Measuring Qualitative Information in Capital Markets Research
Stevens Institute of Technology
Andrew J. Leone
University of Miami
A growing stream of research in accounting and finance tests the extent to which the tone of financial disclosure narrative affects security prices, over and above the disclosed financial performance. These studies measure tone by counting the relative frequency of positive versus negative words in a given disclosure (e.g., earnings press releases). Critical to the analysis is the list of words deemed to be positive or negative. Most studies use general wordlists (GI or Diction) rather than wordlists that are specific to the domain of financial disclosure. General wordlists likely omit words that would be considered positive or negative in the context of financial disclosure and include words that would not. Application of general wordlists to financial disclosure also gives rise to problems with polysemy. For example, the word 'division' is considered a negative word in the GI wordlist, but that word is commonly used in financial disclosure to describe a segment of a company and is thus neither negative nor positive in a domain-specific context. In this study, we compare the predictive validity of these commonly-used wordlists to a wordlist developed specifically for the context of financial disclosure. Using a sample of over 15,000 earnings press releases, we find that the context-specific wordlist developed by Henry (2006, 2008) is more powerful than the general wordlists used in past research. Our findings suggest that capital markets researchers will benefit by using the domain-specific wordlist in the context of financial disclosure. These results will help to establish a firm foundation for research on qualitative information in financial disclosure.
Number of Pages in PDF File: 39
Date posted: September 9, 2009
© 2016 Social Science Electronic Publishing, Inc. All Rights Reserved.
This page was processed by apollobot1 in 0.266 seconds