JIANG Xiorong, LIU Qing, WANG Shengln,＊

a College of Resource Environment and Tourism, Hubei University of Arts and Science, Xiangyang, 441053, China

b School of Geography and Planning, Sun Yat-Sen University, Guangzhou, 510006, China

Keywords：

Primary products

Global trade networks

Structure entropy

Energy security

Import and export

market

A B S T R A C T

The production and trade of primary products had a growing impact on the economic security of all countries and regions, and the strategic position of these products in the global trade network was becoming increasingly prominent. Based on complex network theory, this paper explored the spatial pattern and complex structural evolution of the global primary product trade network (GPPTN) during 1985–2015 by using index methods, such as centrality, Sankey diagram, and structure entropy, focusing on the diversified spatial structure of China’s import and export markets for primary products(with exceptions of Taiwan of China, Hong Kong of China, and Macao of China due to a lack of data) and their geographical implications for China’s energy security. The research offered the following key findings. The GPPTN showed an obvious spatial heterogeneity pattern, and the area of import consumption was more concentrated; however, the overall trend was decentralized. The trade center of gravity shifted eastwards and reflected the rise of emerging markets. The overall flow of the GPPTN was from west to east and from south to north. In terms of the community detection of the GPPTN, North America, Europe, and Asia increasingly presented an unbalanced “tripartite confrontation”. China’s exports of primary products were mainly concentrated in the Association of Southeast Asian Nations(ASEAN) and other peripheral regions of Asia, and its imports undergone a major transformation, gradually expanding from the peripheral regions of Asia to Africa, the Middle East, Latin America, and other parts of the world.Energy fuels also became the largest imported primary products. Based on the changing trend of structural entropy and main market share, the analysis showed that the stable supply of China’s energy diversification was gradually realized. In particular, the cooperation dividend proposed by the Belt and Road initiative became an important turning point and a strong support for the expansion of China’s energy market diversification pattern and guarantee of energy security.

1. Introduction

As an important part of international trade, the production and trade of primary products, especially crude oil and other energy products, have always attracted global attention (Grilli and Yang, 1988; Hanousek and Kocenda, 2014).On the one hand, the production and export of primary products are an important source of national income and foreign exchange income for developing countries and even serve as an economic lifeline for some nations (Tegene,1990). On the other hand, primary products are the most basic production factors of industrial products (Li et al.,2008), and more natural resources are needed to support the large global population base. Due to a limited growth of supply and increasing demand, the production and trade of primary products have an increasingly stronger impact on the economic security of all countries and regions, and their strategic position in the global trade network is becoming more prominent (Hanousek and Kocenda, 2014). The basic attributes of primary products are related to not only their natural materiality and economic characteristics, but also their historical extension and political nature, their effect on a national sustainable development and foreign strategy, and global geo-economic patterns and their changes, which influence global political patterns (Danilov-Danil’ yan, 2013).

The complex network analysis method has become a research hotspot in academic circles since it was initially proposed (Watts and Strogatz, 1998). It has gradually extended from sociology and psychology into other disciplines,such as economics, statistical physics, city science, biology, and geography (Zhong et al., 2016; de Andrade and Rêgo,2018; Angelidis et al., 2020; Jia et al., 2020) and became an effective interdisciplinary research tool. International trade involved a social network of relationships among many countries and regions. These relationships can be complicated and involve different volumes of trade. In the construction of the global trade network model, the social network analysis method was applied with countries and regions as nodes, trade exchanges as sidings, import and export as flows, and trade volume as weights, which could more comprehensively reflect relative position, network characteristics, and evolution trend of a country or a region (Garlaschelli et al., 2007; Alves et al., 2018).

Early studies on global trade networks mainly focused on the core edge structure of the network, especially the analysis of its topological properties. For example, Wilhite (2001) and Serrano and Boguã (2003) verified the “small world”, “scale-free”, and local heterogeneity of the global trade network by identifying the trade network of major countries and regions. Some scholars have also introduced the concept of space in the study of trade networks,revealing that today’s international trade presents a typical core-marginal structure, “Rich club”, and significant“Matthew effect” (Mahutga, 2006; Fagiolo, 2007; Zhu et al., 2021). All of these studies have discussed the complex topological structure and evolution characteristics of global trade networks. In recent years, researchers have favored more complex representations of trade, such as the multi-layer network (Alves et al., 2018; Ikeda and Lyetomi, 2018).The research results further verified that the trade of industrial products is still highly regionalized and that geographical proximity is still an important factor in promoting international trade (Sturgeon et al., 2008).

