Identifying Environmentally - Important Industrial Clusters Using Multi - way Clustering Method

This paper combines input–output analysis with non-negative matrix factorization analysis widely used in the image segmentation and attempts to find the target industrial groups (industrial clusters) in Japan that have intensive CO 2 emissions. We generalized the industrial cluster analysis proposed by Kagawa et al. (2012) to identify environmentally-important industrial clusters from the entire economy. Furthermore, we estimated the optimal number of industry clusters using the Newman–Girvan modularity index. The empirical results obtained using the 2005 Input–Output Tables of Japan show that for example in automobile supply chain, the optimal number of industry clusters is 19, and 4 industry clusters are playing a key role in CO 2 emission reduction. We also found the CO 2 intensive clusters from the supply chains of other commodities and ranked the identified clusters on the basis of their within-cluster effects.