Evolution and Impacting Factors of Global Renewable Energy Products Trade Network: An Empirical Investigation Based on ERGM Model

Abstract

Global trade of renewable energy products has increased significantly in recent years. This paper constructs an analytical framework of a global trade network for renewable energy products based on bilateral trade data between 2009 and 2019. It analyses its structural evolution at the global and local levels and investigates the influencing factors of the network with the Exponential Random Graph Model. The empirical results indicate that countries in the trade network have become more closely connected, featuring a core-periphery and increasing reciprocity relationship. China, Germany, and Japan have remained in the position of core countries; China has especially been prominent among core countries. Our empirical results verify that the sender-receiver effects can explain the evolution of this global trade network. The empirical results also indicate that the climate change agreement network and the common border network have positive effects on the formation of the trade network. As regards political implications, the core countries in the trade network should optimize the layout of renewable energy development and improve infrastructure accordingly. Countries should also jointly build a more fair and reasonable multilateral system that fulfills their responsibilities.

1. Introduction

The overexploitation and utilization of natural resources have deteriorated the global environment, and natural resources are undergoing shortages or even depletion. To maintain a country’s energy security, cope with climate change and protect the ecological environment, it is necessary to adjust energy consumption strategies in production systems [1,2]. The use of low-carbon energy sources and renewable energy should be prioritized to reduce carbon emissions [3]. Meanwhile, the vigorous expansion of new energy has become an important goal for major countries as a means to boost investment, promote employment and facilitate green economic recovery [4,5,6,7,8]. For example, the “Outline of the 14th Five-Year Plan (2021–2025) for National Economic and Social Development and Vision 2035 of the People’s Republic of China” state that China should promote green development and harmonious coexistence of man and nature. China will strive to reach peak CO₂ emissions by 2030 and achieve carbon neutrality by 2060. China is setting more aggressive renewable energy development goals for the future [9]. The development and use of renewable energy are conducive to the transformation of traditional energy sources and the achievement of the goal of sustainable energy development. They can also help to keep pace with global trends and integration with the global energy system.

The global renewable energy industry has undergone tremendous changes. In addition to the wind and solar industries, some renewable energy industries that were once relatively left behind (bioenergy, geothermal energy, ocean energy) have now taken a share in the market. Different types of renewable energy products play similar roles, which can not only alleviate the pressure of the energy crisis to a certain extent [10,11,12] and bring considerable economic benefits [13] but also contribute to the solution of global environmental problems and sustainable development [14,15,16,17,18,19]. Studying only a single renewable energy product does not provide a comprehensive understanding of the industry’s current development. Meanwhile, according to the CEPII database, the value of global trade in renewable energy products exceeded USD 320 billion in 2019, with a total of 224 countries and regions involved. With the participation of many countries in the global production specialization, the trade networks have become increasingly complex. Therefore, it is necessary to study the trade in this field and provide suggestions for the development of global trade in renewable energy products and the stability of the trade network on this basis. Only through in-depth analysis of the overall structural characteristics of the trade network and the characteristics of each node in the network correlation, then can we see what role each country plays in the trade network.

In recent years, complex social network methodology has become a major focus of research in the field of international trade [20,21,22,23,24]. The complex social network analysis framework is used to explore the macro- and micro-level characteristics of network structure evolution [25]. Complex networks have also been applied to trade involving renewable energy products [26,27]. These studies have offered a preliminary understanding of the drivers of renewable energy products trade among economies, but most of them assumed that renewable energy products trades between the economies are independent of each other, particularly when using the gravity model to analyze the influencing factors of global renewable energy products trade. In fact, the renewable energy products trade relationships between economies are not as simple as dyad but instead becoming even more complex and interdependent than ever under the accelerating process of globalization [28]. More importantly, studies that examined the impact of national attributes (e.g., income, environmental regulations, transport costs, and economic policies) on renewable energy products trade largely ignored the endogenous dependence of renewable energy products trade (e.g., reciprocity, transitivity, and connectivity) or the external network relationships. The exponential random graph model (hereafter abbreviated as the ERGM model) is introduced into the formation mechanism of trade networks to study the dependencies between network relationships, i.e., whether the occurrence of one relationship affects the probability of the occurrence of other relationships. The ERGM model is an effective tool for the analysis of variables in terms of network structure and relationship at multiple levels and considers the reasons behind the evolution of the network structure comprehensively [29,30].

To remedy the aforementioned deficiencies, we use the bilateral trade data for each country from 2009 to 2019 according to the CEPII database and filter the bilateral trade of renewable energy products for each country regarding the HS 6-digit codes summarized by Jha [31]. Then, we construct the ERGM model to analyze the influencing factors of the renewable energy products trade network. This study combines endogenous and exogenous factors to characterize the formation of a global renewable energy products trade network. To be more specific, the primary research questions of this paper are as follows:

• (1). What are the macro- and micro-level characteristics of network structure evolution relating to global trade in renewable energy products? To answer this question, this paper introduces the analytical framework of complex networks in the field of renewable energy product trade and constructs a global renewable energy products trade network model.

• (2). What are the dependencies between network relationships, i.e., whether the occurrence of one relationship affects the probability of the occurrence of other relationships? To answer this question, the exponential random graph model (hereafter abbreviated as the ERGM model) is introduced into the formation mechanism of trade networks.

• (3). What are the factors that influence the evolution of renewable energy products trade networks? What are the reasons behind the evolution of the network structure comprehensively? To answer this question, the ERGM model is introduced to simultaneously take into account variables in terms of network structure and relationship at multiple levels.

By addressing the above issues, this paper contributes to the existing literature in the following three respects. First, this paper enriches the research on the global trade of renewable energy products. Few academic articles study the trade of renewable energy products from the perspective of global flows and the overall situation. This paper targets the global context and general renewable energy products to achieve a more comprehensive analysis. Second, this study introduces the analytical framework of complex networks to the field of renewable energy product trade, which broadens the research method and paradigm of renewable energy product trade. It provides a more objective, scientific, and clear understanding of the world’s renewable energy products trade flows and patterns. Global renewable energy products trade reflects trade flows among economies, which can be mapped as a network by viewing the economies as the nodes and trade flows between the economies as the edges. Finally, this paper applies the ERGM model to analyze the factors that influence the evolution of renewable energy product trade networks. It explores the effects of environmental regulations and climate change agreement networks on the network strength and stability in order to provide empirical evidence for a deeper understanding of the mechanisms that shape the global renewable energy product trade networks.

The rest of the paper is organized as follows. Section 2 conducts a comprehensive literature review on global renewable energy products and trade networks. Section 3 is devoted to the theoretical mechanisms and hypothesis on influencing factors of renewable energy products trade network. Section 4 describes the methodologies, including the process of constructing global renewable energy products trade networks and the ERGM model. Section 5 presents the findings of the analysis of the global renewable energy product trade networks and explores the network formation using ERGM. Conclusions and policy implications are drawn in Section 6.

2. Literature Review

Previous studies on renewable energy products focused mainly on the relationships between renewable energy development and CO₂ emissions [32,33], environmental pollution [11,34], energy supply and energy security [35], economic development [36], and other social benefits [37]. In addition, there have been systematic and in-depth explorations of regional renewable energy development [38,39]. Few academic articles study the global trade of renewable energies. The issue of renewable energy products trade is not broadly covered. Some of them focused on the current state of renewable energy product exports and related influencing factors, the measurement of export competitiveness, and the study of micro-enterprises [38,39,40]. Others subdivided renewable energy products and studied trade in solar, wind, or hydro power energy industries and products [41,42,43,44]. However, these studies pertaining to the renewable energy products trade pattern lacked comprehensiveness, focusing on specific economies or relying on the perspective of a single type of product. While the renewable energy industry has changed and trade flows in renewable energy products are large and growing, studies that comprehensively and systematically analyze the global trade in renewable energy products are rare.

Complex network analysis methods have been widely used in biology, physics, mathematics, computer science, sociology, economics, management, and library intelligence [45,46,47]. Guided by the complex network theory, several have applied this approach to the fields of international trade [48,49,50,51]. Global trade in renewable energy products reflects the trade flows between economies, which can be graphed as a network by considering the economies as nodes and the trade flows between economies as edges. In recent years, complex networks have also been applied to trade involving renewable energy products, such as solar energy-related product trade networks [26,27], wind energy product trade networks [52,53], and hydro energy product trade networks [54]. However, there is a lack of studies on the evolutionary patterns and structural characteristics of the overall global trade in renewable energy products from a network perspective, mostly focusing on the study of a segmented renewable energy product trade network. More importantly, most of these studies analyze the overall characteristics and evolutionary patterns of trade through descriptive statistics of network indicators and fail to explain the formation of their network patterns [55].

As complex network analysis methods continue to mature and grow, some scholars no longer limit themselves to descriptive statistical analysis of the constructed networks but involve network model building and fitting problems, focusing on network hypothesis testing and network model fitting. A large number of researchers have used Quadratic Assignment Procedure (abbreviated as QAP) [56,57], Fuzzy logic [58,59], etc., for model fitting. Among them, the QAP model is widely used to explore the factors and mechanisms that influence the formation and evolution of international trade networks. The QAP method is a good way to examine the influence of national attributes on international web-based networks, but these models are only linear and nonlinear fits using traditional statistical methods, while the most important feature of social networks is the network structure. Therefore, in order to analyze the structural characteristics of networks, exponential random graph models (ERGM) have been applied to the analysis of complex networks [56,57,58,59,60,61]. The ERGM model can take into account network structure and relationship variables at multiple levels simultaneously and consider the causes of network structure evolution in a comprehensive manner [60]. Therefore, we can utilize the ERGM model to investigate why the global renewable energy trade network has emerged and why its structure has changed. At present, the model mainly uses Markov Chain Monte Carlo maximum likelihood estimation to estimate the parameters, which can identify, test, and explain the relationships and structural characteristics of the network in a statistical form similar to that of logistic regression [61].

3. Theoretical Mechanisms and Hypothesis on Influencing Factors of Renewable Energy Products Trade Network

Economies that bring their own comparative advantages, resource endowments, and implementation policies are also important for the formation of global renewable energy product trade networks [62], which is known as the economic attribute effect. For the formation and evolution of global renewable energy products trade networks, economic attribute effects include not only congruence but also sender effects and receiver effects. Assortativity is the tendency of individuals to interact with others with similar characteristics [63], while the sender effect and receiver effect are attributes that may encourage individuals to be more proactive (expressed as higher out) and more popular (expressed as higher in), respectively [64].

The new trade theory believes that the motivation and basis of international trade can be analyzed from different perspectives, such as supply and demand. As the economic scale of the exporting country (region) becomes larger and larger, it is easier for manufacturers to achieve economies of scale, thereby reducing production costs, lowering export prices, and increasing the export volume. The larger the economic scale of the importing country (region), the greater the demand for product diversity and quantity. According to Krugman’s home market effect theory [65], as the economic scale of the exporting country (region) becomes larger and larger, its production and demand capacity become great. Domestic consumers’ demand for diversified products continues to increase, stimulating countries and enterprises to increase investment in R&D and innovation of the product. Thus, the scale of the industry has considerably expanded to form economies of scale, reduce production costs, and produce in large quantities, forming a virtuous circle with product exports.

Given the particularity of renewable energy products, this study considers both the demand- and supply-side of the renewable energy sector in proposing its hypotheses:

 Hypothesis 1.

Countries or regions with a high level of economic development and high consumption of renewable energy are inclined to trade renewable energy products with each other.

 Hypothesis 2.

Countries with high renewable energy production are inclined to export and import renewable energy products, reflecting the receiver and sender effects.

There are two classic hypotheses about the impact of environmental regulation on a country’s industrial competitiveness: the pollution haven hypothesis and the Porter hypothesis. The pollution haven hypothesis believes that strict environmental regulations reduce export volume [66]. The Porter hypothesis believes that countries with strict environmental regulations can profit from environmental benefits and obtain comparative advantages. Fu and Wu [67] regard the CO₂ emission intensity of importing and exporting countries as an indicator of environmental regulation. The lower the CO₂ emission intensity is, the stronger the environmental regulation will be. The research results show that countries with strict environmental regulations profit from the export of solar and wind energy industries, motivating them to export these renewable energies. Based on this, Hypothesis 3 is proposed.

 Hypothesis 3.

Countries with weak environmental regulations are inclined to export renewable energy products, reflecting the sender effect.

