The Large - Scale Network of a Tokyo Industrial District : Small - World , Scale - Free , or Depth Hierarchy ?

The large-scale networks of suppliers and buyers in industrial districts have rarely if ever been studied as social networks due to analytical complexity and rarity of datasets. We quantitatively analyzed such a complex system to identify its mechanisms of integration. Tests of the small-world model failed because of a power-law degree distribution, shorter-thanrandom average distances, and lack of local clustering. The scale-free network model was also rejected because primarily hubs organized the network not preferences of suppliers. A directed acyclic graph (DAG) model explained the structural properties. Finally, in lieu of small-world or scale-free models, we offer statistical evidence that the DAG should be a general property for the complex production networks, as modeled by Harrison White.