Effects of Agricultural Trade on Reducing Carbon Emissions under the “Dual Carbon” Target: Evidence from China

Abstract

Within the “dual carbon” framework, sustainable agriculture is pivotal for climate change mitigation and long-term agricultural health. To explore the environmental effects of agricultural trade, this study assesses the carbon emissions from agriculture using information from 30 provinces in China (excluding Tibet, Hong Kong, Macao, and Taiwan due to a lack of available data) from 2001 to 2021. Then, we analyzed the impact of agricultural trade on agricultural carbon emissions and tested for the possible existence of mechanisms. Finally, we validate the synergistic effects of agricultural trade on pollution and carbon abatement. The findings were as follows: (1) Agricultural trade significantly suppresses agricultural carbon emissions. (2) Agricultural technological progress and the rationalization of the structure of the agricultural industry are the two influencing mechanisms. (3) The inhibitory effect of agricultural trade on agricultural carbon emissions is more pronounced in southern regions and regions with lower degree of topographic relief and higher agricultural industrial agglomeration. (4) Agricultural trade fosters the collaborative advancement of mitigating agricultural pollution and reducing carbon emissions. As a result, this study confirms that agricultural trade has a role in reducing carbon emissions and that giving full play to the environmental effects of trade in agricultural products contributes to the high-quality development of agriculture.

1. Introduction

Environmental problems have become a significant global challenge. Environmental pollution is a crucial constraint to sustainable economic growth and social development [1]. Extreme weather, aridification of soil, and other environmental problems have brought great harm to people’s lives [2]. According to the United Nations Environment Programme’s Emissions Gap Report 2023: Broken Record, Temperatures hit new highs, yet world fails to cut emissions (again), there were a total of 86 days in which temperatures were more than 1.5 °C above pre-industrial levels up to the beginning of October 2023, with September emerging as the hottest month on record. The average global temperature exceeded pre-industrial levels by 1.8 °C throughout this timeframe. Unless countries step up their actions and meet their 2030 commitments under the Paris Agreement, the world will be heading toward a temperature rise of between 2.5 and 2.9 degrees Celsius from pre-industrial levels [3]. These clear indications demonstrate that reducing carbon emissions has emerged as a crucial instrument in the ongoing battle against climate change. It not only helps to mitigate the escalating pace of global warming but also plays a vital role in safeguarding the Earth’s ecological environment and promoting sustainable development. At the 75th session of the United Nations General Assembly, General Secretary Xi proposed the “dual carbon” target, stating in particular that China will strive to achieve carbon peaking by 2030 and work towards carbon neutrality by 2060 [4]. As the world’s second-largest economy and a major carbon emitter, China is taking action to achieve its “dual carbon” goal. According to relevant data, approximately 25% of global carbon emissions come from the AFOLU sector [5]. China, a significant agricultural producer, produces large quantities of food and essential agricultural products and has long ranked first in the world in the production of cereals in particular. At the same time, it is responsible for around 29% of agricultural carbon emissions in Asia and approximately 12% of global carbon emissions in 2017 [6]. Given that China’s agricultural sector produces 17% of the nation’s overall carbon emissions, the imperative to explore low-carbon strategies for developing China’s agriculture has become paramount [7].

The agricultural sector is central to achieving net-zero emissions as a substantial contributor to greenhouse gas emissions worldwide and a considerable carbon sink system [8]. Agricultural emissions reduction and carbon sequestration are necessary to achieve a peak carbon, and their potential is enormous. As the primary sector of the national economy, agriculture is highly vulnerable to external climate change, and its production process inevitably generates negative externalities [9]. Agriculture plays a crucial role in China’s national economy [10], and a stable supply of food and essential agricultural products for a population of more than 1.4 billion is always a top priority. Sustainable agricultural development addresses the challenges of enhancing crop production, boosting farmers’ earnings, and improving the quality of life in rural communities while also holding substantial importance in safeguarding the environment.

From a global trade perspective, China is not only a significant importer of agricultural products but also a major exporter and has now firmly established itself as the world’s second-largest agricultural trading nation. With the signing of agreements such as the Belt and Road Initiative and the Regional Comprehensive Economic Partnership, the opening up of China’s trade has been dramatically enhanced—especially in agriculture. According to data released by the Ministry of Commerce of the People’s Republic of China, China’s commerce in agricultural products reached USD 304.17 billion in 2021, with a year-on-year increase of 23.2%, accounting for the continued rise in the proportion of total trade in goods. Agricultural trade is widely regarded as one of the crucial elements affecting the agricultural environment. The structural, scale, and technological impacts contribute to its intricate influence on reducing carbon emissions [11]. Comprehensively analyzing the impact of China’s agricultural trade on the environment within the framework of the “dual carbon” goal, better clarifying the environmental effects exerted by agricultural trade, and clarifying the intrinsic influence mechanism therein will be conducive to exploring the potential driving force for the sustainable development of agriculture from the perspective of agricultural trade in the context of increasingly serious environmental problems. This will be crucial to the modernization and sustainable development of agriculture.

The existing literature on agricultural trade and agricultural carbon emissions primarily concentrates on the quantification and determinants of agricultural carbon emissions, as well as the influencing factors and environmental effects of agricultural trade. Scholars have primarily focused on studying the measurement of agricultural carbon emissions and the factors that influence them. When measuring these emissions, scholars predominantly consider aspects such as the handling of agricultural waste, methods of raising livestock and poultry, the utilization of energy in agriculture, and the cultivation of rice [12,13]. Regarding the determinants of carbon emissions in agriculture, scholars have studied them from the perspective of natural and social factors. Natural factors mainly include energy poverty, soil, and water resources [14,15]. At the level of social factors, the main aspects include economic growth [10], digitization [16], and urbanization [17]. Regarding the research on the factors influencing agricultural trade, scholars have explored them from external environmental perspectives such as trade facilitation [18], regional bias [19], coronavirus [20], etc. There are also many scholars who have based their research on internal national factors such as economic and social development status, including infrastructure [21] and land abundance [22]. Moreover, with the continuous development of open trade, its environmental effects have gradually attracted the research attention of scholars. Most scholars believe that the environmental effects of agricultural trade are complex. Several researchers have employed spatial panel modeling to find that agricultural trade exerts a restraining influence on agricultural carbon emissions [23].

In the long term, trade can reduce carbon emissions [24]. The trade–environment general equilibrium analytical model of Grossman and Krueger (1991) [25] and the general equilibrium pollution–trade model of Antweiler et al. (2001) [26] provide theoretical references for studying the environmental effects of agricultural trade. First, the environmental consequences of trade can be assessed by examining three fundamental aspects: magnitude, composition, and technological advancements [27]. The scale effect means that any increase or reduction in agricultural trade leads to changes in the input factors of agricultural pollutants, leading to dynamic changes in agricultural carbon emissions. The effect of technology is to promote the adoption of advanced green production methods through trade in agricultural products, enhancing technical efficiency [28] and forcing a technological revolution in agriculture to reduce agricultural carbon emissions. At the level of structural effects, agricultural trade contributes to enhancing the quality of agricultural produce, using more environmentally friendly factors of production, and adjusting the factor input structure to improve the framework of the agricultural sector, which will impact carbon dioxide emissions from agricultural activities. Second, agricultural trade has effects on the extent to which natural resources are utilized in different regions, the pollution generated by changing relative output prices, and the production process. Trade-induced changes in consumption, which lead to changes in the demand for environmental goods, further lead to the need to regulate pollution and natural resources [29]. In addition, there are green barriers in the agricultural trade process. To meet the importing country’s food green standards, agricultural producers must take the initiative to improve the green attributes of their products and reduce environmental pollution. This study explores the impacts of agricultural trade on agricultural carbon emissions from the aspects of scale and technology effects. The mechanistic analysis is shown in Figure 1.

