The Fed's stress tests should be more stressful. Photographer: Andrew Harrer/Bloomberg
The Fed's stress tests should be more stressful. Photographer: Andrew Harrer/Bloomberg

This week, the Federal Reserve will present the results of stress tests designed to ensure that the largest U.S. banks won't turn the next financial crisis into an economic disaster. There's just one problem: If the tests were realistic, most of the banks would fail.

Previous stress tests helped to restore confidence in the U.S. banking sector after it almost collapsed in 2008. The idea is to show what would happen to standard measures of financial strength, such as capital ratios, in the worst of times. The tests force banks to consider risks that, in their quest for quarterly profits, they might otherwise ignore.

The exercise, however, is only as good as its design. If the scenarios don't reflect what would happen in a real crisis, or if the bar for passing is set too low, they can create a false sense of security. Regulators are sometimes inclined to be permissive, because flunking too many banks might trigger a crisis of its own.

What, then, would a rigorous stress test say about U.S. banks today? Researchers at New York University have created a tool to answer the question. Drawing on the historical relationship between banks' stock prices and the market as a whole, it estimates what the value of the banks' equity capital would be if the market fell 40 percent over six months -- similar to what happened in 2008. It then calculates, based on past crises, how much capital the banks would have to raise to be financially sound.

The results aren’t pretty. Using a start date of Sept. 30, 2013, the same as that of the Fed's latest round of stress tests, the NYU model gives only one of the six largest U.S. banks -- Wells Fargo & Co., Inc. -- a passing grade. The other five -- JPMorgan Chase & Co., Bank of America Corp., Citigroup Inc., Goldman Sachs Group Inc. and Morgan Stanley -- would have a combined capital shortfall of more than $300 billion. That's not much less than they needed to get themselves out of the last crisis.

On the surface, the Fed's latest stress test looks at least as tough as the NYU researchers'. It envisions a 50 percent drop in the stock market, and, in a new twist, subjects the biggest banks to the default of a major counterparty. With access to much more detailed information on the banks' holdings than the folks at NYU could ever hope to see, the central bank should be able to come up with a much more accurate and thorough assessment.

Unfortunately, the Fed's approach ignores a lot of the horrible things that actually happen in crises -- things that NYU's simpler approach implicitly captures. Banks' borrowing costs, for example, tend to rise, killing profits that could offset their losses. Trouble at one bank can spread as investors wonder which others will be affected. Credit freezes can force financial institutions to sell assets at a loss, setting in motion downward spirals in which falling prices and banks' woes reinforce each other.

The Fed also makes a passing grade too easy to achieve. Participating banks, for example, must maintain a leverage ratio of at least 4 percent, or $4 in capital for each $100 in assets. They should not be allowed to get anywhere near such a level, which research and experience suggest is well below what's needed to avoid distress. In a crisis, banks should be a source of strength, not propagators of panic.

It doesn't have to be this way. The Bank of Canada, for example, includes the effects of credit freezes and contagion in its stress tests. If the Fed did the same, and checked its results against a market-based approach like that of NYU, its conclusions would be a lot more believable.

To contact the senior editor responsible for Bloomberg View's editorials: David Shipley at davidshipley@bloomberg.net.