Michelle Bowman makes a compelling case in her op-ed “How to Make Life Hard for Small Banks” (Sept. 9). In 2016 guidance on the current expected credit losses methodology (CECL), federal financial regulators anticipated that most smaller institutions wouldn’t need “costly and complex models.” As Ms. Bowman documents, reality hasn’t borne this out. To her arguments, we would add the historical arc that shaped this approach.

Under the incurred-loss methodology that the CECL replaced, a bank waited for evidence that a loss was probable before setting aside reserves to cover it. After the 2008 financial crisis, that practice drew blame from critics partly because many banks had faced substantially higher provisioning needs only as the downturn deepened. Yet, as Ms. Bowman rightly points out, the incurred-loss methodology was better “aligned with how credit risk actually works.” We concur. The mere existence of credit risk, which is inherent in lending, isn’t the same thing as an incurred credit loss.

The Financial Accounting Standards Board responded to the post-crisis criticism with CECL, which obliges banks to apply forward-looking estimates of expected credit losses. However, provisioning for losses that may never materialize is a questionable precaution, not least because economic and financial conditions are notoriously difficult to predict. For example, economists polled in the Philadelphia Fed’s Livingston Survey, published in December 2007 as the recession began, projected that the economy would keep expanding through 2008, and put unemployment at 4.9% for 2009. Instead, unemployment averaged 9.3%.

It bears remembering that community banks are fundamentally in the business of lending, not forecasting.