The Impact of Climate Change on China and Brazil’s Soybean Trade

Abstract

In the recent past, China has expanded its grain production to achieve high food security and increased its partial dependence on imported agricultural commodities, of which soybean supply is highly import-dependent. This study systematically reviews the past trends in China’s soybean demand, Brazil’s soybean production and export, factors contributing to the soybean trade between China and Brazil, and future uncertainty in China’s demand and Brazil’s supply under climate change. We find that recently China imported ~64% of soybean from Brazil, while ~73% of Brazil’s soybean exports were destined for China, making them key stakeholders in their international soybean trade. China’s accession to the World Trade Organization, China–Brazil trade cooperation, and diversion from trade with the USA have played a pivotal role in China’s increasing soybean imports from Brazil. China’s soybean import has brought increasing virtual land to China (from 3.57 million hectares (Mha) in 2005 to 19.63 mha in 2020). This growing virtual land import could be one of the reasons behind Brazil’s soybean harvested area, which increased from 22.95 Mha in 2005 to 37.19 Mha in 2020. In the future, climate change impacts on soybean production in Brazil can seriously affect China’s soybean imports from Brazil and its domestic food security. We analyze these effects using a climate-crop–economic modeling approach, where yield changes from the crop model are incorporated into the economic model as lower land productivity. Our results show that Brazil’s future soybean production and gross exports can drop up to 13.1% and 15.2% under the highest emissions scenario (RCP8.5). Consequently, China would face a decrease in its soybean imports from Brazil (−9.94 Mt). Due to these import reductions, China’s domestic soybean supply will be reduced (−9.94 Mt). There would also be some reduction in China’s meat supply and a drop in China’s consumer welfare. Our results can contribute to devising policies to ensure China’s food security and promote global sustainable development goals.

1. Introduction

With the rapid development of China’s economy and the continuous increase in population, the country’s food demand has grown fast in recent decades. China’s per capita GDP grew by 12.3% annually in the last two decades. Although the population growth rate slowed, with an average annual growth rate of 0.536%, due to the vast population base, the population still increased from 1.267 billion in 2000 to 1.400 billion in 2019 [1]. At the same time, China’s per capita availability of agricultural inputs such as water and land is among the lowest globally [2]. Despite these imbalances, China has achieved complete self-sufficiency in staple crops and high self-sufficiency in many other food commodities [3]. However, due to land and water resource limitations, China increasingly depends on the international market to maintain a stable soybean supply. Today, China consumes more soybean than anywhere in the world and at any time in history. Most imported soybeans are not consumed directly but are mainly used as oilseeds and animal feed. Due to changing dietary preferences, soybean has been transformed from protein-rich human food in the domestic agricultural ecosystem into the most critical imported agricultural product for animal husbandry [4].

On the global trade front, Brazil’s role in the soybean international trade market has increased remarkably in recent years. With more than 70% of its total soybean output being exported, Brazil has become the largest soybean exporter and the second-largest soybean meal and soybean oil exporter globally. China is the largest buyer of Brazilian soybean. In Brazil, most of the soybean are produced in the Cerrado region. A large body of scientific literature suggests that climate change already affects rainfall patterns in the Cerrado and that this weather change impacts soybean production [5,6,7,8]. Despite Brazil’s increasing supply, there is still a question of whether Brazil will be able to meet China’s future demand for soybean under climate change.

Notwithstanding the rapid development of the China–Brazil soybean trade, there is a severe lack of coherent research from both China’s and Brazil’s perspectives on the uncertainty of soybean trade in the future. In this study, we take Brazil as a case study of China’s exporting partners, who provide the bulk of soybean. We start by reviewing the historical process of China’s soybean import and Brazil’s soybean export perspectives. In the final part, we analyze the impacts of Brazil’s soybean supply potential on China’s soybean demand under the background of climate change in the future and then compare China’s increasing virtual land (VL) imports through soybean trade with the soybean acreage expansion in Brazil.

For these analyses, we use the historical data from FAO [9] and the literature review to analyze the evolution and driving forces of soybean demand in China, supply in Brazil, and expansion in Brazil’s soybean acreage. We use the virtual land content (VLC) concept to estimate China’s VL imports through soybean imports from Brazil and compare them with Brazil’s soybean acreage expansion. For the climate change impacts, we use the data from the Inter-Sectoral Impact Model Intercomparison Project [10] on crop yield, for which the climate scenarios data were taken from the Coupled Model Intercomparison Project Phase 5 (CMIP5) [11]. We employ the yield changes under RCP2.6 and RCP8.5 scenarios to showcase the least and most severe climate change scenarios. Finally, the GTAP model is used to analyze the trade, economic, and food supply outcomes of soybean yield changes.

Our results show that China and Brazil have strengthened their mutual soybean trade in recent years. In the future, climate change will significantly affect Brazil’s soybean crop production, China’s soybean imports, and domestic food supply. Under the extreme shock of the RCP8.5 scenario, Brazilian soybean yield will decline by 17.66% by 2030. Under these extreme conditions, Brazil’s soybean production will drop by 13.06%, causing China’s imports to decrease by 9.94 million tonnes (Mt). China’s domestic soybean prices will increase, while the consumer welfare and livestock sectors will be adversely affected. China’s soybean imports from Brazil can be partially related to Brazil’s soybean acreage expansion in the last decade. China’s soybean trade with Brazil has saved increasing quantities of domestic and global land resources, which may decline under climate change. The study concludes that China and other import-dependent countries on foreign food production should make necessary arrangements to diversify their trade to cope with transboundary climate change effects.

