Abstract
The paper studies the relationship between human capital spillovers and productivity using a unique longitudinal matched employer–employee dataset of Israeli manufacturing plants that contains individual records on all plant employees. I focus on the within-plant diversity of employees’ higher-education diplomas (university degrees). The variance decomposition shows that most knowledge diversity takes place within the industries. Using a semi-parametric approach, the study finds that hiring workers who are diversified in their specific knowledge is beneficial for plants’ productivity—the knowledge diversity elasticity is about 0.2–0.25 and is robust—and that the benefit of knowledge diversity increase with the size of the plant. This suggests that for each allocation of labor in the production process it is beneficial for plants to diversify their skilled labor. The findings also suggest that the conventional way of estimating plant-level production function using Ordinary Least Squares or Fixed-Effects method is biased upward due to simultaneity of the inputs and the unobserved productivity shock.