Compared with industrial products, primary products have their own particularities that determine whether their production and supply are highly dependent on natural endowment conditions (Hong et al., 2018). This also makes the frequent circulation of primary products in the international trading system necessary for different countries and regions. Therefore, it can also be said that guaranteeing the supply of primary products in the international trade is a major new national security issue related to the lifeline of the national economy. Generally, primary products can be divided into two categories: “energy” and “non-energy”. Most of the research on the global trade network focuses on energy products, such as crude oil, natural gas, and coal. For instance, An et al. (2014) established a trade-based international crude oil network model and studied the evolution of their scale, stability, hierarchical structure, and division over time. Chen et al. (2018) evaluated the structural risks of global natural gas trade based on simulation models of complex network theory. Wang et al. (2021a) constructed a complex network model to comprehensively explore the dynamic relationship and intensity of the global coal import trade. In addition, certain areas of the nonenergy product trade, such as grain (Chen and Li., 2019; Qiang et al., 2020) and woody forest products (Wang et al.,2021b), have become research hotspots in recent years. As a national strategic resource, ore has also become the focus of research on primary products (Hao et al., 2018; wang et al., 2021c).

After 40 years of rapid economic growth, China has become a large consumer of natural resources but is in shortage of these types of resources. Consequently, China’s energy security is often seriously affected by global economic and political factors (Xu et al., 2017). Previous studies have focused on the quantitative assessment of energy security,competition and cooperation with other major powers, and policy formulation (Zha, 2015; Kulkarni and Nathan,2016). A lot of research has also been produced on energy, agricultural products, and other commodities using historical analysis, case studies, and other methods that focus on induction. This research mainly focused on energy,power, and politics, however, and failed to consider geographical factors, such as geographical environment,geographical relations, and spatial structure. The research objects were mostly single products, such as crude oil, iron ore, and agricultural products, which means that a variety of natural resources are not comprehensively considered.

The purposes of this study are to describe the network topology of the global primary product trade network(GPPTN) and to reveal the temporal and spatial differences and evolution stages of the global trade network. This study provides a good reference value for China to correctly understand and grasp the supply guarantee of primary products in the post epidemic period. The instability and uncertainty of the global industrial chain supply chain have increased significantly, and the supply guarantee of primary products has become an unavoidable practical problem.

2. Measurement model and data processing

2.1. Data sources

The bilateral trade data used in this paper were from United Nations (UN) Comtrade (https://comtrade.un.org/).The database was classified in terms of the Standard International Trade Classification (SITC) (UNS Division, 2006).The database contains bilateral trade data for more than 200 countries and regions in the world, such as the United States, Russia, China (with exceptions of Taiwan of China, Hong Kong of China, and Macao of China due to a lack of data), India, etc. Primary products mainly covered five types of goods: (1) food and live animals that were used primarily for food (SITC-0); (2) beverages and cigarettes (SITC-1); (3) non-edible raw materials other than fuel(SITC-2); (4) fossil fuels, lubricants, and related raw materials (SITC-3); and (5) oils and grease from animals and vegetables (SITC-4). Primary products have not been processed or were only slightly processed due to the marketing habits, such as natural rubber, crude oil, iron ore, and other agricultural, forestry, husbandry, fishery, and mineral products. Based on the availability of Chinese records in the database, we decided to study the records from 1985,1990, 1995, 2000, 2005, 2009, and 2015. Social network analysis method requires the establishment of adjacency matrix between nodes, so it is necessary to convert bilateral trade data intoN×Nadjacency matrix (whereNis the total number of network notes).

2.2. Complex networks analysis

2.2.1. Distribution and cumulative probability

In the trade network, the role and degree of a country or a region determine its position in the network. Considering the dislocation distribution characteristics of the supply and demand of primary products in geographic space, we divided the center degree of each node into the out-degree and in-degree, respectively representing the export and the import. The cumulative probability distribution function of the degree value is:

whereP(k)refers to the cumulative probability of the degree value of network nodek;Nis the total number of networks; andnk’is the number of nodes with a degree value ofk’. If the node degree distribution had the law of power function distribution, it can be understood that the network conforms to the characteristics of scale-free network(Barabási and Albert, 1999).

2.2.2. Complex network indicators

In order to calculate the total network connection of a node (total volume of trade), we further defined the outdegree and in-degree of nodeias follows:

Network density can be used to describe the density of edges between nodes in the network (Barabási and Albert,1999). The higher the the network density is, the closer the relationship between network nodes. The network density is defined as follows:

whereDis the network density;nis the number of nodes in the GPPTN; andlrepresents the number of actual links to this network.

