The Basel Committee on Banking Supervision’s final revisions to the FRTB guidelines aim to address industry concerns around complexity and capital implications. A forum of industry leaders discusses whether the changes have been effective and how banks are coping with potential variations in regional implementation, as well as the technological, operational and financial challenges of the post-Libor transition
- Bruno Castor, Head of Market Risk, Murex
- Eugene Stern, Global Market Risk Product Manager, Bloomberg
- Hany Farag, Senior Director, CIBC
- Hjalmar Schröder, Head of Market Risk, Zürcher Kantonalbank
- Suman Datta, Head of Portfolio Quantitative Research, Lloyds Bank Commercial Banking
- Azar Khurshid, Director, Global Market Risk Management, Mizuho International
To what extent have the Basel Committee on Banking Supervision’s final revisions to the FRTB guidelines addressed banks’ previous concerns? Which areas still need clarification?
Bruno Castor, Murex: The amendments to FRTB released in January are a large step in the right direction. The complete revamp of the profit and loss attribution (PLA) test – including the introduction of more robust metrics and new green, amber and red zones – has made the internal models approach (IMA) a realistic option for banks that were previously constrained in adopting the standardised approach (SA) because of temporary breaches.
Additionally, the simplification of the risk factor eligibility test (RFET) – in terms of both detail and breadth – has been received positively. The diversification benefit correlation introduced in the stressed capital add-on charge should also have a positive impact.
While progress has been made, certain aspects of this regulation remain unclear at a jurisdictional level, and further regional guidelines are expected in the coming weeks.
Eugene Stern, Bloomberg: Two key areas saw changes welcomed by the industry:
- The PLA test: the Basel Committee provided the industry with a number of revisions, including the metrics that will now apply (Spearman correlation and Kolmogorov-Smirnov tests) and how frequently the test is applied (quarterly instead of monthly). The Basel Committee also introduced the new amber zone as part of the traffic-light approach to mitigate potential cliff-edge effects of failing the test.
- The RFET: the Basel Committee made changes to the test for classification of risk factors between modellable and non-modellable to alleviate the effects of seasonality which the industry previously identified. While the Basel Committee chose to retain the minimum number of 24 real price observations, it replaced the maximum one-month gap threshold with a relaxed requirement of evidencing a minimum of four observations in any 90-day period. For risk factors that fail this test, the revised rules include an alternative criteria enabling banks to satisfy the RFET by evidencing a minimum of 100 observations without meeting any gap or interval requirements.
The committee also confirmed that only one observation should be counted a day, and then separately clarified the use of committed quotes.
Despite the changes incorporated in the finalised framework, many in the industry argue that the changes do not go far enough in addressing seasonality, and while those risk factors are relatively liquid, they may still be at risk of failing the test as currently specified.
The Basel Committee incorporated seven qualitative criteria for the modellability of risk factors that pass the RFET. These much-discussed principles will form an integral part of the data governance required for FRTB. For risk factors that fail the test – non-modellable risk factors (NMRFs) – the Basel Committee accommodated a shortening of the maximum liquidity horizon to be taken into account when computing stressed capital add-on.
Despite the finalisation, there remain areas where the industry will require further clarification to facilitate a smooth implementation. This includes where a look-through approach is required for determining trading book eligibility and treatment of positions linked to funds, indexes or with multi-underlyings.
Hjalmar Schröder, Zürcher Kantonalbank: The Basel Committee has struck a fair balance between banks’ valid concerns and the need to keep the framework coherent, sufficiently conservative and not overly complex. With regard to the SA, the calibration of the low correlation scenario and the ambiguity around the treatment of equity index options have been addressed. However, one of the few remaining shortcomings is the lack of recognition for non-linear hedges against linear risks, which is the price to pay for the segregation between delta and curvature aggregation.
Hany Farag, CIBC: The Basel Committee has addressed many concerns among market participants, and the framework has become quite reasonable. There is always an appetite for reducing capital implications further or simplifying the framework even more. However, the Basel Committee’s response to industry concerns has been constructive and quite reasonable. We have a framework that is risk-sensitive and manageable.
Azar Khurshid, Mizuho International: General consensus is that the final revisions are well aligned to the concerns of the industry and of national regulators. For instance, the inclusion of index buckets for equity and credit risk classes, the relaxation of the look-through requirements and changes to the stressed expected shortfall (SES) risk charge calculations have been helpful. So too have the reconfiguration of the RFETs, the introduction of the base currency approach for calculating foreign exchange risk class capital, and the adoption of lower risk weights for general interest rate risk and credit risk classes for the SA.
