Test-bedding the new reporting standards for loan loss reserves

Abstract

We test-bed the differences in properties of the Expected Credit Loss (ECL) and Current Expected Credit Loss (CECL) models in a bank loan setting with respect to their impacts on the adequacy, comparability, and predictability of loan-loss reserves and the volatility of reported profit. To do so, we develop a stylized bank-loan setting in a controlled laboratory environment with a series of eight different secured personalloan portfolios. Fifty-six senior university accounting students take the role of loan managers responsible for making annual loan-loss reserve decisions. We find that the laboratory ECL and CECL reserves are greater than the baseline reserves for the Incurred Credit Loss (ICL) model (which was in place before the introduction of ECL and CECL models), but both regimes lead to lower reserves in the laboratory setting than we find for our risk-neutral baselines. In the laboratory, CECL displays fewer uncovered reserves and greater excessive reserves than ECL. The comparability of reserves deteriorates under the two new regimes relative to ICL, and CECL shows relatively more comparable reserves than ECL. While the laboratory ECL exhibits a modest increase in predictability over ICL, CECL shows a significant decline. Finally, profit volatility falls significantly under both ECL and CECL relative to the ICL baseline, and the laboratory results show less volatility than the risk-neutral baseline values.