In complex network theory, the clustering coefficient was used to measure the degree that nodes tend to gather together in the network (Saramäki et al., 2007). There are trade relations between trading partners, so different trading groups are formed in the GPPTN. A high clustering coefficient indicates good connectivity between trade partners around the country (or region) node. In this study, we used the average clustering coefficient to measure the overall level of clustering in the network. The average clustering coefficient is defined as follows:

whereCis the average clustering coefficient andnis the number of nodes in the GPPTN. If and only if the edge connects nodesiandj,aij=1; otherwise,aij=0. The same is true for values ofajh,ahj,ahi,andaih.kiis the degree of nodei;

The average path length of the networkLis defined as the average of the distance between any two nodes (Fronczak et al., 2004). In the GPPTN, a shorter average path length indicates higher interoperability between nodes in the network. The average path length is defined as:

whereLis the average path length;dijis the distance between nodesiandj, whichis defined as the number of edges on the shortest path connecting these two nodes; andnis the number of nodes in the network, without considering the weighted network.

2.2.3. Community detection

The community reflects the local characteristics of individual behavior in the network and their mutual relationships. We used Gephi 0.92 (an open source software for exploring and manipulating network is available at http:/lgephi.org) to detect three network communities. Community detection can be introduced to identify potential communities in network. It is generally believed that the trade exchanges between countries (regions) within an association are relatively close in the GPPTN, while the connections between countries (regions) in different associations are relatively much less. Usually modularity is a common method to measure the quality of community division in community testing. The maximization of modularity is considered to be the currently definitive method of community detection (Newman, 2006), and the modularityQis defined as:

Here, we introduced an introduction symbolδ(i,j). If and only if nodesiandjare in the same community,δ(i,j)=1.In Equation 7,mis the number of lines in the GPPTN. If and only if an edge connects nodesiandj,aij=1; otherwise,aij=0. Further,kiandkjrepresent the degree values of nodeiandj, respectively. The termkikj/2min Equation 7 represents the expected value of the number of edges between nodeiand nodejwhen edges are randomly placed.

2.2.4. Structure entropy

Entropy, as a measure of the disorder degree of a network system, is an important index in the measurement of the complexity and randomness of structure networks (Bianconi, 2008). Taking “nodes” as the research object, the structure entropy reflects the heterogeneity of the network—point disorder according to the uncertainty of the probability distribution of the (relative) number of edges of nodes. The formula of structure entropy (SE) is:

wheredkanddiare the degree values of nodeskandi,respectively, andNis the total number of networks.

It can be seen that structure entropy is an index that can reflect the diversification of trade markets (Cai et al., 2017).In the trade network, the higher the structural entropy of nodes is, to some extent, the more balanced the spatial distribution of the national trade market; this is completely consistent with the expanding equilibrium state pursued by the energy diversification strategy.

3. Complex structural evolution of the GPPTN

3.1. Network topology

In order to better demonstrate the evolution of country (or region) status in the GPPTN, we calculated the centrality of all countries and regions in each cross-section year (standardized processing was adopted for convenience of comparison). The cumulative probability distribution is shown in Figure 1.

Fig. 1. Degree cumulative probability distribution of the global primary products trade network (GPPTN). P(k) refers to the cumulative probability of the degree value of network node k.

It can be seen that the cumulative probability distribution of both the in-degree and out-degree conforms to Pareto’s distribution (Newman, 2005), indicating that less than 20% of countries and regions occupied a dominant position in the GPPTN, which reflected the spatial heterogeneity of the network. When comparing the in-degree and out-degree curves, it appeared that the spatial heterogeneity of the in-degree network was more serious (the turning point of the in-degree cumulative probability curve in Figure 1 was further to the lower left corner), indicating that the in-degree values in less than 10% of countries and regions were more than 0.10, which will be at a disadvantage in the global trade network. Compared with producing and exporting countries and regions of primary products, the distribution of importing countries and regions and consumers was more concentrated. In the early stage, they were mainly the United States, Japan, Germany, and developed countries in Western European. After 2005, China and India, the two largest emerging economies, also became the major consumers of primary products. The distribution was more extensive for exporting countries and regions, including developed countries such as the United States, Britain, the Netherlands, and France that export agricultural products, and developing countries such as the former Soviet Union,Saudi Arabia, Mexico, Indonesia, Nigeria, and Brazil. With the passage of time, the distribution of exporting countries and regions has become more centralized (the turning point of the curve has shifted to the lower left corner), indicating that the production of primary products (especially fossil energy) was mainly determined by countries and regions with natural endowment, which was relatively fixed during 1985-2015, and the few resource-rich countries increasingly came to occupy the dominant position in the GPPTN.

