By Mike Gullette
Credit goes to the Department of Treasury with their conclusion, documented in the Congressionally mandated study “The Current Expected Credit Loss Accounting Standard and Financial Institution Regulatory Capital” released on Sept. 15. Tasked with assessing CECL’s impact on bank capital, the study concedes that it is too early to make conclusions related to CECL.
ABA’s take? Of course, it’s too early!
Even without a pandemic-induced spontaneous recession, CECL’s impact really can’t be measured unless CECL-based forecasts are studied over an entire economic cycle. Past studies have had significant flaws: They study only residential mortgages, limit the study only to prerecession periods (instead of including the trough and recovery periods), and/or apply “perfect foresight” notions of loss given default estimates (in other words, steep and sudden drops in collateral values are foreseen far in advance). Layer in the pandemic and it’s not surprising Treasury’s study punted on an opinion.
Some believe the pandemic represents a “perfect storm” for CECL. Some say CECL is better under these conditions. Who’s right?
Both FASB and the OCC have recently presented graphs showing how the allowance for credit losses by CECL adopters are supposedly stronger and more responsive to credit risk than those of non-adopters. How should bankers view the CECL implementation thus far?
1. The effectiveness of CECL cannot be credibly assessed after only two quarters. Bankers opposing CECL will cite pro-cyclicality stemming from the unreliability of economic forecasting before a downturn (forecasts are too late), during a downturn (spikes in ACLs are too big), and after a downturn (which are also forecasted late). With this in mind, no one could have reasonably forecast this sudden and deep recession and, as of October 2020, no one knows whether we are in the trough or in recovery. As the Treasury report indicates, it is too early to make such an assessment.
2. CECL ACLs are product-sensitive and most banks concentrate in specific lending segments. Any analysis relying on total allowances throughout the industry, therefore, will be misleading. During the first two quarters of 2020, ACLs for consumer loans increased far more than those for commercial loans (both C&I and CRE). Consumer lending is mostly dominated by larger banks, almost all of which have adopted CECL.
More importantly, however, due to the stability in real estate prices, the increase in ACLs for real estate-secured loans (both residential and CRE) was far less than the increase related to unsecured loans. Large banks generally have only about one-third of their loan portfolios in real estate-secured lending, whereby small banks generally have two-thirds of their portfolios in real estate. This means that smaller banks (most likely not to have implemented CECL) will be expected to have smaller increases in their ACLs. The larger the non-real estate portfolio (in other words, the larger the bank), the larger the expected ACL increase. Judging ACLs based only totals distorts this relationship.
3. CECL ACL ranges throughout the industry are huge. Use of medians and averages therefore, can often distort CECL practice. For example, of the 15 largest banks reporting credit card CECL allowances, range of reserve levels ranged from a high of 1,479 basis points (bps―that’s a loss rate of 14.79 percent) with a low of 403 bps.
Similar ranges occur for virtually all consumer segments. The OCC points out that median 2Q 2020 ACLs for residential real estate were 104 bps for CECL banks versus 74 bps for incurred loss banks. Of the largest 35 banks, however, the CECL range was between 40 bps and 267 bps. The 74 bps level reported by the OCC for incurred loss banks is certainly within the CECL range. Further, two non-bank organizations that report CECL ACLs hold the majority of residential mortgages in the country: Fannie Mae and Freddie Mac. They recorded 37 and 32 bps, respectively. From this perspective, incurred loss banks may appear to be heavily reserved.
Ranges of loss estimates are tighter for commercial loans, yet incurred loss banks seem generally to fit within the CECL range. The OCC median for CECL for commercial real estate loans was 160 bps vs. 120 bps for incurred loss. However, the range of CECL-based CRE ACLs for the largest 35 banks was between 106 bps and 349 bps. So many incurred-loss banks are still within their CECL range. Further, it is helpful to know that the OCC acknowledges that median incurred loss reserves during 1Q 2020 were actually higher than median CECL reserves.
With all this in mind, CECL reserves don’t appear to be appreciably “stronger” than incurred loss reserves.
4. These are unprecedented times. Almost all banks are closely monitoring specific borrower situations and assessing loss content. Given these efforts, credit loss provisioning has likely never been so detailed across the industry – CECL or incurred. The uncertainty of forecasting local economic conditions, however, often requires a coin flip on how quickly and how successful epidemiological efforts will be.
Until we get through this current pandemic period, it is way too early to tell if CECL is resulting in significantly greater or earlier loss recognition. What we do know, however, is that auditing CECL will become more stringent. In a recently issued report, the PCAOB specifically pointed out that deficiencies in 2019 inspections were significant in the auditing of the ACL, especially as they relate to the reasonableness of qualitative factors used in the estimate. As bankers know, qualitative factors are prevalent in today’s ACL (CECL as well as incurred loss).
Going forward, especially considering the new auditing standards that address accounting estimates have been issued by both the PCAOB and AICPA, preparing detailed quantitative analysis of Q-factor adjustments is a reasonable expectation going forward for all banks. This is not only a challenge to those banks that have implemented CECL, but also for those planning and implementing CECL systems for their 2023 effective date.
