We analyze input-output matrices for a wide set of countries as weighted directed networks. These graphs contain only 47 nodes, but they are almost fully connected and many have nodes with strong self-loops. We apply two measures: random walk centrality and one based on count-betweenness. Our findings are intuitive. For example, in Luxembourg the most central sector is "Finance and Insurance" and the analog in Germany is "Wholesale and Retail Trade" or "Motor Vehicles", according to the measure. Rankings of sectoral centrality vary by country. Some sectors are often highly central, while others never are. Hierarchical clustering reveals geographical proximity and similar development status.     «

We analyze input-output matrices for a wide set of countries as weighted directed networks. These graphs contain only 47 nodes, but they are almost fully connected and many have nodes with strong self-loops. We apply two measures: random walk centrality and one based on count-betweenness. Our findings are intuitive. For example, in Luxembourg the most central sector is "Finance and Insurance" and the analog in Germany is "Wholesale and Retail Trade" or "Motor Vehicles", according to the measure....     »