Note: We are providing a more comprehensive Executive Summary with this article, as the main body is a bit more detailed and accounting technical; we expect the Executive Summary provides a comprehensive overview for most readers, and the main body goes into detail for those with a more technical interest.
Additionally, this article provides an overview of IFRS 9 and its potential impacts. We intend to follow this up with an article in the second quarter focussing on operational and implementation challenges.
Canadian Banks and credit unions are in the midst of implementing one of the most challenging accounting changes in recent memory, when they move from an Incurred Loss (IAS 39) to an Expected Credit Loss (“ECL”) (IFRS 9) accounting standard.
The Canadian banks will adopt the new standard on November 1st of this year (the start of the 2018 fiscal year), while the European banks will adopt in 2018; the US Financial Accounting Standards Board (FASB) published its final standard on Current Expected Credit Losses (CECL) in June 2016. The FASB’s new standard will take effect on January 1, 2020 for certain public banks, and in 2021 for all other banks, with early adoption permitted in 2019. These changes comes into effect after years of global review and debate, with the goal of having a more appropriate level of credit losses recognized earlier in the credit cycle (following concerns raised during the financial crisis that provisions were “too little, too late.”)
As a result, loan loss provisions (i.e. the income statement charge) and its volatility could increase significantly, as they will respond much more directly to changes in credit risk trends (as indicated by a combination of credit models, loan ratings/other characteristics, economic trends and stress testing analysis – and thereby introducing significant judgement into the process). A Deloitte survey of global banks indicates that their Allowance for Credit Losses (i.e. the balance sheet reserve – which is built up via loan loss charges through the income statement, and drawn down by write-offs) could increase by up to 50% for some banks. Additionally, a recent survey carried out by the European Central Bank reported that loan loss provisions (i.e. the charge that goes through the income statement) could increase in a range from 18%-30%. And KPMG has reported that “credit losses are expected to increase and become more volatile under the new expected credit loss model. The number and complexity of judgements is also expected to increase.” The following graphic, provided by KPMG, outlines the significant change in provision for credit loss charges under the two methodologies:
Loss allowance under IFRS 9 and IAS 39
While there are significant details and changes in these new standards (see full article, below), the most fundamental change is that banks will now be required to take a provision on all loans in their banking book, as opposed to the current standard which requires objective evidence that the loan has become impaired before a reserve is established. So as new loans are originated an expected credit loss is calculated, based on a 12 month time horizon (i.e. considering the probability that the loan will default within one year – often a pretty low number). But then the banks will have to do a second analysis, as at each reporting period; they will have to assess each loan for a “significant increase in credit risk”, and when that occurs they will have to increase the provision to a Lifetime expected credit loss, where the probability of default time horizon extends from 12 months to the full term of the loan (and therefore more similar to a provision taken today against a defaulted credit). So one can see how loan loss provisions will increase and become more volatile under the new standard.
The changes will also significantly increase the complexity of the Bank’s credit risk systems and models. Interestingly, just as regulators are considering a fundamental overhaul and limitation on the use of credit risk models and parameters for capital management purposes, such models will become a key component in the new loan loss accounting standard. The expected loss model requires banks to build a new set of credit models (for the larger banks we observe mostly as extensions of their capital and stress-testing models; for smaller banks and credit unions who are not on the advanced Basel capital approach, this is a bigger challenge) and exercise significant judgement to determine loan losses at each reporting period. The IFRS 9 implementation also introduces operational risks, as complex models need to be built, vetted and maintained, and then coupled with significant estimations and judgement, in order to calculate the new allowance and loan loss numbers. In fact, although the ECL models may be built as extensions of the Basel capital models at some banks, the standards and requirements are different from the capital standards and therefore another set of books needs to be maintained.
IFRS will also likely have significant impacts on the banks capital levels. As higher and more volatile provisions are recognized through the income statement, they will flow through to and reduce retained earnings and thereby the capital ratios. There is a significant pro-cyclicality involved here, as a downturn in the economic/credit cycle will cause retained earnings to fall just as risk weighted assets are increasing (i.e. the capital rules also require risk weighted assets models to increase with a turn in the credit cycle, albeit more slowly); banks will therefore be pressured to curtail lending to protect their capital ratios, leading to a self-fulfilling cycle. Also, there is an asymmetric treatment of credit allowance reserves in the capital standards, where under provisioning (i.e. allowance reserve less capital model expected losses) leads to a haircut for the crucial Tier 1 Common Ratio, whereas an over provisioning (which could well occur under IFRS 9) does not provide a similar benefit (the benefit is recognized in the Total Capital Ratio, which is less of a binding constraint.)
So, banks and their shareholders have some significant challenges and operational questions ahead of them. Who will own the models and processes? Who will reconcile the various sets of books, and manage the asymmetry between capital, stress test and accounting treatment? Who will maintain the talent and systems to operate these complex systems going forward? How will the significant estimates and judgements be governed? And finally, who will educate the user community regarding the significant change in financial disclosure?
As this is such a significant change for both the banks and their shareholders, we plan to continue our research in this area, considering next the operational and implementation issues.