The predictive ability of corporate narrative disclosures
Emerald Group Publishing Limited
Faculty of Business and Law
School of Accounting, Finance and Economics
Purpose – The purpose of this paper is to investigate the relationship between narrative disclosures and corporate performance based on Australian evidence. In particular it builds a model which discriminates between good and poor performing companies based on their corporate narratives. Design/methodology/approach – A sample of Australian manufacturing companies is classified into two groups based on earnings per share (EPS) movement between 2008 and 2009. A content analysis of their discretionary narrative disclosures is used to classify and predict group membership. Findings – This study finds that the word-based variables based on discretionary disclosures are significantly correlated with corporate performance. Word-based variables can successfully classify companies between “good” performers and “poor” performers with an accuracy of 86 percent. Research limitations/implications – The relatively small sample size, for Australian manufacturing companies, limits both the predictive ability of the model and its generalisability elsewhere. Practical implications – The findings of the paper demonstrate that certain keywords, notably the use of “high/highest” and “dividends” are significantly and positively associated with superior performance. Originality/value – The study builds a classification model for continuing Australian companies, whereas prior research focuses on UK and US companies and is based on a healthy/failed distinction.