The rest of the article is organized as follows: Section 2 reviews the historical evolution and driving forces for China’s soybean demand and Brazil’s supply. Section 3 introduces the climate change data and its impacts on soybean yield in Brazil. Section 4 describes methods and data. Section 5 presents the results of climate change’s impacts on soybean production in Brazil and its implications for China’s food security, and a comparison of Brazilian soybean acreage expansion and China’s virtual land imports. Section 6 concludes the study with some policy recommendations.

2. Historical Evolution and Future Projections of Soybean Demand in China and Supply in Brazil

In reviewing the historical data, we have mainly relied on FAO data sources [9] to analyze China’s soybean imports and Brazil’s soybean exports. Since Brazil, the USA, and Argentina have been China’s top three soybean import source countries, we mainly focus on China’s soybean imports from these countries. Brazil’s soybean is mainly exported to China and Europe, so we mainly analyze the data on Brazil’s soybean exported to China and Europe. Based on these data, we review the previous literature and analyze the reasons behind China’s soybean demand and Brazil’s soybean production and export.

2.1. China’s Demand and Import

In the past 25 years (1995–2020), China’s soybean imports from the world have grown rapidly (Figure 1a). In 1995, China’s soybean import volume was only about 0.29 Mt, which reached a record high of 100.32 Mt in 2020, recording an increase of 340 times. In the recent ten years (2011–2020), China has almost doubled (+91%) its soybean import volume from 52.45 Mt. The numbers are staggering compared to Europe—the second-largest global soybean importer. For example, between 1995 and 2020, Europe’s import of soybean increased by only 26%, from 17.2 Mt to 21.6 Mt. China has undoubtedly become the country with the greatest soybean demand globally. However, the number of China’s soybean supply partners has not increased over the years and are concentrated mainly in Brazil, the USA, and Argentina (Figure 1b). Since 1997, the share of these three countries in China’s gross soybean import has always been more than 94%. Among them, Brazil has gradually replaced the USA as China’s largest soybean supplier. Starting from a meager 2.4% in 1995, China’s soybean imports from Brazil accounted for 75% of China’s total imports in 2018. The USA, China’s top source of imported soybean for the last two decades, is now in second place. Compared with the top two countries, Argentina’s market share in China has gradually declined, accounting for an average of ~7% of the total imports in the past five years.

China’s accession to the World Trade Organization (WTO) and various bilateral trade agreements are vital for soybean trade liberalization [12]. After joining the WTO in 2001, China’s soybean imports remained at over 10 Mt per annum. In 2003, China accepted Brazil’s soybean export application and imported 6.47 Mt of soybean from Brazil, 160% higher than the previous year (3.9 Mt). This figure exceeded 10 Mt in 2006 for the first time, reaching 11.62 Mt. In the later years, Brazil predominantly remained the top soybean supplier to China, surpassing the USA. More recently, the trade conflict between China and the USA in 2017 led to a significant reduction in the quantity of soybean China imported from the USA, whereas Brazil filled the gap.

Although the pattern of China’s high dependence on soybean imports will not change in the near future, many studies point out that China’s soybean imports will continue to grow in the next ten years, but the growth rate will gradually slow down. Huang et al. [13] projected that by 2025, China’s soybean production would reach 11.3 Mt. Compared with that, soybean imports may exceed 100 Mt, accounting for more than 80% of domestic consumption. The total consumption will also accelerate, reaching 116.19 Mt in 2025 and 119.82 Mt in 2029. Under the improved China–US trade relations, China will no longer levy countervailing duty on qualified soybeans for a certain period. China is expected to increase soybean imports from the USA in the future. In the next ten years, China will maintain a market pattern dominated by world soybean imports. It is estimated that China’s global soybean imports will increase to 96.62 Mt in 2025 and 99.52 Mt in 2029 [14]. The results of USDA’s prediction are even higher. It is predicted that by 2029, China’s soybean production will be close to 20 Mt, while the import volume and consumption volume will continue to grow rapidly, reaching 112.5 Mt and 132 Mt, respectively [15]. Some of these predictions were already realized in 2020. For example, China imported >100 Mt of soybean in 2020, and the US exports to China had already started to accelerate, moving to 25.88 Mt in 2020 compared to 16.64 Mt in 2018.

2.2. Brazil’s Production and Export

Brazil’s soybean production has expanded tremendously since the 1990s. From 1995 to 2020, Brazil’s soybean exports soared from 3.48 Mt to 82.95 Mt, increasing 2281% (Figure 2a). On average, Brazil exports 10 Mt more soybean to the world every five years. In 2000, Brazil’s export volume exceeded 10 Mt for the first time, reaching 11.48 Mt and soaring to 22.43 Mt in 2005 and 32.98 Mt in 2011. Subsequently, Brazil’s soybean exports increased significantly, and the export growth rate was much higher than that of the previous ten years. In just two years, i.e., between 2012 and 2013, the export volume exceeded 42.78 Mt. In 2018, Brazil’s soybean export volume reached 83.46 Mt, a record high and ten times the 1998 level. During the same period, Brazil’s soybean share in total global exports rose continuously, moving from 11.6% in 1995 to 48.2% in 2020.

