An investigation of interregional trade network structures

We provide empirical evidence on the network structure of trade flows between European regions and discuss the theoretical underpinning of such a structure. First, we analyze EU regional trade data using Social Network Analysis. We describe the topology of this network and compute local and global centrality measures. Finally, we consider the distribution of higher order statistics, through the analysis of local clustering and main triadic structures in the triad census of interregional trade flows. In the theoretical part, we explore the relationship between trade costs and trade links. As shown by Behrens (2004, 2005a, 2005b) in a two-region linear new economic geography (NEG) model, trade costs and the local market size determine, even with finite trade costs, unconditional autarky and unilateral trade, that is, a one-directional flow from one region to the other. Following these contributions and guided by the empirical evidence, we clarify the relationship between market competition, trade costs and the patterns of trade in a three-region NEG model. We identify a larger set of trade network configurations other the three elementary ones that occur at the dyadic level between two regions (no trade, one-way trade, reciprocated two-way trade), and relate the model with the triad census.