U.S. and European banking regulators are conducting stress tests to determine whether financial institutions have enough capital to sustain losses as a result of adverse economic conditions.
A critical question is whether these results should be made public, and if so, at what level of detail. The answer is that regulators should be mindful of significant pitfalls.
There is a compelling case for publicly releasing stress-test results. Disclosure would enhance market discipline by allowing outsiders to better price a bank’s risks. This, in turn, would prevent bank insiders from engaging in excessive risk-taking behavior.
Disclosure also would improve financial stability by reassuring outsiders that systemically important banks are adequately capitalized and have the ability to survive a future crisis. And it would bolster the discipline of regulators by exposing them to greater external scrutiny.
These benefits, however, may not be fully realized if stress-test results aren’t properly handled.
Banks have opaque portfolios whose underlying risks are hard to evaluate. Simply disclosing the performance of these portfolios under adverse scenarios without any understanding of the underlying risk exposures of the bank may induce perverse incentives. This is because a bank’s insiders could choose inefficient portfolios to maximize their institution’s chances of passing the stress test rather than maximizing the bank’s long-term value.
To minimize such incentives, my colleague Itay Goldstein, at the University of Pennsylvania’s Wharton School, and I have devised a few recommendations.
For market discipline to work, stress-test results should be accompanied by detailed disclosures of a bank’s exposure by asset class, country and maturity. The market can then do its own math by evaluating the stress tests’ outcomes in light of the bank’s risk exposure.
If such detailed disclosure isn’t feasible, we argue that it might be better not to release the results of individual banks, but only to disclose results aggregated across the banks. Of course, the benefit of market discipline would be lost. However, aggregated disclosures would minimize perverse incentives while still providing the benefit of financial stability.
Perverse incentives could also be mitigated by not revealing to a bank’s insiders the stress-test model used by the regulator. Not knowing how they would be graded would make it harder for insiders to game the test.
Disclosure also may harm market discipline by causing market participants to overreact to bad news. Banks operate in environments that are inherently fragile and prone to contagion in the sense that if some creditors lose confidence and “run,” others might do so, too.
In such environments, disclosure of stress-test results -- especially if the news is bad -- may serve as a signal that coordinates the actions of all the market participants. Suppose a bank receives a poor grade on the stress test and this result is disclosed to the public. This poor grade probably reflects the bank’s poor condition, yet, given that this poor grade is observed by all market participants, it also provides information about how they might react to the bad news.
For example, if a market participant thinks that after seeing the bank’s poor grade other market participants might withdraw their money from the bank, she might withdraw her money. If all market participants think the same way about the other participants, such a “run” becomes self-fulfilling. Note that this overreaction wouldn’t occur if the stress test perfectly reflected the bank’s condition. In that case, market participants wouldn’t care about the behavior of others as they would know exactly how to react to the bank’s poor grade. The potential for overreaction to the disclosure only occurs because stress tests are inherently imperfect measurements of a bank’s condition.
To alleviate such overreaction, regulators should complement the disclosure of any bad news about a bank with a description of the corrective actions to be taken so that the coordinating role of disclosure is minimized and panic isn’t triggered.
Furthermore, individual bank results must be disclosed only when the results are sufficiently precise and reliable. If this isn’t feasible, the regulator should disclose aggregated results across banks, which have the advantage of being less likely to be wrong as idiosyncratic errors are averaged away in estimating bank conditions. Again, if the goal is to promote financial stability, this is a viable solution.
Disclosure can harm market discipline by reducing the incentives of market participants to produce private information. Banking regulators rely on two sources of information in determining the financial condition of a bank: the data gleaned from the market price of a security of the bank, and that collected by the regulator during the supervisory process.
The market price of a bank’s security aggregates the information of market participants who have monetary incentives to trade on their information and opinions. Disclosing stress-test results might adversely reduce the incentives of market participants because they lose some of their informational advantage and therefore produce less information and trade less aggressively. This, in turn, degrades the information role of the stock price and impedes the ability of the regulator to learn from market prices.
Disclosing aggregated information of stress tests will also alleviate this problem. Such disclosure still conveys the benefit of financial stability without drowning out signals about individual banks from the market participants.
Today’s highlights: the View editors on Spain’s banking crisis and turmoil at the Nuclear Regulatory Commission; Michael Kinsley on intervening in Syria; Amity Shlaes on corporate sexual harassment; Luigi Zingales on competition and inequality; Steven Greenhut on California referendums; Matthew Schoenfeld on staving off the next AIG.
To contact the writer of this article: Haresh Sapra at Haresh.Sapra@chicagobooth.edu.
To contact the editor responsible for this article: Max Berley at firstname.lastname@example.org.