The rapid growth in China’s soybean demand was one of the reasons for the recent surge in Brazil’s soybean exports over the past decade (Figure 2b). Before 2006, Brazil’s leading export destination for soybean was Europe. During 1995–2006, Brazil’s annual average soybean export to Europe was 8.1 Mt, accounting for 66.5% of Brazil’s total exports. During the same period, Brazil’s export to China was 3.4 Mt annually, accounting for 18.2% of Brazil’s total exports. In 2006, China overtook Europe to become Brazil’s largest importing partner in soybean export. From 2006 to 2020, soybean imports from Brazil by China and Europe show opposite trends. China’s average share of Brazil’s export market reached 68.2%, while Europe accounted for 19.1% in the same period. In the past ten years, Brazil’s soybean export volume has increased by leaps and bounds, mainly due to the rapid growth in China’s demand. In 2010, Brazil exported 15.85 Mt of soybean to China and 6.67 Mt to Europe. By 2020, it was 60.59 Mt for China and 9.77 Mt for Europe. In 2020, China accounted for 73% of Brazil’s total soybean exports, with the highest value of 82.5% in 2018. These numbers show that Brazil’s soybean export to Europe has remained stable in the past ten years, meaning that the demand for soybean in Europe has been saturated, while China’s demand is still rising.

Based on the current situation, it is expected that Brazil’s soybean production will continue to rise in the future. With domestic output projected to reach 140 Mt by 2029, Brazil is expected to become the world’s largest producer, ahead of the USA [16]. It is estimated that by 2029, Brazil’s soybean planting area will increase from 35.8 million hectares (Mha) to 45.3 Mha, i.e., a 26.6% hike [17]. The MAPA’s (Ministério da Agricultura, Pecuária e Abastecimento) projection shows that Brazil’s soybean planted area will increase by about 9.7 Mha over the next decade, reaching 46.6 Mha in 2030. Soybean production in 2029 is projected at 156.5 Mt. This number represents an increase of 30.1% over the production of 2019. The soybean export volume will increase by 23.1% from 84 Mt to 103.4 Mt [18].

Not only Brazil’s soybean production (121.97 Mt in 2020) has surpassed that of the USA (112.54 Mt), but also the expected growth rate in Brazil (1.5% per year) will be stronger than that in the USA (0.6% per year) in the next decade, mainly due to the possible increase in planting intensity of soybean under double cropping system. From the perspective of trade, it is expected that the USA’s dominance in global soybean trade will slow down significantly in the next decade, which is directly related to China’s expected slowdown in the growth of imported soybean demand from the USA. At the same time, Brazil will consolidate its position as the world’s largest soybean exporter. By 2029, it is projected that Brazil will account for 48% of total global exports [16], similar to its current level of 48.2% in 2020 [9]. USDA predicts that Brazil’s soybean yield and planting area will continue to expand in the next ten years, increasing the soybean yield steadily at an average annual rate of 3%. According to USDA projections, by 2029, Brazil will produce 158 Mt of soybean and export 97.4 Mt of soybean to the world [15].

3. Climate Change and Changes in Biophysical Yields of Soybean in Brazil

3.1. Climate Change Shocks for Biophysical Yields of Soybean in Brazil

This study focuses on soybean because it is among the top crops produced in Brazil that can be affected by climate change, China’s top imported agricultural commodity, and the most highly traded crop between China and Brazil. The Inter-Sectoral Impact Model Intercomparison Project (ISI-MIP, http://www.isi-mip.org; accessed 10 December 2021) [10] has assessed the global climate change impacts on crop yield in 0.5° × 0.5°grid cells by using multiple global gridded crop models (GGCMs) that were forced with various climate scenarios over 1980–2099. The climate scenarios data were taken from the Coupled Model Intercomparison Project Phase 5 (CMIP5) [11] with five Global Climate Models (GCMs), including GFDL-ESM2M, HadGEM2-ES, IPSL-CM5A-LR, MIROC-ESM-CHEM, and NorESM1-M, under four Representative Concentration Pathways (RCPs). For this study, we applied the Brazil soybean yield simulated by two GGCMs (LPJmL and pDSSAT) without considering the effect of CO₂ concentration under RCP2.6 and RCP8.5 to assess climate-related yield impacts in 2030 (average over 2020–2039) relative to 2011. We used the RCP8.5 and RCP2.6 scenarios to showcase the worst and best climate change scenarios. Thus, a total of 10 GGCM–GCM combinations under each RCP were used in our analysis to consider the uncertainty of climate change shocks on Brazil’s soybean yield in the future.

The climate model projections show that compared to 2011, during the soybean growing season, annual average temperature and precipitation will vary significantly in 2020–2039, with considerably wide variations (Figure 3). Under RCP8.5, the temperature would fluctuate more from negative values (−2.9 °C) to positive values (+4.0 °C), which is much larger than that under RCP2.6 with a range of +0.7 °C to +3.4 °C. The precipitation will vary even more, ranging over −14.2 mm to +10.8 mm under RCP2.6 and −22.6 mm to +46.2 mm under RCP8.5.

For each 0.5° × 0.5°grid, soybean yield is calculated as the weighted yield, where the weights are the proportions of irrigated and rain-fed areas in the grid. Then, we estimate the national soybean yield, equal to an area-weighted average over grids. As shown in

Table 1. The annual impacts of climate change on soybean yield in Brazil (average over 2020–2039) in percentage.

	RCP2.6	RCP8.5
Minimum	−16.05	−17.66
25th percentile	−6.76	−11.95
Median	−3.03	−7.94
75th percentile	1.13	−0.34
Maximum	16.39	39.29
Mean	−1.51	−2.47

Note: The base year is 2011. Source: Simulation results from ISI-MIP [10].

