The U.S. banking agencies recently reaffirmed their commitment to implement the most recent Basel Accord, a sweeping series of changes to bank capital requirements. If implemented without care, and specifically without avoiding duplication of capital charges unique to the existing U.S. regime, those changes could reduce U.S. economic growth, further diminish capital markets liquidity, increase systemic risk and further disincentivize bank lending to small businesses and low- to moderate-income households.
While the Basel Accord has many parts requiring calibration, at a macro level, the question is whether these costs of higher bank capital requirements are offset by corresponding benefits — in other words, whether there is evidence that U.S. banks require additional capital to reduce the danger of insolvency or financial instability. While many years ago, that question was a theoretical one about which economists were free to theorize, there is now a considerable amount of real evidence. And it all points one way.
Aggregate Capital Levels
The worthy goal of the Basel Accord is to ensure that all countries impose similar capital requirements on their banks, and more particularly, that none allows its banks to hold significantly less capital in order to gain competitive advantage. In practice, the United States has adopted a unique capital regime with not only Basel components but also multiple additional layers of stringency. First, U.S. regulators have doubled the capital surcharge that Basel imposes on the largest banks. Second, for those banks, the leverage ratio (capital as a percentage of total assets, regardless of the risk of those assets) has been raised from 3 percent to 5 percent at the holding company level, and doubled to 6 percent at the bank level. Third, through its annual capital stress test, the Federal Reserve has imposed another capital surcharge on all large U.S. banks that is 1.6 times higher than its parallel in the Basel Accord as applied in non-U.S. jurisdictions. Fourth, the U.S. approach produces significant spurious volatility in these requirements, with only three months for a bank to remedy a projected shortfall, effectively requiring U.S. banks to hold still more capital to guard against an unexpected increase.
The question then becomes whether the current U.S. capital requirements are appropriately calibrated. When the U.S. banking agencies enacted the latest round of capital rules on the heels of the Global Financial Crisis, they placed a heavy emphasis on academic papers to determine optimal capital levels, with an upward bias (always referred to by regulators as “conservatism”) built into each component of the regime. This approach was understandable given the experience of 2008–09 and the lack of any practical way to calibrate the contemplated changes. But now we have more than a decade of actual experience with the regime, in good times, bad times and times of crisis. And it all points one way.
Real-world Experience
Since enactment of post-crisis reforms, there has never been the slightest indication that any major U.S. bank was in danger of insolvency. No large U.S. bank has ever fallen below any minimum capital standard or come close to it.
A prime example is the performance of the banking industry during the COVID crisis. As American businesses scrambled to raise cash at the outset of the crisis, banks increased commercial and industrial loans by $482 billion between March 11 and April 1, 2020 – that is, by 6 percent per week, or 50 times the weekly average.[1] The unprecedented increase in loans was based on existing capital levels and raised no safety and soundness concerns.
A similar story played out in capital markets. Large banks made more than $40 billion in trading revenues in the first half of 2020 as they performed the core function of market making — posting bids to those looking to sell and offers to those wishing to buy.[2] The Federal Reserve was subsequently forced to intervene in those markets because of the broad economic dislocations created by the pandemic, not because the banks were losing money or were at risk of doing so.
Bank critics have attempted to downplay the extraordinary performance of banks in 2020 by noting that the government assisted the economy significantly, which lessened bank losses. This narrative ignores multiple important facts.
First, governmental assistance to the economy did not come immediately. Banks funded the largest increase in corporate liquidity demand while the government was considering whether and how to act. The lending and trading numbers cited above came before enactment of the CARES Act and establishment of the Federal Reserve’s crisis programs. Banks stood strong on their own as the storm began and intensified. (At the same time, fintech and other nonbank lenders largely stopped lending.)
