Number Processing Constraints and Earnings News

Abstract

Neuroscience shows that human brains are neurologically constrained to process small numbers linearly and large numbers logarithmically, leading to underreactions to larger numbers as their perceived difference becomes smaller. We test this hypothesis in the context of earnings announcements and find that investors respond less in the short term to earnings news for stocks with high earnings per share magnitudes, exacerbating postearnings announcement drift for these stocks. These findings are distinct from and incremental to several risk-based and behavioral explanations, attenuated by robot presence and present in a quasi-experimental design using stock splits. Our evidence suggests that number processing constraints have implications for stock price efficiency. This paper was accepted by Will Cong, finance. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.01722 .