Network density refers to the ratio between the actual number of connections between nodes in a network and the theoretical maximum possible number of edges, reflecting the closeness of connections among network members (Cai et al., 2017). The GPPTN increased gradually from 1985 to 2010, and the increase rate was basically consistent with the average degree (Fig. 2a and b). This showed that with the deepening of economic globalization and trade liberalization, countries and regions in the world primary products trade exchange needed goods with each other, and the connection of the whole network was increasingly close. However, after the financial crisis, the growth of the average degree and density of the network tended to slow down or even slightly decrease, reflecting the emergence of anti-globalization challenges in the world, mainly due to the continued economic downturn and the escalation of protectionism in the post-crisis era.

Fig. 2. Evolution of the GPPTN’ networks topology during the 1985-2015. (a), average degree; (b), network density;(c), average path length; (d), average clustering coefficient.

The average path length of the trade network of primary products showed a downward trend from 1990 to 2000,reflecting that the accessibility of the network was gradually enhanced (Fig. 2c). Although the clustering coefficient fluctuated, it remained within the range from 0.70 to 0.74 (Fig. 2d), indicating a high level of local agglomeration. A network with a low average path length and a high clustering coefficient was called the “small-world network”, so the global trade network for primary products was a typical “small-world network”. These findings also showed that most national nodes can be connected through a few core countries.

3.2. Pattern and evolution of the GPPTN

In order to facilitate the analysis of the cross-regional evolution of the trade network, we divided the world into seven continents: Asia (excluding West Asia), Europe, North America, South America (including Central America and the Caribbean), Africa (excluding North Africa), Oceania, and the Middle East (West Asia and North Africa). It should be noted that these geographical divisions did not strictly adhere to continental boundaries, in order to facilitate statistical analysis. West Asia (a total of 19 countries) and North Africa (6 countries) were categorized as the Middle East. Mexico was included in the North American Free Trade Area, so it belonged to the North America. The remaining countries in the Americas were all classified as South America. Due to its unique endowment of oil and gas resources, the Middle East has always been an important location for major trading countries based on factors such as geopolitics and economic interests. Therefore, they were listed for analysis.

As shown in Figure 3, with the deepening of economic globalization, the scale of global primary product trade continued to expand, and presented a transcontinental network model. Spatial imbalance has also always been a prominent feature of the global trade network. A few large trading countries and regions occupied the central position of the global trade network and had a significant “core-periphery” structure. In order to shed light on the evolution of global dominance, we listed the top ten countries in the in-degree and the out-degree of the GPPTN for seven time periods (see Tables 1 and 2).

Fig. 3. Community detection of the GPPTN in 2015. Node size indicates the total trade in the country or region. Line segments between nodes represent the size of trade flows between them. The national and regional International Organization for Standardization (ISO) codes were used in the diagram. The specific full name of the abbreviations is shown in Appendix (Table S1).

Table 1Top ten countries in the total in-degree of the global primary products trade network (GPPTN) in seven time periods.

Table 2Top ten countries in total out-degree of the GPPTN in seven time periods.

In the 1980s, the leading importing countries were mainly distributed in North America, Western Europe, and Japan, which were all developed countries. Primary products were the basic raw materials used in modern industrial production, and during this period, major manufacturing countries were all developed countries because industrial production required a large amount of fossil energy, minerals, agricultural products, and other primary products, which directly increased the import demand for primary products in developed countries, such as the United States, Japan,and countries in Europe. The United States and Japan took the first place in turn. As the birthplace of the industrial revolution, the industrial belt of Western Europe occupied a strong position in the network. In fact, seven of the top ten manufacturing countries were located in this area and formed the intra-European trade network.

The United States, Japan, and Germany were the top three importing countries during the 1980s. Imports that came into the United States originated from countries in North America and South America, such as Canada, Mexico, and Venezuela, as well as oil producing countries, such as Saudi Arabia and Nigeria. Japan needed to purchase raw materials globally because of its poorer stock of the resources, and the United States, Saudi Arabia, the United Arab Emirates, Indonesia, and Australia were its main sources of imports. Germany received its raw materials mainly from Europe. Countries that dominated exports mainly included the United States, the former Soviet Union, Saudi Arabia,and other Western European countries that primarily exported agricultural products, such as Britain, the Netherlands,and France, which ranked in the top five exporting countries. These findings indicated that during this period, the global export trade for primary products mainly coexisted with energy and agricultural products. The Gulf War in 1991 severely damaged oil resources in the Middle East. The ranking of Saudi Arabia, a major producer, fluctuated widely, cementing the dominance of developed countries in Europe and the United States. After the collapse of the Soviet Union, Russia’s dominance weakened, and it fell from the second place before shifting to the ninth place. The fluctuations in the two countries’ rankings also reflected the significant impact of geopolitical events on trade. Among the seven continental networks, only Western Europe and North America have formed a network prototype dominated by horizontal connections, while the rest of the continents was dominated by hierarchical single-center vertical connections.

