Initial Proposals for Reform to Reduce Duplicative Risk Capture and Maintain the Credibility of the Collins Floor Requirement: Part VI in Our Series on US Bank Capital Requirements

• The US Basel 3 Endgame proposal released on July 27, 2023 creates a “three-stack” approach to risk-based capital requirements, replacing the current two-stack approach: the revised US standardized approach, pursuant to the Collins Amendment of the Dodd-Frank Act (“Collins Floor”); a new expanded risk-based approach (“ERBA”); and a separate standardized output floor.
• This blog, part VI in our series on capital requirements for U.S. banks, examines which of the proposal’s three capital stacks are most likely to be binding for the US GSIBs, as well as their interactions with the Federal Reserve’s Stress Capital Buffer (“SCB”). There are two key design flaws that ought to be addressed in the final rule.
• First, the proposal’s design makes the statutorily mandated Collins Floor, as well as the standardized output floor, effectively obsolete (i.e., not binding requirements for banks), reducing them to costly compliance exercises for banks. Second, applying the SCB to the ERBA leads to material duplicative risk capture (i.e., capital requirements in excess of the underlying risks) for market, operational and credit valuation adjustment (“CVA”) risks.
• The over-capitalization resulting from these two design issues will likely constrain the largest banks’ ability to provide key capital markets products and services, translating into higher costs for end-users and the broader US economy. Moreover, it could have negative implications for liquidity in the US Treasury markets, US financial stability, and the broader economy.
• These design issues could be mitigated through two simple adjustments. First, capital buffers should be applied to the Collins Floor and the ERBA in such a way that either could realistically act as the binding constraint for the largest banks.  This would involve a) applying the SCB and Method 2 GSIB surcharge (“Method 2”) to the Collins Floor, and b) applying the Capital Conservation Buffer (“CCB”) and the Method 1 GSIB surcharge (“Method 1”) to the ERBA.  Second, the standardized output floor should apply only to banks with sufficiently large trading activities.

Background on The Existing US Capital Framework

The Collins Amendment of the 2010 Dodd-Frank Act requires large US banks (those subject to “advanced approaches” treatment, generally the US GSIBs) to calculate their capital requirements in two ways: first, using regulator-set standardized approaches and second using internal models approaches, with the higher of the two capital requirements acting as the firm’s binding capital constraint.^[1]  This creates a “two-stack” approach to calculating capital requirements (Figure 1 provides a high-level schematic depiction of this framework).^[2]

The Federal Reserve then applies the Stress Capital Buffer (“SCB”) to the Collins Floor-mandated standardized approaches and the Capital Conservation Buffer (“CCB”) to the advanced approaches capital calculations. The SCB is no lower and can be significantly higher than the CCB.  As a result, the combination of the Collins Floor and SCB often generally acts as the binding capital constraint for the largest banks.  As of Q4 2021, the binding risk-based capital constraint for the eight US GSIBs was the Collins Floor,^[3] though some institutions “escaped” the Floor in Q1 2023.^[4]

Figure 1. A high-level schematic overview of the two-stack risk-based capital framework under the current capital rules.^[5]

How Does the US Proposal Change the Existing Capital Framework?

On July 27, 2023, the Federal Reserve Board, the Office of the Comptroller of the Currency, and the Federal Deposit Insurance Corporation (“Agencies”) jointly released a proposal that would implement the internationally agreed Basel III Endgame standards into the US capital rules.  The proposal contains several requirements that are super equivalent to the minimum Basel standards – an example of long-standing US “gold plating” of international standards.^[6]  Amongst other elements, the proposal contains a “three-stack” design for U.S. risk-based capital requirements: the revised US standardized approach (i.e., the Collins Floor); the expanded risk-based approach (“ERBA”); and the standardized output floor, which is designed to ensure that banks’ modeled capital requirements do not fall below a defined percentage (72.5%) of the capital requirements calculated using the regulator-set standardized approach.  The SCB would, in turn, be applied to all three capital stacks.  As a result, the CCB will be eliminated – another instance of the US “gold plating”.  This new proposed framework is represented in Figure 2 below.

