Date of Graduation

1980

Document Type

Dissertation/Thesis

Abstract

The foundation of interdependence among industries is the system of interindustry linkages. An interindustry linkage is the operational connection of flows of material and information among the industrial units of an economic system. By analyzing linkages, economists are able to work out impacts of changes in industrial activity on an economy. Entries of input-output tables provide measurements of inter-industry linkages. The rows of a table represent forward linkages, or connections further along the production process. The columns represent backward linkages, or connections with input suppliers. In recent years, regional and urban analysts have focused on clusters of industries for explaining economic processes. A cluster is a group of industries with strong intragroup linkages but weak ties otherwise. Clusters can be generated from input-output tables by using methods that group industries according to strengths in their interindustry linkages. Clustering methods developed to date have a diverse mathematical background. They are based on matrix algebra, graph theory, and multivariate analysis. The purpose of this research was to analyze critically five clustering methods as applied to input-output tables. The methods were triangularization, graph theory, network analysis, and two multivariate methods. The hypothesis under test was that the five methods will yield similar results when applied to the same input-output table. Two input-output tables were used in this test. One was an illustrative transactions table, while the other was the 1975 West Virginia transactions table. To differentiate between clusters generated by each method, a chi-square contingency test was used. The results of applying the five clustering methods to these tables showed that the methods do not produce similar results.

Download

Share

COinS