A Closer Look at CECL: What Can We Learn from the European Implementation of IFRS 9?

By | January 29, 2020

Courtesy of Arndt-Gerrit Kund and Daniel Rugilo

Introduction

The sub-prime crisis of 2008 drew attention to the fundamental drawbacks of the incurred loss accounting model, which determined the loan loss provisions for troubled exposures. The subsequent financial crisis revealed in particular that the recognition of impairments was not only too late, but also incomplete. As illustrated in Figure 1 below, this trend was not only observable for U.S. banks, as shown in the left panel with FDIC data, but also on a global scale as shown in the Bankscope subset on the right. In response to this universal problem, the IASB and the FASB urged a comprehensive revision of their accounting standards for financial instruments, which culminated in the release of IFRS 9 and ASC 326-20, respectively.

Figure 1: Annual Impairments for different subsets of banks.

The new accounting standard constitutes a paradigm shift in the calculation of impairments for financial institutions. Instead of recognizing impairments only in the aftermath of a loss event, the new expected credit loss model mandates the buildup of loan loss provisions ex ante based on a bank’s internal estimates. A key element of the new IFRS 9 impairment model is the so-called three stages approach; based on default probability (PD), amongst other factors, a loan is assigned to one of three stages. The expected credit loss is then calculated as the product of the PD, the outstanding balance (EAD), and the percental loss on this balance in case of default (LGD).

Stage 1 contains the most resilient assets, such that the bank only has to recognize the expected credit losses that occur within the following 12 months. Hence, PD and LGD estimates are only on a 12-month horizon. Should the credit quality further deteriorate, the exposure is assigned to Stages 2 and 3, respectively, where lifetime expected credit losses have to be recognized, such that PD and LGD are estimated over the whole lifetime of the loan. This granular staging model constitutes one of the major differences between the lessons learned in Europe and the U.S. Under the FASB Current Expected Credit Losses (CECL) model, U.S. banks have to recognize the lifetime expected credit losses immediately, irrespective of asset quality, which makes it more expensive. This divergence in implementation opens up an interesting line of thought concerning the implications of the new models on either side of the Atlantic. We intend to shed a little more light on CECL in this blog post by assessing the lessons learned from IFRS 9 in Europe and relating them to the U.S. case.

Empirical Results

In our recent paper, we studied the effect of IFRS 9 in Europe. In doing so, we faced the central challenge of data scarcity, as the implementation of IFRS 9 has only been mandatory since 2018. However, we were able to overcome this challenge by reverting to bank stress tests from the European Central Bank and derived a meaningful panel of banks to study. Our study covers all publicly available European stress test results from 2014 onward, which relates to 43 banks from 15 different European countries, representing approximately 70% of all exposures in the Eurozone. Furthermore, the panel contains both the old and the new accounting standard, together with two calculations for a baseline and an adverse scenario, such that meaningful inferences can be derived from the data.

In our study, we show that the shift from the old incurred loss to the new expected loss model has released two opposing forces in terms of bank stability. First, the “cliff-effect”, which refers to sudden increases in impairments. It occurred under the old accounting standard, as credit losses were only recognized once they were identified. As a result, impairments came late and abruptly. Consequently, the “cliff-effect” represented a major source of procyclicality, as the occurrence of impairments necessitated a second wave of impairment, which culminated in a reinforcing feedback mechanism. This relationship is especially problematic during financial crises and has been addressed through the new expected credit loss model for both CECL and IFRS 9.

However, one drawback emerges from the direct recognition of expected credit losses from an instruments’ inception. The “front-loading” effect leads to an earlier recognition of losses, which may adversely impact bank resilience by lowering their capital levels, thereby making them more failure prone. Thus, while the timelier recognition of credit losses under CECL and IFRS 9 fosters financial stability by mitigating procyclical effects, it also weakens capital adequacy, potentially setting off this benefit. The effect is probably even more pronounced for CECL because it continuously requires the lifetime expected credit losses to be recognized, whereas Stage 1 exposures under IFRS 9 only apply a forward-looking 12-month window.

In our study, we proceed to analyze this inherent trade-off in terms of financial stability. We show that the new IFRS 9 accounting standard increases the likelihood of bank failure in normal times due to the “front-loading” effect. At the same time though, it makes banks more resilient when the economy is in shambles, as the funding gap between the baseline and adverse scenario narrows significantly. Taken together, one can conclude that financial stability during an economic downturn has been increased, at the cost of making a normal economic growth scenario more expensive. Figure 2 below illustrates these deliberations by showing the huge spike in impairments in 2018 that is due to the described “front-loading” effect. At the same time, one can see that only unsubstantial additional impairments are required thenceforth, due to the transfer from Stage 1 to Stage 2 or 3, and the subsequent recognition of lifetime expected credit losses.

Figure 2: Average impairments over all banks in the respective European stress test scenarios.

What can we learn from that?

Against the background of these findings the question arises: What can be learned from the European case to aid a smooth transition to the CECL framework in the U.S? A central challenge concerns the optimal implementation policy. As shown before, the “front-loading” effect leads to an earlier recognition of losses in the first place. In conjunction with an economic downturn, this might create the conditions for a perfect storm, where credit losses are inherently more severe, and also occur earlier. However, the case of the EU has shown that bank resilience grows during economic downturns, as banks have already realized the credit losses that were otherwise to come. Taken together, the initial costs may be higher, but constitute a net benefit beyond the transition period. In this light, the critique of a level playing field can be attenuated. Figure 3 below illustrates the differences between the two accounting regimes.

Figure 3: Impairments under IFRS 9 and CECL.

The horizontal dashed line shows the cumulative impairments under the U.S. CECL model for a high quality 30-year mortgage. You can see that the realized impairments are initially higher, compared to the European IFRS 9 implementation, which is plotted as the full line. This difference, as highlighted by the shaded area, manifests due to the staging approach of the IFRS model. While the loan is initially performing, it is assigned to Stage 1, where only the expected losses of the next 12 months are realized. The CECL framework though already incorporates the impairments over the full lifetime of the loan, or said differently, the next 360 months. It is only when the credit quality deteriorates and the loan migrates to Stage 2 or higher under the IFRS framework that this gap closes. Thus, while it is true that European banks may be better off initially, one can see that the aggregate losses realized on the loan are, ceteris paribus, the same in Europe and the U.S. The only difference lies in their distribution over time. Given the record profitability of U.S. banks relative to their European peers, this observation seems like a reasonable burden in the quest for a more resilient financial system.

 

Leave a Reply

Your email address will not be published. Required fields are marked *