Friday, May 8, 2009

Stress Test Math Makes Absolutely No Sense

I just can't get too excited about these stress test results. It's the mathematician in me.

Sure, it sounds great: the 19 largest banks need only $75 billion, total. They emerged from their Geithner shakedown cruise not in tip-top shape, but looking far better than anyone thought.

Sure, it sounds great, if you don't think too hard about the numbers.

It's simple really. New York University Professor Nouriel Roubini, one of the few economists who was eerily prescient about the magnitude of this crisis, forecasts $3.6 trillion of losses to come on U.S. loans and securities, more than half of that coming from banks and broker-dealers.

What do the stress test results imply about losses to come? The 19 banks that were examined control two-thirds of the banking system's assets. If you assume for simplicity that they hold two-thirds of the "losses to come" (though in truth they probably hold more, having dabbled in riskier assets more recklessly), that seems to imply the U.S. banking system will lose no more than $113 billion (since these banks total must raise no more than $75 billion).

And that makes absolutely no sense. Roubini's math (and not just his -- but the calculations of others as well, including the International Monetary Fund) suggests the U.S. banking system is looking at another trillion dollars or so of value evaporating off the books. If these stress-tested banks shoulder about $700 billion of that -- at least -- how does $75 billion in capital stuff that hole?

And who is dumb enough to provide that capital, knowing this?

We already know the answer to that last question. Look for more government aid, probably through the backdoor, to prop up the banks. This seems to be the Obama Team's strategy for 2009 for dealing with the financial crisis: try to minimize problems, paper over gaping holes, and silently slip the banks (through cheap funds, multiple lending facilities, bailouts galore) enough money to keep them tottering along.

No comments:

Post a Comment