The Impact of Agricultural Outward Foreign Direct Investment on Agricultural Imports: Evidence from China

Abstract

This study uses provincial panel data from China (2014–2022) to examine the effect of agricultural outward foreign direct investment (OFDI) on agricultural imports. Employing panel regression, mediation effect, and spatial Durbin models, it explores the underlying mechanisms, spatial spillover effects, and regional heterogeneity, while emphasizing the role of OFDI in building sustainable agricultural supply chains. The results show that: (1) OFDI significantly promotes agricultural imports, enhancing the stability and diversity of the domestic supply, supporting food security, and facilitating the sustainable allocation of resources. (2) Mechanism analysis reveals that OFDI affects imports through reverse technology spillovers and improved international relations. (3) Heterogeneity analysis indicates that OFDI exerts stronger influence in major grain-marketing areas, production–marketing balance regions, and provinces along the Belt and Road, compared with grain-producing areas and non-Belt and Road provinces. (4) Spatial analysis based on the 0–1 adjacency matrix reveals that agricultural imports across Chinese provinces exhibit significant positive spatial autocorrelation. Furthermore, OFDI not only directly promotes agricultural imports within a given province but also generates notable positive spatial spillover effects, whereby OFDI in neighboring provinces likewise exert a positive influence on the province’s agricultural imports. To enhance the import effect of agricultural OFDI and stabilize the domestic supply of agricultural products, policy implications suggest that the government should adhere to the agricultural “going global” strategy, enhance enterprises’ capacity to absorb reverse technology spillovers, and explore regionally differentiated pathways for agricultural OFDI, among other policy recommendations.

1. Introduction

Against the backdrop of economic globalization, agricultural trade has emerged as a crucial channel linking regions abundant in agricultural resources with those facing resource scarcity. Such trade facilitates the transnational flow of agricultural products and helps satisfy domestic demand in resource-scarce countries [1]. According to the FAO, the value of global food and agricultural trade reached USD 1.5 trillion in 2018, having more than doubling since 1995 [2], thereby underscoring the vital role of international trade in safeguarding global food security. However, amid Sino–U.S. trade frictions, the Russia–Ukraine conflict, and the resurgence of trade protectionism, many countries have implemented restrictions on agricultural exports. These measures have disrupted trade chains and imposed severe shocks on resource-scarce countries that depend heavily on imports [3,4]. In response, the international community has advocated for strategies such as agricultural OFDI to optimize the global allocation of agricultural resources [5]. As a proactive strategy, agricultural OFDI enables investing countries to engage deeply in the host country’s agricultural production by transferring capital and technology, utilizing local resources to produce agricultural goods, and subsequently importing them back to the home country. In this way, diversified supply channels can be established, product quality and variety enhanced, and both supply chain resilience and food security strengthened [6]. Moreover, agricultural OFDI can mitigate import risks and improve trade environments through the integration of agricultural science and technology, as well as investment in storage, logistics, processing, and information systems. Consequently, resource-scarce countries such as Japan and Saudi Arabia have widely adopted agricultural OFDI as a means to stabilize domestic supplies [7].

In China, economic development, increased household consumption, population growth, and expanding food demand—coupled with frequent conversion of farmland to non-agricultural uses and land abandonment—have collectively intensified the resource pressures faced by agricultural production [8]. Against this backdrop, agricultural OFDI has emerged as an important means of alleviating domestic resource constraints and reconciling the contradiction between supply and demand. By supporting agricultural development in host countries, China is able to utilize overseas resources to produce agricultural commodities and subsequently import them, thereby enhancing the stability of its domestic agricultural supply [9]. Since the launch of the “Going Global” agricultural strategy in 2006, the Chinese government has consistently encouraged agricultural OFDI, repeatedly emphasizing in the central government’s “No. 1 Documents” the need to coordinate the use of both domestic and international markets and resources, and to raise the level of international cooperation in agriculture. Under the impetus of such policies, China’s agricultural OFDI has expanded significantly, with the stock reaching USD 19.43 billion in 2020—an increase of approximately 23 times compared with that in 2006. Case studies illustrate this trend. For instance, COFCO successfully secured the concession for the Santos Port grain terminal, strengthening the construction of logistics facilities to enhance transshipment capacity and expand China’s imports of soybeans, corn, and sugar from Brazil. Similarly, WH Group integrated vertically into the global agricultural supply chain, internalizing import-related trade risks through enhanced supply chain management. Although existing cases demonstrate the positive impact of agricultural OFDI on agricultural imports and thus support its practical feasibility, direct research on the specific mechanisms of action and overall effects remains relatively scarce. Moreover, prior studies have suggested that agricultural products generated through outward investment typically enter the investor country’s market via imports, thereby stimulating import growth [10]. Consequently, it is necessary to conduct systematic and in-depth empirical research on both the effect and the mechanisms through which agricultural OFDI influences China’s agricultural imports.

