Another reason not to believe the Bank of England’s stress tests

This posting is the third in a series on the 2016 Bank of England stress tests. A fuller report, “No Stress III: the Flaws in the Bank of England’s 2016 Stress Tests”, will be published later in the year by the Adam Smith Institute. 

The previous posting is here.

The Bank of England repeatedly reassures us that its stress tests demonstrate the resilience of the UK banking system.

Well, let’s put the stress tests to a stress test.

We have the performance measure, the leverage ratio at the peak of the stress scenario[1] and we have the pass standard. A bank passes the stress test if its leverage ratio at the peak of the stress is at least as high as the pass standard, and it fails the test if the leverage ratio at the peak of the stress falls short of the pass standard.

Let’s consider the five biggest banks: Barclays, HSBC, Lloyds, RBS and Standard Chartered.

In its 2016 stress tests, the BoE used the ratio of Tier 1 capital to leverage exposure as its leverage ratio. The Bank refers to this leverage ratio as the ‘Tier 1 leverage ratio’. The leverage exposure is a measure of the amount at risk and will be of a similar order of magnitude to, and for UK banks will typically be a little smaller than, total assets.

Across the big five, the average Tier 1 leverage ratio at the peak of the stress was 3.95 percent.

The pass standard used in the test was based on Basel III rules and was 3 percent.

By this test, the UK banking system looks to be in reasonable shape and only RBS failed to meet the 3 percent pass standard.

It would, however, be premature to get the champagne out just yet.

On July 8th this year I wrote to Governor Carney about the stress tests and one question I put to him was “How does the Bank justify the 3% Tier 1 minimum required leverage ratio?”

On August 3rd the Bank’s Executive Director for Financial Stability Strategy and Risk, Alex Brazier, wrote back to me with the following answer:

Our minimum leverage requirement for the major UK banks is now 3.25% of assets excluding central bank reserves. … But this is a minimum. On top of that the systemic and countercyclical leverage ratio buffers will, once phased in, add around 0.75% to the average leverage requirement of the largest UK banks.[2] Furthermore, to pass stress tests, firms typically need to hold a buffer of around 1 percentage point on top of this. (My italics)

I am grateful to Mr. Brazier for the clarification, which I interpret as an authoritative statement that the largest UK banks will typically face a minimum required leverage ratio of around 5 percent once the new buffers are phased in.

I am however puzzled why the Bank did not use this higher minimum required leverage ratio as the pass standard in its stress tests. After all, what is the point of the Bank using a 3 percent pass standard in the stress tests whilst simultaneously arguing that the actual minimum required leverage ratio is, or will be, considerably higher than 3 percent? The reason this is a problem is that it opens up the incongruous possibility that a bank might be deemed to pass the stress test whilst simultaneously failing to meet the minimum required leverage ratio.

I am even more puzzled when Mr. Brazier writes that the banks need to meet these higher standards in order to pass the stress tests. Whatever is one to say when the Bank of England official in charge of the stress tests maintains that to pass the stress tests the banks must meet a higher pass standard than the pass standard used in the stress tests?

So the question then arises: how would UK banks have performed in the stress test had the BoE used a minimum required leverage ratio of around 5 percent as its pass standard, instead of the 3 percent pass standard that it did use?

Recall that across the five big banks, the average Tier 1 leverage ratio at the peak of the stress was 3.95 percent. Since 3.95 percent is nowhere near close to 5 percent, then it would appear that, taken as a group, the big five UK banks would have failed the stress test.[3]

The “incongruous possibility” mentioned earlier would appear to be a reality:  taken as a group, the big five banks passed the stress tests even though they did not meet minimum regulatory requirements during the projected stress.

In fact, it would appear that they passed the stress tests even though they did not meet the pass standard required to, er, pass the stress tests. 

End Notes

[1] The Bank’s headline capital ratio, the ratio of CET1 capital to Risk-Weighted Assets, is not considered here because the denominator is deeply flawed to the point of being discredited. See, e.g., K. Dowd, Math Gone Mad: Regulatory Risk Modeling by the Federal Reserve, Cato Policy Analysis No. 754, Cato Institute, Washington D.C., September 2015 or No Stress II: The Flaws in the Bank of England’s Stress Testing Programme, Adam Smith Institute, London, August 3rd 2016. 

[2] At this point. Mr. Brazier inserted a flag to the following footnote: “See the Governor’s letter to Andrew Tyrie of 5 April 2016 for a fuller explanation of the impact of buffers on leverage requirement available here: https://www.parliament.uk/documents/commons-committees/treasury/Correspondence/Mark-Carney-Governor-Bank-of-England-to-Rt-Hon-Andrew-Tyrie-MP-5-04-16.pdf”.

[3] When I replace the leverage exposure measure in the denominator of the leverage ratio with total assets, I estimate that the average leverage ratio across the big 5 banks at the peak of the stress would have been in the region of 3.7 percent, a comfortable fail.