FRTB Selection: Standard Approach vs. Internal Models

The FRTB Standardised Approach (A-SA) uses regulatory risk weights and correlation parameters across seven risk classes. Capital requirements are the sum of the Sensitivity-Based Method (SbM), Default Risk Charge (DRC) and Residual Risk Add-On (RRAO). The Internal Models Approach (A-IMA) calculates capital based on Expected Shortfall using proprietary models with individual liquidity horizons per risk factor, plus the Stressed Expected Shortfall component and a separate DRC calculation. In practice, IMA can deliver 30–50% lower capital charges for diversified portfolios, but only for desks that pass the P&L Attribution Test and backtesting.

Each trading desk seeking IMA approval must continuously pass two quantitative tests: First, the P&L Attribution Test (PLAT), which checks the alignment between hypothetical and actual profit and loss at desk level. Second, backtesting with a maximum of four breaches in twelve months (traffic light approach). Additionally, desks must have sufficient modellable risk factors (MRF) – non-modellable risk factors (NMRF) receive a separate capital add-on via the Stressed Expected Shortfall. Full model documentation, an independent validation function and regulatory approval from the national supervisor are also required.

A hybrid strategy is beneficial when an institution operates desks with complex, diversified portfolios alongside desks with simpler products. Desks with high diversification and liquid risk factors benefit from IMA through lower capital charges. Desks with many non-modellable risk factors (NMRF) or low trading volumes fare better under the Standardised Approach, as NMRF surcharges can eliminate IMA advantages. The decision requires desk-specific capital impact analysis comparing both direct capital costs and ongoing implementation and validation expenses per approach.

The Sensitivity-Based Method (SbM) is the core component of the FRTB Standardised Approach. It replaces the former simplified standard approach with a more risk-sensitive calculation. Positions are assessed using delta, vega and curvature sensitivities and aggregated via regulatory correlation matrices within and across risk classes (interest rate, credit, equity, commodity, FX, basis risk). Three correlation scenarios (low, medium, high) are calculated and the highest value is applied. Additionally, the Default Risk Charge covers default risk while the Residual Risk Add-On addresses non-standardisable risks.

The Standardised Approach requires lower upfront investment: main costs cover adapting calculation systems for SbM, DRC and RRAO plus data provisioning for the seven risk classes. Typical timeframe: 12–18 months. The IMA demands significantly higher investment in risk modelling, Expected Shortfall infrastructure, NMRF calculation, P&L attribution systems and an independent validation function. Regulatory approval adds another 6–12 months. Total timeframe: 18–36 months. Ongoing IMA operating costs substantially exceed SA costs but must be weighed against capital savings.

CRR 3 makes FRTB mandatory in the EU from

1 January 2027. Key changes: SA becomes a compulsory calculation for all institutions and serves as the capital floor (output floor) for IMA users. The EU has introduced some adjustments compared to the Basel standard, including treatment of certain bond structures and the Residual Risk Add-On. BaFin and EBA are developing regulatory technical standards (RTS) for detailed specifications. Institutions should not delay implementation, as parallel reporting under old and new regimes is expected.

ADVISORI guides financial institutions through the entire decision and implementation process. During the analysis phase, we quantify capital impacts of both approaches at trading desk level and identify the optimal SA/IMA allocation. For implementation, we provide gap analyses, process design, system adaptation and preparation for model approval. Our experience from multiple FRTB projects at universal banks and securities firms enables proven solutions for typical challenges such as NMRF treatment, desk restructuring and parallel reporting.