, climate change is projected to affect soybean yield in Brazil differently under both scenarios. As though the mean effect is a reduction in yield under both RCP2.6 (−1.51%) and RCP8.5 (−2.47%), the range is spread over a large area. There are significant differences between most shocks under RCP2.6 and RCP8.5; however, the most severe (minimum) effects are quite similar, i.e., RCP2.6 (−16.05%) and RCP8.5 (−17.66%). Although Brazil’s soybean yield is also projected to expand (positive effects) under some outcomes of the GGCM–GCM combinations, here we only use the mean and minimum effects to showcase the risk that China’s domestic food security might face due to soybean production reduction in Brazil.

3.2. Incorporating Brazil Soybean Biophysical Yield Shocks into Economic Models

Different studies have used a slightly different methods to incorporate crop yield shocks. For example, Roson and Mensbrugghe [19] used ENVISAGE CGE model to simulate changes in multifactor productivity for agricultural activities to simulate the varying agricultural yield. In this approach, the crop output levels can vary while they use the same mix of production factors. Robinson et al. [20] discussed differences in the outcomes of general/partial equilibrium models using different methods of yield shocks. In a globally representative study comparing a wide range of CGE models, Nelson et al. [21] implemented the climate change shocks by changing the land efficiency parameters of the affected crop in a general equilibrium economic model and by additive shifters in a yield or supply equation in the partial equilibrium model setup. Thus, we can summarize that some general equilibrium CGE modeling studies incorporate climate change as shocks to total factor productivity while others use shocks to land efficiency. In this study, we chose to shock the land use efficiency in the CGE model.

4. Methods and Data

4.1. Economic Modeling Methods for Simulating the Impact of Climate Change on China and Brazil’s Soybean Trade

This study uses the GTAP (Global Trade Analysis Project), a multi-region, multi-sector, computable general equilibrium model with perfect competition and constant returns to scale, to analyze the impacts of climate change on soybean production in Brazil and its implications for China’s food security. The model assumes cost minimization by producers and utility maximization by consumers. In a competitive market setup, prices adjust until the supplies and demands of all commodities equalize. The model and database have been extensively used in different research areas such as climate change, food security policy, energy, poverty, migration, and so on [22,23,24,25].

We extended the GTAP model and database (version 9) of 2011 to include the soybean and maize sectors for this study. Two original GTAP sectors were split into four new sectors, following routines from Horridge [26]. Specifically, using production, price, and trade data (from [9,27]), the ‘other oilseeds’ (osd) sector was split into soybean and ‘othosd’, while ‘other coarse grains’ (ogr) was split into maize and ‘othgro’ sectors. We also introduced some structural changes into the standard GTAP database to better represent agricultural input–output relations for China. For developing a reliable database, the key players in global soybean production, consumption, and trade were carefully examined in light of existing data and literature. Then, the original 141 countries/regions of GTAP were aggregated into 12 regions of strategic importance to the research objectives. Our updated GTAP database contains 59 sectors; however, not all sectors are relevant to this research. Therefore, these sectors were aggregated into 24 sectors considering the input–output relations of these sectors with soybean (see Appendix A

Table A1. Sectoral aggregation of the GTAP database.

Aggregation	Original Sectors
Rice	Paddy rice; Processed rice
Wheat	Wheat
Maize	Maize
Other grains	Other cereal grains n.e.c
Vegfru	Vegetables, fruit, nuts
Soybean	Soybean
Other oilseeds	Other Oilseeds
Sugar	Sugar cane, sugar beet, Sugar
Plant-based fibers	Plant-based fibers
Other crops	Crops n.e.c.
Cattle	Bovine cattle, sheep and goats, horses; Bovine meat products
Pork and Chicken	Pork and chicken; Pork and chicken meat products n.e.c.
Milk	Raw milk; Dairy products
Wool	Wool, silk-worm cocoons
Vegetable oils	Vegetable oils and fats
Beverage and Tobacco	Beverages and tobacco products
Processed food products	Processed food products
Fishing	Fishing
Extraction	Forestry; Fishing; Coal, Oil; Gas; Minerals; Mineral products; Ferrous metals; Metals; Metal products
Light Manufacturing	Textiles; Wearing apparel; Leather products; Wood products; Paper products, publishing; Petroleum, coal products; Chemical, rubber, plastic products; Motor vehicles and parts; Transport equipment; Manufactures
Heavy Manufacturing	Electronic equipment; Machinery, and equipment n.e.c.
Utilities and Construction	Electricity; Gas manufacture distribution; Water; Construction
Transport and Communication	Trade; Transport others; Water transport; Air transport; Communication
Services	Financial services; Insurance; Business services; Recreational and other services; Public Administration, Defense, Education, Health; Dwellings

and

Table A2. Regional Aggregation of GTAP database.

