Home > Global Futures for Agriculture, International trade, Potatoes, Simulation modeling > Really a nontraded commodity? A look at the international potato trade network

Really a nontraded commodity? A look at the international potato trade network

by Ulrich Kleinwechter and Victor Suarez.

At times it is said that the potato is a largely nontraded crop which is produced, processed, traded and consumed mostly locally (CGIAR, 2011) and which is less affected by international price movements (FAO, 2008). In contrast to this view, the IMPACT model used, among others, in the Global Futures Project incorporates the assumption of a single world market for this commodity which is characterized by perfect price transmission and in which all country level prices are determined by a single world market price. Understandable that we were somewhat unhappy with this assumption…

So, we decided to dig a bit deeper into the problem. The first step was to obtain a global perspective of international potato trade. How are potatoes traded internationally? How do international trade flows look like? What is the structure of international potato trade?


We choose a network approach for this undertaking, borrowing tools from social network analysis (SNA). SNA has been frequently used to analyze international trade networks (Fagiolo et al., 2009; Garlaschelli & Loffredo, 2005; Serrano & Boguñá, 2003). To our knowledge, however, it has not been applied to trade networks of single commodities.

We start out with constructing bilateral trade matrices are constructed from trade data provided by the FAOSTAT trade database (FAO, 2012) and the COMTRADE database (UN, 2012). To be able to distinguish between the three most important types of potato products separately and to detect possible structural differences between these three
types of trade flows, separate matrices are generated for fresh, seed and frozen potatoes. Country level averages of exports and imports of each country for the years 2005-2009 are used. Speaking in network jargon, the bilateral trade matrices based on the value of the trade flows between any pair of countries represent weight matrices. These weight matrices are converted into adjacency matrices, setting each entry to 1 if a trade flow between a pair of countries exist and to 0 otherwise. From the adjacency matrices we draw maps illustrating the international potato trade networks. All SNA analysis is carried out with the sna package for R.


We produce three maps which provide nice visualizations of the international potato trade networks (Figures 1-3). The maps illustrate the existence and direction of trade links between each pair of countries in the world separately for trade in fresh, seed and frozen potatoes. In order to avoid the maps to become overloaded a threshold is introduced such that a tie is drawn between two nodes if the average value of imports or exports exceeded 100,000 US$ during the period from 2006-2009.

Figure 1: Trade network for fresh potatoes (total average trade volume: 2.14 bn US$).

Figure 2: Trade network for frozen potatoes (total average trade volume: 3.90 bn US$).

Figure 3: Trade network for seed potatoes (total average trade volume: 0.65 bn US$).

The graphs convey some interesting insights about the structure of global trade in potatoes. Comparing the three graphs, it becomes evident that the trade in
each of the potato commodities is of similar geographic structure. While there are trade links spanning the entire globe, in all three networks a tendency
towards regional concentration of trade can be observed, with regional export players supplying surrounding countries. Most striking is the strong
concentration of trade connections in Europe and between Europe, Northern Africa and the Middle East, which represents one of the “hotspots of international
trade” in fresh potatoes previously identified by Walker et al. (2011).

The USA and Canada, beside trading between themselves, are the core exporters to supply the Central American and Caribbean region. Parts of these flows represent
trade between the United States and Canada where potatoes are traded across the border for processing, and, further south, the trade of fresh potatoes from the
USA to Mexico and potato chips into both directions, triggered by NAFTA (Walker et al., 2011).

In South America, Argentina appears to have a central position. In Eastern Asia, China has many export links to South East and Central Asian countries, in
particular for fresh potatoes. In the Oceania region Australia and New Zealand appear as export supplier for South-East Asian countries and the Pacific island
states. In Southern Africa, finally, South Africa is a central supplier of fresh and seed potatoes. Thereby, a few of the regionally central exporters are also
players of global outreach. European exports are destined to the countries without significant potato production in Western Africa and the Caribbean. The
USA and Canada export to countries in the Western Pacific Ocean.

