A Review on the Multidimensional Analysis of Earnings Quality

There is a generalized consensus among accounting researchers about the multidimensional nature of earnings quality: Earnings quality depends on a series of characteristics that enhance the usefulness of the earnings figure for decision making. In this paper, we undertake a literature revision on empirical research on earnings quality that reveals that, although earnings quality is probably the most recurrent topic in accounting, empirical research that have treated earnings quality as a multidimensional concept is almost inexistent. In this sense, we document that: (1) Most of the empirical papers on earnings quality deal with just one earnings characteristic, not including the potential effect of the other characteristics related to earnings quality. (2) Some characteristics (particularly, accruals quality and, in a lesser way, conservatism) are widely employed for representing earnings quality, whereas other characteristics (smoothness, persistence) are much less used by researchers. (3) The research on the relationships among the different earnings quality characteristics is scant and with mixed results. (4) Only a few papers develop multidimensional measures of earnings quality, but these measures are based on too restrictive assumptions and there is no evidence of superiority over single-property measures. We complement our bibliometric analysis by discussing the limitations of both the single-dimension approach and the multidimensional approaches used to date, illustrating our arguments with a simulation process. Finally, we conclude that the research on earnings quality can benefit from the use of more sophisticated statistical methods, particularly Structural Equation Modelling, for overcoming these issues