Future Earnings Information in Business Text
Keywords
Textual analysis, Bag of words, Sentiment analysis, Information in business text
Abstract
We show that measures of business text sentiment are highly affected by transitory events. These transitory events lead to a strong positive association between sentiment and contemporaneous earnings. However, the association between sentiment and future earnings is insignificant. We develop a quantification of business text that is less affected by transitory events and is therefore better suited for prediction. Applied to 10-K filings, our proposed measure is strongly associated with both contemporaneous and future earnings. Further, applied to earnings announcements, our measure is strongly associated with future returns. Other tests verify and confirm the value of our measurement approach.
Original Publication Citation
Green, Jeffrey and Green, Jeremiah and Kartapanis, Antonis, Future Earnings Information in Business Text (January 8, 2021). Available at SSRN: https://ssrn.com/abstract=3762794 or http://dx.doi.org/10.2139/ssrn.3762794
BYU ScholarsArchive Citation
Green, Jeffrey Jack; Green, Jeremiah; and Kartapanis, Antonis, "Future Earnings Information in Business Text" (2021). Faculty Publications. 6143.
https://scholarsarchive.byu.edu/facpub/6143
Document Type
Peer-Reviewed Article
Publication Date
2021-03-12
Permanent URL
http://hdl.lib.byu.edu/1877/8872
Publisher
Elsevier
Language
English
College
Humanities
Department
Linguistics
Copyright Use Information
https://lib.byu.edu/about/copyright/