The Risk Factor Eligibility Test (RFET) uses the availability of market data as a proxy for liquidity risk. It is being introduced following the Basel Committee’s Fundamental Review of the Trading Book (FRTB) – along with other changes to capital adequacy requirements that now form part of Basel III – all of which are due to be implemented in 2023 (following a recent one-year delay to implementation triggered by the Covid-19 crisis).
In layman’s terms, the Basel Committee is saying that banks need a minimum amount of ‘real’ price data to accurately model risks. Without enough ‘real’ data, they will not be allowed to use their own models and must apply punitive capital charges based on stressed market scenarios.
FRTB quantifies what it deems to be ‘enough’ data and the attributes required for data to be considered ‘real’. For a risk factor to be modellable it needs at least 24 real prices per year, with at least four pricing events in each 90-day period.
As for what makes a price ‘real’, it either must stem from an actual transaction or a committed quote. To ensure that this definition cannot be gamed, regulators have specified some caveats. For example, trades must be between arms’ length participants (which excludes intra-group transactions or potential wash trades). Also, if data has been sourced from a vendor, then that vendor must have played a role in processing the transaction and be able to prove it took place. Also, only one pricing event can be counted per day, which should help to reduce instances of double-counting.
In terms of what constitutes a “committed quote”, the Basel Committee notes that these must be prices “at which the provider of the quote must buy or sell the financial instrument” – suggesting that they must be tradable and not merely indicative. Equally, real prices (including both trades and quotes) should be of “non-negligible volume, as compared to usual transaction sizes for the bank” and “reflective of normal market conditions.”
The European Banking Authority recently published technical implementation standards that impact the RFET. It has added further caveats regarding the use of committed quotes in the RFET. Such ‘committed quotes’ can only be counted as a verifiable price should the following criteria be met:
- “[The quote] must have both a firm bid and a firm offer price.”
- “The bid and offer quotes may not necessarily emanate from the same entity, but they must be collected on the same date.”
- “[The quote] should have a reasonably small bid–offer spread”
The EBA has also provided clarification on mapping real prices to risk factors, noting that “any verifiable price may be counted as an observation for all of the risk factors for which it is representative” as long as the institution is capable of extracting the value of the risk factor from the price (for example, extracting implied volatility from the price of an option). Guidance was also provided on how prices for a fixed income instrument could be reallocated (under certain conditions) as its maturity shortens, and how to apply the RFET to objects that are defined parametrically (curves, surfaces and cubes).
Finally, guidance was provided on how banks can deal with the fact that some price data is only made available on a delayed basis. Under these conditions, the period for calculating the RFET can be shifted by up to a month to accommodate the lag in data provision, although this will need to be documented.
Why is the RFET Important?
In an era where exchange-traded instruments trade at the speed of light, one could be mistaken for thinking that the RFET criteria (at least four prices in a 3-month period) would be relatively easy to meet. However, across OTC markets it is clear a large range of risk factors will struggle to be eligible for internal models.
A number of industry studies conducted over the last few of years have looked to quantify the challenge of meeting the RFET. These studies are based on slightly outdated information, given that the RFET parameters have changed over time. However, they all emphasize the seriousness of the challenge ahead, as illustrated by the following quotes:
- “The NMRF (non-modellable risk factor) charge has the potential to overwhelm other components of IMA” – ISDA Industry Response to BCBS Consultation
- “[Only] half of bond issues would fulfil requirements for continuously available ‘real’ prices” – European Banking Association Report which studied 80 corporate bonds that were part of the ITRAXX 125 index
- “An industry survey conducted by Oliver Wyman has suggested that banks expect the NMRF SES charge to account for 30% to 50% of their total internal market risk capital.”
- A study of USD-linked Vanilla FX options found that only 22 percent passed the RFET; another of the US Corporate bond market found that only 36 percent passed the RFET – research by Thomson Reuters
Certain test criteria have eased since these studies were conducted. For example, the need for at least four real prices for each 90-day period had originally been a requirement to have no more than 30-days between pricing events (which would have been difficult to meet for markets that experience seasonal dips in trading activity).
However, other developments may make the RFET harder to pass. One example is the EBA requirement for committed quotes to be 2-sided and with a reasonable bid-offer spread. Another aspect that is more difficult to predict is the transition away from Libor as a key interest rate benchmark, which could lead to fragmentation and fewer ‘real’ prices per risk factor/bucket.
How Can Banks Minimize the Impact of the RFET?
The key to solving the RFET and minimizing the cost of non-modellable risk factors (NMRFs) lies in better data management. The tests encompass a wide variety of processes.
Banks will need to source all relevant price data; normalise and map that data to risk factors / buckets; run relevant validation rules to ensure ‘real’ price criteria are met; run the RFET itself; monitor test results; alert when risk factors approach non-modellable status; and source additional committed quote data, where possible, to prevent that from happening.
The criteria for determining ‘real’ prices have yet to be tested in anger, particularly when it comes to ‘committed quotes’, which will be a key factor in determining risk factor eligibility. Given the cost implications of failing the RFET, much attention is likely to go into solving this challenge over the coming years.
How Does the Recent Market Turmoil (Relating to Covid-19) Impact Implementation of Basel III and FRTB?
The most direct implication of the recent market turmoil has been the decision to delay implementation of Basel III by a year to January 2023. The Governors and Heads of Supervision (GHOS) of the Basel Committee announced this delay on the 27th of March, explaining that the decision was taken to ensure the financial system could focus their full attention on responding to the impact of Covid-19.
An ancillary impact of the turmoil is that the market data from this crisis will prove relevant to stressed market model calibration (another requirement of FRTB/Basel III). Prior to this recent downturn, the most recent period of significant market stress had been the financial crisis of 2007/2008. The availability of more recent data to represent ‘stressed market’ conditions may make things easier from an operational perspective. However, the speed and severity of the downturn could have a punitive impact on capital requirements – although we have yet to see the results of quantitative impact studies to confirm this.
Can Data Used for the RFET be applied to Calibrate Models?
Under the EBA technical standards consultation, one question asked to respondents was whether data used for the risk factor eligibility test (RFET) could be re-applied to calibrate models. The response was unanimous in finding that there is little overlap between data used for assessing modellability (and running the RFET) and that used for model calibration. Model calibration data typically requires a higher frequency of updates (daily) measured at consistent cut-off times, so may therefore continue to be managed separately, with the RFET conducted as a standalone test on its own separate data set.
Xenomorph specializes in risk data management solutions and can help minimize the impact of the RFET and address other data management challenges posed by FRTB (as well as other regulations).