Charles W Swenson

A considerable and growing literature examines the information content of text in financial accounting disclosures. Much of this literature attempts to tie the sentiment / tone words contained in the text of those disclosures to specific outcomes, even though those sentiment and tone words have been largely generated for use in general accounting and finance settings. We add to this literature by using tax-specific and expert-derived tone words to predict the effective tax rates of publicly-traded firms. We find that such specificity and expertise provides more information than simply using the generic terms. In addition, contrary to more generic accounting and finance word-based text analysis, we find that the signs on positive and negative tone words are different and are more consistent with theoretical expectations. We propose that text analysis may be enhanced with the use of context and domain -specific and expert-derived tone terms.