IPSR Mini grant awarded to Paul Gouvard
IPSR has awarded $1,000 to support Paul Gouvard’s research on unexpected exchanges which occur during quarterly earnings calls. Paul is a visiting PhD student at BerkeleyHaas working with Sameer Srivastava to incorporate computational linguistics, natural language processing, and machine learning into this independent research project.
Quarterly earnings calls are highly scripted and ceremonial events where substantial information is rarely disclosed. However, sometimes events go off-script – an unexpected question is asked, a manager makes an unplanned revelation – and it is then that substantial information on the firm may be revealed. Offscript events have been identified in a dataset of earnings call transcripts utilizing natural language processing. Based on this analysis, the degree to which a given call has gone astray is correlated with 1) the degree to which the call went astray, and 2) earnings surprise for the next quarter after the call.
Gouvard and Srivastava hypothesize that the content of off-script events during a given call will impact the direction of the earnings surprise in the next call, either upward if analysts and managers remained aligned in the type of topical content discussed during the off-script event, or downward if they did not. They believe the degree to which a call went astray impacts the magnitude of this effect. Beyond creating a measure of ‘off-scriptness’, they hope to demonstrate that off-script events, generally seen as a negative thing, might actually have a positive impact on earnings surprise under certain conditions.