The Evolution of Global Food Trade Systems and Their Resilience in Response to COVID-19: Performance Across Nations

Abstract

A resilient food trade system is crucial for global food security. The spatiotemporal changes in the trade of four main cereals (soybean, wheat, rice, and maize) and their responses to COVID-19 may serve as an efficient indicator of system resilience but remain underexplored. Using the United Nations Comtrade dataset and the COVID-19 dataset, this paper analyzed the evolution of the Global Trade Network for Four Cereals (GTN4) over 21 years and assessed their trade responses to COVID-19. The findings are as follows: (1) The GTN4 underwent a significant shift after 2019. Between 2000 and 2019, the network steadily expanded in size and became more interconnected, both overall and within groups of developing and developed countries. However, following 2019, its overall accessibility declined, with the extent of deterioration varying between these two groups. (2) COVID-19 influenced the cereal trade in 44–69% of countries, with developed nations exhibiting greater resilience. (3) Wheat exports from Germany, rice from Italy, and maize from the United States demonstrated the highest resilience, while Spain’s soybean trade played a key role in global imports. This research provides new insights into global food security and pandemic resilience, informing sustainable development at the national, group, and global levels.

1. Introduction

Food security is of great significance for maintaining social stability across countries [1,2]. Ensuring food security is a key objective within the Sustainable Development Goals (SDGs) outlined in the 2030 Agenda for Sustainable Development [3,4].

However, achieving food security remains a global challenge [5,6]. Different regions and countries face distinct food insecurity issues [2]. Developing nations often grapple with challenges related to food availability and stability, whereas developed countries prioritize issues of food utilization and accessibility [7,8]. Moreover, the status of food security evolves due to changes in internal and external environments within specific countries or regions [9]. For instance, over the past 40 years, China has experienced a rapid decrease in its rural population, leading to challenges such as aging farmers, a decrease in arable land area, and degraded soil and water resources, which exacerbate food insecurity [10,11]. In India, annual salinization rates have increased by up to 10%, foreshadowing a potential food crisis over the next 30 years [12]. Ethiopia faced severe food insecurity in 2023 due to regional conflicts and extreme dry weather since 2022 [13]. In addition, the COVID-19 pandemic resulted in increased household food insecurity in the United States and Canada [14,15]. This evidence illustrates that food security is influenced by various factors with strong spatiotemporal heterogeneity [16,17]. Exploring spatiotemporal evolution and the underlying mechanisms of global food security is essential for building a resilient global food system to achieve SDG goals [18,19].

The food trade serves as an essential way of achieving food security [20,21] and is intricately linked to all four key dimensions of food security, namely, the availability, access, utilization, and stability of food [22]. The international food trade system plays a vital role in ensuring food supply, stabilizing food prices, promoting the diversification of food sources, and responding to both domestic and external shocks and crises [6,20,21]. For instance, Porkka et al. [23] argued that global food trade enables many regions to ensure food supplies and overcome the constraints imposed by scarce natural resources or underdeveloped agricultural practices. Olufemi-Phillips et al. [24] found that open trade policies positively influence food security by enhancing market stability and price flexibility. Similarly, through empirical analysis, Dithmer et al. [25] demonstrated that trade openness improves dietary diversity and quality-related aspects of food security, exerting a highly significant positive effect overall. The spatiotemporal evolution of the global food trade system provides evidence for changes in food dependencies among countries or regions [26,27], which are essential for understanding the multidimensional characteristics of global food security [28].

A resilient food system can withstand shocks and maintain food security throughout its evolution and response [29,30,31]. Examining the reactions of global food trade systems or group/national subsystems to specific global crises offers valuable perspectives for understanding the heterogeneity of food system resilience [18,32,33]. Crises such as violent conflicts [34,35], extreme weather events [5,36], agricultural input shocks [37], and environmental shocks [16] can influence food trade flow to varying extents. The COVID-19 pandemic, which has lasted for 3 years to date, has been a typical global shock that has severely affected economies worldwide, leading to a substantial recession and exacerbating global food insecurity [38,39,40]. Specifically, Gebeyehu et al. [41] reported that the COVID-19 pandemic both directly and indirectly undermined individual food security. Similarly, Boratyńska [42] noted that the pandemic posed multiple risks to the sustainable food security system. Subsequent studies have further highlighted its impact on the meat market [43], seafood systems [44], food prices [39], and smallholders and farm enterprises [45,46,47] in different regions. Yet, how it has influenced the food trade systems of the four main cereals worldwide is still unknown. Identifying this information can provide a fundamental insight into the different resilience mechanisms of food systems worldwide, helping to address global food insecurity issues.

In previous studies, global value chain (GVC) flows were usually used to capture the characteristics of global food trade [48]. The national input–output tables they relied on are highly aggregated, based on broad metrics, and only updated every five years, with the most recent version released in 2016 [49,50,51]. Therefore, this model cannot be used to capture recent network effects centered on direct trade flows of specific food types. However, the complex network model has a unique advantage in capturing the supply and demand characteristics of the food trade [52,53]. This approach has been used to identify the spatiotemporal dynamics of food trade networks at both the national and global scales [54,55]. For example, researchers used this model to explore the dynamics of the global rice trade network from 2000 to 2016 [56] and the evolution pattern of the global cereal trade network from 2000 to 2018 [57]. However, these studies mainly focused on the trade dynamics of limited types of food before the COVID-19 pandemic [54,58]. The spatiotemporal evolution of international trade covering four main cereals—soybean, wheat, rice, and maize—over the long term, including the pandemic period, remains unclear.

Regarding research on how COVID-19 has influenced the global food trade, previous studies have focused on observing changes in different components of the system, such as food production [59], supply and demand dynamics [60], prices [39,61], and nutrition [62]. For instance, some scholars have conducted economic modeling analysis to uncover the negative impacts of COVID-19 on poverty, food security, and diets [63], while others have used a supply model to explore the relationship between global food prices and trade restrictions induced by the pandemic [61]. However, many of these studies were limited to country-level analyses [62] or focused on specific types of food [64]. Only a small group of studies has aimed to link COVID-19 with food trade at a global scale; however, they have tended to conduct qualitative analyses based on surveys [33,65] or comments [66]. Quantifying the impact of the COVID-19 pandemic on the trade of four main cereals at a global scale to evaluate varying levels of resilience across different food trade systems remains a significant challenge.

To address these issues, this study aimed to analyze the evolution of trade in the four main cereals from 2000 to 2021 and assess the resilience of the global food trade system in response to COVID-19, using data from the United Nations Comtrade dataset and the COVID-19 dataset. We aimed to answer three questions: (1) How have global food trade networks evolved over the last two decades? (2) Are the post-2019 changes observed in global trade networks attributable to the COVID-19 pandemic? (3) How did food trade systems perform in response to the pandemic across countries? This paper shows the heterogeneous evolution and resilience mechanisms in the global food trade systems across the main cereals and across nations. More importantly, we highlight the diverse performances of the four main cereals in developing and developed country groups. The results offer significant insights that can aid in addressing food insecurity and achieving the SDG goals at national, group, and global scales.

2. Materials and Methods

2.1. Study Area and Data Sources

This study considered 160 countries, comprising 36 developed countries and 124 developing countries (Appendix A,

Table A1. Classification of developed and developing economies.