Figure 2. A high-level schematic overview of the two-stack risk-based capital framework under the proposal

For the purpose of the Collins Floor, the proposal retains the current US standardized approach to credit risk and replaces the current internal models approach to calculating market risk with the Fundamental Review of the Trading Book’s internal models approach (“FRTB-IMA”), meaning that the Collins Floor capital requirement is now equal to the sum of the current US non-modeled credit risk capital and FRTB-IMA market risk capital.^[7]  The ERBA stack is determined as the sum of non-modeled credit risk, operational risk and CVA risk capital – all calculated using the new standardized approaches, plus FRTB-IMA market risk capital.^[8]  The standardized output floor is calculated based on 72.5% of the ERBA capital requirements, though market risk capital must be calculated using the FRTB’s standardized approach (“FRTB-SA”) for these purposes.^[9]  In short, the result of these changes is that regulator-set standardized approaches will now be the basis for determining the US risk-based capital requirements rather than internal models approaches.

Why is the New Standardized Output Floor Unlikely to be a Binding Capital Constraint?

For the standardized output floor to be the binding capital constraint, a bank’s trading activities need to be sufficiently large that at a minimum its non-modelled capital requirements account for more than 40% of total risk-based capital requirements.^[10]

The eight US GSIBs collectively have the largest trading activities amongst all the Category I-IV banks under the Agencies’ tiering rule and account for 94% of the $94bn aggregate trading losses under the Federal Reserve’s 2023 supervisory stress test.^[11]  The standardized output floor only has the potential to be a binding constraint for this group of firms.  Figure 3 plots the modeled market risk capital calculated as a percentage of the total capital requirements under the ERBA, in term of risk weighted asset (“RWA”) amount, for the eight US GSIBs, assuming the ERBA was effective as of Q1 2019 (recalling that the new standardized output floor equals 72.5% of the EBRA).^[12]  For all eight US GSIBs, the modeled market risk capital accounts for no more than 20% of the total risk-based capital requirements under the ERBA since Q1 2019 (as of Q1 2023, the highest percentage is 15% across all eight banks).

Figure 3. The ratio of FRTB-IMA RWA to total ERBA RWA for the eight US GSIBs since Q1 2019

Assuming the modeled market risk capital equals 20% (15%) of total capital, for the standardized output floor to be binding, the corresponding non-modeled market risk capital needs to exceed 2.9x (3.6x) of the modeled market risk capital.^[13]  Table 1 presents the ratio of the non-modeled to the modeled market risk capital requirements by risk classes as reported in the Basel Committee’s 2019 explanatory note.  Given the US GSIBs’ diverse trading activities, it is reasonable to expect their non-modeled to modeled market risk capital ratio be no larger than 2x which is significantly smaller than 2.9x (3.6x).  Thus, the new standardized output floor has little chance of being binding for the US GSIBs.^[14]

Table 1. The estimated capital requirements under FRTB SA relative to FRTB IMA by risk classes

The Collins Floor Will No Longer Be a Binding Constraint for the Largest Banks Under the Proposal

The Collins floor is binding on firms whenever non-modeled credit risk capital savings (i.e., the difference in the US non-modeled vs the ERBA non-modeled credit risk capital) exceed 100% of the aggregate operational and CVA risk capital under the ERBA.^[15]  Figure 4 plots the estimated credit risk capital savings as the percentage of the aggregate operational and CVA risk capital, in terms of RWA amount, for the eight US GSIBs.

Figure 4. The ratio of non-modeled credit risk capital savings relative to the aggregate operational and CVA risk capital for the eight US GSIBs since Q1 2019

The non-modeled credit risk capital savings are no larger than 10% of the aggregate operational and CVA risk capital under the ERBA for the eight US GSIBs.  This means that the Collins Floor will not act as the binding constraint for the US GSIBs, effectively making this statutory requirement redundant.

Therefore, both the Collins Floor and the new standardized output floor effectively become compliance exercises that create unnecessary operational burdens for banks, while the binding capital requirement will almost always be the ERBA.

The ERBA and the SCB Effectively Double Count Similar Market and Operational Risks

Under the Federal Reserve’s supervisory stress test, market and CVA risks are captured via the global market shock (“GMS”) losses.  In addition, the ERBA capitalizes market and CVA risks under the FRTB and the FRTB-like standardized approach for CVA (“SA-CVA”) respectively.  The designs of the FRTB and the GMS share many key similarities – e.g., measure deep tail losses, large risk factor shocks and severely constrain diversification benefits.^[16]

Additionally, operational-risk losses are captured by the supervisory stress test’s pre-provision net revenue (“PPNR”) component.  The Federal Reserve’s PPNR “model projects losses stemming from operational-risk events using information about the size and historical operational-risk losses of the firms”.^[17]  The historical operational-risk losses directly feed into the internal loss multiplier (“ILM”) of the ERBA standardized operational risk framework which the proposal relies on to increase operational risk capital requirements.  Consequently, applying the Federal Reserve’s SCB to the ERBA will lead to material duplicative risk capture between the two frameworks (something approximating a “double counting” of risks).

Two Simple Design Modifications Can Address These Issues

There are two ways in which the proposal could be adjusted to remediate the redundancy built into the three stack-approach and mitigate the duplicative risk capture between the SCB and the ERBA.

a) Apply Additional Buffers to the Collins Floor and ERBA so that Collins can Act as a Credible Floor: We recommend that the Agencies apply both the SCB and the Method 2 GSIB surcharge to the Collins Floor – all three are US specific requirements– and both the CCB and the Method 1 GSIB surcharge to the ERBA – the three components that are consistent with the Basel standards. As noted above, the risk-based capital ratio for the largest US banks equals the sum of their minimum capital requirements plus buffers such as the SCB, the CCB and, as applicable, the countercyclical capital buffer (CCyB)^[18] and the GSIB surcharge.  The ERBA as proposed always produces significantly higher minimum capital requirements than the Collins Floor, which appears to run counter to statutory intent.  To allow the Collins Floor to have a reasonable chance of acting as the binding capital constraint for the US GSIBs, the buffer requirements added to the Collins Floor should be higher than those added to the ERBA.

The SCB is dependent on the outcome of the Federal Reserve’s supervisory stress test.  For some banks, the SCB is significantly higher than the CCB.^[19]  As a result, adding the SCB to the Collins Floor and the CCB to the ERBA (without applying the SCB) would increase the likelihood of the Collins Floor being binding and would remove the duplicative risk capture for operational and CVA risks (but not for market risk).  Additionally, as part of the buffer requirements, the applicable GSIB surcharge can be adjusted.  US GSIBs needs to calculate GSIB surcharges using two methods: the Basel Method 1 (“Method 1”) and the US-specific Method 2 (“Method 2”).^[20]  The Method 2 almost always produces a higher surcharge than the Method 1 approach. Therefore, adding the Method 2 GSIB surcharge to the Collins Floor and the Method 1 GSIB surcharge to the ERBA would further enhance the credibility of the Collins Floor.

b) Reduce the Compliance Burden on Banks and Only Apply the Standardized Output Floor to Banks with Large Trading Operations. With a non-modeled to modeled market risk capital ratio of 1.8x, FRTB-IMA capital effectively needs to exceed 50% of the total ERBA capital for the standardized output floor to bind. Given that no banks’ trading activities are anywhere close to this threshold, the Agencies should reduce the compliance burden associated with calculating the standardized output floor capital requirement and only apply it to banks with sufficiently large trading activities, e.g., where the FRTB-IMA capital exceeds a certain regulators-set percentage of the total ERBA