These improvements mean the new regulatory regime will be closer to expectations than before. It is also helpful that the text now includes frequently asked questions (FAQs) and a fully hyperlinked web version.
Areas that need further clarification include the IMA; SA treatment of equity investment funds and regulated collective investment undertakings; treatment of curvature adjustment and the base currency approach for forex risk share for the SA; and certainty over RFET conditions, which could mean a significant narrowing of the group of eligible risk factors.
Most of these topics could and should be addressed in subsequent clarifications through FAQs and regulatory technical standards at Basel and national regulatory levels.
Suman Datta, Lloyds Bank Commercial Banking: The Basel Committee has addressed a lot of concerns and there has been a healthy dialogue via mechanisms such as quantitative impact studies. The main regulatory agenda is now set so now the question is about implementation. The question of timing is crucial. The Basel Committee has given its view, so it is now up to national regulators to confirm the timelines and give banks clarity over their own planning.
The other challenge is Libor. Regulators will need to understand the added complexity and support the industry in Libor transitioning alongside FRTB.
What will determine banks’ choice between the SA or IMA?
Eugene Stern: Banks considering the IMA are comparing its benefits – such as better risk management tools and capital relief – with its costs. These include both the direct cost of implementation and the uncertainty of being able to pass the tests around price observations (RFET) and ongoing front-middle office alignment (PLAT).
The cost of uncertainty is hard to quantify, but is a significant component of many banks’ thinking. Banks don’t like being unable to predict what capital add-ons may suddenly arise from any failed tests.
The direct costs begin with risk infrastructure, as banks will need to align data and analytics across multiple asset classes and risk types on a consistent platform. Banks running the IMA may also need to enhance throughput on their calculation grid, as they might need to run up to 15 or 20 times as many simulations as at present. The RFET – which requires banks to prove that the time series used in their risk models are based on observed prices – pushes banks to upgrade the infrastructure they use to track their own trades and connect it with the risk workflow, source data from outside, or more likely both.
There is also a cost around knowledge and/or staffing, and many smaller banks in particular don’t think they have the expertise to assess and address the issues implied by IMA.
On the other hand, banks also have several incentives to seek IMA approval beyond just capital relief. Some see IMA status as a competitive signal to the marketplace, connoting sophisticated risk management in line with best practices. Others feel less competitive pressure around IMA, but expect regulatory pressure to implement it – at least for core businesses or the most material risks. One issue here is that the books for which IMA might be most needed are also the books where implementation may be the most challenging.
Hany Farag: There is significant benefit to be derived from the IMA. The regulators are serious in their calibration of a healthy ratio of SA to IMA – believed to be around 1.5x. It is possible that some banks might be bound by the global output floor but, for the most part, banks should be able to benefit from the IMA. In the long term, however, it is not clear how to run a sophisticated industry with trillions of dollars in trading without the IMA. There may be inertia for some banks to opt for the SA to begin with, to avoid the operational burden. In the long term, however, the curve will be less steep and I predict the percentage of banks using the IMA will likely be comparable to what it is today.
Bruno Castor: For financial institutions that need to comply with FRTB, cost will have the most significant influence on the decision-making process. Following the release of the final FRTB guidelines in January, many banks that previously considered the IMA out of reach are reassessing their situations.
Comparing the 2016 guidelines with those published this year, there has been a significant increase in the SA capital charge. The capital requirements for foreign exchange risk outlined in the most recent iteration of FRTB have, in some cases, almost doubled. With regard to equity trading, there has been a rise in the risk charges for equity risk factors, particularly for large emerging market economy class 2 stocks. Moreover, correlation trading desks will face higher capital charges as they will be prevented from offsetting credit default swap indexes against single-named constituents.
On top of these changes from the regulator, banks are now benefiting from the knowledge gained by first-movers in the market. Looking at banks that are advanced on their FRTB journey, it is interesting to note that the incremental cost of being IMA-ready is less than market participants predicted, particularly when included as part of an overall review of the enterprise market risk framework.
The leading enterprise risk management systems on the market today already include the main components required to calculate FRTB-IMA. To gain IMA approval, the next step is risk factor classification and modellability. If the prerequisites are – or are soon to be – in place, pre‑packaged FRTB-IMA business solutions can be implemented within a matter of months to provide global expected shortfall and SES. This enables banks to perform what-if analyses and to have the business option of switching to the IMA for an individual desk, for example, should an attractive business opportunity emerge.