Aggregation	Original Regions
Australia New Zealand	Australia; New Zealand; Rest of Oceania
China	China
Southeast Asia	Brunei Darussalam; Cambodia; Indonesia; Japan, Lao PDR; Malaysia; Philippines; Singapore; Thailand; Viet Nam; Rest of Southeast Asia
Canada	Canada
USA	United States of America
Mexico	Mexico
Argentina	Argentina
Brazil	Brazil
European Union	Austria; Belgium; Cyprus; Czech Republic; Denmark; Estonia; Finland; France; Germany; Greece; Hungary; Ireland; Italy; Latvia; Lithuania; Luxembourg; Malta; Netherlands; Poland; Portugal; Slovakia; Slovenia; Spain; Sweden; United Kingdom; Bulgaria;
Russia	Russian Federation
Africa	Benin; Botswana; Burkina Faso; Cameroon; Côte d’Ivoire; Egypt; Ethiopia; Ghana; Guinea; Kenya; Madagascar; Malawi; Mauritius; Morocco; Mozambique; Namibia; Nigeria; Rest of Central Africa; Rest of Eastern Africa; Rest of North Africa; Rest of South African Customs Union; Rest of Western Africa; Rwanda; Senegal; South Africa; South Central Africa; Tanzania, the United Republic of; Togo; Tunisia; Uganda; Zambia; Zimbabwe
Rest of the world	Albania; Armenia; Azerbaijan; Bahrain; Bangladesh; Belarus; Bolivia; Chile; Colombia; Costa Rica; Croatia; Dominican Republic P; Ecuador; El Salvador; Georgia; Guatemala; Honduras; Hong Kong; India; Iran, the Islamic Republic of; Israel; Jamaica; Jordan; Kazakhstan; Korea, Republic of; Kuwait; Kyrgyzstan; Mongolia; Nepal; Nicaragua; Norway; Oman; Pakistan; Panama; Paraguay; Peru; Puerto Rico; Qatar; Rest of Caribbean; Rest of Central America; Rest of East Asia; Rest of Eastern Europe; Rest of Europe; Rest of European Free Trade Association; Rest of Former Soviet Union; Rest of North America; Rest of South America; Rest of South Asia; Rest of the World; Rest of Western Asia; Romania; Saudi Arabia; Sri Lanka; Switzerland; Taiwan; Trinidad and Tobago P; Turkey; Ukraine; United Arab Emirates; Uruguay; Venezuela

for sectoral and regional aggregation schemes).

4.2. Baseline Scenario

We have used the recursive-dynamic simulations method on the GTAP model for developing the business-as-usual scenario from 2011 into 2030. The multi-period simulation results are computed one period at a time in these simulations, using the previous period’s data as base data for the next period. In this approach, the given GDP targets are met through exogenous estimates of factor endowments—population, unskilled labor, skilled labor, capital, and natural resources. Hertel [28] and Walmsley et al. [29] discuss the procedure and the exogenous macro assumptions for this approach.

4.3. Climate Change Scenario

In our study, the yield changes due to climate change were incorporated in the GTAP model by shocking the land use efficiency for the land used by soybean production in Brazil (parameter ‘afe’ in Equation (1). This is a generally accepted method for incorporating yield changes into economic models [21,30,31].

$$pv{a}_{j,r}={{\displaystyle \sum }}_{k=1}^n(SV{A}_{k,j,r}*\left(pf{e}_{k,j,r}-af{e}_{k,j,r}\right)),$$

(1)

where $j=$ production commodity (industry); $r=$ region; $k=$ endowment commodity; $pva=$ firm price of value added in industry $j$ of region $r$; $pfe=$ firm price for endowment commodity $k$ in industry $j$, region $r$; $SVA=$ share of k in total value added in $j$ in $r$; $afe=$ sector/region-specific average rate of primary factor $k$ augmenting technology change. In order to isolate the influences of climate impacts of soybean production in Brazil on China, we fixed China’s domestic production, other countries’ production, and imports of soybean from other countries to the pre-climate change levels.

5. Results and Discussion

5.1. Climate Change Impacts on Soybean Production in Brazil and Implications for China’s Food Security

Among many risks facing soybean supply, climate change is perhaps the most frequently debated in the literature that can affect Brazil’s soybean production and disrupt China’s soybean supply by reducing its imports. The reduced soybean supply may impact both food supply and prices in China. Below, we use our climate-crop–economic modeling approach to analyze these effects in detail.

In 2030, with the assumption that there is no effect of CO₂ fertilization on yield, the mean effect of the climate change shock on Brazil’s soybean yield would be a 1.51% decline under the scenario of lowest emissions (RCP2.6) and 2.47% decline under the scenario of highest emissions (RCP8.5). However, these climate effects will be pretty significant under the worst-case scenario, which we call ‘extreme shocks.’ Under this scenario, Brazilian soybean yield will drop by 17.66% under RCP8.5 and 16.05% under RCP2.6.

Our economic model results indicate that soybean yield shocks of −17.66% under RCP8.5 and −16.05% under RCP2.6 would cause Brazil’s soybean production to drop by 13.06% and 11.78%, respectively (

Table 2. Impacts of climate change on Brazil’s soybean production and China’s food security in 2030.

	Indicator	Baseline Quantity (Mt)	Climate change Impacts
			Scenarios	Percentage	Quantity (Mt)
Brazil	Production	147.5	RCP8.5	−13.06 * (−1.7) **	−19.26 (−2.507)
			RCP2.6	−11.78 (−1.04)	−17.373 (−1.534)
	Total export	85.4	RCP8.5	−15.25 (−1.91)	−13.028 (−1.632)
			RCP2.6	−13.7 (−1.16)	−11.704 (−0.991)
China	Total import	107.6	RCP8.5	−9.244 (−1.131)	−9.942 (−1.217)
			RCP2.6	−8.282 (−0.689)	−8.908 (−0.741)
	Import from Brazil	69.9	RCP8.5	−14.22 (−1.74)	−9.942 (−1.217)
			RCP2.6	−12.74 (−1.06)	−8.908 (−0.741)
	Total domestic supply	119.1	RCP8.5	−8.352 (−1.022)	−9.947 (−1.217)
			RCP2.6	−7.482 (−0.623)	−8.912 (−0.741)
	Soybean price (%)	-	RCP8.5	0.24 (0.03)	-
			RCP2.6	0.21 (0.02)	-
	Welfare (billion US$)	-	RCP8.5	-	−4.53 (−0.579)
			RCP2.6	-	−4.077 (−0.352)

* Impacts of most adverse shocks ** Impacts of average shocks.

