The author’s findings are based on a novel metric that captures every word spoken during a conference call. In contrast, prior research on information production has used such metrics as tone and word counting. The use of tone ignores a large number of the words spoken during conference calls, and word counting overlooks the nuances of the idiosyncratic word choices made by management. Therefore, the author’s conclusions are based on a more holistic analysis of conference call transcripts.

This research can help practitioners understand the determinants of information production in corporate conference calls. One such determinant is the degree of analyst engagement. The results of this research provide empirical support for the role of analysts in information production. The author finds that analyst participation increases information production in corporate conference calls, which is intuitive because one would expect material information to be discussed when analysts are more engaged.

Another determinant is the words used by management in the Q&A session, which are found to improve information production when they are different from the words used during the prepared management presentation session. Because the Q&A session is less scripted, it is not surprising that unique responses by management during the Q&A session improve information production.

The author identifies conference calls in the Fair Disclosure database for firms in the S&P 1500 Index and sorts the call transcripts into managerial sessions and Q&A sessions and then by speaker name, speaker title and position, and speaker comments. He keeps only the transcripts in which at least one manager speaks in both the first and second sessions, yielding 24,399 observations.

Next, the author screens for firms with sufficient data on CRSP, Compustat, and I/B/E/S. He then implements an event study using EVENTUS and a cosine similarity analysis, which is a measure of how similar the words spoken during the presentation are to the words spoken during the Q&A session. The final sample contains 19,309 observations.

The author develops a metric called the information measure, which is equal to 1 minus the cosine similarity variable. Regressions are then performed using three models, with the information measure as the dependent variable and analyst recommendations, dividend payer, abnormal trading volume, and return on assets as control variables. Each of the three models includes one additional independent variable: the CEO’s participation in the Q&A session, analysts’ participation in the Q&A, or other managers’ participation in the Q&A. The author concludes that analyst participation in the Q&A session increases information production and that the participation of others does not.

Corporate conference calls are an opportunity for investors to get more information on the content of corporate press releases. Some of these conference calls are lengthy and cover multiple topics. Therefore, navigating the maze of information to extract the valuable pieces may be challenging. The author has developed a novel metric that helps identify what types of information are important. Although the results are not surprising, the metric developed by the author helps validate quantitatively the types of activities during conference calls that produce more information. For example, investors know that regurgitated or stale information does not add value, so repeating the same information multiple times does not increase the amount of information investors should trade on.

In addition, the author’s results confirm the important role analysts play in information production. Analysts’ time is valuable, so their participation in the Q&A session provides a signal on topics that are relevant to investors. Analysts then turn around and disseminate such information to their clients through published reports. Conversely, a low level of analyst engagement acts as a signal that the content of the corporate conference call may not be valuable.