Does the Story Matter? Putting Financial Statement Numbers into Context using XBRL Data

Abstract

We create a measure of “discussed numbers” that captures the narrative context of numbers reported in financial statements. Our algorithm matches words within XBRL tags to sentences in 10-Ks. We show that greater volume of discussed numbers improves analyst ETR forecast accuracy and reduces dispersion. We find that greater discussion of complex tax numbers, including the UTB, deferred taxes, and the valuation allowance, are helpful to analysts in forecasting the ETR. We confirm that the benefits to analyst forecast properties manifest in sentences that explain tabulated numbers in financial statements, and not only in sentences that simply contain or repeat numbers. Additionally, the extent to which tax numbers are discussed is positively associated with GAAP ETR and materiality of tax accounts. Overall, this study enhances our understanding of corporate disclosure practices by identifying and finding that the interrelation of numerical values and narrative discussion significantly improves the information environment.

Keywords: XBRL, Tax, Financial Analyst, ETR Forecast

JEL Classification: M40, M41, D80, D83