). Compared to the extreme shocks, Brazilian soybean production will decrease only moderately due to the average shocks, i.e., 1.70% under RCP8.5 and 1.04% under RCP2.6. Production changes are usually lower than yield shocks because some of the adverse effects of climate change are absorbed by farmers through their resilience in terms of higher inputs and improved management practices. Brazil’s total soybean production of ~147.5 Mt under the baseline by 2030 will drop by 19.3 Mt under RCP8.5 and 17.4 Mt under RCP2.6 due to extreme climate shocks. Other studies, such as Feng et al. [32], have also predicted that suitable soybean cultivation areas in Brazil will decline significantly in the future. Zilli et al. [5] found that Brazil’s future production of soybean and corn will decline under RCP2.6 and RCP8.5 scenarios.

Because a big chunk of Brazil’s soybean production is exported to other countries, the decrease in soybean production will directly affect Brazil’s soybean exports. Due to the extreme shocks, the country’s gross soybean export to other countries will decline by ~15.2% under RCP8.5 and ~13.7% under RCP2.6. The volume effects will be equivalent to ~13 Mt under RCP8.5 and ~11.7 Mt under RCP2.6, which are quite large compared to the total baseline exports of 85.4 Mt. The average climate shocks will moderately affect Brazil’s soybean exports −1.9% (−1.6 Mt) under RCP8.5 and −1.2% (−1 Mt) under RCP2.6. Zilli et al. [5] also indicated that Brazil’s soybean export volume and its share of the international market would decrease under climate change in the future.

Our baseline projections show that China will import a large volume of soybean in the future, tallying 107.6 Mt of global imports (Table 2). High dependence on Brazil will continue for this import, and China will import 69.9 Mt (65% of China’s total soybean imports) from Brazil in 2030. A large part (90%) of China’s total domestic supply of soybean (119.1 Mt) will be met from global imports, where imports from Brazil will contribute a substantial (59%) share. High soybean imports mean China will have a low self-sufficiency ratio (SSR) of ~10% in soybean supply. Under the extreme climate change shocks on soybean production in Brazil, China’s global soybean imports will decrease by 9.2% under RCP8.5 and 8.3% under RCP2.6. The corresponding volumetric impacts will be −9.9 Mt under RCP8.5 and −8.9 Mt under RCP2.6. These import reduction results are similar to the ones found by Xie et al. [33], who showed that under RCP8.5, China’s net import for soybean will decline by 13.71% (or ~10 Mt) in 2050 due to global impacts of climate change. Our results show that the average climate shock will have minimal effects on China’s total soybean imports, i.e., −1.2 Mt under RCP8.5 and −0.68 Mt under RCP2.6. To isolate the impacts of climate change on soybean production in Brazil on China, we have set up our model to keep China’s imports from other countries static even though imports from Brazil will fluctuate. The percent changes in China’s soybean imports from Brazil will be much larger than global impacts, i.e., −14.2% under RCP8.5 and −12.7% under RCP2.6. However, the quantity effects will be the same size, i.e., −9.9 Mt under RCP8.5 and −8.9 Mt under RCP2.6, because we have only perturbed Brazil’s soybean yield under the climate change scenario and left the yield in other countries unchanged. China’s soybean imports from Brazil will also follow similar trends under the average climate shocks, albeit with smaller sizes.

As China heavily depends on foreign markets for soybean supply, the decline in Brazil’s soybean exports to China will reduce China’s domestic supply significantly. From the baseline level of 119.1 Mt, China’s domestic soybean supply will drop 8.35% (9.9 Mt) under RCP8.5 and 7.5% (8.9 Mt) under RCP2.6. Notice that despite smaller percentage changes, the quantity changes in China’s domestic soybean supply are much the same as its imports. The reason is that although we have added climate shock to China’s domestic soybean production, the change in China’s domestic soybean production plays a minimal role due to the large share of imports in its domestic supply. China’s domestic soybean supply will only drop moderately under the average climate shocks, i.e., 1% (1.22 Mt) under RCP8.5 and 0.6% (0.74 Mt) under RCP2.6. China’s domestic soybean price will also rise moderately, i.e., 0.24% under RCP8.5 and 0.21% under RCP2.6 (Table 2). Mosnier et al. [34] have also shown that adverse impacts of climate change on trading partners lead to higher import prices.

The second-order effect of reduced soybean import in China would be on the meat supply, which uses soybean as feedstock. The effects on China’s meat supply are marginal because China can substitute other feedstocks for meat production. However, the effects of reduced meat supply and higher soybean prices will be felt through reduced consumer welfare, which will drop by 4.53 billion US$ under RCP8.5 and 4.02 billion US$ under RCP2.6. The lower welfare highlights reduced options for China’s consumers to purchase meat and related products that depend heavily on imported soybean.

5.2. Comparison of Brazil’s Soybean Harvested Area and China’s Virtual Land Imports from Brazil

In this section, first, we analyze the historical changes in Brazil’s soybean-harvested area and China’s virtual land (VL) imports through soybean for Brazil and any correlation between them. Then, we analyze the future expansion of Brazil’s soybean acreage and China’s VL imports from Brazil.

