The Relevance of Expected Credit Losses: The effect of IFRS 9 on analyst forecast accuracy

Abstract: This study examines how the adoption of the expected loss model under IFRS 9 has affected the forecast accuracy of credit losses. Specifically, we investigate the effect on absolute forecast errors and forecast dispersion. To establish the effect, we employ a difference-in-differences analysis using a dataset that includes 39 European banks that adopted the standard on January 1, 2018. To control for the observed effect on the European data set, we employ a control group consisting of U.S. banks reporting under U.S. GAAP. The study covers 24 quarters between 2014 and 2019. Our results suggest that the absolute forecast errors and forecast dispersion increased more for the European banks than the U.S. banks after the IFRS 9 mandatory adoption date. Accordingly, we conclude that the forecast accuracy of credit losses has decreased. In relation to IASB's Conceptual Framework, our results imply that the relevance of credit losses has likewise deteriorated. However, we assert that the deterioration may only reflect a temporary effect as analysts adapt to the new information environment. Our study makes two primary contributions. First, we provide early evidence that the forecast accuracy of credit losses has deteriorated after the adoption of IFRS 9. Secondly, we add to the existing literature on forecast accuracy in relation to the adoption of new accounting standards by shedding light on analysts' adaption to a new information environment.