The advancement in technology in the field of agriculture plays a crucial role in attaining higher levels of agricultural output while simultaneously reducing carbon emissions. Based on endogenous growth theory, advancements in technology have the potential to enhance resource utilization and decrease energy consumption [30]. For instance, technological advances in agriculture not only change energy consumption patterns but also improve resource efficiency in the agricultural sector, thereby promoting sustainable agricultural development [31]. Moreover, advancements in technology not only decrease the cost of reducing carbon emissions but also impact the distribution of agricultural resources, alter the composition of energy resources, and enhance the societal benefits of reducing carbon emissions [32]. Trade can aid in the dissemination of production technologies. Through trade, technology can be transferred indirectly through knowledge inputs, allowing importers to benefit from advanced technology. The industrial structure is an important indicator reflecting the allocation of factors among industries and sectors, and upgrading the industrial structure is conducive to reducing carbon emissions [33]. Based on the theory of trade liberalization, nations engaged in the free trade of agricultural goods will modify their agricultural industrial structure in accordance with their comparative advantage in export trade, driven by their trade interests. Agricultural trade causes factor resets, prompts the internal structural adjustment of agriculture, and reduces the proportion of energy-intensive production sectors to upgrade the industry. On the one hand, import competition can significantly enhance the optimization of the agricultural industrial structure. With the evolution of the agricultural industry’s structure, there is an increasing need for cost reduction and enhanced management practices, ultimately leading to improved agricultural energy efficiency. On the other hand, from the point of view of the income effect, export diversification facilitates capital accumulation [34], and capital accumulation through trade activities will have an indispensable impact on upgrading the industrial structure. In this process, the digestion, absorption, and transformation of advanced technology via the “technology spillover” phenomenon and the “learning by doing” approach can further facilitate the internal enhancement of the industrial framework. The industrial structure and carbon emissions share a two-way dynamic relationship, where enhancing the agricultural industry can effectively decrease agricultural carbon emissions by restructuring agriculture internally and transitioning traditional high-energy-consuming and high-polluting industries into energy-efficient and eco-friendly sectors. Accelerating the upgrading of the industrial structure is conducive to improving the efficiency of energy utilization in agricultural production, which helps reduce carbon emissions and mitigate the greenhouse effect [35]. Based on the above analysis, hypotheses are proposed:

Hypothesis 1. 

Trade in agricultural products has a dampening effect on agricultural carbon emissions.

Hypothesis 2. 

Agricultural trade curbs agricultural carbon emissions through technological advances in agriculture.

Hypothesis 3. 

Agricultural trade suppresses agricultural carbon emissions by promoting the structural upgrading of the agricultural industry.

Scholars in the fields of agricultural carbon emissions and agricultural trade have carried out relatively extensive research [36,37,38], the results of which constitute the theoretical basis for this study. However, there is still room for improvement. This study’s innovations primarily consist of the following three points: (1) it investigates the impact of agricultural trade on environmental outcomes and the dual advantages of decreasing both pollution and carbon emissions. To date, the analysis of agricultural trade from the perspective of environmental effects has not yet been perfected, especially the study of the synergistic effect of agricultural trade on pollution reduction and carbon reduction, which still has a more fantastic research space and potential. The present study further verifies the combined impact of pollution reduction and carbon reduction, broadening the relevant research on the trade of agricultural products. (2) It explores the effect of trade in agricultural products on agricultural carbon emissions from a mechanistic perspective. Empirical research on the structural and technological effects brought about by agricultural trade provides theoretical references for the direction of green development of the agricultural sector. (3) It enriches the literature on the factors that impact agricultural carbon emissions. Figure A1 in Appendix A shows the research framework of this paper.

2. Methods

2.1. Modeling

In order to test the impact of agricultural trade on agricultural carbon emissions, we chose a time-region double fixed-effects model to reduce the endogeneity problem and absorb the effects of common regional trends and temporal fluctuations [39]. On this basis, standard errors were used to further eliminate the impact of heteroskedasticity on the test results. The model was constructed as follows:

$$LnCa{r}_{it}={\alpha }_0+{\alpha }_{1}AgTr{a}_{it}+{\alpha }_2{Control}_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(1)

where the subscripts i and t represent the region and year, respectively; LnCar represents the level of agricultural carbon emissions; AgTra represents the level of import and export trade of agricultural products; Control represents the control variable; ${\mu }_i$ represents the regional fixed effect; ${\lambda }_t$ represents the temporal fixed effect; and ${\epsilon }_{it}$ represents the attendant error term.

2.2. Variable Selection

The primary dependent variable chosen for examination in this study was the extent of agricultural carbon emissions. Despite agriculture being a significant contributor to carbon emissions, China has not yet implemented official channels to count and publicize the carbon emissions stemming from agricultural activities. Using the available data, and drawing on relevant studies from the literature [40,41], this study mainly defines the sources of agricultural carbon emissions in four categories: agricultural factor inputs, agricultural cultivation, biomass combustion, and animal breeding in the agricultural production process. On this basis, agricultural carbon emissions in each province were calculated using the available data, with the help of Formula (2), and a logarithmic treatment was carried out.

$$C_t={\displaystyle \sum _{i=1}^5{\delta }_{1i}{\omega }_{1i}}+{\displaystyle \sum _{i=1}^2{\delta }_{2i}{\omega }_{2i}}+{\displaystyle \sum _{i=1}^2{\delta }_{3i}{\omega }_{3i}}+{\displaystyle \sum _{i=1}^8{\delta }_{4i}{\omega }_{4i}}$$

(2)

Here, C_t represents the carbon emissions from agriculture in year t. ${\omega }_{1i}$ represents four agricultural factor inputs, including amount of agricultural plastic film used, the amount of pesticides used, the amount of pure fertilizer applied, the effective irrigated area, and the amount of agricultural diesel oil used; ${\omega }_{2i}$ represents agricultural cultivation, including the area of crops sown and the area of rice sown; ${\omega }_{3i}$ represents biomass burning, including the outputs of paddy rice, wheat, corn, soybeans, rapeseed, and cotton; and ${\omega }_{4i}$ represents animal breeding, including the end-of-period numbers of cows, horses, donkeys, mules, camels, pigs, goats, and sheep. δ represents the corresponding carbon source type. Because of the unpredictable nature of the rice production cycle, a median of 130 days was established as the standard for accounting for production cycles. Both methane and nitrogen oxides were considered in measuring carbon emissions from animal farming.