Starting in 1990, the soybean harvested area was 11.5 Mha, accounting for 22.4% of the country’s total harvested area of 51.4 Mha (Figure 4a). The soybean harvested area grew gradually over the next two decades, reaching 23.3 Mha in 2010, or 35.5% of Brazil’s total harvested area. One of the major reasons behind Brazilian soybean expansion in this era was the US corn ethanol production that pushed global soybean prices up [35,36]. In the recent decade leading up to 2020, the soybean harvested area in Brazil increased at a much faster pace when it moved to 37.2 Mha and occupied the historical high point of 44.4% in the national total harvested area. China’s soybean imports have been cited as one of the main driving factors behind this increase in soybean acreage in the last few years [37].

We estimate China’s virtual land imports through soybean trade using the term virtual land content (VLC), i.e., the quantity of land required to produce one tonne of crop biomass. VLC is akin to the inverse of per unit crop yield, usually expressed in ha tonne⁻¹. Several studies have used this concept to estimate virtual land trade and the ensuing land savings (or dissavings) due to regional differences in crop yields (e.g., [3,38,39]). Our estimations, based on FAO’s yield and crop trade data [9], show that due to very low imports of soybean from Brazil in the early 1990s, China’s net virtual land (VL) imports from Brazil were also negligible (Figure 4a). It was in 2000 that China’s VL imports through soybean trade from Brazil reached some considerable level with a value of 0.88 Mha, roughly 6.5% of Brazil’s soybean harvested area in the same year. Between 2000 and 2005, China’s soybean-based VL imports from Brazil recorded the fastest-ever growth of 304%, reaching 15.5 Mha, with the VL imports accounting for 15.5% of Brazil’s soybean harvested area. This period’s growth in China’s VL imports from Brazil was several times faster than that of the Brazilian soybean harvested area (68%). After 2005, China continuously imported high quantities of VL through soybean trade from Brazil, which stood at 6.31 Mha in 2010 and 13.23 Mha in 2015. In the later period, 2018 was a significant year such that, for the first time, China’s soybean-based VL imports from Brazil (19.49 Mha) were more than half (56%) of Brazil’s soybean harvested area (34.78 Mha). In 2020, the Brazilian soybean harvested area had reached 37.19 Mha, of which China imported 19.63 Mha in the virtual form (VL).

The difference in soybean per unit yield in China and Brazil can also save land resources in China and globally through China’s imports of VL from Brazil, i.e., if China imports VL from Brazil in a situation where Brazil has a higher yield, it would save the global land resources. China can save its domestic land resource by importing soybean from Brazil, for which the former does not have to use domestic land to grow the same quantity of land. Since the beginning of the 21st century, China’s domestic land savings through soybean imports from Brazil has grown steadily, i.e., from 1.28 Mha in 2000 to 10.49 Mha in 2010. In the later decade, domestic savings grew much faster, i.e., from 11.23 Mha in 2011 to 32.36 Mha in 2020. China’s most recent domestic savings were 328% higher than the soybean harvested area in 2020. The VL imports through soybean imports from Brazil by China also contributed significantly to global land savings, as China’s soybean yield has always been lower than Brazil’s. For example, global land savings related to China’s soybean imports from Brazil grew 304%, from 4.19 Mha in 2010 to 12.73 Mha in 2020. These findings are similar to several other studies [3,39].

The above analysis shows that although China’s soybean imports did not play any significant role in Brazilian soybean acreage in the 1990s, they began to contribute to the Brazilian soybean acreage expansion in the early 2010s. With the fast-rising soybean imports by China from Brazil in the early 2010s and the significant diversion away from the USA in the late 2010s, China’s imports started playing an increasingly significant role in the soybean acreage expansion in Brazil. As projected by other studies, China’s future soybean imports will be kept at a high level, and most of these imports will be sourced from Brazil [13]; the increasing VL imports will continue to play a significant role in Brazilian soybean acreage.

It is noted here that several factors might cause the acreage expansion in Brazil in recent decades, and we find that China’s VL imports are one the most critical factors behind this expansion. Brazil’s soybean acreage and exports have continuously increased since the 1990s. However, the share of Brazil’s domestic use of soybean (described as ‘domestic supply = production + imports − exports + stock variation’ by FAO, 2022) in its total production has decreased during the same time. Specifically, the share of Brazil’s domestic use of soybean decreased from 85% in 1990 to 68% in 2000, 55% in 2010, and 42% in 2019 [9]. The share of Brazil’s soybean exports in total production increased from 20% in 1990 to 35% in 2000, 42% in 2010, and 65% in 2019. In the initial years, most of Brazil’s soybean was exported to Europe; starting in the mid-2010s, China became the dominant destination for Brazil’s soybean exports. Additionally, Brazil’s soybean production has increased faster (62.8%) than its yield (4.9%) in the last decade, indicating that much of Brazil’s soybean production has come from area expansion. We use the concept of VL imports to show that Brazil’s soybean acreage increased with China’s VL imports in recent years. The VL trade approach has been used in several other studies that showed China’s increasing imports of VL, particularly through soybean trade, e.g., [3,38,39,40,41,42].

