Trading Book Capital: Basel III Implementation

Thank you for this opportunity to speak with you all today.1

In my July dissent and in an October speech, I made the case that our Endgame proposal amounts to a big leap of faith in the Basel Committee. In developing the Endgame reforms, the Basel Committee made some key design decisions with little or no explanation.2 This lack of rationale has left the U.S. banking regulators unable to defend or even understand important aspects of the Endgame reforms that we now propose to implement. The lack of rationale has raised doubt about the merits of key aspects of the Endgame reforms. The lack of rationale has hindered the public's ability to comment on our proposal. Finally, the lack of rationale has raised doubt that we can achieve legitimacy and consensus in the eventual final rule.

Today, I'll pose the question whether the U.S. banking regulators could get a better outcome—one that earns legitimacy and consensus—by adopting a phased approach that punts on the underdeveloped aspects of the Endgame reforms pending further data collection, analysis, and debate, while proceeding with implementation of the less contested aspects.

Our Objectives 

Backing up a bit, there has been lots of debate about why, or even whether, our Endgame proposal is needed. This proposal has been years in the making and has little, if anything, to do with the events of March. As its “Endgame” moniker suggests, the proposal's basic idea is to finally put an end to 15 years—yes, somehow 15 years—of efforts to solve problems identified by the 2007-2008 financial crisis.3

Of greatest relevance to this audience, our Endgame proposal aims to address weaknesses in the trading book framework,4 and it does seem to me there is work to be done to that end.

One lesson of the financial crisis was that the value-at-risk (VaR) measure, which had been designed to measure the risk of short-term fluctuations in market prices, did not appropriately capitalize low probability tail events, market liquidity risk, or credit risk and, more generally, was not calibrated to a period of significant stress.

Another lesson was that regulators did not have a credible option for withdrawing approval of internal market risk models in part because the standardized approaches did not pose a credible backstop and, in part, because model approval was done at the banking organization-level instead of some subdivision, like the trading desk.

Yet another lesson was that the boundary between the trading and banking books created opportunities for capital arbitrage by incentivizing banks to classify instruments as “held with trading intent” even where there was no regular trading so as to benefit from the reduced capital requirements on the trading book.

The Basel 2.5 reforms patched up some of the most serious weaknesses in the framework,5 but even at the time, the Basel Committee recognized the need for a fundamental review of the trading book.6 After more than a decade of work, the Committee's review culminated in a new market risk capital framework that our Endgame proposal now seeks to implement.7

While the details do really matter (and on that aspect, I have serious concerns), at a high level, I view aspects of the Endgame market risk reforms as potentially addressing lessons learned from the financial crisis.

At a high level, it generally makes sense to me to adopt a risk measure that better estimates tail risk by replacing the current VaR measure with an expected shortfall (ES) measure calibrated to a period of significant stress.

At a high level, it also generally makes sense to me to endeavor to more appropriately capitalize market liquidity risk by replacing the VaR framework's static 10-day liquidity horizon with a set of different liquidity horizons that vary based on the underlying risk factors.

At a high level, it also seems sensible to me to endeavor to give regulators more credible options to disqualify specific models by: (i) shifting the model approval process from the banking organization level to individual trading desks; and (ii) revamping the standardized approach to more closely link it to the models-based approach so that it provides a credible floor to, and fallback for, the internal models.

At a high level, I also see some merit in implementing a revised boundary that seeks to reduce the possibility of arbitrage through stricter limits on switching between banking and trading books and reducing the differences in capital requirements on either side of the boundary.

All in all, at a high level, if the details are done right, these aspects of the Endgame reforms offer the potential to better align capital requirements with the underlying market risks and reduce incentives to take on tail risk or arbitrage the trading book boundary. These strike me as appropriate objectives for our Endgame proposal.8

The Potential for a Phased Approach 

But I think I said “at a high level” six times. Again, the details really matter. And as I said in my July dissent and my October speech, I have deep reservations about certain design decisions made by the Basel Committee with little or no explanation.

I've previously noted concerns with the profit-and-loss attribution test, or what is called the “PLA test.” A bank would use the PLA-test metrics to measure the extent to which its risk management models align with its front office models. Depending on the test metrics, each trading desk would fall into a “green,” “amber” or “red” zone. A desk in the amber zone would have metrics that suggest some moderate divergence between the models. That would then trigger an additional capital requirement for the desk called the “PLA add-on.” A desk in the red zone would be disqualified from using its models. And so, the quantitative thresholds for this traffic light approach do really matter.

