“It is clear that the banks have too much capital.” Jamie Dimon (CEO, JPMorgan), Annual Letter to Shareholders, April 4, 2017.
“If JPMorgan really had demand for additional loans from creditworthy borrowers, why did it turn those customers away and instead choose to buy back its stock?” Neel Kashkari (President, Federal Reserve Bank of Minneapolis), Jamie Dimon’s Shareholder (Advocacy) Letter, April 6, 2017
Over the past 40 years, U.S. capital markets have grown much faster than banks, so that banks’ share of credit to the private nonfinancial sector has dropped from 55% to 34% (see BIS statistics here). Nevertheless, banks remain a critical part of the financial system. They operate the payments system, supply credit, and serve as agents and catalysts for a wide range of other financial transactions. As a result, their well-being remains a key concern. A resilient banking system is, above all, one that has sufficient capital to weather the loan defaults and declines in asset values that will inevitably come.
In this primer, we explain the nature of bank capital, highlighting its role as a form of self-insurance providing both a buffer against unforeseen losses and an incentive to manage risk-taking. We describe some of the challenges in measuring capital and briefly discuss a range of approaches for setting capital requirements. While we do not know the optimal level of capital that banks (or other intermediaries) should be required to hold, we suggest a practical approach for setting requirements that would promote the safety of the financial system without diminishing its efficiency.
What is bank capital? There are several consistent definitions of a bank’s capital (or, equivalently, its net worth). First, capital is the accounting residual that remains after subtracting a bank’s fixed liabilities from its assets. Second, it is what is owed to the banks’ owners—its shareholders—after liquidating all the assets at their accounting value. Third, it is the buffer that separates the bank from insolvency: the point at which its liabilities exceed the value of assets.
The following figure shows the balance sheet of a simple bank that finances its assets (composed of cash, securities, loans, and other instruments) with deposits and other debts, as well as the equity and retained earnings that constitute its net worth. The proportions shown correspond to the average shares of these components in the U.S. commercial banking system at the end of 2017 (see here). In this example, the bank’s capital is 11.3% of assets, corresponding to the gap between total assets (100%) on the one hand and the combination of deposits and other fixed liabilities (88.7%) on the other. This fraction is also known as the bank’s leverage ratio: the ratio of capital to assets. For comparison, the leverage ratio a decade earlier (amid the financial crisis) was 7.2% (see data here).
A Simple Bank: Percent Shares of Assets and of Liabilities and Net Worth (Capital)
Source: FRED (based on Federal Reserve Board H.8 for U.S. Commercial Banks, December 2017).
Importantly, capital is a source of funds that the bank uses to acquire assets. This means that, if a bank were to issue an extra dollar worth of equity or retain an additional dollar of earnings, it can use this to increase its holding of cash, securities, loans, or any other asset. When the bank finances additional assets with capital, its leverage ratio rises.
Banks (and many other financial intermediaries) issue a far larger proportion of debt (relative to equity) than nonfinancial firms. Recent data show that nonfinancial firms have between $0.80 and $1.50 worth of debt liabilities for each dollar of equity (see here and here). By contrast, as we can see from the figure above, the average U.S. commercial bank has a debt-to-equity ratio of roughly 8. This reliance on debt boosts both the expected return on and the riskiness of bank equity, and makes banks vulnerable to insolvency.
In addition to their balance-sheet risks, banks also tend to have a variety of large off-balance-sheet exposures. The most prominent are derivatives positions, which have gross notional value in the trillions of dollars for the biggest global banks, and credit commitments (for a fee), which appear on the balance sheet only after the borrower exercises their option to draw down the loan. As a result, simple balance sheet information understates the riskiness of banks, especially large ones.
Role of bank capital. Bank capital acts as self-insurance, providing a buffer against insolvency and, so long as it is sufficiently positive, giving bank management an incentive to manage risk prudently. Automobile insurance is designed to create a similar incentive: auto owners bear part of the risk of accidents through deductibles and co-pays, which also motivate them to keep their vehicles road-ready and to drive safely.
