The Fundamental Review of the Trading Book (FRTB): An Introductory Guide

  • This blog, the first in a series on topics related to the Fundamental Review of the Trading Book (“FRTB”), provides an overview of the Basel Committee’s sweeping changes to bank capital standards for market risk, changes that are likely to be the subject of a proposed rulemaking by U.S. banking regulators in the coming months.
  • Specifically, we discuss the origins of the FRTB and the objectives of the regulators that designed it, why the subject is receiving increasing attention as implementation moves ever closer and provide an overview of the core elements of the reform package.

The Fundamental Review of the Trading Book (“FRTB”) was initiated by the Basel Committee on Banking Supervision (“BCBS”) in the years following the Great Financial Crisis (“GFC”) of 2007-2009, with the aim of completely revising the approach to calculating risk-based capital requirements for trading activities (i.e., “market risk capital”).

As we will discuss in forthcoming blogs, this complex series of reforms will likely have far-reaching impacts not only on the trading business models of large banks, but also on liquidity provision in key funding markets and therefore the ability of certain non-financial end-users to raise funds.  These impacts are inevitable given that 73 percent of the funding for U.S. non-financial corporations is generated by the U.S. capital markets[1], and because banking organizations remain crucial providers of capital markets services to non-financial end-users.  Given these potentially far-reaching impacts, it is important for us to first understand what the FRTB was designed to do, why we are hearing more about it now, and what is contained in the package of reforms.

What were regulators’ goals in undertaking the FRTB?

In the immediate wake of the GFC, the BCBS introduced the so-called “Basel II.5” package of reforms, which included a significant increase in the market-risk capital standards for banks. While increasing the overall quantum of capital required for market risk and making other modifications to the calculation of market risk capital (particularly the introduction of the Incremental Risk Charge and the Stressed Value-at-Risk “VaR”) , the reforms were seen mostly as a stop gap measure until the underlying structural problems with market risk capital standards could be properly addressed by the Committee, which would come later in the form of the more comprehensive FRTB.

Those structural problems included the lack of a clearly defined boundary between the trading book and the banking book, which provided opportunities for arbitrage between books to obtain more favorable capital treatment for specific instruments or portfolios. Another important concern involved the weaknesses associated with the existing VaR approach to modelling risk. While this approach did a good job of modeling risk under normal market conditions, it performed poorly during periods of market volatility, such as those that occurred during the GFC (i.e., put differently, it did a poor job of capturing so-called “tail risks”). The addition of a “stressed VaR” measure in the Basel II.5 framework was seen as insufficient to compensate for these weaknesses.

A further structural deficiency of the existing market risk capital framework was its failure to consider the relative liquidity of trading book positions and the risks of market illiquidity.  The existing framework was based on liquidity horizon assumption of ten days, ignoring the fact that many positions may take far longer to liquidate in practice (and particularly during periods of market stress).  Other identified problems with the existing framework included a lack of transparency and comparability between the internal models and standardized approaches, particularly in areas such as hedging and diversification, as well as the more favorable treatment of credit risk in the trading book, which had led banks putting an excessive number of their credit positions in the trading rather than the banking book.

To address these deficiencies, the BCBS issued a consultation on the FRTB in 2012. This was followed by further consultation exercises in 2013 and 2014, an initial set of standards in 2016 and subsequent modifications to those standards that were ultimately finalized in early 2019 (see timeline in Figure 1 below).

Throughout, the core goals of the FRTB initiative were to:

  1. a) create a clear regulatory boundary between the trading and banking books;
  2. b) replace the VaR approach to risk measurement with a more comprehensive alternative known as Expected Shortfall (“ES”);
  3. c) revise the Standardized Approach (“SA”) to make it more risk-sensitive and allowing it to act as a credible fallback Internal Models Approach (“IMA”);
  4. d) replace the static 10-day liquidity horizon assumed under the VaR framework with varying liquidity horizons in the IMA;
  5. e) introduce a new capital add-on for risk factors that fail modelability tests, known as Non-Modellable Risk Factors (“NMRFs”); and
  6. f) create a new and more robust approvals processes for obtaining regulatory approval for IMA use and requiring these approvals to occur at the level of individual desks rather than granting firm-wide approval.

Although not a stated objective of the BCBS, industry quantitative impact study (“QIS”) exercises have demonstrated[2], that the market risk changes will also likely increase the aggregate level of market related RWA in the US banking system if implemented without any modifications.  This impact would be distributed unevenly, imposing a significant capital penalty on banks with large trading books. Depending on how it is implemented, the FRTB could also have a disproportionate impact on some types of products/markets versus others. For example, the way securitization exposures are calculated under the Basel FRTB standards could discourage market making in that sector. Similarly, less liquid markets (e.g., certain corporate bonds and emerging markets) could also see a reduction in market making activity; this is because these types of products/markets are more like to fail the FRTB’s modelability tests, resulting in more punitive capital treatment under the Basel standards. We will discuss these potential impacts among others in future blogs.