Finally, for some financial institutions, the market risk capital charge seems less significant compared with the credit or liquidity capital charges – and is therefore arguably less important for this group.
Suman Datta: FRTB is trying to connect risk-taking activity with capital consumption, so it’s a business decision based on risk versus return cost dynamics as to which desks should run on the SA or IMA. Each bank will have its own strategy based on its range of trading desks, asset classes and individual competitive advantages.
The IMA is more complex and has extra operational overhead costs, but it makes sense to have it for risk management purposes and peace of mind.
There’s also a perception that it’s all or nothing in terms of applications for model approval. The SA is the baseline requirement, but not every desk has to run the IMA from day one. You can operate a staggered introduction based on risk profile, materiality and strategic business priorities.
Hjalmar Schröder: The first step is an assessment of whether the risk profile of a bank’s trading operations can be adequately captured by the methodology of the SA. If that is not the case, there isn’t really a choice.
Where the SA is a viable option – as Zürcher Kantonalbank concluded for itself – it should be approached as a sober business case consideration, where the advantages of superior risk insight and efficient allocation of capital need to be weighed against the investment and operational costs of an internal model.
Azar Khurshid: This will be based on each bank’s own view, the make-up of its portfolios, current regulatory approval, quality of models, infrastructure and national regulators. This is likely to evolve over time, depending on any transition arrangements at national level.
While current regulatory approval for the internal model is not a guide, we would expect most current IMA waiver banks to seriously consider adding FRTB IMA application to significant parts of their portfolio.
What are the main challenges around implementation and what will be the impact of post-Libor transition?
Hany Farag: The NMRF framework has been dramatically improved. I do not believe the post-Libor transition will be a major challenge. The challenge is a technological build, operational sophistication and certainly a financial investment. For a multi-trillion dollar industry, this seems a reasonable investment for the next decades. It is challenging but achievable.
Suman Datta: Libor represents a major structural change in the market, and the proposed deadlines are close to those for FRTB. There has been a lot of talk of NMRFs and the potential for higher capital charges, but there’s also a logistical challenge: how do you embed such a structural change in transitioning from present state to future state under FRTB? At some point you will need to make decisions such as how you transition the value-at-risk (VAR) model. If you’re running FRTB and Libor as two different streams, you need to work out when they need to converge.
Eugene Stern: The convergence of timelines for the Libor/risk-free rate (RFR) transition and implementation of FRTB around 2022 compound the challenges faced by banks in planning for these changes.
For a trading desk planning to seek approval for the IMA, FRTB challenges include:
- Where to obtain sufficient time-series data for the RFR risk factors and method for backfilling based on proxies where necessary
- The remaining trading book risks exposed to legacy Libor-based risk factors the desk may be exposed to, and the risk of not being able to source sufficient real price observations to avoid NMRF consequences.
Beyond Libor replacement and RFET/NMRF, the RFET poses both systemic and organisational challenges. We’ve heard from a number of banks considering the IMA that the organisational challenges should not be underestimated.
For the SA, a challenge for banks that may have all the analytics in place is the required risk bucketing. Bloomberg has received a number of questions around this – even from some larger banks.
Hjalmar Schröder: From a project perspective, one of the key challenges has been working with a fluid set of requirements that changed as soon as the answers and revisions to the FAQs were published. Adopting an agile approach during development has proven very helpful in this environment.
As far as content is concerned, key challenges have been the alignment between profit and loss (P&L) and risk systems as well as managing the data needed for correctly mapping our positions to the risk factors of the SA.
While the transition away from Libor simplifies the risk factor landscapes by consolidating the different reference tenors, banks must deal with the lack of historical time series for the new benchmark rates.
Bruno Castor: To meet the FRTB implementation deadline, the first step is investing in the right technology. Murex is seeing more and more financial institutions take a holistic approach to compliance, partnering with technology vendors that offer a single platform to address regulatory demands and risk management requirements. The key challenges facing banks include a gap in knowledge and experience, the decommissioning of legacy systems and an evolving capital markets landscape.
Murex has set up a global team of FRTB experts to share the learnings from client projects that are already live, as well as those coming down the road. The more FRTB projects we undertake, the more we discover issues of mutual interest across our client base – whether it be interpretation of the detailed rules, data mapping and classification approaches, or the day-to-day operational process setup.
