The accounting quality of the incurred versus expected credit loss allowances

Abstract

The impairment model in International Accounting Standard 39 Financial Instruments: Recognition and Measurement (IAS 39) is an incurred credit losses (ICL) model, and it was replaced by an expected credit losses (ECL) model under International Financial Reporting Standard 9 Financial Instruments (IFRS 9). The ECL model incorporates greater management discretion, but the impact of this on accounting quality in general, and on credit losses in particular, is uncertain.

Samples in existing literature on this topic have been dominated by code law countries. Legal systems govern the legal rules which ultimately influence financial markets (La Porta, Lopez-de-Silanes, Shleifer & Vishny, 2000). Accounting quality often differs between code law and common law countries, as legal systems have a direct impact on various elements of financial markets, management discretion, management incentives, laws and investor protection in a country. Thus, the impact of the change from the ICL model to the ECL model under the different legal and institutional environments in common law countries is unclear. This study therefore investigates the impact of the shift from an ICL model to an ECL model on accounting quality of listed banks in four common law countries, namely Australia, Canada, South Africa and the United Kingdom, and develops theoretical expectations specific for these countries. It focuses on three measures of accounting quality: income smoothing, timely loss recognition, and value relevance. In line with theoretical expectations, findings from a multivariate regression approach show that overall, accounting quality has not changed or that, at best, it has only improved slightly in common law countries since the adoption of the ECL model.

This study also challenges the theoretical argument that stage 3 credit losses under IFRS 9 are the equivalent of the ICL model under IAS 39. The findings provide empirical evidence that the accounting properties of stage 3 impairments in the ECL model are not the same as in the ICL model. Furthermore, additional analyses of the stages of the ECL model provide preliminary evidence of the drivers behind changing accounting quality metrics and the differences in accounting quality between the stages.

The sample years used for the post-ECL implementation period are limited. Future researchers may therefore wish to repeat the study when data are available for more sample years, as the limited sample period precludes the potential impact of a learning effect. The findings of the current study may be of interest to standard setters, academics, regulators, preparers of financial statements, investors, and other participants in the banking industry.