This section features projects that are relatively developed. My research agenda studies financial disclosure through the lens of capital market imperfections. These imperfections have first-order effects on disclosure. Below I highlight three papers that use unique empirical settings for identification and econometric techniques to investigate fundamental questions related to disclosure.
An area of tension exists between extant disclosure theory and empirical evidence of earnings guidance. Empirically, managers strategically guide market expectations to avoid negative earnings surprises. The primary contribution of this paper is to provide a theoretical model of this behavior. We incorporate investors with Prospect Theory (Kahneman and Tversky, 1992) preferences into the classic voluntary disclosure setting of Dye (1985). This refinement is micro-founded by empirical evidence of a discontinuity in returns about earnings that narrowly beat vs. miss expectations. The model characterizes a novel, three-threshold equilibrium where managers voluntarily disclose 'mildly bad' but withhold 'mildly good' news. Managers strategically disclose to avoid investor loss aversion associated with missing earnings expectations. Note that the traditional prediction carries over: managers disclose 'very good' and withhold 'very bad' news.
Do managers attempt to obfuscate weak performance with complex disclosures? A significant challenge in addressing this question is controlling for non-discretionary disclosure complexity driven by the underlying firm and its economic transactions. We examine the “manager obfuscation” hypothesis in the context of homogenous S&P 500 index funds. This allows us to hold non-discretionary complexity (e.g., investments and risks) largely constant in order to examine how funds’ disclosure choices covary with net performance (as measured by expenses or, equivalently, post-expense returns). We have three findings that are relevant to both the mutual fund and corporate disclosure literatures. First, funds with weaker net performance have more complex disclosures, which is compelling evidence of managerial obfuscation. Second, funds obfuscate weak performance by ex-ante creating unnecessarily complex within-fund class structures. This indicates that seemingly non-discretionary firm characteristics may be part of a discretionary obfuscation strategy. Third, we find that funds simultaneously choose both their expenses and complexity, which is a departure from most studies’ assumption that managers choose disclosure complexity to obfuscate non-discretionary poor performance.
Prior research establishes the media as an important information intermediary. In this study, we investigate whether sell-side research analysts efficiently process the qualitative information from media sources. We quantify this qualitative information through sentiment: aggregate tone of textual information. Across both earnings and target price forecasts, analysts are inefficient at incorporating media sentiment. This inefficiency is with respect to sentiment in both news (e.g. business press) and social media (e.g. TwitterTM). On average, analysts overweight aggregate news and social media sentiment, but underweight firm-specific news and social media sentiment. In effect, analysts are inefficiently incorporating both types of information, but in opposite directions. Taken together, this study contributes to our understanding of how analysts process textual information.
Research in Progress
These projects are relatively early-stage. Some started more recently. Some have required two years of intensive data collection. These projects are a continuation of my interdisciplinary focus with projects related to machine learning, psychology, labour markets, and finance. I look forward to seeing these projects to fruition.