Finally, we assume that Brazil’s current soybean yield will hold for 2030 while the harvested area will expand to 45.04 Mha (Figure 4b), which would be required to produce 147.5 Mt of soybean under the baseline, as shown in Table 1. Due to the data limitations, we cannot project whether this additional land will come by reducing the harvested area for other crops, forest, livestock production, or any combination of these. The same harvested area is assumed under the climate change scenarios, and Figure 4b shows this for RCP8.5, where soybean yield will decline due to the lower land productivity; thus, the same quantity of land will produce less soybean. We also project a ‘counterfactual’ scenario where Brazil will try to ensure that its baseline soybean production of 147.5 Mt is maintained under climate change, which will require 54.69 Mha of harvested area. For China’s VL imports, 21.34 Mha of VL will be imported by China under the baseline scenario in 2030. Under the extreme case of RCP8.5 scenarios, China’s VL imports will drop to 18.3 Mha, i.e., a 14.2 reduction. Under the ‘counterfactual’ scenario, China’s VL imports will be higher (25.11 Mha) due to an increase in Brazil’s VLC under climate change. Due to China’s reduced soybean imports from Brazil under RCP8.5, the lower VL imports would mean that not only China’s domestic savings of land would decline, but it will also reduce the global savings of land resources that were happening due to the differences between Brazil and China’s virtual land contents. Under the ‘counterfactual’ scenario, China’s domestic land savings will increase (assuming that China’s soybean yield will not be affected by climate change), while global land savings will decrease (due to lower yield in Brazil).

Here, we discuss three (of many) possible outcomes of climate change and soybean area changes in Brazil in the future. We have shown a range of links between China’s VL imports and Brazil’s soybean area expansion, while we note that other scenarios could also emerge, such as faster improvement in soybean yield and higher pressure on soybean acreage from climate impacts on other crops, among others. These scenarios would yield varying results that might differ from ours.

6. Conclusions

China and several other developing countries partially rely on imported food to feed their population. For China, soybean is among the main imported food items whose imports have gone through several phases in the past, and several factors pose substantial risks to China’s future soybean supply. This study has analyzed the historical evolution of China’s soybean demand and import and Brazil’s soybean production and export. It also analyzed China’s food security vulnerability arising from future climate change effects on Brazil’s soybean production. Based on our results, we have three main conclusions. First, in recent decades, China’s soybean demand has expanded tremendously, and the country has increasingly relied on imported soybean to feed its livestock, making it the world’s largest soybean importer in recent years. China’s increasing soybean imports have been driven mainly due to its open trade policies. The future soybean demand and imports will likely rise to high levels due to higher domestic demand for meat and dairy products.

Second, China initially started to expand its soybean imports from the USA. However, in recent years, Brazil has overtaken the USA as a major soybean supplier to China. Moreover, most of Brazil’s soybean is destined for China’s market, replacing the European Union from a decade ago. Soybean has become a major trading commodity between China and Brazil’s markets. This has led to rapid expansion in Brazil’s soybean production that relied on both yield improvement and area expansion. China’s VL imports through soybean imports from Brazil have increased over the years and contributed significantly to China’s domestic land savings (due to land area spared from cultivating the same quantity of imported soybean) and global savings of land resources (because China imported relatively high yield soybean from Brazil).

Third, China’s soybean supply and food security will face several risks in the future. Among these risks, future uncertainty in Brazil’s supply capacity due to climate change is one of the most serious. Our analysis shows that due to adverse climate impacts on Brazil’s soybean production, China’s soybean imports will decrease significantly in the future. The lower soybean imports will cause the domestic supply to dwindle. Consequently, China’s meat supply will decrease while meat and soybean prices increase. Moreover, China’s historically increasing trend of large VL imports from Brazil will likely slow down under climate change.

Quantitatively analyzing the future climate change impacts sheds new light on the ways in which climate change risks can have transboundary food security impacts on a global scale. Free trade has partially helped China meet its domestic food needs by switching to key food crops (wheat and rice) and importing other crops such as soybean and maize. However, meeting the SDG2 targets of food security would not be possible without carefully evaluating climate change risks and food trade pathways. Such evaluation should be at the core of food security and sustainable agriculture policies. Based on our results, we suggest some policy options. China should opt for yield improvement by using better soybean varieties such as genetically modified seeds to increase domestic production and reduce reliance on unstable foreign food supplies. China should benefit from the enhanced cooperation in agriculture between its trading partners, such as countries in the Belt and Road Initiative, to efficiently use the collective resources for improved agricultural production and reduce transboundary risks. China can also leverage trade diversification to reduce the risks related to highly concentrated imports of soybean and similar food commodities. Brazil can reduce its domestic risks to soybean production by adopting climate-resilient soybean varieties and technologies. China’s soybean trade links are based on historical, regional, and geopolitical ties between its trading partners. Like many other close trading partners, China and Brazil should jointly enhance robust bilateral cooperation to address these shared, systemic risks facing their food systems.

Similar to many studies, our study has some limitations that need to be mentioned here. Our climate change results should be interpreted cautiously as they are based on only two climate scenarios. Future studies should add more scenarios to provide a full range of climate change impacts on soybean production. Future studies should incorporate the impacts of extreme weather events on agricultural production and critical trade-related infrastructure. In more detailed studies, the parametric settings of the economic model should also be tested for sensitivity. Analyzing the impacts of climate change on land use change in both exporting and importing countries is another area where our study had limited analysis and should be studied in future research. Comparing the dynamics of areas under crops with the dynamics of China’s imports might not show a more direct correlation between China’s virtual land (VL) imports and Brazil’s soy acreage increase. Future studies should further expand on this area of research.