local volatility, local volatility combined with the one-factor Hull & White (H&W1F) model, as well as the Local Stochastic Volatility (“LSV”) models. The local volatility model treats volatility as a function of time and the price of the underlying. Its main property is that it considers the implied volatility of the option (derived from market data) relative to its exercise price. The hybrid local volatility combined with H&W1F consists of combining the local volatility model described above with a one-factor Hull & White model, described below (see fixed-income products) . The LSV Model is based on joint diffusion of the underlying asset and its volatility (two factors in total), with a local volatility function (called a “decorator”) in order to be consistent with all of the vanilla options; fixed-income products: fixed-income products generally have V specific characteristics which justify the choice of model. The valuation of the payoff will take into account all underlying risk factors. The main models used to value and manage fixed-income products are the one-factor (HW1F) and two-factor (HW2F) Hull & White models, or the one-factor Hull & White stochastic volatility model (HW1FVS). The HW1F model makes it possible to model the yield curve with a single Gaussian factor and one calibration of the vanilla rate options. The HW2F model makes it possible to model the yield curve with two factors and one calibration of the vanilla rate options and spread-option instruments. The HW1VS model makes it possible to jointly model the Gaussian factor representing the yield curve and its volatility (like the LSV model for equity); currency products: currency products generally have specific V characteristics which justify the choice of model.

5

331

www.natixis.com

NATIXIS UNIVERSAL REGISTRATION DOCUMENT 2021