The institutional agreement network is established by signing free trade agreements among countries or regions to reduce trade costs and promote free trade. Wu [68] considers three dimensions (i.e., the social, cultural, and institutional agreement relationship network) to select the humanistic relationship network in his study about the influence of the humanistic relationship network on the international trade network. In addition, the transaction cost theory proposed by Coase [69] attaches importance to the effect of institutional agreement relationships on trade to a certain extent. The institutional agreement network reduces the cost of negotiation information and barriers to communication and trade, promoting the trade of renewable energy products. Therefore, Hypothesis 4 is as follows:

 Hypothesis 4.

The climate change agreement network paves the way for establishing a renewable energy trade network.

Geographical distance has a significant impact on trade between economies. From theoretical and empirical analysis, Eaton and Kortum [70] and Anderson and van Wincoop [71] show that trade volume is inversely related to geographical distance. If two neighboring countries trade with each other, they can significantly reduce transportation costs and increase profits. In addition, having a common border can often reduce the cost of information acquisition and facilitate trade between the two countries. Therefore, the Common Border Network (CBN) impacts trade and the embodied carbon flows between the two countries. Also, the “new economic geography” proposed by Krugman [72] supports the relationship between geographical location and economic trade. Therefore, economies with common geographical boundaries are more likely to trade renewable energy products. Based on this, Hypothesis 5 is proposed.

 Hypothesis 5.

Countries or regions having a common border network are inclined to establish trade relations for renewable energy products.

4. Methodology

4.1. Construction of Global Trade Networks for Renewable Energy Products

In this paper, we take the countries (regions) involved in the global trade of renewable energy products as the nodes of the network, the trade links between the two countries (regions) as the edges, and the trade volume of renewable energy between countries (regions) as the strength of links to construct complex network Model G. According to the description method of complex networks, let vector ${\mathrm{V}}_{\mathrm{i}}$ = [${\mathrm{v}}_{\mathrm{i}}$] (i = 1, 2, 3, …, n) and vector ${\mathrm{V}}_{\mathrm{j}}$ = [${\mathrm{v}}_{\mathrm{j}}$] (j = 1, 2, 3, …, n) be the exporting and importing countries, respectively. Adjacency matrix ${\mathrm{W}}_{\mathrm{ij}}$ = [${\mathrm{w}}_{\mathrm{ij}}$] (i = 1, 2, 3, …, n; j = 1, 2, 3, …, n) is used to represent the weighted edge for trade volume between ${\mathrm{V}}_{\mathrm{i}}$ and ${\mathrm{V}}_{\mathrm{j}}$. If there is export trade volume from country i to country j, then ${\mathrm{W}}_{\mathrm{ij}}$ = 1; otherwise, it is 0. ${\mathrm{V}}_{\mathrm{i}}$, ${\mathrm{V}}_{\mathrm{j}}$ and ${\mathrm{W}}_{\mathrm{ij}}$ constitute the directed weighted global renewable energy products trade network. The specific network representation is as follows:

(1)$G=(V_i,V_j,W_{ij},T)$

where T denotes a set of years of trade networks for renewable energy products.

We capture the features of the world renewable energy product trade network using such network statistics as the number of nodes (N), number of edges, average path length, degree and strength of nodes, clustering coefficient, and reciprocity coefficient (see Appendix A).

This paper utilizes bilateral trade data from the CEPII-BACI database, which provides disaggregated data on bilateral trade flows for more than 5000 products and 200 countries. It applies Jha’s classification method for renewable energy products, referring to the 6-digit HS codes it summarizes for a total of 95 commodities (See Table A1 of Appendix B) to filter the trade data on renewable energy products from 2009 to 2019. More than 200 countries are involved in trade, and the number of countries involved in trade varies slightly from year to year.

4.2. Exponential Random Graph Models (ERGM)

The ERGM, known as the P* (P-star) model, assumes a wide conditional dependency among the connections in the network. Specifically, suppose the other connections in the network have been determined; in that case, there is a conditional dependency between two specific links, and the conditional probability of the simultaneous existence of the two links is unequal to the product of the marginal conditional probabilities of the two links [73]. The statistical items that are considered in the P* model are not only for network features but also for attribute features of network members. These attributes in each given network include reciprocity, assortativity, and transitivity. It helps understand whether a given network can be formed by the local attribute characteristics of network members and the structural characteristics of the overall network.

Suppose that there are n nodes in the network: V = {1, 2, 3, …, n}, M = {(i, j); i ∈ V, i ≠ j} are all possible edges between nodes. G = (V, E) is the actual network, and E is the edge of the real network, then E is a subset of M. First, create a random variable, Y, to represent the elements of M. When (i, j) ∈ E, ${\mathrm{y}}_{\mathrm{i},\mathrm{j}}=1$; otherwise, ${\mathrm{y}}_{\mathrm{i},\mathrm{j}}=0$. Second, build a random adjacency based on the variable Y, matrix y = ${\mathrm{y}}_{\mathrm{i},\mathrm{j}}$. All random adjacency matrices constitute the feasible set Y of the network adjacency matrix. The general form of ERGM type is as follows:

(2)$P\left(Y=y\right)=(\frac{1}{c})exp\{{{\displaystyle \sum }}_{k=1}^k{\theta }_k{z}_k\left(y\right)\}$

The formula can be expressed as

(3)$Pr\left(Y=y|\theta \right)=P_{\theta }(\mathrm{y})=(\frac{1}{c})exp\{{\theta }^{T}z\left(y\right)+{\theta }_a^T{z}_a\left(y,x\right)+{\theta }_b^T{z}_b\left(y,h\right)\}$

$\mathrm{Pr}\left(\mathrm{Y}=\mathrm{y}|\mathsf{\theta }\right)$ is the explained variable, representing the probability of the network appearance; $\mathrm{z}\left(\mathrm{y}\right)$ represents the pure structural effect, that is, the statistics of the structural characteristics of the network itself; ${\mathrm{z}}_{\mathrm{a}}\left(\mathrm{y},\mathrm{x}\right)$ is the network statistics after adding node attributes; ${\mathrm{z}}_{\mathrm{b}}\left(\mathrm{y},\mathrm{h}\right)$ shows the network statistics after adding the external covariates network; and ${\mathsf{\theta }}^{\mathrm{T}}$ is the parameter vector of the three network structure statistics. If these parameters are statistically significant, the three network structure statistics significantly impact the formation and establishment of a specific network, y. If the estimated value of this parameter is positive, it indicates that under the control of other conditions, the probability of this kind of structure appearing in the network is more than random expected. If the estimated value of this parameter is negative, it indicates the opposite. $\frac{1}{\mathrm{c}}$ is a standardized constant, mainly used to ensure that the model has an appropriate probability distribution.

4.3. ERGM Variables and Data

The ERGM model mainly estimates the influence of different factors on network formation, including endogenous and exogenous structural variables, actor-relational variables, and network covariates.

As shown in Table 1, the endogenous structural variables mainly select edges and reciprocity. Edges play an intercept effect in the ERGM model, that is, the basic tendency or probability of the occurrence in the network structure under the condition that other variables are controlled. Especially in the renewable energy product network, it refers to the trade possibility of renewable energy products among countries if other factors remain unchanged. The reciprocity effect is common in most directed networks, which we also considered. Since it is an endogenous structural variable of the network, no assumption is made.

Actor-relational attribute variables are mainly measured by homophily (Hypothesis 1) and the sender and receiver effect (Hypothesis 2 and Hypothesis 3). In terms of homophily, to test whether countries with the same economic level are more prone to trade renewable energy products, we divide the GDP of each country into three categories: high (GDPhigh), medium (GDPmid), and low (GDPlow). Among more than 200 countries, GDPhigh, GDPmid, and GDPlow show the top 25%, 25~75%, and bottom 25%, respectively. Similarly, the renewable energy consumption of various countries is in three categories: REhigh, REmid, and RElow. The sender effect indicates whether a node with a particular attribute is more active than a node without this attribute and establishes more outgoing connections. The receiver effect can be explained in the same way, that is, whether a node with a particular attribute has established more inbound connections. This paper uses per capita emissions of greenhouse gases to measure environmental regulations. The smaller the emissions, the stronger the environmental regulations.

Network covariates represent the impact of external environmental networks on the global trade of renewable energy products (Hypothesis 4 and Hypothesis 5). If the estimated parameter of this indicator is significantly positive, countries or regions with embedded covariates networks tend to build relations for trading renewable energy products. Data sources of these variables are listed in Table 2.

5. Results

5.1. Evolution of Structural Properties of Global Renewable Energy Products Networks

5.1.1. Global Network Analysis

Descriptive statistics of the network are calculated for the years 2009–2019 to facilitate the network evolution analysis of renewable energy products trade, as shown in Table 3. The density of this trade network varied in stability between 2009 and 2019, and density is the ratio of observed relationships (also called edges) in a network to the maximum number of possible relationships. Thus, the density level ranges from 0 to 1. The closer the density level is to 1, the more connected the network is. From 2009 to 2017, the network’s density gradually increased, rising from 0.268 to 0.304, and there was some fluctuation in network density from 2017 to 2019; overall, the network’s density remained above 0.26 over these 11 years, and the network was somewhat closely connected. The number of nodes in the network has remained above 220, indicating that most countries or regions of the world are involved in the trade of renewable energy products. The number of edges in the network fluctuated from 37,089 in 2009 to 36,676 in 2019, with the number of edges in the network reaching 37,199 in 2012 over 11 years and then dropping to 35,689 in 2017 before increasing again. The average path length of the renewable energy product trade network ranges from 1.7110 to 1.7560 and shows minor fluctuations over this period, which suggests that the chain of renewable energy product trade between countries or regions has been relatively stable, indicating that any country (or region) i trading renewable energy products with another country j is no more than two steps away. The clustering coefficient of the global trade network of renewable energy products remains about 0.7 or above, and the closer the clustering coefficient is to 1, the more clustering in the trade network of renewable energy products there is. In addition, if a network has a small average path length and a large clustering coefficient, the network has small-world characteristics, and it is inferred that the trade network of renewable energy products has small-world characteristics. The reciprocity coefficient of this network is about 0.67, showing that countries prefer the bilateral trade of renewable energy products.

5.1.2. Local Network Analysis

In this paper, we define the out-degree and in-degree as node attributes, with the out-degree representing the number of countries a country exports to and the in-degree representing the number of countries that a country imports from. According to Table 4, as the out-degree of the global renewable energy products trade networks is concerned, in 2009, the top 10 renewable energy products exporters were Germany, the USA, China, Japan, France, the UK, Italy, Canada, The Netherlands, and Finland. While in 2019, the top 10 exporters were China, Germany, the USA, The Netherlands, the UK, France, Italy, Spain, Belgium, and India. In the past 11 years, overall, the number of export partners of the world’s top 10 exporters are fluctuating and increasing and is above 185, and some countries’ rankings have changed significantly. For example, Japan ranked in the top 10 in 2009 and then gradually left the list, and India’s ranking also improved, achieving tenth place in 2019.

China’s out-degree has been maintained at above 200 and has also remained among the top 3 in the world, indicating that China’s renewable energy products are exported to a broad collection of countries and regions and that China is a strong participant in the world trade of renewable energy products. In terms of in-degree, the top 10 countries with the number of export partners in 2009 were Germany, France, Canada, the United States, The Netherlands, Mexico, the UK, South Africa, China, and Belgium. In 2019, the top 10 countries on this indicator were Canada, France, the United States, Germany, South Africa, The Netherlands, the UK, Singapore, China, and Spain (See Table 5). As a whole, the number of import partners of the top 10 importers is less than the number of export partners of the top 10 exporters, and countries’ compositions are also more variable, showing that the imported networks of renewable energy products are looser than the corresponding export networks.

The out (in)-strength, on the other hand, differs from the out (in)-degree, with the out-strength and in-strength, respectively, denoting the total amount of renewable energy products exported by a country and the total amount of renewable energy products imported by a country. In terms of out-strength, the top 10 countries in the export volume of renewable energy products were China, Germany, the United States, Japan, etc., as shown in Table 6. The comparison of out-degree shows that the countries with the largest export partners of renewable energy products are not necessarily the largest exporters of renewable energy products, such as South Korea, Denmark, and Malaysia. This indicates that the markets that these countries export renewable energy products to are less extensive than those of other top-rank countries, but the export concentrations are relatively high. The global trade volume of renewable energy products is increasing, and the market is expanding. China’s out-strength ranks first in the world, and its export value is also increasing, meaning that China is the world’s largest importer of renewable energy products. The sum of renewable energy product exports of the top 10 countries from 2009 to 2019 accounted for 60% and more of the total global renewable energy product exports of these years, showing that the network presents a pronounced core-periphery feature. In terms of in-strength, shown in Table 7, the top countries were China, Germany, the United States, and Japan. China’s import volumes ranked first or second in these 11 years and maintained growth from 2009 to 2018, increasing from USD 27.7 billion to USD 42 billion, but its import volumes are lower than export volumes in all cases, indicating a large surplus in China’s trade in renewable energy products.

