Duke Research Advantage

By Paul Zipkin

Here we are in early fall, 2009. It was a fairly calm summer. There were no major financial disasters. The stock market revived somewhat. The economy is still troubled, but compared to the unremittingly awful news during the previous year, it’s been a good time. I believe some credit for this happy turn of events is due to the bank stress tests conducted by the government during the spring.

The tests calculated what would happen over the next couple of years under two scenarios, a base-case scenario and a bad scenario reflecting worse economic conditions. They considered each bank’s entire position, not just standard deposits and loans but also fancy mortgage-backed securities and swaps. The question was, how much capital would each bank need to continue operating normally under the bad scenario?

The method itself is not at all novel. Many businesses, military forces and other organizations do this sort of scenario planning all the time. No one knows exactly what will happen in the future, so it makes sense to simulate several possible futures, to estimate whether the organization’s plans are robust.

But, amazingly, this was the first time the government conducted such a test of our financial system. We do have elaborate financial regulations, but those specify procedures and check whether they have been followed. In other words, they look backwards, not forwards. To look ahead at several plausible scenarios, the exact same ones for all banks, and to include all the banks’ exposures – that was unprecedented. Of course, the inspiration for the stress tests was the sequence of ugly surprises in 2008. Nobody knew how dire the situation was at Lehman and the others until it was too late.

Banks are supposed to do this sort of thing for themselves, of course, but it’s not clear how seriously they take these “risk management” exercises. Anyway, they are not required to use the same methods and certainly not to make the results public.

Before the results were revealed, many observers criticized the tests. The tests were not rigorous enough, the bad scenario was too tame, they said. (The Treasury invited some of this carping by calling the bad scenario the “worst-case” scenario. It was not that.) But the criticism largely fell silent once the tests were complete and the results announced on May 7.

And what were the results? In a word, bad, but not as bad as some feared. Many banks needed a lot of additional capital, but none was near disaster. Furthermore, the banks promptly raised the capital they needed. It was pleasantly surprising that they were so easily able to do so.

This was about the best outcome that could be imagined. If the tests had found that everything was fine, that all banks were adequately capitalized, I doubt anyone would have believed them. If all the banks were hopelessly insolvent, well, it’s hard to imagine anything but a calamity. The moderate problems the tests actually revealed were, in this sense, perfect.

Did the Treasury rig the tests to come out this way? I don’t believe that for a moment, and if you do, you should be ashamed of yourself.

Anyway, once the test results were announced, the overall economic mood seemed to lift. That set the stage for the relatively happy summer we enjoyed.

I suggest the government do similar stress tests regularly, say once per year. If so, the banks should be grateful. If not, they should hire somebody else to do it. I think we need more than just two scenarios, but that’s a quibble.

It is worth pointing out that this sort of intricate quantitative analysis has received much criticism lately. Some even blame it for the whole financial crisis. But finance is all about numbers, and we really can’t do without them.

Heaven knows, another disaster could erupt any day. But for now, let’s enjoy the pleasant chill in the air and the first of the brilliant fall colors.