Influence of earnings quality dimensions on the peception of earnings quality: an empirical application of composite PLS using archival data

Abstract

Despite the fact that empirical research on Earnings Quality (EQ) has used a wide range of earnings properties that are expected to be related to EQ, research on how these properties affect investors’ perception of earnings quality is scarce, as most of the papers on EQ focus on a single EQ dimension. Moreover, extant research presents some limitations, as most studies rely on first-generation statistical methods (mainly OLS), without empirically testing the validity of the indicators used for capturing the underlying EQ dimension. This paper aims to explore how the different EQ properties described by previous literature map onto stockholders’ perceptions of EQ. Using partial least squares – PLS – our results show that some of the properties more widely studied by accounting research (such as accruals quality) have little influence on stockholders’ perceptions of earnings quality, whereas other, less studied properties (such as persistence and smoothing) exhibit a stronger relationship with stockholders’ perceptions of EQ. Our results also show that the most usual indicators previously used in empirical research to represent accounting conservatism do not converge in a single construct, possibly indicating that those indicators may represent different underlying concepts.