Developed Economies	Developing Economies
Australia	Afghanistan	El Salvador	Mauritius	Thailand
Austria	Algeria	Equatorial Guinea	Mexico	Timor-Leste
Belgium	Angola	Eritrea	Mongolia	Togo
Bulgaria	Argentina	Eswatini	Morocco	Trinidad and Tobago
Canada	Bahamas	Ethiopia	Mozambique	Tunisia
Croatia	Bahrain	Fiji	Myanmar	Türkiye
Cyprus	Bangladesh	Gabon	Namibia	Uganda
Czechia	Barbados	Gambia	Nepal	United Arab Emirates
Denmark	Belize	Ghana	Nicaragua	United Republic of Tanzania
Estonia	Benin	Guatemala	Niger	Uruguay
Finland	Bhutan	Guinea	Nigeria	Vanuatu
France	Bolivia (Plurinational State of)	Guinea-Bissau	Oman	Venezuela (Bolivarian Republic of)
Germany	Botswana	Guyana	Pakistan	Viet Nam
Greece	Brazil	Haiti	Panama	Yemen
Hungary	Brunei Darussalam	Honduras	Papua New Guinea	Zambia
Iceland	Burkina Faso	India	Paraguay	Zimbabwe
Ireland	Burundi	Indonesia	Peru
Italy	Cabo Verde	Iran (Islamic Republic of)	Philippines
Japan	Cambodia	Iraq	Qatar
Latvia	Cameroon	Israel	Republic of Korea
Lithuania	Central African Republic	Jamaica	Rwanda
Luxembourg	Chad	Jordan	Samoa
Malta	Chile	Kenya	Sao Tome and Principe
Netherlands	China a	Kiribati	Saudi Arabia
New Zealand	Colombia	Kuwait	Senegal
Norway	Comoros	Lao People’s Democratic Republic	Sierra Leone
Poland	Congo	Lebanon	Singapore
Portugal	Costa Rica	Lesotho	Solomon Islands
Romania	Côte d’Ivoire	Liberia	Somalia
Slovakia	Cuba	Libya	South Africa
Slovenia	Democratic People’s Republic of Korea	Madagascar	South Sudan
Spain	Democratic Republic of the Congo	Malawi	Sri Lanka
Sweden	Djibouti	Malaysia	State of Palestine
Switzerland	Dominican Republic	Maldives	Sudan
United Kingdom	Ecuador	Mali	Suriname
United States	Egypt	Mauritania	Syrian Arab Republic

^a All Chinese data in this paper include data from Mainland China, Hong Kong SAR, Macau SAR, and Taiwan Province of China.

). The classification of developed and developing countries was conducted according to the definitions provided by the United Nations in the World Economic Situation and Prospects 2022 report (https://www.un.org/development/desa/dpad/publication/world-economic-situation-and-prospects-2022/, accessed on 7 April 2023) [67]. It is important to note that 17 transition countries (e.g., Russia, Ukraine, and Serbia) were not included in the study.

The global food data for soya beans (hereafter named soybeans), wheat, milled rice (hereafter named rice), and maize across 160 countries during the last 20 years (2000–2021) was sourced from the United Nations Commodity Trade Database (UN Comtrade) [68]. The global COVID-19 data from 21 January 2020 to 31 December 2021 was obtained from the Center for Systems Science and Engineering at Johns Hopkins University (https://github.com/CSSEGISandData/COVID-19, accessed on 20 May 2023) [69]. This dataset records the number of COVID-19 infections and deaths in different countries and regions. The global population data used to calculate the Epidemic Severity Index was sourced from the World Population Prospects 2022 published by the United Nations (https://population.un.org/wpp/downloads?folder=Standard%20Projections&group=Most%20used, accessed on 13 August 2025) [70]. Additional datasets used in the regression model were obtained from five key data sources. Gravity-type control variables—including geographical distance, shared borders, official language similarity, colonial ties, and landlocked status—were sourced from the GeoDist database provided by CEPII [71]. The indicator variable representing whether countries have signed a regional trade agreement was derived from the dataset constructed by Egger and Larch (2008) [72]. WTO membership status was determined based on official membership information available on the WTO website [73]. Country-level GDP data were retrieved from the World Bank database [74].

2.2. Methodology

In this study, we focused on analyzing the trade of soybeans, wheat, rice, and maize due to their nutritional importance and dominance in global agricultural markets. Together, these four crops represent the majority of global cereal and oilseed production and trade, serving as staple foods for a substantial portion of the world’s population. They provide essential calories and protein, contributing significantly to human diets and livestock feed [75,76]. Given their critical role in global food security, this study aimed to investigate the dynamics of the food trade system for these crops from 2000 to 2021 and assess its performance in response to COVID-19 during the period from 2019 to 2021. Specifically, we analyzed the food trade systems of soybeans, wheat, rice, and maize at global, developed/developing country groups, and individual country scales.

The framework for this work (Figure 1) encompassed three main steps. First, complex network analysis was used to examine the long-term spatiotemporal evolution of the Global Trade Network of Four Cereals (GTN4) at both the global and group levels. Second, a baseline regression model with various control variables was conducted to assess the impact of COVID-19 on the import/export trade of four cereals (IE4) at both the global and group levels. Third, we analyzed the performances of certain countries, i.e., the top 10 countries and outlier countries, by examining changes in their weighted in/out degree and epidemic severity index (ESI) after the COVID outbreaks to highlight critical structures within each cereal’s trade system. The top 10 countries were determined by initially ranking the average import/export volumes for four main cereals in 2000, 2010, and 2018, and then selecting the top 10 countries for each cereal. The outlier countries were determined using a threshold of weighted in/out-degree set at 200%, assuming that a country’s import/export activities would not double continuously over 2 years under normal conditions.

2.2.1. Complex Network Analysis

The complex network model is widely used to profile the structure and behavior of networks in complex systems [43,44]. It offers a comprehensive approach to understanding the characteristics and dynamics of food trade supply and demand at various scales, from national to global levels [45,46].

In this paper, a weighted complex network model was used to construct a global food trade network, denoted as G = (V, E, W), where V represents the set of all nodes (countries or regions N), E is the set of all edges (trade links between countries or regions M), and W denotes the set of trade volumes between countries/regions. The weighted complex network accounts for the strength of relationships between nodes and has proved to be effective in capturing the characteristics of food trade networks in previous research [57,77]. The model was constructed on the Gephi platform with version 0.10.1 [78]. Five key indicators, including average degree, weighted degree, graph density, average clustering coefficient, and average path length, were derived. The definitions and equations of these indicators are explained in the following sections and are consistent with descriptions from previous studies [65,79,80].

(1)

Graph Density

The network density refers to the proportion of actual trade linkages in the network relative to the maximum possible trade linkages. In a directed trade network with N nodes, the maximum possible number of trade linkages is N (N − 1). Network density is a measure of the degree of interconnectedness within the network. A higher density indicates closer linkages between members of the network, as expressed in Equation (1):

$$\mathrm{D} = {\displaystyle \frac{M}{N(N-1)}}$$

(1)

where M represents the actual trade linkages, and N represents the number of nodes.

