What Can Community Bankers Learn from Large Bank CECL Disclosures?

Last month, the four largest US banks began releasing their second quarter 2020 earnings reports and many analysts were focused on their updated estimates for credit losses anxious to see the effects of the Covid-19 crisis on these leading financial institutions. These second quarter 2020 earnings releases represent only the second time large bank provision expenses have been recorded using CECL, (Current Expected Credit Losses – ASU 2016-13), the new accounting standard for credit losses. This new credit loss standard was first implemented January 1st, 2020 by approximately 90 of the largest US banks meeting the FASB requirements for mandatory adoption, effectively placing nearly 80% of all US bank loans under the new CECL standard.

While many community bankers are still hoping that the effects of the new guidance will never apply to them or that the standard will be repealed in whole, the reality is that most of the banking industry, as measured by loan portfolio dollars (rather than by number of financial institutions) has already adopted the standard. According to current FASB rules, community banks and other smaller financial institutions aren’t scheduled to adopt the CECL standard until January 1st, 2023.

Implementation of the CECL standard in normal times would have been a daunting enough challenge even for the largest US banks that have abundant resources available to build the large data warehouses, assemble the complex loss forecasting models and link meaningful economic forecasts to those models, as required under the CECL standard. However, towards the end of the first calendar quarter for CECL adoption, the Coronavirus pandemic struck, beginning a downward slide in business activity fueled by a rapid rise in unemployment caused by the lockdown of a significant portion of the US economy.

As a result, financial institutions of all sizes, whether using CECL-based metrics, (large banks), or the traditional Annual Incurred Loss methodologies, (ALLL), (smaller institutions), began increasing their provision for credit loss estimates.  Despite the various forms of stimulus being provided by the Federal government, and lower interest rates and increased liquidity supplied by the Federal Reserve, some consumers and businesses are struggling to remain current on their financial obligations. Forbearances granted, as well as PPP loans soon to be forgiven may be helping to delay the onset of actual loan delinquencies. However, banks large and small are still preparing their balance sheets for what may become an expanded credit crisis in the second half of 2020, especially given the risk of a significant spike in coronavirus cases possibly leading to the need for further restrictions on businesses and individuals.

The chart above traces the recent trends in Net Charge-offs and the combined Provision for loan losses recorded by all US commercial and savings banks over the last 10 quarters, including a forecast for Q2, 2020 based on the early results released by the “Big 4” banks, (JPMorgan Chase, Bank of America, Wells Fargo and Citibank). The forecast applies the rate of increase in each of NCO’s and PFCL’s taken from the Big 4 and applied to the Q1 2020 actuals for the entire industry; this may or may not provide an accurate forecast. Actual results for the industry won’t be known until all banks have reported Q2 results, which likely won’t be available until approximately August 15th, 2020.

What the data clearly show is that from year end 2017 through year end 2019, Net Charge-offs and the recorded Provision for Credit Losses remained essentially equal, at an historically low level, consistent with the results of most of the last five years.  For most of the recent past, over periods of time long enough to form a complete business cycle, these two data points, (NCO’s and the PFCL’s), have remained essentially equal. Starting in Q1 2020, the Provision for Credit Losses rose significantly as a function of two factors; first, the adoption of CECL by the 90 or so largest banks, and secondly, due to an early expectation that the Coronavirus pandemic would eventually cause an increase in loan losses as a result of escalating unemployment and severely negative GDP results. To date, with what can be seen regarding Q2 2020 based on the recent earnings releases of the Big 4 banks, the increase in expected loan losses is accelerating in a significant manner. It is interesting to note that Net Charge-offs remain very low in comparison to the recognized Provision Expense, an expected outcome of the transition to CECL during periods of economic contractions.

The table above captures the activity of the Big 4 banks as it relates to their total Q1 recognition of loan losses, including their one-time CECL adoption impacts. These four banks combined hold approximately $3.6 Trillion of loans, which equals approximately 35% of the total loans held by all US Commercial and Savings Banks. During the first quarter of 2020, these banks recorded a total Provision for Credit Losses of approximately $21.2 Billion, which represented a very significant increase from Q4 2019, ($4.9 B), or any recent prior periods, during which time PFCL’s have been at historically low levels, and were reported only under the previous annual incurred loss rate method.

