How to manage the impact of the PRA’s new stress testing principles
How AI solves the three biggest collections challenges
Overcoming the costs and challenges of IRB implementation
17 October 2018
Traditionally, for a firm implementing IRB, there have been many challenges and heavy costs attached. However, recent regulation changes and refinements have caused a shift in the cost-benefit ratio in favour of IRB for many firms. The changes mean that many of the costs and challenges previously associated with an IRB development can now be overcome more easily, especially for small and medium sized firms. Ben Clark explores the costs and challenges of IRB implementation in this blog.
1. The data problem
Implementation of IRB models requires a long history of data and, for some models, the data must capture the effects of a full economic cycle. For example, in the case of UK mortgages, this would be defined as including the early 1990’s recession.
In other cases, modellers must evidence that the models represent a reliable worst-case component. This suggests that a data sample which includes 20 years is required to build IRB models; most challenger firms simply won’t have this long history of data.
Given the low interest rate environment experienced over the last 10 years and the tightening of risk appetite, many firms have little to no defaults to build models to assess the level of risk in their book today. As these models are required as a starting point for the IRB calculation, this has previously been a major stumbling block for these firms.
The solution to this is to utilise external data sources, which can come in a range of forms, and is now more acceptable thanks to recent changes to the regulations. The challenge of sourcing the right data and evidencing its relevance remains, however, the recent changes have opened up the possibility of IRB to firms who had previously discounted it due to insufficient data.
2. The model development requirement
Successful IRB model development programmes require substantial knowledge, expertise and experience. This requires many firms to have a team who can develop and oversee the modelling as well as the resource to challenge and review the models in the wider management teams in order to utilise the model efficiently.
The cost of developing an internal modelling function can be large as modellers are a skilled resource and are in high demand due to a need for these niche skills in the market. This has historically been a major problem for smaller firms.
There is no quick fix for this as accurate modelling requires years of experience, trial and error and training in order to develop the right approach to managing a model focused institution. Management teams also need the expertise to know when to trust the models, when to add judgement and when to redevelop.
However, unintentionally, most firms have been building this expertise through the development and usage of IFRS 9 models. For all but the smallest firms, a form of PD model will already be in use within the institution and key components for IRB modelling will already be in place.
3. The cost is high, and the benefits are uncertain
When the capital benefits are uncertain, it can be difficult to justify and commit to an IRB project. Financial benefits are likely to be significant, but the size will be dependent on several factors, including the availability to source relevant data and the quality of the models at the end of the development.
Within most firms, costs are tightly controlled, and large projects are only commissioned when the benefits are certain. This can make achieving internal buy-in for an IRB development challenging, especially when it will involve substantial upfront costs such as the purchase of external data or an increase in modelling resources.
The solution to this is two-fold:
o Where the benefits are uncertain, the move into IRB can be gradual. Initial impact assessments can be carried out through IRB simulations using very basic models and internal data. It may not be possible to accurately estimate the capital requirement at this stage, but a range can be given with confidence intervals to support the cost benefit analysis
o Synergies from the IRB development can be maximised to optimise work to ensure the benefits are recognised as quickly as possible. For example, the initial modelling work can be used to support the firm’s Pillar capital assessment providing an alternative calculation to the standardised approach
4. The elapsed time
Designing, developing, validating, implementing and internally approving an IRB model is likely to take over a year, with a requirement for firms to evidence their ability to use an IRB approach effectively over at least a 3-year period. This means that the elapsed time from the commencement of an IRB programme to approval of the waiver can be several years.
Recent changes mean this approval process has been dramatically shortened which means that a successful IRB waiver can be achieved in under three years of a development starting.
For more information on overcoming the costs and challenges of IRB implementation, download our whitepaper ‘IRB: Six reasons to transition amidst a world of changing regulation.’