Abstract. Economists use the term “human capital” to represent everything that an individual brings to a productive process: their skills, abilities, experience, and knowledge. Traditionally, human capital has been modeled as a simple, one dimensional measure. This model was intended to capture the value of labor in manufacturing, and it largely makes sense in that context, where workers are valued according to their speed, education, or tenure. However, that simple model does not translate well when it comes to the production of knowledge. When people solve problems, they often use their skills in combination, and thus a person's contribution to knowledge production may be greater than the sum of their individual skills. Moreover, the value of an additional skill might depend on the skills they already have. A theoretical model of skill recombination bears this out: even when skills are assigned at random, the relationship between skill sets and outcomes is highly non-linear, and small differences in skill sets are exaggerated to create large differences in outcomes. In the context of academic collaboration, this creates superstars. In the labor force, it generates dramatic wage inequality.
Skill networks provide a method of summarizing the interactions between skills in a given labor market or knowledge-generating community. Nodes in the network are skills, and two skills are connected if an individual has both. In an academic community, the dynamics of this network reflect the evolution of knowledge over time. In a labor market, it provides a picture of the supply and demand for different skill combinations. On an individual level, the distribution of a person's skills across the network provides measures of skill specialization, diversity, and coherence–aspects of human capital that cannot be captured by simple, one-dimensional measures.