But there is very little in the public domain as to how the Basel Committee came up with these thresholds. Indeed, it appears that these thresholds were calibrated using simulated data and that the Basel Committee itself regarded these as an initial guesstimate that remained a work in progress.9

Another underdeveloped design decision was the approach to non-modellable risk factors or “NMRFs.” A risk factor would be modellable only if, among other things, the bank has a sufficient number of real price observations that are representative of the risk factor.10 Specifically, a risk factor would be modellable only if the bank can identify on a quarterly basis either (i) at least one hundred real prices in the previous twelve-month period or (ii) at least twenty-four real prices in the previous twelve-month period if each 90-day period contains at least four real prices.

A risk factor that fails this part of the test would be deemed an NMRF and have its capital requirement determined under a separate process that would provide significantly less recognition for hedging and portfolio diversification relative to the models-based approach.11 This approach to NMRFs generally results in greater capital requirements and so the quantitative thresholds for the risk factor eligibility test do really matter. Here again, there is little in the public domain as to how the Basel Committee came up with these thresholds.

Pulling this together, aspects of the Endgame market risk reforms have the potential to address weaknesses in the current framework. But at the same time, important aspects of the Endgame market risk reforms are underdeveloped and perhaps even fundamentally flawed.

I would be interested in hearing the thoughts of commenters as to whether we should delay implementation of those aspects of the Endgame market risk reforms that are underdeveloped while proceeding with an implementation of the less contested aspects.12

Delaying implementation of the underdeveloped aspects would allow time for a trial run of the rest of the framework. That would afford us an opportunity to gather data and conduct analysis to potentially re-calibrate or otherwise improve the underdeveloped aspects. That also could permit us to disclose through subsequent rulemakings the rationale and evidence for our eventual U.S. implementation.

For example, it might make sense to delay implementation of the PLA test pending further data collection. The UK regulator has already proposed a delayed implementation,13 and, as I noted, the Basel Committee itself suggested it would be important to see how the PLA test worked in practice. To that end, we might consider waiting to calibrate the amber and red zone triggers until after we have collected data on how the new framework actually works. In the meantime, monitoring of each trading desk's models could be done through the back-testing requirements.

Similarly, relating to the concerns regarding NMRFs, it might make sense to delay implementation of the risk factor eligibility test or, alternatively, temporarily adopt reduced quantitative thresholds for triggering an NMRF determination. One particular area of concern has been that the seasonality of markets and low volumes of trading during holiday periods could result in otherwise liquid risk factors inappropriately being deemed NMRFs.14 In an attempt to address these concerns, the Basel Committee replaced the requirement that there be no more than a 30-day gap between real price observations with a requirement of a minimum of four real price observations in every 90-day period.15 But the Basel Committee offered no rationale for why four observations was the right number. There could be merit to using fewer than four observations as the minimum during a trial run, with data collected to potentially re-calibrate the thresholds for the risk factor eligibility test.

Besides the PLA and NMRF parameters, other aspects of the Endgame reforms might also benefit from at least a temporary departure from the Basel Committee's standards to permit a trial period and potential re-calibration. The 0.6 correlation factor for the NMRF's stressed expected shortfall formula appears to lack much in the way of public domain rationale as to its sizing. The same can be said of the 1.0 supervisory parameter, or p factor, for securitization exposures. I'm still unclear as to why client cleared derivatives should be subject to a CVA capital requirement. It might be worth taking a closer look at the calibration of the 10 percent credit conversion factor for unconditionally cancellable lines of credit. It might also make sense to delay implementation of the minimum haircut floors for securities financing transactions pending a rulemaking by the markets regulators that can more directly address the risks associated with excessive leverage outside the banking system.

I say all this well aware that this might not be a welcomed perspective by all and that there would be considerable drawbacks to further delaying a reform effort that is already 15 years in the making. Finality and predictability are important goals that are in tension with a phased implementation of the Endgame market risk framework.

The problem, though, is that the fundamental review of the trading book was, well, quite fundamental. And it seems quite difficult, perhaps impossible, to revamp such a complex framework all at once. In particular, the calibration of some of the key parameters should depend on how the framework actually works in practice.16 If that's right, it appears that something like a trial run, or some other iterative process, is worth serious consideration.

The Rest of Our Endgame Proposal 

I've so far focused on our proposal's approach to the trading book. I am skeptical that my concerns with the proposal's approach to the banking book could be addressed through a phased approach.17

There is also some doubt whether a phased approach would solve problems posed by the operational risk framework, which appears to have more fundamental problems.