Second, over the course of the COVID-19 crisis, the Federal Reserve subjected large banks to multiple stress tests. A key test conducted in June 2020 had the unemployment rate peaking at 15½ percent. Moreover, the Fed adjusted bank balance sheets by applying widespread downgrades of business borrowers in industries with substantial demand declines after the COVID-19 event. These adjustments were sizable and effectively corresponded to doubling the peak unemployment rate to 30 percent.[3] Furthermore, the Fed explicitly assumed, counterfactually, that there was no fiscal support that would lower bank losses: for example, no increase or extension of unemployment insurance, no Paycheck Protection Program and no Economic Impact Payments, even though those programs had been established in April 2020. The Fed also assumed that the banks would take no actions to mitigate losses in the event its doomsday scenario occurred.[4]
Despite the extreme assumptions in the Fed’s sensitivity analysis conducted in June 2020, 75 percent of banks were projected to remain well above their minimum capital requirements at the end of the stress planning horizon and the remaining 25 percent were near minimum capital ratios — and nowhere near insolvency.
Third, as large banks issued long-term debt in 2020, spreads widened appropriately given greater risk, but banks had no difficulty selling this debt, demonstrating that the market concurred with the Fed that they had ample capital.
And we know for certain that markets were relying on bank resilience and not government support when they bought this debt, because banks were the only companies disqualified from participation in the Federal Reserve’s Primary Market Corporate Credit Facility and Secondary Market Corporate Credit Facility. Through these programs, the Fed directly supported corporate debt issuance by buying corporate bonds – except bank bonds.
(Of course, even if banks had required government support for the economy to avoid large losses, it would seem odd to say that the optimal capital level for banks should be set by assuming that the government would not support the economy in a pandemic—or not fight a war if the country were invaded, or not offer unemployment assistance under current law, etc. [5] It would be akin to insurance regulators demanding that property and casualty insurers set home insurance rates assuming there were no fire departments.)
The Federal Reserve’s Annual Assessment
More broadly, and leaving aside the events of 2020, the Federal Reserve has conducted annual stress tests of bank capital levels each year since the Global Financial Crisis. Those results have shown banks to have ample capital to weather a market shock and deep recession more severe than 2008-09: the typical stress scenario assumes an unemployment rate of at least 10 percent, a 50 percent drop in equity prices, and a 25 percent drop in home prices – again, assuming that banks take no action to mitigate losses in the midst of such a scenario. While the scenarios are hypothetical, the tests utilize banks’ actual balance sheets.
For 2022, total loss absorbency for the 33 banks included in the stress tests results — equity plus allowances for credit losses and bail-in debt — was in excess of $2.8 trillion, while total net stress losses under that scenario were approximately $300 billion. Thus, absorbency was more than nine times net losses predicted under an extreme stress, as shown in the following chart:
It would be remarkably capricious if, in calibrating the new Basel Accord, the Federal Reserve ignored its own consistent analysis over the past dozen years.
The Relevant Parts of the Latest Accord
Higher capital requirements for U.S. banks appear unjustified not only in aggregate but also with respect to the particular risks that are the focus of changes made in the Basel Accord.
One important change made by the recent Basel Accord is to add a substantial capital charge for so-called operational risk — meaning losses not from credit or market risk but from cyber-attacks or other breakdowns in process. The Basel Committee has estimated that the share of operational risk in total capital requirements is nearly 15 percent for GSIBs, on average.[6] Meanwhile in the United States, regulators have previously taken the position that a separate operational risk charge was unnecessary under its standardized approach to capital. The capital charges for credit and market risk were set sufficiently high to cover that risk. After all, unless these risks are perfectly correlated (which they clearly are not), a dollar of capital can guard against multiple risks, just as a single airbag can guard against a variety of collisions.
Furthermore, the Federal Reserve already requires banks to hold capital against operational risk by imposing significant losses associated with operational risk events in its stress tests. In the 2022 stress tests, the Fed estimated that operational risk losses would reach nearly $200 billion (or nearly ¼ of aggregate losses in the stress tests). Because the stress test results determine banks’ current capital requirements through the stress capital buffer, U.S. banks are already subject to a substantial operational risk capital charge – many multiples of the charge applied to any non-U.S. bank. And many multiples still higher than actual loss experience.
