The parallel run is a critical period for institutions to evaluate and refine their IFRS 9 and CECL implementations. Unfortunately, many will find little time in their implementation schedules to accommodate it, say Laurent Birade and Martim Rocha of SAS
With the impending transition to IFRS 9, and the cutover to CECL in the US starting in 2020, financial institutions worldwide are redesigning their loss allowance estimation processes to meet these new accounting standards. Those in the midst of this are finding it no easy task. Interpretations of the standards, and their lack of prescriptiveness, leave institutions to find their own way – and the impacts of implementation decisions can be massive. A recent survey by the European Banking Authority (EBA) found that banks expect an average increase of 13% in loss provisions under IFRS 9, with 72% of respondents also expecting an increase in income statement volatility under the new standard. Given this backdrop, it is critical that institutions rigorously test their new allowance models and processes prior to going live.
Unfortunately, planned durations of parallel runs for IFRS 9 have been shortening over time. This is a result of implementation schedules that have been slipping for a variety of reasons, including lack of management focus, unforeseen data and modelling challenges, and persistent underestimates of the time and effort required to satisfy the new standard. Unfortunately, the EBA reports that 19% of respondents to its survey will now not be performing such parallel tests at all. Post-transition, these institutions will face a heightened risk of encountering issues that require restatement. Given the importance of the loss allowance on reported financials, this could have significant market ramifications.
But the quality of the parallel run is just as important as its duration. Resources will be challenged during this time because legacy production must be maintained for financial reporting while the new procedures are learned. Here, institutions that have proactively built an efficient, production-ready IFRS 9 and CECL process – with a modular architecture, workflow automation and powerful processing capabilities – will be at a distinct advantage. Rather than struggling with a manually intensive process to produce estimates, they will be able to focus their limited resources on the outcomes. This will give them the opportunity to experiment with alternative models, calibrations and assumptions and spend time understanding, documenting and explaining their consequences.
For example, the IFRS 9 stage-allocation rules can have a significant impact on the magnitude and volatility of allowance estimates – as movement from stage one to stage two requires the allowance to increase from a one-year to a lifetime expected loss. Both IFRS 9 and CECL require loss forecasts to be extended over an exposure’s lifetime, and the method and timing of transitioning to a long-term loss rate can have a material impact on results. Given the potential consequences of these types of analytical decisions on the financial statements, it is highly advantageous to assess multiple alternatives over time, prior to the cutover date. These types of analyses will prove essential in developing the knowledge base within the institution, and ensure that the allowance process instituted at cutover is sound and sustainable.
Meanwhile, institutions with inefficient, piecemeal processes may find that the resource demands of running a parallel test strain their organisations and limit their ability to conduct rich analyses. They may also lack the ability to further refine and automate their processes, leaving them with a suboptimal workflow at the time of cutover to the new standard.
Institutions reporting under US Generally Accepted Accounting Principles can learn from the IFRS 9 experience. Though they have until 2020 to begin reporting under CECL – and this may seem far away today – they should not underestimate the task at hand, and should include a rigorous and extended period of parallel testing as a critical component of their project plans. While this compresses the implementation timeline, testing will help institutions ensure that their estimates are accurate and unbiased, and remain stable over time. It will also provide essential insights into the potential impact to capital relative to the current standard, allowing management to evaluate strategic changes in response. A production-paced parallel run will also allow the institution to assess the efficiency of the workflow and the effectiveness of its controls.
The parallel run provides a critical window of time in which to refine processes, further automation and experiment with design options away from the bright lights of investor scrutiny. Institutions should not squander this opportunity.