Reciprocal links are of minor importance in the world potato trade network, i.e. there is little intra-industry trade. Some spots of intra-industry trade,
however, can be found. In case of fresh potatoes, reciprocal trade links exist between Mexico, Canada and the USA, between Argentian, Brazil and Paraguay, and
in Europe. Often, this reciprocal trade is linked to the seasonality of trade flows, such as seasonal exports of table and processed potatoes between
Argentina and Brazil (Scott, 2002; Scott & Maldonado, 1999; Walker et al., 2011).

A phenomenon which can also be observed is that some countries are exporters of fresh potatoes, but import frozen potatoes. Examples include China or South
Africa. Likewise (but less well visible in the graph), countries of the North African and Middle East region import seed potatoes from Europe and export back
fresh early potatoes (see also Scott, 2002; Walker et al., 1999).

Judging from the graphs, not all potato producing countries actually engage in international trade. In South America, for example, trade links in fresh
potatoes from, to or between the Andean countries are scarce. Only Peru has export links to Ecuador and Bolivia, but this link vanishes when raising the
threshold for the graph to 500,000 US (not shown). Similarly, the potato producing countries of the highland regions of Eastern Sub-Saharan Africa have
no in- nor outgoing links. India’s trade is to a large extent limited to the island states of the Indian Ocean. This observation points towards a tendency
towards autarky in fresh potatoes among these countries.

Said that, some part of this tendency towards autarky may be explained from underreporting of trade in FAO and COMTRADE data, in particular in Sub-Saharan
Africa (Walker et al., 2011). Some authors, for example, observe cross-border trade in particular of fresh potatoes occuring between Mozambique and Malawi, Uganda and Kenya,
Uganda and Rwanda and Kenya and Tanzania (Walker et al., 2011), and trade is reported to occur between Colombia and Ecuador (Scott, 2002; Scott & Maldonado, 1999).
Neither of these flows shows up in the maps, and their actual extent and importance for the potato sectors of the respective countries remain to be

While the trade networks of the three potato commodities look similar in their overall structure, they differ in their density. The least dense of the three
networks is the seed trade web. The web of fresh potato trade features more connections, but the trade web of frozen potatoes is the densest one. Thus, the differences in densities reflect the differences in total traded volumes. This shows that globalization is most advanced in trade in frozen potatoes and underlines the importance of trade in processed potato products (of which frozen potatoes are the most important representative) already observed by Keijbets (2008), Scott & Maldonado (1999 ) and Walker et al. (2011). The main explanation for the stronger development of trade in frozen potatoes is the better suitability of frozen potatoes for international trade, mainly due to a higher ratio of product value to transportation cost and the lower importance of non-tariff barriers which can be easily erected to block trade in fresh potatoes (Walker et al., 2011).


Graphing the networks of international potato trade definitely provides interesting insights into the structure of global potato trade. But what does it tell us about the initial question of whether potatoes are actually nontraded commodities or whether we can speak of a single world market with perfect price transmission?

First of all, there exists a substantial amount of trade linkages among countries, and these linkages in parts span across the globe. Thus, many potato producing countries actually are connected to international markets. Although there are countries, in particular in South America and Africa which appear to be less well connected or even isolated, this points at least to some degree of tradability and market integration, and the existence of a world market for potatoes. This would be even more the case if the autarky observed in case of some countries does not reflect the reality, but rather the underreporting of cross-border trade in official statistics.

Said that, the existence of trade flows does not guarantee price transmission, let alone the existence of a single world potato market. Regarding the latter, our analysis shows that there are at least trade networks at the regional level – which are connected with each other. Regarding the former, it ultimately is an empirical matter to determine the degree of price transmission across international potato markets. The share of potato production which is traded internationally may matter here. FAO data reveals that during the period from 2003-2007 average shares of exports in total production quantities reached around 7% for the world as a whole (FAO, 2012; expressed in fresh potato equivalents). This is high, for example, compared to rice, where the share of exports in production is around 5% (and rice is rarely described as a nontraded commodity).

The type of potato commodity and the degree of internationalization of trade in that commodity plays a role, too. The different densities of the trade networks of fresh, frozen and seed potatoes lead to the hypothesis that the market for frozen potatoes is the most globalized one, with the highest market integration and highest price transmission.

