Setting Bank Capital Requirements

Bank capital requirements are the focus of contentious and heated debates. Since they limit banks’ ability to take on risk and leverage, owners and managers almost always argue for lowering them. To reduce the likelihood of using public funds for further bailouts, both libertarians and progressives argue strenuously that they should be higher. Focusing on the balance between the social benefits of a more resilient financial system and the social costs of curtailing liquidity and loan provision, academicians usually conclude that current levels are too low. So, with well-financed banks and their lobbyists on one side, and a cohort of advocates armed with academic research on the other, regulators are caught in the middle. To whom should they listen?

The answer to this question is an empirical one, so it is important to base any conclusions on a fair and balanced reading of the evidence. Regular readers of this blog will be unsurprised that we continue to maintain that bank capital requirements should be higher than they were even before the Federal Reserve began its quiet campaign to relax them several years ago. If we were to pick a number, we would start with a leverage ratio—the ratio of common equity to total assets (including off-balance sheet exposures)—that is in the range of 10 to 15 percent, and possibly higher. The risk-weighted equivalent would be about twice as high in the United States (or three times as high in Europe). (The exact numbers depend on the intricacies of accounting standards.) The one thing we would not be arguing for is a further erosion of capital requirements from their current level.

We start with a short reminder about why we need capital requirements in the first place. In our primer, we describe how bank capital acts as self-insurance, providing a buffer against insolvency and giving bank managers an incentive to manage risk prudently. A system of well-capitalized banks provides resources to creditworthy borrowers, lets zombie firms die, and reduces the likelihood of systemic crises. As we emphasize in an earlier post, resilient banks enhance growth and stability, poorly capitalized banks do not.

Continuing, it is worth a few words on whether debt really is cheaper than equity. The famous Modigliani-Miller theorem establishes that in the absence of taxes and bankruptcy, the value of a firm is independent of its capital structure, so firms should be indifferent between debt and equity finance. Adding taxes and bankruptcy alters the balance, leading businesses to prefer debt.

In the case of banks, a variety of government subsidies tilt the financing bias substantially further towards debt. Without getting into too much detail, underpriced deposit insurance and the too-big-to-fail guarantee give owners (and managers) of large banks an incentive to leverage their firms as much as possible. This means that there are clear private costs to increasing capital—owners are right to believe that equity is more expensive than debt. From their perspective, it is. But, as Admati and Hellwig persuasively argue, raising capital requirements to reduce these subsidies is exactly the right policy response to bring social costs and benefits back into balance

Unsurprisingly, banks continue to argue for lower capital requirements. A recent piece by Francisco Covas of the Bank Policy Institute is an example. Looking back at analysis that helped calibrate the Basel III reforms a decade ago, as well as some more recent work by the Minneapolis Fed, he suggests that, if anything, capital requirements are too high. There are at least three problems with this conclusion. First, there is the framing of the Basel Committee’s work. Second, there is the research of the last decade. And third, the claim flies in the face of the evidence.

Starting with the first, the Basel Committee’s 2010 study makes a set of assumptions that lead costs and benefits to balance at a low level of capital. It is important to keep in mind that the study maximizes costs by assuming that bank lending rates make the entire adjustment as capital requirements increase, that there is no decline in the cost of equity finance for safer banks, and that there is no other policy response (such as lower policy rates in equilibrium). The result is to overstate the degree to which borrowers will reduce investment and lower growth. Furthermore, the study ignores the cyclical stabilization benefits of having more resilient banks and less indebted borrowers. Even so, a careful reading of the decade-old report leads to the conclusion that costs and benefits balance at a risk-weighted capital requirement of at least 16%—far more than the Basel III minimum of 7%.

Second, over the past few years, researchers have been actively working to examine the costs and benefits of bank capital. The Minneapolis Plan recommends capital requirements of 24%, consistent with a leverage ratio requirement in the range of 12-15%. Looking at recent academic work, we point to two pieces of excellent work by Juliane Begenau. The first shows that if there is a high demand for safe assets, then increasing bank capital will reduce banking funding costs and raise bank lending. In the second,  Begenau and Landvoigt build a quantitative model that takes explicit account of the fact that risk can migrate from a safer banking system to riskier nonbanks. That is, raising bank capital requirements changes the relative costs of banks versus nonbanks, shifting activity from the former to the latter. Despite this risk migration, they find that the optimal leverage ratio is around 17%. In other words, a decade’s worth of progress in analyzing the question leads us to increase our estimate of required capital further.

Finally, we come to some facts. Increases in bank capital levels have accompanied the rise in requirements. The Basel Committee’s quantitative impact studies show this decisively: for the roughly 100 largest banks in the world, common equity has gone from less than 3% of total assets at the end of 2009 to 6% in mid-2019 (the latest available). Focusing on the United States, the following chart shows the evolution of the average supplementary leverage ratio (equivalent to the Basel leverage exposure ratio) for the eight U.S. institutions designated as global systemically important banks (G-SIBs). Information is incomplete prior to 2015, so we approximate the earlier numbers. Full implementation of the new requirements is designated by the vertical black line, and the horizontal red line is the 5% requirement. We note that the leverage ratio rose steadily from 2013 to 2015, at which point it stabilizes before starting to fall in the last few years. (This decline is a measure of the Fed’s recent supervisory easing.)