We also use weighted network data to construct a continuous core-periphery model to analyze core-periphery changes in the global international trade network for renewable energy products for 2009–2019. If we assume the trade network with core-periphery structure as a single point cloud in Euclidean space, “coreness” is the distance between node countries (regions) and the centroid of the single point cloud. In the economic sense, coreness mainly reflects the control and influence of a country (region) on other countries (regions) in the trade network. The greater the coreness, the greater the country’s participation in the trade of renewable energy products and the more important its position in the whole network. In the continuous model, each point needs to be given a measure of coreness. If the items in the network are continuous data and represent the strength of the relationship, core-periphery can be used to evaluate the degree of fitting, but the following ideal structural matrix should be constructed:

(4)${\delta }_{ij}=c_i{c}_j$

where c is a non-negative vector representing the degree of the core at each point. This allows each point to be divided into core, semi-core, and peripheral regions. Running the Continuous command in Core/Periphery through the Ucinet6.0 software assignment matrix gives an estimated annual intensity Core for each country. We measure each country’s participation in the international trade market for renewable energy products expressed quantitatively in terms of coreness. Countries with coreness values of 0.1 or more are classified as core regions, those with coreness of 0.01–0.1 are classified as semi-core regions, and those with coreness values of less than 0.01 are classified as peripheral regions [64]. The number of countries in each classification, as shown in Table 8, remained at 7 or 8 core countries from 2009 to 2019; the number of semi-core countries remained between 25 and 29, and the number of peripheral countries was approximately 190. To further investigate the core countries of the global renewable energy products trade network of this period, we study the three countries with the highest core degrees (China, Germany, and Japan) in a detailed analysis (see Figure 1). China’s overall coreness increases and then decreases and then maintains a stable trend, increasing from 0.594 in 2009 to 0.8 in 2010 and then decreasing and remaining at 0.65 year-round. Overall, China’s coreness has always ranked first in the world, and its coreness value maintains a certain gap with those of Germany and Japan, which ranked 2nd and 3rd, respectively. Figures of network structures for 2009 and 2019 show that China’s core position in the global renewable energy products trade network improved over the 11 years, but it is noteworthy that China’s coreness is also declining (See Appendix C).

5.2. Influencing Factors of Renewable Energy Products Trade Network

5.2.1. Basic Results of ERGM Estimation

We take the international trade network of renewable energy products in 2019 as an example to analyze the results. Table 9 shows that in the case of considering only pure network effects and reciprocity (model 1). In model 1, the coefficient of the statistical term of Edges is negative, showing that the density of the network is below 50%, which confirms the network density in Section 5.1.1; the coefficient of Reciprocity is significant and as high as 2.75 at 0.1%, indicating that the international trade network of renewable energy products has high reciprocity, and bilateral trade tends to occur between countries, which is also consistent with the previous analysis in Section 5.1.1. Although the pure network effects model can well characterize the density characteristics of the observed network, it does not reveal other characteristics of the network; therefore, we consider adding node attributes of the network for further analysis. Therefore, Model 2 brings GDP and renewable energy consumption to the total energy consumption into the analysis. The results show that the coefficients of both a country’s GDP and renewable energy consumption share are significant at 0.1%, indicating that both the level of a country’s economic development and the scale of renewable energy consumption is strongly associated with the formation of a world trade network for renewable energy products.

We test model 3 based on the homophily of GDP and the proportion of renewable energy consumption in countries (regions). The statistical coefficient of Homophily (GDPhigh) is significantly negative at the 5% level, which indicates that countries with high GDP are not so inclined to trade in renewable energy products among themselves, and the trade among them is dishomophily; the statistical coefficient of Homophily (GDPmid) is positive but not significant, which indicates that countries with medium GDP do not tend to trade renewable energy products with countries at a similar level of development to their own. The coefficient of Homophily (GDPlow) is positive and significant at 0.1%, which indicates that countries with low GDP levels tend to generate renewable energy products and trade with each other. The coefficient of the statistical term of Homophily (REmid) is significantly negative, which indicates that the trade of renewable energy products among countries with medium renewable energy consumption is dishomophily, and they do not tend to trade with each other. Homophily (RElow) is significantly positive at a 0.1% confidence interval, which indicates that countries with low renewable energy consumption tend to trade renewable energy products with each other and have some dishomophily. Overall, the international trade of renewable energy products exhibits some heterogeneity, which confirms the previous Core/Periphery structure in Section 5.1.2, i.e., the central country both exports and imports renewable energy products and is more active and central in the network. The AIC and BIC of model 3 are lower than those of both model 1 and model 2, indicating that model 3 is optimized on the basis of model 1 and model 2.

We further consider the receiver effect of GDP and the proportion of renewable energy consumption in model 4. The results show that the coefficients of reciprocity are significant at 0.1%, indicating high reciprocity of the international trade network of renewable energy products and motivation of countries to make bilateral trade, which is consistent with the previous analysis in Section 5.1.1. In model 4, the Homophily coefficients of countries with high GDP are positive and statistically significant at the 0.1% level, while it is negative in model 3. Other results of homophily are also more significant than those in model 3. The coefficient of the receiver (GDPhigh) is positive and significant, indicating that countries with higher GDP levels have high inbound connections. In addition, the coefficients of the receiver (GDPmid and GDPlow) are positive and significant at the 5% level, which indicates that countries with medium and low GDP levels are also popular in the network. Especially for countries with a medium GDP level, they show an obvious receiver effect. Bai (2018) [74] and Qi et al. (2021) [75] also use GDP as a measure of a country’s economic development and use the QAP model to explore the impact of a country’s economic development on the evolution of the trade network structure of renewable energy products, and their results both conclude that there is a significant impact of economic development on the formation of the trade network structure of renewable energy products, and countries with higher economic development are more likely to export renewable energy products. However, the results of this paper are not completely consistent compared with their results. The ERGM model used in this paper is better from different levels of data, respectively, to explore the differences that exist in the impact of different income levels on the pattern of import and export trade rather than generalizing that economic levels have an impact on the global trade of renewable energy products. The coefficient of the receiver (REhigh) is significantly positive at the 0.1% level, indicating that countries with a high renewable energy consumption are more popular in the trade network of renewable energy products and have more inbound connections, showing a noticeable receiver effect. Receiver (REmid) is significantly negative at the 0.1% level, indicating that countries with a moderate proportion of renewable energy consumption are not very popular in the consumption network of renewable energy products. The coefficient of the receiver (RElow) is significantly positive at the 0.1% level, indicating that the countries with low renewable energy consumption are inclined to import renewable energy products. The results of model 4 show that the impact of renewable energy consumption on the formation of renewable energy product trade networks is more direct, but the AIC and BIC of this model increase compared to model 3, and this model is not more optimal than model 3.

Then, we take environmental regulations measured by per capita greenhouse gas emissions and the sender effect of renewable energy production into the model (see model 5 and model 6). The estimated coefficients of greenhouse gas emissions (ghg) and renewable energy production are positive and statistically significant at the 0.1% level. It indicates that countries with high greenhouse gas emissions per capita are active in the network and inclined to export renewable energy products. The results are similar to Bai (2018) [74], showing that the greater the difference in CO₂ emissions between countries, the greater their trade flows. It implies that countries with weak environmental regulations export high renewable energy products, and countries with high renewable energy production tend to export high renewable energy products, showing an obvious sender effect. The AIC and BIC of both model 5 and model 6 are reduced compared to the previous model, and the model that takes into account the sender effect is more optimized.

Finally, we add network covariates, including the climate change agreement network and the common border network, to study the influence of exogenous networks on the trade network of renewable energy products (See model 7). It turns out that their coefficients are positive and statistically significant at 0.1% level, which indicates that the countries that have signed the Paris Agreement are more likely to trade renewable energy products under the control of other factors, and countries with common borders are inclined to trade renewable energy products. The results are similar to Qi et al. (2021) [75], where most renewable energy products trade between economies with geographical proximity. Although consistent findings are obtained, Qi et al. (2021) assumes that economies’ trade in renewable energy products is independent of each other and does not consider the complexity of network structure. The AIC and BIC of model 7 are smaller than those of model 6, indicating that the model has been further optimized. Model 8 takes into account the node attributes, homophily, sender effect, receiver effect, and network covariates, and the AIC and BIC are found to be further optimized based on model 7.

5.2.2. Dynamic Analysis

We analyze the trade network dynamic effects of renewable energy products based on model 8. We consider the contracting parties in Annex B of the “Kyoto Protocol” as the climate change agreement network in the network covariates during 2009–2015. Then, we consider contracting parties of the Paris Agreement signed in 2016 because it is stronger and newer than the previous “United Nations Framework Convention on Climate Change” and the “Kyoto Protocol”. In addition, countries mainly join this agreement to participate in global climate governance. Based on Table 10 and Table 11, the trade of renewable energy products shows homophily in countries with medium and low GDP levels from 2009 to 2018. Countries with relatively high consumption of renewable energy tend to trade renewable energy products with each other, showing homophily. Countries with low renewable energy consumption show homophily in 2011, 2012, 2016, and 2018, while the performance in other years is obscure.

The countries with higher GDP levels did not have more inbound connections in 2010 and 2012. On the contrary, in 2012, the countries with medium and low GDP levels were more popular in the network, especially countries with low GDP levels showing a more obvious receiver effect. In most years from 2013 to 2018, countries with high GDP levels in these years have high connections and are popular in the trade network of renewable energy products. At the same time, other countries also show a certain receiver effect. During 2009–2018, the countries with a relatively high proportion of renewable energy consumption were popular in the trade network of renewable energy products and had high inbound connections, showing an obvious receiver effect. Although countries with a moderate proportion of renewable energy consumption show an obvious receiver effect in 2009, 2010, 2017, and 2018, no obvious receiver effect exists in the rest of the year. The coefficient is significantly negative in 2014, which indicates that countries with a medium share of renewable energy consumption in this year are not very popular in the trade network of renewable energy products; the statistical coefficient of Receiver (RE low) is significantly positive at the level of 0.1%, which indicates that countries with low renewable energy consumption also import more renewable energy products.

Statistical coefficients for per capita greenhouse gas emissions (ghg) from 2009 to 2011 are significantly positive (0.1% confidence interval), showing that countries with high greenhouse gas emissions per capita in recent years are active in the network and inclined to export renewable energy products. It implies that countries with weak environmental regulations export a high volume of renewable energy products. This situation also existed in 2013 and 2017. In 2012, however, the situation changed, and countries with high per capita greenhouse gas emissions were disinclined to export renewable energy products in the network. Countries with strong environmental regulations exported a high volume of renewable energy products in 2012. There is no obvious sender effect in other years. The coefficients of renewable energy production from 2009 to 2018 are all significantly positive, which shows that countries with higher renewable energy production tend to export renewable energy products, showing an obvious sender effect.

On the exogenous network covariates, the statistical coefficients of the climate agreement networks are significantly positive except for 2009, which indicates that trade in renewable energy products is more likely to occur among the parties to either the Kyoto Protocol or the Paris Agreement. When examining the common border network of countries, the coefficient of the statistic coefficient of the common border network is significantly positive from 2011 to 2017, and in general, countries will consider transportation costs when trading renewable energy products.

5.2.3. Analysis of Each Kind of Renewable Energy Product

In this section, we study the international trade network by dividing renewable energy products into six sub-categories of solar, wind, biomass, water, ocean, and geothermal energy products according to HS codes. Table 12 shows the results of 2019.

We find that the economic development and renewable energy consumption of a country is of great significance to the formation of the segmented trade network of renewable energy products. When renewable energy products are further subdivided, the density of the network is lower than 50% and shows strong reciprocity. The trade network of each kind of renewable energy product also shows Core/Periphery characteristics. Countries with a high level of renewable energy consumption show obvious homophily in the trade of various products, while countries with a medium level of renewable energy consumption show homophily in the overall network of renewable energy products. At the same time, they show dishomophily in the trade of wind energy products, hydro energy products, and geothermal energy products after subdividing.