In that context it also matters how the different potato products – in particular fresh and frozen potatoes – can be substituted for each other. To which extent, for example, do imports of frozen french fries into a given country influence domestic producer prices for fresh potatoes? Similarly, differences in quality and how these influence the substitutability between imports, exports and domestic production play a role.

Finally, it is important to take into account the time perspective in which domestic prices are expected to react to some international reference price. If price transmission is only weak, a given domestic market may indeed be shielded from short term price fluctuations. However, there may be a convergence of international prices in the long run. This second aspect is particularly relevant for a modeling exercise as done with the IMPACT model, which simulates only yearly price averages with a projection horizon of 50 years. In this specific case, the assumption of a unified world market may be warranted.

More generally speaking, it remains an open question whether (and where) potatoes can actually be seen as largely nontraded crops which are traded only in local or national markets isolated from each other. The most likely case that the reality corresponds to something in between: an internationally traded crop with moderate degrees of international price transmission. But clarifying this requires more empirical research.

Amendment (05 December, 2013): We have now made the data and the R code used for this analysis publicly available. Take a look here.


CGIAR. (2011). CRP-RTB 3.4 – Roots, Tubers, and Bananas for Food Security and Income; Revised proposal 9 September, 2011.

Fagiolo, G., Reyes, J., & Schiavo, S. (2009). The evolution of the world trade web: a weighted-network analysis. Journal of Evolutionary Economics, 20(4), 479–514. (Preprint here)

FAO. (2008). Food Outlook – Global Market Analysis.

FAO. (2012). FAOSTAT database.

Garlaschelli, D., & Loffredo, M. I. (2005). Structure and evolution of the world trade network. Physica A: Statistical Mechanics and its Applications, 355(1), 138–144.

Keijbets, M. J. H. (2008). Potato Processing for the Consumer: Developments and Future Challenges. Potato Research, 51(3-4), 271–281.

Scott, G. (2002). Maps, models, and muddles: World trends and patterns in potatoes revisited. Potato Research, 45, 45–77.

Scott, G., & Maldonado, L. (1999). Globalization takes root: Potato trade in Latin America. CIP Program Report 1997-98 (pp. 221–228). Lima, Peru: International Potato Center (CIP).

Serrano, M., & Boguñá, M. (2003). Topology of the world trade web. Physical Review E, 68(1), 1–4.

United Nations. (2012). UN comtrade database.

Walker, T. S., Schmiediche, P. E., & Hijmans, R. J. (1999). World trends and patterns in the potato crop: An economic and geographic survey. Potato Research, 42(2), 241–264.

Walker, T., Thiele, G., Suarez, V., & Crissman, C. (2011). Hindsight and foresight about potato production and consumption. International Potato Center (CIP), Lima, Peru.

  1. CH
    12/11/2012 at 4:38 pm

    To our knowledge, however, it [SNA] has not been applied to trade networks of single commodities.

    There are few papers about single commodities. For example “The evolution of knowledge and trade networks in the global wine sector: a longitudinal study using social network analysis” by Cassi, Morrison, and ter Wal (2009). But at large you are totally right: there are still a lot of questions to ask.

    PS: Your visualizations might be even more meaningful if you would map the node size to the amount of traffic.

    • Ulrich K.
      12/11/2012 at 4:53 pm

      Thanks for the hint about the wine paper – that’s why we’ve said “to our knowledge”…

      And yes, mapping the node sizes to the traded volumes definitely would help. We keep it in mind for the work ahead.

  2. 10/04/2014 at 3:01 am

    Great job! Here’s another single commodity paper – although published with some formatting troubles. I’d be grateful to receive comments, thanks.

    • Ulrich K.
      11/04/2014 at 6:05 am

      Thank you, Umberto, for pointing me to this paper. It is very interesting to see a SNA application to a time series of an agricultural trade network.

  1. 05/12/2013 at 7:46 am
  2. 30/04/2014 at 9:17 am
  3. 19/11/2015 at 4:37 am

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: