FRTB risk programmes should be split into analytics, data and computation components, and front-office desk heads must act like business leaders and assume the daily tasks of reducing complexity and cost, writes David Kelly, partner at Parker Fitzgerald.
The Fundamental review of the trading book (FRTB) presents methodology, data granularity and computation challenges for all banks, regardless of whether they seek internal model approval (IMA) or focus on the sensitivity-based approach. As the 2019 deadline approaches, there is a tendency for groups within risk, finance and technology teams to design a solution based on current infrastructure. We believe this approach needs refining, with the focus moving to a more distributed approach together with a higher level of engagement with the front office.
Multi-year risk change programmes follow a well-trodden path of acquiring transactional and end-of-day market data from front-office systems, and reference data from multiple sources. The movement of data across the globe is followed by data normalisation to enable aggregate risk calculations – value-at-risk, stress, incremental risk charge and credit valuation adjustment. The front office often struggles to use the output from this approach for daily risk management, which is a key supervisory expectation.
Each desk head will need to decide which products provide a realistic return on capital, are robust during periods of stress and present some opportunity for an orderly exit. This requires forensic analysis of their current portfolio to outline which trades are a priority to unwind, and how the capital will accumulate as it absorbs new trades, including hedges. Such prototyping requires multiple recalculation of the portfolio’s capital charge under different scenarios that accurately predict the impact on new trades as calculated by the official overnight run. If the overnight run involves data mappings and enrichments that are opaque to the front office, this exercise is fraught with inaccuracies, and not only is the opportunity to manage capital lost, but the IMA is at risk.
David Kelly, partner at Parker Fitzgerald
For FRTB, the front office will always prefer the agility that comes with building its own prototype – a desktop approach with minimal technology controls and code originated by its own quants. By flushing out data quality issues, it remains the cheapest way to reduce overall capital charge. But it is not a scalable model for the bank’s entire portfolio. These tools are end-user applications, they are not strategic and therefore cannot compete with a large installation. Enabling the current infrastructure to be both forensic for desktop analysis and available for an entire population is likely to dim the lights when it is switched on. In short, the two requirements are incompatible, so multiple and co-ordinated solutions must prevail.
The solution is to decouple the analytics and data components so that the calculations can be done by multiple users on different portfolios.
- Risk to build the analytics for the capital model and store in a quant library that can be called by all users and systems – it must be agnostic to the user and portfolio-applied.
- Centralise the market data time series and make available for scrutiny and governance. The data should be called by the analytics and not from within the user’s code base.
- The front office to centrally define the instruction set for each product on how each will be calculated for end-of-day profit-and-loss attribution and a standardised version of risk that can then map to the market data time series. The standardised view is distinct from the local view traders use intraday.
- Risk and finance to migrate data tagging – liquidity horizons – into the product as defined above for the front office to manage going forward.
This distribution of calculations enables the front office to check the capital consumption pre-deal and not be surprised when the closed deal is part of an overnight run. The splitting of data, analytics and computation greatly reduces the complexity and running costs of the FRTB programme.
The F word
The front office will have to adopt risk sensitivity methodologies in cases where it is not the main ‘owner’. Each desk head will need to be bilingual, as they switch from their local view to an aggregate view. This is a critical component to evidence an effective risk challenge environment where aggregate reports are treated as a legitimate view of risk and where the difference from the local views is understood.
The additional responsibilities pushed back on the front office will require an increase in non-revenue headcount, which signals a move away from the first line being purely income generation to one where the desk head is running a business. Keeping capital costs down through intelligent origination and portfolio management must become as significant a component of the business model as revenue generation and market share-orientated initiatives. It is this shift in the role of the front office that is really the ‘F’ in FRTB.