CECL model reduces investor processing efficiency in banks
A Banco de España working paper finds that the Current Expected Credit Loss (CECL) model impairs equity investors' ability to efficiently process banks' earnings announcements. This effect is more pronounced when loan loss provisions are driven by new loan origination.
Complexity versus transparency
The CECL model, implemented for U.S. banks in 2020, aimed to improve the timeliness and transparency of loan loss provisions by requiring the recognition of lifetime expected credit losses at loan origination.
This shift followed criticism of the previous Incurred Credit Loss (ICL) model for leading to 'too little, too late' provisions after the 2008 financial crisis.
However, critics argue that CECL has increased the complexity of these provisions, obscuring their information content and limiting their decision-usefulness.
The estimation of lifetime expected losses involves more assumptions and discretion.
Furthermore, a timing mismatch arises as expected losses are recognized upfront, while interest revenue accrues gradually over the loan's life, making it harder for investors to interpret valuation implications.
Academic research remains inconclusive on CECL's effectiveness, with some studies showing improved information production and others finding reduced decision-usefulness.
Slower resolution of investor disagreement
The paper contributes to this debate by examining CECL's effect on equity investor processing efficiency during earnings announcements, when loan loss provisions are disclosed.
The study measures processing efficiency using the speed of resolution of investor disagreement, reflecting how quickly disagreement is resolved after an announcement.
Employing a difference-in-differences methodology, the research compares publicly traded U.S. banks that adopted CECL in January 2020 with those exempt until January 2023.
The main finding is that CECL adoption reduces investor processing efficiency, leading to a 2-percentage point decrease in the speed of resolution of investor disagreement.
This effect is material, reducing the average pre-2020 gap in processing efficiency between CECL and ICL banks by 50 percent.
The results are robust to various controls, including bank fundamentals and the information environment, and persist beyond the initial COVID-19 shock.
A warning for forward-looking models
This study provides robust empirical evidence supporting concerns about CECL's practical implications for market transparency.
While methodologically sound, its focus on US banks limits direct generalizability to other regulatory regimes.
Nevertheless, the findings offer a crucial warning to regulators considering similar forward-looking provisioning models.