Stress tests for banks, performed by mighty regulators, signify dangerous systemic risks, as well as useless predictors

Sir, Caroline Binham reports on how “Andrea Enria, the outgoing head of the European Banking Authority, who is set to become the Eurozone’s top banking regulator, has questioned the value of its stress tests of lenders’ balance sheets, arguing that elements of them are no longer ‘tenable’ and need a redesign” “European regulator questions value of stress tests” November 16.

I could not agree more for two reasons:

First: Stress tests introduce a systemic risk. The fact that banker know their banks will be the object of stress tests causes them to distract their attention from what they might think to be more dangerous, in order to concentrate more on what they think regulators might think more dangerous.

Second: The stress tests are useless since they avoid stress testing many real stresses. In 2003 the United States General Accounting Office (GAO), in its study of the IMF’s capacity to predict crisis concluded, among other things, that of 134 recessions occurring between 1991 and 2001, IMF was able to forecast correctly only 11 percent of them. Moreover, when using their Early Warning Systems Models (EWS), in 80 percent of the cases where a crisis over the next 24 months was predicted by IMF no crisis occurred. Furthermore, in about 9 percent of the cases where no crisis was predicted, there was a crisis.

Much of that has to be a consequence of that if IMF forecasts a crisis; it could quite possibly be blamed for detonating that crisis. Similarly, regulators will avoid to stress test the risks they might be blamed for having produced. For instance when will they stress test the banks on the possibility that their risk weight of 0% to sovereign would have to be increased, and the market reactions to that news. Never! They have painted themselves into a corner.

Sir, when it comes to banks, and their regulations, worry much more about what might be perceived as safe than about what is perceived as risky. In that respect, if I were to perform stress tests on banks, I would look to stress test the risks that seemingly would least need to be stress tested.

@PerKurowski