Ben O’Brien, Risk Practice Director, Jaywing
The direction that IFRS 9, the much awaited legislation for financial assets, took was born out of the recent financial crisis. Existing legislation had allowed lenders to postpone recognition of credit losses and this may have contributed to the slow fiscal recovery.
IFRS 9 provides a unifying approach to classification and measurement of financial assets and changes to hedge accounting rules but the most significant change for UK lenders surrounds the treatment of loan impairments. IFRS9 will replace the IAS 39 model with a forward-looking view of expected loss so that losses are recognised much sooner. The legislation itself presents lending organisations with the challenge of calculating a lifetime view of their customers’ future losses, before they have been incurred, and comes in to effect on 1st January 2018.
There are many factors to consider when implementing the new IFRS 9 requirements. However, despite perceived complexities to navigate, the process can be broken down into nine manageable steps:
- Interpret standards and guidance
The first tip is to build a collective interpretation of the IFRS 9 standards across Finance and Risk departments. Guidance is offered by Guidance on Accounting for Expected Credit Losses (GAECL) that provides useful insight into the deeper implications of the regulation. However, interpretation work must be underpinned by strong ties between the risk and finance worlds, given the wide reach of the new regulations.
After the standards have been interpreted, the next step is to establish a set of requirements for your organisation. Many of these requirements can be addressed by harnessing tools already available within financial institutions, although there is likely to be gaps that need filling before a fully compliant solution can be produced. By measuring the requirements against existing capabilities, gaps can be identified against the requirements the standards will have on your business. Knowing these requirements will also provide a foundation for the remainder of the project.
- Asses current state, historical data and existing risk models
The many references to using existing credit risk tools means it is important to conduct a thorough examination of your existing data, system and model infrastructure for suitability within a forward-looking expected loss model.
There are two fundamental steps in conforming to the standards. One of these is understanding the implications of historical data and reporting system, while the other is comprehending how the existing risk models suite can be leveraged to solve IFRS 9.
Shortfalls in the data infrastructure can be pinpointed through early assessment.This allows time to engage the business to organise and implement changes.
Quality of data is one of the most important factors in the quality of a model. Given that stage allocation requires measuring change in credit risk from origination, data gaps or inconsistencies may exist for some assets. Questions are also raised on how far back historically, and with what level of assurance, origination data spans? This is where an audit is necessary. It should uncover issues that result in a wider project to enhance data infrastructures as part of a longer-term vision for financial organisations.
Not everything needs to be done from scratch and some organisations will have existing IRB models built on a one year expected loss outcome. Not having this means leveraging existing risk models or building new models to forecast loss over different outcome periods. Therefore, ensuring definitions or assumptions made during the model development are compatible with IFRS 9 is vital.
Long-term forecasts of expected loss are what IFRS 9 is looking to promote – aligning this process by using stress testing and loss forecasting outputs to drive IFRS 9 lifetime expected loss is key to success.
Businesses should also be aware of credit loss forecasting models that have methodological differences from other credit risk models. Given that a forward-looking five-year view is similar to lifetime expected, then forecasting, stress testing and IFRS 9 processes should be aligned, underpinned by the same models.
- Develop your methodology
A natural next step from identifying the gaps between requirements and capabilities is to identify ways of addressing these gaps. There may be a single methodology or, more likely, an organisation will need multiple methodologies across different asset classes or portfolio segments. What is important is finding the right methodology to suit your organisation. Existing methodologies may not solve the problem, so sometimes alternative approaches need to be explored and the rationale behind methodological choices explained.
- Work up a prototype model and assess initial impact
Once methodologies have been identified, then modelling can begin. Without a concrete methodology then a full-blown modelling project may not be the most efficient course of action.
This is where prototype models come in as they enable the pros and cons of multiple methodologies to be identified in a manageable time frame. In many model developments, particularly with new methodologies, there will be unexpected challenges. Therefore, prototyping helps highlight any issues and facilitates remediation early on. This approach also helps approximate the framework required for an end-to-end solution. This is vital information for planning for implementation and testing.
Another advantage of building prototype models at the beginning of the process is that they provide early estimates of the provision figures under the new regime. It is expected that provision cover will generally increase under IFRS 9 but getting an early view of the magnitude will help manage expectations with senior management ahead of full implementation in 2018.
- Refine your methodology
Once the methodology and prototype are in place, reviewing figures from stage four may uncover issues with provision adequacy or stability. These issues can be addressed by revising the modelling approach, so it’s important to have the opportunity to refine the model methodology. If multiple methods are being researched, a quantitative and qualitative assessment will facilitate the selection of the optimal method for the business. These actions provide a good foundation before moving into the critical final stages of the wider project.
- Pay attention to your final modelling
Modelling should adhere to already established standards to ensure they are robust, predictive and relevant. The data inputs must prove to be representative of the portfolio prior to commencing any development.
It is widely touted that adherence to IFRS 9 will result in a suite of models across multiple portfolios. In my view, planning and preparation is key to avoiding over-engineering the solution and leaving a model legacy that is difficult to maintain. It is clear that modelling is a core component of the project. Businesses should keep records of the model development processes that have been followed as this is critical to developing a fully compliant solution that concludes with audit and Prudential Regulation Authority (PRA) approval.
To ensure acceptable performance during validation, organisations should include the level of acceptable discriminatory power, stress testing and backtesting criteria and thresholds and any other relevant validation standards.
- Model Governance
Fundamental to a successful process is a final model that has undergone appropriate independent validation and challenge. Comprehensive model documentation is critical to demonstrating compliance.
The documentation should include information around the methodologies employed and judgements made throughout the process. It should also describe the model development journey in the agreed model documentation format. The regulators will view the model documentation as evidence of compliance to the IFRS 9 Standards, so investing time in producing high quality documentation is paramount.
Also essential is independent validation of the modelling approach. This will be obtained during the course of the development and comply with agreed model governance standards.
- Implement and test
Development and approval of the models is not the end of the journey. Models must fit within a system and run alongside IAS 39 models for at least one year. This is long but proves that the implementation is correct and also allows financial institutions to demonstrate adequacy and stability in the provision estimates.
At this stage, the structure of the financial reports should be agreed and the means to populate these reports automatically from the model estimates developed.
- Monitor your model
Demonstrating that the models are predictive and stable at development stage is only the first step in validation of the lifespan of a model. The models will also need regular monitoring.
This is challenging as lifetime expected loss models have complexities that need to be considered early in their development. A framework is needed that ensures models continue to meet both IFRS 9 and internal standards in the longer term. By following the above steps and putting the time in to develop the right processes from the start, the necessary requirements can be met effectively.