Human Readability of Disclosures in a Machine-Readable World

Abstract

While regulators emphasize the need for machine-readable corporate disclosures, we examine how improvements in machine readability of textual and numerical information affect the human readability of these disclosures. Relative to the 2009 XBRL mandate that required a separate XBRL exhibit of financial statement numbers and footnotes, the 2019 Inline XBRL (iXBRL) regulation improves the machine readability of both textual and numerical content throughout corporate filings. Utilizing the iXBRL mandate as a quasi-exogenous shock to machine readability, we observe a negative effect of machine readability on human readability. In addition, we document that following the iXBRL regulation, disclosures become less informative to retail investors, who generally have less ability to process corporate disclosures with machines and who are more reliant on human readability, and that they reduce ownership in stocks impacted by the iXBRL regulation. Further evidence suggests the reduction in human readability is driven by both lower incentive to allocate effort toward making disclosures human-readable and reduced attention to human readability. Our results are robust to a regression discontinuity design and an alternative difference-in-differences design. Overall, our findings indicate that improved machine readability has implications for the human processing of disclosures.

Keywords: Machine readability, human readability, retail investor, capital market consequences

JEL Classification: G14, G18, M41