Journal Issue

2. Centrality in the Global and Regional Trade Network

Céline Allard, Jorge Canales Kriljenko, Jesus Gonzalez-Garcia, Emmanouil Kitsios, Juan Trevino, and Wenjie Chen
Published Date:
March 2016
  • ShareShare
Show Summary Details

Sub-Saharan Africa’s overall progress in trade integration, however, needs to be put in perspective with developments in global trade over the same period. Global trade took off following the implementation of the Uruguay Round, the creation of the World Trade Organization (WTO) in 1995, and China’s entry in the WTO in 2001. This rapid expansion was characterized by the emergence of new trade giants and the decline in advanced economies’ contribution to world trade. In fact, it is only to the extent that sub-Saharan Africa was able to redirect trade toward these new trade players, particularly China, that it managed to keep its place in world trade—a place that nonetheless remains small in the global scene. As a simple illustration of this fact, export ratios at the global level rose by about as much as in sub-Saharan Africa, from 17 percent of GDP in 1995 to 25 percent of GDP in 2013 (versus 20½ percent and 27½ percent of GDP in the region, respectively).

A more granular measure of the region’s integration in global trade—its centrality in the global trade network—paints a similar picture. This measure takes into account not only the size of exports for a given country but also the number of its trade partners, as well as the relative weight of these trade partners in global trade, therefore better capturing the country’s interconnectedness within the web of global trade (De Benedictis and others 2014).

The centrality measure is estimated here using the PageRank algorithm created by Brin and Page (1998). The centrality score of each exporting country is computed as the probability that the country is chosen in the trading network by an importing country. The PageRank algorithm treats the trading network as a stationary distribution of a random walk defined on a set of trading countries.

The random walk tracks a randomly chosen importer with probability p at each step, and with probability 1 – p, the walk starts afresh from an exporter chosen at random according to a uniform distribution. More specifically, the centrality values satisfy the following equation:

where C(i) is the centrality of country i, dj is the number of countries that country i is exporting to, and p represents the probability of exporting to a country when abandoning the random walk. The sum in the first term on the right-hand side of the equation above is taken over all those countries j that are importing from country i. The centrality is then calculated using an iterative algorithm that corresponds to the principal eigenvector of the normalized bilateral trade matrix. By construction, the sum of C(i) for all countries considered in the network is 1.

By that measure, sub-Saharan Africa remains the least integrated region in the world, with an average centrality of only about half of that observed in other emerging and developing economies (Figure 6). Of course, this partly reflects a relatively lower level of development than in other regions. But even South Africa, the most interconnected and one of the highest-income countries in the region, has a relative position that is substantially lower than other emerging markets such as Brazil or Mexico. And outside of Angola and Nigeria—where the large role of oil exports has led to an increase in centrality—sub-Saharan Africa’s most globally integrated members have only maintained their relative foothold in the global trade network between 2000 and 2013. By contrast, countries such as China, India, Poland, Turkey, and Vietnam saw their relative centrality score double over the same period. All in all, this points to substantial potential for a larger role of trade in sub-Saharan African economies.

Figure 6.World Centrality per Region, 2000–131

Source: Staff calculations based on data from IMF, Direction of Trade Statistics.

Note: Only emerging market and developing countries with 2013 GDP per capita below US$ 20,000 from each region are considered. China is excluded from the Asia group as its centrality measure is about 28 times higher than the average for that region and five times the second largest centrality measure, corresponding to India.

1 This measure is the PageRank centrality and takes into account the size of exports for any given country, the number of its trade partners, and the relative weight of these partners in global trade (see Brin and Page 1998 for a description of the computation).

One bright spot has been the increase in regional trade. As mentioned earlier, the share of regional trade almost doubled over the past 20 years, although from a low base of 2 percent of GDP to 3½ percent of GDP. Measuring centrality at the intraregional level reveals the emergence of trade subregions, with hubs such as Côte d’Ivoire, Nigeria, and, to a lesser extent, Senegal in West Africa, Kenya in East Africa, and South Africa in the southern part of the region (Figure 7).

Figure 7.Regional Centrality, 2013

Source: IMF staff calculation based on data from IMF, Direction of Trade Statistics.

    Other Resources Citing This Publication