(2)

Average Degree

The average degree represents the average number of connections per node in the network, reflecting the level of trade linkages between countries and the overall complexity of the network. A higher average degree indicates more frequent trade connections between countries and greater network complexity, as expressed in Equation (2):

$$\mathit{AD}= {\displaystyle \frac{M}{N}}$$

(2)

where M represents the number of trade linkages between nodes, and N represents the number of nodes.

(3)

Average Clustering Coefficient

The clustering coefficient reflects the level of connectivity between trading partners. The average clustering coefficient measures the average degree of clustering around the nodes in the network, capturing the group characteristics of the network, such as the overall level of trade relationship clustering between countries in a specific group. A higher average clustering coefficient indicates denser direct trade relationships among a country’s trading partners, as expressed in Equation (3):

$$\mathit{CL}= {\displaystyle \frac{1}{N}}{\sum }_{i=1}^N{\displaystyle \frac{e_i}{k_i(k_i-1)}}$$

(3)

where $k_i$ denotes the node degree of the node, $i$, and $e_i$ denotes the number of edges between all neighbors of the node, $i$.

(4)

Average Path Length

The average path length is the average number of steps in the shortest path between two countries in the trade network. It is used to measure the efficiency of trade transmission within the network. A shorter average path length indicates that nodes are more closely connected, leading to higher transport efficiency, as expressed in Equation (4):

$$L= {\displaystyle \frac{1}{N(N-1)}}{\sum }_{i\ne j}{d}_{ij}$$

(4)

where $d_{ij}$ denotes the shortest path between node $i$ and node $j$ in the network, and N represents the number of nodes.

To explore the evolution of the global food trade network, we initially obtained four sets of values for six indicators (including the abovementioned four indicators, the nodes, and the edges) for the years 2000, 2010, 2019, and 2021. For each type of cereal, the dynamics of its food trade system during specific periods were depicted by the annual average change rate of each indicator over the same timeframe, as expressed in Equation (5):

$$R(A) = {\displaystyle \frac{A_{(b)}-A_{(a)}}{A_{(a)}}}·{\displaystyle \frac{1}{b-a}}·100\%$$

(5)

where A represents the indicator, a and b respectively refer to the start and end year of a period, and $A_{(a)}$ and $A_{(b)}$ respectively represent the values of the indicators in a and b.

(5)

Weighted In/Out Degree

The weighted degree ($c_i^{sum}$) is the sum of the trade volume between the node, $i$, and all the nodes it is connected to. A higher weighted degree value indicates a larger trade volume. The weighted out degree $(c_i^{out})$ and the weighted in degree $(c_i^{in})$ indicate the export and import volume of a country, as expressed in Equations (6)–(8), respectively:

$$c_i^{sum}=c_i^{out}+c_i^{in}$$

(6)

$$c_i^{out}={\sum }_{j=1,i\ne j}^N{W}_{ij}$$

(7)

$$c_i^{in}={\sum }_{j=1,i\ne j}^N{W}_{ji}$$

(8)

where $W_{ij}$ represents the trade volume from node $i$ to node $j$, $W_{ji}$ represents the trade volume from node $j$ to node $i$, and N represents the number of nodes.

2.2.2. The Change Rate of the Weighted In/Out Degree

The change rate of the weighted out degree and the change rate of the weighted in degree are used to quantify changes in the exports and imports of a country, respectively. A negative change rate of the weighted in/out degree indicates a decreasing trend in imports/exports, while a positive rate indicates an increasing trend. The equations are expressed as follows:

$$\mathit{R-WID}= {\displaystyle \frac{WID\left(b\right)-WID\left(a\right)}{WID(a)}}·100\%$$

(9)

$$\mathit{R-WOD}={\displaystyle \frac{WOD\left(b\right)-WOD\left(a\right)}{WOD(a)}}·100\%$$

(10)

where WID denotes the weighted in degree ($c_i^{in}$ in Equation (8)), R-WID denotes the change rate of the weighted in degree, WOD denotes the weighted out degree ($c_i^{out}$ in Equation (7)), and R-WOD denotes the change rate of weighted out degree. A larger value for a weighted in/out degree denotes a greater volume of imports/exports. a and b represent the start and end years of the pandemic period, respectively.

If the explanatory capacity of R-WID/R-WOD in 2019 is unreliable owing to instances where the value of WID/WOD in 2019 reached zero, we replace the abnormal zero WID/ WOD values in 2019 with data from 2018. Countries with consistent zero WID/WOD values in both 2018 and 2019 are considered noise and are subsequently excluded from the analysis. As a result, 28 and 6 countries are excluded when calculating the R-WID and R-WOD of soybeans, respectively; 34 and 5 countries are excluded when calculating the R-WID and R-WOD of wheat, respectively; 13 and 12 countries are excluded when calculating the R-WID and R-WOD of rice, respectively; and 12 and 7 countries are excluded when calculating the R-WID and R-WOD of maize, respectively.

2.2.3. Epidemic Severity Index

The epidemic severity index (ESI) reveals the extent to which countries have been affected by COVID-19. As a relative measure, the ESI enables more comparable assessments of these pressures across countries. It represents the proportion of infections and deaths induced by COVID-19 compared with the total population of a country. A larger ESI value indicates a more severe impact caused by the epidemic, as expressed in Equation (11):

$$\mathit{ESI} = {\displaystyle \frac{I_{(a)}+D_{(a)}+I_{(b)}+D_{(b)}}{P_{(b)}}}·100\%$$

(11)

where $I_{(a)}$ and $D_{(a)}$ represent the number of infections and deaths in year a, respectively, $I_{(b)}$ and $D_{(b)}$ represent the number of infections and deaths in year b, respectively, and $P_{(b)}$ represents the total population size in year b. Here, a and b were set to 2020 and 2021, respectively.

2.2.4. Regression Analysis

The baseline regression model is a fundamental econometric tool used to examine statistical associations between variables. In trade analysis, the gravity model is often utilized as its theoretical framework. By including various control variables, this model facilitates empirical assessments of trade flows and is particularly suitable for quantifying the initial impact of external shocks, such as the COVID-19 pandemic. Numerous scholars have applied this approach to investigate COVID-19’s effects on global and regional trade [78,79,80]. In this study, the baseline regression model incorporates monthly trade data for soybeans, wheat, rice, and maize from 2019 to 2021, alongside monthly data on COVID-19-affected populations, while controlling for several other relevant variables. The regression equation is specified as follows:

$$ln{Trade}_{ij}=\beta 0+\beta 1ln{COVID}_i+\beta 2ln{COVID}_j+\beta 3{Control}_{ij}+\epsilon$$

