Those who may chuckle at the cutting phrase “no good deed goes unpunished” might appreciate how it fits today’s banking crisis in one respect: Regulators who, post-crisis, devised so-called stress tests to uncover – and ultimately repair – bank weaknesses, could inadvertently spark a bank run by disclosing the test results. Yet, without some level of public disclosure, the stress tests might not have the intended market disciplinary effect.

But there is a path away from this dilemma, according to a recent paper by Wharton finance professor Itay Goldstein and his colleague, Haresh Sapra of the University of Chicago Booth School of Business. In the paper, titled “Should Banks’ Stress Test Results be Disclosed? An Analysis of the Costs and Benefits,” the two argue that some fixes to the stress test process could beef up weak banks by identifying problems before they blow up and also guard against damaging disclosures that could trigger a bank run.

“The disclosure of stress test results can be beneficial because they promote financial stability,” says Goldstein, and they can be “quite useful in providing market discipline for individual banks and helping with the accountability of regulators.” That all provides the right incentives.

But given banks’ deep vulnerability to the disclosure of financial deficiencies, indiscriminant disclosure can cause problems. If depositors and investors lose confidence, they will withdraw their money and possibly cause a bank to fail. And when disclosure makes bad news public, markets tend to amplify it, Goldstein points out. Even those who may believe a bank under fire is actually sound will abandon the institution because they must account for “what others may think.” A bank-run scenario can become a self-fulfilling prophecy, so regulators must carefully consider what information to release.

What is more, publicly disclosing stress test results can have an adverse effect on the behavior of bank managers, who may game the system by putting too many resources toward passing the stress test, rather than doing what is best long term. “If a manager does not know [the test criteria] there is less room for manipulation or gaming,” Goldstein says.

The authors’ recommendations? Any disclosure of problems with an individual bank should be precise – to avoid misunderstandings – and should come with a prescription for corrective action, so that markets can see that the situation is not necessarily dire. “The danger arises if we just say we found that a bank is in trouble, and don’t follow up with action,” Goldstein says. Instead, the message should be “We found a problem and now we are taking these steps to make things better.”

If regulators cannot meet that bar, then they should instead issue a system-wide evaluation offering aggregate results, and avoid singling out any one bank. While stress testing can be good for the banking system as a whole, regulators have to be “more careful with individual banks,” Goldstein points out.

System-wide disclosure can also moderate a more subtle challenge, where disclosure can adversely affect the inner workings of the market – specifically, how investors trade based on information. Market prices reflect the total information held by market participants, who trade on it. Once regulators provide more information via disclosure, some participants may feel they are losing their information advantage and thus could lose interest in trading in the area. That makes it more difficult for regulators to learn about the market from price signals, which in turn could degrade the regulatory effort. Aggregated disclosure can also help sidestep that problem.

So, could stress tests, introduced after the financial meltdown, have prevented the disaster? “They potentially could have made banks more prepared and reduced the extent” of the crash, Goldstein says. “But it’s always a challenge. We are prepared for things that happened in the past,” not in the present. “People talked about subprime and fragility for some time before the crisis, but the common thought was that it was a local problem that could affect some banks, but any damage would be contained and so it was not a big problem. They failed to understand the macro or general equilibrium implications — once one banks folds, it affects others that are connected to it and this contagion was hard to predict…. You can’t predict everything.”

Once a crisis is underway, however, full disclosure is the best course. “There is no choice but to disclose during a crisis.” There already is so little confidence that not disclosing at that point could cause more damage, though disclosure probably “will not help much.”