A few thoughts after a conference dedicated to the FRTB
Last week I took part in the 4th “Impact of Fundamental Review of the Trading Book” conference in London, and I would like to share with you some feedback and insight we gathered on this occasion. The conference was well attended by European and Japanese banks, with high-level executives and engineers from risk management and modelling teams. American banks however were conspicuously absent. Perhaps they hope the new US Administration will abandon FRTB or at least ease the pain. Among very interesting discussions, I thought a few points stood out as particularly insightful.
Should we wait for another year before doing anything?
Giles Artaud (CA-CIB), chair of the seminar, announced that European regulators have pushed back the implementation of FRTB by 15 months. I could immediately see a sense of relief among the delegates whose organisations were obviously late in their planning. This was short lived. Early estimations of the impact of the capital risk charge at a number of banks’ using the Standard Approach showed a fourfold increase when compared to current reporting.
As a result, there are intense discussions underway to amend the rules and come up with an improved calibration methodology that would stay within the factor 2 that the Basel Committee initially aimed for. In the meantime however, it seems clear that banks need to be ready to change their capital allocations or reorganise their business way ahead of the deadline, as if those discussions do not bear fruit they might be hit by a large increase in their capital risk charge. It means that, in any case, banks need to have an infrastructure to run the SA model on their actual business by early 2018 at the latest.
Further detailed presentations about the challenges of defining the desk granularity and boundaries made it clear that even implementing the Standard Approach required long periods of testing and simulations, particularly if the bank wishes to use the Internal Model Approach as a ratio of the SA Capital risk charge will be used as a floor for IMA. The exact ratio is still the subject of debate, but the consensus is that there will definitely be one. One particular presentation focused on the probability of a desk losing their IMA eligibility. Long periods of analysis with real data are required to feed a proper model and help decide if it is worth investing in the IMA as opposed to just sticking with the Standard Approach, as losing a desk eligibility and having to revert to SA could be a very costly event indeed.
What about the hardware infrastructure?
When sizing up the effort required to implement the FRTB, one has to consider that it won’t be like a normal project, where you have design, development and UAT phases as well as a short parallel period before you can replace the current system. For the reasons explained above, one has to factor in a long period of parallel run – I estimate between one year and 18 months. This raises the question of infrastructure: should you replicate the entire architecture on duplicate hardware with grid for sensitivity and P&L scenarios calculations and hardware for aggregation and reporting? To that my reply would be that this looks like a perfect case for trying a Cloud deployment.
For a start, you won’t need to run the new calculations every day, at least at the beginning. You may start with a few ad-hoc weekly runs until you have all your risk factors ready, then you can progressively make them more frequent. You also need to be able to run scenarios on demands to test new organisation or business changes. A Cloud deployment is the perfect solution for that, providing the elasticity you need at a moderate cost, especially when you do not use it 24×7.
Needless to say, at ActiveViam we are ready to help you by delivering a short time-to-market solution which implies minimum changes to your risk infrastructure with our FRTB Accelerator package. Furthermore, our full solution is already available on the Cloud.