The Value of Human Capital for Firm Performance: Roles of Individual and Group Expertise

Abstract

We develop and validate a novel measure of human capital inflow using machine learning techniques applied to online job postings. Our measure captures two key dimensions of firm-level human capital: individual expertise and group joint expertise, the latter reflecting synergies from employee teamwork. We evaluate the predictive power of this measure for future firm performance, specifically two-year-ahead earnings. In out-of-sample tests, our human capital inflow measure outperforms existing human capital proxies in predicting future earnings. We also document systematic cross-sectional variation in the measure’s predictive power, with stronger results for firms characterized by complex tasks and effective employee communication. This variation is consistent across both individual and group expertise components, and predictive power increases when we aggregate three years of human capital inflow. Our findings demonstrate the effectiveness of our measurement approach in capturing meaningful aspects of human capital that are relevant for firm performance prediction.

Keywords: human capital, teamwork, complementarity, machine learning, XGBoost JEL classification: J23, J24, M41, M54