The independent variable was the level of trade in agricultural products. First, the import and export trade of agricultural products of each province in the sample period was converted to be measured in RMB units through the exchange rate of USD to RMB. Then, the level of agricultural trade was characterized by the proportion of total agricultural trade to the value added by the primary industry.

The control variables selected in this study mainly include the replanting index, financial support for agriculture, degree of disaster, land quality, education level of rural residents, and consumption level of rural residents. All these variables influence agricultural carbon emissions. Resident consumption is one of the important factors influencing carbon emissions [42]. An increase in the consumption level of rural residents implies an increase in rural living standards. This is conducive to increased investment in technological advances in agricultural production, creating a more favorable environment for sustainable low-carbon agricultural practices. Degree of disaster includes critical disasters affecting agriculture, such as droughts, floods, wind and hailstorms, freezes, and typhoons suffered during the year. Higher levels of damage may affect the sown area and yields of crops, thereby reducing agricultural carbon emissions. Financial support may stimulate scientific and technological advancements in agriculture and enhance the efficiency of large-scale operations, ultimately aiding in reducing carbon emissions in agriculture [43]. Human capital is conducive to reducing environmental degradation [44]. The replanting index plays a role in agricultural carbon emissions by influencing the scale of agricultural cultivation. Improvements in land quality can reduce agricultural carbon emissions by improving the storage of organic carbon in the soil and increasing the efficiency of nitrogen fertilizer use [45,46]. The specific measurements are shown in

Table A1. Measurement of control variables.

Control Variables	Measurements
Replanting index	The ratio of the sown area of crops to the arable land area.
Financial support for agriculture	Proportion of local finance expenditure on agriculture, forestry and water in the general budget expenditure of local finance.
Degree of disaster	The ratio between the affected area and the total sown area of crops.
Land quality	The share of effectively irrigated area in the total sown area of crops.
Education level of rural residents	Years of schooling per capita in rural areas.
Consumption level of rural residents	The ratio of per capita consumption expenditure by rural residents to the per capita consumption expenditure by urban residents.

in Appendix A.

2.3. Visual Analytics Methodology

For feature fact analysis, we used Origin 2021, MATLAB R2022b, and ArcGIS 10.8 for visualization and analysis. First, we used Origin 2021 software to depict the trend in the level of trade in agricultural products from 2001 to 2023 to visualize the current status of trade in agricultural products. Secondly, we used MATLAB R2022b software to estimate kernel density and analyze the spatial distribution and trends of agricultural carbon emission levels in specific years. Finally, we used ArcGIS 10.8 software to characterize the trends and spatial distribution of agricultural trade and agricultural carbon emissions in different regions for specific years. Specifically, we categorized the levels of agricultural trade and agricultural carbon emissions into four groups: low level, lower level, higher level, and high level according to the 25th percentile, 50th percentile, and 75th percentile, respectively. On this basis, we used ArcGIS 10.8 to visualize and analyzed the spatial evolution trends of agricultural trade and agricultural carbon emissions in China in 2010 and 2021.

2.4. Theil Index Method and Location Quotient Method

With reference to the work of Zhang et al. (2022) [47], we used the Theil index to measure the level of rationalization of the agricultural industrial structure so as to explore whether the trade in agricultural products can promote the upgrading of the agricultural industrial structure. The specific formula is shown in Equation (3):

$$TL={\displaystyle \sum _{i=1}^n\frac{Y_i}{Y}}ln\left({\displaystyle \frac{Y_i}{L_i}}/{\displaystyle \frac{Y}{L}}\right)$$

(3)

where ${\displaystyle \frac{Y_i}{L_i}}$ reflects the production efficiency of the industry. Considering the data availability, we chose the value added by the unit intermediate consumption of agriculture, forestry, animal husbandry, and fishery (i.e., value-added/intermediate consumption) to measure the production efficiency of each industry, and we expressed Y_i/Y in terms of the proportion of the industry’s output value to the total output value of agriculture, forestry, animal husbandry, and fishery. If the agricultural economy is in the state of equilibrium development, the Theil index is 0. However, if the Theil index deviates from 0, the agricultural industrial structure strays from the equilibrium state, and the industrial structure is not reasonable.

Furthermore, we measured the locational entropy of 30 provinces to characterize their levels of agricultural industrial agglomeration from 2001 to 2021 using the panel data of agricultural output value and GDP with the following formula:

$$AI{A}_{it}={\displaystyle \frac{A{G}_{it}}{G_{it}}}/{\displaystyle \frac{A{G}_t}{G_t}}$$

(4)

where $AI{A}_{it}$ represents the agricultural industrial agglomeration level of province i in year t, $A{G}_{it}$ represents the primary output value of province i in year t, $G_{it}$ represents the GDP of province i in year t, $A{G}_t$ represents the primary output value of the whole country in year t, and $G_t$ represents the national GDP in year t.

2.5. Robustness Test Methods

To verify the reliability of the benchmark regression results, we conducted the following: replacing the dependent variable with agricultural carbon intensity, endogeneity testing, removing outlier interference, replacement of independent variable, changing the time window, and adding control variables in robustness tests. First, considering that a single carbon emission indicator fails to reveal comprehensive regional carbon emission characteristics, this study adopts agricultural carbon emission intensity as an alternative characterization of the explanatory variables to verify the robustness of the baseline regression conclusions [7]. Second, in order to solve the possible endogeneity problem, referring to the studies of Huang and Li (2006) [48] and Chen (2022) [39], we used the average distance between provincial capital cities and ports, which reflects the trade facilitation level of the sample area, and the degree of topographic relief, which reflects the topographic characteristics of the sample area, to construct interaction terms with lagged one-period trade in agricultural products as instrumental variables. The two-stage least squares method was applied to conduct the endogeneity test. On the one hand, the geographic location of each city remains constant, and its distance to the nearest port is an objective fact that is not disturbed by any factors. Therefore, the distance from each city to the coastal port is a strictly exogenous variable, and satisfying the instrumental variables is the essential condition of homogeneity. On the other hand, there is a correlation between the distance from cities to the coastal port and the volume of traded products. At the same time, the degree of topographic relief directly affects the trade of agricultural products, thereby fulfilling the relevance condition while not directly affecting the agricultural carbon emissions, so it also fulfills the exogeneity condition. Considering that these indicators are cross-sectional data rather than time-series variables, we constructed separate interaction terms between these two variables and trade in agricultural products in the lagged period as instrumental variables for testing. Third, given that the outliers within the dataset could potentially influence the outcomes of the regression analysis, we further performed the upper and lower 1% shrinkage treatment on all variables to reduce the influence of extreme values and outliers on model estimation. Fourth, we examined agricultural import trade and agricultural export trade separately. Fifth, we changed the time window to eliminate some of the macro-environmental effects for the test. Sixth, in order to verify the robustness of the benchmark results, production efficiency, urbanization, level of infrastructure, and level of environmental regulation were added into the regression model gradually as control variables.

