Why the Bank of England should end its stress tests: ‘No Stress’ in a nutshell

In December last year the Bank of England reported the results of its first set of annual stress tests of the capital adequacy of the UK banking system. Its message was reassuring: our banks are safe.

Don’t believe a word of it. The banking system is actually highly vulnerable and the stress tests merely hide that fact. Indeed, they even make the banking system less safe than it already is, by pressuring banks to take hidden risks that the risk models cannot see. 

The stress tests are flawed in three distinct ways. 

The first is methodological. The tests are based on one scenario and cannot possibly give us confidence the system is safe against all the other possible scenarios they did not consider. They are based on a ‘risk-weighted’ asset metric that is unreliable because it is dependent on gameable risk models that under-estimate banks’ risks – and the system incentivizes banks to game the models to get lower capital requirements and thence higher distributions of false profits. They create systemic instability by pressuring banks to use the same models that are blind to the same risks. They lack credibility because even if the central bank thinks there are major banking problems, it cannot publicly admit to them – to do so would undermine confidence and lead to questions about its own past competence in rebuilding the banking system. The results then have all the credibility of a rigged election. 

Even if we ignore these problems, the stress tests are fatally flawed because they use a very low ‘pass’ standard, a 4.5 percent minimum ratio of capital to risk-weighted assets. This minimum is well below those coming through under Basel III. Had the Bank carried out a test using these latter minima, the banking system would have failed the test: same exercise, higher safety standard, opposite result. 

The Bank also failed to carry out any tests based on a minimum ratio of capital to leverage – a test that would have been more reliable because it is much less dependent on unreliable risk models. Even the most undemanding such test – one based on a minimum leverage ratio of only 3% - would have revealed that the banking system was very weak. The Bank’s failure to apply this latter test is all the more puzzling because the Bank expects UK banks to meet this minimum standard. Thus, the Bank asks us to believe the banking system is sound, despite the fact that a 3 percent leverage ratio test based on what it currently expects from banks and based on its own stress scenario would have suggested otherwise. One might add that many experts recommend a minimum leverage ratio of 15 percent, five times larger than the leverage test that the Bank did not conduct, or at least report. You can imagine the results. 

One can only speculate why the Bank did not report the results of these alternative tests, but one thing is sure: the comforting headlines last December would have been rather different had they done so. 

Overseas experience also indicates that stress tests are useless as indicators of bank vulnerability and can go catastrophically awry. Recent stress tests failed to notice the impending collapse, not just of one but of three national banking systems: Iceland in 2008, Ireland in 2010 and Cyprus in 2013, all of which collapsed shortly after being signed off as safe by regulatory stress tests. The European stress tests missed the latter two collapses, but also have a history of being captured by powerful banks and the Euro elite – and of using very unstressful stress scenarios to produce loss projections that turned out to be dramatically short of subsequently realized losses. 

The most recent European stress tests would appear to be no exception: the ECB party line was that all was well in the European core despite problems in the fringe countries, but these were dismissed on publication by a variety of experts who pointed out that it was the big French and German banks that were most vulnerable. However, the European stress tests overlooked their vulnerability because they used ‘risk-weighted’ asset measures that were blind to their main risks instead of leverage measures that would have revealed them – the same mistakes that were made by the Bank of England. 

Stress tests operate like a radar that cannot see any hazards. We wouldn’t dream of sending out a ship or plane reliant on a radar that didn’t work. We really shouldn’t do that with our banking system either: the Bank of England should abort its stress testing program forthwith. 

Kevin Dowd is the author of the Adam Smith Institute's most recent report 'No Stress', which is available here.