Data, Markups, and Asset Prices

Abstract

This paper investigates how data technology affects firms' market power and asset prices. Using a novel dataset tracking firms' employment of data scientists, we document three key empirical findings: firms with higher proportions of data scientists exhibit larger markups, have higher information quality proxied by lower sales forecast errors, and earn higher stock returns. Specifically, a long-short portfolio strategy based on firms' data scientist ratios generates significant annual excess returns of approximately 4%. To quantitatively rationalize these empirical findings, we develop a heterogeneous firm model in which firms optimally hire data scientists to learn about unobserved consumer tastes. The model demonstrates how data enables firms to improve demand forecasting accuracy and extract higher markups. Importantly, supply-constrained firms have stronger incentives to hire data scientists, leading to countercyclical data scientist hiring that amplifies their exposures to aggregate risk through an operating leverage channel. We provide empirical evidence supporting our model mechanism.

Keywords: E2, E3, G12