(12)

where lnTrade_ij denotes the bilateral import trade volume, specifically the trade volume of imports by country i from country j; lnCOVID represents the impact of COVID-19 on the population of an economy during the study period, measured primarily by the logarithmic value of the cumulative number of confirmed cases or deaths [81]; lnCOVID_i represents the degree of epidemic severity in the importing country, i, while lnCOVID_j indicates the same for the exporting country, j; and Control_ij denotes a set of control variables, following the empirical practices of previous studies [79,82]. Specifically, GDP_i and GDP_j reflect the level of economic development in country i and country j, respectively, and are expressed in logarithmic form. Border_ij is a binary indicator variable that is equal to 1 if countries i and j share a border and is equal to 0 otherwise. WTO_ij is also a binary indicator that indicates whether both country i and country j are members of the WTO (1 if both are members, 0 otherwise). Language_ij indicates the two countries share a common official language (1 if yes, 0 if not). Distance_ij denotes the geographical distance between the capitals of country i and country j, expressed in logarithmic form. Rta_ij is an indicator of whether the two countries have signed a regional trade agreement (1 if yes, 0 if not). Landlocked_ij equals 1 if at least one of the two countries is landlocked, and it equals 0 otherwise. Colony_ij indicates whether the countries have had colonial ties in the past (1 if yes, 0 otherwise). Concol_ij represents whether both countries shared the same colonizer after 1945 (1 if yes, 0 otherwise). Comcol_ij indicates whether the importing and exporting countries share a common colonizer after 1945. Curcol_ij indicates whether the importing and exporting countries currently have a colonial or dependent relationship, while Col45_ij reflects whether countries maintain or have historically maintained a colonial relationship since 1945. If a colonial relationship exists as defined by any of the abovementioned variables, the corresponding indicator takes a value of 1; otherwise, the value is 0. ε represents the random term. To assess the impact of the COVID-19 epidemic on the agricultural trade sector and to mitigate potential heteroskedasticity in the regression analysis, this study applies logarithmic transformations to most of the variables used in the model [82].

3. Results

3.1. Spatiotemporal Evolution of the GTN4 from 2000 to 2021

3.1.1. Dynamics of the GTN4

Between 2000 and 2021, trade volumes grew by 230% (soybeans), 53% (wheat), 223% (rice), and 92% (maize), with soybean and rice volumes more than doubling. Over this period, the average network size ranged from 122 to 141 nodes, with densities between 0.058 and 0.089, average degrees from 7.08 to 12.58, clustering coefficients around 0.37–0.41, and path lengths between 2.34 and 2.75. However, from 2000 to 2021, the GTN4 underwent a notable shift, transitioning from an overall trend of improvement to partial deterioration.

Throughout 2000–2019, the global trade networks of the four main cereals generally exhibited a positive expansion trend. The values and periodical average annual growth rates of indicators such as nodes, trade linkages, average degree, and average clustering coefficient in the global trade networks of the four main cereals showed overall growth (Figure 2a,b,d,e), reflecting the integration of an increasing number of countries into the global food trading system over this period. There was a noticeable rise in inter-country trade flows, expansion in the size of trade networks, closer ties between member countries, and a growing level of aggregation within the networks. The overall decrease in the average path length of trade networks for soybean, rice, and maize suggests increased efficiency, with greater accessibility among countries. It is noteworthy that the previously accelerating pace of the average path length in the global wheat trade network experienced a slowdown, signaling a concerning change in network accessibility.

Specifically, the indicators for the wheat trade network during 2019–2021 continued to develop as in the pre-2019 period, while the trade networks of soybeans, rice, and maize exhibited signs of deterioration to varying degrees (Figure 2a,b,d,e). An increasing trend in the average path length and a decline in trade accessibility were observed across these three cereals. However, the three cereals evolved in different ways: (1) the average clustering coefficient of the soybean trade network decreased; (2) the number of nodes in the rice trade network decreased, indicating a reduction in countries involved in rice trade; and (3) the maize trade network showed decreases in the number of nodes, trade linkages, and average degree, indicating a shrinking network size with weakened linkages among countries.

3.1.2. Dynamics of the GTN4 in the Developed Country Group

From 2000 to 2021, the GTN4 in the developed group underwent a similar shift to that in the total GTN4, with an overall trend of improvement in the first phase (2000–2019) and a partial deterioration in the second phase (2019–2021).

Specifically, during the period from 2000 to 2019, the number of countries, trade linkages, average number of trading partners, and average clustering coefficients within the trade networks for the four main cereals in developed countries showed consistent increases (Figure 3a,b,d,e), while their average path lengths generally decreased (Figure 3f). These findings demonstrated that developed countries, as a group, had established robust trade connections for the four main cereals, involving an expanding number of countries and improving trade accessibility. Notably, the significant increase in the average annual change rate between these two phases indicated that the linkages and complexity of the trade networks for the four main cereals within the developed group were accelerating (Figure 3a).

From 2019 to 2021, although the number of nodes within the trade networks for soybeans, wheat, and rice in the developed group gradually increased (Figure 3a), their network density and average clustering coefficients decreased. In addition, their average annual change rates were significantly lower compared with the earlier phase. These findings suggest that while the size of these three trade networks within the developed group expanded during this period, their cohesion and clustering weakened (Figure 3c,e). In the case of maize trade within the developed group, the change rate of trade linkages and average clustering coefficient showed an upward trend, indicating a closer maize trade relationship among countries during this period. Moreover, the increasing clustering within the maize trade network suggests a continuously accelerating trend from 2019 to 2021.

3.1.3. Dynamics of the GTN4 in the Developing Group

From 2000 to 2021, the GTN4 in the developing group underwent a sudden deceleration, transitioning from an overall improvement in the first phase (2000–2019) to an overall deterioration in the later phase (2019–2021).

From 2000 to 2019, the nodes, edges, network density, average degree, and average clustering coefficient within the GTN4 in the developing group showed an overall increasing trend (Figure 4a–e), indicating a progressively closer network structure. Furthermore, the average annual change rate of almost all indicators within the GTN4 also increased during this period (Figure 3a,b,d), except for the network density within the trade network for rice (Figure 4c).

However, during the period from 2019 to 2021, the change rates for nodes within networks for soybeans, rice, and maize in the developing group, as well as trade linkages, network density, and average degree, significantly declined (Figure 4a–d). These findings show a slowing trend within the GTN4 in developing countries. In addition, there are certain differences in the change patterns of the four major food trade networks in developing countries. The trade networks for soybeans, rice, and maize exhibited similar change patterns, characterized by an inverted-V shape in the annual change rates of most indicators (Figure 4a,b,d,e), showing a sharp deceleration trend between the two phases. In contrast, the wheat trade network remained relatively stable. For instance, changes in trade linkages and the average degree were relatively small (Figure 4b–d), with increases observed in nodes and average clustering coefficients, as well as network clustering. In addition, the change rate of the average path length sharply decreased, indicating a more frequent and convenient trade network for wheat (Figure 4f).

3.2. Dynamics of IE4 and Its Relationship with COVID-19 from 2019 to 2021

3.2.1. Impact of the Pandemic on Global IE4

Approximately 44–69% of the world’s countries were affected by COVID-19 in terms of their IE4. Export trade proved more sensitive than import trade in response to the pandemic. Notably, more than half of the world’s countries (60–69%) witnessed a decline in exports in 2021 compared with 2019. Specifically, 42, 50, 67, and 74 countries saw decreasing exports of soybeans, wheat, rice, and maize, respectively, accounting for 60.9%, 64.1%, 60.4%, and 68.5% of the world’s countries, respectively. Meanwhile, 44–60% of countries globally experienced a decrease in imports in 2021 compared with 2019. A total of 61, 52, 71, and 75 countries saw declining imports of soybean, wheat, rice, and maize, respectively, accounting for 59.9%, 44%, 56.3%, and 58.5% of all countries, respectively.