2.6. Data Sources

This study mainly takes the panel data of 30 provinces in China (excluding Tibet, Hong Kong, Macao, and Taiwan) from 2001 to 2021 as the research sample. In terms of the original data sources for the relevant variables, the original data on the measurement of agricultural carbon emissions, the value added by the primary industry, the level of economic development, the rural population numbers, the total power of agricultural machinery, the urban population and regional resident population, the sown area of grains, the sown area of crops, and the arable land area were sourced from the database of the National Bureau of Statistics of China (NBS); affected area was sourced from China Rural Statistical Yearbook; the information regarding the trade in agricultural products was sourced from the China Agricultural Yearbook and the China Monthly Statistical Report on Import and Export of Agricultural Products; the data on employees in the primary industry were sourced from the statistical yearbooks published by provincial statistical bureaus. The moving-average method was utilized to fill in individual missing data. The descriptive statistics of the main variables are shown in

Table 1. Descriptive statistics of the main variables.

Variable	Sample Size	Mean	Standard Deviation	25th Percentile	75th Percentile
Agricultural carbon emissions	630	7.386	0.981	6.973	8.026
Agricultural trade	630	0.685	2.268	0.0320	0.289
Replanting index	630	1.271	0.366	0.950	1.536
Financial support for agriculture	630	0.099	0.035	0.070	0.124
Degree of disaster	630	0.215	0.154	0.0950	0.298
Land quality	630	0.410	0.161	0.301	0.490
Education level of rural residents	630	7.599	0.688	7.214	8.034
Consumption level of rural residents	630	0.421	0.0860	0.357	0.484

.

3. Results

3.1. Characterization Facts Analysis

3.1.1. Analysis of the Spatial and Temporal Evolution of the Level of Agricultural Trade

Upon finishing our assessment of agricultural trade levels, we plotted the trends in agricultural imports, exports, and import/export trade levels from 2001 to 2023, as shown in Figure A2 in Appendix A. During the sample period, it can be observed in the figure that China’s agricultural trade exhibited a predominantly upward trajectory. With growth from USD 27.9 billion in 2001 to USD 304.17 billion in 2021, China has become a major player in agricultural trade, ranking second worldwide. At the same time, the deficit trend of China’s agricultural trade gradually started to appear around 2004. Via the red and gray lines, we can see that the growth of agricultural trade imports is continuously increasing, whereas the growth of agricultural trade exports is very small. With agricultural trade imports accounting for the continued increase in the proportion of total imports and exports, China’s agricultural trade deficit pattern is gradually expanding. This shows that China’s agricultural exports are facing greater pressure. It will be necessary to accelerate the transformation and development of agriculture and enhance the diversification of export products to adapt to the international agricultural trade market.

3.1.2. Analysis of the Spatial and Temporal Evolution of Agricultural Carbon Emission Levels

We conducted kernel density estimation and analyze the spatial distribution and trends of agricultural carbon emission levels in specific years. The findings are illustrated in Figure A3 in Appendix A. It can be seen that the overall distribution curve of the kernel density of China’s agricultural carbon emissions from 2001 to 2021 first moves to the right, then moves to the left, and finally comes to a standstill, which intuitively reflects the primary trend of China’s agricultural carbon emissions first increasing and then remaining stable. In the change in the peak value of the kernel density distribution, the “multi-peak” pattern was most evident in 2006, while other years not obvious “multi-peak” pattern, indicating that the regional polarization of China’s agricultural carbon emission levels is not great.

Figure 2 and Figure 3 show the trends and spatial distribution characteristics of agricultural trade and agricultural carbon emissions in China in 2010 and 2021, respectively. Among them, according to quartiles, agricultural trade and carbon emissions are categorized into four different levels. As can be seen in the figure, the regions with higher agricultural carbon emissions are mainly concentrated in grain-producing provinces such as Heilongjiang, Henan, Shandong, and Hunan. This trend may be attributed to the factor endowment of these regions, where favorable climatic conditions and abundant land resources provide favorable conditions for agricultural activities. In contrast, coastal provinces such as Hainan, Fujian, Shanghai, and Zhejiang have relatively low agricultural carbon emissions. A comparison of the two distribution maps reveals that there is a spatially staggered relationship between the level of trade in agricultural products and agricultural carbon emissions and that there may be some interaction between the two, which will be verified in the subsequent sections of this paper.

3.2. Analysis of Benchmark Regression Results

Table 2. Baseline regression results.

Variables	Agricultural Carbon Emissions
	(1)	(2)	(3)	(4)
Agricultural trade	−0.0366 ***	−0.0224 ***	−0.0230 ***	−0.0238 ***
	(0.0049)	(0.0040)	(0.0044)	(0.0043)
Replanting index		0.3100 ***	0.3023 ***	0.2792 ***
		(0.0415)	(0.0534)	(0.0522)
Financial support for agriculture		2.0864 ***	0.5834 **	2.1815 ***
		(0.3293)	(0.2299)	(0.3313)
Degree of disaster			−0.0514	−0.1245 ***
			(0.0395)	(0.0463)
Land quality			−0.2492 *	−0.0929
			(0.1315)	(0.1450)
Education level of rural residents				−0.0065
				(0.0324)
Consumption level of rural residents				−0.3459 **
				(0.1681)
Provincial FE	Y	Y	Y	Y
Year FE	Y	Y	Y	Y
Constant term	7.4114 ***	6.8012 ***	7.0733 ***	7.0917 ***
	(0.0058)	(0.0594)	(0.1148)	(0.2614)
Observations	630	630	630	630
R-squared	0.9880	0.9901	0.9888	0.9904

Note: ***, **, and * denote 1%, 5%, and 10% significance levels, respectively. Y = yes, N = no. Standard errors are in parentheses. In addition to columns (2)–(3), which are the results of gradually increasing the control variables, we verify that the explanatory variables remain robust to the other results of gradually increasing the control variables.

demonstrates the test results of the fixed-effects model. The table reveals that when considering the effects of year and province simultaneously, the impact coefficient of agricultural trade on China’s agricultural carbon emissions consistently shows a significant negative correlation at the 1% level. This suggests that an increase in agricultural import and export trade can effectively decrease agricultural carbon emissions, thereby confirming Hypothesis 1. The degree of the disaster has a significant inhibitory effect on the trade of agricultural products. Increasing the replanting index promotes agricultural carbon emissions. Improvements in the education level of rural residents and the quality of land reduce carbon emissions from agriculture. The rising consumption level of rural residents has a dampening effect on agricultural carbon emissions. This may be due to the fact that when rural residents have higher spending power, they may pay more attention to the quality of their living environment, and reduce the use of chemical fertilizers and pesticides in agricultural production. The impacts of the control variables on agricultural carbon emissions mostly conform to the theoretical expectations. But the rise in financial support for agriculture has a promoting impact on agricultural carbon emissions, particularly at the 1% significance level. This is likely because the increased financial support affects the planting and breeding practices within the agricultural sector.

3.3. Robustness Test

3.3.1. Replacing the Dependent Variable with Agricultural Carbon Intensity

The carbon intensity can reflect the level of carbon emissions from agricultural production activities. The results of replacing the dependent variable with agricultural carbon intensity are shown in columns (1)–(4) of

Table 3. Results of the replacement of the dependent variable.