Among this small group of the largest banks in the country, only Wells Fargo recorded a decrease in its provision expense during Q1 related to its transition to CECL. During the initial quarter, Wells decreased their PFCL’s by $1.3 Billion due to CECL implementation, while still recognizing credit deterioration from other sources. Excluding Wells, the remaining three of the Big 4 banks recognized a Provision Expense in the first quarter of $18.0 Billion, of which, $10.5 Billion, or 58% was related to the one-time impacts of transitioning to the new CECL accounting standard. For these three banks, the remaining $7.5 Billion increase represented their current assessment of credit losses during the Q1 period. The range of Provision Expense as a percentage of average loans for just these four banks, excluding the CECL one-time transition adjustment, has risen from Q4 2019 from a Low of 0.07% to a High of 0.27%, to a range for Q1 2020 of a Low of 0.15% to a High of 0.47%, a relatively significant increase.  As we’ll see below, these ratios continued to increase in Q2 2020.

The table above provides an early forecast of what might be in store for the banking industry for Q2 2020 and beyond, based on the disclosures of the Big 4 banks, which further increased their reserves in Q2 by recognizing even higher levels of Provision for Credit Losses. Total Provision Expense for the four banks for Q2 was $33.1 Billion, an increase of $21.1 Billion over the normalized level of Q1, (excluding one-time CECL transition costs), or 275% higher (please note that while this is a very large increase, it is coming off of a very low starting point).

As of Q2 2020, the Provision for Credit Losses / Average Loans ratio now ranges from a Low of 0.50% to a High of 1.12%.  Most of the banks cited in their earnings releases that the increase in Provision Expense was mostly a function of a deterioration in their economic forecast for the remainder of 2020, and NOT a function of increased Net Charge-offs, which remain relatively low. Certain banks suggested that these increases in Provision Expense and Reserves may be enough to carry them through the end of 2020.

Impact for Community Bankers

The economic consequences of the coronavirus pandemic will impact all US banks, but it is unlikely that it will impact all banks in the same way.  The effects currently being recognized by large institutions likely will not impact community banks in a similar manner, for at least 3 major reasons:

  1. The difference in accounting methodologies – CECL for larger banks, and the Annual Incurred Loss Rate method for community banks;
  2. The product mix of large bank’s loan portfolios include more consumer loans and credit card balances which have much higher loss rates;
  3. Community bank loss rates have traditionally been much lower than larger banks, in addition to the product mix differences cited above, also because community banks have fewer customers whom they are closer to and know better.
The chart above tracks the differences in the Provision for Loan Losses as a Percentage of Average Loan balances over the past 10 quarters and provides an early forecast of the possible direction of this ratio for Q2, 2020.  During this two-and-a-half-year period, larger banks (> $10 B in Total Assets) have consistently incurred a higher provision loss rate than smaller banks (< $10 B in Total Assets).  During this extremely low loss rate period, large banks averaged roughly double the provision rate of small banks, with an average of 0.14% vs 0.07% for smaller banks.  Beginning in 2020, this spread between larger and small banks will be widening, due to both the difference in accounting methods, and real underlying credit loss differences.
The graph above summarizes the losses incurred by the Big 4 banks in the years following the great recession.  Some analysts have suggested that while the current Covid-19 induced recession and the Great Recession are different in many ways, this prior recession may still provide some indication of the type of losses that might occur during the current recession, if only because the two recessions share somewhat similar unemployment consequences.

The unemployment rate in June of 2020 moderated to 11.1% from a high in April of 14.7%.  The unemployment rate during the Great Recession maxed out at 10.0% in October of 2009, however unemployment averaged above 8% consistently for four years from January of 2009 through December of 2012.  Hopefully, losses during this recession will be much lower, assisted by the unemployment rate returning to a more moderate level much more quickly.

Significant differences in the loan portfolios of larger vs. smaller banks are a major component of the higher loss rates experienced by large banks, as is also illustrated in the NCO chart above.  Using these Big 4 banks as a relative proxy for larger institutions, nearly half of all Net Charge-offs recorded during the six years following the onset of the Great Recession, were incurred on Credit Card Loans (35%) and on Consumer Loans (14%), loan types which are typically not found, or not found in concentrations, on community bank balance sheets.  This single difference explains much of the difference in loss rates between larger and smaller banks.  The chart also discloses that large banks also incurred significant losses on 1-4 Family residential mortgage loans and on Commercial & Industrial loan types, products which community banks share with the larger institutions.

