Network Trade
This dataset explores international trade data through the lenses of Network Analysis, in order to visualize the World Trade Network and describe the topology of the network of world trade. The details of such analysis are provided in a companion paper De Benedictis et al (2013).
The main advantage in using network analysis to explore international trade flows relies on the piece of information that networks provide. The network represents a dyad of counties ij; not the monads i or j, but the relationship between them. However, the specificity of networks is that the relation between i and j is not analyzed in isolation, but it is studied focusing on its structural dimension, that is taking into account the effect of z in the relation between i and j. Extending the effect of others, or in our case the z country effect, to the many z included in the set of possible trade relations, the resulting image is a network in its essence. The implication of this “structural view” is that the relation between i and j cannot be considered independent from the relation between i and z, and between j and z. Therefore the characteristic of interdependence is the hinge of networks.
The file “CEPII_Centrality_measures.dta” contains the Stata 12 dataset with the main network centrality measures described in De Benedictis Luca, Nenci Silvia, Santoni Gianluca, Tajoli Lucia and Vicarelli Claudio (2013), Network Analysis of World Trade using the BACI-CEPII dataset, Cepii wp.
The main advantage in using network analysis to explore international trade flows relies on the piece of information that networks provide. The network represents a dyad of counties ij; not the monads i or j, but the relationship between them. However, the specificity of networks is that the relation between i and j is not analyzed in isolation, but it is studied focusing on its structural dimension, that is taking into account the effect of z in the relation between i and j. Extending the effect of others, or in our case the z country effect, to the many z included in the set of possible trade relations, the resulting image is a network in its essence. The implication of this “structural view” is that the relation between i and j cannot be considered independent from the relation between i and z, and between j and z. Therefore the characteristic of interdependence is the hinge of networks.
The file “CEPII_Centrality_measures.dta” contains the Stata 12 dataset with the main network centrality measures described in De Benedictis Luca, Nenci Silvia, Santoni Gianluca, Tajoli Lucia and Vicarelli Claudio (2013), Network Analysis of World Trade using the BACI-CEPII dataset, Cepii wp.
Méthodologie : une brève description
The file “CEPII_Centrality_measures.csv” contains the comma-separated equivalent of the previously indicated file.The centrality measures are computed yearly for 178 countries covering the period 1995-2010.
In detail the variable included in the csv and Stata file are:
- t: year
- id: country indicator
- i: BACI-CEPII country indicator
- iso3: ISO3 (International Organization for Standardization) country code
- country: country name
- Out-degree/In-degree: measures the number of arcs exiting from a given country (out) / and the number of arcs pointing to a given country (in).
- Out-strength/In-strength: measures the total strength of the arcs exiting from a given country (out) / and the total strength of the arcs pointing to a given country (in), in terns of trade volumes. The measure is normalized by the number of possible trade partners (N-1).
- Out-degree percent/In-degree percent: measures the number of arcs exiting from a given country (out) / and the number of arcs pointing to a given country (in) normalized by the total number on links M in the network.
- Out-strength percent/In-strength percent: measures the total strength of the arcs exiting from a given country (out) / and the total strength of the arcs pointing to a given country (in), in terns of trade volumes. The measure is normalized by total world trade.
- Out-closeness/In-closeness: Closeness centrality It is a measure of how close (in terms of topological distance) a node is with respect to all other nodes. Taking the inverse of the average geodesic distance as a measure of proximity, closeness centrality provides high centrality scores to nodes that are located closer to the set of reachable nodes.
- W-Out-closeness/W-In-closeness: Weighted closeness centrality is measured weighting the unweighted closeness centrality with the average bilateral trade volume in the Dijkstra (1959) algorithm.
- Out-eigenvector/In-eigenvector: The basic idea is that a node's eigenvector centrality is determined by the eigenvector centrality of its neighbors. It is not the country's position In the network that determines the country’s centrality but is the position of the countries linked to him.
- W-Out-eigenvector/W-In-eigenvector: Weighted eigenvector centrality is measured weighting the unweighted eigevector centrality with the average bilateral trade volume in the Dijkstra (1959) algorithm.