On June 25, the Federal Reserve announced the results of its annual stress test, supplemented by an ad hoc sensitivity analysis that was run in parallel. These results were accompanied by a requirement for firms to update their capital plans and the announcement of two new rules for the third quarter 2020 capital distributions – share repurchases are prohibited, and common dividends may not increase from second-quarter levels and are capped at an amount equal to the average of the bank’s past four quarters of earnings.
This note analyzes the policy rationales for the Federal Reserve’s new sensitivity analysis and capital distribution rules, including how they fit within the existing regulatory regime. The note describes an unavoidable tension between dynamism and procyclicality in capital requirements and concludes that the Federal Reserve should commit to implementing the stress capital buffer (SCB) to govern capital distributions for the fourth quarter. If the Fed should decide that a further degradation in economic conditions warrants another sensitivity analysis – whether as an examination tool, a public disclosure tool or as the basis for share repurchases or dividend limits or a retooled SCB – it should be clear about its purpose, and revisit all the assumptions of the analysis to conform to that purpose.
Recent Federal Reserve Actions
The Federal Reserve’s new requirements provided a transition from the 2019 Comprehensive Capital Analysis and Review (CCAR) cycle to the first SCB requirement in October of this year. The Fed calculated the SCB in accordance with its established design framework, using the stress scenario it published in February. While the new rules on capital distributions were a surprise, most banks had voluntarily halted share repurchases, and tying holding company dividend capacity to past earnings has over eighty years of precedent at the bank level. The short-term effects have been limited: all but two of the 33 covered bank holding companies (BHCs) are expected to maintain their current dividends. Perhaps most significantly, the new restrictions and the forthcoming SCB served as an antidote to proposals to ban all dividends at bank holding companies regardless of their existing capital position, recent earnings and future prospects.
The medium and longer-term implications, however, are potentially more significant. If the Fed extends the provisions on capital distributions into the fourth quarter and beyond, it would effectively constitute the Fed abandoning the current capital framework in the midst of a first wartime test that appeared to be demonstrating its wisdom. More broadly, uncertainty about capital requirements and dividend capacity has contributed to a downdraft in bank equity prices and may make banks more reluctant to expand their balance sheets. In the longer-term, uncertainty about capital requirements will result in a higher cost of bank capital, which necessarily translates into lower long-term economic growth.
The Federal Reserve’s requirement for banks to update their capital plans also highlights an unavoidable conflict within capital regulation: the fact that dynamism in capital requirements is synonymous with procyclicality in capital requirements. The Federal Reserve’s June 25 actions, in essence, reflect a conclusion that the existing capital regime (including its own SCB) is insufficiently dynamic and therefore must be updated to reflect current and potential future degradations in economic performance. At the same time, though, the Fed has expressed concerns about procyclicality and earlier had provided (together with the other banking agencies) some regulatory relief designed to allow banks to deploy the hundreds of billions in dollars of capital they hold above regulatory minimums in order to support lending and market intermediation.
The Existing Regulatory Capital Regime
During the 2007-09 financial crisis, many banks paid dividends when they should have been retaining their capital. As a result, a central focus of post-crisis regulation was restricting the ability of banks to pay dividends, and its centerpiece was Basel III, which not only made capital requirements more stringent in terms of both the quantity and quality of capital, but also introduced a series of buffer requirements the exact purpose of which was to determine at what point banks should be required to cease paying dividends, including in a stressed economic environment.
What are those existing restrictions? First, and most obviously, the CCAR process was constructed to govern capital distributions: the sole sanction for failure to meet a capital requirement on a post-stress basis was a set of limitations on dividends and share repurchases. This year the process converted the stress test results from a pass-fail quantitative grade to a buffer that must be maintained over minimum requirements at the holding company level and, if applicable, any Global Systemically Important Bank (GSIB) surcharge. Notably, the SCB requires each bank to pre-fund four quarters of dividends on top of each bank’s projected stress losses as part of its buffer requirement. The Federal Reserve commendably operated that process as designed and published the stress losses for each of the 33 banks on June 25.