In addition, 36 countries experienced zero IE4 in 2021. Specifically, the pandemic caused exports to drop to zero for two countries (soybeans), nine countries (wheat), twelve countries (rice), and eleven countries (maize). It also resulted in imports falling to zero for four countries (soybeans), one country (rice), and seven countries (maize).

The results from the baseline regression model indicated that the COVID-19 pandemic significantly inhibited global grain trade. After controlling for various gravity model variables, the regression coefficients for COVID-19 demonstrated notably negative values for soybean and wheat trade in both importing and exporting countries, rice trade in exporting countries, and maize trade in importing countries, suggesting a detrimental impact on global grain trade (

Table 1. Results of the impact of the COVID-19 pandemic on global trade of soybeans, wheat, rice, and maize.

Variant	Global	Soybean	Wheat	Rice	Maize
	Trade_ij	Trade_ij	Trade_ij	Trade_ij	Trade_ij
lnGRCOVID_i	−0.019 **	−0.042 **	−0.033 *	0.030	−0.182 ***
	(0.008)	(0.016)	(0.023)	(0.009)	(0.015)
lnGRCOVID_j	−0.003 *	−0.027 *	−0.059 ***	−0.051 ***	0.155
	(0.008)	(0.014)	(0.022)	(0.008)	(0.014)
lnGDP_i	0.454 ***	0.941 ***	0.192 ***	0.314 ***	0.361 ***
	(0.010)	(0.024)	(0.027)	(0.010)	(0.018)
lnGDP_j	0.236 ***	0.176 ***	0.161 ***	0.024 **	0.055 ***
	(0.011)	(0.021)	(0.028)	(0.011)	(0.018)
WTO_ij	0.774 ***	−0.691 ***	1.167 ***	0.793 ***	0.082
	(0.074)	(0.166)	(0.161)	(0.074)	(0.124)
Border_ij	2.826 ***	2.468 ***	2.849 ***	2.415 ***	2.869 ***
	(0.068)	(0.115)	(0.131)	(0.070)	(0.091)
Language_ij	0.083 *	1.498 ***	−0.332 ***	0.189 ***	−0.172 **
	(0.048)	(0.096)	(0.111)	(0.051)	(0.072)
lnDistance_ij	−0.401 ***	0.037	0.170	0.024	−0.387 ***
	(0.020)	(0.039)	(0.044)	(0.021)	(0.031)
Rta_ij	−0.628 ***	−1.478 ***	0.675 ***	−1.138 ***	0.236 ***
	(0.039)	(0.085)	(0.112)	(0.041)	(0.070)
landlocked_ij	−1.029 ***	−0.570 ***	−0.963 ***	−1.377 ***	−1.190 ***
	(0.046)	(0.099)	(0.112)	(0.048)	(0.076)
Colony_ij	−0.663 ***	0.169	−1.410 ***	−1.167 ***	−0.349 **
	(0.112)	(0.184)	(0.201)	(0.115)	(0.153)
Comcol_ij	0.890 ***	0.870 ***	−1.205 ***	1.537 ***	−0.054
	(0.075)	(0.176)	(0.226)	(0.074)	(0.133)
Curcol_ij	1.594 ***	−0.948	−0.071	2.056 ***	3.741 ***
	(0.509)	(1.426)	(0.848)	(0.507)	(0.694)
Col45_ij	−0.014	−1.968 ***	1.943 ***	0.298 *	−1.981 ***
	(0.154)	(0.301)	(0.324)	(0.157)	(0.237)
constant	−4.370 ***	−19.804 ***	−0.879	0.826 **	1.557 **
	(0.396)	(0.913)	(1.021)	(0.402)	(0.674)
N	69638.000	17416.000	18209.000	49366.000	28625.000
R2	0.089	0.153	0.053	0.098	0.094

Note: ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively; values in parentheses are standard errors.

). Specifically, in the soybean trade, a 1% increase in COVID-19 cases in importing countries led to a significant 0.042% reduction in trade volume, while an increase in cases in exporting countries resulted in a 0.027% decrease in trade volume. Similarly, in the wheat trade, a 1% increase in COVID-19 cases resulted in a significant reduction in trade volume of 0.042% for importing countries and 0.027% for exporting countries. In the wheat trade, a similar 1% increase led to a significant reduction of 0.033% in importing countries and 0.059% in exporting countries. Moreover, for rice and maize, a 1% rise in COVID-19 cases significantly decreased trade volumes by 0.051% (exporting countries for rice) and 0.182% (importing countries for maize), respectively.

To verify the robustness of our conclusions from the baseline estimation, we conducted additional tests to confirm whether the findings hold when the measurement of the core explanatory variable—the occurrence of the COVID-19 pandemic—is altered. Specifically, we replaced the original variable (cumulative confirmed COVID-19 cases) with the cumulative death tolls. We then re-ran the regression analysis based on Equation (12), and the results are presented in Appendix A,

Table A2. Results of the impact of the COVID-19 pandemic on the global trade of soybeans, wheat, rice, and maize (after replacing core variables).

Variant	The Global	Soybean	Wheat	Rice	Maize
	Trade_ij	Trade_ij	Trade_ij	Trade_ij	Trade_ij
lnSWCOVID_i	−0.011 *	−0.017 *	−0.041 **	0.035	−0.134 ***
	(0.008)	(0.014)	(0.021)	(0.008)	(0.014)
lnSWCOVID_j	−0.021 ***	−0.026 *	−0.076 ***	−0.060 ***	0.109
	(0.007)	(0.014)	(0.019)	(0.007)	(0.012)
lnGDP_i	0.444 ***	0.952 ***	0.186 ***	0.311 ***	0.337 ***
	(0.010)	(0.024)	(0.027)	(0.010)	(0.018)
lnGDP_j	0.247 ***	0.164 ***	0.173 ***	0.032 ***	0.079 ***
	(0.010)	(0.021)	(0.027)	(0.011)	(0.018)
WTO_ij	0.777 ***	−0.676 ***	1.172 ***	0.797 ***	0.091
	(0.074)	(0.166)	(0.161)	(0.074)	(0.124)
Border_ij	2.825 ***	2.477 ***	2.852 ***	2.407 ***	2.874 ***
	(0.068)	(0.115)	(0.131)	(0.070)	(0.091)
Language_ij	0.087 *	1.477 ***	−0.331 ***	0.192 ***	−0.180 **
	(0.048)	(0.096)	(0.111)	(0.051)	(0.072)
lnDistance_ij	−0.403 ***	0.035	0.171	0.018	−0.390 ***
	(0.020)	(0.039)	(0.045)	(0.021)	(0.031)
Rta_ij	−0.627 ***	−1.492 ***	0.681 ***	−1.135 ***	0.215 ***
	(0.039)	(0.085)	(0.112)	(0.041)	(0.070)
landlocked_ij	−1.031 ***	−0.558 ***	−0.962 ***	−1.377 ***	−1.193 ***
	(0.046)	(0.099)	(0.111)	(0.048)	(0.076)
Colony_ij	−0.665 ***	0.186	−1.412 ***	−1.166 ***	−0.346 **
	(0.112)	(0.184)	(0.200)	(0.115)	(0.154)
Comcol_ij	0.886 ***	0.857 ***	−1.202 ***	1.536 ***	−0.054
	(0.075)	(0.176)	(0.226)	(0.074)	(0.134)
Curcol_ij	1.574 ***	−0.999	−0.089	2.040 ***	3.676 ***
	(0.509)	(1.425)	(0.847)	(0.507)	(0.695)
Col45_ij	−0.011	−1.946 ***	1.959 ***	0.304 *	−1.976 ***
	(0.154)	(0.301)	(0.324)	(0.157)	(0.237)
constant	−4.450 ***	−19.828 ***	−1.091	0.703 *	1.532 **
	(0.398)	(0.915)	(1.024)	(0.404)	(0.677)
N	69,638.000	17,416.000	18,209.000	49,366.000	28,625.000
R2	0.089	0.154	0.053	0.098	0.092

Note: ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively; values in parentheses are standard errors.

