The Informativeness in Earnings Press Release, A Topic Modeling Approach

Details

Supervisors

Thewissen, James

Faculty

Louvain School of Management

Degree label

Master [120] en sciences de gestion, à finalité spécialisée

Abstract

The objective of this paper is to show the impact of the thematic content of earnings press releases on the probability of companies meeting or beating analysts’ expectations. Using an algorithm that extracts a set of words representing a main topic from a collection of documents, we empirically determined and quantified 6 themes in a sample of 56,071 earnings press releases between 2004 and 2021. We find that the topics produced by this algorithm are semantically meaningful and that this gives us clear indications about the topics discussed in the earnings press of companies. We show in this paper which topics positively or negatively impact the probability of firms to meet or beat expectations. In addition to the meet or beat expectations variable, we focus on earnings manipulation to examine potential similarities in the impact of topic content on these two variables. In addition, we analyze the diversity of topics present in earnings press. We find that firms that meet or beat expectations tend to cover more topics in their earnings press. On the contrary, firms that practice earnings manipulation tend to cover less. Overall, this paper sheds light on the topics covered in earnings press and how investors are sensitive to them.