The Design of Industry

We study the division of labor within an industry, formulate it as a generalized make-or-buy problem, and characterize the optimal allocation of work as that minimizing the sum of adjustment-costs within and between firms. Using a unique dataset on eight segments of the global automobile industry, we test the theory from several angles: We first show that any two tasks are more likely to be performed by the same firm if mutual adjustments between them are needed on a sufficiently frequent basis. To take indirect effects into account, we then look at the entire industry and find that a disproportionate number of adjustments are managed inside firms. We finally use simulated GMM to estimate a structural model in which industry design is portrayed as the solution to an integer program aimed at minimizing industry-wide adjustment-costs. The program is extremely complex and while there is a significant heterogeneity in the eight actual designs, they all fit the model very well. The main substantive contribution of the paper is thus to present robust evidence consistent with the view that the firm is a low variable, but high fixed cost way to govern adjustments. A more methodological contribution is to introduce industry-level estimates and show that they outperform the firm-level estimates used in other studies of make-or-buy decisions