When capital is too low relative to assets, however, bank managers have an incentive to take risk. The reason is straightforward. Shareholders’ downside risk is limited to their initial investment, while their upside opportunity is unlimited. As capital deteriorates, potential further losses shrink, but possible gains do not. Because shareholders face a one-way bet, they will encourage bank managers to gamble for redemption. This problem goes away as the level of capital rises. That is, when shareholders have more skin in the game, they will be exposed to greater losses and will encourage the bank managers to act more prudently. (See Myers for a discussion of this debt overhang problem).
The role of self-insurance is most important for those banks that are too big to fail (TBTF). As we have discussed in a recent post, governments cannot credibly promise to avoid future bailouts if the alternative is economic disaster (see the primer on time consistency). Consequently, anticipating a bailout, TBTF banks have an incentive to take risks that will spill over to the financial system as a whole. Making TBTF banks resilient through increased self-insurance both ensures their shareholders will bear losses and prompts these firms to internalize the spillovers that otherwise would occur.
Finally, a banking system that is short of capital can damage the broader economy in three ways. First, an undercapitalized bank is less able to supply credit to healthy borrowers. Second, weak banks may evergreen loans to zombie firms, adding unpaid interest to a loan’s principal to avoid taking losses and further undermining their already weak capital position (see here). Finally, in the presence of a widespread capital shortfall, the system is more vulnerable to widespread panic, reflecting fears that some banks may be lemons (see the primer on adverse selection).
Measuring bank capital and exposures. The definition of bank capital makes it seem deceptively simple to measure: just subtract liabilities from assets. Unfortunately, it is often very difficult to measure the value of assets. (And even more difficult to figure out how to treat off-balance sheet exposures.)
At any moment in time, assets are worth what buyers will pay for them. Determining the value of a liquid instrument, like a U.S. Treasury bond, is easy. However, most securities—like corporate, municipal, and emerging market bonds, are significantly less liquid than Treasuries (see here). And since most bank loans, which represent more than one-half of U.S. commercial bank assets, do not trade at all, no one knows their market price. Finally, in periods of financial strain, even active markets can freeze, making the value of a bank’s assets even more difficult to value.
Aside from liquidity, the value of an asset may depend on the solvency of the bank. At one extreme, some intangible assets only have value when the bank is a going concern. For example, when one bank acquires another, the excess of the purchase price over the accounting value of the target becomes goodwill on the balance sheet of the newly merged entity. Another example is deferred tax assets (DTAs). A bank is allowed to use past losses to reduce future tax payments, assuming that they become profitable and would otherwise owe taxes. Neither goodwill nor DTAs typically have value if the bank fails.
We should emphasize that this is not a small matter. As of mid-2017, for the eight U.S. global systemically important banks (G-SIBs), goodwill plus DTAs corresponded to 26% of tangible equity (see here). Five years, earlier, that ratio was 39% (including a whopping 48% for Bank of America).
The presence of intangibles means that the book value of capital may tell us relatively little about the ability of a bank’s balance sheet to absorb unforeseen losses on its assets (on- and off-balance sheet) without becoming insolvent. For that purpose, regulators often exclude things like DTAs from their computation of net worth.
In addition to capital, we also need to compute the assets or exposures against which net worth provides a buffer. Derivatives and accounting conventions complicate this calculation. The five largest U.S. banks held more than $200 trillion of gross notional value as of end-2015 (see here). To take account of these off-balance-sheet exposures, regulators apply credit conversion factors (CCFs) to translate derivatives exposures into asset equivalents that they then use to calculate capital ratios. But, accounting frameworks differ significantly: under U.S. GAAP, a bank may use collateral it receives from a counterparty to offset (or net) a derivatives exposure. Under International Financing Reporting Standards (IFRS), which apply outside the United States, it cannot.
Doubts about GAAP netting—which generously assumes that collateral is of high quality, has not been re-lent, and can always be sold—lead us to prefer IFRS measures of capital adequacy. For the largest banks that dominate global derivatives trading, the difference is enormous. As the chart below shows, for the U.S. G-SIBs, in 2017 the leverage ratio was 8.24% under GAAP, but only 6.62% under IFRS. Back in 2012, the levels were lower and the disparity even larger: 6.17% vs. 3.88%. Put differently, under IFRS in 2012, the effective debt of the biggest banks was nearly 25 times their capital. That ratio is still greater than 15.
