By Josh Stein and Mike Gullette
ABA Viewpoint
When Federal Reserve Vice Chair for Supervision Michelle Bowman repeated her recent call for a repeal of CECL for community banks, most discussion immediately focused on burden reduction. That’s understandable. Community banks have spent years buying models, hiring consultants, documenting forecasts, supporting qualitative factors, and responding to auditor and examiner questions.
Josh Stein is VP for accounting policy, and Michael Gullette is SVP for accounting policy, at ABA.
But we wonder if we’re looking at the issue too narrowly. What if the biggest opportunity isn’t just simplifying credit loss estimates? What if the biggest opportunity is also improving them?
When we look at how CECL is being performed today, it seems that many institutions spend an enormous amount of time and resources implementing and defending models, even though that estimation process often does not align with the credit analysis that matters most.
Community bankers are still reviewing loans, monitoring borrowers, analyzing concentrations and tracking local economic conditions — the work that really matters. The opportunity is to better connect the allowance estimate to that bread-and-butter work, so that the process reinforces — not distracts from — the day-to-day identification and management of credit risk.
Repealing CECL could well leave us with the same onerous documentation and validation requirements for banks. A practical expedient offers an opportunity to rethink that entire process.
Thousands of banks, repeatedly solving the same problem
Consider what happens today. Across the country, thousands of institutions are independently deciding:
- Which historical periods to use.
- How to convert historical losses into lifetime losses.
- What economic forecasts to use.
- How long forecasts should remain supportable.
- How reversion should work.
- Whether qualitative adjustments are needed and how to quantify them.
- How to document and validate all the above.
In many cases, institutions are answering essentially the same questions using different methodologies and the result is an incredible amount of duplicated effort.
A practical expedient could recognize something very simple: Not everything needs to be reinvented by every bank every quarter. Instead of requiring thousands of institutions to independently develop estimation frameworks, agencies could standardize significant portions of the process using publicly available data, transparent assumptions and long-term industry experience.
That would not eliminate judgment. It would simply standardize the portions of CECL that often produce little incremental value.
The real credit work already happens elsewhere
One of the ironies of CECL is that the most valuable credit analysis often has little to do with the model itself.
Community banks know which loans are attracting attention. They know which borrowers are showing signs of weakness. They know which credits are criticized, classified, modified or collateral-dependent. They know which industries are facing pressure. That work happens through loan review, credit administration and conversations with customers.
In other words, the most important risk management activities are already occurring.
A practical expedient would allow banks to devote more resources to those activities and fewer resources to constructing forecasting methodologies for performing loans whose long-term credit performance is relatively predictable.
A better starting point already exists
Another uncomfortable truth is that most community bank portfolios are simply too small to produce statistically reliable estimates.
Recent loss experience is often exceptionally low. Individual charge-offs are infrequent. A single loss can dramatically change calculated loss rates. Yet at the same time, decades of industry-wide charge-off experience already exist through publicly available sources. Those data show how loan portfolios have behaved during stable periods, recessions, recoveries, and everything in between. Why should a bank with 10 years of limited loss experience start by placing greater reliance on that small sample than on decades of industry-wide performance data?
A practical expedient could start with a more reliable foundation: long-term industry-wide experience. Instead of asking every institution to determine its own historical period and conversion methodology, regulators could publish transparent benchmark assumptions that effectively convert annualized losses into lifetime CECL estimates. That alone could eliminate a significant amount of recurring complexity.
The most interesting part may be economic forecasting
When people think about CECL burden, economic forecasting often comes to mind first. But economic forecasting may also be where a practical expedient creates the greatest opportunity.
Consider a community bank with a meaningful concentration in commercial real estate. Today, management may be expected to support views about vacancy rates, rental growth, property values, capitalization rates, and broader economic conditions.
Now imagine a different approach. Suppose that same bank primarily lends on office and multifamily properties in a specific metropolitan market. Rather than creating its own economic forecast, the bank could look at publicly available market information. The Federal Reserve Bank of Atlanta’s Commercial Real Estate Market Index, for example, evaluates commercial real estate conditions by property type and market, providing an objective view of sector-specific conditions. A practical expedient could allow institutions to use those standardized indicators to determine whether local office or multifamily markets should be considered stable, transitioning or stressed.
The same concept could apply to residential mortgage portfolios. A bank operating in a single market might review local unemployment trends, home-price trends and mortgage performance indicators. Those factors could drive an economic-state determination through a standardized decision framework rather than through a bespoke forecast.
The objective would not be to eliminate forward-looking information. The objective would be to standardize how it is evaluated. And perhaps most importantly, two banks looking at the same local economy would be far more likely to reach similar conclusions. Under such an expedient, rather than debating dozens of modeling decisions, bankers could focus on benchmark comparisons, peer relationships, economic-state assignments and observable credit trends.
Practical steps to take now
The good news is that most of the building blocks already exist. The industry has decades of loss data. The Federal Reserve System publishes extensive economic research. Local labor-market data are readily available. Housing market indicators are readily available and CRE market information is increasingly available at the property-type and market level. And the review framework we just described would be relatively easy to build — and to automate!
The question therefore may not be whether a practical expedient is possible. The question is whether the industry is willing to move beyond the traditional assumption that every institution must independently build, support, validate and defend its own estimation framework. Bowman’s call for repeal should not be viewed merely as a request for simplification. It should be viewed as an invitation to build something better.
ABA Viewpoint is the source for analysis, commentary and perspective from the American Bankers Association on the policy issues shaping banking today and into the future. Click here to view all posts in this series.
