Analysis of Agri-Food Trade Structures
To Promote Agri-Food Trade Networks
In the Islamic Countries
20
2.3. Networks in World Agricultural Trade
The previous subsections have focused on analyzing trade flows in the traditional ways, looking
at sectoral and geographical breakdowns, and identifying the directions of trade at an aggregate
level. An alternative way of looking at the data is through concepts drawn from network science.
As set out in Section 1, the analysis applies key findings from the applied mathematics literature
on networks to examine some key characteristics of agri-food trade networks at the global level.
The analysis proceeds at the title level of the Annex 1 classification, because the metrics are best
adapted to tradematrices that are relatively dense, and the number of zeros in the bilateral trade
matrix increases rapidly at higher levels of disaggregation.11
There is no unique visual representation of network data, but it is possible to use a standard
technique, the prefuse force directed algorithm, to produce images that summarize the relevant
data. The full trade network is made up of thousands of bilateral connections, and is impractical
to present visually: the mass of individual lines in the end prevents any meaningful
interpretation. The analysis therefore and analyzes the network using only each country’s
largest export flow.12 Countries are represented by the three letter ISO codes, which are
reproduced for reference in Annex 4.
Figure 9 looks at the global network of trade in agri-food products. Blue lines indicate trading
relationships between World Bank regional groupings, while red lines indicate trading
relationships within World Bank regional groupings. Arrowheads indicate the direction of trade
flows. The figure makes clear that there is a strong regional dimension to trade in agri-food
products, with most countries having their largest trade link with a regional partner (i.e., red
lines predominate). This finding is consistent with standard models of trade in which bilateral
trade costs are in part determined by the geographical distance between countries, which means
that more distant countries tend to trade less, keeping other factors constant. However, there
are important cases in which extra-regional trade linkages are important as well, so the finding
is by no means a cut and dried one. Rather than presenting a single integrated global network,
the figure makes clear that there are subnetworks: a primarily European one, centered on
Germany, a primarily Asian one centered on China and India, an American one centered on the
United States, and a smaller West Asian network involving Turkey, the UAE, and Saudi Arabia.
Each network also has extra-regional members, so trade clearly crosses regional boundaries in
this sector, but the salience of the sub-networks is an important feature to note from the
perspective of future trade development opportunities.
11 Haveman, J., and D. Hummels. (2004). “Alternative Hypotheses and the Volume of Trade: The Gravity Equation and the
Extent of Specialization.” Canadian Journal of Economics, 37(1): 199-218.
12 Shepherd, B., and L. Archanskaia. (2014). “Evaluation of Value Chain Connectedness in the APEC Region.” Report Prepared
for the APEC Policy Support Unit. Singapore: APEC.