The Listenability of Disclosures and Firms’ Information Environment

Abstract

We introduce a new construct, listenability, defined as the ease with which humans comprehend spoken language, and examine the association between the listenability of corporate vocal disclosures and firms’ information environments. Using a Large Speech Model (LSM) trained on about 680,000 hours of labeled audio data, we analyze audio recordings of 50,280 earnings calls and find that the listenability of managers during conference calls is higher among firms with strong analyst coverage and institutional ownership, but lower when reporting a large earnings surprise. In addition, higher listenability of managers’ presentation is associated with higher-quality interactions with analysts during the Q&A session of the call and higher-quality information production by analysts shortly following the call. Further, higher listenability is associated with stronger price responsiveness in the short window after the call as well as lower bid-ask spreads. These findings suggest clear and effective corporate oral communications facilitate a better information environment.

Keywords: listenability, vocal disclosure, analyst forecast, information environment, artificial intelligence

JEL Classification: G14, G18, M41