Similarly, for banks engaged in trading, the Basel Accord dramatically increases capital requirements for market risk—what it labels the Fundamental Review of the Trading Book (FRTB)—by basing the charge on market risk under stressed rather than static conditions. But stress risk is exactly what the Global Market Shock (GMS) in the Fed’s stress tests is intended to capture. Here again, the consequences of double counting would be significant. The GMS already results in market risk capital requirements for the largest U.S. banks more than double those of the current Basel framework.[7] The FRTB would further increase those requirements and make these banks less willing to engage in essential market-making activities.
The problem is all the worse because evidence suggests that the existing Global Market Shock massively overstates market risk. As one Fed governor has observed, the actual global market shock of March 2020 was similar to the Global Market Shock included in DFAST 2020. But, as noted above, banks made more than $40 billion in trading revenues in the first half of 2020, yet were projected to lose $87 billion in trading losses in DFAST 2020 — when the Global Market Shock requires them to assume massive losses for capital setting purposes. The trend has continued to the present, as market volatility has generated earnings for banks, but the Fed’s models in this year’s stress test again predicted — and now require capital to be held sufficient to cover — massive losses. One major outside study demonstrated numerous flaws in the Fed’s approach (to the extent that its approach could be deduced from what little the Fed publicly discloses) that could be producing such an outcome.[8] No one has suggested that the study was incorrect.
Meanwhile, evidence that non-economic levels of capital are being required has presented itself through declining market depth in Treasury and other fixed-income markets. While regulators have proven reluctant to confess any role, market participants and analysts regularly report increasing illiquidity to the effects of the enhanced supplementary leverage ratio, the GSIB surcharge and the U.S. stress tests.[9] In 2020, a temporary deduction of reserve balances and U.S. Treasuries from the denominator of the leverage ratio enhanced market liquidity; its removal in 2021 produced a deterioration. As shown in the chart below, dealer banks’ balance sheet capacity has not kept up with the growing amount of Treasury debt.
The recent Basel Accord also introduces some offsetting changes to the higher capital requirements via an increase in the granularity and risk sensitivity of the standardized approach for credit risk. For example, loans to publicly traded investment-grade corporates and mortgage loans with reduced loan-to-value ratios would receive lower risk weights compared with those under the current Basel framework. The European Commission legislative proposal removed the requirement that investment-grade corporates have securities listed on recognized exchanges to receive a lower risk weight. In addition, the EU is also planning to maintain certain exemptions to the credit risk valuation framework and postpone the implementation of minimum requirements on secured financing transactions.
A Wise Implementation
Concerns about economic growth and financial stability currently abound. High capital requirements have raised the cost of credit and pushed more and more risk outside the banking system to unregulated places that lack dependable, through-the-cycle funding and a customer incentive to keep supporting customers and markets. The relevant performance of large bank stocks to the S&P 500 does not indicate an industry earning attractive returns on equity:
Bank capital can be seen increasingly as a modern-day Maginot line, an impressive fortress that financial risk will simply bypass. And it turns out that the southern terminus of the Maginot line was in a little town called Basel. The question is whether construction will continue, or whether regulatory attention will focus on the costs to the economy of constructing this fortress and forcing the risk to move elsewhere.
[1] See https://www.nber.org/papers/w27256
[2] Trading revenues for the 11 banks subject to the Global Market Shock in the Federal Reserve’s stress tests were $41.5 billion in the first half of 2020, based on the FR Y-9C “Consolidated Financial Statements for Holding Companies,” Schedule HI line item 5.c. “Trading revenue.”
[3] See https://bpi.com/how-tough-was-the-feds-sensitivity-analysis/.
[4] See https://www.federalreserve.gov/publications/files/2020-sensitivity-analysis-20200625.pdf, p. 14.
[5] https://bpi.com/no-banks-were-not-bailed-out-in-2020/
[6] See Basel Committee on Banking Supervision, “Basel III Monitoring Report,” September 2022. Available at https://www.bis.org/bcbs/publ/d541.pdf
[8] See Global Market Shock and Large Counterparty Default Study – SIFMA. Aug. 2019, https://www.sifma.org/wp-content/uploads/2019/09/SIFMA-GMS-LCD-Study-FINAL.pdf. Also, see https://bpi.com/trading-revenue-and-stress-tests/.