The Challenges of MCRMR-2016 (FRTB)
As a result of the FRTB and the resulting MCRMR-2016, there will necessarily be a significant impact on risk reporting operations and risk calculations – with the latter likely to require rather weighty changes to current systems.
Operationally-speaking, risk reporting will be based on an entirely new risk dataset of regulatory sensitivities and weightings, including a new DRC which replaces the old IRC. This will be for the revised SA which will always be required to be produced daily. If the client is using IMA, then there are further changes to cover the ES and other new IMA requirements.
Systems-wise – if a client is currently using IMA, the system will have to be adapted to include and calculate the mandatory revised SA – then further modifications will be required to meet ES and the rest of the FRTB requirements. If the client is only on the Standardised Approach, the systems will still have to be significantly revamped to cater for the new revised SA.
The other consideration is that model justification and calibration will require at least 10 years of historical data for 75% of the asset classes, according to the new rules for IMA. The issue of data backfilling, linear and scaled proxies will require extensive justification to meet regulatory scrutiny. Additionally, justification for an IMA DRC model will require indirect methods to establish viability, along with a horizon calculator for non-modellable risk factors.
Another major impact is on raw processing power, especially for the IMA. Unlike VaR, FRTB does not promote diversification – it ultimately requires the computation of ES for five liquidity horizons and seven risk classes (IR, CS – non-securitised, CS – securitised, CS – correlations, EQ, commodity and FX), which will need one set of unconstrained runs and several sets of constrained runs in total through the entire client risk portfolio, even allowing for filters. In reality, the number of runs through the system will be rather higher due to other requirements such as Stressed ES, DRC, 1-Day Hypothetical Backtesting and 1-Day Actual Backtesting.
The CVA calculation processes are also likely to be affected – these are already computationally-intensive and the enhanced horizons for FRTB add to the regulatory processing workload due to the extended scenarios.
Furthermore, there may be an impact on trading limits and risk appetites if the organisation elects to standardise on ES as the limit management metric.