2025 年 55 巻 2 号 p. 259-277
This study proposed an assortative Minimum Spanning Tree （MST） visualization method that represents each country using feature vectors based on industry shares considering both upstreamness and downstreamness. MST was constructed by linking only highly similar country pairs. Regions such as Western Europe and Brazil, Russia, India, Indonesia, Australia, and Turkey （BRIIAT） were treated as categories to analyze whether countries within the same group show significantly higher industrial structure similarity.
Using World Input-Output Tables for 2000 and 2014, we evaluated the visual detectability of assortative components, Global Value Chain （GVC） lengths, anomalies, and their temporal changes. The analysis showed that in 2014, Western, Southern, Northern, and Eastern Europe, as well as East Asia, exhibited significantly larger numbers of assortative links, while in 2000, only Western Europe and East Asia did. This indicates that regional groupings are more similar than cross-regional ones, and that regional homophily has grown over time.
The visualization also revealed a pattern in which countries with standard GVC feature vectors are likely to be positioned centrally, while those with distinct or anomalous ones are periphal. Moreover, most countries’ industrial characteristics were captured using the top three GVC-based feature values.
These findings demonstrate the potential of the assortative MST visualization method for structural comparison, anomaly detection, and temporal analysis in global industrial networks.
JEL Classifications:C1, C5, F2
