The Bank of England uses its stress tests to reassure the public that the UK banking system is safe. However, the Bank’s reassurances lack credibility and are contradicted by the evidence.
Rightly interpreted, the stress tests demonstrate the opposite of what the Bank claims they do: they demonstrate that UK banks are still financially weak and far from resilient. The UK banking system is an accident waiting to happen.
The conclusion that UK banks are weak is confirmed by an analysis of their capital positions and is further confirmed by banks’ market values being less than their book values. Low market values indicate problems with the banks that the stress tests did not pick up.
The Bank made a number of mistakes in its stress tests. Among these: it relied on book values instead of market values, relied on unreliable metrics such as risk-weighted assets and Tier 1 capital, relief on a single stress scenario and used insufficiently demanding pass standards. The Bank’s stress model also produced implausibly low projected losses and so failed a basic reality check.
More generally, the stress tests are based on a series of imprudent judgments that led the Bank to miss obvious problems with UK banks.
Regulatory stress testing is a highly imperfect tool with a track record of repeated failure in other countries, is compromised by conflicting objectives and by the Bank’s poor forecasting record. It is also compromised by basic Public Choice economics, i.e., that public agencies act in accordance with their own interest.
It also creates invisible systemic risks by pressuring banks to standardise their risk management practices to conform to the Bank’s view of the risks they face.
Far from providing a credible assurance that the banking system is safe, the stress tests are worse than useless because they provide false comfort, suggesting that the UK banking system is safe when it is clearly not. In this sense, the stress tests are like a ship’s radar system that cannot detect an iceberg in plain view.
The stress test programme is therefore dangerous and should be scrapped.

Read the full paper here.