A country’s GDP development level and renewable energy consumption level are of great significance to the formation of a subdivided renewable energy products trade network. Countries with high GDP have a receiver effect on the trade of wind energy products, and at the same time, it has the nature of homophily, indicating that large countries are particularly active in the wind energy products market. The rest of the countries show no noticeable receiver effect on market segments, which is consistent with the estimation of the overall market. Although there is no sender effect in the biomass and water energy products, this effect exists in overall renewable energy products. Countries that have signed the Paris Agreement are inclined to trade solar, wind, hydro, and ocean energy products, but they are disinclined to trade biomass energy products and geothermal energy products. Countries with common borders are also inclined to trade solar energy products, wind energy products, and geothermal energy products, but they disincline to trade hydro energy products, biomass energy products, and ocean energy products. Generally, countries should consider the factors such as transportation costs when trading renewable energy products.

6. Conclusions and Policy Implications

This work builds a global trade network of renewable energy products based on the bilateral trade of renewable energy products from 2009 to 2019. It analyzes the evolution of its structure from the overall and node level and employs an exponential random graph model to analyze influencing factors of the trade network of renewable energy products.

The features of the evolution of renewable energy products are as follows: (i) Over the past 11 years, the network connections of renewable energy products have become tighter and tighter, and countries have generally participated in the trade of renewable energy products, showing the characteristic of a small world. (ii) The reciprocity of countries has also continued to be strengthened, and they are inclined to increase bilateral trade. (iii) The network shows obvious core-peripheral characteristics. China, Germany, and Japan have always been in the position of core countries. It confirms China’s tremendous achievements in reform and construction to a certain extent. Eleven years of continuous development in China have strengthened its core position, making it an emerging market country with a great voice in the international trade of renewable energy products.

By using ERGM to analyze the influencing factors of the trade network of renewable energy products from the pure structure effect of the network, the actor-relational attribute variable, and the network covariates. Specifically, during 2009–2019, GDP and renewable energy consumption have a remarkable correlation with the trade network of renewable energy products of the world. There is homogeneity in the trade of renewable energy products in the big GDP countries. The trade of renewable energy products in renewable energy-consuming countries also has homogeneity. They tend to trade more renewable energy products among themselves, which indicates that the trade network of renewable energy products shows some characteristics of the central periphery, with the central countries showing more active performance. Moreover, the large renewable energy-producing countries have more export trade partners. Countries with large renewable energy consumption have more import trading partners, which shows noticeable receiver effects. Countries with high per capita greenhouse gas emissions have high outward connections, indicating that countries with low environmental regulations are more likely to export renewable energy products. The climate change agreement network and the common border network have a specific impact on the trade network of renewable energy products. Countries sharing a common geographic boundary and signing climate change agreements are more likely to trade renewable energy products. The results above are similar to that of Qi and Bai, who use the QAP regression analysis method to consider partial influencing factors. This paper confirms their finding using ERGM capable of exploring network structural factors.

The above conclusions provide rich policy implications. First, the core countries in the trade network of renewable energy products should fully display their impacts and renewable energy resource advantages. They should continue to promote the transformation of energy consumption to renewable energy consumption, such as optimizing the layout of renewable energy development and improving infrastructure in the areas. Second, the climate change agreement is conducive to forming the world trade network of renewable energy products. This shows the importance of global climate governance in guiding the formulation of rules. Countries around the world should also jointly build a more fair and reasonable multilateral system that fulfills their responsibilities. They should create and share development opportunities through solidarity and cooperation and promote the building of a community with a shared future for mankind, relying on multilateral environmental conventions and related mechanisms.

In this paper, when using ERGM to explore the dynamic structural factors of the network, the model fails to incorporate data from different times into the same analytical model. Thus, the impact of temporal factors on the global renewable energy products trade network is still unclear. Two possibilities come to mind. One is that the global renewable energy products trade network is temporally stable, i.e., the current network structure characteristics do not change over time. The other possibility is that the network is temporally volatile, e.g., do unconnected nodes in the previous network have a tendency to be linked in the current network? Have the links in the previous network been unlinked and no longer exist in the current network? However, this investigation will be left for future research. Additionally, more work is needed to understand the mechanisms behind the evolution of each kind of renewable energy products networks. For each network, the covariates in attribute levels should be prepared separately because countries involved in each type of renewable energy are different. In general, the heterogeneity of impacting factors of global renewable energy products would seem to be an interesting subject for further research. In addition, future research directions could explore the possible economic consequences of the formation and evolution of global renewable energy product networks. For example, the impact of the evolution of global renewable energy product trade networks on global or regional energy security.

Author Contributions

Conceptualization, J.L. and K.L.; data curation, K.L.; formal analysis, J.L. and K.L.; funding acquisition, J.L.; investigation, Z.Y.; methodology, J.L. and K.L.; project administration, J.L. and Y.Q.; resources, Z.Y. and Y.Q.; software, K.L.; supervision, J.L. and Y.Q.; validation, J.L.; visualization, K.L.; writing—original draft, J.L. and Z.Y.; writing—review and editing, J.L. and Y.Q. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data on bilateral trade across the world comes from the CEPII-BACI database. These data can be found here: http://www.cepii.fr/CEPII/fr/bdd_modele/bdd_modele_item.asp?id=37 (accessed on 5 September 2022). Other data applied in this study can be downloaded freely. Their sources can be referred to in Table 2 in this paper.

Acknowledgments

We thank the editor and the reviewers. We also thank the reviewers and the participants of the International Conference on Climate and Energy Finance, Nanjing, China (2022), for their constructive comments and suggestions. The authors acknowledge the Humanities and Social Science Youth Foundation of the Ministry of Education of China for funding support.

Conflicts of Interest

The authors declare no conflict of interest.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figure and Tables

Figure 1. Coreness of China, Japan, and Germany.

Table 1

Influencing factors.

Variables	Implication	Schematic Diagram	Statistical Expression	Corresponding Hypothesis
Edges	edges	[Image omitted. Please see PDF.]	${\displaystyle {\displaystyle \sum }_{i,j}}{y}_{ij}$	Is the basic tendency to generate trade in renewable energy products between national sectors stronger? Is the network denser?
Reciprocity	reciprocity	[Image omitted. Please see PDF.]	${\displaystyle {\displaystyle \sum }_{i,j}}{y}_{ji}{y}_{ij}$	Are countries more inclined to have reciprocal trade?
Homophily(x)	homophily	[Image omitted. Please see PDF.]	${\displaystyle {\displaystyle \sum }_{i,j}}\left|x_i-x_j\right|y_{ij}$	Are national partners that all have X attributes more inclined to establish strong trade relations for renewable energy products?
Sender(x)	the sender effect	[Image omitted. Please see PDF.]	${{\displaystyle \sum }}_{i,j}{x}_i{y}_{ij}$	Are national partners with X attributes more active in the trade network of renewable energy products and have more connections?
Receiver(x)	the receiver effect	[Image omitted. Please see PDF.]	${\displaystyle {\displaystyle \sum }_{i,j}}{x}_j{y}_{ij}$	Are national partners with X attributes more popular in the trade network of renewable energy products and have more connections?
NCov(g)	network covariates	[Image omitted. Please see PDF.]	${\displaystyle {\displaystyle \sum }_{i,j}}{y}_{ij}{g}_{ij}$	Are countries with relationships in other networks more inclined to establish strong trade relations for renewable energy products?

Table 2

Actor-relational attributes and network covariates data sources.

	Variables	Sources of Data
actor-relational attributes	GDP	World Bank WDI database
	RE Production	International Renewable Energy Agency Database
	RE Consumption	UNdata
	ghg emission	UNdata
network covariates	Common Border Network(CBN)	CEPII (Gravdata)
	Network of Parties to the Paris Climate Agreement (2016–2019)Kyoto Protocol State Party Network(2009–2015)	Official Document of the United Nations Paris Agreement“Kyoto Protocol”-Annex B

Table 3

Descriptive statistics of the global renewable energy products trade network.

Year	Number of Nodes	Number of Edges	Density	Average Path Length	Clustering Coefficient	Reciprocity Coefficient
2009	222	37,089	0.268	1.756	0.733	0.67
2010	221	36,565	0.281	1.752	0.672	0.579
2011	224	36,788	0.278	1.743	0.692	0.668
2012	225	37,199	0.284	1.741	0.728	0.673
2013	225	36,783	0.291	1.733	0.656	0.567
2014	225	36,399	0.291	1.729	0.72	0.665
2015	225	36,092	0.3	1.716	0.723	0.668
2016	225	36,113	0.3	1.717	0.726	0.673
2017	225	35,689	0.304	1.711	0.725	0.671
2018	226	36,385	0.301	1.719	0.73	0.678
2019	224	36,676	0.29	1.728	0.751	0.692

Notes: Results are calculated via UCINET.

Table 4

Top 10 countries in terms of out-degree, 2009–2019.

2009		2011		2013		2015		2017		2019
Germany	212	USA	210	USA	211	Netherlands	209	Germany	210	China	209
USA	204	China	204	Germany	211	Germany	208	Netherlands	206	Germany	204
China	200	UK	201	Netherlands	207	China	206	China	205	USA	203
Japan	198	France	200	China	205	France	203	USA	204	Netherlands	203
France	195	Japan	197	France	202	USA	203	France	201	UK	200
UK	195	Germany	196	Italy	201	Switzerland	201	UK	201	France	199
Italy	190	Netherlands	196	Japan	198	UK	198	Belgium	197	Italy	197
Canada	189	Italy	192	UK	197	Japan	197	Italy	196	Spain	196
Netherlands	187	Canada	191	Belgium	194	Italy	197	India	195	Belgium	195
Finland	186	Finland	191	Canada	193	Sweden	196	Sweden	194	India	194

Notes: Results are calculated via UCINET. To save space, results of year 2010, 2012, 2014, 2016, and 2018 are not reported.

Table 5

Top 10 countries in the world by in-degree, 2009–2019.

2009		2011		2013		2015		2017		2019
Germany	168	France	172	Canada	175	Netherlands	173	Canada	180	Canada	181
Netherlands	163	Canada	167	Netherlands	174	Canada	171	France	178	France	178
Canada	163	USA	163	France	171	USA	168	USA	172	USA	175
USA	161	Netherlands	161	USA	164	France	163	Netherlands	170	Germany	160
France	153	Mexico	150	Somalia	151	UK	152	South Africa	156	South Africa	159
UK	146	UK	147	Mexico	149	China	152	UK	151	Netherlands	158
China	145	China	145	South Sudan	148	South Africa	152	China	145	UK	154
South Africa	142	Germany	142	UK	146	Germany	149	Spain	144	Singapore	153
Germany	138	Spain	141	Germany	144	Spain	147	Italy	142	China	152
Spain	135	South Africa	138	China	142	Italy	143	Belgium	141	Spain	147
Italy	127	Belgium	136	Sierra Leone	141	Thailand	143	Australia	140	Belgium	147

Notes: Results are calculated via UCINET. To save space, results of year 2010, 2012, 2014, 2016, and 2018 are not reported.

Table 6

Top 10 countries in the world in terms of out- strength, 2009–2019.

2009		2011		2013		2015		2017		2019
China	38.8	China	71.1	China	64.1	China	68.9	China	69.3	China	80.6
Germany	32.4	Germany	41.9	Germany	34.9	Germany	29.6	Germany	31.0	Germany	33.1
USA	21.8	USA	28.4	USA	28.0	USA	28.1	USA	25.5	USA	26.0
Japan	21.6	Japan	27.4	Japan	23.0	Japan	20.2	Japan	20.3	Japan	20.3
Italy	11.1	Italy	12.9	Italy	13.0	Korea	12.3	Korea	12.4	Italy	12.0
France	9.4	Korea	11.8	Korea	11.7	Italy	11.2	Italy	11.5	Korea	10.8
Korea	8.2	France	10.5	France	9.2	Malaysia	8.9	Malaysia	9.0	Malaysia	9.3
Denmark	6.6	Malaysia	7.8	Malaysia	7.3	France	8.2	France	8.3	France	8.2
UK	5.5	Denmark	7.2	UK	5.9	Mexico	6.4	Mexico	5.8	Mexico	7.1
Netherlands	4.9	Netherlands	6.9	Netherlands	5.9	Spain	5.4	Spain	5.7	India	6.5

Notes: Results are calculated via UCINET. To save space, results of year 2010, 2012, 2014, 2016, and 2018 are not reported.

Table 7

Top 10 countries in the world in terms of in-strength, 2009–2019.

