In early 2009, the Federal Reserve "stress tested" major banks, modeling how much each could lose if the economy got worse, to ensure they had enough capital to stay afloat.
The Fed's assumptions assumed a worst-case scenario of unemployment hitting 8.9% in 2009. Within days of the test's completion, the actual unemployment rate was 9.4%, and it hit 10% later that year.
Hence the flaw with forward-looking stress tests: They're only as useful as the assumptions are realistic.
The Fed released another round of stress tests earlier this week. Most major banks passed, although some, including Citigroup
That's great. But just like 2009, we have to ask: How realistic was the test?
It looks like the Fed learned a lesson and was more punishing on the downside this time. The worst-case scenario (not a forecast, mind you) assumes the economy slips back into a major recession this year that would be far worse than the one we just went through. It assumes real gross domestic product falls nearly 8%, incomes fall 7%, and unemployment spikes to 13%.
It also assumes the Dow falls to 5,500, or a 60% decrease from current prices, before rebounding sharply:
Source: Federal Reserve.
And it assumes nationwide home prices fall another 20%, without a rebound:
Source: Federal Reserve.
So far, these are tough assumptions that truly test a bank's resilience against losses. That most major banks can withstand this kind of shock underscores a key difference between today and four years ago: Banks are in much better shape.
But another part of the Fed's assumptions are more suspicious. The hypothetical recession assumes interest rates drop to an all-time low, with 10-year Treasury yields falling to 1.6%, and short-term interest rates staying pegged at 0%.
That might happen. It's close to what happened over the past few years, after all.
But such thinking demonstrates one of the most common flaws in forecasting: assuming the next recession will be just like the last one.
It probably won't. Every postwar recession has been unique. In the early 1980s, inflation surged and interest rates spiked, but stocks actually rose. In the early 1990s, real estate and stocks fell, but interest rates didn't do much at all. After 9/11, stocks plunged, real estate surged, and interest rates fell only moderately.
The point of a stress test is not to incorporate every possible scenario, but it's helpful to ask what might happen if the deep recession the Fed assumes is instead accompanied by rising inflation and higher interest rates. With the amount of quantitative easing the Fed has engaged in, that's hardly a bold assumption. And it's particularly important when forecasting bank losses, since interest rates are instrumental in dictating bank profits and loan defaults.
It's hard to know exactly what might happen if interest rates rise, but it would almost certainly be more stressful on banks than the Fed currently assumes. Loan defaults would likely be higher, profit margins might shrink, and loan growth would probably stall, curtailing earnings. For banks that don't have a strong deposit base and rely on short-term funding, things could get really ugly.
That isn't a forecast. But it shows how meaningless the stress test could be even if we head back into a deep recession like the Fed tested for, but one that looks different than it assumes. The biggest risk -- the worst-case scenario -- is almost never something you can think of, and isn't likely to mimic what just happened, like the last recession. "What looks like tomorrow's problem is rarely the real problem when tomorrow rolls around" James Fallows once wrote. Alas, that's not something any stress test can model.
For more on the recession's impact on the economy, check out my latest ebook, 50 Years in the Making: The Great Recession and Its Aftermath, for your iPad or Kindle, from Amazon or Barnes & Noble. It's short, packed with information, and costs less than a buck.
Check back every Tuesday and Friday for Morgan Housel's columns on finance and economics.