In the early 1990s, Japan surpassed the United States and became the world’s largest importer of primary products,a position it held until 1995. Some other key changes also took place in the global trade pattern of primary products.Though the United States and Japan were still the leading importing countries in 2015, South Korea and China in East Asia have gradually risen to the forefront. For example, South Korea ranked the third and China ranked the seventh in 2000. China rose to the fourth place, while Western Europe fell to the sixth place in 2005. The import trade center of gravity has gradually begun to shift to Asia. In addition, the Asian financial crisis in 1998 had a serious impact on Japan, and Japan fell from its first place position to the third place in 2000. The network has become increasingly complex, with a small growth in cross-regional trade links from 76.69% in 1995 to 77.37% in 2005. Large trade(≥1.0×109USD) grew even more significantly from 48.00% in 1995 to 53.83% in 2005, with more than half of large trade occurring between countries (regions). In terms of exports, oil producers dominated, with Saudi Arabia, Russia,and Canada topping the list. The establishment of the North American Free Trade Agreement and the European Union has greatly enhanced the closeness of the two sub-regional networks in North America and Europe and formed network communities with close internal connections. Driven by economic radiation from Japan, China, and South Korea, trade links among Asian countries have gradually strengthened, forming a sub-regional network among major countries in East Asia, Southeast Asia, and South Asia.

As economic globalization and trade liberalization continued to advance in depth and breadth, the flow of primary products was more frequent and free. At the same time, much has changed in the dominant core countries of the global trade network. Since the beginning of the 20thcentury, major adjustments and changes have taken place in the international economic system, and the economies of western developed countries have grown slowly. The rose of emerging economies, such as those in China, India, and Brazil, has been a prominent feature that affected both imports and exports. China, in particular, has overtaken Germany, Japan, and the United States to become the world’s largest net importer of primary products. India was also the seventh leading importer, and Brazil’s rich natural resources,such as iron ore and oil, determined its status as a major power in the global trade network. Since 2009, North America,Europe, and Asia have formed an unbalanced “tripartite confrontation” on import and export structure. The network structure in North America was the most balanced, forming a three-node and two-way interactive transmission mode.The one-way transport mode of Russia was apparent in the European region, which showed that the European Union had a high dependence on Russian energy. Due to its rapidly developing levels of industrialization, huge population base, and relatively scarcity of resources, Asia was rapidly becoming the world’s “pump” of resources, continuously importing the world’s energy, minerals, agricultural products, and other resources and materials to the four consumer powers: China, Japan, India, and South Korea.

In terms of exports, the Middle East and South America formed sub-regional networks with Saudi Arabia and Brazil as the core, respectively, while Africa and Australia mainly participated in the global trade network through the one-way flow of exporting oil, minerals, and agricultural products, and network phenomenon was relatively insignificant. Globally, the great “bow” zone formed by countries and regions in Latin America, North Africa, the Middle East, Australia, and Association of Southeast Asian Nations (ASEAN) constituted the central region of global primary product production. Brazil and Australia on either side were the world’s dominant suppliers of iron ore, and the gulf region in the middle was the main source of oil, gas, and other energy. The global trade network presented an unbalanced flow pattern.

It was worth noting that compared with data in 2005, the streamline density and edge weights (representing trade volume) of the global trade network showed a trend towards regression in some countries and regions in 2009,indicating that the financial crisis in 2008 directly led to a sharp decline in global trade links of primary products. The financial crisis that took place in 2008 caused a widespread aversion to globalization and triggered a new wave of trade protectionism, leaving global trade in a crisis of sluggish growth, stagnation, and even backsliding. As of 2015,countries around the world had not fully recovered from the effects of the economic crisis and demonstrated limited growth in trade volume and even signs of recession. Most obviously, the United States and developed countries in Europe, which were hit the hardest by the financial crisis, have seen their economy stagnating and domestic consumption shrinking. Weak demand was also reflected in a shortfall in their imports. The imports of Italy, Japan,the United States, and France, which were all part of the Group of Seven (G-7), fell by more than 15.00%. Export-led countries have also been serious affected. Members of the Organization of Petroleum Exporting Countries (OPEC),in particular, have seen their exports fall sharply, including Nigeria, Angola, Venezuela, Iran, and Saudi Arabia, whose exports fell by more than 30.00%.

