Industries as in a Network: Micro Evidence from Job Search
Abstract: In this paper, we propose the concept of cross-industry skills (CRISs) as the labor market linkages across industries. These skills are productive to multiple industries, and therefore, they affect labor mobility intensity across industries. Using online job board data, we first employ machine learning to investigate industries as a network. We then construct worker-job skill match index as the measure of CRISs of worker-job pair in applications. We further apply it to empirically study the effect of CRISs in job search. We find that the closer the industries are in the industry network, the more intensive CRISs they require. Additionally, we find that a one standard deviation increase in CRISs is associated with 0.51 percentage points increase in callback probabilities across occupations. The economic magnitude is 1.16 times of that of over-experienced and 2.68 times of that of over-education. We also find the important role of CRISs in explaining earnings gaps. Finally, our approach is used to predict various ripple effects of an industry-specific labor demand shock.