2009		2011		2013		2015		2017		2019
China	27.8	China	37.3	China	35.5	China	42.7	China	40.8	USA	46.6
Germany	25.2	Germany	34.3	Germany	23.0	Germany	21.7	Germany	22.2	Germany	23.1
USA	17.3	USA	24.4	USA	22.6	USA	19.7	USA	21.2	China	21.0
Japan	8.2	Japan	16.7	Japan	14.3	Japan	13.3	Japan	11.7	Hong Kong, China	11.4
Italy	8.1	Italy	13.2	Italy	13.8	Korea	12.8	Korea	11.2	Japan	11.1
France	8.0	Korea	11.8	Korea	11.1	Italy	10.0	Italy	9.3	Netherlands	10.5
Korea	7.4	France	10.8	France	9.2	Malaysia	9.3	Malaysia	9.2	Mexico	9.8
Denmark	6.7	Malaysia	10.0	Malaysia	8.4	France	9.1	France	8.9	Korea	9.4
UK	6.6	Denmark	9.1	UK	8.1	Mexico	7.4	Mexico	8.3	France	9.1
Netherlands	6.2	Netherlands	8.6	Netherlands	7.9	Spain	7.2	Spain	7.9	UK	9.0

Notes: Results are calculated via UCINET. To save space, results of year 2010, 2012, 2014, 2016, and 2018 are not reported.

Table 8

Top 10 countries in the world in terms of in-intensity, 2009–2019.

	2009	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019
core	8	8	7	7	7	7	7	8	8	8	8
semi-core	25	24	26	29	29	28	26	26	26	27	26
peripheral	189	189	191	189	189	190	192	191	191	191	190

Notes: Results are calculated via UCINET.

Table 9

Results of ERGM model.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8
Edges	−1.9396 ***	−4.8100 ***	−6.8865 ***	−4.7135 ***	−4.8229 ***	−5.3275 ***	−5.4821 ***	−7.0536 ***
	(0.0171)	(0.0605)	(0.1971)	(0.1280)	(0.1246)	(0.1456)	(0.1155)	(0.2013)
Reciprocity	2.7497 ***	1.6119 ***	1.5767 ***	2.4157 ***	2.3952 ***	2.3531 ***	2.2336 ***	1.9931 ***
	(0.0343)	(0.0392)	(0.0406)	(0.0424)	(0.0498)	(0.0480)	(0.0431)	(0.0445)
nodecov.GDPhigh		0.2099 ***	0.2115 ***					0.2360 ***
		(0.0028)	(0.0048)					(0.0058)
nodecov.GDPmid		0.1434 ***	0.1665 ***					0.1322 ***
		(0.0028)	(0.0045)					(0.0058)
nodecov.GDPlow		0.0712 ***	0.3084 ***					0.2332 ***
		(0.0068)	(0.0240)					(0.0252)
nodecov.RE high		−0.0044 ***	−0.0068 ***					−0.0076 ***
		(0.0003)	(0.0007)					(0.0009)
nodecov.RE mid		0.0020 *	−0.0060 ***					0.0061 ***
		(0.0008)	(0.0010)					(0.0016)
nodecov.RE low		−0.0149 **	0.0283 ***					0.0503 ***
		(0.0052)	(0.0062)					(0.01)
Homophily (GDPhigh)			−0.1046 *	−2.0667 ***	−2.0568 ***	−1.9405 ***	−1.9395 ***	−0.0972
			(0.0470)	(0.0378)	(0.0485)	(0.0341)	(0.0347)	(0.0769)
Homophily (GDPmid)			0.0363	0.6907 ***	0.6784 ***	0.6942 ***	0.7839 ***	0.2982 ***
			(0.0343)	(0.0404)	(0.0559)	(0.0402)	(0.0345)	(0.0540)
Homophily(GDPlow)			−0.4686 **	1.6290 ***	1.6054 ***	1.5618 ***	1.6905 ***	1.6618 ***
			(0.1439)	(0.0839)	(0.0823)	(0.0879)	(0.0809)	(0.1722)
Homophily (REhigh)			−0.2438 ***	0.4848 ***	0.3786 ***	0.4062 ***	0.4170 ***	0.0186
			(0.0421)	(0.0392)	(0.0392)	(0.0343)	(0.0353)	(0.0531)
Homophily (REmid)			−0.1633 ***	−0.2530 ***	−0.2663 ***	−0.2680 ***	−0.2789 ***	−0.1087 **
			(0.0273)	(0.0355)	(0.0462)	(0.0368)	(0.0289)	(0.0380)
Homophily (RElow)			0.4155 ***	0.2998 ***	0.3824 ***	0.3716 ***	0.3769 ***	0.5389 ***
			(0.0266)	(0.0394)	(0.0424)	(0.0434)	(0.0295)	(0.0367)
Receiver (GDPhigh)				0.0807 ***	0.0775 ***	0.0918 ***	0.0849 ***	−0.0841 ***
				(0.0051)	(0.0052)	(0.0056)	(0.0042)	(0.0057)
Receiver (GDPmid)				0.2445 ***	0.2370 ***	0.2455 ***	0.2325 ***	0.0228 **
				(0.0063)	(0.0075)	(0.0079)	(0.0061)	(0.0074)
Receiver (GDPlow)				0.4247 ***	0.4135 ***	0.4170 ***	0.4072 ***	0.0959 ***
				(0.0188)	(0.0211)	(0.0209)	(0.0184)	(0.0192)
Receiver (RE high)				0.0027 ***	0.0018 *	0.0019 *	0.0011	0.0035 ***
				(0.0008)	(0.0008)	(0.0009)	(0.0008)	(0.0010)
Receiver (RE mid)				−0.0152 ***	−0.0153 ***	−0.0151 ***	−0.0185 ***	−0.0250 ***
				(0.0016)	(0.0018)	(0.0018)	(0.0015)	(0.0023)
Receiver (RE low)				0.0120	0.0244 *	0.0192	0.0086	−0.0429 **
				(0.0106)	(0.0106)	(0.0109)	(0.0089)	(0.0155)
Sender(ghg)					0.0287 ***	0.0305 ***	0.0287 ***	0.0229 ***
					(0.0020)	(0.0019)	(0.0017)	(0.0019)
Sender(REP)						0.0603 ***	0.0499 ***	0.0082 *
						(0.0048)	(0.0043)	(0.0041)
NCov (agreement)							0.4991 ***	0.2898 ***
							(0.0237)	(0.0253)
NCov (CBN)							0.6594 ***	0.5873 ***
							(0.1155)	(0.1134)
AIC	52,935.2033	39,469.2773	39,121.8544	40,969.9747	40,635.8601	40,334.4190	39,792.9552	37,156.1451
BIC	52,952.859	39,539.8992	39,245.4429	41,093.5631	40,768.2763	40,475.6630	39,951.8547	37,368.0110
Log Likelihood	−26,465.6	−19,726.6386	−19,546.9272	−20,470.9873	−20,302.9301	−20,151.2095	−19,878.4776	−18,554.0726

Notes: Standard deviations are reported in parentheses. ***, **, and * represent statistical significance at the 0.1%, 1%, and 5% levels, respectively.

Table 10

ERGM results of renewable energy product trading network of 2009–2014.

Variables	2009	2010	2011	2012	2013	2014
Edges	−7.5181 ***	−7.0474 ***	−7.7539 ***	−8.2897 ***	−5.3742 ***	−8.1608 ***
	(0.2100)	(0.1872)	(0.2398)	(0.0060)	(0.1497)	(0.1972)
Reciprocity	1.9274 ***	1.1840 ***	1.7183 ***	1.7452 ***	2.2620 ***	1.6777 ***
	(0.0494)	(0.0453)	(0.0476)	(0.0011)	(0.0435)	(0.0473)
nodecov.GDPhigh	0.1995 ***	0.2014 ***	0.2099 ***	0.2824 ***	0.0962 ***	0.2067 ***
	(0.0078)	(0.0076)	(0.0078)	(0.0020)	(0.0062)	(0.0082)
nodecov.GDPmid	0.1385 ***	0.1396 ***	0.1250 ***	0.2065 ***	0.0350 ***	0.1503 ***
	(0.0081)	(0.0078)	(0.0077)	(0.0017)	(0.0059)	(0.0085)
nodecov.GDPlow	0.2510 ***	0.2405 ***	0.2609 ***	0.2774 ***	0.1161 ***	0.1318 ***
	(0.0241)	(0.0217)	(0.0298)	(0.0036)	(0.0179)	(0.0221)
nodecov.RE high	−0.0152 ***	−0.0115 ***	−0.0139 ***	−0.0121 ***	−0.0131 ***	−0.0127 ***
	(0.0009)	(0.0009)	(0.0009)	(0.0006)	(0.0008)	(0.0009)
nodecov.RE mid	−0.020 ***	−0.008 ***	−0.0160 ***	−0.0032 *	−0.0096 ***	−0.0014
	(0.0015)	(0.0015)	(0.0016)	(0.0013)	(0.0014)	(0.0015)
nodecov.RE low	−1.0871 ***	−0.4713 ***	−0.7354 ***	−0.6509 ***	−0.6893 ***	−0.7350 ***
	(0.0721)	(0.0476)	(0.0724)	(0.0017)	(0.0436)	(0.0548)
Homophily (GDPhigh)	−0.0137	0.0188	−0.0129	−0.2345 ***	−0.1830 **	−0.2491 *
	(0.0831)	(0.0761)	(0.0897)	(0.0035)	(0.0631)	(0.1074)
Homophily (GDPmid)	0.2154 ***	−0.0298	0.1342 *	0.2965 ***	0.0348	0.3849 ***
	(0.0637)	(0.0594)	(0.0640)	(0.0019)	(0.0474)	(0.0822)
Homophily(GDPlow)	1.3452 ***	1.3201 ***	1.4457 ***	1.7260 ***	0.9433 ***	1.4750 ***
	(0.1704)	(0.1521)	(0.2080)	(0.0042)	(0.1244)	(0.1644)
Homophily (REhigh)	0.5258 ***	0.5480 ***	0.4412 ***	0.3320 ***	0.4955 ***	0.4020 ***
	(0.0595)	(0.0604	(0.0595)	(0.0020)	(0.0507)	(0.0580)
Homophily (REmid)	0.0498	−0.0611	0.0457	−0.0670 ***	0.2265 ***	0.0411
	(0.0464)	(0.0459)	(0.0466)	(0.0015)	(0.0391)	(0.0466)
Homophily (RElow)	0.0279	0.0643	0.2204 ***	0.2591 ***	0.0276	0.0634
	(0.0492)	(0.0486)	(0.0504)	(0.0017)	(0.0376)	(0.0445)
Receiver (GDPhigh)	−0.0115	−0.0415 ***	−0.0059	−0.0572 ***	0.0175 *	0.0152
	(0.0085)	(0.0078)	(0.0078)	(0.0024)	(0.0069)	(0.0081)
Receiver (GDPmid)	0.0440 ***	0.0054	0.0664 ***	0.0371 ***	0.1033 ***	0.0914 ***
	(0.0101)	(0.0091)	(0.0092)	(0.0019)	(0.0081)	(0.0097)
Receiver (GDPlow)	0.0385 *	0.0181	0.0526 **	0.1540 ***	0.1371 ***	0.0646 ***
	(0.0192)	(0.0162)	(0.0187)	(0.0024)	(0.0152)	(0.0141)
Receiver (RE high)	0.0190 ***	0.0159 ***	0.0187 ***	0.0164 ***	0.0174 ***	0.0168 ***
	(0.0010)	(0.0009)	(0.0010)	(0.0009)	(0.0009)	(0.0010)
Receiver (RE mid)	0.0150 ***	0.0026	0.0124 ***	−0.0026	−0.0009	−0.0048 *
	(0.0023)	(0.0021)	(0.0024)	(0.0021)	(0.0023)	(0.0023)
Receiver (RE low)	0.9084 ***	0.4314 ***	0.8892 ***	0.6474 ***	0.6370 ***	0.7958 ***
	(0.0893)	(0.0611)	(0.0953)	(0.0019)	(0.0612)	(0.0709)
Sender(ghg)	0.0052 *	0.0137 ***	0.0041 *	−0.0077 ***	0.0045 *	0.0037
	(0.0021)	(0.0020)	(0.0021)	(0.0019)	(0.0019)	(0.0020)
Sender(REP)	0.1927 ***	0.1854 ***	0.1698 ***	0.1435 ***	0.1590 ***	0.1606 ***
	(0.0067)	(0.0062)	(0.0066)	(0.0000)	(0.0065)	(0.0072)
NCov (agreement)	0.0153	0.4279 ***	1.1119 ***	0.7045 ***	0.7885 ***	0.7536 ***
	(0.0703)	(0.0643)	(0.0727)	(0.0026)	(0.0560)	(0.0635)
NCov (CBN)	0.2026	0.1258	0.7830 ***	0.8401 ***	0.7932 ***	0.7910 ***
	(0.1213)	(0.1290)	(0.1226)	(0.0038)	(0.1058)	(0.1191)
AIC	32,457.5589	36,223.0961	33,046.6343	33,899.3276	39,972.4418	34,165.5319
BIC	32,668.5619	36,433.8809	33,258.0706	34,110.9792	40,184.0934	34,377.1835

Notes: Standard deviations are reported in parentheses. ***, **, and * represent statistical significance at the 0.1%, 1%, and 5% levels, respectively.

