Large banks (think of Wells Fargo, Citicorp, JP Morgan and Bank of America) today hold equity capital roughly equal to about ten percent of total assets. Inverting that ratio we get assets to equity of ten times or leverage as commonly defined (debt divided by equity, instead of assets divided by equity) of nine times. While this would be enormous leverage for a non-financial company, it is not unusual for financial institutions and leaves the big banks “well-capitalized” from a regulatory point of view. Even though regulatory capital requirements have increased a lot post the Global Financial Crisis (GFC) of 2008, the current standard calls for equity of just seven percent of Risk Weighted Assets (RWA). Since RWA tend to be quite a bit less than total assets, the big banks are easily in compliance.
Nevertheless, many people believe that nine times leverage is too much and banks, particularly large banks, should be forced to further increase capital levels, perhaps dramatically so.
In their book “The Bankers’ New Clothes” economists Admati and Hellwig strongly make the case for higher bank capital requirements in order to reduce the likelihood and severity of financial crises. They argue that banks can fulfill their important economic mission even while operating at much lower leverage levels. Their suggested strategy for migrating to this new state is to require banks to stop paying dividends or buying back stock until higher capital levels are achieved.
A counterargument is that higher bank capital levels will necessitate less bank lending and therefore lesser economic growth.
What is the appropriate level of capital for a large bank? To a great degree it depends on the nature of the bank’s assets. As my colleague Ray Hawkins has argued in an unpublished paper, the appropriate leverage varies significantly with the volatility of bank assets. He produces a measure of the distance from default that varies positively with the ratio of equity to assets and inversely with asset volatility. For example, if bank portfolios were comprised solely of common stock, then in order to achieve investment grade status, the equity to asset ratio would have to be greater than 60%. On the other hand, if asset volatility was lower as it would be with a well-underwritten mortgage portfolio, then the appropriate equity to asset ratio would be lower as well.
In theory, asset volatility can be reduced by well-designed hedging or risk management strategies. For example, a well-underwritten portfolio of fixed rate mortgage loans is a lot riskier if funded with short-term deposits or borrowings than if funded by longer duration liabilities that roughly match the duration of the mortgage portfolio.
Since 1988, the “go-to” driver of quantitative bank capital requirements has been the Basel Committee on Banking Supervision at the Bank for International Settlements (Basel for short). The first iteration, known as Basel I, was unveiled in 1988. Basel I introduced the concept of Risk Weighted Assets whereby relatively low risk assets were accorded low risk weights (for example, 0% for Sovereign debt and 20% for highly rated mortgage securities) and higher risk assets had higher risk weight (for example, 100% for business loans). Given subsequent enormous losses on some Sovereign debts and some previously highly rated mortgage securities, the Basel I risk weighting method clearly required some modification.
Basel II introduced the concept of the “99.9% one-year Value at Risk” as a metric for estimating minimum capital levels. This concept, applied to both credit risk and operational risk, represents the size of the annual loss that would be experienced once in a thousand years.
Effectively what Basel II was aiming at is the concept of “economic capital” defined as the “unexpected” or “worst case” loss minus the expected loss. The idea is that the expected loss should be covered by expected profits and economic capital is a reserve to absorb the unexpected loss. The “one in a thousand year” loss idea reflects the notion that large banks should strive to be rated AA, and the historical failure experience of such highly rated companies is a very low annual default rate.
A major problem with this notion of economic capital is that it is impossible to measure with any degree of accuracy. Rather than defining the “worst case” as the event of failure, a more operational definition would be to define “worst case” as the event of ratings downgrade of two full notches (to my knowledge, this idea was first proposed by Ray Hawkins in the above mentioned unpublished paper). A major bank that suffered such a downgrade would find that its operations would be adversely affected, even though it may still be solvent. Historical data suggest that the probability of a two full notch downgrade is unlikely but not remote. Ratings migration data suggest that the probability of a downgrade from AA to BBB or from A to BB in one year is about 1%. Using this standard the idea would be to investigate whether a bank can survive the “one in one hundred” extreme scenario, and still have sufficient capital to avoid a major ratings downgrade. Operationally, this would mean maintaining a high equity to asset
ratio after any losses associated with the extreme scenario, or suffering only a modest increase in the probability of default.
A good way to implement this idea is through stress testing. The basic idea is to conjure up a few stress scenarios, estimate the bank’s likely losses in each of the scenarios, and make sure that the equity to asset ratio after taking the losses is still sufficient to maintain the target credit rating. The Federal Reserve has implemented a stress test regime for large banks. Under this regime (called the Comprehensive Capital Adequacy Review (CCAR)) the Fed prescribes a small set of economic scenarios designed to ascertain whether banks have adequate capital to absorb losses in stressful environments. Banks are required to estimate their losses under the prescribed economic scenarios. The Fed does their own calculations as well and may not approve share buybacks or dividends if a bank does poorly per the Fed’s calculation.
Since stress testing seems like a good idea, one would think that banks would have come up with the idea without the aid of the regulators. And they did. Just to take one example, in the late 1990s Chase Manhattan Bank (now part of JP Morgan) regularly reported on the effects of a series of stressed economic scenarios on bank earnings and capital. The regulatory stress test requirement dates back to the financial crisis year of 2008 when a stress test (SCAP for Supervisory Capital Adequacy Program) was applied to the largest banks in an attempt to convince the public that the banking system was sound (after injection of hundreds of billions of capital through the government Troubled Asset Repurchase Program (TARP)).
I like the idea that the regulators design a set of stressed economic scenarios and then request that banks calculate their earnings and capital in those scenarios. I also like the idea that the regulators do their own calculations to compare against the banks’ results. I do think a major improvement in the process would be for the regulators to be more transparent with respect to their internal calculations. Currently, it is pretty much a black box and that is surely a source of concern among bankers. More transparency could be an important factor in pushing both sides to improve the quality of their models and methodologies. Maybe the science of financial risk management will have advanced before advent of the next financial crisis.
What are the growth consequences of higher capital requirements? I think there are arguments on both sides. First, bank management teams focusing on the return on equity may feel pressure to raise lending rates, or exit businesses where that is not feasible. But I don’t think risky lending will be terminated. After all, the majority of investments in equity securities are funded with zero leverage (think of mutual funds and pension funds). Economist John Cochran is calling for risky loans to be financed in a similar way, with 100% equity or long-term debt. The effect of higher bank capital requirements would probably be two-fold – somewhat higher lending rates and lesser chance of financial panics. The net effect may well be positive for economic growth.