Variables	Intensity of Agricultural Carbon Emissions	Intensity of Agricultural Carbon Emissions	Intensity of Agricultural Carbon Emissions	Intensity of Agricultural Carbon Emissions	Per capita Agricultural Carbon Emissions	Agricultural Carbon Emissions (Including Transportation Carbon Emissions)
	(1)	(2)	(3)	(4)	(5)	(6)
Agricultural trade	−0.0132 ***	−0.0047 **	−0.0053 **	−0.0064 **	−0.0365 ***	−0.0714 ***
	(0.0035)	(0.0024)	(0.0023)	(0.0025)	(0.0079)	(0.0101)
Control variables	N	Y	Y	Y	Y	Y
Year FE	Y	Y	Y	Y	Y	Y
Provincial FE	Y	Y	Y	Y	Y	Y
Constant term	0.4699 ***	0.1947 ***	0.2212 ***	0.2124 ***	0.8932 ***	8.2244 ***
	(0.0028)	(0.0254)	(0.0332)	(0.0789)	(0.2995)	(0.5098)
Observations	630	630	630	630	630	630
R-squared	0.9622	0.9750	0.9710	0.9762	0.9411	0.9565

Note: *** and ** denote 1% and 5% significance levels. Other details are provided in Table 2.

. Columns (2)–(3) are the result of adding the control variables stepwise. In column (2), the control variables to be added are the replanting index and financial support to agriculture, and in column (3), the control variables are the replanting index, financial support to agriculture, the degree of damage, and land quality. We also validate additional results for the stepwise addition of control variables, and the explanatory variable remains robust.

In addition, per capita agricultural carbon emissions are also an important indicator of agricultural carbon emissions. We also tested this, and the results, as shown in column (5) of Table 3, show that the result is robust. Transportation in agricultural trade also generates carbon emissions. In order to further verify the robustness of the carbon reduction effect of agricultural trade, we try to include transportation-related carbon emissions into the analytical framework, and the results are shown in column (6) of Table 3, with the coefficients still significantly negative.

3.3.2. Endogeneity Test

The rationale for the choice of instrumental variables has been discussed previously. The outcomes of this test are presented in

Table 4. Endogeneity test results.

Variables	Agricultural Trade	Agricultural Carbon Emissions	Agricultural Trade	Agricultural Carbon Emissions
	(1)	(2)	(3)	(4)
Cross-multiplier with topographic relief	1.1388 ***
	(0.1598)
Cross-multiplier with distance to port city			0.0011 ***
			(0.0001)
Agricultural trade		−0.1746 ***		−0.0278 ***
		(0.0487)		(0.0048)
Kleibergen-Paap rk Wald F	50.757		79.678
Kleibergen-Paap rk LM	10.56 [0.0012]		22.425 [0.0000]
Cragg-Donald Wald F	161.441		1544.254
Control variables	Y	Y	Y	Y
Year FE	Y	Y	Y	Y
Provincial FE	Y	Y	Y	Y
Constant term	0.3665	6.1820 ***	−2.2899	5.1349 ***
	(0.6981)	(0.4286)	(2.6244)	(0.2670)
Observations	600	600	600	600
R-squared	0.5322	0.5347	0.9469	0.9915

Note: *** denotes 1% significance. Other details are provided in Table 2.

. Columns (1) and (2) show the results of the first- and second-stage regressions using the cross-multiplier terms of the topographic relief and lagged first-order agricultural trade constructs as instrumental variables, respectively. Columns (3) and (4) present the results of the first- and second-stage regressions using the cross-multiplier terms of the average distance to port cities and the lagged first-order agricultural trade construct as instrumental variables, respectively.

The test for all instrumental variables suggested that both instrumental variables are justified. In the test results in the table, it can be seen that the levels of agricultural import and export trade with agricultural carbon emissions after refitting using different instrumental variables are all negative. These results are consistent with previous regression results, which further confirms the robustness of the benchmark regression results.

3.3.3. Removing Outliers

The results are shown in

Table 5. Robustness test results for outlier removal interference.

Variables	Agricultural Carbon Emissions
	(1)	(2)	(3)	(4)
Agricultural trade	−0.3186 ***	−0.0497 ***	−0.2464 ***	−0.0358 ***
	(0.0147)	(0.0053)	(0.0170)	(0.0050)
Control variables	N	N	Y	Y
Year FE	N	Y	N	Y
Provincial FE	N	Y	N	Y
Constant term	7.5939 ***	7.4187 ***	6.7042 ***	7.2153 ***
	(0.0311)	(0.0059)	(0.4054)	(0.2610)
Observations	630	630	630	630
R-squared	0.4281	0.9882	0.5400	0.9905

Note: *** denotes 1% significance. Other details are provided in Table 2.

. After controlling for temporal and regional fixed effects, the suppression effect of agricultural import and export trade on agricultural carbon emissions was significant at the 1% level regardless of controlling for other variables, which proves that the estimation results of the baseline model are robust.

3.3.4. Replacement of Independent Variable

Regarding the independent variable, we used the method of Xu et al. (2023) [49] and adopted the sum of import and export trade in agricultural products to represent the level of agricultural trade. The results are shown in column (1) of

Table 6. Robustness test results for the replacement of the independent variable.

Variables	Agricultural Carbon Emissions
	(1)	(2)	(3)
Total agricultural trade	−0.0002 ***
	(0.0000)
Agricultural import trade		−0.0245 ***
		(0.0044)
Agricultural export trade			−0.1835 **
			(0.0917)
Control variables	Y	Y	Y
Year FE	Y	Y	Y
Provincial FE	Y	Y	Y
Constant term	7.2818 ***	7.0862 ***	7.0084 ***
	(0.2823)	(0.2614)	(0.2745)
Observations	630	630	630
R-squared	0.9907	0.9904	0.9899

Note: *** and ** denote 1% and 5% significance levels. Other details are provided in Table 2.

. In these results, it can be seen that the coefficient on the level of agricultural trade is still significantly negative at the 1% level, again indicating that the findings of the benchmark regression are robust. In addition, considering the different channels of action of imports and exports, we again validated these findings with agricultural import trade and export trade as explanatory variables. The results are shown in columns (2) and (3) of Table 6. In the results, we can see that the effects of both agricultural imports and exports on agricultural carbon emissions are significantly negative.

3.3.5. Excluding Other Significant Events from Interfering

Global economic development was significantly affected by the 2008 financial crisis [22]; subsequently, the expansion of the coronavirus outbreak and severe weather conditions, in conjunction with a rapidly growing global economy, have worsened instability and unpredictability in agricultural output, heightening the likelihood of disturbances to the global agricultural supply chain, market instabilities, and the possibility of food crises emerging [50]. Therefore, we excluded the data from 2008 and 2019–2021, respectively, before repeating the test, the results of which are shown in

Table 7. Robustness test results after excluding other significant events.

Variables	Excluding 2008		Excluding 2019 Onwards
	(1)	(2)	(3)	(4)
Agricultural trade	−0.0365 ***	−0.0235 ***	−0.0524 ***	−0.0400 ***
	(0.0050)	(0.0045)	(0.0112)	(0.0096)
Control variables	N	Y	N	Y
Year FE	Y	Y	Y	Y
Provincial FE	Y	Y	Y	Y
Constant term	7.4129 ***	7.0639 ***	7.4228 ***	7.3001 ***
	(0.0060)	(0.2662)	(0.0076)	(0.2769)
Observations	600	600	540	540
R-squared	0.9876	0.9901	0.9894	0.9911

Note: *** denotes 1% significance levels. Other details are provided in Table 2.