I mentioned in my Endgame dissent the Basel Committee's failure to fix the overcapitalization of banks with high fee revenues.18 Since our July vote, I've heard other fundamental problems. For example, there may be some odd implications for client cleared derivatives. Dealers often pass clearinghouse and exchange fees to the client, which are recognized as both revenue and offsetting expenses. Although not generating earnings, these pass-through fees nonetheless would add to the business indicator for service income. There also may be some odd implications for gain-on-sale revenues and other activities that pull forward future revenues into a lump-sum payment.

To pull this to a close, my larger point is that the Endgame debate need not be binary. I support efforts to enhance our regulatory capital framework, including efforts to address known weaknesses in the market risk framework. But I am unable to support rote adoption of the Basel reforms without some validation of the underlying rationale for key design decisions. And I oppose efforts to reverse engineer higher capital requirements without regard to the costs and benefits or the underlying calibration framework. That does, however, seem to leave open an approach that moves to finalize the less contested aspects of the Endgame market risk reforms and then finalizes the rest through future notice-and-comment rulemakings that can rationalize our own U.S. implementations.

• 1

  The views expressed here are my own and not necessarily those of my fellow board members or the FDIC.

• 2

  As many of you know, the final standards issued by the Basel Committee include little, if any, discussion about the public comments on the consultative standards or why the Basel Committee made its key design decisions.

• 3

  Regulatory Capital Rule: Large Banking Organizations and Banking Organizations With Significant Trading Activity, 88 Fed. Reg. 64,028, 64,030 (proposed Sep. 18, 2023) (“The proposal would build on these initial reforms by making additional changes developed in response to the 2007–09 financial crisis and informed by experience since the crisis.”).

• 4

  See generally Basel Committee, Fundamental review of the trading book, Annex 1 (May 2012) (providing a history of trading book capital requirements and one perspective on the lessons from the financial crisis).

• 5

  See Basel Committee, Revisions to the Basel II market risk framework (Feb. 2011). The U.S. banking regulators implemented those Basel 2.5 reforms in August 2012. Risk-Based Capital Guidelines: Market Risk, 77 Fed. Reg. 53,060 (Aug. 30, 2012).

• 6

  See generally Basel Committee, Fundamental review of the trading book, Annex 1, at 50 (May 2012) (“In 2009, the Committee also initiated a fundamental review of the trading book framework (the 'fundamental review') to identify and tackle the deep-rooted, structural weaknesses of the regime. The Committee's Trading Book Group has spent considerable time analysing the lessons learnt from, and before, the crisis. Some of the most pressing deficiencies identified were addressed by the July 2009 revisions. However, the Committee has agreed that a number of the framework's shortcomings remain unaddressed and require further attention.”).

• 7

  The Basel Committee published three consultative documents as part of this fundamental review. See Basel Committee, Fundamental review of the trading book (May 2012); Basel Committee, Fundamental review of the trading book: A revised market risk framework (Oct. 2013); Basel Committee, Fundamental review of the trading book: Outstanding issues (Dec. 2014). In January 2016, the Basel Committee published the new market risk framework. Basel Committee, Minimum capital requirements for market risk (Jan. 2016). New issues came up during the implementation process, resulting in additional rounds of changes proposed in June 2017 and March 2018 and finalized in January 2019. See Basel Committee, Simplified alternative to the standardised approach to market risk capital requirements (June 2017); Basel Committee, Revisions to the minimum capital requirements for market risk (Mar. 2018); Basel Committee, Frequently asked questions on market risk capital requirements (Mar. 2018); Basel Committee, Minimum capital requirements for market risk (Jan. 2019, rev. Feb. 2019); Basel Committee, Explanatory note on minimum capital requirements for market risk (Jan. 2019).

• 8

  I also see potential merit in efforts to enhance the risk sensitivity of the regulatory capital framework by incorporating more credit-risk drivers and explicitly differentiating between more types of risk.

• 9

  It's interesting that in the 2018 consultative document, the Committee said that “upon finalisation of the traffic light approach into the market risk standard, the Committee will continue to monitor the effectiveness of the finalised calibration of the thresholds to ensure their appropriateness.” Basel Committee, Revisions to the minimum capital requirements for market risk, § 2.1.3, at 8 (Mar. 2018).

