What a year! Well, it has certainly been interesting to say the least. While many institutions and stakeholders spent the majority of 2019 refining modeling techniques, reserve processes, and polishing off parallel runs, not many spent time planning for what 2020 had to offer. Let us list them out as if we need a reminder; worldwide pandemic, economic shutdown, natural disasters, unprecedented government stimulus with more to come (maybe), deferrals and modifications that are not Troubled debt restructurings (“TDRs”), another CECL delay (well, sort of), a potentially contested election (probably should have seen that one coming), and of course Netflix’s release of Tiger King. We really wish we could forget that last one, but Carol showed up again on Dancing with the Stars! All kidding aside, in an already challenging environment, bankers were left with even more questions to answer.
Going into the initial adoption of Financial Accounting Standards Board (FASB) Accounting Standard Codification (ASC) 326 Financial Instruments – Credit Losses, commonly referred to as “CECL”, there were significant concerns surrounding comparability across the industry and the uncertainty surrounding what this change would bring. While most thought there would be increases in the Allowance for Credit Losses (“ACL”), the events that unfolded brought the ACL well above prior expectations. This paper will examine the impact of both CECL adoption as well as the reporting periods following. Rather than relying on Federal Deposit Insurance Corporation (“FDIC”) Call Report data which can be misleading in detailed analysis, we have scoured through each relative U.S. Securities and Exchange Commission (“SEC”) Form 10-K and 10-Q to identify the most accurate values and management commentary. In the analysis, consideration will be given to the ACL excluding reserves on unfunded commitments, debt and equity investments, and deferred tax implications. As such, changes will not correspond directly to the net retained earnings adjustment required at CECL adoption.
As a note, this analysis is focused on the FDIC regulated banking sector and does not consider any credit unions (some adopted early), insurance companies, real estate investment trusts, “shadow lenders”, or other companies who may be required to record an allowance for credit losses or reserve for bad debts in accordance with ASC 326. The initial population of banks considered includes regulated financial institutions traded on a Major Public Exchange (NYSE, NASDAQ, OTC), with total assets as of 12/31/2019 of less than $400 Billion. This resulted in a sample of 338 institutions.
Of the Banks analyzed, the CECL standard became effective for 174 banks as of 01/01/20 or roughly 52% of the population. As a result of the COVID-19 pandemic however, 37 of those banks or around 21% elected to delay adoption as provisioned for as a part of the CARES Act enacted by Congress. This legislation permitted banks to delay adoption of the CECL standard until the earlier of the end of the national emergency declaration or December 31, 2020. At the earlier of these two events, these Banks will be required to adopt CECL effective 01/01/2020 (refer to the opening statement on the “sort of” CECL Delay).
The following graphic shows the count of the remaining 137 institutions reporting under CECL to include asset size grouping and Census Bureau region of the institution’s headquarters. By count, the South has the largest representation of 48 institutions followed by the Northeast with 34, West with 31, and the Midwest with 24. Based on count, institutions with less than $10 Billion in assets make up 44% of the population (60 banks), $10-25 Billion showing 28% (38 banks), $25-100 Billion with 23% (32 banks), and >$100 Billion the remaining 5% (7 Banks).
Implementation & Adoption
At the beginning of 2020, economic expectations were strong, and a benign credit cycle was in effect. Therefore, base case forecast scenarios were being utilized for many adopters. If anything, banks may have given a small weighting to a recessionary forecast or would have contemplated it under their qualitative and environmental factor framework. As a result, adoption impacts at 01/01/20 were mainly thought to be related to moving from a standard based on Incurred Loss to a framework based on Lifetime Loss expectations. Overall, the average increase to the reserve at adoption was approximately 37% with an average coverage ratio of around 1.07%. However, the variation of change was significant with the lowest being a decrease of 28% and the highest being an increase of 275%. This translates to the lowest coverage ratio of .31% and the highest being 2.92%.
Exhibit 1 details some of the summary statistics surrounding change in overall allowance at adoption and resulting coverage ratios, respectively.
Overall, the average increase to the reserve at adoption was approximately 37% with an average coverage ratio of around 1.07%. However, the variation of change was significant with the lowest being a decrease of 28% and the highest being an increase of 275%. This translates to the lowest coverage ratio of .31% and the highest being 2.92%.
There was also significant variation in the average change based on both asset size grouping and Census Region. As displayed in Exhibit 2, the largest increase was shown for the participants in the Midwest Region at 48%, followed by the South with 46%, West at 27%, and Northeast at 25%. Additionally, institutions with $10-25 Billion in total assets saw the largest increase at 53% followed by the $25-100 Billion grouping at 39%. Institutions greater than $100 Billion showed a 29% increase and those less than $10 Billion showed an increase of 26%.