Second, a related emphasis of the post-crisis framework was to require large banks to maintain capital buffers above their minimum capital requirements at the bank and holding company level: a 250 basis point capital conservation buffer (CCB), scheduled to be replaced this fall with the SCB; a GSIB buffer; and potentially a countercyclical capital buffer (CCyB). The idea was that these buffers are capital that a bank could hold in good times but then be able to deploy to continue lending in an economic downturn – a countercyclical action. The challenge of buffers, however, is how to provide an incentive to use them in bad times; otherwise, buffers simply become a constant drag on economic growth.
The Federal Reserve and other global regulators decided to impose only a single cost on a bank that chooses to draw down its buffers: a restriction on capital distributions. Thus, in bad times banks could use their capital buffer without suffering the more severe sanctions associated with breaching minimum capital requirements (e.g., growth limits and formal orders to issue more equity); in good times, they would still hold the buffers because their shareholders value their capital distributions. One can debate the logic of that policy – as we have and will – but for present purposes, the relevant point is that banks were already subject to dividend and stock repurchase restrictions to the extent that they drew down capital buffers.
These post-crisis rules supplemented existing constraints on bank capital distributions. A variety of banking agency regulations and guidance already restrict the ability of a bank to pay a dividend to its parent holding company in certain cases, and a longstanding “prompt corrective action” regime limits the ability of a bank to pay dividends if it becomes undercapitalized. Of course, the latter would only be triggered if a bank depleted 100% of its buffers and fell below minimum requirements, so is not a present concern. The former, though, generally includes consideration of a bank’s net income in recent quarters in terms of assessing its ability to pay a dividend. Thus, bank dividend capacity is already restricted both as a matter of the bank’s earnings and its current capital requirements.
Lastly, since the inception of CCAR, the Fed has demanded that large banks develop rigorous internal capital planning processes that consider not only the general macroeconomic risks present in the Supervisory Severely Adverse scenario, but also idiosyncratic risks from the bank’s specific activities and risk profile. Firms must consider these internal stress tests in setting their capital targets and limits (required under Fed CCAR SR letters), and these internal targets and limits frequently exceed regulatory limits or buffers.
Despite this abundance of restrictions on capital distributions at both the bank and holding company, the Federal Reserve decided to impose two more at the holding company level for the third quarter: a flat ban on share repurchases, regardless of individual circumstance, and an income-based restriction on dividends (along with a ban on any increase) without prior Fed approval. The Fed does allow subject firms to request a waiver from the income-based restriction (or a dividend increase), albeit through an opaque and subjective process that may not provide much comfort (which tends to correlate with certainty) to investors. The Fed also conducted an ad hoc sensitivity analysis outside of the normal CCAR/SCB stress testing process; while this analysis was perfectly appropriate as a monitoring tool (and, as discussed below, should have given the Fed considerable assurance about bank resiliency), there is much uncertainty about how it might be structured and what role it might play going forward.
Why the Change?
The Federal Reserve conducted its sensitivity analysis because the stress test scenario that was distributed in February had become outdated as the coronavirus pandemic resulted in a baseline economic outlook that was worse than the Fed’s stress scenario. In other words, the Federal Reserve found itself in a situation where it believed that the current regime did not reflect the extraordinary risks arising from worsening public health and economic conditions – put another way, that the current regime was not sufficiently dynamic.
In several respects, the existing regime does not indeed capture changing risk. Under Basel III, global regulators moved towards standardized approaches to measuring risk – that is, using a one-size-fits-all (and all-times) model designed by the Basel Committee to measure a firm’s credit and other risks, rather than a bank’s model, which is dynamic. For purposes of the macro shock under the Federal Reserve’s CCAR/SCB, the starting point is risk-weighted assets measured under the standardized approach. The post-crisis framework also includes leverage ratios, which ignore risk completely.
