Provisioning in fog

The setting of provision levels (loss allowance) is fraught at this year-end. Attention will be placed on overlays and the extent to which judgement is used. The risk is provision will be released because those amounts sit outside the model and are largely reliant on expert views and their best judgement of what may unfold.

It is clear the setting of provision is hampered by high levels of model risk driven by the uncertainty caused by the pandemic and various interventions in managing the impacts. However, there is an uneasy conflict between reporting based on objective model driven outcomes, which to-date show limited evidence of credit deterioration, and the degree to which models are adjusted to provide adequately to cover the economic risks. At a high level there is not any obvious reason to believe that the economic losses caused by the pandemic have disappeared. In many cases leverage of businesses affected has increased limiting capacity for future investment. While a vaccine is welcome it will not entirely immunise the banking industry from losses that are likely to be already latent within existing portfolios.

Those reviewing and auditing financial accounts are right to push for a robust justification and basis if overlays are made, particularly where this is largely based on expert judgement. But if Banks follow a position based purely on what is known, what is contained in the data and can be objectively established at this time, it is quite likely they will report sizeable releases of provisions, and under report on the impact of key risks yet to be seen. There continues to be a need to reflect a high degree of uncertainty into expectations.

Last week, US Bank chiefs pointed to the fact that setting provisions where the outlook remains so uncertain may be volatile. It will be important during this reporting period to ensure model risks continue to be identified and addressed, which if done robustly should report only minor reductions in allowances.

This does not mean that provisions should not go down at this year-end. There are quite objective reasons for seeing some reduction. It does mean though, that provision is still required to cover an absence of outcome where the risk remains. That may come in the form of an overlay to specific model parameters, or to specific segments of lending where outcomes are particularly unclear or to the weights placed on forward-looking views.

Decision makers and those tasked with reviewing the adequacy of provisions may do well to remember IFRS 9 credit risk models remain relatively new and immature and so have limitations, one being that they are yet to have stood the test of time. Another is they are unlikely to fully reflect the extent of outcomes that will result from this set of events.

A key reference that can help Banks with how they can approach the use of expert judgement over the year-end reporting period is provided by the EBA guidance on credit risk management and accounting for ECL (GL/2017/06). In it they suggest judgement of the impact of forward-looking information is to be incorporated into the ECL.

There are limitations in modelling approaches that can only be dealt with by using expert judgement, and extreme events serve as a reminder that such risks need careful management.

When there exists only a partial view of the outcome of a severe risk event that has wide ranging impacts it would be unsafe to release provision that is set to address such risks if this is because they are not yet reflected in data and are based on expert views.

A final thought, reflecting on the guidance offered by our Regulators early on in March last year. Coming into the crisis they urged caution, to focus on the long-term consideration of events and risks. That logic remains just as relevant now as it did at that time.

ANY QUESTIONS? Please contact Phil at phillip.dransfield@4-most.co.uk or Chris at christopher.warhurst@4-most.co.uk.