The Fed’s Stress-Test Charade

From this morning’s New York Times:

Every year, the largest banks have to undergo a theoretical test to assess how they might fare in a period of extreme stress in the markets and the economy.

Now, that test may become substantially tougher.

A senior official at the Federal Reserve, the bank regulator that administers the so-called stress tests, said on Monday that the Fed was discussing changes to the test that could make it harder for banks to pass. . . .

In an interview with Bloomberg Television, Daniel K. Tarullo, the Fed governor who oversees regulation, said there was “more than a pretty good chance” that banks would have to have a higher amount of minimum capital left after suffering the theoretical losses envisioned in the stress tests. [Emph. added.]

Who can possibly think this is sensible regulation? The large banks have had to undergo annual stress tests since 2007. One would assume that, by now, the geniuses at the Fed have given the matter enough thought that they’ve been able to develop a stress test that, administered prudently, they believe really will help protect the financial system. But no.  Eight years on, Dan Tarullo is tacitly admitting that he still hasn’t gotten things figured out. If Tarullo thinks that, after all this, banks still don’t carry enough capital, why did he even administer the stress tests in the first place?

Or maybe the Fed’s overarching stress-test goal isn’t to protect the system but rather to be to make life miserable for the big banks. How else to explain the lengths the Fed will go to make the entire process utterly opaque? In preparing their tests, banks get little help in the way of what assumptions to use, or even much guidance as to what results will count as “passing.”

We’re thus apparently getting bank regulation by whim. How much capital should the banks have? It seems to depend on how one unelected Fed governor, Dan Tarullo, feels when he wakes up on a given day.

This isn’t just expensive and annoying, it’s dangerous. First, it’s not just a lack of capital that can put the financial system at risk. It’s also a lack of liquidity. In the midst of a panic, remember, no one trusts the value of each other’s assets, and so everyone stops providing liquidity to each other all at once–regardless of how ample stated capital levels might be. That’s why the Fed’s fixation on capital, and its fetish that stress tests will ensure that the banking system has enough of it, is kind of beside the point.

But more important, this whole notion that stress tests can prevent disaster is fundamentally flawed. A stress test is, at bottom, remember, a mathematical model. A very, very complex mathematical model that purports to predict the future. You know, like the models the rating agencies used to justify triple-A ratings for portfolios of subprime mortgages, or that subprime CDO issuers used to map out payout waterfalls to holders. I can’t believe I even have to type this. If anyone—notably financial regulators—ought to have learned anything from 2008, it’s that, when a crunch happens, even the most “sophisticated” predictive models will go haywire, and that any prior reliance on them will likely end up making a bad situation even worse.

Instead, the Fed has responded to a colossal failure of financial modelling with even more complex financial models! Which of course makes no sense. By all means, the banking system should carry more capital than it did before the crunch. (Back then, the geniuses of Wall Street had somehow convinced themselves that they had balance-sheet risk all figured out.) But this stress-test nonsense may be providing a sense of security that’s unwarranted. Rather, Dan Tarullo should figure out how much capital he wants the banks to carry, make sure they carry it, and then get out of the way.

What do you think? Let me know!