However, in important ways, other parts of the existing capital regime – particularly for larger banks – are dynamic, with bank resiliency requirements increasing as the economy turns down. First, large banks in the first quarter of 2020 implemented a new accounting rule, the Current Expected Credit Losses (CECL) methodology, that required them to establish an allowance (a reserve) for every loan, and base that allowance on expected lifetime losses for that loan. The most important determinant of expected losses, and thus of the reserve, is a macroeconomic forecast used by the bank. Thus, the allowance is now dynamic.
Second, the Fed’s annual stress test, now used to calculate the SCB, greatly increases dynamism over the Basel norm, as it responds to changes in bank balance sheets, the economic outlook and financial risks, albeit only at one-year intervals.
Third, under risk-based capital standards, banks subject to the advanced approaches — that is, the largest nine most internationally active banks — are required to model the credit, market and other risks of their assets, updating the risk-weights applied to the assets when calculating capital ratios as the risks change. In the first quarter, these banks increased risk weights to reflect downgrades of borrowers, higher mark-to-market losses and increased requirements for counterparty credit risk.
The problem with dynamism, though, is that in this context it is synonymous with procyclicality. If capital requirements reflect changing reality, and reality is worsening, then capital requirements must be going up. Indeed, concerned about that procyclicality, the Federal Reserve and the other banking agencies in March mitigated the impact of CECL on capital. They did so because CECL was due to have a procyclical effect, which would have caused banks’ capital ratios to decline rapidly, thereby incentivizing them to restrict lending. Dynamism and procyclicality are two sides of the same coin, so the agencies’ move against procyclicality was also a move against dynamism.
The Sensitivity Analysis
The sensitivity analysis conducted by the Federal Reserve was an attempt to measure what a more dynamic capital measure would show about current bank resiliency. As noted, conducting such an analysis made complete sense as an examination tool, and as an input into larger judgments that the Federal Reserve needed to make about the economy. But there is uncertainty about how its results will be used in setting capital requirements.
The exact method the Federal Reserve used for conducting its sensitivity analysis is difficult to determine because the June 25 bank-specific results were not disclosed to the public or the banks. (Disclosure of the results, which were reached through a necessarily less rigorous approach than normal stress tests, would have imposed potentially unwarranted costs on the few banks that did poorly and also on investors in loan types for which the Fed projected high loss rates.) That said, the Federal Reserve has emphasized that the purpose was to test the resilience of large banks to a further deterioration in health and economic conditions. It was a reality check — much like the 2009 SCAP (Supervisory Capital Assessment Program) which is universally considered its noble precedent. In practice, though, the sensitivity analysis appears to have toggled between reflecting reality and sticking to the February stress testing scenarios, albeit with an apparent overriding theme: choosing the option that would lead to larger losses and greater capital depletion.
With respect to unemployment and output, the sensitivity analysis logically includes more severe assumptions than the 2020 CCAR assumptions used to size banks’ SCB: so, not only a V-shaped recovery that is roughly similar to the February scenario, but also a U-shaped recovery and a W-shaped double-dip recession that are worse. The last scenario is also effectively a 13-quarter stress rather than nine under the CCAR/SCB stress test, as it includes a reserve build in the last quarter of the projection horizon to cover the increase in loan losses over the following four quarters.
The most binding, W-shaped, scenario was within a range of forecasts in the market, but the Federal Reserve assumed additional shocks across various lines of business that collectively made its scenario much more severe. All three downside scenarios included additional stresses in specific business sectors and commercial real estate markets. These adjustments were on top of what was already expected by the path of the macroeconomic series in the alternative downside scenarios. Also, the analysis increased the size of shocks in certain markets to capture dislocations seen at the onset of the crisis. Finally, the analysis increased holdings of corporate loans to reflect the surge in revolver draws seen in March and assumed higher market risk-weighted assets due to higher volatility in financial markets.
