The results of the Federal Reserve’s 2013 stress tests are promising, showing that 17 of the 18 largest U.S.-bank holding companies have a resilient base of capital and could withstand three different severe recession scenarios.

Performing well on these tests is highly important to banks and shareholders, because the results will determine whether certain banks may finally return capital to their shareholders. Observers should keep in mind, however, that these stress tests rely on three assumptions that are inconsistent with our experiences during the past financial crisis and limit the real-world relevance of the tests.

First, the tests, based on any of the financial-shock scenarios, assume that the short-term creditors of the largest banks wouldn’t start a run on them. The importance of this assumption is hard to overstate.

According to the Office of Financial Research, the department established by the Dodd-Frank Act within the Treasury to monitor financial stability, this assumption seriously limits the value of these tests for evaluating that stability. It seems entirely plausible that the shock scenarios imposed by the stress tests, which include equity prices falling by 50 percent and the unemployment rate rising to 12 percent, would result in a large bank’s short-term creditors pulling their financing.

Stability Shaken

After the failure of Lehman Brothers Holdings Inc., it was a run by these very creditors that necessitated intervention by the Fed and the Treasury with unprecedented guarantees and buying programs to support the money markets. The stability of the entire financial system was truly at risk when the largest banks stopped lending to one another and sophisticated institutional investors ran on the prime money-market funds that fund these banks.

If this were to occur again, the government would have to intervene or banks would be forced to sell assets at fire-sale prices, almost certainly rendering such banks insolvent.

Second, the stress tests assume that the shock scenario wouldn’t result in banks reducing the credit they make available. In reality, when a bank’s assets are severely reduced in value, that is precisely what it does. The attendant reduction in credit availability magnifies the severity of the shock scenario, creating a vicious cycle. As seen during the crisis, a reduction in large banks’ willingness to lend to other large banks is particularly relevant to financial stability.

Although the Fed’s scenario includes other questionable assumptions, including that banks would continue to pay dividends during a severe recession, a more important point is that the value of troubled assets is not knowable during a crisis.

When trying to value troubled assets, to paraphrase former Defense Secretary Donald Rumsfeld, we’re exclusively dealing with “known unknowns” and “unknown unknowns.” This is precisely why the original stated purpose of the Troubled Asset Relief Program -- to clean up the balance sheets of troubled banks by buying up toxic assets -- couldn’t have been carried out without undue risk to taxpayers.

At the time, there was no way of knowing the actual value of the relevant subprime-mortgage-backed securities and collateralized-debt obligations, and this uncertainty helped drive the general panic that seized up global financial markets. The Federal Reserve stress tests have no way of accounting for this uncertainty.

Thus, although the Dodd-Frank-mandated tests provide an important window into the resiliency of our largest banks, they do not mean that our banks could survive the severe-recession scenarios imposed by the tests.

The Federal Reserve should work toward minimizing the assumptions inherent in the stress tests. By requiring banks to pass these tests in order to redistribute capital to their shareholders, the Fed sends the message that these tests are more relevant to financial stability than they really are.

(John Gulliver has worked for ACA Compliance Group, a regulatory consulting company, and as a research associate for the Massachusetts Institute of Technology Laboratory for Financial Engineering. The opinions expressed are his own.)

To contact the writer of this article: John Gulliver at johnmgulliver@gmail.com.

To contact the editor responsible for this article: Katy Roberts at kroberts29@bloomberg.net.