Sifting Through CECL: Challenges and Implications Beyond Provisioning

By Grigoris Karakoulas
InfoAgora Inc.

The Current Expected Credit Loss (CECL) standard from the Financial Accounting Standards Board (FASB) is considered one of the most challenging accounting change projects in a generation, with significant and widespread impact on banks. Not surprisingly, FASB has been inundated with feedback and pushback since CECL was issued in 2016. In early June, two bills were introduced in the House and Senate with bipartisan support to “stop and study” CECL before it takes effect in January 2020 for SEC filers. At its board meeting on July 17, FASB affirmed the effective date of January 1, 2020, for the SEC filing banks—excluding smaller reporting companies—and extended the deadline for all other banks—the majority of community banks—to January 1, 2023.

Accounting standards implementation is often a finance-only effort, but not CECL. It has many governance, modeling, credit analysis, information technology and financial reporting interdependencies. Due to these interdependencies, the risk of not properly preparing for CECL is significant, especially as the odds of a recession in 2020 have been increasing. How does a bank prepare for a downturn through its CECL provisions?

Financial institutions have been using the incurred loss standard for calculating loan loss provisions. As this standard is backward-looking, it results in provisions that peak during economic downturns. Instead, CECL requires institutions to combine a credit-loss model based on historical information with current conditions and forecasts of future economic conditions and future internal conditions. By doing this, banks can form a forward-looking view of losses.

Due to the non-prescriptive nature of the new standard, the different implementation choices—from data and segmentation to macroeconomic scenarios, reasonable and supportable (R&S) horizon and loss forecasting method—may render the loss provisions of a bank less comparable with its peers. In this article, we examine the effects of some of these design choices on loss reserves and beyond.

Thus, the impact from CECL on loan loss reserve (LLR) varies a lot amongst a dozen of the big banks that disclosed estimates as part of their first and second quarter 2019 reporting. While the average increase in LLR is 26 percent, it varies from a 5 percent reduction (Wells Fargo Bank) to a 62 percent increase (Goldman Sachs). The impact on common equity tier 1 (CET1) capital ratio appears to be manageable with an expected decline in the CET1 ratio of 17bps on average. The regulatory agencies also issued a final rule in December 2018 that provides financial institutions with “an option to phase in over a period of three years the day-one regulatory capital effects” of CECL.

Although the above impact disclosures are from big banks, there are a couple of lessons learned that apply to community banks as well. First, one of the main contributors to LLR variability among banks is going to be portfolio mix. Banks with heavier consumer exposure and longer duration portfolios are expected to be impacted more than ones with shorter duration portfolios, i.e., commercial loans. Thus, most of the increase in LLR for JPMorgan Chase and Citigroup is due to their credit card portfolios. In contrast, the small drop in LLR for Wells Fargo “reflects the expected decrease for commercial loans, given their short contractual maturities and the current economic environment, partially offset by an expected increase for longer-duration consumer loans.”

Second, there is at least one other factor—the length of the R&S horizon—that could be causing LLR variability. In the case of Wells Fargo, it is set at one year and may be contributing to the insignificant impact on its loss reserves from CECL. The duration of the R&S horizon is a judgmental decision reflecting management’s confidence in the forecast of economic conditions that drive the estimate of expected credit loss (ECL). A shorter R&S horizon may not capture peak losses when the economy is expected to deteriorate, for example starting in the third quarter of 2020. The longer the R&S horizon, the less procyclical the ECL will be.

In their April 2019 joint commentary on CECL guidance, the regulatory agencies state that banks should provide evidence for the selection of the R&S horizon independently of their stress-testing process.

Although the CECL standard is meant to be countercyclical by design, the various choices for its implementation may render it procyclical, causing volatility in loss allowance and earnings. Thus, in addition to the R&S horizon, the selection of scenario(s) can also lead to procyclicality. A scenario that is not relevant to the asset class and geography of a bank may inject significant inaccuracy and therefore procyclicality to the CECL allowance simply because it will not reasonably forecast the cycle of that asset.

Scenario relevance is an issue, particularly important for community banks due to their lending specialization and the local nature of their footprint. For this reason, the regulatory agencies point out that the banks should be using “economic variables and other factors relevant to the collectability of an institution’s portfolios.”

Are macroeconomic scenarios at the national level relevant to the collectability of a portfolio of a community bank with a specific geographical footprint? Figures 1 and 2 illustrate the importance of geography in asset prices. Different MSAs within Texas have exhibited different business and credit cycles both statewide and nationwide (see, for example, Austin and Midland). Furthermore, there is a lot of variance in the year-to-year changes of house prices (HPI) and the unemployment rate across the state. For example, in the first quarter of 2018, the unemployment rate varied between 2 percent to just under 6 percent, whereas the year-to-year change in HPI was between 0.0 percent to just over 10 percent.

Are macroeconomic scenarios at the national level relevant to the collectability of a community bank’s multifamily CRE portfolio, given the different credit cycles across asset classes and geographies? Figure 3 illustrates the difference between the multifamily CRE cycle vs. the U.S. economic cycle (light blue bars), and the difference of multifamily CRE across U.S. regions. Multifamily CRE was not affected during the 2001 recession. Although it was more resilient than other CRE asset classes during the Great Recession, it experienced a U-type recession in Midwest and South compared to a V-type in the rest of the regions. Ignoring these differences may result in significant inaccuracy in allowances and earnings volatility.

FASB and the regulators have provided a non-exhaustive list of forecasting methods that would be compliant with the new standard. Although adopting any of those methods would tag a CECL implementation as compliant, each of them incorporates different assumptions and complexity/accuracy trade-offs. For example, the methods of static pool loss rate and weighted average remaining maturity (WARM) have low complexity, but cannot capture changes in loan origination, effects from aging and maturity, and changes due to the forecast of the collateral. They may, therefore, introduce significant inaccuracy in loss estimation—up to three times the error of alternative methods—and hence procyclicality in loss allowance and earnings volatility.

Since expected loss is one of the factors affecting loan pricing, the accuracy in loss estimation can have direct implications for the pricing and lending decisions of a bank. There is certainly a cost associated with implementing, monitoring and reporting with CECL. The benefits from more accurate loss provisioning and better pricing far outweigh such costs.

Given the above design choices and their significance beyond loss reserving, it is imperative for community banks to identify the proper choices according to their asset mix, data granularity and geographical footprint. Because compliance with CECL can be an opportunity for a bank to enhance profitability by more accurately forecasting its expected losses.

Grigoris Karakoulas is president and founder of InfoAgora Inc., which provides risk management consulting, prescriptive analytics and RegTech solutions (CECL/IFRS9/IRRBB/Basel III) to financial services organizations. Contact him at

Published in Bankers Digest October 7, 2019