• 10

  “A price will be considered real if it is (i) a price from an actual transaction conducted by the bank, (ii) a price from an actual transaction between other arm's length parties (eg at an exchange), or (iii) a price taken from a firm quote (ie a price at which the bank could transact with an arm's length party).” Basel Committee, Minimum capital requirements for market risk, § 10.26, at 11 (Jan. 2019, rev. Feb. 2019).

• 11

  “Under the proposal, a banking organization would have to use a stress scenario that is calibrated to be at least as prudent as the expected shortfall-based measure for modellable risk factors and calculate the liquidity horizon-adjusted expected shortfall-based measure for non-modellable risk factors in stress using the same general process as proposed for modellable risk factors, with three key differences. First, the proposal would require a banking organization to separately carry out such calculation for each non-modellable risk factor, as opposed to at the risk class level. Second, the proposal would require a banking organization to apply a minimum liquidity horizon adjustment of at least 20 days, rather than 10 days. Third, the proposal would require a banking organization to separately identify for each risk class the stress period for which its market risk covered positions on model-eligible trading desks would experience the largest cumulative loss, except that a common twelve-month period of stress could be used for all non-modellable risk factors arising from idiosyncratic credit spread or equity risk due to spot, futures and forward prices, equity repo rates, dividends and volatilities.” 88 Fed. Reg. at 64,140 (Sep. 18, 2023).

• 12

  This might not be as much work or as time consuming as it initially seems. The Basel Committee's working groups have prepared numerous work papers and other documents supporting many of the Basel Committee's design decisions. Those are not, however, public domain. The U.S. banking regulators could borrow heavily from these staff analyses in future notice-and-comment rulemakings to develop U.S. implementations of the underdeveloped aspects of the Endgame market risk reforms.

• 13

  See Bank of Eng., Implementation of the Basel 3.1 Standards, Consultation Paper 16/22, § 6.68, at 264 (Nov. 30, 2022).

• 14

  Basel Committee, Revisions to the minimum capital requirements for market risk § 2.2.2, at 10 (Mar. 2018) ("The Committee has received feedback from market participants that the [risk factor eligibility test] requirement for there to be a gap of no longer than one month between any two price observations may result in some risk factors being inappropriately deemed as non-modellable despite other evidence of market liquidity. In particular, challenges in identifying observable prices during certain periods of the year have been cited due to seasonality of markets and low volumes of trading during holiday periods").

• 15

  Basel Committee, Explanatory note on minimum capital requirements for market risk § 3.2, at 8 (Jan. 2019) (“The quantitative conditions for a risk factor to be eligible for modelling have been amended to include risk factors that have sufficient liquidity but may experience extended periods during which there is limited trading (eg agricultural commodities). For example the requirement of no more than a 30-day gap between real price observations has been replaced by a requirement of a minimum of four real price observations in a 90-day period. Where a risk factor fails this risk factor eligibility test, it may still be considered eligible for modelling if there are a minimum of 100 real price observations in the previous 12 months. In both cases banks are permitted to count only one real price observation per day.”).

• 16

  For example, the impact of the risk factor eligibility test might depend in part on the extent to which banking organizations are able to establish data-pooling schemes that improve the availability of real price observations. I understand there have been efforts toward that goal, but at least so far, these schemes have been challenged by confidentiality issues and other obstacles.

• 17

  Among other things, I was not able to support the proposal's large surcharge added to the Basel Committee's credit risk capital requirements for residential real estate exposures. I also was not able to support the proposal's “dual-requirement structure” under which a large bank would determine each risk-based capital ratio using the existing standardized approach (revised to include the new market risk approach) and then separately determine that ratio using the new expanded risk-based approach, with the smaller of the two ratios being the binding constraint. As I have said, the Collins Amendment does not require a dual-requirement structure, and concerns about competitive parity across large and small banks can be better and more directly addressed through other solutions.

• 18

  The Basel Committee acknowledged in its consultative documents that its approach to operational risk capital would overcapitalize banks with high fee revenues. Basel Committee, Operational risk – Revisions to the simpler approaches, § 46, at 16 (Oct. 2014). The Committee then proposed a fix in the second consultation. Basel Committee, Standardised Measurement Approach for operational risk, § 20-22, at 4 (Mar. 2016). But the Basel Committee then dropped that fix from the final standards without explanation. And so that leaves the U.S. banking regulators trying to defend an approach to operational risk capital that its own Basel authors have said does not work for high fee revenue banks. Related to this, as I also mentioned in my July dissent, the specification of the formula and the sizing of the coefficients have a significant impact on operational risk capital requirements. The Basel Committee made considerable changes to the formula in the final standards, again with no explanation.