Analysis of Transparency Report Similarities of Audit Firms Through Plagiarism Detection Systems

Abstract

There has been a rising movement in financial environments towards the adoption of Artificial Intelligence (AI) based tecniques. This study aims first to measure document similarity of transparency reports (TRs) disclosed by Turkish audit firms between 2013-2020 through a unique application of plagiarism detection systems. Further, it also aims to observe firm-based attributes influencing both intra and inter similarity scores. The results show that TRs greatly share intra similarity, providing evidence for boilerplate effect recognized in accounting/finance literature. Moreover, firm age, big four involvement, and numbers-to-words ratio have negative effect on the intra similarity while ownership structure has a positive influence on it. On the other hand, the inter similarity is lower for firms having big four involvement, foreign license and higher female partner and numbers-to-words ratios.