U.S. G-SIB Capital-to-Asset Ratios (Percent): GAAP, IFRS, and Basel Risk-Weighted Assets
Note: The 2012 and 2017 measures for Basel Risk-Weighted Assets were calculated under Basel I and Basel III standards, respectively. Source: FDIC Global Capital Index (2Q 2012 and 2Q 2017).
Furthermore, these exposure measures ignore the riskiness of the assets themselves. A bank that holds Treasury debt will be significantly less risky than one that makes illiquid loans with a comparable duration. For these reasons, regulators also measure risk-weighted assets. The risk weights range from zero (for Treasury debt) to more than 100% (for the riskiest loans).
Using risk-weighted measures, capital ratios appear higher (see chart above). But this ignores the ability of banks to “game” the risk-weighted measures by concentrating their holdings in assets with understated risk weights (including, in some cases, sovereign debt). Moreover, hypothetical portfolio exercises reveal that banks’ internal models generate vastly different measures of risk-weighted assets for the same portfolio. For these reasons, supervisors view leverage ratios as a useful supplement to those that are risk weighted.
Capital requirements. Minimum capital requirements are the leading regulatory tool for ensuring resilience of the banks (and bank-like intermediaries). In addition, regulators use stress tests to limit concealed risk and to measure the capital adequacy of banks in scenarios where asset prices are assumed to decline dramatically, markets cease to operate and funding dries up. The combination of higher capital requirements and stringent stress tests has led to a sharp rise in banking system capital in the decade since the financial crisis. But how much capital is enough?
The fundamental theorem of corporate finance—the Modigliani-Miller (MM) theorem—states that, under a specific set of circumstances, firms (including banks) will be indifferent between debt and equity finance. The MM assumptions are clearly violated. If we let the banks choose, they typically minimize reliance on equity funding, which they perceive as expensive (see the opening quote from Jamie Dimon and the reply from Neel Kashkari).
The question of how to set capital requirements depends in part on the factors causing the MM violations that lead banks to prefer debt to equity. The candidates are numerous, ranging from distortionary government debt subsidies (in the form of explicit and implicit guarantees) to information asymmetries that make collateralized short-term debt finance relatively attractive. Raising capital requirements will raise a bank’s private costs. But, to the extent that higher capital requirements reduce the distortions from subsidies and compensate for banks’ ability to conceal risk off balance sheet, social cost will decline.
MM is not the end of the story. As requirements rise on banks, the opportunity to reduce private costs will encourage a shift of risk-taking to non-banks—beyond the regulatory perimeter. One solution to this is to focus regulation on the economic function rather than the legal form of the intermediary (see here). Absent such a system, there will be a point where higher capital requirements, while making banks more resilient, will make the financial system less safe as a result of risk-shifting.
Accordingly, there is a range of views about the appropriate level for capital requirements. At the top end, proponents of narrow banking call for all risky assets to be 100% financed by equity. Admati and Hellwig advocate a leverage ratio of 20% to 30%. The Minneapolis Plan to End Too Big to Fail would set the number in the 15% to 24% range, while Dagher et al argue that 15% would be sufficient to absorb the losses in 90% of past OECD banking crises. Notably, all these proposals far exceed current capital requirements—which are between 3% and 5%—as well as current leverage ratios (computed using a variant of IFRS) of the 30 largest global banks. The latter range from 3.4% to 7.6% (see G-SIBs "Fully phased-in" column of Annex Table C.35 here). In contrast, the Treasury argued in June 2017 that capital requirements for the largest U.S. banks should be “recalibrated” in cases where they exceed international standards.
We do not know the optimal capital ratio for banks. But, in contrast with the Treasury (and with Jamie Dimon), we believe that current capital requirements are not high enough. Our practical suggestion is to raise capital requirements gradually until we observe either a reduced supply of bank credit or a shift of risk-taking to de facto banks. Absent these negative side effects, more bank capital means a safer financial system without loss of efficiency.