Table 11

ERGM results of renewable energy product trading network of 2015–2018.

Variables	2015	2016	2017	2018
Edges	−7.5132 ***	−7.6157 ***	−7.9182 ***	−7.7598 ***
	(0.1978)	(0.2045)	(0.2045)	(0.1980)
Reciprocity	1.6681 ***	1.7874 ***	1.7200 ***	1.7640 ***
	(0.0462)	(0.0469)	(0.0441)	(0.0447)
nodecov.GDPhigh	0.1948 ***	0.1865 ***	0.1435 ***	0.1540 ***
	(0.0079)	(0.0083)	(0.0073)	(0.0074)
nodecov.GDPmid	0.1325 ***	0.1257 ***	0.0864 ***	0.1066 ***
	(0.0078)	(0.0077)	(0.0073)	(0.0071)
nodecov.GDPlow	0.1868 ***	0.1628 ***	0.2359 ***	0.2038 ***
	(0.0248)	(0.0249)	(0.0246)	(0.0244)
nodecov.RE high	−0.0126 ***	−0.0128 ***	−0.0138 ***	−0.0184 ***
	(0.0009)	(0.0009)	(0.0009)	(0.0007)
nodecov.RE mid	−0.0042 **	−0.0019	−0.0064 ***	−0.0085 ***
	(0.0015)	(0.0016)	(0.0016)	(0.0012)
nodecov.RE low	−0.7387 ***	−0.4334 ***	−0.4271 ***	−0.0248 *
	(0.0525)	(0.0372)	(0.0421)	(0.0099)
Homophily (GDPhigh)	−0.0791	−0.1347	−0.2300 **	−0.2213 **
	(0.0826)	(0.0879)	(0.0822)	(0.0826)
Homophily (GDPmid)	0.2725 ***	0.3272 ***	0.2952 ***	0.3060 ***
	(0.0595)	(0.0607)	(0.0606)	(0.0599)
Homophily(GDPlow)	0.8964 ***	0.8880 ***	1.8317 ***	1.4263 ***
	(0.1701)	(0.1729)	(0.1754)	(0.1713)
Homophily (REhigh)	0.3119 ***	0.3727 ***	0.2483 ***	−0.0763
	(0.0540)	(0.0541)	(0.0547)	(0.0390)
Homophily (REmid)	0.089 *	0.1112 *	0.0728	0.0891 *
	(0.0441)	(0.0441)	(0.0420)	(0.0388)
Homophily (RElow)	0.0610	0.0921 *	0.0781	0.3354 ***
	(0.0443)	(0.0433)	(0.0437)	(0.0358)
Receiver (GDPhigh)	0.0219 **	0.0326 ***	0.0437 ***	0.0452 ***
	(0.0081)	(0.0080)	(0.0077)	(0.0077)
Receiver (GDPmid)	0.0959 ***	0.1103 ***	0.1118 ***	0.1084 ***
	(0.0096)	(0.0095)	(0.0089)	(0.0091)
Receiver (GDPlow)	0.1423 ***	0.1673 ***	0.1577 ***	0.1496 ***
	(0.0178)	(0.0176)	(0.0180)	(0.0176)
Receiver (RE high)	0.0168 ***	0.0189 ***	0.0201 ***	0.0196 ***
	(0.0010)	(0.0010)	(0.0010)	(0.0009)
Receiver (RE mid)	0.0027	−0.0032	0.0058 *	0.0051 **
	(0.0023)	(0.0025)	(0.0024)	(0.0018)
Receiver (RE low)	0.7156 ***	0.4065 ***	0.5058 ***	0.0605 ***
	(0.0710)	(0.0507)	(0.0598)	(0.0149)
Sender(ghg)	0.0029	−0.0005	0.0084 ***	0.0146 ***
	(0.0021)	(0.0021)	(0.0022)	(0.0019)
Sender(REP)	0.1963 ***	0.1964 ***	0.2212 ***	0.2344 ***
	(0.0070)	(0.0076)	(0.0076)	(0.0071)
NCov (agreement)	0.7421 ***	0.0895 ***	0.3025 ***	0.0717 **
	(0.0657)	(0.0242)	(0.0246)	(0.0237)
NCov (CBN)	0.7546 ***	0.3014 *	0.7904 ***	−0.0398
	(0.1146)	(0.1187)	(0.1143)	(0.1173)
AIC	35,124.9615	34,933.5670	35,836.4641	5594.0378
BIC	35,336.6131	35,145.2187	36,048.1157	35,805.9037

Notes: Standard deviations are reported in parentheses. ***, **, and * represent statistical significance at the 0.1%, 1%, and 5% levels, respectively.

Table 12

ERGM results of each kind of renewable energy product.

Variables	(1)Solar	(2)Wind	(3)Biomass	(4)Water	(5)Ocean	(6)Geothermal
Edges	−6.2516 ***	−2.8533 ***	−6.4548 ***	−2.2559 ***	−1.0841 ***	−3.8590 ***
	(0.2173)	(0.1456)	(0.7080)	(0.1735)	(0.2435)	(0.1856)
Reciprocity	2.1417 ***	2.7135 ***	1.9102 ***	0.4659 ***	1.1695 ***	1.7195 ***
	(0.0477)	(0.0406)	(0.0613)	(0.0459)	(0.0454)	(0.0923)
nodecov.GDPhigh	0.2153 ***	0.0044	0.2336 ***	−0.0188 ***	0.0365 ***	0.0354
	(0.0046)	(0.0032)	(0.0074)	(0.0035)	(0.0049)	(0.0021)
nodecov.GDPmid	0.1027 ***	−0.1030 ***	0.1177 ***	−0.0251 ***	0.0466 ***	−0.1340 ***
	(0.0045)	(0.0031)	(0.0100)	(0.0035)	(0.0049)	(0.0034)
nodecov.GDPlow	0.2051 ***	−0.0312	0.0398	0.0419 *	−0.3280 ***	0.0512
	(0.0292)	(0.0188)	(0.0999)	(0.0212)	(0.0377)	(0.0179)
nodecov.RE high	−0.0076 ***	−0.0094 ***	−0.0018	0.0110 ***	−0.0027 ***	−0.0086 **
	(0.0009)	(0.0008)	(0.0010)	(0.0007)	(0.0007)	(0.0006)
nodecov.RE mid	0.0164 ***	0.0125 ***	−0.0120 ***	0.0055 ***	0.0030 *	0.0179 ***
	(0.0016)	(0.0012)	(0.0017)	(0.0012)	(0.0013)	(0.0012)
nodecov.RE low	0.1030 ***	0.0774 ***	−0.0216	0.0842 ***	0.0419 ***	0.0664 ***
	(0.0102)	(0.0089)	(0.0113)	(0.0091)	(0.0099)	(0.0084)
Homophily (GDPhigh)	0.0355	0.0935 *	−0.3988 ***	0.1553 *	−0.1204	0.0735 *
	(0.0658)	(0.0458)	(0.1118)	(0.0604)	(0.0673)	(0.0496)
Homophily (GDPmid)	0.1251 *	−0.0900 *	0.5497 ***	0.0169	0.0705	−0.0972 *
	(0.0508)	(0.0375)	(0.1069)	(0.0506)	(0.0580)	(0.0476)
Homophily(GDPlow)	1.5452 ***	0.4127 **	0.6303	0.2285	−1.2306 ***	0.1235 **
	(0.1995)	(0.1254)	(0.6930)	(0.1466)	(0.2253)	(0.1674)
Homophily (REhigh)	−0.3730 ***	−0.5420 ***	0.0978	0.0439	−0.2674 ***	−0.8450 ***
	(0.0576)	(0.0515)	(0.0686)	(0.0541)	(0.0550)	(0.0416)
Homophily (REmid)	−0.0136	−0.1000 **	0.0472	−0.1765 ***	−0.0700	−0.1400 *
	(0.0398)	(0.0314)	(0.0471)	(0.0421)	(0.0415)	(0.0334)
Homophily (RElow)	0.3289 ***	0.3379 ***	0.0645	0.3026 ***	0.3177 ***	0.3782 ***
	(0.0371)	(0.0278)	(0.0377)	(0.0349)	(0.0355)	(0.0259)
Receiver (GDPhigh)	−0.1272 ***	−0.0043	−0.1052 ***	0.0806 ***	−0.0318 ***	−0.0283
	(0.0045)	(0.0037)	(0.0088)	(0.0033)	(0.0051)	(0.0097)
Receiver (GDPmid)	−0.0287 ***	0.1320 ***	0.0587 ***	0.0187 ***	−0.0425 ***	0.1020 **
	(0.0057)	(0.0047)	(0.0119)	(0.0038)	(0.0061)	(0.0089)
Receiver (GDPlow)	−0.0268	0.0696 ***	0.0900 *	0.0340 **	0.1748 ***	0.0286 ***
	(0.0187)	(0.0146)	(0.0365)	(0.0119)	(0.0203)	(0.0144)
Receiver (RE high)	−0.0001	0.0025 **	0.0050 ***	−0.0086 ***	0.0023 **	0.0038 **
	(0.0011)	(0.0008)	(0.0010)	(0.0006)	(0.0007)	(0.0005)
Receiver (RE mid)	−0.0251 ***	−0.0197 ***	0.0154 ***	−0.0075 ***	−0.0070 ***	−0.0147 ***
	(0.0024)	(0.0019)	(0.0022)	(0.0016)	(0.0018)	(0.0014)
Receiver (RE low)	−0.0534 **	−0.0516 ***	0.0625 ***	−0.0317 **	−0.0021	−0.0346 ***
	(0.0163)	(0.0147)	(0.0165)	(0.0115)	(0.0131)	(0.0167)
Sender(ghg)	0.0294 ***	0.0387 ***	0.0063 **	0.0013	−0.0189 ***	0.0667 ***
	(0.0019)	(0.0015)	(0.0021)	(0.0019)	(0.0021)	(0.0075)
Sender(REP)	0.0319 ***	0.0807 ***	−0.0115 *	−0.0328 ***	0.0225 ***	0.0567 *
	(0.0042)	(0.0035)	(0.0050)	(0.0046)	(0.0047)	(0.0055)
NCov (agreement)	0.1940 ***	0.3460 ***	0.0128	0.0419 **	0.0435 ***	0.0934
	(0.0237)	(0.0198)	(0.0294)	(0.0262)	(0.0255)	(0.0168)
NCov (CBN)	0.5373 ***	0.2858 **	−0.0818	−0.0933	−0.0422	0.2098 **
	(0.1138)	(0.1016)	(0.1593)	(0.1379)	(0.1330)	(0.3316)
AIC	35,000.5713	42,703.6925	23,581.2820	35,111.8744	33,045.1731	32,403.6025
BIC	35,212.0077	42,914.2581	23,786.7536	35,316.3589	33,250.6448	32,514.2981

Notes: Standard deviations are reported in parentheses. ***, **, and * represent statistical significance at the 0.1%, 1%, and 5% levels, respectively.

Appendix A. Description of Network Statistics

(1) Number of nodes

The number of nodes (N) refers to all the nodes in a network. The number of nodes can reflect the size of a network, and the more nodes there are, the larger a network is. Therefore, we use nodes to refer to specific countries, and the number of nodes denotes the number of countries involved in the trade of renewable energy products.

(2) Number of edges

The number of network edges refers to the number of relationships generated between all nodes in a network. In the renewable energy product trade network, if there is a direct edge from country i to country j, it indicates that country i exports renewable energy products to country j, and vice versa.

