Data Visualization in Earnings Conference Calls

Abstract

As data visualization gains popularity in modern society, it is important to understand how this aspect of presentation affects accounting disclosures. By applying deep learning to create customized image recognition algorithms, I investigate firms’ use of data visualization in earnings conference calls. I hypothesize and find that firms tend to use more data visualization when information demand is higher, financial statement processing costs are higher, and operating performance is better. In addition, data visualization is positively associated with analysts’ information acquisition and earnings call informativeness. Overall, my results provide support for dual-coding theory, which predicts that in a situation where verbal stimuli dominate, an increase of imagery stimuli will enhance the audience’s memory and comprehension of the message.