An investigation into earnings per share disclosures in South Africa.

Abstract

This dissertation examines Earning per share (EPS) as a disclosure requirement for listed companies by investigating firstly, EPS disclosures in annual reports of certain selected JSE listed companies and secondly, the attitudes of the preparers of those annual reports to a number of issues relating to EPS. The three mandatory EPS disclosures - Basic EPS, Diluted EPS and Headline EPS - are discussed with a view to determining their information content and reporting framework. This study also considers whether cash based measures of performance are better than earnings based measures. Due to the reliance placed on reported EPS numbers this study attempts, by an examination of annual reports, to provide evidence as to whether or not South African companies are correctly calculating and disclosing the various EPS measures. By means of a questionnaire survey into the attitudes of the preparers of annual reports, this study also attempts to provide evidence as to the importance of the EPS measures as well as the preparers' perceptions on the appropriateness of the Headline earnings definition. The annual report survey into EPS disclosures revealed that South African companies are correctly calculating and disclosing Basic EPS. Even-though all companies correctly calculate Diluted EPS, most companies do not properly disclose Diluted EPS information. As far as Headline EPS is concerned, the annual report survey revealed that many South African companies make disallowed Headline earnings adjustments with most offenders disclosing higher Headline EPS numbers as a result. The survey into the attitudes of preparers of company reports towards various matters concerning EPS revealed that preparers of annual reports consider Headline EPS to be the most important earnings based measure of performance and the adopted Headline earnings definition as being appropriate. It is therefore important that companies calculate and disclose Headline EPS correctly.