(3) Average path length

The average path length measures the average distance between two economies in a renewable energy product trade network, reflecting the closeness of renewable energy product network connections and the effectiveness of information transfer between network nodes. If the shortest path length from country i to country j is ${\mathrm{d}}_{\mathrm{ij}}$ = 1, the trade of renewable energy products occurs directly from country i to country j; if the shortest path length from country i to country j is ${\mathrm{d}}_{\mathrm{ij}}$ = 2, trade between the two countries must pass through intermediate countries to occur. Therefore, the average path length reflects whether the chain of trade in renewable energy products is longer or shorter in each country, which is expressed as follows. (A1)$$L=\frac{1}{n\left(n-1\right)}{{\displaystyle \sum }}_{i=1}^n{{\displaystyle \sum }}_{j=1}^n{d}_{ij}\left(i\ne j\right)$$ where n is the number of nodes in the network.

(4) Degree and intensity of nodes

The node degree measures the number of nodes connected to a specified node, which indicates the number of trading partners of a country. The higher the degree of a node, the more influence it has in the network. In a directionally weighted renewable energy product trade network, the out-degree and in-degree of an economy (region) represent the number of partner countries involved in its exports and imports of renewable energy products, respectively, and the out-intensity and in-intensity are equal to the total exports and imports of renewable energy products of that country, respectively. The specific formulas for out-degree (${\mathrm{K}}_{\mathrm{j}}^{\mathrm{out}})$ and in-degree (${\mathrm{K}}_{\mathrm{j}}^{\mathrm{in}}$) are expressed as follows:(A2)$$K_j^{out}={{\displaystyle \sum }}_{i=1\left(i\ne j\right)}^n{w}_{ij}$$ (A3)$$K_j^{in}={{\displaystyle \sum }}_{i=1\left(i\ne j\right)}^n{w}_{ij }$$

(5) Clustering coefficient

The clustering coefficient Ci measures the probability of the existence of a link between two randomly selected neighbors of node i. Specifically, in the trade network of renewable energy products, the clustering coefficient can be used to analyze whether the relationship between any three nodes, country i, country j, and country s, will tend to form a closed triangular relationship between them, where the clustering coefficient of country i indicates the probability of renewable energy product trade occurring between country j and country s, if there is renewable energy product trade from country i to country j and from country i to country s. The formula is expressed as follows. (A4)$${\mathrm{C}}_{\mathrm{i}}=\frac{{\mathrm{E}}_{\mathrm{i}}}{{\mathrm{m}}_{\mathrm{i}}\ast \left({\mathrm{m}}_{\mathrm{i}}-1\right)}$$ where ${\mathrm{m}}_{\mathrm{i}}$ denotes the number of nodes adjacent to node i. ${\mathrm{m}}_{\mathrm{i}}$∗(${\mathrm{m}}_{\mathrm{i}}$ − 1) represents all possible connections between nodes adjacent to node i in theory, and ${\mathrm{E}}_{\mathrm{i}}$ refers to the actual connections between nodes adjacent to node i.

(6) Reciprocity coefficient

The reciprocity coefficient is measured by dividing the total number of mutual edges between network nodes by the total number of network edges, which mainly reflects the closeness between any two nodes in the network. In the trade network of renewable energy products, a higher reciprocity coefficient indicates that the two countries prefer the two-way trade of renewable energy products. Its formula is expressed as follows. (A5)$$r=\frac{\overleftrightarrow{E}}{NE}$$ where $\overleftrightarrow{\mathrm{E}}$ denotes the number of mutual edges between network nodes, and NE denotes the number of all directed edges in the network.

(7) Density

Network density refers to the ratio of the number of directed edges in a network to the number of directed edges of the maximum possible links. The formula is (A6)$$D=\frac{L}{n\left(n-1\right)}$$

For a trade network containing n member countries, the maximum possible number of directed edges is n(n − 1), where L is the actual number of directed edges.

Appendix B

Table A1

HS codes for renewable energy products.

Solar Energy	Wind Energy		Ocean Energy	Water Power	Biomass Energy	Geothermal Energy
700991	730820	850422	850421	382450	220710	730431
700992	841290	850423	850422	681091	220720	730441
711590	848210	850431	850423	841011	380210	730451
732290	848220	850432	850431	841012	840681	741121
830630	848230	850433	850432	841013	840682	741122
841280	848240	850434	850433	841090	841182	741129
841919	848250	854459	850434	850161	841280	841861
841950	848280	854460	854459	850162	841620	841950
841989	848340	890790	854460	850163	841931	850239
841990	850161	902830		850164	841940
850239	850162	903020		850421	847920
850440	850163	903031		850422	850161
854140	851064	903083		850423	850162
900190	850231			850431	850163
900290	850300			850432	850164
900580	850421			850433
				850434

Appendix C. Renewable Energy Products Network Structures

Figure A1. (a). Global renewable energy products network 2009. (b). Global renewable energy products network 2019.

Figure A2. (a). Renewable energy products trade network of core and semi-core countries 2009. (b). Renewable energy products trade network of core and semi-core countries 2019.

References

1. Abbas, A.; Waseem, M.; Ahmad, R.; Khan, K.A.; Zhao, C.; Zhu, J. Sensitivity analysis of greenhouse gas emissions at farm level: Case study of grain and cash crops. Environ. Sci. Pollut. Res.; 2022; 29, pp. 82559-82573. [DOI: https://dx.doi.org/10.1007/s11356-022-21560-9] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35751727]

2. Elahi, E.; Khalid, Z. Estimating smart energy inputs packages using hybrid optimisation technique to mitigate environmental emissions of commercial fish farms. Appl. Energy; 2022; 326, 119602. [DOI: https://dx.doi.org/10.1016/j.apenergy.2022.119602]

3. Abbas, A.; Zhao, C.; Waseem, M.; Khan, K.A.; Ahmad, R. Analysis of Energy Input-Output of Farms and Assessment of Greenhouse Gas Emissions: A Case Study of Cotton Growers. Front. Environ. Sci.; 2022; 9, 826838. [DOI: https://dx.doi.org/10.3389/fenvs.2021.826838]

4. Zeng, M.; Xue, S.; Ma, M.; Zhu, X. New energy bases and sustainable development in China: A review. Renew. Sustain. Energy Rev.; 2013; 20, pp. 169-185. [DOI: https://dx.doi.org/10.1016/j.rser.2012.11.067]

5. Radovanovic, M.; Filipovic, S. New approach to energy intensity in the EU—Total energy and carbon cost approach. Energy Environ.; 2015; 26, pp. 601-616. [DOI: https://dx.doi.org/10.1260/0958-305X.26.4.601]

6. Xu, B.; Lin, B. Assessing the development of China’s new energy industry. Energy Econ.; 2018; 70, pp. 116-131. [DOI: https://dx.doi.org/10.1016/j.eneco.2018.01.001]

7. Maricic, V.K.; Danilovic, D.; Lekovic, B.; Crnogorac, M. Energy policy reforms in the Serbian oil sector: An update. Energy Policy; 2018; 113, pp. 348-355. [DOI: https://dx.doi.org/10.1016/j.enpol.2017.11.011]

8. Gu, J.; Renwick, N.; Xue, L. The BRICS and Africa’s search for green growth, clean energy and sustainable development. Energy Policy; 2018; 120, pp. 675-683. [DOI: https://dx.doi.org/10.1016/j.enpol.2018.05.028]

9. Huang, B.; Zhang, Y.; Wang, C. New Energy Development and Issues in China during the 14th Five-Year Plan. Electr. Power; 2020; 53, pp. 1-9.

10. Zhang, X.; Ruoshui, W.; Molin, H.; Martinot, E. A study of the role played by renewable energies in China’s sustainable energy supply. Energy; 2010; 35, pp. 4392-4399. [DOI: https://dx.doi.org/10.1016/j.energy.2009.05.030]

11. Wang, B.; Wang, Q.; Wei, Y.-M.; Li, Z.-P. Role of renewable energy in China’s energy security and climate change mitigation: An index decomposition analysis. Renew. Sustain. Energy Rev.; 2018; 90, pp. 187-194. [DOI: https://dx.doi.org/10.1016/j.rser.2018.03.012]

12. Ahmad, T.; Zhang, D. A critical review of comparative global historical energy consumption and future demand: The story told so far. Energy Rep.; 2020; 6, pp. 1973-1991. [DOI: https://dx.doi.org/10.1016/j.egyr.2020.07.020]

13. Amri, F. The relationship amongst energy consumption (renewable and non-renewable), and GDP in Algeria. Renew. Sustain. Energy Rev.; 2017; 76, pp. 62-71. [DOI: https://dx.doi.org/10.1016/j.rser.2017.03.029]

14. Aleixandre-Tudó, J.L.; Castelló-Cogollos, L.; Aleixandre, J.L.; Aleixandre-Benavent, R. Renewable energies: Worldwide trends in research, funding and international collaboration. Renew. Energy; 2019; 139, pp. 268-278. [DOI: https://dx.doi.org/10.1016/j.renene.2019.02.079]

15. Cansino, J.M.; del Pablo-Romer, P.; Román, R.; Yñiguez, R. Tax incentives to promote green electricity: An overview of EU-27 countries. Energy Policy; 2010; 38, pp. 6000-6008. [DOI: https://dx.doi.org/10.1016/j.enpol.2010.05.055]

16. Jagoda, K.; Lonseth, R.; Lonseth, A.; Jackman, T. Development and commercialization of renewable energy technologies in Canada: An innovation system perspective. Renew. Energy; 2011; 36, pp. 1266-1271. [DOI: https://dx.doi.org/10.1016/j.renene.2010.08.022]

17. Sims, R.E.H.; Rogner, H.-H.; Gregory, K. Carbon emission and mitigation cost comparisons between fossil fuel, nuclear and renewable energy resources for electricity generation. Energy Policy; 2003; 31, pp. 1315-1326. [DOI: https://dx.doi.org/10.1016/S0301-4215(02)00192-1]

18. Wang, C.; Huang, X.; Hu, X.; Zhao, L.; Liu, C.; Ghadimi, P. Trade characteristics, competition patterns and COVID-19 related shock propagation in the global solar photovoltaic cell trade. Appl. Energy; 2021; 290, 116744. [DOI: https://dx.doi.org/10.1016/j.apenergy.2021.116744]

19. Charters, W.W.S. Developing markets for renewable energy technologies. Renew. Energy; 2001; 22, pp. 217-222. [DOI: https://dx.doi.org/10.1016/S0960-1481(00)00018-5]

20. Baskaran, T.; Bloechl, F.; Brueck, T.; Theis, F.J. The Heckscher-Ohlin model and the network structure of international trade. Int. Rev. Econ. Financ.; 2011; 20, pp. 135-145. [DOI: https://dx.doi.org/10.1016/j.iref.2010.11.003]

21. An, H.Z.; Zhong, W.Q.; Chen, Y.R.; Li, H.J.; Gao, X.Y. Features and evolution of international crude oil trade relationships: A trading-based network analysis. Energy; 2014; 74, pp. 254-259. [DOI: https://dx.doi.org/10.1016/j.energy.2014.06.095]

22. Zhang, S.; Wang, L.; Liu, Z.; Wang, X. Evolution of international trade and investment networks. Phys. A-Stat. Mech. Appl.; 2016; 462, pp. 752-763. [DOI: https://dx.doi.org/10.1016/j.physa.2016.06.117]

23. Zhou, M.; Wu, G.; Xu, H. Structure and formation of top networks in international trade, 2001–2010. Soc. Netw.; 2016; 44, pp. 9-21. [DOI: https://dx.doi.org/10.1016/j.socnet.2015.07.006]

24. Hou, W.; Liu, H.; Wang, H.; Wu, F. Structure and patterns of the international rare earths trade: A complex network analysis. Resour. Policy; 2018; 55, pp. 133-142. [DOI: https://dx.doi.org/10.1016/j.resourpol.2017.11.008]

25. Watts, D.J.; Strogatz, S.H. Collective dynamics of ‘small-world’ networks. Nature; 1998; 393, pp. 440-442. [DOI: https://dx.doi.org/10.1038/30918] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/9623998]

26. Yang, Y.; Poon, J.P.; Dong, W. East Asia and Solar Energy Trade Network Patterns. Geogr. Rev.; 2017; 107, pp. 276-295. [DOI: https://dx.doi.org/10.1111/j.1931-0846.2015.12155.x]

27. Guan, Q.; An, H.; Hao, X.; Jia, X. The Impact of Countries’ Roles on the International Photovoltaic Trade Pattern: The Complex Networks Analysis. Sustainability; 2016; 8, 313. [DOI: https://dx.doi.org/10.3390/su8040313]

28. Schweitzer, F.; Fagiolo, G.; Sornette, D.; Vega-Redondo, F.; Vespignani, A.; White, D.R.; Networks, E. Economic Networks: The New Challenges. Science; 2009; 325, pp. 422-425. [DOI: https://dx.doi.org/10.1126/science.1173644]

29. Stivala, A.; Robins, G.; Lomi, A. Exponential random graph model parameter estimation for very large directed networks. PLoS ONE; 2020; 15, e0231023. [DOI: https://dx.doi.org/10.1371/journal.pone.0227804]

30. Robins, G.; Pattison, P.; Kalish, Y.; Lusher, D. An introduction to exponential random graph (p*) models for social networks. Soc. Netw.; 2007; 29, pp. 173-191. [DOI: https://dx.doi.org/10.1016/j.socnet.2006.08.002]

31. Jha, V.; Flows, T. Barriers and Market Drivers in Renewable Energy Supply Goods. Available online: https://ictsd.iisd.org/bridges-news/biores/news/trade-flows-barriers-and-market-drivers-in-renewable-energy-supply-goods (accessed on 13 June 2022).

