The Economic Cost of Weakening Capital Requirements for Large Banks

September 28, 2017 1:30 PM
Photo Credit: 
REUTERS/Brendan McDermid

The policy response to the financial crisis of 2007–09 brought new regulations on bank capital and stress tests that have strengthened the ability of the major US financial institutions to withstand future shocks. However, recent proposals by the US Treasury Department could weaken capital requirements, undermining the improvement in financial stability that has been achieved. The consequence would likely be a major cost to the economy, because the increased probability of a banking crisis would impose expected damages that would outweigh gains from freeing up some capital for new investment. My new book The Right Balance for Banks provides an analytical framework for quantifying the resulting net economic losses, which could be large.[1]

I calculate that the optimal capital requirement for the banking system is equity capital of 7 to 8 percent of total assets, corresponding to 12 to 14 percent of risk-weighted assets.[2] If capital is lower, the risk of a banking crisis rises substantially; if capital is higher, it becomes unduly costly to finance new investment projects, and the future productive capacity of the economy is unduly curbed. The current international agreement, Basel III, sets capital requirements at only 9.5 percent of risk-weighted assets for the large global systemically important banks (G-SIBs). So the Basel targets need to be strengthened by about one-third.

As it turns out, the large US banks do currently hold nearly 8 percent of total assets in common equity capital. They do so because they pursue a voluntary cushion above the legally required amounts, and also because the stress tests have tended to push them toward these levels. However, in its June report on the financial system, the US Treasury Department called for lower capital requirements for large banks. Treasury proposed that capital required by the supplementary leverage ratio (SLR) under US implementation of the Basel III agreement should exclude the following safe assets from the "exposure" base: cash on deposit with central banks, US Treasury securities, and initial margin for centrally cleared derivatives.[3]

The Basel III exposure base for the SLR is considerably larger than the balance sheet assets, because it includes off-balance-sheet items.[4] As shown in table 1, in 2013 the combined SLR exposure of the eight large G-SIB bank holding companies amounted to $14.7 trillion, compared to $10.3 trillion for their total assets.

Table 1 Total assets and supplementary leverage “exposure” base, G-SIBs (billions of dollars and ratios, 2013)
Bank holding company Total assets SLR exposure Exposure/assets
JP Morgan Chase 2,416 3,570 1.48
Citigroup 1,880 2,895 1.54
Bank of America 2,102 2,696 1.28
Wells Fargo 1,527 1,961 1.28
Goldman Sachs 912 1,518 1.66
Morgan Stanley 833 1,283 1.54
Bank of New York Mellon 374 410 1.1
State Street 243 345 1.42
Total 10,287 14,678 1.43
G-SIB = global systemically important bank; SLR = supplementary leverage ratio
Sources: Meraj Allahrakha, Paul Glasserman, and H. Peyton Young, “Systemic Importance Indicators for 33 Bank Holding Companies: An Overview of Recent Data,” OFR Brief Series 15-01, February 12, 2015; 10-K Reports of banks.

The general Basel III SLR is tier 1 capital at 3 percent of exposure. In US implementation, for G-SIBs the "enhanced" SLR (eSLR) applicable beginning in 2018 is 5 percent of exposure for bank holding companies, and 6 percent for the depository bank subsidiaries of the bank holding companies.[5] For the eight G-SIBs, the depository bank subsidiaries range from 98 percent of total bank holding assets (State Street) to 16 percent (Goldman Sachs), with the weighted average at 73 percent.[6] The current US treatment will thus require tier 1 capital amounting to 5.73 percent of exposure for the eight G-SIBs as a group. In 2016, total assets of the eight G-SIBs were $10.7 trillion.[7] Applying the bank-specific estimates of table 1 for the ratio of SLR exposure to assets, their combined total exposure was $15.2 trillion. Applying the bank-specific eSLR requirements (ranging from 5 to 6 percent, depending on the share of commercial bank assets in the holding company total), the eSLR would thus require tier 1 capital of $860 billion.[8] Common equity tier 1 capital for these banks amounts to 89 percent of tier 1 capital.[9] Common equity under the eSLR would be $764 billion, or 7.1 percent of total assets.

Table 2 reports actual common equity capital holdings of the eight G-SIBs in 2016. The first column shows their total assets at the end of 2016. The second column shows the ratio of their common equity tier 1 capital to total assets, and the third column, the absolute amounts of that capital. For the group as a whole, common equity tier 1 amounted to 7.9 percent of total assets. These actual holdings are consistent with maintaining a small cushion above their prospective 2018 requirements under the eSLR. The slight excess above the eSLR amount (7.1 percent) likely also reflects the fact that it is the stress tests rather than the prospective supplementary leverage requirement that has recently been the binding constraint.