. The severity of the epidemic (SWCOVID-19) in importing and exporting countries again demonstrated significant inhibitory effects on global trade for soybeans and wheat, as well as on rice trade in importing countries and maize trade in exporting countries. These outcomes closely align with those in Table 1, further confirming the robustness of our regression results.

3.2.2. Performance in the Developed Group

During the study period, the average severity of the pandemic in developed countries (13.09–13.56%) was higher than the global average (6.79–8.36%), and the pandemic had a more pronounced impact on imports than exports in this group. Specifically, for developed countries, only 57.1% (soybean), 43.7% (wheat), 50% (rice), and 43.7% (maize) of exports experienced shrinkage, whereas 55.9% (soybean), 52.8% (wheat), 63.6% (rice), and 57.1% (maize) of imports decreased.

After controlling for a series of gravity variables, the regression coefficients for COVID-19 cases showed significantly negative values in importing countries for soybeans and maize (developed countries) and exporting countries for wheat and rice (trading partners of developed countries). These results indicate that the pandemic had a notable impact on the soybean and maize trades within developed nations and presented significant challenges for the wheat and rice trades. Specifically, as shown in

Table 2. Results of the impact of the COVID-19 pandemic on the trade of soybeans, wheat, rice, and maize in developed countries.

Variant	Soybean	Wheat	Rice	Maize
	Trade_ij	Trade_ij	Trade_ij	Trade_ij
lnCOVID_i	−0.055 **	0.015	0.032	−0.098 ***
	(0.022)	(0.029)	(0.011)	(0.021)
lnCOVID_j	−0.041 *	−0.047 *	−0.052 ***	0.078
	(0.017)	(0.029)	(0.010)	(0.021)
Control variables	yes	yes	yes	yes
N	12057	11878	29868	16710
R2	0.214	0.174	0.172	0.160

Note: ***, **, * indicate significance at the 1%, 5%, and 10% levels, respectively; values in parentheses are standard errors.

, a 1% increase in COVID-19 cases in importing countries led to reductions of approximately 0.055% and 0.098% in soybean and maize trade volumes, respectively. Although the direct impact on wheat- and rice-importing countries was not statistically significant, a 1% increase in COVID-19 cases in exporting countries significantly reduced trade volumes by approximately 0.047% for wheat and 0.052% for rice. These findings highlight the extensive negative impact of the COVID-19 pandemic on the global grain trade in developed nations, demonstrating how disruptions in both importing and exporting countries have hindered trade volumes.

3.2.3. Performance in the Developing Group

During the study period, the average epidemic severity in developing countries (3.9–5.07%) was lower than the global average (6.79–8.36%); however, the pandemic had a significant impact on the export of the four main cereals. Specifically, 63–82% of developing countries experienced reductions in their exports, whereas only 40–62% of developing countries saw decreases in their imports.

After controlling for a series of gravity variables, the regression coefficients for COVID-19 cases in importing countries of soybeans and maize (developing countries), as well as in exporting countries of wheat and rice (trading partners of developing nations), were all significantly negative. The absolute values of these coefficients were larger than those observed for developed countries, indicating that developing nations faced more severe challenges in grain trade during the pandemic. As shown in

Table 3. Results for the impact of the COVID-19 pandemic on the trade of soybeans, wheat, rice, and maize in developing countries.

Variant	Soybean	Wheat	Rice	Maize
	Trade_ij	Trade_ij	Trade_ij	Trade_ij
lnCOVID_i	−0.138 ***	0.071	0.037	−0.275 ***
	(0.029)	(0.039)	(0.015)	(0.023)
lnCOVID_j	0.136	−0.082 **	−0.060 ***	0.231
	(0.026)	(0.034)	(0.012)	(0.020)
Control variables	yes	yes	yes	yes
N	5359	6331	19498	11915
R2	0.169	0.047	0.065	0.075

Note: *** and ** indicate significance at the 1% and 5% levels, respectively; values in parentheses are standard errors.

, a 1% increase in COVID-19 cases in importing countries led to significant reductions in trade: Approximately 0.138% for soybeans and 0.275% for maize. For wheat and rice, although the direct effect of COVID-19 in importing countries was not statistically significant, a 1% increase in cases in exporting countries resulted in significant trade volume reductions of about 0.082% and 0.060%, respectively. These findings suggest that the COVID-19 pandemic has had a substantial negative impact on the global grain trade involving developing countries, one that is more pronounced than in developed nations.

3.3. Impact of COVID-19 on Focal Countries

3.3.1. Response of Top 10 Countries to the Pandemic

Between 2000 and 2021, the top 10 countries for crops experienced substantial shifts in total exports/imports for soybeans, wheat, rice, and maize. For instance, soybean exports surged by over 6100% in Uruguay and more than 100% in the USA, while some exporters, such as the UK, saw declines approaching 99%. Wheat imports rose by over 200% in Germany and Indonesia, with exports soaring by more than 3000% in Romania. Rice imports in Benin and China expanded by over 2700% and 380%, respectively, whereas Indonesia’s imports fell by more than 90%. Maize exports increased dramatically in Brazil (over 6400%) and Romania (nearly 3500%), even as imports declined modestly in countries like Japan and Egypt. Notably, the magnitude of increases generally far exceeded that of decreases, indicating that, overall, global trade in these cereals has seen strong expansion, despite some localized contractions.

However, following the COVID-19 outbreak, considering the top 10 exporting countries as a group (Figure 5), their exports of wheat, rice, and maize were less affected by COVID-19, highlighting their significant roles in stabilizing the global export system. Throughout the pandemic, the export change rates for wheat, rice, and maize within the top 10 group were 14.16%, −8.63%, and −0.75%, respectively, compared with the global averages of −10.44%, −13%, and −33%, respectively. However, soybean exports from the top 10 countries (−17.06%) experienced a greater decline compared with the global average (−15.08%).

Within the group of the top 10 exporters (Figure 5), wheat exports from Germany, rice exports from Italy, and maize exports from the United States notably supported the global export system during the COVID-19 pandemic, while China’s soybean exports were relatively vulnerable. Specifically, Germany and Bulgaria faced severe epidemic situations (8.36% higher than the global average) but significantly increased their wheat exports. Despite a more severe epidemic in Italy (7.35% higher than the global average), the country maintained a higher rate in exporting rice. The United States, France, Hungary, and Romania continued to experience steady growth in maize exports despite the severity of the pandemic. In contrast, China and Canada had a relatively mild pandemic (8.34% lower than the global average), but their change rates in exporting soybeans decreased significantly.

