The Evolution of the World Trade Web

This paper empirically studies the statistical properties of the world trade web (WTW) and its evolution over time using a weighted network approach. Previous works have characterized the WTW as a binary network, where countries play the role of nodes and a link is in place between any two countries if there exists a sufficiently large amount of trade between them. On the contrary, we exploit the heterogeneity of trade relationships and weight each existing link by some measure of the actual amount of trade carried through that link. Our results indicate that the WTW is a strongly symmetric network, where the majority of trade relationships (and their intensities) are reciprocated. We also find that: (i) the majority of countries hold many weak trade relationships and coexist with a few countries holding less but more intense export relationships; (ii) countries that hold more intense trade relationships preferably trade with poorly-connected countries, but are typically more clustered; (iii) rich countries tend to form more intense trade links and to be more clustered. Furthermore, all structural properties of the WTW display a remarkable stationarity across years. From a methodological point of view, our paper suggests that a weighted network approach is able to provide more precise conclusions than a binary analysis. Many implications that are indeed valid in binary networks are reversed in our weighted analysis. Finally, we show that all our main results are robust to alternative weighting procedures.