Often, FRTB compliance requires investing in new technology while also decommissioning legacy risk systems that cannot meet new and evolving regulatory demands. Banks need to balance these two projects, injecting new technology and functionality as soon as possible while untangling a web of existing systems. To overcome this challenge, it is important to partner with a technology provider that includes this additional element in the FRTB project structure from the very beginning.
Many capital markets participants are working on addressing Libor reform at the same time as FRTB. They are currently deeply involved in the impact analysis, and interplay with the FRTB implementation timeline is clear. Having a common interest rate curve framework across trading and risk for all asset classes, along with harmonised pricing, is a good place to start.
Are the timelines realistic? What schedule should banks be working to?
Hjalmar Schröder: Having opted for an early project start, the Basel Committee’s timelines appear realistic for Zürcher Kantonalbank. When setting their road maps, banks should allow sufficient time to familiarise themselves with the behaviour of the new methodology in a real-world environment and to allow the business processes around the new standards settle in.
Hany Farag: Overall, the timelines of 2022 are reasonable, and the fact there is a subsequent year for PLA to become fully binding adds a fair cushion as well. The target should be 2022. Sophisticated programmes of this type cannot afford to aim lower than expected.
Azar Khurshid: While most global systemically important banks seem to have fairly well advanced FRTB programmes, the scale of the challenge is not just in producing the calculation results. There are implied changes to the operating model that will also need to be agreed and bedded down, including increased monitoring of the trading book/banking book boundary, establishing workflows for calibration of NMRFs, SES, and so on, as well as aligning the front office with the risk models and data.
National regulators will also need to be prepared to process a significant number of applications. We are hoping to make a single application for all IMA desks at the same time to reduce the burden and turnaround time.
Bruno Castor: For many financial institutions, the road to FRTB compliance began several years ago. Due to regulatory uncertainty, this path has not been smooth, with some pausing projects while waiting for clarification and guidance. Others decided to move forward steadily, using FRTB as an initial driver for more large-scale IT transformation projects. Following the publication of the final guidelines, the market is heading towards the 2022 deadline with increased urgency and, although there are still challenges ahead, I believe the timeline is realistic for most.
At this stage, banks should be aiming to not just tick the boxes of the minimum compliance threshold. Rather, they should be thinking long term. With the right technology partner, FRTB is not only manageable, it can bring a lot of value to your business.
Suman Datta: The timeline is, again, connected with business decisions. If you want to be an IMA bank from day one, it’s a case of deciding where you want to be and working back – ensuring you have the necessary infrastructure and governance in place in good time to have your model application validated by regulators. If you’re at the back of the application queue, it’s unlikely you’ll be a day-one IMA bank, but the dates should be achievable – particularly if you are phasing the introduction of IMA.
What steps are banks taking to adapt existing systems and processes?
Azar Khurshid: We split this into separate workstreams where possible. Each stream then determines which systems or infrastructure needs to adapt to be compliant. This can be achieved through regular updates – such as for the policy workstream – in-flight projects or FRTB-specific projects. FRTB impacts several traditional functions and departments, so these are multidiscipline workstreams.
Hany Farag: A redesign of existing systems and processes is advisable, and close realignment of front office, risk and finance needs to be implemented. FRTB is quite binary; either you do it well and reap the rewards, or you will wish you had just opted for the fallback SA approach for the time being.
Eugene Stern: Moving to FRTB is a huge project for many banks, and in many cases it starts with assessing budget and putting dedicated teams in place. Once in place, the team has a lot of decisions to make early on: whether to go for IMA approval, and for which desks? Will the bank need to change some of its business structures? Banks will also need to decide how much of their current risk current risk infrastructure will be fit for purpose going forward. They’ll also need to determine what can be adapted versus what needs to be scrapped and replaced altogether. Lastly, banks will have to evaluate whether they can build the necessary solutions internally or must seek out vendors.
The fact that some banks that use the IMA today may move to the SA under FRTB because of the complexity of FRTB-IMA has some interesting consequences for the maintenance and adaptation of risk systems. Banks using the IMA today typically also use VAR/expected shortfall for internal risk management and reporting, and they don’t expect to give up VAR models internally even if they move to FRTB‑SA for regulatory capital. This category, which may include many regional banks, will have to decide how much to align their regulatory and internal risk models and platforms going forward. In particular, some banks are now deciding to stick to the SA for regulatory capital but are simultaneously taking the opportunity to upgrade their VAR systems for internal risk management. They have the budget to upgrade their risk infrastructure because of FRTB, and are taking the opportunity to kill two birds with one stone.