Considering the top 10 importing countries as a group (Figure 6), their imports of wheat, rice, and soybeans were less affected by COVID-19, indicating support for the global import system. Specifically, the import change rates for wheat, rice, and soybeans were 3.98%, 0.36%, and 1.75%, respectively, compared with the global average of 10.83%, 2.97%, and −9.61%, respectively. However, the import change rate for maize (−16.57%) was lower than the global average (−10.35%), indicating its limited contribution to the stability of the global maize import system.

Among the top 10 importing countries (Figure 6), soybean imports from Spain exhibited a robust performance despite moderate epidemic conditions, while wheat imports from Japan, rice imports from Saudi Arabia, and maize imports from Vietnam made limited contributions to stabilizing the global import system. Specifically, Spain and Argentina faced relatively severe epidemic situations, yet their import change rates in soybeans were higher than the global average (−9.61%). In contrast, Japan and Mexico experienced moderate epidemics (7.06% lower than the global average), resulting in import change rates in wheat that were lower than the global average (10.83%). Saudi Arabia, the Philippines, and the United States experienced mild epidemics (6.74% lower than the global average) and notably reduced their rice imports. Vietnam and Egypt, with mild epidemic situations (6.69% lower than the global average), also reduced their maize imports.

3.3.2. Response of Outlier Countries to the Pandemic

During the COVID-19 pandemic, most outlier countries with exceptionally high increases in imports/exports (above 200%) were developing nations that experienced lower pandemic severity compared with the global average. Specifically, there were a total of 33, 19, 21, and 24 outlier countries in terms of exports of soybeans, wheat, rice, and maize, respectively. Their average pandemic severity was 5.05%, 4.30%, 2.87%, and 4.96%, respectively, which was lower than the global average (8.34%, 8.36%, 7.35%, and 6.79%, respectively). In addition, there were a total of 22, 8, 7, and 8 outlier countries in terms of imports of soybeans, wheat, rice, and maize, respectively. Their average pandemic severity was 5.12%, 3.98%, 2.23%, and 5.19%, respectively, which was lower than the global average (7.49%, 7.06%, 6.74%, and 6.69%, respectively). Furthermore, in the import and export trade of soybean, wheat, rice, and maize, certain countries, such as Honduras (ESI = 3.79%), Syria (ESI = 0.25%), Rwanda (ESI = 0.84%), and Oman (ESI = 6.85%), significantly increased their exports in 2021 after minimizing them in 2019. Similarly, Guatemala (ESI = 3.66%), Benin (ESI = 0.19%), Vietnam (ESI = 1.81%), and Kiribati (ESI = 0%) experienced minimal imports in 2019 but showed significant growth in 2021.

4. Discussion

4.1. Heterogeneous Spatiotemporal Evolution in the GTN4

On the global scale, few studies have comprehensively analyzed the dynamics of the GTN4 over the past 21 years, including the pandemic period. Although previous studies have typically used bilateral trade data or global value chain (GVC) flows to capture the characteristics of global food trade [48], these approaches have primarily focused on value distribution across multi-stage production and rely heavily on national input–output tables. However, these tables are highly aggregated, based on broad metrics, and only updated every five years, with the most recent version released in 2016 [49,50,51]. Therefore, these studies have rarely captured the network effects centered on the direct trade flows of specific food types. In this paper, we applied complex network analysis to examine the spatiotemporal evolution of trade networks for four grains, i.e., soybeans, wheat, rice, and maize, over the past two decades. The results showed that the obtained network indicators effectively characterize the topological relationships and structural dynamics of grain trade [79,80]. Notably, we observed a significant temporal shift in GTN4 since 2019, both globally and among developed and developing groups. Our findings on size expansion and complexity increase in the GTN4 before 2019 were consistent with previous studies [54,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,83]. Among them, Nie et al. (2021) analyzed the global grain trade network using grain trade data from 2000–2018 and found that it tended to become more complex, with the degree of network development gradually increasing [54]. Puma et al. (2015) used data from 1992–2009 to study the global food system and found that its complexity is increasing and that the number of trade links for wheat and rice has doubled [83]. The results after 2019 provide further insights into the heterogeneous temporal evolution of food security at the global level.

Moreover, this paper identified significant differences in the evolution patterns of the GTN4 between developed and developing countries. While some previous studies have analyzed the food trade of countries along the Belt and Road [57,84,85], they have rarely reported on the different evolution patterns between developed and developing countries. Our study found distinct evolutions of these two groups before and after the pivotal year of 2019. Before 2019, the trade networks for the four main cereals showed continuous expansion characterized by increasingly close linkages in both developed and developing countries, with the latter demonstrating faster growth rates than the former. However, after 2019, the GTN4 in the developing countries exhibited more pronounced and comprehensive deterioration compared with their developed counterparts. This highlights the fragility and inequality of global food security and demonstrates that countries should strive not only to enhance their self-sufficiency in food production but also to actively strengthen regional cooperation to jointly improve the level of food security.

4.2. COVID-19 and Global IE4

It is worthwhile to explore the relationship between COVID-19 and IE4. Previous studies have usually used qualitative methods [10,38,59], and the use of quantitative methods to analyze the impact of COVID-19 on global IE4 is not common.

This paper quantified the impact of the COVID-19 pandemic on global IE4 by examining changes in the weighted in/out degrees of the four main cereals during the pandemic period and their relationships with epidemic severity. Building on previous research, this paper not only employs indicators such as infection numbers and death tolls—which measure the absolute impact of COVID-19 on the population size of a given economy during the study period [86]—but also constructs a pandemic severity index. This index integrates data on infection numbers, death tolls, and total population. As a comparative standard, it allows for a more consistent assessment of the pressures faced by different countries during the COVID-19 pandemic, thus providing a more comprehensive evaluation of the varying degrees of impact that COVID-19 has had on specific economies over the study period. Here, we found that the pandemic directly decreased food imports and exports, with a greater impact on the export trade than the import trade. IE4 in developing countries was more affected by the pandemic than in developed countries. These findings align with previous studies conducted in some developing countries [8,60,87] and underscore the threat posed by COVID-19 to food security.

The different resilience and responses of developed and developing groups to the COVID-19 pandemic can be attributed to several possible reasons. First, developing countries rely more heavily on food imports [45,88], and trade bans on imports/exports can exacerbate food insecurity in these nations [8,89,90]. Second, factors such as labor-intensive industries, inefficient transportation, limited healthcare resources, poor access to resources, and fragile social security systems contribute to a weaker capacity to withstand epidemics [38,87,91]. Finally, food inflation and currency depreciation result in income loss, threatening food access for the poorer populations in developing countries [46,60]. Therefore, this study offers a case study perspective that can inform future research on the impacts of other global crises, such as climate change and geopolitical conflicts, on food trade and security. By examining the resilience and vulnerabilities exposed during the COVID-19 pandemic, policymakers and researchers can formulate strategies to mitigate the impact of similar events and ensure that food systems are more robust and adaptable in the face of future challenges.