The above changes would also help to mitigate some of the potential negative effects of the proposal on the US Treasury markets specifically and the US capital markets more generally. A recent report from staff at the New York Federal Reserve finds that dealer banks’ capacity to intermediate in the US Treasury markets is heavily tied to their VaR model estimates.“^[21] The proposal will replace this VaR-based market risk measure with a new Expected Shortfall (“ES”)-based market risk measure, a change that contributes heavily to an estimated 75% increase in market capital RWA for the largest US banks. This capital increase will significantly constrain the capacity of the US GSIBs to support the US Treasury markets and other key funding markets. Moreover, it effects would be further compounded by the SCB – due to the duplicative risk capture – and the proposed changes to the GSIB score indicators^[22] under the Agencies’ GSIB surcharge proposal.^[23]  The proposed design modifications would mitigate the duplicative risk capture with the SCB and moderate the proposal’s excessive increases in capital requirements for the largest US banks, allowing them to continue to provide support to the US Treasury markets and the capital markets overall. In turn, this would help promote greater US financial stability.

Conclusion

The Basel III Endgame completely rewrites the risk-based capital standards for banks.  The US proposal adopts the Basel standards but with significant gold plating in many areas, including in the adoption of a three-stack capital framework, with only one stack – the ERBA – likely to be binding for the largest US banks.  When the SCB is applied to the ERBA, it leads to material duplicative risk capture (effectively “double counting” of risks) for market, operational and CVA risks, results in excessively higher capital requirements than would otherwise be the case.  This could increase funding costs and reduce market liquidity in key capital markets such as the US Treasury markets, with significant negative consequences for US financial stability and the broader economy.

These design issues could be mitigated through two simple adjustments. First, capital buffers should be applied to the Collins Floor and the ERBA in such a way that either could realistically act as the binding constraint for the largest banks. This would involve a) applying the SCB and Method 2 GSIB Surcharge to the Collins Floor, and b) applying the CCB and the Method 1 GSIB surcharge to the ERBA.  Second, to minimize the duplicative capture of market-related and operational risks, the standardized output floor should apply only to banks with large trading activities.  Moreover, to mitigate the duplicative risk capture for market risk, the SCB’s GMS component needs to be redesigned, as outlined in  Part IV of this blog series.

–

Dr. Guowei Zhang is Managing Director and Head of Capital Policy for SIFMA

Dr. Peter Ryan is Managing Director and Head of International Capital Markets and Strategic Initiatives for SIFMA

Mr. Carter McDowell is Managing Director and Associate General Counsel for SIFMA

Appendix

This Appendix provides details of the data and methodologies used to estimate the US GSIBs’ risk-weighted asset (“RWA”) amounts under the expanded risk-based approach (“ERBA”) sets out in the US proposal.

1. Data Description

The data used in the analysis for the 8 US GSIBs were downloaded from the Federal Financial Institutions Examination Council’s website.  The data includes:

• quarterly credit and market risk RWAs under the US standardized approach (FR Y9-C); and
• quarterly credit, market, operational and CVA risk RWAs under the advanced approaches (FFIEC 101 and 102).
• The time period coverage is Q1 2019 – Q1 2023.

Additionally, the proposal provides the following impacts estimates:

2. Estimation Methodology

We first calculate a set of scaling factors using the impact estimates the regulators provided in the proposal.  For example, the FRTB IMA to the current market risk rules capital scaling factor equals 760/430 = 1.77  for Cat. I and II bank holding companies, and 220/130 = 1.69 for Cat. III and IV bank holding companies.

To estimate each bank’s RWA by risk category under the ERBA and the revised US standardized approach from the RWA amounts under the current capital rules, the same scaling factors are applied to all banks in the sample.  For example, if a bank’s current market risk RWA is $100, the FRTB IMA RWA is estimated to be $100 * 1.77 = $177.

^[1] The Collins Floor capital requirements equal the sum of non-modeled credit risk capital and modeled market risk capital.  The advanced approaches calculate capital requirements equal the sum of modeled credit risk capital, market risk capital, operational risk capital and CVA risk capital.

