Complex systems approach for international agricultural trade network

Mechanization and the use of fossil fuels and fertilizers in agriculture since the early 20th century are the catalyst of the current global food production growth and worldwide trading. For food supply, the massive... [ view full abstract ]

Mechanization and the use of fossil fuels and fertilizers in agriculture since the early 20^th century are the catalyst of the current global food production growth and worldwide trading. For food supply, the massive development of international agricultural trade networks is a turning point in history, as it has changed the food system into a highly interconnected global system. The trading network growth is built on a dependency trajectory constrained by finite resources and geopolitical policies. Food security worldwide and in particular of urban population which is more than half of today global population, increasingly relies on global network supply fidelity.

Modeling the global agricultural system is necessary to understand the main patterns of the system and to anticipate environmental and food security issues. The graph mathematical theory allows modeling the collective properties of a complex system composed of many interconnected parts. The graphs are made of nodes and links that are relevant to describe the system. The graph nodes correspond to the countries and the edges to product flows.They provide a modeling framework to assess the robustness and the resilience of such complex networks. The evolution of global wheat and rice networks has been studied for the period 1992-2009 (Puma et al., 2014) based on FAOSTAT data-bank that reports national production and origin-destination food and feed flows since 1986. The network framework has been also used by Lin et al. (2014) to study the food flows within the USA for the year 2007 based on the Commodity Flow Survey (CFS).

Our work examines the growth of international trade agri-food networks over the last 27 years (1986-2013) from FAOSTAT data for product categories such as cereals (other than rice) and coffee for comparison. The objective is to characterize the evolution of key features of the network quantified by its topological complexity and the networks metrics. Here, the international origin-destination flows are defined as a weighted and directed network. The nodes are characterized by different metrics such as their weights (inherent features as national production, ability of export or import, urban population) and also by their centrality in the graph (clustering coefficient, connectedness, paths and distances).

These key network features underline the network stability and its resilience and in the case of food they have highly evolved overtime. The results show the changing heterogeneousness of the global trading network: the distribution of exchanged volumes is highly uneven across nodes and displays scaling properties. Most of the trading volume is concentrated in a few number of nodes and edges, while nodes display large disparities between net exporting and importing countries. Moreover, the agricultural trade graph is evaluated as a disassortative network where high degree nodes are preferentially connected with lower degree nodes.