Stress Testing Is What Got Us Here

Treasury Secretary Geithner’s plan to repair the banking system fell flat because, while he promised something bold and forceful, he delivered, “a carefully designed and comprehensive stress test, to use the medical term.” A beleaguered banking industry awaiting decisive leadership is asked to hop on the treadmill.

Geithner, in the spirit of the Obama administration, began with the requisite fear-mongering. The Great Depression makes a cameo appearance as does Japan in the 1990s. Then he went technical, talking about what is apparently a favorite subject of his: stress testing.

For the uninitiated, stress testing is a process whereby banks calculate their capital adequacy at statistically unlikely but possible extremes. It has been done for decades at most banks. So it was odd to me, having worked at two large banks that employ thousands of quantitative analysts to do this “value at risk” analysis, to hear the Treasury Secretary talking about it like it was something he just thought up.

Worse, one would hope the man charged with fixing the crises would have some appreciation for what started it. And for this crisis, it is fair to say stress testing is what got us here.

Legions of quantitative analysts at the big banks stress test all forms of market risks: foreign exchange, interest rates, credit, volatility, etc. Most of them missed the current crisis because their work is backward looking. Stress testing involves looking at the volatility of assets and their correlations to determine a bank’s exposure to “worst case” moves. The most toxic assets the banks now own had shown very little volatility before they fell off a cliff. This is why the banks owned so much of them.

Stress testing for interest rates or foreign exchange is easily done because there are long and continuous time series of data. One can calculate all the statistical variables and the correlations among them. Banks do this well, which is why it rarely happens in modern banking, unless by fraud or error, that a large bank takes a big hit on generic market risks.

Enter the super-senior tranches of CDOs backed by subprime mortgages. All that was really known about these securities was that they were rated AAA and before the crises traded very close to par. The total history was about 3 years. Banks used ratings as a proxy for risk and got burned.

The problem was not one of insufficient testing but of insufficient imagination. The model might indicate that these AAA tranches should never move by more than a few points. Some banks took that as reason to buy tens of billions of dollars of them. Some who asked “What if?” have fared better.

Perhaps Geithner is naïve as to the role of stress testing in the crisis thus far, but as former head of the Federal Reserve Bank of New York, he should not be. What he proposes now is even more bizarre, using stress testing to determine capital needs. The assets that are most troublesome are also the ones most difficult to price, the so-called “toxic assets.” They cannot be priced, but Geithner somehow believes they can be stress tested?

Simplified, a stress test requires very good information about at least two things: the current price and the historical prices. Ideally one would know something about an asset’s correlation with other asset classes. We have none of the three.

Strangest of all was Geithner’s punchline: We will employ stress testing to determine which institutions are most in need of additional capital. And then what?

One might naively have thought that the reason for mentioning Japan in the 1990s was because Geithner had chosen some form of receivership for those revealed to have the weakest capital positions. Oddly, he proposes to give them the capital they need. In effect, do what Japan did and hope for a better outcome.

Some of the smartest minds on the planet have been noodling for 2 years on this banking crisis, and there are many ideas floating around. One might point out that the marketplace of ideas has been doing its job in thrashing around different alternatives. Not surprisingly, there is not a consensus on what to do next.

Then, in the middle of this maelstrom, the young new Treasury secretary tells the world that next Monday he’s going to announce a comprehensive plan to fix this whole thing. Check that. Tuesday.

It is not surprising that the market fell sharply after this underwhelming plan was announced. Part of the problem may be that Geithner’s plan showed a misunderstanding as to what caused the crises; but part of it may be that his remedy seems so unlikely to work. It’s a fair bet that those banks which stress testing shows need more capital will be the banks who have already taken the most from TARP. Geithner announced that to whom much has been given, even more will be given.

The plan does not seem serious because seriousness will require making choices about which banks to let fail. It may require admitting that those most in need of capital are most in need of shuttering. The approach of saving everybody cannot work. Even if stress testing could determine which are which, it would still take leadership to make the right choices.