G-Money: Earning Calls Predict Stock Performance

Abstract

Efficient market theory denies the possibility of being able to profit using public information. Yet Wall Street practitioners devote considerable resources to doing such a thing. Brokerage houses and banks invest heavily in stock analysts who are paid to research firms in order to generate superior returns for their clients and studies have validated their claims. This provides motivation for identifying a link between analyst information acquisition and the updating of their beliefs. We use conference call transcripts, a key information gathering activity, as a proxy for the the belief updates of these analysts. Researchers have documented that this real-time setting is crucial to increasing the quality of analyst forecasts and recommendations, thus analysts must be updating their beliefs using soft information obtained during these calls. Studying conference call transcripts, while beneficial in its own right because it enhances the understanding of using text data for quantitative prediction, is also motivated by the theory that stock valuations are conditional on the beliefs of informed investors. Prior literature already shows that institutions or individual traders (or both) trade on analyst beliefs, affecting the stock price . This study’s main contribution to the literature is that the beliefs of informed investors can be learned during earnings conference calls through the use of sentiment, directly linking analyst beliefs to information acquisition and this can be used to explain variations in stock price performance not accountable by current asset pricing theory.

Download PDF