NATIXIS - Universal registration document and financial report 2019

3 RISK FACTORS, RISK MANAGEMENT AND PILLAR III Risk management

Methodology for measuring 3.2.5.3 market risks (Data certified by the Statutory Auditors in accordance with IFRS 7) Natixis’ market risk management is based on a risk metrics model that measures the risks incurred by each entity of the bank. Different techniques are used to measure market risk: Value at Risk (VaR) Natixis’ internal VaR model was approved by the Autorité de Contrôle Prudentiel et de Résolution (ACPR — French Prudential Supervisory Authority) in January 2009. Natixis uses VaR to calculate capital requirements for market risks within approved scopes, and to manage and supervise market risks. The model is based on a calculation by computer simulation, based on Monte Carlo-type methodology, taking into account a portfolio’s possible non-linear characteristics with respect to different risk factors. VaR is calculated and monitored daily for all the Natixis trading portfolios. Market data used in the valuation of portfolios (share prices, indices, interest rates, exchange rates, commodity prices and the related volatility) are updated on a daily basis when available, and the statistical data used (standard deviation and correlations) are updated weekly. All the trading portfolios are subject to adequate risk monitoring and supervision systems, in accordance with the market risk policies in force. A VaR limit is set at an overall level and for each business. These measurements give a snapshot of VaR and help identify potential losses in each business, based on a predetermined confidence level (99%) and time period (1 day). To this end, a statistical model has been constructed to track the combined behavior of market parameters affecting portfolio value. The calculation method is based on an econometric model whose standard deviations are calculated as being the maximum (risk factor by risk factor) standard deviations calculated over rolling 12-month and 3-month periods. This method makes VaR more responsive if the markets suddenly become more volatile. For the calculation of VaR, the portfolio’s holding horizon is set at one day for risk monitoring and 10 days for the calculation of capital. The 10-day holding period involves extrapolating from the one-day VaR by multiplying it by the square root of 10. Portfolio valuation methods vary according to product and are either based on a total revaluation or on sensitivity to first or second order market inputs to factor in both linear and non-linear effects (in particular for derivative financial instruments). Yields used by Natixis to simulate potential changes in risk factors are absolute yields for most risk factors. The exceptions to this rule are exchange rates, share prices and indices, precious metals prices and indices, commodity indices and commodity futures. Moreover, the reliability of the VaR is measured regularly through comparison with the changes in the daily trading results, a process also known as backtesting. This exercise allows an ex-post comparison of the potential losses, as projected ex-ante by the VaR, with the actual losses.

Stressed Value at Risk (SVaR) Due to changing regulatory standards (Basel 2.5), Natixis implemented a daily stressed VaR model (SVaR), which is calculated based on a fixed econometric model over a continuous 12-month period that defines the charge that the bank’s current VaR model would generate under a representative crisis scenario relevant to its portfolio. The calculation method is based on an historical simulation for a one-day horizon and a confidence level of 99%. However, unlike VaR, which uses 260 daily fluctuation scenarios on a sliding one-year period, stressed VaR uses a one-year historical window corresponding to a period of significant financial tension. The stressed period currently used by Natixis covers the period between September 1, 2008 and August 30, 2009, as it is the most conservative for calculating stressed VaR. As is the case for VaR, the 10-day holding period used by Natixis involves extrapolating from the one-day stressed VaR by multiplying it by the square root of 10. Performance monitoring and testing of VaR and SVaR Natixis’ internal model is based on a calculation of VaR and stressed VaR. To check that the calculation assumptions are respected, particularly the confidence interval of 99%, backtesting calculations are performed on the P&L on a daily basis. These involve comparing 99% VaR levels with actual and hypothetical daily P&L levels. The measurements are conducted on Natixis’ consolidated accounts and across all of its main activities. In the event that negative real or hypothetical P&L levels exceed the VaR calculation for the same day, an analysis is performed and explanations provided according to internally-defined criteria. The regulator is subsequently contacted to inform them of any such exceptions within Natixis’ authorized consolidation scope. In the event of too many exceptions over a period of one year, on a sliding annual basis, an add-on is made to the multiplying factor applied in calculating VaR and SVaR during measurements of Natixis’ capital. Incremental Risk Charge (IRC) The IRC is the capital charge required to cover rating migration risk and the default within one year of issuers for approved products in terms of specific interest rate risk. Calculated using a Monte Carlo internal simulation model, the IRC is a 99.9% value at risk which corresponds to the largest risk after eliminating the 0.1% of the worst outcomes over a period of one year. Rating and default migrations are simulated using an issuer correlation model and migration matrices over a capital horizon of one year. Positions are remeasured based on various scenarios. Thus, for each scenario, positions may be downgraded, upgraded or go into default. The liquidity horizon, which represents the time required to sell a position or hedge it in unfavorable market conditions, used in the IRC calculation model is one year for all positions and all issuers. The calibration of the transition matrix is based on Standard & Poor’s historical transition data. For both corporates and sovereigns, the historical depth exceeds 20 years. For issuers not rated by S&P, restatements are performed internally.

128

NATIXIS UNIVERSAL REGISTRATION DOCUMENT 2019

Made with FlippingBook Annual report