When will the FRTB be implemented?

While the global standards have been finalized at the Basel level, important outstanding questions remain about the implementation of the final rules in each jurisdiction. Perhaps the most obvious uncertainty concerns the timing of FRTB implementation.  The final BCBS standards set January 1, 2022, as the deadline for national implementation, but in March of last year the BCBS agreed to push back that date to January 1, 2023, to give banks additional time to respond to the COVID-19 crisis.  Given the complexity of the FRTB changes (along with the remaining Basel III Endgame reforms to other RWA requirements), it was widely assumed that proposed rulemakings would be issued in all the major jurisdictions by the end of 2020 at the latest.  This timeframe was seen as necessary to provide for significant public input on the proposed changes and to ensure that there was an adequate amount of time between finalization of national rules and the final go-live date for banks to build, test and validate the necessary systems and models required to implement the FRTB.

However, as of late August 2021, only a few jurisdictions (such as Hong Kong) appear to be on track for full implementation by the 2023 deadline.  The EU has issued implementing regulations relating to the FRTB, with standardized reporting requirements due to come into effect by September 2021 and full implementation pushed out to 2024.  Of note, the U.S. banking regulators had yet to issue a formal FRTB proposal, or indeed a proposal to implement other elements of the Basel III Endgame reforms, having focused much of their attention over the past year on dealing with the response to the pandemic.  This strongly suggests that the timeline for U.S. implementation (and likely most other jurisdictions) will extend into 2024 or possibly later.  While a further delay is viewed as almost inevitable, neither the BCBS nor U.S. regulators have, as of this writing, confirmed a new implementation date for the FRTB.

Although much of the focus has been on the implementation timeline, important substantive issues remain undecided.  While the FRTB is an internationally agreed minimum standard, national regulators are afforded flexibility and discretion in how they implement it (discretion that has often been used by U.S. regulators to “gold plate” other Basel Accords).  In recent public comments made at the BPI-SIFMA Prudential Conference[3], Federal Reserve General Counsel Mark Van Der Weide committed to a “robust and faithful” implementation of the FRTB and the related Basel III endgame proposals, but also acknowledged that the agencies retained the flexibility to modify the standards to fit unique U.S. legal requirements and business/market structures, as well as to determine the scope of applicability of the reforms.  We will discuss some of the outstanding substantive issues later in this blog series.

What are the core components of the FRTB?

The finalized BCBS framework outlines two approaches that firms can adopt to calculate their market risk capital requirements – the Standardized Approach (“SA”) and Internal Models Approach (“IMA”).  All banks must implement the SA; however, some firms may additionally opt for the IMA subject to initial and continued conformance to stringent model performance standards.  Consistent with the goal of establishing a clearer boundary between the credit and trading books, the final rule also significantly revised the approach to classifying firm positions as either trading or banking exposures.  Below is a high-level description and graphic of each of the key elements of the final FRTB standards.

Data, Reporting and Testing Changes

In the FRTB, the BCBS makes changes to the way market risk data is gathered, reported, monitored, and verified for accuracy. At a high-level, these changes include:

  • Implementing a more granular desk level model review and approval process;
  • Requiring desk level profit and loss attribution tests (“PLAT”) and back testing to be performed on a daily basis;
  • Expanding the substance of intra-day monitoring and measurement of market risk; and
  • Applying more granular assessments of model performance.

Trading Book Definitions

As noted above, one of the main concerns about the existing market risk framework was that it was insufficiently prescriptive in its classification of trading instruments, allowing for significant arbitrage to occur between the trading and banking books, and leading to inconsistent interpretations between firms.  The BCBS addressed their concerns in the FRTB framework by:

  • Defining a new boundary based on whether the bank intends to trade an asset or hold it to maturity;
  • Providing a presumptive list of trading book assets and Regulatory Trading Desks (“RTDs”);
  • Imposing more restrictive rules for internal transfers between bank and trading booking, including prohibiting any capital relief as a result of reclassification;
  • Placing a restriction on recognition of internal risk transfers (only external risk transfers are recognized under the framework); and
  • Imposing a more robust framework for hedging recognition to focus on their effectiveness during periods of stress.