3.3. Community detection of the GPPTN

According to the calculation produced by Gephi’s modularity community discovery model, three major network communities in Asia, Europe, and America were formed globally with China, Germany, and the United States as the cores (Fig. 3). In terms of community size, the Asian community was the largest and included 84 countries and regions. The European community followed with 74 countries and regions, and the American community was the smallest with only 25 countries and regions. From the perspective of geographical location, the members of the Asian community were relatively diverse and included some countries and regions in East Asia, Southeast Asia, Oceania,the Arabian Peninsula, and even South America, such as Brazil, Argentina, Chile, and Peru, which should have been included in the America community due to their closer geographic locations. The Asian community, which included many countries and regions with emerging economies such as China, India, South Korea, and the ASEAN, has become the engine that led the recovery and growth of global trade in the post financial crisis period, and has established close trade ties with resource powers such as South America. The member states of the European community were mainly concentrated on Europe and the nearby Middle East. The core region consisted of European Union member states,including Germany, Britain, France, Italy, and the Netherlands. They have carried out in-depth and close resource trade cooperation with other European countries and regions, such as Russia, Ukraine, and Turkey. The American community included two African countries (Gabon and Chad), and the rest were located in the Americas. Although there were few member countries and regions in this community, three of the top four trade flows were located within this community: bilateral trade flowed between the United States and Canada and flowed from Mexico to the United States.

Based on community detection, we found the spatial organization characteristics of the GPPTN. First of all,community monitoring showed that the GPPTN had a typical “core-periphery” hierarchy. China, Germany, and the United States, as the core nodes of the three major community, played a leading and driving role in the trade cooperation of primary products within their associations. Secondly, geographical proximity was an important factor in the formation of global trading communities, and spatial proximity promoted the formation of trade relations and increased trade flows. Most countries and regions in Asia, Europe, and North America were in the same network community, while Africa, Oceania, and South America had no core powers of their own and were attached to the three neighboring communities by strong trade attraction. There were a few exceptions, however, such as Brazil, Argentina,and the other countries or regions in South America mentioned above. Thirdly, large country effect was also a distinctive feature of the GPPTN, which reflected the preferential attachment of small world networks. This strong connection was reflected not only within the community, but also between the large countries of different communities.The size of a national economy reflected its economic capacity and potential trade needs to a certain extent, and it was an important source of attraction for bilateral trade.

4. Diversification of China’s primary products import trade and its energy security

4.1. Spatial evolution of China’s primary products import trade

The production of primary products relied on geographical advantages and rich natural resources. Supply-side markets tended to be concentrated in a few countries and regions, which caused sellers to monopolize the global market of primary products. In addition, commodity trading was monopolized by a few international oligopolies, and the pricing power of almost all primary products was firmly controlled by a few developed countries and regions.China’s large population base and its economic growth in industry have led to a situation where more food and fossil energy were needed to meet the increasing demand. Therefore, under the joint action of supply and demand, primary products had a growing influence on China’s economic security, energy security, and food security. In other words,the trade in primary products played an increasingly prominent strategic role in China.

In order to provide a clearer impression of the flow path of China’s primary products, we drew the Sankey diagrams of import and export in the three study periods (1985-2000, 2001-2008, and 2009-2015) to help to visualize the flow and proportion of China’s import and export of primary products (Fig. 4).

Fig. 4. Sankey diagram of import flow of China’s primary products during the 1985-2015.

China’s entrance in the World Trade Organization (WTO) in 2001 and the global financial crisis in 2008 were two important time nodes for China’s foreign trade development and the formation of the current international trade pattern. For the convenience of analysis, we divided 1985-2015 into three periods: early stage (1985-2000), middle period (2001-2000), and post-financial crisis period (2009-2015).

From the diagram above, we can see that during 1985-2000, non-edible raw materials surpassed energy fuels as the largest import commodities, and the sources of imports changed considerably. The ASEAN, the United States, the Middle East, Australia, and Latin America became the major import regions and countries. This trend reflected that China was in a stage of rapid industrial development and urbanization and thus needed to consume large amounts of energy and resources from around the world to meet growing resource demands. After China accession to the WTO in 2001, the geographical structure of imports of China’s primary product underwent great changes, with the Middle East, Africa, Russia, and other oil powers being the main suppliers of China’s energy products. Brazil also rose sharply in the rankings because it supplied China with a large amount of iron ore. China’s steel production grew rapidly after 2001. From 2001 to 2007, the average annual growth rate of steel was 21.11%, compared with 6.65% in the early stage period (1985-2000).

