The financial services industry has recently undergone a major change due to the introduction of IFRS 9 impairment requirements. This has come generally at increased costs due to either the redirection of internal resource or engagement of third parties to develop compliant models. In practice we mainly see that these models are account level lifetime loss predictions with the ability to mechanically adjust to use probability weighted economic scenarios. So far, in our experience, this is all been done with one sole aim – to ensure the impairment allowance contained in the company accounts is compliant with the IASB standard and therefore satisfy auditors.
However, let’s look again at what is being built:
“…account level lifetime loss predictions with the ability to mechanically adjust to use probability weighted economic scenarios.”
In our experience this kind of predictive modelling is a significant enhancement on anything most organisations currently possess. Therefore, why not use it to drive better credit decision and increase bottom line profit and returns through a range of uses including:
optimising the use of liquidity/funding
optimising the use of capital
identify growth and withdraw segments
identify concentration risk
set risk appetite
enhancing forward looking forecasts and scenario testing.
The first two uses are achieved through enhanced credit approval processes.
Credit decisioning through scoring has long been the standard within the industry. However, how these scores are used varies across organisations. We generally see three broad categories of increasing sophistication:
Cut-offs set on a perceived level of acceptable risk e.g. we want default rates less than x%
Cut-offs set on simple break even modelling based on average loss and income expectations
Cut-offs set using fully modelled expectations of returns by targeted segments (against a range of metrics e.g. RWA, EC, investment etc…).
The last category requires a significant level of additional modelling to allow granular expectations of losses, income (net of funding costs), costs and capital consumption.
The most complex of these requirements are losses and capital consumption (for IRB portfolios). However, the models developed for IFRS 9 impairment can largely satisfy these requirements with minor adjustments. This just leaves the income and costs – generally quite simply modelled through simple assumptions and deterministic outcomes, and the model to tie all these pieces together.
Once this model has been produced, the returns and profit expectations of new originations and existing loans under varying economic scenarios can be modelled. Analysis of these estimates by key segments and/or critical scenarios can then be used to identify key segments, concentration risk and help set risk appetite.
At 4most we are already engaging with organisations to develop these models and our clients are seeing benefits with both their financial performance and modelling sophistication, opening doors to further sophistication in a range of areas including marketing, customer retention and portfolio acquisitions. Additionally, for any organisation with IRB aspirations, these models should form the cornerstone for use test requirements and the requirement to embed IRB models and metrics in all areas of the organisation.
For further information please contact Chris Warhurst, Technical Director at email@example.com.