^[2] Part I of our Basel capital blog series provides a more detailed overview of the various components of the current regulatory capital rules.

^[3] See “All top US banks below Collins floor”, Risk.net, January 27, 2021 (https://www.risk.net/risk-quantum/7922981/all-top-us-banks-below-collins-floor).

^[4] See “Citi, BNY Mellon escape Collins floor”, Ris.net, July 19, 2022 (https://www.risk.net/risk-quantum/7952111/citi-bny-mellon-escape-collins-floor).

^[5] The risk weighted assets (“RWA”) amount under the US standardized approach equals the sum of US standardized credit risk RWA and the internal models-based market risk RWA, except that the specific risk charge must be calculated using the standardized approach.  The RWA amount under the advanced approaches equals the sum of internal models-based credit, market, operational and CVA risks.

^[6] For example, the proposal eliminates internal models for credit risk in its entirety whereas the Basel standards permit it.  The proposal also removes the modeled default risk charge in the Basel standards for the FRTB.  The proposal floors the Internal Loss Multiplier of the standardized operational risk framework at 1, whereas the Basel standards have no such floor to incentivize banks’ prudent operational risk management practices.  Additionally, the proposal adopts the SFT minimum haircut floor which wasn’t implemented in several other major jurisdictions due to the concerns that the framework could have detrimental impacts on SFT markets.

^[7] Although the proposal allows modeled market risk for the purpose of the Collins floor, because of its complexity and the stringent model approval process for the FRTB-IMA some banks may choose calculate market risk capital using FRTB-SA instead. In this case, the Collins floor capital requirements equal the sum of the current US non-modeled credit risk capital and FRTB-SA market risk capital.

^[8] where RWA ERBA -SA denotes the risk-weighted asset (“RWA”) for credit risk calculated using the ERBA standardized approach (“ERBA-SA”) credit risk.  Similarly, the other terms denote corresponding risk types and calculation approaches.  Banks that do not implement the FRTB-IMA or lose the regulatory approval for it must calculate market risk capital using the FRTB-SA.

^[9]

^[10] The binding condition, i.e., , can be re-written as  .

^[11] See https://www.federalreserve.gov/supervisionreg/dfa-stress-tests-2023.htm

^[12] See Appendix for details on the estimation of the ratios.

^[13] For simplicity, assume the total ERBA capital is $1.  The modeled market risk capital is $p and the aggregate non-modeled credit risk, operational risk and CVA risk capital is $(1-p).  In addition, assume non-modeled market risk capital equals  times the modeled market risk capital.  The standardized output floor is binding if: 0.725 * (a * p + (1 – p)) = 1.  When  p= 20% , a= 2.9.

^[14] For a the FRTB SA to FRTB IMA capital ratio of 1.5x and 1.8x, the corresponding percentage exceeds 76% and 50%.

^[15]

^[16] Part IV of our Basel capital blog series provides an in-depth examination of the similarities.

^[17] See https://www.federalreserve.gov/publications/files/2023-june-supervisory-stress-test-methodology.pdf

^[18] The CCyB is 0% under the current US capital rules.

^[19] See https://www.federalreserve.gov/publications/files/large-bank-capital-requirements-20220804.pdf

^[20] See https://www.ecfr.gov/current/title-12/chapter-II/subchapter-A/part-217/subpart-H

^[21] The paper notes that capacity utilization measures based on estimated VaR … is an important explanatory factor for the tail behavior of Treasury market liquidity.” See Dealer Capacity and US Treasury Market Functionality.

^[22] Specifically the proposal’s treatment of derivatives in the interconnectedness and complexity indictors and trading volume in the substitutability indictor.

^[23] The proposed changes in GSIB surcharge are expected to increase capital requirement for the 8 US GIBs by $13bn.

Related Resources

• Pennsylvania + Wall Blog

  Blog Series on US Bank Capital Requirements