Rosengren cited a soon to be issued paper by the Boston Fed estimating the losses at 26 of the largest banks during the crisis. Half of those banks would have suffered losses equal or greater to 200 basis points of their capital. Eight had losses greater than 450 basis points.
What this means is that a bank starting with a seven percent capital ratio—which is where Basel III sets the minimum—would have been reduced to a ratio of just 3.5 percent during the crisis.
At that level, it's likely that even without regulatory action a bank would fail, because investors would panic, counter-parties wouldn't be willing to take the risk of doing business with a bank so close to insolvency, and lenders would balk at providing financing.
"Given the losses in excess of 450 basis points at a number of large institutions during the last financial crisis, minimum standards for large institutions may be too low," Rosengren said.
It helps when considering these arguments to get back to the basics about what capital requirements are meant to do. Although the common description of the requirements as "capital cushions" that "absorb losses" is accurate as far as it goes, it can be a bit misleading.
Required capital cannot actually absorb any losses of an operating bank because if you fall below the requirement, you either get seized by the government or bailed out.
The real purpose of a required capital level is to trigger a series of actions—selling assets, raising new capital—when a bank suffers losses that bring it near the regulatory threshold. It's sort of a flashing yellow light signaling that its time for a bank to take steps to avoid any further capital erosion.
There's a slightly strange recursiveness about the way markets treat required capital levels. On the one hand, regulators attempt to set the required capital threshold high enough that counter-parties and investors will not worry that any operating bank will have trouble meeting its liabilities.
So, for instance, the Federal Reserve's annual Comprehensive Capital Adequacy Review (CCAR) required that bank capital exceed 5 percent Tier 1 common equity, because the regulators at the Fed think that is high enough to prevent counter-parties and investors from panicking and starting a run on the bank.
That's why Rosengren points out that half of the 26 banks in the Fed study saw their capital ratios fall by 200 basis points or more—that's exactly the distance between the standard Basel requirement and the five percent level the Fed figures will trigger a run on the bank.
On the other hand, the threshold's themselves create incentives for runs. If a bank approaches a regulatory threshold, investors may sell sharesin the bank for fear of being diluted by new issuances, or seeing their dividends cut.
Meanwhile, counter-parties may worry that the new capital may not be raised fast enough to prevent a regulatory seizure—so they will attempt to withdraw funding before they are made to wait through a bankruptcy or resolution process.
This concept of a regulatory run on the bank is one reason bank stocks rally when they do things like monkey with their risk-weightings, as Deutsche Bank did last month. This doesn't do one thing to improve a bank's underlying financial condition.
However, so long as it signals that regulators won't push the bank to dilute current shareholders or prevent dividends from being paid, investors will bid up the stock of a bank when better numbers signal regulatory compliance rather than financial health.
A run on the bank isn't a crisis with a smaller bank. So you don't need the extra capital cushion. If it's losses bring it down below a level deemed by markets or regulators to be tolerable, it gets seized by the FDIC and has its assets sold off to a healthier rival.
The crux of Rosengren's argument is that capital requirements for the biggest banks are different. They need to be set higher not just to provide a cushion in the case of an actual base failure but to prevent a run on the bank because their failures are too costly for the system.
So should we panic? Are the big banks still undercapitalized? After all, during the crisis 8 banks suffered losses equal to 527 basis points of the capital ratios. This means that even if they had Basel III's Teir 1 seven percent plus the full 2.5 percent SIFI surcharge—for a total capital ratio of 9.5 percent—they would still have fallen down below the run-on-the-bank trigger of five percent.
Rosengren thinks that other tools such as stress tests, tougher liquidity requirements and resolution plans might be effective supplements to capital requirements. Because they may act to reassure the markets about the health of a bank (stress tests, liquidity requirements) and about access to capital in the event of a bank failure (resolution), they may effectively lower the market trigger for a run.
Notice that these new supervisory tools depend critically on the credibility of regulators.
If the market loses confidence in the effectiveness of stress tests or regulation—neither of which has actually been tested in a crisis—they could backfire. The failure of a bank that had been given a regulatory pass on a stress test, for example, could trigger doubts about even healthy institutions that passed the test. A chaotic resolution of a large bank might cause a panic in creditors to other banks with approved resolution plans.
In short, there's no real question about whether our capital requirements are too high. It's a question of whether the new supervisory authorities of bank regulators are effective enough to allow capital requirements to work when they are too low.