Société Générale / Risk report 2021 - Pillar 3 2020

9 MARKET RISK MARKET RISK MAIN MEASURES PILLAR 3 - 2021 | SOCIETE GENERALE GROUP | 176 Within the framework described above, the one-day 99% VaR, calculated according to the 260 scenarios, corresponds to the mean of the second and third largest losses computed, without applying any weighting to the scenarios. The day-to-day follow-up of market risk is performed via the one-day VaR, which is calculated on a daily basis at various granularity levels. Regulatory capital requirements, however, oblige us to take into account a ten-day horizon, thus we also calculate a ten-day VaR, which is obtained by multiplying the one-day VaR aggregated at Group level by the square root of 10. This methodology complies with regulatory requirements and has been reviewed and validated by the regulator. The VaR assessment is based on a model and a certain number of conventional assumptions, the main limitations of which are as follows: by definition, the use of a 99% confidence interval does not take into p account losses arising beyond this point; VaR is therefore an indicator of the risk of loss under normal market conditions and does not take into account exceptionally significant fluctuations; VaR is computed using closing prices, meaning that intraday p fluctuations are not taken into account; the use of a historical model is based on the assumption that past p events are representative of future events and may not capture all potential events. The Market Risk Department mitigates the limitations of the VaR model by performing stress tests and other additional measurements. The same model is used for the VaR computation for almost all of Global Banking & Investor Solution’s activities (including those related to the most complex products) and the main market activities of Retail Banking and Private Banking. The few activities not covered by the VaR method, either for technical reasons or because the stakes are too low, are monitored using stress tests, and capital charges are calculated using the standard method or through alternative in-house methods. The main market risk not covered by an internal model is the exchange risk of the banking book, which is not subject to a daily revaluation by construction and therefore cannot be taken into account in a VaR calculation. Backtesting The relevance of the model is checked through continuous backtesting in order to verify whether the number of days for which the negative result exceeds the VaR complies with the 99% confidence interval. The results of the backtesting are audited by the Risk Department in charge of the validation of internal models, which, as second line of defence, also assesses the theoretical robustness (from a design and development standpoint), the correctness of the implementation and the adequacy of the model use. The independent review process ends with (i) review and approval Committees and (ii) an Audit Report detailing the scope of the review, the tests performed and their outcomes, the recommendations and the conclusion of the review. The model control mechanism gives rise to reporting to the appropriate authorities. In compliance with regulations, backtesting compares the VaR to the (i) actual and (ii) hypothetical change in the portfolio’s value: in the first case (backtesting against “actual P&L”), the daily P&L (1) p includes the change in book value, the impact of new transactions and of transactions modified during the day (including their sales margins), refinancing costs, the various related commissions (brokerage fees, custody fees, etc.), as well as provisions and parameter adjustments made for market risk; in the second case (backtesting against “hypothetical P&L”), the p daily P&L (2) includes only the change in book value related to changes in market parameters and excludes all other factors. As a result of the Covid-19 crisis, markets were impacted, particularly in the first half of 2020, by high volatility across all asset classes. Daily losses on market activities were recorded 44 times. A breach of VaR backtesting was observed at Societe Generale Group level on 13 days. Following the decision of the European Parliament to authorise the exclusion of backtesting events occurring between 1 January 2020 and 31 December 2021 not attributable to internal model, the competent authorities have validated the exclusion of such events from the calculation of Societe Generale’s risk-weighted assets for market risk. Three VaR backtesting breaches were finally maintained at Group level, against hypothetical result, in particular as a result of the VaR model not taking into account movements on share dividends once they are announced. “Actual P&L” by agreement hereinafter. (1) “Hypothetical P&L” by agreement hereinafter. (2)