Leverage exposure ratio for the largest U.S. banks, 2013 to 2019

Sources: Federal Reserve Bank of Kansas City, Federal Deposit Insurance Corporation and authors’ calculations. The solid black line is the average supplementary leverage ratio (SLR) as computed by the FRBKC. We estimate the dashed line using information from editions of the FDIC Global Capital Index.

With capital levels going up, what happened to bank lending and bank profitability? Prior to 2007, the average return on equity for U.S. banks with more than $15 billion in assets was roughly 15%. In the past few years, it has been between 9% and 12%. Not only that, but (prior to the pandemic) the market price of G-SIB bank equity rose relative to the book value of their assets, so that the end-2019 price-to-book ratio exceeded one for all eight U.S. G-SIBs.

Did the rise in capital and capital requirements diminish U.S. banks’ supply of credit? Not according to the data. The following chart shows the evolution of lending to the non-financial sector from 2013 to 2019—the same period for which we show bank leverage ratios in the previous chart. The black line (using the left-hand scale) represents total credit from all sources as a fraction of GDP, and the red bars show the fraction of total credit accounted for by banks. We draw two conclusions. First, as bank capital increased, credit remained relatively stable at around 150% of GDP. Second, banks accounted for an increasing share of credit. And, while we do not show it here, higher capital requirements have been accompanied by lower, not higher net interest margins. (We note that the Financial Stability Board’s monitoring of nonbank intermediation confirms these results—U.S. banks’ share of lending has been increasing in recent years.)

Quantity and sources of credit to the nonfinancial sector, 2013-2019

Source: BIS.

To be as clear as we can possibly be, higher capital requirements have not hurt banks, they have not hurt borrowers, and, if there was any macroeconomic impact, it was probably offset by monetary and fiscal policy. In other words, it is difficult to find any social costs associated with increasing capital requirements and improving the resilience of the financial system.

Now, some people might take this as evidence that we should raise risk-weighted capital requirements to 100% and force all commercial banks to be narrow banks. That is, we should require that any bank liabilities not deposited at the central bank would have to be funded 100% by equity. In our analysis of this proposal several years ago, we conclude that it is extremely unlikely to meet its advocates’ objectives of a more stable financial system. First, pushing risk out of the banking system does not get rid of it. In fact, if risk migrates to parts of the economy with no capital requirements at all, systemic risk could very well increase. Second, banks do two things that are difficult to reproduce in other structures. First, they provide liquidity both to depositors and to borrowers. Second, they have expertise in screening potential borrowers and then monitoring those to whom they make loans. While it is surely possible to replicate a part of each of these without having the risk on the balance sheets of banks, we doubt that all of it can move without a migration of systemic risk.

Before concluding, we should mention that in addition to capital requirements, there are other rules constraining bank behavior. Of the many other regulatory practices that influence the structure of bank liabilities, the two most important are stress tests and the requirement that large banks issue long-term debt that is convertible into equity at resolution.

We have discussed the importance of well-structured stress tests on numerous occasions in the past. (For our most recent views, see here and here.) During normal times, they provide authorities and banks with information to set capital buffers. And, during periods of heightened stress, the tests help resolve uncertainty and measure the need for additional equity finance. Indeed, stress tests can be an essential element of a system in which intermediaries facing capital shortfalls are automatically recapitalized through the conversion of subordinated debt into equity. In such a framework, continuity of service does not suffer even if a bank fails the test, making it all the more credible.

With regard to the requirement for total loss-absorbing capacity (TLAC), it is set in the range of 20% of risk-weighted assets—roughly double the equity requirement (see our earlier post). This means that bailing in TLAC-eligible long-term subordinated debt would be sufficient to recapitalize a bank at the required level should a shock wipe out its equity and force the bank into resolution. But TLAC only comes into play in resolution. If a bank incurs losses but remains marginally above the threshold for restructuring its TLAC debt, it must rebuild its capital buffer by retaining earnings. We think it would be easier and cleaner to simply require additional equity (rather than TLAC debt).

To return to where we started, the debate over the appropriate level of capital requirements probably will never end. By focusing on the very real private costs these impose, banks and their representatives will continue to push to lower them. But their complaints ring hollow. The past decade shows that more than doubling capital requirements had no clear impact on the supply of bank credit. Surely private returns have fallen as subsidies declined, but that is a feature, not a flaw, of good public policy.

As we have emphasized before, we do not know exactly where to set capital requirements. But the experience of the past decade continues to suggest that they should be higher than they are now. So, why not raise them until we start to see some solid evidence of real social costs: higher lending rates, lower lending volumes, or a shifting of systemic risk outside the banking system?