Federal Reserve's Stress Testing Shows Politics, Not Math

Photograph by JB Reed/Bloomberg

The Federal Reserve released the results of stress tests of 19 of America’s largest financial institutions at 4:30 p.m. on Thursday. But at any time this morning you could have already known which banks contribute most to systemic risk: Bank of America, JPMorgan Chase, and Citigroup, in that order.

You could have found them listed by name in the V-Lab data from the Volatility Institute at the NYU Stern School of Business. The institute, run by Robert Engle, a financial economist who won a Nobel Prize in 2003, has come up with a way to estimate how much equity each firm would lose in a crisis, and how much each would contribute to the total expected capital shortfall of the entire financial sector. This month, the potential capital shortfall stands at $500 billion. Of that, 18 percent belongs to Bank of America.

The V-Lab’s method isn’t the only way to estimate systemic risk by using publicly available data. We don’t need the Fed to tell us what the danger is—and which firms are the most dangerous. The people who run the Fed’s stress tests are very, very smart. They believe in what they do. They have done a lot of work since 2008 to understand financial risk, and they have hinted that they’re using the V-Lab model, among others. But the task they’ve been set to fulfill is not one of math. It’s political.

In January of this year, Ed Kane, a professor at Boston College and a lonely voice of caution on bank risk before the crisis, gave a talk called “Gaps and Wishful Thinking in the Theory and Practice of Central-Bank Policymaking.” Bank regulators, he said, have not progressed sufficiently through the Kübler-Ross stages of grief over the failure of the assumptions that guided them before 2008. They are still bargaining, he believes. As he said in his talk:

Their treatment plans inevitably misperceive capital and misweight risk. Authorities acknowledge that financial crises are socially costly to cure, but they pretend that crises can be avoided by aligning a firm’s deceptively understated leverage with politically negotiated conceptions of its exposure to risk.

The parameters of the stress tests are all subject to intense negotiation. They allow for “risk-weighting,” for example, which presumes that some asset classes are more acceptable and require less capital as a buffer. Weighting assumes that we can know what’s going to go wrong in the future and that we can trust banks not to shuffle assets around and rename them to beat a rule.

The stress tests, mandated by Dodd-Frank, are supposed to produce a result. For the Fed, it’s to show that the banks fall plausibly within a negotiated definition of “safe.” For the banks, it’s to be allowed to distribute some of their capital to shareholders, rather than hold on to it. For taxpayers, though, the goal of a stress test should be to prove beyond a reasonable doubt that the banks will never, ever have to come back to us for money.

The V-Lab tests, far simpler and without any of the proprietary data the Fed gets, shows that the banks aren’t nearly there. Several additional approaches that use publicly available data show that, despite all the regulation that has been put in place in the last five years, markets still price in a likelihood that large banks will be bailed out. The stress-test results won’t change that, so their release is not news. It’s just an event.

    Before it's here, it's on the Bloomberg Terminal.