. After changing the time window, the regression results are still significantly negative, and the robustness of the benchmark regression is further enhanced.

3.3.6. Adding Control Variables

Increased production efficiency may enhance resource utilization efficiency, thereby reducing agricultural carbon emissions. The conversion of agricultural land to urban land in the process of increasing urbanization level may lead to the destruction of soil and ecosystems, thus releasing more carbon. Improved infrastructure facilitates the flow of factor resources and promotes the greening of agriculture [51]. Environmental regulations may increase the cost of carbon emissions, thereby contributing to standardized production behavior [52]. The model may suffer from omitted variables. The results after adding control variables are shown in

Table 8. Robustness test results after adding control variables.

Variables	Agricultural Carbon Emissions
	(1)	(2)	(3)	(4)
Agricultural trade	−0.0395 ***	−0.0353 ***	−0.0294 ***	−0.0281 ***
	(0.0058)	(0.0062)	(0.0055)	(0.0054)
Control variables	Y	Y	Y	Y
Year FE	Y	Y	Y	Y
Provincial FE	Y	Y	Y	Y
Production efficiency	−0.7903 ***	−0.7721 ***	−0.6821 ***	−0.5998 ***
	(0.1126)	(0.1137)	(0.1038)	(0.1097)
Urbanization		0.3447 *	0.5420 ***	0.4219 **
		(0.2015)	(0.1907)	(0.1925)
Level of infrastructure			−0.2145 ***	−0.2052 ***
			(0.0289)	(0.0285)
Level of environmental regulation				−2.0584 ***
				(0.6415)
Constant term	7.1978 ***	7.1142 ***	7.2044 ***	7.3672 ***
	(0.2514)	(0.2639)	(0.2416)	(0.2461)
Observations	630	630	630	630
R-squared	0.9914	0.9915	0.9924	0.9925

Note: ***, **, and * denote 1%, 5%, and 10% significance levels. “Control variables” represents the control variables used in the article’s baseline regression. Other details are provided in Table 2.

below. After the gradual addition of control variables, agricultural trade still significantly reduces agricultural carbon emissions. The robustness of the benchmark regression results is again verified.

3.4. Mechanism Test

The previous benchmark regression results show that the trade of agricultural products has a significant inhibitory effect on agricultural carbon emissions. In this case, through what channels does agricultural trade affect agricultural carbon emissions? Synthesizing Hypotheses 2 and 3, we carried out mechanistic tests of the technology effects and structural effects of agricultural trade.

3.4.1. Technical Progress in Agriculture: Technical Effect

Import competition in agricultural trade drives agricultural producers to adopt advanced foreign technology and update their agricultural technology. As previously analyzed, agricultural trade may affect agricultural production by promoting the learning effect, reducing the cost of reducing carbon emissions, influencing the allocation of agricultural factors, and changing the structure of energy factors, thereby facilitating reductions in agricultural carbon emissions. The Malmquist index can be decomposed into technical progress and technical efficiency [53,54]. The technological change index reflects the technological progress in production. In this subsection, we carried out a mechanistic test from the perspective of the effect of technology. To verify Hypothesis 3, we used the agricultural technological progress index measured by the DEA model to measure the technical progress in agriculture. Output variables are expressed in terms of the value of agricultural, forestry, livestock, and fishery output; labor inputs are expressed in terms of the number of people employed in the primary sector; agricultural machinery inputs are expressed in terms of the amount of electricity used in the countryside; irrigation inputs are expressed in terms of the effective irrigated area; and fertilizer inputs are expressed in terms of the net amount of fertilizer applied to the farm. The results of the test are shown in columns (1)–(2) of

Table 9. Test results of mechanisms.

Variables	Technical Progress in Agriculture		Agricultural Industrial Structure
	(1)	(2)	(3)	(4)
Agricultural trade	0.0043 *	0.0060 **	−0.0398 ***	−0.0686 ***
	(0.0026)	(0.0029)	(0.0064)	(0.0086)
Control variables	N	Y	N	Y
Year FE	Y	Y	Y	Y
Provincial FE	Y	Y	Y	Y
Constant term	1.0250 ***	1.0761 ***	0.9984 ***	3.7637 ***
	(0.0033)	(0.1466)	(0.0135)	(0.7271)
Observations	600	600	624	624
R-squared	0.2973	0.3027	0.7692	0.7920

Note: ***, **, and * denote 1%, 5%, and 10% significance levels. Other details are provided in Table 2.

below. We found that the level of agricultural import and export trade significantly promotes the increase in agricultural machinery density, which indicates that agricultural trade can effectively improve the technical conditions of agricultural production and strengthen the level of agricultural science and technology. The enhancement of the level of agricultural science and technology promotes the specialization and standardization of agricultural production [55,56], making the agricultural production process greener and low-carbon so that the agricultural trade achieves the effect of agricultural carbon reduction.

3.4.2. Rationalization of Agricultural Industrial Structure: Structural Effect

China’s industrial upgrading mainly stems from resource reallocation among industries in a particular region. The import competition caused by the trade in agricultural products forces the regional internal resources to be allocated to higher value-added industries to transfer resources to high-end industries, thereby allowing for the “survival of the fittest”. As a measure of “unevenness”, the Theil index can better measure the rationality of industrial structure [57].

The test results are shown in Table 9. The improvement in the level of agricultural trade can significantly promote the rationalization of the agricultural industrial structure, thus exerting a structural effect. Accompanied by the rationalization of the agricultural and industrial structure, the regional economic development level and increased investment in environmental management strengthen the high-quality development of agriculture; thus, agriculture will tend to be modernized, organic, and green [47].

3.5. Heterogeneity Analysis

3.5.1. Heterogeneity of Agricultural Industrial Agglomeration Level

According to the agglomeration effect, it is known that the process of industrial agglomeration generates positive environmental externalities, promotes technological innovation and spillover effects, and increases the promotion and application of more environmentally friendly technologies, thereby reducing pollution. Agricultural industrial agglomeration plays a crucial role in reducing agricultural carbon emissions. Naturally, the effects of import and export trade in agricultural products on agricultural carbon emissions at different levels of agricultural industrial agglomeration should be different. According to the measured agricultural industrial agglomeration levels, the samples were divided into two groups—high and low agricultural industrial agglomeration (agglomeration levels of >1 and <1, respectively)—and regressed, and the results are shown in

Table 10. Results of the heterogeneity test.

Variables	Level of Agro-Industrial Agglomeration		Level of Topographic Relief		Geographical Position
	(1) Low	(2) High	(3) Low	(4) High	(5) South	(6) North
Agricultural trade	−0.0209 ***	−1.1083 ***	−0.0174 ***	0.2342	−0.0218 ***	−0.0229 **
	(0.0039)	(0.2177)	(0.0044)	(0.3707)	(0.0041)	(0.0095)
Control variables	Y	Y	Y	Y	Y	Y
Year FE	Y	Y	Y	Y	Y	Y
Provincial FE	Y	Y	Y	Y	Y	Y
Constant term	6.9888 ***	7.3361 ***	7.4959 ***	7.8413 ***	7.2858 ***	6.2295 ***
	(0.3786)	(0.3110)	(0.3491)	(0.4078)	(0.2674)	(0.4262)
Observations	312	314	399	231	315	315
R-squared	0.9940	0.9783	0.9934	0.9813	0.9946	0.9892

Note: *** and ** denote 1% and 5% significance levels. Other details are provided in Table 2.