Recommendations for Community Bankers

As the largest institutions in the country adopted CECL in Q1 2020 and in Q2 are now adjusting their rolling “life-of-loan” loss estimates under the back-drop of the Coronavirus pandemic, community bankers may well be asking themselves, what does all of this mean for me and my financial institution?

Even though community banks are not required to adopt CECL until 2020, board members, shareholders, regulators and audit firms are likely to be asking community bank leaders what the CECL impact is on your institution and will be expecting you to have knowledgeable and insightful responses to their questions.  In addition, you will be taking emerging information about the effects of the current economic situation into account in setting the Qualitative Adjustment factors you are using to adjust your current loss estimates under the existing Annual Incurred Loss methodology.

CECL Model Development Activities Within Community Banks

Most community banks by now have begun to develop at least rudimentary CECL based loss forecasting models appropriate in scope to the size and complexity of their bank’s loan portfolios.  Normally included in this preparation is the collection of the necessary supporting data required to project future losses based on previous relevant experience under any of the existing CECL methodologies.

Many community banks with smaller non-complex loan portfolios will not find it necessary to go beyond deployment of some of the simpler loss forecasting methods such as the Enhanced Historical Loss Rate method, or perhaps the Weighted Average Remaining Maturing (WARM) method.

Larger community banks may also build out one or more of the more advanced CECL loss forecasting methodologies, including Vintage Analysis, Migration Analysis, Probability of Default / Loss Given Default, (PD/LGD), or the Discounted Cashflow method.

Completion of these preparatory steps establishes your community bank as well positioned for eventual CECL implementation, but more importantly, provides you with a reasonable tool with which to answer incoming questions regarding not only your institutional readiness for CECL, but also the predicted effects of CECL on your future provision expense, reserve balances and the potential impact to your bank’s capital.

Parallel Runs – CECL vs ALLL

Once your institution has collected enough historical data and has an appropriate CECL loss forecasting methodology in place, best practices indicate that you should be running parallel loan loss estimates from your CECL model and comparing these to the results of your existing ALLL results, noting the amount and location of significant differences.  Each model (CECL and ALLL) should be simultaneously updated using your institutions current economic forecast.

As in the current environment, these Qualitative Adjustment factors will play a significant role in determining your final loss estimates, as has been pointed out by the experiences of the larger institutions already functioning under the CECL guidance.  Performing these parallel runs once a year, or more frequently when conditions are changing materially, would be considered a minimum standard.

Loan & Capital Stress Testing

Whether your institution has a high or low degree of confidence in the accuracy of the parallel runs currently being produced by your existing CECL model, another valuable tool for assessing your institutions readiness for CECL is a Loan & Capital Stress Test.

The stress testing process normally follows the recommendations of the banking supervisory agencies Supervisory Capital Assessment Program (SCAP) and the more recent Comprehensive Capital Analysis and Review (CCAR), as is most generally applicable to larger institutions, but valuable to community banks, as well.

Stress testing your loan portfolio using a number of different potential loss rate scenarios, most of which will be in excess of the loss levels you might expect to experience based on your CECL model’s predictions, can be a very helpful exercise in assessing the impact of an underestimation in your CECL model loss forecast, or an unfavorable change in the unemployment rate or the general performance of the economy.


Much of the banking industry has already adopted CECL and many community bankers have invested time and energy in preparing for CECL’s eventual application to their institutions though the development of models, the collection of data, etc.

Both the eventual application of the CECL standard to community banks, and the accurate reporting of loan losses under currently existing methodologies in the meantime are management tasks made much more difficult by the current Coronavirus pandemic and its related economic consequences.

Community bankers wishing to remain connected to the trends impacting the industry in general must continue to be aware of the trends affecting other relevant institutions in the marketplace, be cognizant of the significant differences between those institutions and theirs, and have keen insights into their own portfolios using metrics that are consistent with industry best practices.

Should you have any questions regarding the application of any of the ideas and recommendations included in this article, or if you would like to have a personal discussion of your institutions unique situations related to CECL, loan losses, credit management, or other issues, please feel free to contact the author at  jmorris@probank.com.