Stress tests should be subject to stress tests. Would the banks that passed the version of the test imposed by Europe's financial regulators be able to fund themselves adequately if implicit or explicit state guarantees of their non-deposit liabilities were withdrawn? This further test should be applied to each bank individually, and for the group of systemically important banks as a whole. The outcome would indicate how far Europe's banks have progressed in being able to survive without public subsidy. Not far, I suspect.

The concept of stress tests is derived from the procedures used to ensure the robustness of complex engineering structures. There are three stages. You begin by testing each component in conditions considerably more demanding than it is likely to encounter. Then, you review system design to ensure that, even if several elements break down simultaneously, this does not jeopardise the integrity of the whole structure. Third, and most importantly, you test the total system for outcomes far outside the range of experience. You do not ask, “Will the bridge survive a strong gust of wind?” You ask, “Will it survive a gale worse than any at this site in the last century?”

There is much that the finance sector could learn from this, but no indication it has done so. The adverse scenario of the bank stress tests, far from being outside the range of experience or expectation, is not far from the mean. Sovereign default is not considered, since politicians have decreed it will not happen, although allowance is made for ?the ?possibility ?that ?pesky ?markets might not believe that assertion. Any risk manager in a bank who has not considered far more extreme developments than those in the stress tests should be fired.