Standardized Approach (SA)

The SA is the default market capital requirement for all banks that fall within the scope of the FRTB, regardless of whether a firm pursues IMA accreditation for the permitted asset classes on a desk-by-desk basis.  Importantly, because the IMA approach is subject to initial and continual performance testing, the SA serves as a fallback if a firm fails to meet the performance goals and is the only approach available for correlation trading.[4] As noted above, the SA is significantly more risk sensitive than current standardized approaches.  However, that enhanced risk sensitivity – which occurs primarily through the introduction of a new Sensitivities-Based Approach (“SBA”) – is reliant on data obtained from a bank’s pricing models to derive capital requirements. That means that many firms will have to build significant new modelling capacities to meet the requirements.

The SA capital risk charge is comprised of three components: the SBA, the Default Risk Charge (“DRC”) and the Residual Risk Add-on (“RRA”).  The SA capital charge is calculated using the sum of the SBA, DRC and, if applicable, the RRA charge.

Sensitivities-Based Method (“SBA”)

The SBA is a form of Parametric VaR with regulatory weights and correlations.  Banks are required to compute risk factors (i.e., observable or measurable market data that is likely to influence the valuation and therefore the profit and loss (“P&L”) generated by a financial instrument) sensitivities (Delta, Vega, and Curvature)[5] for seven regulatory defined risk classes.[6]  These risks are then aggregated across buckets, which are sets of instruments of the same risk class sharing the same characteristics and therefore a similar risk profile (a bucket may mean, in this context, a currency or commodities).  Moreover, to capture changes in correlations due to financial stresses, banks are required to use three different correlation scenarios (low, medium, and high) to calculate three separate risk charges for each of the regulatory defined risk classes.  Following the calculation methodology, the firm sums the all the risk class charges and then uses the largest as the SBA charge (see Figure 3 below for a visual representation of this process).

Default Risk Charge (“DRC”)

The framework includes a DRC to capture “jump-to-default risk” i.e., credit spread risk.  Only certain portfolio types are exposed to this type of risk, so instead of seven risk classes under the SBA, the risk classes are reduced to three under the DRC: debt instruments, equity products, and securitizations.  To determine the aggregate DRC, firms are required to apply differing risk weights, formulas, and expectations regarding netting across the three asset classes.

Residual Risk Add-on (“RRA”)

The final component of the SA is the RRA, which is designed to capitalize exposures that are not adequately captured by the SBA.  Generally, this charge is applied to more complex transactions where the exposure is difficult to quantify (e.g., products in incomplete markets, products having correlation risk).  The methodology applies a simplistic formula that aggregates all residual instruments’ gross notional amount multiplied by a risk weight of 0.1 percent of or in the case of exotic (i.e., highly complex) instruments, 1.0 percent.

The Internal Models Approach (IMA)

The IMA fundamentally changes the approach to modeling market risk by reforming the measurement methodology and expanding model performance requirements.  Most importantly, IMA replaces the long standing VaR-based approach for quantifying market risk with an Expected Shortfall (“ES”) approach.  As noted above, while VaR does a good job of capturing risk in normal markets, an ES approach is more effective in capturing the “fatter tail” distributions of risk typical of stressed markets when multiple asset classes move in tandem.

The IMA also removes the Incremental Risk Charge introduced in Basel 2.5 and replaces it with a Default Risk Charge (referred to here as an “IMA DRC”), which acts (as it does under the SA) to capture losses that stem from an obliger defaulting.  It places limits on the benefits of hedging, netting and diversification and no longer permits correlation trading positions capital to be estimated with internal models.  Lastly, the framework implements asset class-specific liquidity horizons which more rigorously incorporates the impact of stress periods; and introduces the concept and approach for non-modellable risk factors.

The use of internal models and their on-going performance is subject to considerable supervisory scrutiny under the FRTB as well as subject to quantitative requirements.  Internal models are also now subject to regulatory approval at the trading desk level. For a desk to qualify for IMA, they must demonstrate that a model is supported by adequate P&L attribution tests and back testing results.

P&L attribution is not a new concept to market risk management; however, it is new to the capital framework in the level of specificity that is included in the new framework.  For initial and continued approval, the framework requires daily comparison and analysis of P&L statements generated by a bank’s front-office, also known as hypothetical P&L (“HPL”), with the results produced by middle/back office, also known as risk-theoretical P&L.  The framework applies quantitative limitations on the amount of unexplained P&L and the variability of the unexplained between the two systems.  Failure to meet those requirements results in loss of IMA approval and application of the standardized approach.