4.3. Different Performance of Four Main Cereals

Generally, the size and complexity of the rice trade network were prominent and characterized by a higher number of nodes, more trade linkages, greater network density, a greater average degree, and a greater average clustering coefficient. This paper further observed the distinct evolutionary performances of four main cereals. From 2000 to 2019, the global soybean trade networks exhibited improvements in size, complexity, internal agglomeration, and trade accessibility, which is consistent with previous findings [80,92]. Similarly, the global wheat trade network experienced enhancements in size, complexity, and internal agglomeration but showed a decline in trade accessibility. These characteristics were also noted in prior research [93]. The global rice trade network showed improvements in size, complexity, internal agglomeration, and trade accessibility, aligning with previous work [56,79]. The maize trade network exhibited expansion in size, strong trade linkages, and increased trade accessibility, as confirmed by earlier research [93].

In addition, we observed varying degrees of resilience in trade systems across the four main cereals by examining the dynamics of IE4 and their relationships with COVID-19. From a global perspective, the COVID-19 pandemic has had a significant inhibitory effect on the trade activities of countries exporting soybeans and wheat, as well as those exporting rice and importing maize. Among them, the soybean and wheat trades were the most severely impacted, while the rice exports and maize imports were affected to a lesser extent. Developed and developing countries showed marked differences in trade for these four major food crops. Among developed countries, the pandemic has shown a significant negative correlation with soybean and maize imports, as well as wheat and rice exports. This suggests that their soybean and maize trading systems have been under considerable pressure, while the wheat and rice trades have also been affected. The situation is even more severe for developing countries. The absolute values of the negative coefficients that represent the pandemic’s impact on their soybean and maize imports, as well as wheat and rice exports, are higher than those for developed countries—highlighting the greater fragility of food trading systems in developing nations. During the COVID-19 period, the top exporters and importers of soybeans, wheat, rice, and maize exhibited diverse trends. For example, Germany showed strong resilience in its wheat trade, with exports and imports increasing by 57.98% and 23.16%, respectively, despite an ESI of 8.66%. Italy increased its rice exports, with an ESI of 10.57%, while the USA’s maize exports rose sharply by 56.99%, with an ESI of 16.54%. Regarding Spain’s significant role in soybean imports, between 2000, 2010, and 2018, Spain’s average annual soybean imports were 3.04 million tons, ranking sixth globally. Despite a high ESI of 13.44%, Spain’s soybean import volume increased by 11.96% during this period, indicating strong trade resilience and adaptability in meeting domestic demand amid pandemic challenges. These findings underscore the urgent need for countries to strategically adjust their crop planting structures to give priority to the cultivation of resilient staple crops such as corn and enhance the self-sufficiency rates of key imported commodities like wheat and rice, thereby reducing their reliance on the globally fluctuating supply chain.

4.4. Limitations and Outlook

This work has several limitations:

First, only four types of grains (soybeans, wheat, rice, and maize) were selected for this study, and other important food categories were not included, which may affect the generalizability of the study’s conclusions. Future research could include potatoes, miscellaneous grains, and so on, to explore the characteristics of the trade network structure and the spatial and temporal evolution of different crops.

Second, the analysis of the food trade only covered one aspect of the multidimensional issue of food security [19,94]. Understanding the trends in domestic trade and household food accessibility is also critical for comprehending global food security [17,95]. Future research investigating these trends will undoubtedly contribute to a more comprehensive understanding of this complex issue.

Third, this study employed baseline regression analysis to examine the impact of the COVID-19 pandemic’s severity on global food networks. While the model is widely used in econometrics to test statistical associations between variables and has been validated by previous related studies [86], it may not fully capture the complex relationships influenced by confounding factors. This study serves as a starting point for providing new insights into this issue, while future research could build upon this foundation by employing more advanced models that incorporate additional variables such as global production levels and extreme climate events.

Finally, this study did not explore potential reasons why different nations responded differently to COVID-19. Future research exploring factors such as supply chain disruptions, demand fluctuations, and policy changes, such as export strategies across countries, would offer valuable insights and effective recommendations for addressing global food insecurity.

5. Conclusions

On the basis of data from the FAOSTAT and COVID-19 datasets, this study examined the evolution of global food trade systems and their resilience in response to the COVID-19 pandemic. Specifically, we focused on four main cereals—soybeans, wheat, rice, and maize—and conducted our research at the global level, as well as within developing and developed country groups.

The main conclusions were as follows:

(1)

Between 2000 and 2021, the global food trade network underwent significant changes, particularly after 2019. In the initial phase, the GTN4 exhibited increasing internal clustering, size, and trade linkages. This trend was similar in both developing and developed countries. However, in the latter phase, the accessibility of trade networks for all four main cereals weakened. Specifically, the internal clustering of the soybean trade network decreased, the size of the rice trade network declined, and the size and accessibility of the maize trade network decreased, along with its linkages. In the developed country group, there was a reduction in the degree of clustering and agglomeration for soybeans, wheat, and rice. In the developing country group, the growth rate in size and trade linkages significantly decreased for the soybean, rice, and maize trade networks.

(2)

Global IE4 was significantly influenced by the COVID-19 pandemic, as evidenced by approximately 44% to 69% of global nations experiencing declines in their imports and exports during the pandemic period. Furthermore, as shown in the records, the import/export activities of the four main cereals dropped to zero in 38 countries in the year 2021, likely due to strict trade policies, high tariffs, international challenges, or domestic economic restructuring. For both developed countries and developing countries, the COVID-19 pandemic had a significant negative effect on soybean and maize imports, as well as rice and wheat exports. The pandemic’s effect on IE4 was more pronounced in developing countries compared with developed ones, indicating that the developing countries’ food systems were more vulnerable to the pandemic.

(3)

During the pandemic, Germany, Italy, and the United States demonstrated robust exports of wheat, rice, and maize, respectively, providing critical support to the global food trade. However, China’s soybean exports made limited contributions to the stability of global food exports. Conversely, Spain’s robust soybean imports—ranking sixth globally, with 3.04 million tons annually between 2000 and 2018—increased by 11.96%, despite a high ESI of 13.44%, highlighting its key role and trade resilience during the pandemic. In contrast, Japan, Saudi Arabia, and Vietnam were relatively vulnerable in importing wheat, rice, and maize, respectively.

This study not only revealed heterogeneous long-term evolution within global food trades for the four main cereals but also explored the impacts of short-term shocks (i.e., COVID-19) on the global food trade. The findings offer new perspectives and insights for addressing global food insecurity and achieving the SDGs on a global scale. On the one hand, the research highlights the vulnerability of the global food trade system to external shocks, underscoring the urgent need to strengthen international cooperation and policy coordination to ensure food security and stable trade flows. On the other hand, it emphasizes that developing countries must build more resilient food systems and actively diversify their trade partnerships. Moreover, countries should assume roles commensurate with their positions in the global food trade—where dominant nations lead efforts and bolster the capacities of more vulnerable countries—to collectively safeguard the security of global food trade. Future studies should focus on household food accessibility and model improvements to enhance understanding. Investigating the underlying factors behind global differences in resilience to COVID-19 would also be valuable.