In an op-ed in last week’s Wall Street Journal, Thomas Hoenig, Vice Chairman of the FDIC, and Sheila Bair, former Chairman of the FDIC, argue that potential changes to bank capital requirements proposed by the Fed and OCC on April 11 would not increase bank lending and would increase the likelihood of another financial crisis. Their argument is based on two false premises: first, that leverage measures are superior to risk-based measures of capital (including stress tests) for evaluating bank risk, and second, that lowering capital requirements does not boost bank lending.
Their article begins, though, with an ad hominem argument that deserves a quick rebuttal. They note that profits for America’s biggest banks “are breaking records – for some, soaring 35% to an all-time high,” and state that they are therefore concerned that the Federal Reserve and OCC would allow greater leverage. One does not generally tie regulation to profitability – do OSHA and the FAA inspect profitable companies more than unprofitable ones? And would Mr. Hoenig and Ms. Bair be suggesting a relaxation of the leverage ratio if large banks were suffering large losses – or does their regulatory ratchet only turn one way? Finally, even if profitability were relevant, any real investor or analyst would never look to the level of profits not growth; after 8 years of GDP growth, every industry is earning record profits. Rather they would look first to return on equity (ROE), and it turns out that large bank ROE still trails pre-crisis levels (see graph) and most other industries, as we point out here.
But on to the meat (or fat) of their argument. A leverage ratio is calculated by dividing equity by assets. When used as a means to evaluate the risk that a bank will fail, it assumes that all assets – whether a Treasury security or a delinquent construction loan – have the same risk. A risk-based capital ratio instead divides equity by risk-weighted assets, where the weight for each asset category depends in the risk of that type of asset. Thus, the less risky the assets, the higher the capital ratio. Similarly, when the Fed conducts stress tests, it estimates bank losses for each type of asset in a severe economic downturn and requires banks to hold enough capital to cover those losses and still meet its minimum capital requirements. Because, in general, riskier assets make more losses, stress testing ends up requiring banks to hold more capital against risky than safe assets.
Certainly, risk-weights can sometimes be wrong. Banks underestimated the risk or subprime mortgages in the crisis, though they got cash and Treasuries and corporate debt pretty much right. But a leverage ratio will always be wrong. There has never been a time when the loss experience for cash was the same as for subprime loans.
Thus, not surprisingly, and in contrast to Bair and Hoenig’s assertions, risk-weighted measures do better than simple leverage measures predicting bank failures. As we show in a previous research note using analysis that is easy to replicate, when a risk-based measure and a leverage measure are used side-by-side to predict which banks failed in the crisis, both measures are statistically significant, but the leverage measure enters with the wrong sign. That is, holding constant a bank’s risk-based capital ratio, the bank is less likely to fail if it has a lower, not higher, leverage ratio. Why? Because the bank with the lower leverage measure has a greater proportion of low risk (and probably highly liquid) assets.
Which leads naturally to the next point: Hoenig and Bair emphasize that the largest banks have relatively low leverage ratios compared to smaller banks. What they fail to mention is that, as reported on the table of capital ratios released by Hoenig last week, larger banks have higher risk-weighted capital ratios than smaller banks. The reason for this profile – lower leverage ratios but higher risk-based measures—is that larger banks are exactly in the situation highlighted by the statistical analysis just discussed: they hold disproportionately large amounts of cash, Treasuries, and other low risk and highly liquid assets. Correcting for the relatively high share of those assets held by large banks, they are less, not more, risky than smaller banks.
Leverage ratios are not only poor measures of bank risk; they can also be highly counterproductive. In particular, when leverage requirements are high relative to risk-based requirements, they create undesirable incentives for banks to increase the riskiness of their portfolios. We explain this incentive more completely in the research note cited above, but in a nutshell, a bank bound by its leverage requirement will have an incentive to increase the risk of its risk-weighted assets until it is also bound by its risk-based requirements; until it does, the relative risks it chooses to take have no impact whatsoever on how much capital it must hold.
This is not a controversial point. If you read the transcript of the public Board meeting in which the Fed approved the current leverage ratio requirement (which is about twice the international norm), each Board member emphasized that he or she did not want the leverage ratio to be the binding requirement because of the unwanted incentives that that configuration creates.
Hoenig and Bair provide an excellent example of the danger of focusing exclusively on the leverage ratio when they state that the Fed and OCC proposal is flawed because the banking agencies should be requiring building capital in these good economic times. That is exactly what the agencies do. In the annual stress tests, the Fed and the banks project capital ratios under the assumption of an extraordinarily severe recession and market collapse, projected economic calamities that become worse as the actual economy strengthens. Specifically, because the scenario design principles require the unemployment rate to always quickly rise to at least 10 percent, as the actual unemployment rate falls, the projected increase becomes larger and more rapid. As we described in a recent post, the stress test scenario this year is more severe than last year and includes declines in equity, house, and CRE prices, and a widening in BBB spreads, that are materially more sudden and severe than occurred in the recent financial crisis. While we have written extensively on the severe unintended consequences on the Fed’s reliance on a single implausible scenario each year (see here, here, here, and here), the general design principle that the tests get tougher as the financial cycle peaks is sound and bakes in the countercyclicality that Hoenig and Bair call for.
Hoenig and Bair also maintain that “lowering bank capital requirements boosts lending is urban legend.” If so, it is a legend that is apparently believed by virtually the entire economics profession. Every evaluation of the cost and benefits—including those done by the Bank for International Settlements (the parent of the Basel Committee on Banking Supervision), the Federal Reserve, the Bank of England, the International Monetary Fund, and the Minneapolis Fed – begin by estimating the decline in lending and economic activity that are caused by higher capital requirements.
The Fed and OCC proposed modifications to the capital framework that rotate capital requirements away from the leverage ratio and toward stress tests. The Fed and OCC jointly proposed that the agencies’ leverage ratio requirement for GSIBs (the supplemental leverage ratio or SLR) be reduced to the international norm (rather than about twice the norm). And the Fed proposed that stress tests results be knit more closely into large banks’ ongoing capital requirements. (One wonders, by the way, how Mr. Hoenig and Ms. Bair would approach stress testing: assume that every asset behaves the same way under stress?) The proposals continue the larger policy shift in the United States toward the use of stress tests, as opposed to simpler capital measures, as the key determinant of how much capital larger banks must hold. At the same time, the proposals recalibrate existing leverage requirements from a sometimes binding capital constraint to a strong backstop to risk-based requirements and stress tests—their intended role.
By increasing the emphasis on more risk-sensitive measures, the Fed and OCC proposal will lead to stronger, safer banks. Building on their two false premises, Hoenig and Bair instead argue for a leverage-ratio centric approach that will inevitably lead to riskier banks. Moreover, by denying the implications for economic growth at the core of regulatory design, they demonstrate that they should not be trusted participants in the current effort to build a more efficient and more robust capital framework.
Disclaimer: The views expressed in this post are those of the author(s) and do not necessarily reflect the position of The Clearing House or its membership, and are not intended to be, and should not be construed as, legal advice of any kind.