Bank stress tests should not be open to debate

Informed debate is a crucial part of public policy development. But the behind the scenes tug-of-war between banks and the US government over the results of their recent stress tests strains the exercise’s already tenuous credibility. It also shows that banks have become too powerful.

How so? First, banks and their overseers run stress tests all the time, on individual products, divisions and the institutions as a whole. Without them, it would be very difficult to manage risk or allocate capital among business lines. The current crisis proved these tests were inadequate – or in some cases, ignored. But that’s largely because of management incentives to take outsized risks, and the fact that the scenarios used in the tests were not sufficiently grim.

So it’s curious that regulators have put so much stock in the tests they unveiled in February. The release of their results has been delayed until next week, while banks ask for clemency. Since the results will determine which institutions will be forced to raise private capital or take further government infusions, the stakes are high.

BUT like the banks’ earlier and insufficient stress tests, the government’s worst-case scenarios aren’t all that far-fetched. They also use banks’ own estimates – meaning unscrupulous managers could tweak them to get a better grade. And bankers say they’ll produce very little information that regulators don’t already have.

Because of this, bank risk managers (admittedly, not the most credible group these days) tend to view these tests as a public relations stunt that regulators will use to force their institutions to toe Uncle Sam’s line. That, in itself, is worrying. Regulators shouldn’t have to invent justifications for regulating properly. The right response by a bank when its overseer says jump is “how high?” That regulators are wrangling with banks over the results of these tests shows that they are not confident in their ability to understand the institutions. That gives banks too much power. It would be better for watchdogs to demand that they reduce their complexity to comprehensible levels. Otherwise they’ll retain the upper hand – and no amount of testing will be sufficient to diagnose their problems.