Thursday, February 7, 2013

Global Trade Network

. Thursday, February 7, 2013

This is the network structure of global trade in 2000.* We do not often employ network models to characterize the global economy, so I thought I would offer one that I generated in connection with a book I am writing on the political economy of American hegemony. The visualization depicts the 31 largest trading economies in 2000. As a group, their imports accounted for 91% of total world trade, while trade among these 31 countries accounted for 76% of total world trade. Node size is total imports, and thus gives a sense of relative importance of each national market. Ties are directed, and the weight is import value. One can appreciate the US trade deficit, for instance, by comparing the relative weights of the two ties that connect the US to China (or Japan). I used the Force Atlas 2 algorithm in Gephi to generate the layout and made small adjustments by hand to move overlapping nodes. The algorithm approximates the gravitational pull among nodes based on tie structure and weights. Hence, the clustering present in the figure is an attempt by this algorithm to represent the underlying structure of the network, not a structure I imposed on the data. The underlying data are from the Gleditsch dyadic trade data.

Three patterns seem worth emphasizing.

As we know, trade has structure that is not fully captured by geographic proximity. The cluster at the top of the graph nicely illustrates the regional nature of some EU trade, especially among the original six plus the UK. The cluster at the bottom left, centered on the United States, in contrast, highlights the degree to which trade of the major East Asian economies is oriented around the US economy rather than primarily intra-regional.

Second, countries that we might expect to orient their trade more toward the EU than toward the US, due to geography, seem to orient more toward the US than we would expect. The central European countries, the Scandinavian countries, Ireland, as well as Israel, UAE, and India, appear to be pulled toward the US and away from the EU cluster, in spite of the geographic proximity of the latter. This appears to nicely illustrate the importance of economic mass as a counter to geographic proximity.

Third, overall, the system is strongly hierarchical. The US is central--it trades larger volumes with more partners than any other country. The remaining countries fall into three tiers: large economies that engage in substantial trade with multiple parties, but are (at best) regionally rather than globally central. Germany, for instance, is large and regionally central; Japan and China both are large but not central. Second, medium sized countries who engage in an appreciable amount of trade but are not regionally central. These are, for instance, the small open economies of Scandinavia and Eastern Europe. The fourth tier is the set of countries not represented--the 140 or so for which trade may be important but which as individual economies account for an imperceptible amount of world trade.

Nothing terribly surprising here, perhaps, but one interesting question arises--to what extent will Asia evolve toward a regional cluster that is more independent of the US? Are we likely to observe a future Asia that resembles contemporary Europe, or will future East Asia look like East Asia of 2000? This is obviously the decoupling question--but the network visualization might suggest why decoupling has not occurred to the extent people believed it would. Decoupling requires a fundamental reorientation of trade relationships throughout the region. And if much of this trade is intra-firm, then the underlying production network must also be reoriented. These are substantial changes; and it is not obvious how the global economy treats sunk costs. Are sunk costs sunk, or are important path dependencies at work?

*If you would like an SVG version of the image, leave a comment. 


Kindred Winecoff said...

Obviously I find this interesting, but my question is: What has happened since 2000? Esp w/r/t China -- which has acceded to the WTO since then -- and Europe -- which has admitted many new members and integrated the euro. Nobody was really talking about decoupling in 2000... they were talking about it in 2006.

As I'm sure you know, I personally doubt much has changed in the overall structure. But it'd be nice to see for sure. Is the Gleditsch data available for more recent years?

Thomas Oatley said...

The Gleditsch data I have ends in 2000. Yes, I need to update but it isn't central to what I am doing at present. I will say that the topography changed hardly a bit between 1960 and 2000--stronger ties, but the weighted indegree distribution basically stable.

I don't know that this persists through 2010, but I suspect it does.

Unknown said...

Excellent. May I request a higher quality copy of the image? I'd like to throw it into a lecture next week. Thanks.

Thomas Oatley said...

Hi Justin, thanks for your interest.
svg file at this link:

png file here:

Unknown said...

Brilliant, thanks!

Unknown said...

Thank you so much !
It is very interesting !
May I ask you that would you mind sharing the gephi file for me.

I dont know how to transform trade activities into gephi.

Thank you!!

Global Trade Network




Add to Technorati Favorites