Suman Datta: First, you need to take a step back. FRTB has an impact on systems, analytics, models, methodologies and operating models. Once you have identified the main areas that need attention, you need to get the key people involved to come together. Lloyds decided a couple of years ago to have front-office pricing model-based calculation for our capital under FRTB, and we built a team around it with pricing and risk model experience. Likewise, when you look at the technology stack, you need a blend of skill sets that covers the necessary areas. And when it comes to the operating model, closer alignment between the front office, finance and risk makes it easier to resolve any issues, such as comparing numbers for the PLA test.
My portfolio quantitative research team is a good example of the added value this can bring. There are many other portfolio models in use within the bank – such as those in initial margin, PLA, prudential valuation, independent price verification and stress-testing. Having the same team working across these models to test risk scenarios and calculate P&L for the various portfolios means we can develop a unified platform and consistent analytics and metrics.
Hjalmar Schröder: The investment need that accompanies FRTB has been a catalyst for consolidating the system landscape and pooling consistently defined data across systems.
Another focus is the alignment between risk and P&L – not only on the systems side but also along the axes of organisational structure and business processes.
What problems will banks face in optimising pricing models?
Hany Farag: Some products and pricing models will represent real challenges because of the large number of simulations used for pricing and the number of scenarios generated by risk.
Additionally, some historical shocks – when combined and applied to the current state of the market – can produce scenarios that are difficult or impossible for front-office pricers to consume directly.
Suman Datta: Pricing models are typically built with a certain set of parameters and typically operate within a certain range of these inputs. But they may become stretched under new scenarios in FRTB to a point where they fail to calibrate, or lead to a state of being arbitrageable. Empirical testing is needed upfront to understand the dynamics and model behaviour and address the problem.
The other problem comes with the scenarios you’re generating. Depending on the available data – such as historical time series – scenarios might lead to arbitrageable states or inconsistent and spurious market data states. So how do you generate a realistic scenario? This is again a major piece of work, and you need to do it well in advance, allowing enough time for model owners to work on generating realistic scenarios and potentially address issues in historical time-series data.
Finally, the number of calculations needed means the computational challenge is much higher. You may need to give advance notice to technology partners to ensure capacity is available. This planning needs to happen now.
Azar Khurshid: In the FRTB sense, the optimisation means aligning the risk-theoretical P&L with the hypothetical and actual P&L. This is only possible with a very good understanding of both the pricing models, as well as the PLA. Unfortunately, the more complex products tend to have high number of inputs, some of which may be non-modellable or may become so under stressed periods. Alignment of the front-office models and data with risk also presents its own challenges. For example, the front-office team may not update data or model outputs for no-risk positions, which can lead to gaps in historical data for simulation and calibration purposes.
Hjalmar Schröder: Pricing models can be divided into three categories serving distinct purposes: quoting, marking-to-market and risk management. Banks are in a constant race to optimise their quoting models to win the right transactions at the right price. Mark-to-market pricing models have followed the increase in sophistication to avoid significant differences in transaction pricing. In risk management, however, the primary drivers are consistent dynamics and modellable inputs. Thus, these pricing models have remained more standard. The FRTB requirements for PLA will limit how far models used for mark-to-market can deviate from those used in risk management. Therefore, optimisation of pricing models used in the P&L process will have to be matched to the models used in the risk engine. Where this is not possible, trading desks might have to get used to more P&L noise between quoting and revaluation.
How will banks cope with potential variations in regional implementation?
Hjalmar Schröder: While a globally homogenous implementation of standards would be highly desirable, the degree of regional variation that will be introduced by national regulators is hard to predict. Banks should therefore make the adaptability of their processes a key criteria when choosing a vendor product or building an in-house solution.
Azar Khurshid: One of the key motivations for the Basel Committee’s FRTB proposals is to create a level playing field and ensure no regulatory arbitrage. One could therefore argue that regional variation is likely to be minimal. Parameter and treatment interpretation will need to be closely watched by banks that face off with multiple regulators. While variation in interpretation and parameters may be less likely, one would expect differences in the implementation schedule. This is potentially tricky to manage, with banks having to maintain multiple systems and calculations for different regulators at the same time.
Hany Farag: My belief is that the regulators are trying to converge rather than diverge and are not likely to allow for too much variation. It is, however, too soon to judge as regional implementations are not yet fully announced.
The panellists’ responses are in a personal capacity, and the views expressed herein do not necessarily reflect or represent the views of their employing institutions