Over the past decade, the extreme prosperity of China’s real estate market and construction of large-scale infrastructure have made it the country with the greatest demand for iron ore. The main source of import was in Australia and Brazil, which has led to the countries’ rise in the rankings of primary product imports to China. Australia became the second importing country for China in primary products. We found that after the financial crisis, energy fuel surpassed non-edible raw materials and became the largest imported product in China during this period. This trend may be related to the increase of China’s energy consumption and energy reserves in recent years, leading to a growth in China’s foreign dependence for energy year by year. The OPEC countries, such as the countries in the Middle East and Africa, were still the major sources of China’s energy imports, among which crude oil accounted for 70.63% of the total imports of energy fuel, reflecting the diversification of China’s imports of crude oil.

4.2. Diversification of China’s energy imports

In the global market, oil and gas were the primary products that received the most attention. As important strategic energy, oil was the cornerstone of a national industrial economy, and it was most often used by major powers of interest in the management of geopolitics. As a major oil consumer, China has become increasingly dependent on oil imports in recent years. By 2018, crude oil imports in China had reached up to 4.62×109 t, with external dependence rising to 70.99%, which has far exceeded the general international standard of 50.00%. Therefore, the possibility of exogenous impact and the damage degree caused by impact has also increased. In the case of war or another sudden political event, the Chinese government was highly dependent on the energy import pattern in the strait of Malacca(some people call it the “Malacca dilemma”) due to its inability to control the major sources and shipping routes of primary products across the world, which will bring huge risks to China’s energy security. Therefore, in order to solve the long-term high external dependence on crude oil, it is necessary to expand the import source market and establish strategic and commercial reserves of important primary products.

We can describe the degree of import market diversification by structure entropy. Figure 5 intuitively showed that China’s energy import channels have become increasingly diversified, with the structure entropy climbing from 1.20 in 1985 to 3.31 in 2015, indicating that China’s stable supply of energy diversification has gradually been realized. From 1985 to 2000, the growth of structure entropy was the fastest. After the breakthrough of 3.00 in 1999, the structure entropy was basically stable at around 3.00 until 2008. Since the end of the Cold War, the financial crisis has intensified the fierce turbulence and profound evolution between the world economic pattern and the comparison of international strategic forces. The importance of energy trade has become increasingly prominent. China has also further expanded the energy import channels, resulting in a slight increase in structure entropy. The structure entropy has thus been maintained at a high level. The shale gas revolution in the United States broke its dependence on oil imports and led to energy independence after 2004, resulting in a slow decline in structure entropy. It was also the result of a combination of efficiency measures, increased use of natural gas by the United States, and the financial crisis.

Fig. 5. Trend of structure entropy of energy products imports in China and the United States during the 1985-2015.

We selected ten major importing countries (Saudi Arabia, Angola, Iran, Iraq, South Korea, Oman, Russia,Singapore, Venezuela, and Yemen) to analyze the regional pattern of China’s crude oil imports. The total import value of these ten countries accounted for about two-thirds of China’s total import value of crude oil, so it had good representativeness for a long time (Fig. 6). The trend of trade volume among major importing countries showed that China’s demand for crude oil began to soar within one or two years after China joining the WTO in 2001. The demand for crude oil in Saudi Arabia, Angola, Russia, and Iran also increased rapidly. In 2008, the financial crisis had a major impact on China’s crude oil import volume. In 2009, China’s imports of crude oil declined sharply, rebounded quickly after the crisis, and then began to decline after 2012. Only Oman and Iraq continued to grow with respect to crude oil imports. In 2015, due to the continuous deterioration of the international trade environment, China’s total crude oil imports decreased sharply, and the import volume of all countries dropped significantly.

Fig. 6. Trend of China’s trade volume from the top ten crude oil importers during the 1985-2015.

Over the past 30 years, a major shift has occurred from the perspective of the change trend of the share (Fig. 7).This change indicated that the major markets of China’s crude oil imports were becoming increasingly diversified.The diversification pattern in the middle and later stages was significantly better than the one observed in the early stage. During the early stage, Singapore, Oman, and South Korea had a large share. In the 1980s, Singapore’s share was as high as 45.00%-60.00%, reflecting Singapore’s dominant position as the entrepot of world crude oil trade.Compared with the middle and later stage, we found that although the share of ten major importing countries tended to be even (the lines in the Figure 7 tended to be close), the direction of the line showed that in the period of post financial crisis, China’s market choices have gradually adjusted. Countries and regions that had long maintained good relations with China, such as Russia and Venezuela, have experienced a share increase (9.63% for Russia and 4.72% for Venezuela in 2008, and 15.76% for Russia and 5.42% for Venezuela in 2015).

