It’s not what you say, but how you say it – Charismatic rhetoric in earnings conference calls

Abstract

Using a commercially available pre-trained neural network for natural language processing, this paper proposes a measure of the extent of charisma in the rhetoric of managers in the question-and-answer (Q&A) part of earnings conference calls. Our empirical results shows that more charismatic communication by managers leads to more favorable stock market reactions, to more positive analyst recommendations, and to higher trading activity. In fact, the effects of charismatic rhetoric are stronger than those of tone measures based on previously established business-specific word lists. Contrary to these business-specific tone measures, charismatic rhetoric does however not truthfully reveal information on the firm’s future performance. This indicates that charismatic rhetoric is only a rhetorical means for impression management. In order to remedy the black box character of the neural network underlying our measurement, we identify phrases typical for (un)charismatic rhetoric in conference call. We use these phrases to construct a fully transparent wordcount-based measure of charisma in rhetoric, which produces similar results as the neural network approach. Overall, our results demonstrate that market participants are highly susceptible to purely rhetorical means.

Keywords: Textual Analysis, Managerial Rhetoric, Managerial Affective States, Stock Market Reactions, Earnings Management