Table 2 Total assets and common equity tier 1 capital of US G-SIBs, 2016 (billions of dollars and percentages)
Bank holding company Total Assets (TA) Common equity/TA Common equity
JP Morgan Chase 2,510 7.39 185.5
Citigroup 1,800 9.52 171.4
Bank of America 2,200 7.92 174.2
Wells Fargo 1,940 7.72 149.8
Goldman Sachs 860 8.21 70.6
Morgan Stanley 816 7.48 61
Bank of New York Mellon 333 5.6 18.6
State Street 243 5.13 12.5
Total 10,702 7.88 843.6
Sources: MarketWatch and author’s calculations based on Federal Reserve, Dodd-Frank Act Stress Test 2017.

How much would the Treasury's proposed exemptions from the exposure base reduce the effective capital requirement for the US G-SIB banks? One indicator is that the Treasury considers that "Large U.S. banks hold nearly 24% of their assets in high-quality liquid assets such as cash, U.S. Treasury securities, and agency securities".[10] More specifically, at the end of 2016 commercial banks had total assets of $16.07 trillion, of which Treasury and agency obligations amounted to $2.43 trillion and cash (including reserves at the Federal Reserve) stood at $2.21 trillion.[11] On this basis, and assuming that the eight G-SIB holding companies have holding profiles similar to those of the commercial banking system as a whole, 29 percent of the G-SIBs' total assets would be excludable from exposure under the Treasury proposal, without taking account of initial margin for centrally cleared derivatives.[12] This exclusion would reduce the exposure base by $3.1 trillion (29 percent of $10.7 trillion). Exposure would shrink from $15.2 trillion to $12.1 trillion (1.13 times total assets instead of 1.43). The eSLR tier 1 capital required would amount to $684 billion.[13] The corresponding common equity requirement would be $609 billion, or 5.69 percent of total assets.

However, exempting Treasury obligations and cash (including reserves at the Federal Reserve) could provide an incentive for the G-SIBs to shift away from other assets (such as corporate bonds) toward Treasury obligations. On this basis, it is useful to consider a range in which the common equity tier 1 capital might fall to as low as, say, 5 percent of total assets.

The starting point for an analysis of the impact of weakening the eSLR is to consider the effect of lower capital on the probability of a financial crisis. The three benchmark ratios of common equity to total assets ("k") considered here yield the following probabilities of financial crisis, applying the estimates developed in my book: For k = 7.1 percent, annual crisis probability (Pcr) of 0.33 percent per year (central) or 0.58 percent per year (adverse); for k = 5.69 percent, Pcr = 0.71 percent (central) or 1.26 percent (adverse); and for k = 5.0 percent, Pcr = 1.12 percent (central) or 1.97 percent (adverse).[14]

A financial crisis is estimated to cost 64 percent of one year's GDP in the central case, and 100 percent in the adverse case. The expected annual damage from reducing the capital requirement can then be calculated as the product of the increase in the crisis probability times 64 percent of GDP, for the central case, or times 100 percent of GDP, for the adverse case. On this basis, when the capital requirement is cut from 7.1 percent under the current eSLR to 5.69 percent, the central estimate is that annual damage from the change would amount to [0.71 – 0.33] percent x 64 percent of GDP = 0.25 percent of GDP. In the adverse case, with both a larger increase in the crisis probability and application of the larger loss from a financial crisis, the corresponding expected annual damage from the change would be: [1.26 – 0.58] x 100 = 0.68 percent of GDP. The corresponding expected damages from reducing the capital ratio from 7.1 percent to 5.0 percent would be 0.51 percent of GDP (central) and 1.40 percent of GDP (adverse).

There would be some partial offset from the lower opportunity cost of holding equity capital. In the central case, each percentage point of additional equity capital relative to total assets reduces long-term output by 0.15 percent.[15] In the adverse case, this saving is much smaller, at only 0.03 percent for each percentage point change in the capital requirement.[16] Reducing the capital requirement from 7.1 percent of total assets to 5.69 percent would thus provide a saving equal to [7.1 – 5.69] x 0.15 = 0.21 percent of GDP in the central case but only 0.04 percent of GDP in the adverse case.

Table 3 summarizes the economic impact that could be expected from reducing the capital requirement in the eSLR from 7.1 percent of total assets (common equity) to either 5.69 percent in a narrow interpretation of the Treasury proposal or to 5.0 percent allowing for additional induced shifts in asset composition. After taking account of both the increased expected crisis damage and reduced opportunity costs of equity capital, the net effect amounts to an annual loss of 0.04 percent (central) to 0.63 percent of GDP (adverse) if the new capital requirement is 5.69 percent. The net loss reaches a range of 0.19 percent of GDP (central) to 1.33 percent (adverse) if the capital ratio instead falls to 5.0 percent. The final portion of the table translates these damages to cumulative totals over 10 years, based on a midpoint GDP of $20.5 trillion (2016 dollars). Although this loss is quite modest for the central estimate and the intermediate capital level, at $74 billion, it is large in the adverse case even for this intermediate capital ratio, reaching a net loss of $1.3 trillion over the decade. The corresponding net loss is far larger if instead the capital ratio falls to only 5.0 percent, as cumulative net losses would range from $400 billion (central) to $2.7 trillion (adverse).