. Compared with the samples with lower levels of agricultural industrial agglomeration, the absolute effect of agricultural trade in reducing agricultural carbon emissions is greater in the case of higher levels of agglomeration, and the inhibitory effect on agricultural carbon emissions is stronger. This may be due to the fact that the agricultural industrial agglomeration effect can inhibit agricultural carbon emissions through the scale effect, technology spillover, and imperfect competition effect. Along with industrial agglomeration, agricultural production tends to be gradually standardized and scaled, which makes it easier to optimize the allocation of various types of resources. This is conducive to a better utilization of the technological effects induced by agricultural trade, which in turn promotes the improvement in the efficiency of energy consumption and the efficiency of the use of agricultural production materials and realizes the goal of reducing agricultural carbon emissions.

3.5.2. Heterogeneity in the Level of Topographic Relief

There may be differences in agricultural production methods under different terrain conditions. In order to further analyze the heterogeneous impact of such differences on the carbon-reduction effect of agricultural trade, referring to the study by You et al. (2018) [58], we divided the total sample into two types of landscapes with less topographic relief (≤1) and more topographic relief (>1), according to the degree of topographic relief, in order to identify the different impacts of agricultural trade on agricultural carbon emissions under different terrain conditions. The results are shown in Table 10. The regression coefficients of agricultural trade in the group with lower topographic relief are significantly negative. In contrast, the coefficients for the group with high topographic relief are not significant. This may be because trade in agricultural products can promote the specialization and scale of agricultural production and reduce agricultural carbon emissions by centralizing production and transportation. Areas with lower topographic relief tend to be more suitable for agricultural specialization and large-scale operations. In contrast, areas with higher level of topographic relief may be constrained by land-use limitations and transportation difficulties, making efficient large-scale agricultural production and trade complex to achieve.

3.5.3. Regional Heterogeneity

China is a vast country with large differences in climate between the north and the south, resulting in very different agricultural production methods and crop cultivation patterns. Accordingly, based on geographic location, the sample was divided into two groups (south and north), and group regression was conducted to explore the heterogeneity of the impact of trade in agricultural products on agricultural carbon emissions in different geographic locations; the test results are shown in Table 10. As can be seen in the table, compared with the north, the suppressive effect of agricultural trade on agricultural carbon emissions is more significant in the south. This may be due to the warm and humid climate in the south, which is more suitable for the growth of crops, whereas the north has a short growing season and limited production due to climatic conditions. Thus, the scale effect of agricultural trade is more easily realized in the south. At the same time, in the north, under the constraints of natural geographical conditions, the utilization and upgrading of new technologies face more difficulties in the process of exerting the technological effects brought about by agricultural trade. In the south, due to the sustained development of the agriculture and animal husbandry industry over a long time, the traditional production mode may face a development bottleneck, and the technological effect brought by agricultural trade is more conducive to breaking through this agricultural bottleneck, improving the production efficiency so as to reduce carbon emissions. Therefore, compared with the north, the effect of agricultural trade on reducing agricultural carbon emissions is more significant in the south.

3.6. Extensive Analysis: The Synergistic Effect of Agricultural Trade in Reducing Pollution and Carbon Emissions

The previous section confirms that agricultural trade has an inhibitory effect on agricultural carbon emissions. In addition to the carbon reduction effect, the reduction of environmental pollution is also an essential aspect of the environmental effect. Agricultural surface pollution is increasing and has become a global problem [59]. The U.S. Environmental Protection Agency (EPA) has determined that agricultural surface pollution is the leading cause of surface water pollution, including rivers and lakes, being responsible for about 66% of the total pollution of this kind [60]. Agricultural land is the primary source of global nitrogen pollution; more than half of the nitrogen input from agricultural land is lost to air and water, resulting in severe environmental pollution, while fertilizers, pesticides, agricultural films, and diesel fuel used in agricultural production constitute the primary sources of agricultural surface pollution [61,62].

Many air pollutants and carbon dioxide often have the same root causes, sources, and processes, and previous studies have verified the existence of synergistic effects by examining the distributional characteristics of greenhouse gases and air pollutant emissions, the synergistic development of which has received increasing attention of late [63,64,65]. However, little research has yet been conducted on the synergistic effects of pollution and carbon reduction in agricultural production, and even fewer studies have been conducted on the synergistic effects of agricultural trade on pollution and carbon reduction. The previous chapters have demonstrated the impact of agricultural trade on agricultural carbon emissions, along with its mechanism of action. This study argues that the competition effect and income effect of agricultural trade are the main reasons for its inhibition of agricultural surface pollution. On the one hand, the green barriers in international trade and the competition effect will eliminate products with high pollution and high emissions in the process of their production. In order to adapt to the market demand, people will pay more attention to the environmental effects of these products, including pollution [66]. On the other hand, the “knowledge spillover” and “learning by doing” induced by agricultural trade are conducive to promoting the improvement in agricultural production efficiency, bringing income growth, and giving producers more capital to update their production equipment, thereby achieving de-pollution at the source [67]. Synthesizing the analysis of the synergistic mechanism of agricultural-trade-driven pollution reduction and carbon reduction in agriculture, we constructed a mechanism map, as shown in Figure A4.

Is there a synergistic effect of pollution reduction and carbon reduction produced by agricultural trade, based on the theoretical analysis of its effects on these two aspects? We will use the econometric method to verify this empirically. Drawing on relevant studies, we took fertilizers, pesticides, agricultural films, and diesel as the main sources of agricultural surface pollution [68], using the United Nations Human Development Index and Economic Vulnerability Index for the weighting process. Using an equal-weighting method to assign values to the fertilizer, pesticide, agricultural film, and diesel use, their weights were set to 0.25 and then divided by the sown area of crops to measure the level of pollution of agricultural surface sources. The synergistic effect of agricultural pollution and carbon reduction was measured on this basis. We mainly used two methods to measure the synergistic pollution reduction and carbon reduction: the first draws on the study of Lu et al. (2022) [69] to construct the interaction term between the level of agricultural carbon emissions and the degree of agricultural surface source pollution; the second draws on the work of Chen et al. (2023) [70] to construct a coupled coordination model to measure the degree of coordination between the level of agricultural carbon emissions and the degree of agricultural surface source pollution. The empirical test was then conducted using a double fixed-effects model.

The results are shown in

Table A2. Regression results of agricultural trade on the synergistic reduction in agricultural pollution and carbon emissions.

