Profitability Context and the Cross-Section of Stock Returns

Abstract

Asset pricing models implicitly assume that firm characteristics are context-free. At the same time, companies provide a substantial narrative context that helps investors to put numeric information in perspective. Management may discuss non-quantitative factors that influence performance, such as changes in competitive strategies, future plans, etc. We study the importance of contextual information for asset pricing by focusing on the narrative context surrounding profitability numbers. We use machine learning to incorporate contextual information into the measurement of profitability. Context-adjusted profitability has a superior ability to explain expected returns, both statistically and economically, compared to conventional operating profitability. Further, the context-adjusted profitability factor performs better in portfolio tests and helps to resolve the biggest challenge facing the five-factor model (Fama and French [2015]). Overall, we find that accounting for context adds significant value for investors and can improve the asset pricing models.

Keywords: Contextual information, context-based profitability, asset pricing, machine learning, natural language processing, operating profitability, factor models

JEL Classification: C13, C45, C55, C58, G11, G12, M41