However, the Federal Reserve did not update its 2020 CCAR scenario to take account of many other existing conditions that had proven less severe (and thus less demanding of capital) than projected in the stress scenario for the SCB.
With respect to the macro scenario, draws on credit lines were projected to remain outstanding over the projection horizon, but banks have reported that borrowers have already substantially repaid their coronavirus-related credit-line drawdowns. The sensitivity analysis also ignored the impact of depressed consumer spending on credit card loan balances.
Furthermore, all three downside scenarios still assume a 50 percent drop in equity prices, a sharp widening of corporate bond spreads, a 25 percent decline in house prices, the global market shock, the counterparty failure scenario and huge operational risk losses. None of these assumptions approximates what has happened this year. The Federal Reserve’s approach, by assuming the worst case for every line of business, effectively punished the diversified business model of larger banks that the current crisis has thus far rewarded.
The Global Market Shock (GMS) deserves special mention. While the GMS accounts for a significant portion of the SCB of each bank with significant trading activities, income from such activities has served as an important source of revenue during the current crisis, as shown in the chart below:
In sum, under a stressful second quarter, banks earned a large amount of trading income when the GMS predicted banks would report huge trading and counterparty losses. This error was predictable, not only by a detailed analysis of the severity of the GMS but also based on general evidence that a variety of changes in regulations and industry behavior have made trading a less risky and more profitable business for those firms that have been able to reach the scale necessary to bear its compliance burdens.
A final reason that losses might be lower in the current stress than what would be projected using the Fed’s models is the unprecedented fiscal, regulatory and liquidity policy response to the crisis. In terms of an ex-ante stress test, one would not wish banks to choose capital levels on the assumption of government support for the economy; however, in terms of a real-world look at bank resiliency, and with concerns about procyclicality looming, some acknowledgment of stimulus actions to date seems sensible. While projecting the shape of fiscal support is difficult, its likelihood at the very least argued for the Fed noting that its sensitivity analysis had a substantial upward bias in terms of predicted losses.
All that said, large U.S. banks hold so much capital that even with dire GDP and unemployment scenarios, combined with add-ons and hold-backs, covered banks still performed quite well. Under the most extreme, W-shaped scenario, the Federal Reserve reported that the aggregate CET1 capital ratio for the covered banks would fall from 12 percent in the fourth quarter of 2019 to its minimum of 7.7 percent, with eight of the 33 banks reporting a minimum ratio above 10.5 percent, another eight below 4.8 percent and the remaining 17 banks somewhere between 4.8 and 10.5 percent.
In sum, the sensitivity analysis taught us that even given a dire outcome across all their lines of business, almost all large banks operating in the U.S. continued to meet all regulatory minimum capital requirements imposed by the post-crisis regulatory regime, and most held buffers well above the minimum capital requirements.
Despite this outcome, the Fed banned repurchases and restricted dividends for all large banks without its prior approval for the third quarter. In addition, it required banks to reassess their capital needs and resubmit capital plans later in the year, increasing uncertainty around the level of banks’ capital requirements, particularly as little information is available on what the resubmission will entail or how (or whether) it will be used to determine capital requirements.
The Implications of the Federal Reserve’s Action
In taking these actions, the Federal Reserve appears to have drawn a distinction between capital requirements, which it has not raised and from which it has provided limited relief, and restrictions on capital distributions, which it has now imposed. The apparent logic is that raising capital requirements is procyclical (bad) because it incentivizes banks to shrink their lending to maintain a higher capital ratio, but ordering banks to limit dividends is neutral or countercyclical (good) because it provides banks more capital with which to lend and to absorb losses. Thus, under this logic, the Fed can have its cake (maintain high capital levels to ensure resiliency) and eat it too (allow banks to deploy capital while maintaining not only minimum capital but also buffers).