Variables	Level of Agricultural Surface Pollution		Cross-Multipliers of Agricultural Surface Pollution and Carbon Emissions	Coordination of Agricultural Pollution and Carbon Emissions
	(1)	(2)	(3)	(4)
Agricultural trade	−0.0240 ***	−0.0190 **	−0.2530 ***	0.0086 ***
	(0.0064)	(0.0074)	(0.0627)	(0.0016)
Control variables	N	Y	Y	Y
Year FE	Y	Y	Y	Y
Provincial FE	Y	Y	Y	Y
Constant term	2.9029 ***	4.7752 ***	41.5977 ***	0.6126 ***
	(0.2040)	(0.2388)	(2.3508)	(0.0721)
Observations	630	630	630	630
R-squared	0.4351	0.9539	0.9468	0.9057

Note: *** and ** denote 1% and 5% significance levels. Other details are provided in Table 2.

. Columns (1)–(2) in Table A2 show the impact of agricultural trade on agricultural surface source pollution, while columns (3)–(4) show the impact of agricultural trade on the degree of synergy of agricultural pollution reduction and carbon reduction measured in the two ways mentioned above. As can be seen in the table, after controlling for the effects of land quality, financial support for agriculture, replanting index, consumption level of rural residents, education level of rural residents, and degree of disaster, trade in agricultural products significantly inhibits agricultural surface pollution. And it contributes to the synergistic effect of agricultural pollution reduction and carbon reduction, which is significant at the 5% level.

4. Discussion

Based on theoretical analysis and with the help of econometric methods, we studied the trade in agricultural products and the agricultural carbon emissions in 30 provinces in China from 2001 to 2021. We argue that the trade in agricultural products is conducive to mitigating the problem of agricultural carbon emissions. At the same time, it drives agriculture to achieve synergistic reductions in pollution and carbon emissions.

Compared with the existing literature, despite the differences in the scope and region of this paper, similar results are obtained, namely, that trade in agricultural products has a significant carbon emission reduction effect. Secondly, compared with previous studies that solely considered the impact of agricultural trade on agricultural carbon emissions or agricultural pollution [71,72], this study considers the synergistic benefits based on these two impacts; it also provides some theoretical discussion on the possible impact mechanisms. The synergistic effect of reducing pollution and carbon is to contribute to the realization of economic, social, and environmental synergies [63,64]. Thirdly, our heterogeneity analysis is not only based on a single geographic location feature, but also analyzes different topographic and geomorphic features, as well as agricultural industrial agglomeration levels. Our results provide a theoretical reference for guiding the transition of agriculture to green production under the complex situation of different farming methods and patterns of agricultural production.

From a global perspective, the environmental effect of agricultural trade has been widely recognized. The upgrading of agricultural technology and the upgrading of the agricultural and industrial structure are the key driving factors for the achievement of agricultural and environmental benefits, and the optimized development of agricultural trade may play an essential role in the future of pollution reduction and carbon reduction in the agricultural sector. It is essential to clarify the relationship between agricultural trade and agricultural carbon emissions, analyze the intrinsic influence mechanisms, and determine the optimal direction to cope with global climate change. Based on this, we propose the following suggestions.

First, adapting the agricultural development model to local conditions is an important means of promoting sustainable agricultural development. Due to the heterogeneous impact of agricultural trade on agricultural carbon emissions in different regions, the government should adjust and improve the policies and structures of agricultural trade in each region according to local conditions and continue to play a positive regulatory role in the impact of agricultural trade on agricultural carbon emissions. Second, progress in agricultural technology and upgrading the structure of the agricultural industry are important driving forces for high-quality agricultural development. Agricultural producers should organically combine green technologies, scientifically adjust the planting structure, and realize the complementary advantages of multiple emission reduction paths. Third, it is essential to play a full role in the synergistic effects of pollution reduction and carbon reduction in the agricultural sector. The transformation of land intensification and large-scale operation is conducive to transforming the traditional mode of agricultural production to a modernized mode of development, thereby realizing the synergistic effects of pollution reduction and carbon reduction in agricultural trade.

5. Conclusions

Using panel data from 30 provinces in China (excluding Tibet, Hong Kong, Macao, and Taiwan) from 2001 to 2021, this study explored the impact of import and export trade of agricultural products on agricultural carbon emissions using a double fixed-effects model. Firstly, we measured agricultural carbon emissions from four perspectives: agricultural factor inputs, agricultural cultivation, biomass combustion, and animal breeding; secondly, we utilized a fixed-effects model, and through a series of robustness tests, we examined the direct effect of agricultural trade in curbing agricultural carbon emissions, providing a mechanistic model of the technological and structural effects of agricultural trade. Subsequently, the heterogeneity test was conducted from three perspectives—the level of geographic location, level of agricultural industrial agglomeration, and topographic features. Finally, the synergistic effect of agricultural trade on pollution reduction and carbon reduction was analyzed expansively, its internal mechanisms were theoretically explored, and empirical testing was carried out. We found that, firstly, trade in agricultural products significantly inhibits agricultural carbon emissions, and the robustness test confirmed this conclusion. Second, advancements in agricultural technology and upgrading the agricultural industry structure are critical paths for agricultural trade to promote reductions in agricultural carbon emissions. Furthermore, agricultural trade shows a more pronounced suppressive effect on agricultural carbon emissions in the south, as well as in areas with lesser topographical relief and higher levels of agricultural industrial agglomeration. Finally, agricultural trade fosters the collaborative advancement of mitigating agricultural pollution and reducing carbon emissions. As a result, this study confirms that agricultural trade has a role in reducing carbon emissions and that giving full play to the environmental effects of trade in agricultural products contributes to the high-quality development of agriculture.

Although this study adopted scientific research methods and meticulous research programs, some limitations should be acknowledged. First, this study describes and examines the impact mechanism of agricultural trade on agricultural carbon emissions from the perspective of advancements in agricultural technology and upgrading of the agricultural and industrial structure, deepening the analysis of the intrinsic mechanisms of agricultural trade in mitigating agricultural carbon emissions. However, the multiple effects of agricultural trade make the mechanisms of its impact on agricultural carbon emissions more complicated, which should be more comprehensively explored in the future. In the future, the impact of agricultural trade on agricultural carbon emissions will be analyzed in more detail. Second, due to data limitations, we explored the relationship between agricultural trade and agricultural carbon emissions at the provincial level. In the future, if the improvement in statistical technology and methods can enhance the refinement of data in the field of agriculture, it will be possible to conduct a more detailed study on the impact of agricultural trade on agricultural carbon emissions at the city level or lower. At the same time, affected by the availability of data and the need to maintain the consistency of the study period, this study focuses more on changes in the utilization of cropland when measuring agricultural carbon emissions. Land use carbon emissions are also an essential part of carbon emissions from agricultural trade [73]. In the future, the inclusion of land-use carbon emissions in the accounting system will contribute to a richer analysis of the environmental effects of trade in agricultural products if data availability improves. In addition, since agricultural structures, policies, and even climatic conditions may vary significantly from country to country, exploring more generalizable models of agricultural trade and environmental effects in the context of big data and trade globalization will also be an important research direction. Third, regarding the effects of agricultural trade on agricultural pollution reduction and carbon reduction, this study empirically verified the main effects based on theoretical analysis; however, due to the limitations of the research theme and page, it could not verify the internal mechanisms from the measurement point of view. As a hot topic in the study of environmental effects, the impact mechanism of agricultural trade on the synergistic reduction in agricultural pollution and carbon emissions will be analyzed from a multidimensional perspective in the future.