Paper forthcoming at European Accoutning Review – https://www.tandfonline.com/doi/full/10.1080/09638180.2024.2410892
Kristian D. Alleea*, Chuong Dob, and Huy N. Doc
aWalton College of Business, University of Arkansas, Fayetteville, USA; bCollege of Business, University of Nevada – Reno, Reno, USA; cCollins College of Business, The University of Tulsa, Tulsa, USA
We examine whether analysts can gather useful information based on their public interactions with firm managers. While analysts’ forecasts and price targets are often a “black box,” earnings conference calls offer a unique opportunity to see their information gathering processes through the questions they ask. We design a measure to capture the “uniqueness” of a question by comparing how similar it is to other conference call content or questions. Specifically, we use cosine similarity to compare an individual analyst’s question(s) in a firm’s quarterly earnings conference call against three benchmarks: 1) questions from the other analysts on the same call, 2) the same analyst’s questions on the same firm’s prior calls, and 3) the management-prepared narrative on the same call. The underlying intuition of our method is that the conformity of a question on the call, to one of those three reference points, reveals what information is desired and how the questioner intends to extract the information. Through analyses built around this uniqueness measure, we aim to enhance our understanding of how analysts publicly gather information and the impact of these information gathering activities.
Our results show that the uniqueness of individual analysts’ questions varies with their individual characteristics, controlling for firm and call characteristics. Analysts with superior private information ask more unique questions on the call. Conversely, more experienced analysts, analysts from larger brokers, and analysts who forecast more frequently ask less unique questions. Turning to the consequences of analysts’ information gathering, the results show that unique questions increase the overall number of topics (as measured by Latent Dirichlet Allocation topic modeling) of analysts’ questions as well as managers’ answers in an earnings call. Thus, analysts’ unique questions can elicit new and different topics of discussion. Furthermore, when analysts ask more unique questions, their likelihood of revising their forecasts after the call and the magnitude of the revision increase with the size of the earnings surprise. Finally, the accuracy of an analyst’s forecast appears to be associated with the uniqueness of their questions on conference calls, again conditioned on the size of the earnings surprise.
We next examine how the market responds to the uniqueness of analysts’ questions in conference calls. First, using “real-time,” millisecond trading data from conference calls occurring during market hours, we find that the Q&A session on an earnings conference call is more informative in the presence of unique analyst questions. This result suggests that analysts’ information gathering activities on conference calls, as captured through the uniqueness of their questions, yields value relevant information for both analysts and investors. Second, controlling for the size of the revision, investors appear to react more to forecasts made by analysts who ask more unique questions on the preceding conference call. This result supports the idea that analysts’ questions can serve as public signals of their information gathering capabilities to the market, leading to a larger market reaction to their revised forecasts.
Our study extends the literature on analysts’ behavior by shifting the focus from their forecast and reports to their information gathering behaviors, specifically their (unique) questions on conference calls, as an indication as to whether/if this behavior will then influence their earnings forecast revisions. We also provide insights into analysts’ forecasting behavior based on earnings conference calls and the market’s use of analysts’ information in interpreting earnings information. Our paper documents the significant nuance in the interactions between managers and analysts, as well as within analysts, and there is an association between analysts’ interactions and analysts’ forecasting behavior as well as the market’s reaction. Additionally, our study contributes to the literature that examines earnings conference calls at the conversation level.