Fig. 7. Share trend of China’s top ten crude oil importers during the 1985-2015.

Energy cooperation was a key area in the joint construction of the “Belt and Road”. China has established an energy cooperation organization with more than 30 countries and regions along the “Belt and Road”, which was called the“Belt and Road” Energy Partnership. Because of the “Belt and Road” initiative proposed by China, major oil and gas countries along the “Belt and Road” have seen real cooperation dividends. The “Belt and Road” region was the main source and channel of China’s energy imports. It has been a priority area of ensuring China’s energy supply and international energy and resource cooperation. At the same time, China’s huge advantages in market demand,industrial manufacturing, and infrastructure construction also provided possibilities and great potentials for China and the “Belt and Road” in energy trade, infrastructure construction, and two-way investment. For example, Oman has responded positively, with its share nearly doubling. As one of the first potential founding members of the Asian Infrastructure Investment Bank, Oman has become an important partner in jointly implementing the “Belt and Road”initiative with China.

5. Discussion and conclusions

This paper analyzed the pattern evolution process of the GPPTN and China’s energy diversification and inspiration of geographical cooperation by using the social network analysis method. The results pointed to the following key findings. Based on the analysis of network topology, we found that the cumulative probability distribution of the center degree of the global trade network for primary products conformed to Pareto’s distribution.

The trade network presented typical “small world” characteristics, indicating that the global circulation of primary products had high accessibility and patency. Over the past 30 years, the evolution process of the spatial pattern of the GPPTN has experienced three development stages. These changes have shifted the consumer trade center eastward and led to the rise of emerging markets. Community detection showed that the GPPTN presented three communities in Asia, Europe, and America, respectively. Hierarchy, geographic proximity, and preferential attachment were the three salient features of the GPPTN. In recent years, the dependence on foreign oil has increased in China. By 2018,the dependence on foreign crude oil was as high as 70.99%, posing huge risks to its energy security in China. However,the calculation based on network structure entropy showed that China was gradually realizing the stable supply of energy diversification, and the share of countries and regions along the “Belt and Road” that maintained good international relations with China has increased.

Based on the above conclusions, we offered several policy recommendations to ensure China’s energy security.First, realizing the diversification of import sources was an important guarantee for the security of energy imports.China should rely on its core position within the GPPTN and build a base in Asia before establishing a global primary supply network that radiated outward to Europe, Africa, and America. Second, China need to make good use of the platform of the “Belt and Road” initiative, continue to expand its market globally, strengthen comprehensive cooperation in the fields of politics, economy, energy and security, and gradually increase the initiative of energy security through multiple channels and security forces. Finally, China should give full play to its advantages in manufacturing and infrastructure construction and comprehensively consider cooperation with resource countries (or regions). These efforts should involve contracting major infrastructure projects, industrial investment, and solving local employment, which were also the factors that were most valued by the governments of countries and regions along the “Belt and Road”.

The existing studies mainly focued on the basic geographical pattern, energy trade and channel control, the game between producer and consumer countries (or regions), energy security evaluation and so on (Yang et al., 2015). This paper investigated the network relationship of primary product trade in 227 countries and regions based on bilateral trade data, which offered a new perspective for understanding the evolution of economic globalization. Of course,there are some limitations of research and further study is needed in the future. A depiction of the overall network structure might not detect differences between sub-products, where production and consumption trends differed significantly between the two countries (regions), such as agricultural products and energy products. This oversight could result in very different network topologies (Wang et al., 2021a). The GPPTN and its changes were systematic and complex, and involved many aspects, such as factors related to economy, industry, politics, and culture (Kitamura and Managi, 2017). The network structure and its changes were only one of its many explicit “results”. The formation and evolution of the global trade network were also influenced by special geographical, historical, cultural, and political factors (Coe and Yeung, 2019). Therefore, on the basis of the role of the market, the formulation of trade policies must also take the institutional and cultural differences and political relations between countries (regions).

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

This work was financially supported by the MOE (Ministry of Education in China) Project of Humanities and Social Sciences, China (20YJCZH057), the Hubei Province Social Science Fund General Project, China (2021147),and the Xiangyang City Science and Technology Planning Project, Hubei Province, China (2021rkx04). We are grateful to the anonymous reviewers for their insightful comments.

Appendix

oduction and trade of primary products had a groS1Country and region International Organization for Standardization (ISO) code.