32. Jebli, M.B.; Youssef, S.B. The environmental Kuznets curve, economic growth, renewable and non-renewable energy, and trade in Tunisia. Renew. Sustain. Energy Rev.; 2015; 47, pp. 173-185. [DOI: https://dx.doi.org/10.1016/j.rser.2015.02.049]

33. Barbato, L.; Centi, G.; Iaquaniello, G.; Mangiapane, A.; Perathoner, S. Trading Renewable Energy by using CO2: An Effective Option to Mitigate Climate Change and Increase the use of Renewable Energy Sources. Energy Technol.; 2014; 2, pp. 453-461. [DOI: https://dx.doi.org/10.1002/ente.201300182]

34. Varun,; Prakash, R.; Prakash, R.; Bhat, I.K. Energy, economics and environmental impacts of renewable energy systems. Renew. Sustain. Energy Rev.; 2009; 13, pp. 2716-2721. [DOI: https://dx.doi.org/10.1016/j.rser.2009.05.007]

35. Lilliestam, J.; Ellenbeck, S. Energy security and renewable electricity trade—Will Desertec make Europe vulnerable to the “energy weapon”?. Energy Policy; 2011; 39, pp. 3380-3391. [DOI: https://dx.doi.org/10.1016/j.enpol.2011.03.035]

36. Dent, C.M. China’s renewable energy development: Policy, industry and business perspectives. Asia Pac. Bus. Rev.; 2015; 21, pp. 26-43. [DOI: https://dx.doi.org/10.1080/13602381.2014.939892]

37. Tsai, W.T.; Chou, Y.H. Overview of environmental impacts, prospects and policies for renewable energy in Taiwan. Renew. Sustain. Energy Rev.; 2005; 9, pp. 119-147. [DOI: https://dx.doi.org/10.1016/j.rser.2004.01.014]

38. Leng, Z.; Shuai, J.; Sun, H.; Shi, Z.; Wang, Z. Do China’s wind energy products have potentials for trade with the “Belt and Road” countries?—A gravity model approach. Energy Policy; 2020; 137, 111172. [DOI: https://dx.doi.org/10.1016/j.enpol.2019.111172]

39. Kim, K.; Kim, Y. Role of policy in innovation and international trade of renewable energy technology: Empirical study of solar PV and wind power technology. Renew. Sustain. Energy Rev.; 2015; 44, pp. 717-727. [DOI: https://dx.doi.org/10.1016/j.rser.2015.01.033]

40. Zhao, Y.; Wang, L.; Yu, Y. Trade Liberalization and China’s Exports of Renewable Energy Products: Evidence from Product Level Data. Emerg. Mark. Financ. Trade; 2016; 52, pp. 1281-1297. [DOI: https://dx.doi.org/10.1080/1540496X.2016.1152788]

41. Boz, D.E.; Sanli, B.; Berument, M.H. The effects of cross-border electricity trade on power production from different energy sources. Electr. J.; 2021; 34, 106953. [DOI: https://dx.doi.org/10.1016/j.tej.2021.106953]

42. Ortega, M.; Río, P.D.; Ruiz, P.; Nijs, W.; Politis, S. Analysing the influence of trade, technology learning and policy on the employment prospects of wind and solar energy deployment: The EU case. Renew. Sustain. Energy Rev.; 2020; 122, 109657. [DOI: https://dx.doi.org/10.1016/j.rser.2019.109657]

43. Xiao, Y.; Wang, X.; Wang, X.; Wu, Z. Trading wind power with barrier option. Appl. Energy; 2016; 182, pp. 232-242. [DOI: https://dx.doi.org/10.1016/j.apenergy.2016.08.123]

44. Abhinav, R.; Pindoriya, N.M. Opportunities and key challenges for wind energy trading with high penetration in Indian power market. Energy Sustain. Dev.; 2018; 47, pp. 53-61. [DOI: https://dx.doi.org/10.1016/j.esd.2018.08.007]

45. Liu, S.; Caporin, M.; Paterlini, S. Dynamic network analysis of North American financial institutions. Financ. Res. Lett.; 2021; 42, 101921. [DOI: https://dx.doi.org/10.1016/j.frl.2021.101921]

46. Wang, B.; Pu, Y.; Li, S.; Xu, L. The influence of regional preferential trade agreements on international manufacturing trade in value-added: Based on the complex network method. PLoS ONE; 2021; 16, e0246250. [DOI: https://dx.doi.org/10.1371/journal.pone.0246250] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33606719]

47. Zhang, W.; Chong, Z.; Li, X.; Nie, G. Spatial patterns and determinant factors of population flow networks in China: Analysis on Tencent Location Big Data. Cities; 2020; 99, 102640. [DOI: https://dx.doi.org/10.1016/j.cities.2020.102640]

48. Gutierrez-Moya, E.; Lozano, S.; Adenso-Diaz, B. Analysing the Structure of the Global Wheat Trade Network: An ERGM Approach. Agronomy; 2020; 10, 1967. [DOI: https://dx.doi.org/10.3390/agronomy10121967]

49. Du, R.; Wang, Y.; Dong, G.; Tian, L.; Liu, Y.; Wang, M.; Fang, G. A complex network perspective on interrelations and evolution features crossMark of international oil trade, 2002–2013. Appl. Energy; 2017; 196, pp. 142-151. [DOI: https://dx.doi.org/10.1016/j.apenergy.2016.12.042]

50. Geng, J.-B.; Ji, Q.; Fan, Y. A dynamic analysis on global natural gas trade network. Appl. Energy; 2014; 132, pp. 23-33. [DOI: https://dx.doi.org/10.1016/j.apenergy.2014.06.064]

51. Bernard, A.B.; Moxnes, A. Networks and Trade. Annu. Rev. Econ.; 2018; 10, pp. 65-85. [DOI: https://dx.doi.org/10.1146/annurev-economics-080217-053506]

52. Baker, L.; Sovacool, B.K. The political economy of technological capabilities and global production networks in South Africa’s wind and solar photovoltaic (PV) industries. Political Geogr.; 2017; 60, pp. 1-12. [DOI: https://dx.doi.org/10.1016/j.polgeo.2017.03.003]

53. Zhang, G.; Duan, H. How does international trade network affect multinational diffusion of wind power technology?. J. Clean. Prod.; 2020; 276, 123245. [DOI: https://dx.doi.org/10.1016/j.jclepro.2020.123245]

54. Fu, X.; Yang, Y.; Dong, W.; Wang, C.; Liu, Y. Spatial structure, inequality and trading community of renewable energy networks: A comparative study of solar and hydro energy product trades. Energy Policy; 2017; 106, pp. 22-31. [DOI: https://dx.doi.org/10.1016/j.enpol.2017.03.038]

55. Feng, L.; Xu, H.; Wu, G.; Zhao, Y.; Xu, J. Exploring the structure and influence factors of trade competitive advantage network along the Belt and Road. Phys. A-Stat. Mech. Appl.; 2020; 559, 125057. [DOI: https://dx.doi.org/10.1016/j.physa.2020.125057]

56. Xu, H.; Cheng, L. The study of the influence of common humanistic relations on international services trade-from the perspective of multi-networks. Phys. A-Stat. Mech. Appl.; 2019; 523, pp. 642-651. [DOI: https://dx.doi.org/10.1016/j.physa.2019.02.055]

57. Ball, M.O.; Kaku, B.K.; Vakhutinsky, A. Network-based formulations of the quadratic assignment problem. Eur. J. Oper. Res.; 1998; 104, pp. 241-249. [DOI: https://dx.doi.org/10.1016/S0377-2217(96)00330-X]

58. Hansen, J.U. Reasoning about opinion dynamics in social networks. J. Log. Comput.; 2019; 29, pp. 1121-1137. [DOI: https://dx.doi.org/10.1093/logcom/exv083]

59. Lee, M.R.; Lee, J.W. A fuzzy inference network model for search strategy using neural logic network. J. Intell. Robot. Syst.; 2003; 36, pp. 209-221. [DOI: https://dx.doi.org/10.1023/A:1022644001499]

60. Brughmans, T.; Keay, S.; Earl, G. Introducing exponential random graph models for visibility networks. J. Archaeol. Sci.; 2014; 49, pp. 442-454. [DOI: https://dx.doi.org/10.1016/j.jas.2014.05.027]

61. Pattison, P.; Wasserman, S. Logit models and logistic regressions for social networks: II. Multivariate relations. Br. J. Math. Stat. Psychol.; 2019; 52, pp. 169-193. [DOI: https://dx.doi.org/10.1348/000711099159053]

62. Kilduff, M.; Krackhardt, D. Interpersonal Networks in Organizations; Cambridge University Press: New York, NY, USA, 2008.

63. Jackson, M.O.; Rogers, B.W.; Zenou, Y. The Economic Consequences of Social-Network Structure. J. Econ. Lit.; 2017; 55, pp. 49-95. [DOI: https://dx.doi.org/10.1257/jel.20150694]

64. Lusher, D.; Koskinen, J.; Robbins, G. Exponential Random Graph Models for Social Networks; Cambridge University Press: New York, NY, USA, 2012.

65. Krugman, P. Scale Economies, Product Differentiation, and the Pattern of Trade. Am. Econ. Rev.; 1980; 70, pp. 950-959.

66. Hering, L.; Poncet, S. Environmental policy and exports: Evidence from Chinese cities. J. Environ. Econ. Manag.; 2014; 68, pp. 296-318. [DOI: https://dx.doi.org/10.1016/j.jeem.2014.06.005]

67. Fu, J.; Wu, L. Effects of Institutions and Environmental Policies on Export of Renewable Energy: A Perspective of Intensive Margin and Extensive Margin. J. Int. Trade; 2015; 12, pp. 85-95. [DOI: https://dx.doi.org/10.13510/j.cnki.jit.2015.12.008] (In Chinese)

68. Wu, G. The Mechanism and Effect of the Influence of International Humanistic Relation Network on International Trade Network. Ph.D. Thesis; Hunan University: Changsha, China, 2013.

69. Coase, R.H. The Nature of the Firm. Economica; 1937; 4, pp. 386-405. [DOI: https://dx.doi.org/10.1111/j.1468-0335.1937.tb00002.x]

70. Eaton, J.; Kortum, S. Technology, Geography, and Trade. Econometrica; 2002; 70, pp. 1741-1779. [DOI: https://dx.doi.org/10.1111/1468-0262.00352]

71. Anderson, J.E.; van Wincoop, E. Gravity with Gravitas: A Solution to the Border Puzzle. Am. Econ. Rev.; 2003; 93, pp. 170-192. [DOI: https://dx.doi.org/10.1257/000282803321455214]

72. Krugman, P. Increasing Returns and Economic-Geography. J. Political Econ.; 1991; 99, pp. 483-499. [DOI: https://dx.doi.org/10.1086/261763]

73. Ghafouri, S.; Khasteh, S.H. A survey on exponential random graph models: An application perspective. PeerJ Comput. Sci.; 2020; 6, e269. [DOI: https://dx.doi.org/10.7717/peerj-cs.269]

74. Bai, Y. Structural characteristics and correlation factors analysis of global trade Structural characteristics and correlation factors analysis of global trade network of solar energy industry. Master Thesis; Shandong University: Jinan, China, 2018.

75. Qi, W.; Xiong, H.; Peng, X. How to Promote International Competitiveness of China’s Renewable Energy Products?—Based on SNA Theory. Pol. J. Environ. Stud.; 2021; 30, pp. 4163-4175. [DOI: https://dx.doi.org/10.15244/pjoes/130635]