Do Visuals Crowd Out Attention to Text? Evidence from Earnings Calls
Abstract:
Visual is now ubiquitous in modern business communication. Using GPT-4o, we systematically identify its presence in more than one million slide pages from earnings calls of U.S. public firms over the last 15-year. We show that it diverts investor attention away from less relevant textual content, whose sentiment is significantly associated with future stock returns. A long–short strategy based on this sentiment earns an annualized alpha of approximately 14.6%. This delayed pricing effect is stronger when investor attention is more constrained on earnings call days. Mechanism tests suggest that visual aids capture investor attention during earnings calls via a top-down (goal-driven) attentional pathway, rather than a bottom-up (stimulus-driven) process. Last, although we find that managers are more likely to visualize favorable news, we do not find evidence that this choice varies systematically with firm performance. Overall, our findings suggest that, while prior research emphasizes the attention-grabbing role of visuals, they can also crowd out attention to other information within the disclosure, leading to delayed pricing.
Contact Emails:
zlynne@ceibs.edu