This paper investigates how institutional investor distraction affects the assimilation of narrative content in the MD&A section of the 10-K filing. We introduce the Aggregate Attribute Index (AAI) and an alternative formulation (AltAAI), which capture linguistic features beyond tone to provide a broader measure of corporate narrative richness. Using machine learning and natural language processing, we analyze U.S. firms that follow a staggered reporting strategy, releasing quantitative results before full narrative disclosures. This design isolates the incremental effects of complex language when other portfolio events distract investors. We find that narrative complexity does not trigger short-term return responses but significantly affects stock prices over longer horizons. Complexity moderates how and when attention-constrained investors adjust prices. These effects are not captured by dictionary-based tone or readability metrics, underscoring the distinct role of multi-dimensional attributes in shaping delayed market reactions and price discovery

We consider how images (i.e., photos but not graphs, charts, or infographics) in annual reports provide users with information and use machine-learning algorithms to assess their informativeness. We develop a metric of content reinforcement, defined as the degree to which information investors extract from images complements and reinforces details in textual narratives. We find that firms are more likely to use images when they experience greater asset growth, have greater business complexity, and provide less readable textual disclosures—suggesting images are used more often when information processing costs are high. Our main results indicate that increases in visual prevalence and the extent to which images reinforce text are associated with greater analyst forecast accuracy and lower dispersion, suggesting that images improve users’ information processing. Firms also increase image use after an exogenous decline in analyst coverage. Overall, firms use images when their information environment is poorer, and visual informativeness facilitates information assimilation

We examine whether information processing constraints limit managers’ ability to effectively integrate tax planning and core business strategies (i.e., effective tax planning). We propose that artificial intelligence (AI) tools, such as machine learning, can mitigate these constraints by providing enhanced predictive information for key business decisions (e.g., customer demand, supply chain), thereby reducing processing costs. Using a recently developed firm-year measure of investment in AI-related human capital for a broad sample of U.S. nontechnology firms between 2010 and 2018, we find that AI investment is positively associated with tax effectiveness. This effect is concentrated among more complex firms and those where the tax function holds a higher status. Consistent with AI reducing information processing costs, we find that it improves tax effectiveness by enhancing internal information quality and internal capital management. We provide novel evidence that processing constraints hinder effective tax planning and show that AI can mitigate these constraints

This study examines the investment value of information provided by crypto-influencers, that is, social media influencers covering crypto assets on Twitter. We examine the returns associated with approximately 36,000 tweets issued by 180 of the most prominent crypto social media influencers covering over 1,600 crypto assets for the two years spanning through December 2022. Our primary results indicate that crypto-influencers’ tweets are initially associated with positive returns. However, these tweets are followed by significant negative longer-horizon returns, suggesting they generate minimal long-term investment value. These effects are most pronounced for tweets issued by crypto-influencers proclaiming to be crypto experts, for smaller cap crypto asset securities and for self-described experts with many Twitter followers. In an additional analysis, we use machine-learning methods to classify tweets and find that this pattern of results strengthens when the tweets have a more positive sentiment or relate to buy recommendations