While there was certainly a significant variation in the change from 12/31 reserve levels, there appears to be some consistency in the overall coverage ratio. However, the underlying average coverage ratio varied significantly based on US region as demonstrated in the below chart:
When considering changes in allowance due to adoption, the results can be significantly different across banks for several reasons. Even among a thoroughly defined peer group, comparability can become difficult. First, portfolio composition is a strong driver in the overall ACL Coverage Ratio for several reasons. As the CECL standard requires an estimation for lifetime expected losses, the expected life of an asset class can be significantly different than another (think longer term Mortgage loan vs. a short term Commercial Real Estate balloon structure). Generally, the longer the life of the asset, the higher the lifetime reserve ratio needed.
Additionally, certain loan types inherently carry more risk than others. When analyzing long term charge off rates, it is easy to see and logically explain that Construction and Land Development loans show higher losses than permanent financing Owner-Occupied Commercial Real Estate. Credit cards are an asset class especially impacted through the CECL Standard based on commonly accepted modeling techniques and resulting expected lives. Therefore, readers of financial statements should be extremely cognizant of portfolio composition for this asset class as total ACL coverage ratios can be significantly distorted.
Furthermore, many institutions have loans that carry some sort of principal loss guarantee where an underlying reserve may not be needed (Student Loan, Government Mortgage Program, or Small Business Administration (SBA) Loans). Lastly, Loans Held for Sale and Other Loans where the Fair Value Election was selected are carried at Fair Value thus having no corresponding ACL shown on the balance sheet (loss contemplated in the fair value presented). In the data set utilized, variation by asset class is significant based on both the asset size of the institution and the region in which the institution is headquartered. Overall, the average composition of each institution’s portfolio is shown in Exhibit 3 below. As can be seen, Commercial Real Estate and Multifamily Loans make up on average 38% of each bank’s individual portfolio. This is followed by Commercial Loans at 24%, 1-4 Family Loans at 16%, and the remaining asset classes of Consumer, Construction and Land Development, HELOCs, and Other Loans making up the remaining compositions (on average 22%).
Lastly, we display the portfolio composition by Census Region. As shown in the chart below, Commercial and Multifamily Real Estate define a much larger segment in the Northeast institutions as compared to other regions in the US. In the Midwest, Commercial loans represent a much larger portion of the overall books as compared to other regions.
It is important when considering the change in allowance at adoption to differentiate between changes related to modeling (different methodology, the use of a forecast etc.) and changes due to the underlying accounting. Of the banks analyzed, 78 of the 137 banks who adopted CECL as of 01/01/2020 have completed at least one acquisition after 2017. The graph below details the composition of the data set based on those completing acquisition(s) as compared to those that have not.
The comparison between the change inACL forBanks with recent acquisitions and those without is striking. Those banks with a recent acquisition recorded on average an increase in allowance that was double the increase in allowance of those banks without an acquisition. On average,those institutions without an acquisition since 2017 resulted in an increase of 21% while those with an acquisition showed an increase of 49%.
This comparison is logical when considering the requirements of the CECL standard described above and is an important consideration when analyzing the impact on adoption. Under the CECL standard, any previously purchased credit impaired (PCI) loan became a purchased credit deteriorated (PCD) loan. As such, a PCD Gross Up calculation must be performed at adoption. Essentially, this allows an institution to move any existing purchased discount to the ACL without impacting retained earnings. Alternatively, under prior guidance, the purchased discount for non-PCI loans was commonly utilized to offset any required allowance for those loans.
In conclusion, not all portfolios are created equally. While this has been reiterated many times during the years of CECL preparation, the analysis above clearly illustrates this point. The analyst should carefully consider the portfolio composition, acquisition makeup, and other factors. While peer group analysis and benchmarking techniques have been widely used in
The past, one needs to thoroughly work through the information at hand to determine if reserve ratios are in line with expectations.
As the first quarter of 2020 progressed, the COVID-19 Pandemic impacted the US economy significantly. While businesses closed or moved to remote work environments, unemployment rose significantly, markets took a hit, GDP contracted, and the level of uncertainty increased to unprecedented levels. The timing of the pandemic’s impact in late March and April coincided with first quarter financial reporting. Under the CECL standard, Banks had to take this economic uncertainty into consideration under the required use of a reasonable and supportable forecast. However, operational challenges presented themselves with such a significant change in the forecast so close to a reporting period. Some institutions had already completed runs of their ACL processes and were forced to restart the process.
Whereas at adoption, many relied on a baseline forecast, at Q1 2020 end, many banks disclosed utilizing some form of a stressed forecast scenario. With economic forecasts suggesting record high unemployment and GDP contraction among other recessionary behaviors, the impact to the estimated ACL was significant. At Q1 2020, the average increase to the ACL was around26% resulting in an average coverage ratio of around 1.33%. Considering the average increase at adoption, this means that the population had increased reserves around 63% compared to year end 2019. In comparison to adoption changes, the overall change was more uniformly distributed across the dataset. Exhibit 4 below display summary statistics around coverage ratios and changes from adoption date reserves.