Table 3 Economic impact of reducing the enhanced supplementary leverage requirement for G-SIB capital
New capital requirementa 5.69 5
Change from current requirement -1.41 -2.1
Increase in expected crisis damage (percent of GDP)
Central 0.25 0.51
Adverse 0.68 1.4
Reduction in capital opportunity cost (percent of GDP)
Central 0.21 0.31
Adverse 0.04 0.06
Net impact (percentage points)
Central -0.04 -0.19
Adverse -0.63 -1.33
Cumulative 10-year net impact (billions of dollars)
Central -74 -399
Adverse -1,300 -2,729
a. Common equity tier 1 as percent of total assets. Current rules: equivalent to 7.1 percent.
Source: Author's calculations.

These losses could be overstated to the extent that the eight G-SIBs are not the entire banking system. In 2016 total commercial bank assets were $16.1 trillion. The nonbank assets of the G-SIB holding companies amounted to $2.9 trillion.[17] Of the total assets of $19 trillion, the eight G-SIBs accounted for 56 percent. However, in the political economy of determining bank capital, it seems likely that intermediate and smaller banks would be given alleviation on required capital that is at least proportional to that accorded to the large banks. Moreover, exclusion of the three safe asset categories from the SLR exposure base, thereby turning the ratio into a hybrid that is neither a leverage ratio nor a risk-weighted ratio, is only one of the ways the Trump administration might weaken capital requirements. In principle, it could simply revert to the Basel III SLR of 3 percent rather than applying the 5-6 percent range for G-SIBs (which critics call US "gold-plating"). It can also apply more lenient stress tests than before. Overall, the calculations here are unlikely to overstate the stakes involved in the Trump administration's consideration of weaker capital requirements.

Author's note: For comments on an earlier version, I thank without implicating Morris Goldstein.


1. The Right Balance for Banks: Theory and Evidence on Optimal Capital Requirements. Washington: Peterson Institute for International Economics, June 2017.

2. Risk weights are zero for US government bonds, and can be 50 percent or lower for residential mortgages, for example.

3. US Department of the Treasury, A Financial System that Creates Economic Opportunities: Banks and Credit Unions,Washington, June 2017, p. 14.

4. These off-balance-sheet items include "the net value of certain securities financing transactions plus credit derivatives and commitments as well as counterparty risk exposures." Meraj Allahrakha, Paul Glasserman, and H. Peyton Young, "Systemic Importance Indicators for 33 Bank Holding Companies: An Overview of Recent Data," OFR Brief Series 15-01, February 12, 2015, p. 2.

5. See for example Luigi L. De Ghenghi and Andrew S. Fei, "US Basel III Supplementary Leverage Ratio," Harvard Law School Forum on Corporate Governance and Financial Regulation, October 5, 2014.

6. The individual ratios of commercial bank to total assets in the sequence of table 1 and indicating the market symbol for each entity are as follows: JPM, 0.88; C, 0.75; BAC,0.76; WFC, 0.92; GS, 0.16; MS, 0.23; BK, 0.81; STT, 0.98. Calculated from 2015 10-K reports and from Federal Reserve, Large Commercial Banks.

7. Individual bank financial data are summarized by MarketWatch. (For Bank of America, for example: see

8. This amount is 5.66 percent of the combined group exposure for 2016.

9. Calculated from Federal Reserve, Dodd-Frank Act Stress Test 2017: Supervisory Stress Test Methodology and Results, Washington, June 2017.

10. US Department of the Treasury, A Financial System that Creates Economic Opportunities: Banks and Credit Unions, Washington, June 2017, p. 37.

11. Federal Reserve, Assets and Liabilities of Commercial Banks in the United States – H.8., Washington.

12. The initial margin amounts appear to be relatively small. The total "required house initial margin" amounted to $107 billion in 2016, and the eight G-SIBs would have accounted for perhaps no more than two-thirds. Office of Financial Research, 2016 Financial Stability Report, Washington (p. 52).

13. Based on the 5.66 percent of exposure requirement for the weighted ratio of bank subsidiary versus other bank holding company assets in 2016.

14. Based on a survey of studies by the Basel Committee, I estimate the crisis probability as Pcr = Akγ. In the central case, A = 3.14E – 07. In the adverse case, A = 5.52E – 07. For both cases, γ = –3.5. See The Right Balance for Banks, pp. 99–100 and p. 105.

15. See estimation of this parameter ψ in The Right Balance for Banks, pp. 106–07.

16. This is the estimate of ψ when the adverse "high optimal capital ratio" alternative values are applied for capital share (α), elasticity of substitution (σ), spillover to nonbank credit sources (θ), equity capital cost to banks (ρB), and Modigliani-Miller offset (μ). See The Right Balance for Banks, p. 105.

17. Estimated as 27 percent of the G-SIB total assets of $10.7 trillion.