Bank testing requirements are equally stringent.  The framework requires a comparison of the VaR measure calibrated to a one-day holding period against each of the Actual P&L (“APL”) and HPL over the prior 12 months.  Specific requirements to be applied at the bank-wide level and trading desk level are set out by the BCBS and back testing of the bank-wide risk model must be based on a VaR measure calibrated at a 99th percentile confidence level.  Finally, the scope of the portfolio subject to bank-wide back testing must be updated quarterly based on the results of the latest trading desk-level back testing, risk factor eligibility and P&L attribution tests.  Again, failure to meet the performance requirements will result in the loss of IMA status.

In short, these new requirements make qualifying for use of IMA an extraordinarily difficult and resource intensive exercise.  For those trading desks that do succeed in obtaining approval to use this approach, there are three components that are used to calculate the IMA capital charge: Expected Shortfall or ES; modellable risks and non-modellable risks; and a default risk charge.  For desks where the bank cannot or chooses not to pursue IMA qualification, they must calculate capital charges using the SA.

Expected Shortfall (“ES”) Method

The ES and VaR are both used to measure portfolio risk.  However, the ES is a significantly more conservative measure, since it measures the expected value of all changes in the portfolio value in the tail of the P&L distribution that exceed the VaR.  To account for this additional conservatism, the BCBS lowered the confidence interval from 99 percent requirement under the VaR approach to 97.5 percent under the ES.  Additionally, the ES must be calculated for a maximum stress period over the bank’s total history of observations verses the current look back period of one year or 270 days.  Consequently, this approach, while superior to VaR for capturing tail risk, is also far more data and resource intensive.

Modellable and Non-Modellable Risk Factors (“NMRF”)

The path to determining the capital charge under the IMA is dependent on the ability of a firm to model risk factors.  A risk factor’s model-ability is determined by a series of quantitative criteria regarding the frequency, observability, and durability of a risk factor’s pricing information.  Importantly, the firm must continually assess the eligibility of these risk factors to be included in the IMA models.  If a firm cannot or no longer can support the model-ability of a risk factor, the firm must apply a capital add on which is calibrated under a stress scenario. Instances of NMRF are more likely in more thinly traded and less liquid markets, and the issue of how NMRF are defined continues to be a subject of significant debate (and will be discussed in a future blog).

IMA Default Risk Charge (“IMA DRC”)

While migration risk is captured in ES, the IMA framework separately capitalizes default risk. The IMA DRC replaces the Incremental Risk Charge and will apply to credit positions as well as equity positions.  The IMA DRC will also include a floor for the probability of default.

Conclusion

The FRTB represents a sweeping overhaul of the way banks calculate their trading risk capital charges and will have wide ranging impacts on the business models of banking organizations and funding markets for many years to come. As U.S. regulators move closer to a proposed rulemaking, we will release a series of blogs on some important outstanding issues that still need to be addressed prior to implementation. These include ongoing concerns around the design of the P&L attribution tests, the NMRF framework, and the interaction between the FRTB and other existing domestic capital requirements.

Dr. Peter Ryan is a Managing Director and Head of International Capital Markets and Prudential Policy at SIFMA.

[1] SIFMA, “2021 Capital Markets Fact Book,” July 2021. Available at: 2021 SIFMA Capital Markets Fact Book.

[2] See, for example, QIS data reported in the International Swaps and Derivatives Association (“ISDA”), Global Financial Markets Association (“GFMA”) and Institute of International Finance (“IIF”) response to the BCBS Consultation Paper on Revisions to the Minimum Capital Requirements for Market Risk, June 2018. Available at: Microsoft Word – BCBS-FRTB_CP_March_2018_Response_Final_wo_sigs (isda.org). The QIS, which was based on a proposed version of the FRTB, found that in the aggregate capital for trading desks under the IMA was 3.21 times larger than capital based on current rules. Subsequent QIS exercises continue to show increased capital requirements arising from the final Basel FRTB standards, though the ultimate impact will be determined by national rules implementing the standards.

[3] Remarks made as part of a panel discussion entitled “the Implementation of the Final Basel III Reforms,” at the 8th Annual SIFMA and BPI Prudential Regulation Conference (held virtually).

[4] An example of this would a credit correlation trade in which traders sell credit default swap indexes (e.g., CDX or ITraxx) and buy offsetting single-name swaps, or vice versa.

[5] The Delta corresponds to the sensitivity of the value of a position to a variation in its spread and is applicable to all products. Vega is equal to the product of the vega and the implied volatility of the option. Curvature risk consists of applying a significant change (i.e., a shock) upwards and downwards to each risk factor.

[6] These risk classes are: General interest rate risk; credit spread risk; credit spread risk: non-correlated securitization; credit spread risk: correlated securitization; equities; commodities; and FX.