Do Images Provide Relevant Information to Investors? An Exploratory Study

Abstract

We introduce the concept of “visual-readability” in annual reports and use novel machine learning
algorithms to construct visual-readability metrics. We innovate by creating a novel measure of content reinforcement, representing the information content investors can extract from images, complementing and reinforcing particulars contained in the textual narrative. An increase in visual prevalence and in the degree to which images convey reinforcing information is associated with greater (lower) analyst forecast accuracy (disagreement) in subsequent quarters. Effects of visual readability are distinct from those of textual readability. Using Kelly and Ljungqvist (2012)’s identification, we find that firms increase the use of visuals when facing an exogenous drop in analyst coverage. Our metrics are further associated with lower risk, lower cost-of-equity, and higher credit ratings during the subsequent year. In the age of information overflow, our results highlight the importance of visual readability for information assimilation.

Keywords: Visual readability, annual reports, images, information dissemination, information reinforcement, textual readability, analyst forecast accuracy, analyst disagreement

JEL Classification: D83, G12, G14, M41