2020 Universal registration document and annual financial report - BNP PARIBAS 435

5risks and CaPital adequaCy Pillar 3

5

Market risk

Stressed VaR

Stressed VaR is calibrated over a specified full twelve-month period, including a crisis period. This period applies across the Group, which must have comprehensive market data to calculate the risk measurements and remain relevant when applied to the current trading book. An expert committee reviews the period on a quarterly basis in accordance with a quantitatively informed approach among the three scenarios that generate the maximum stressed risk measures.

The current reference period for calibrating stressed VaR is from 2 July 2008 to 30 June 2009.

BNP Paribas uses the same calculation method as for calculating VaR, with market parameters determined based on this reference period.

The SVaRs presented below are based on a one-day time horizon and a 99% confidence interval and correspond to measurements taken into account within the framework of monitoring market limits.

The SVaR (1-day, 99%) was not impacted by the health crisis and remained stable at around EUR 88 million throughout 2020. The increase in the SVaR compared to 2019 was due to the change in the reference period that occurred at the end of 2019.

➤ TABLE 85: STRESSED VALUE AT RISK (1-DAY, 99%)

In millions of euros

Year to 31 Dec. 2020 Year to 31 Dec. 2019

Minimum Average Maximum Last measure Average Last measure

Stressed Value at Risk 64 88 120 84 63 63

Incremental Risk Charge (IRC)

The IRC approach measures losses due to default and ratings migration at the 99.9% confidence interval (i.e. the maximum loss incurred after eliminating the 0.1% worst events) over a capital and liquidity horizon or rebalancing frequency of one year, assuming a constant level of risk over this horizon. The IRC scope mainly includes plain vanilla credit products (bonds and CDS, excluding securitised products) from the trading book.

This approach is used to capture the incremental default and migration risks on all non-securitised products.

The model is currently used in the risk management processes. This model was approved by the supervisor.

The calculation of IRC is based on the assumption of a constant level of risk over the one-year capital horizon, implying that the trading positions or sets of positions can be rebalanced during the one-year capital horizon in a manner that maintains the initial risk level, measured by the VaR or by the profile exposure by credit rating and concentration. This rebalancing frequency is called the liquidity horizon.

The model is built around a rating-based simulation for each obligor, which captures both the risk of the default as well as the risk of rating migration. The dependency between debtors is based on a multi-factor asset return model. The valuation of the portfolios is performed in each simulated scenario. The model uses a constant one-year liquidity horizon.

The IRC remains stable at around EUR 200 million throughout 2020, but with an increased risk profile resulting from an increase in positions and hedges.

Comprehensive Risk Measure (CRM) for credit correlation portfolio

CRM is an additional capital charge to the IRC which applies to the credit correlation portfolio (excluding securitisation products) from the trading book. It measures potential losses from a variety of specific price change risks (spread, correlation, recovery, credit migration, etc.) at the 99.9% confidence interval (i.e. the maximum loss incurred after eliminating the 0.1% worst events) over a capital and liquidity horizon or rebalancing frequency of one year, assuming a constant level of risk over this horizon.

The corporate correlation activity is an activity that consists of trading and risk managing mainly bespoke corporate CDOs and their hedges using single name CDS, CDS indices and index tranches. This activity falls under the structured credit activity trading within the Credit business line of Global Markets.

The valuation framework uses both market observable prices (particularly used for CDS, index and index tranches) and data established based on models for the implicit correlations and recovery rates, using the models of dependency between debtors used for the IRC.