Bank of England to adjust bank stress tests for IFRS 9
14 Feb 2020
The Bank of England is looking to adjust its stress tests for banks, following the introduction of IFRS 9 expected credit loss accounting (ECL), which has led to impairment losses being recognised earlier than previously
14 Feb 2020
The potential changes were flagged up in a speech by Martin Taylor, external member of the Financial Policy Committee (FPC), to Strathclyde Business School, in which he discussed the way the credit losses are accounted for.
He noted that impairments mattered because they were the biggest single cause of bank failure.
Taylor pointed out that the effect of IFRS 9 is to bring forward the impact of credit losses to the point at which trouble first manifests itself. This is a marked difference from the previous approach, which in his words ‘recognised impairments only when they could no longer be denied’.
‘In terms of our stress tests, this means that some losses which would have been spread by a bank over the three years of a downturn will now overwhelmingly occur in the first year: the quantum of overall loss is little affected, but it all hits rather earlier,’ he explained.
Describing the earlier timing and greater depth of the trough as serious issues, Taylor said that if the Financial Policy Committee took no action, some major banks which had not breached their regulatory minimum capital requirements in a stress under the old standard might do so under the new one.
Given the rigorous way that the committee uses the results of stress tests, this would require it to raise capital requirements, perhaps sharply.
However, it wants to avoid unwarranted capital consequences arising from the impact on the stress test of the change in accounting standards, although Taylor said this has proved challenging.
He described the broad shape of the potential solutions as ‘uncomfortably ugly’. Taylor pointed out that ‘permanently lowering the hurdles that the banks had to clear in the stress test would do the job, but would be unlikely to enhance confidence in the system’.
Alternatively, making adjustments to the banks’ accounts for stress test purposes would effectively undo the effects of IFRS 9 and so rewrite accounting standards.
Instead, Taylor highlighted the committee’s December 2019 Financial Stability Report, containing an outline solution to the IFRS 9 problem which he labelled ‘elegant and sound’.
It suggests that the new provisions made by a bank in the first year of stress under IFRS 9, but not yet written off because the loans to which they relate are not yet utterly compromised, should be formally recognised as loss-absorbing (which they are), allowing the Bank of England to adjust the capital requirements set through stress tests accordingly.
Taylor said: ‘We have been in the past reluctant to treat provisions as capital, because they relate to individual assets and are not fungible across a bank’s book.
‘Not fungible [mutually interchangeable], that is, until or unless a bank is in resolution, which is the state of affairs we most worry about: at that point they would be reliably loss-absorbing.’
Under this approach, the Bank would treat the impairments almost as if they formed part of a bank’s capital stack.
There will be a pilot for the 2020 stress test and a consultation with industry over the next year.
He identified two further issues to be addressed. One concerns the point in time at which IFRS 9 provisions are acknowledged.
‘The obvious answer is - as and when they formally appear in the accounts, but this is likely to be a panicky time, and banks and their investors may not be sure that supervisors will hold their nerve at this point and allow a capital offset, whatever they may have promised earlier.
‘The alternative is to proceed by some form of anticipation derived from a stress-test judgement about the quantum of IFRS 9 provisions that a firm may expect to incur, and recognise them early,’ Taylor said.
This would involve a reallocation of capital, perhaps with some capital initially moving from minimum to buffer, with this buffer becoming an IFRS 9 provision in the stress test, and then absorbing losses as bad credits are written off.
Admitting this solution has ‘elements of supervisory ballet’, Taylor said it was ‘good for supervisors to have explicit conversations with firms in peacetime about what may be expected when economic war breaks out.’
The second issue under discussion is from which capital pot this new IFRS 9-prequel buffer should be carved out.
At the moment a bank’s minimum capital requirements are matched one-to-one by debt that converts to equity in resolution, to replace the capital that will then have been lost.
The IFRS 9-prequel could be funded half and half from minima and bail-in debt, or entirely from minimum requirements.
Taylor argued that doing it from minima would be a coherent way forward, insofar as it would treat new IFRS 9 provisions in the stress test as a substitute for the piece of the capital stack that absorbs losses in going concern.
However, this option would require the Bank to alter the one-to-one formula, which has the virtue of simplicity.
‘But the option of taking the adjustment entirely in going-concern minima leaves you more highly geared than the half-and-half alternative at the point of crisis, as you would have relatively more debt compared to equity capital, and more of that equity in buffer form,’ he noted.
Concluding, Taylor said: ‘The decision will not be mine to take, since I shall have left the Financial Policy Committee before it comes back to the table. I have no doubt that my colleagues will weigh the options carefully; this set of choices is unusually finely balanced.’