With respect to its restrictions for the third quarter, procyclicality appears limited. The restriction is based on historical earnings, so cannot be altered by changes to bank balance sheets. (The same presumably would be true of a permanent such rule.) Going forward, however, banks seeking an exception from these restrictions, or the end of these restrictions as a general rule, will wish to demonstrate high capital ratios. Indeed, a consistent and primary focus of analysts (and the investors they represent) for the second-quarter earnings season was whether each bank would be able to make a case to the Federal Reserve to allow it to continue paying dividends after the third quarter.
As noted above, the Federal Reserve recognized this market reality as recently as March 2020, when it allowed banks greater flexibility to continue paying dividends as they drew down their buffers. In doing so, the Federal Reserve recognized that because dividends are quite valuable to bank shareholders – that is, the people who own banks and whom the directors of the banks serve – banks would be reluctant to draw down buffers if that meant choking off dividends. Thus, the Fed in March stated, “By modifying the definition of eligible retained income and thereby allowing banking organizations to more freely use their capital buffers, this interim final rule should help promote lending activity and other financial intermediation activities by banking organizations and avoid compounding negative impacts on financial markets.
Alternatively, if the dividend ban were imposed independent of a bank’s capital condition — for instance if it were based on GDP or employment or even bank earnings — with no option for a bank to get an exception, then a bank would not have an incentive to hasten its end by boosting its capital ratio. However, if bank investors know they will not receive dividends from banks in economic downturns or if earnings suffer a temporary downturn (or most likely both), they will place a much lower value on bank stocks; that is, banks’ costs of capital will rise, and the supply of bank credit and bank deposits will fall in good times and bad. Moreover, if a bank knows that it will lose the ability to pay out dividends in bad times regardless of its condition, then it will have an incentive to take risks and generate earnings to the fullest extent possible when it foresees bad times — that is, act procyclically.
The Federal Reserve wisely gave itself time to consider how it would proceed next. Theirs is not an enviable task.
To provide market certainty and fundamental fairness, the Federal Reserve should commit to moving to the SCB to govern capital distributions in the fourth quarter. Furthermore, if the Fed plans to conduct another sensitivity analysis – whether as an examination tool, a public disclosure tool or as the basis for share repurchases or dividend limits or a retooled SCB – it should be clear about its purpose, and revisit all the assumptions of the analysis to conform to that purpose.
 Unless otherwise noted, “bank” for purposes of this note includes bank holding companies.
 12 U.S.C. § 60.
 85 Fed. Reg. 15,909 (Mar. 20, 2020). Using balance sheets as of the end of the first quarter, the largest banks — those with more than $100bn in assets — held more than $750bn in common equity tier 1 capital above the minimum requirement.
 12 C.F.R. § 3.11; 12 C.F.R § 217.11; 12 C.F.R § 324.11.
 See Greg Baer, “The Next Financial Crisis and the Great Buffer Fallacy,” Banking Perspectives First Quarter 2019, (Mar. 12, 2019), https://www.bankingperspectives.com/the-next-financial-crisis-and-the-great-buffer-fallacy/. The one exception is the countercyclical capital buffer, which is designed to be lowered in stress, and in fact has been lowered in other countries during the current crisis. For more, see Bill Nelson, “So, What Have We Just Learned About…the Countercyclical Capital Buffer,” Bank Policy Institute (Apr. 24, 2020), https://bpi.com/so-what-have-we-just-learned-aboutthe-countercyclical-capital-buffer/
 12 C.F.R. part 3, subparts H and I; 12 C.F.R. § 5.46; 12 C.F.R. part 5, subpart E; 12 C.F.R. part 6; 12 C.F.R. part 208, subparts A and D; 12 C.F.R. part 303, subpart K and § 303.241; 12 C.F.R. § 324.405. Statutory limits on dividends may apply as well. See, e.g., 12 U.S.C. §§ 56, 59 and 60; 12 U.S.C. §324; and 12 U.S.C. §1828(i)(1).