At the close of the second quarter of 2020, there was more knowledge in the market and less uncertainty at Q1. Multiple government stimulus packages became fully operational including but not limited to the Paycheck Protection Program (“PPP”), Expanded Unemployment Benefits, Taxpayer Stimulus Checks, Mandated Foreclosure and Eviction Moratoriums, and almost unlimited Monetary Easing Programs (Fed Funds rates cut of 150 Basis Points (“BPS”), Debt Purchases including even backstops to the Corporate Debt Market, and Federal Reserve Lending Facilities). In total, federal assistance accumulated to approximately $3.3 Trillion. Furthermore, a provision in the CARES Act allowed for any modification or deferral made in relation to the pandemic to not be disclosed and accounted for as a Troubled Debt Restructuring (“TDR”); assuming the loan was paying in accordance with contractual terms prior to the pandemic. These actions led to unprecedented levels of loans entering deferrals and modifications that typically ranged from 3-12 months.
As time went on, assistance programs took effect and states began to re-open for business. Prior expectations of 25% Unemployment and 20% GDP contractions turned out to be less severe based on reported falling unemployment rates of 14.7% in April, 13.3% in May, and 11.1% in June. Although COVID-19 was certainly a national pandemic, each states’ response during Q2 2020 was significantly different thus leading to wide array of impact across the country.
At the end of Q2, fears over a second shut-down due to another wave of cases loomed and the economic state was still showing reported values combating those at the height of the Great Recession. Banks for the most part maintained their same approach for reserves as at Q1. Allowances stayed high, with minimal release. However, the level of reserve builds did begin to level off. The average ACL coverage ratio climbed to 1.44% showing another 10% increase over Q1 2020 reserve levels. It should be noted that in some cases, the headline coverage ratio was shown to decrease in many instances. Due to the level of PPP loans in some institution’s portfolios, coverage ratios were depressed as compared to prior periods given the higher composition of the PPP loans being fully guaranteed by the SBA.
Therefore, by looking at Q2provision charges as compared to Q1 2020 provision charges, one can glean a clearer picture of industry expectations.Across each asset size grouping,the average Q2 provision charge was approximately 128% of the Q1provision charged demonstrating the continued build in reserves.
The following charts demonstrate the impact to ACL and provision during the second quarter.
As we move into Q3 reporting season, reserve levels are expected to stabilize as compared to prior periods and some of the larger institutions have begun to offer more detailed insights. Many institutions have reported significant improvements in the level of their portfolios that are under deferrals or modification programs and have also provided insight into more limited populations requesting further assistance. Unemployment levels have continued to fall to reported levels of 10.2% in June, 8.4% in July, and 7.9% in September. However, improvement in the labor markets has begun to slow and there is increased uncertainty around Congress’ ability to pass another comprehensive stimulus bill. As we move into year end and close out the first annual period of reporting under the CECL standard, 37 more banks will be included as the CARES Act deferral expires. For all institutions, significant reserve levels have been built. Assuming the losses that have been provisioned for begin to come to fruition, one would expect future reserve levels to fall. In the full spirit of the CECL standard, expected losses should be provisioned for in a more timely fashion than under the prior standard thus leading to much lower coverage ratios while charge off rates are climbing. While it can be argued CECL performed as expected during times of crisis, it will be interesting to see how the industry responds as we hopefully move into a season of economic recovery.
About the Author
Mikayla Spurlock, CPA, Senior Analyst
Mikayla has over three years of accounting and consulting experience with financial institutions ranging in asset size from $300 Million to $19 Billion. She works on projects related to due diligence and Day 1 valuations and Day 2 Accounting. Mikayla also works with clients on ALLL and CECL implementation and ongoing allowance advisory consulting. Mikayla earned her Master of Accountancy with a concentration in Business Measurement and Assurance from the University of South Carolina. She also earned her Bachelor of Science in Accounting and Finance from the University of South Carolina, where she was a member of the Capstone Scholars program. She is a licensed CPA in the state of South Carolina.
Derek Hipp, CPA, Director of Consulting
Derek is a Co-Founder of Valuant and leads the Client Service division. Valuant assists financial institutions with various financial modeling and data analytical services including Acquisition Accounting, Valuation, and Credit Risk Analysis (Current Expected Credit Loss or “CECL”) through delivery of a proprietary suite of Software as a Service (SaaS) solutions, ValuCast™. The Client Service team assists customers with onboarding, training, and execution of
these products and services. Derek is also a member of the Valuant leadership group focusing on team development, business development, product delivery, software feature development and exploration, and other strategic initiatives.