 12 U.S.C. § 1831o; see also 12 C.F.R. part 6; 12 C.F.R. part 208, subpart D; 12 C.F.R. part 303, subpart K; 12 C.F.R. part 324, subpart H.
 SR 15-19 (Dec. 18, 2015); SR 15-18 (Dec. 18, 2015).
 These restrictions applied only to large banks. It remains to be seen whether the Federal Reserve or other regulators have similar concerns about smaller banks, which are not currently subject to stress testing, CECL and other restrictions.
 Randal K. Quarles, Vice Chairman for Supervision, Federal Reserve Board, “The Adaptability of Stress Testing,” At Women in Housing and Finance, Washington, D.C. (June 19, 2020) [hereinafter “The Adaptability of Stress Testing”], https://www.federalreserve.gov/newsevents/speech/quarles20200619a.htm
 See Francisco Covas, “How Tough was the Fed’s Sensitivity Analysis,” Bank Policy Institute (July 14, 2020), https://bpi.com/how-tough-was-the-feds-sensitivity-analysis/ which estimated that the added assumptions included in the sensitivity analysis effectively would require the unemployment rate to reach 25 percent, the peak level attained in the Great Depression and at the top end of baseline projections from the 2020 Q2 Survey of Professional Forecasters.
 Perhaps this was coincidental: in a speech a week before the results were announced, Vice Chair Quarles indicated that there was not sufficient time to update these factors. But there will be time between now and the fall, when the Federal Reserve may rerun the exercise. See “The Adaptability of Stress Testing.”
 For example, with respect to operational risk, see Greg Baer and Anna Harrington, “Operational Risk and the Federal Reserve’s Stress Test: Time for a Reality Check?” Bank Policy Institute (May 27, 2020), https://bpi.com/operational-risk-and-the-federal-reserves-stress-test-time-for-a-reality-check/
 See SIFMA, “Global Market Shock and Large Counterparty Default Study,” (Aug. 2019),
 See, Diana Iercosan et al., “Trading Activities at Systemically Important Banks, Part 1: Recent Trends in Trading Performance,” FEDS Notes (July 10, 2017),
 While the Fed may avoid taking positions on fiscal policy, it recognizes that assumptions about the path of taxes and spending are necessary for a reasonable projection for the economy as well as for consideration of risks around that outlook: staff projections prepared for the FOMC invariably include such assumptions.
 The dissonance here is evident in the Federal Reserve announcement. It quotes Vice Chair Quarles to say that “the results of our sensitivity analyses show that our banks can remain strong in the face of even the harshest shocks,” and then summarizes those results. It then states, “In light of these results, the Board took several actions following its stress tests to ensure large banks remain resilient despite the economic uncertainty from the coronavirus event.” Press Release, Federal Reserve Board (June 25, 2020), https://www.federalreserve.gov/newsevents/pressreleases/bcreg20200625c.htm. But the results clearly demonstrated that banks would be resilient even in the absence of the Board’s actions; thus, “In spite of these results….” would have been a more accurate statement.
 There are arguments against such a rule at the holding company level, which are beyond the scope of this note.
 See supra note  and accompanying text; see also federal banking agencies provide banks additional flexibility to support households and businesses: https://www.federalreserve.gov/newsevents/pressreleases/bcreg20200317a.htm; SR 20-5: Questions and Answers (Q&As) on Statement Regarding the Use of Capital and Liquidity Buffers (March 19, 2020): https://www.federalreserve.gov/supervisionreg/srletters/sr2005.htm
 85 Fed. Reg. 15,909, 15910-11 (March 20, 2020).
 For a further discussion of problems with a blanket ban on bank capital distributions see Francisco Covas and Bill Nelson, “A Few Observations on Professor Stein’s Remarks Last Week at the Brookings Institution,” Bank Policy Institute (June 11, 2020), https://bpi.com/a-few-observations-on-professor-steins-remarks-last-week-at-the-brookings-institution/ “