The discussion surrounding agricultural foreign investment and agricultural imports has been explored by many scholars from different perspectives. First, agricultural foreign investment can bring capital to the host country through knowledge and technology spillover effects, thereby improving its food production efficiency. This helps feed more people in the host country, reduces its dependence on imported agricultural products [11], and promotes its participation in international trade [12]. Second, existing research on the relationship between agricultural OFDI and the agricultural imports of the investing country has not reached a unified view. Three main perspectives exist: The First Perspective: This view argues that although agricultural foreign investment can improve the global resource allocation efficiency and profitability of agricultural enterprises in the investing country, it has no significant impact on the imports of agricultural products by the investing country. These are seen as relatively independent economic activities [13]. The Second Perspective: This view emphasizes that agricultural foreign investment may have a negative effect on the importing country’s agricultural imports. The core mechanism is the “production transfer” and “industry chain extension,” where enterprises in the investing country localize production in the host country and extend the processing chain. As a result, primary agricultural products that could originally be exported are consumed locally or transformed into high-value-added products to be resold, reducing the export supply from the host country to the investing country [14]. The Third Perspective: This view advocates for a positive complementary relationship between agricultural foreign investment and agricultural imports. Multinational agricultural investments improve the agricultural production capacity and efficiency of the host country, expanding the global supply of agricultural products. This, in turn, provides more sources of imports for the global market, including the investing country [15,16]. Empirical studies support this perspective. Chiappini (2019) used an augmented standard gravity model to study the relationship between OFDI and Japan’s imports and exports, finding that the complementary relationship between foreign direct investment and trade is dominant in industries such as Japan’s food sector [17]. Holzman (2004) also showed that there is a significant complementarity between Canada’s agricultural foreign investment and its agricultural imports [18]. Lu (2018) and other scholars further pointed out that agricultural foreign investment has a particularly strong promoting effect on the import of staple products like rice [19]. However, the current literature mainly remains descriptive of correlations, lacking differentiation of heterogeneous investment motivations and mechanism analysis. For instance, “resource-seeking” investments might enhance the stability of imports by controlling upstream production, while “market-seeking” investments are more likely to focus on local sales, suppressing import flows. “Strategic asset-seeking” investments have more complex influence pathways, but existing research lacks systematic classification and comparative analysis, which results in unclear policy implications.

In recent years, the reverse technology spillover effects of OFDI have attracted widespread attention from scholars. However, compared to the manufacturing sector, research in the agricultural field on this topic is still significantly lagging, with substantial gaps in both theoretical framework construction and empirical analysis. Existing studies indicate that agricultural foreign investment generates reverse technology spillovers mainly through three paths. The first one is direct learning. This occurs through the direct acquisition of advanced agricultural technologies and management experiences from the host country. The second one is human resource mobility. Knowledge transfer is achieved through the movement of human resources. And the last one is industry chain integration. Technologies are applied within global production networks through industry chain integration [20]. In terms of empirical research, Michée (2025) conducted a study on Latin American countries, which showed that, despite differing levels of agricultural productivity development across countries, knowledge spillovers and research and development (R&D) collaboration can significantly narrow the technological gap between countries in the agricultural sector [21]. Similarly, studies based on Chinese cases also pointed out that knowledge and technology can positively impact cross-regional agricultural total factor productivity through spatial spillover effects [22,23]. Han and Kim (2023) further conducted an empirical analysis based on data from South Korea, proposing that the effect of reverse technology spillover on total factor productivity enhancement highly depends on the absorptive capacity of firms in the home country [24]. Additionally, some literature suggests that this spillover effect helps improve domestic human capital levels, stimulates R&D innovation, and thus systematically drives progress in agricultural productivity [25]. However, not all outward investments generate reverse technology spillover effects. Enhanced communication between investors and host country enterprises is necessary to foster bilateral knowledge and technology flows [26].

Despite notable progress in the existing literature, several significant limitations remain. First, most studies focus on examining the direct effects of outward investment on productivity or technological progress, while offering limited exploration of the underlying mechanisms and heterogeneity. In particular, research on the pathways through which agricultural investment influences agricultural trade is severely underdeveloped. Second, existing empirical studies are largely based on national-level data, overlooking substantial interregional differences in economic development, resource endowments, and policy environments, which constrains the generalizability of their conclusions. Finally, although some studies mention spatial spillover effects, most fail to establish a comprehensive spatial econometric framework, making it difficult to identify and quantify the cross-regional impacts of agricultural outward investment on agricultural imports.

Considering the previous discussion, our study makes some noteworthy contributions. First, this study examines the direct impact and mechanisms of agricultural OFDI on China’s agricultural imports. While most existing research emphasizes the spatial distribution and driving factors of agricultural OFDI, few studies have explored its effects on the agricultural product markets of investing countries. This study therefore fills that gap, offering new insights into the influence of agricultural OFDI on international trade patterns and China’s agricultural import trade, and providing practical implications for promoting the development of China’s agricultural imports. Second, this study examines the heterogeneous impacts of agricultural OFDI on agricultural imports across regions with different levels of economic development and resource endowments in China. By avoiding the homogenization of diverse regional conditions, this analysis provides a more nuanced and comprehensive perspective than existing research. Moreover, the findings can help the government formulate more precise and targeted policies to maximize the positive effects of agricultural OFDI on agricultural imports. Third, this study further explores the spatial spillover effects of agricultural OFDI affecting agricultural imports, which contributes to a more comprehensive assessment of the integrated impact of agricultural outward investment and provides a richer perspective for policy formulation.

Therefore, this study seeks to answer the following questions: Does agricultural OFDI significantly influence China’s agricultural imports? If so, through what specific mechanisms does agricultural OFDI exert its impact? Which groups are more likely to increase agricultural imports by conducting agricultural OFDI? Whether there are spatial spillovers from agricultural OFDI affecting agricultural imports? Answering these questions could help the Chinese government explore feasible and effective ways to stabilize the supply of agricultural products. Accordingly, based on the relevant data of China’s provincial agricultural imports and agricultural OFDI from 2014 to 2022, this study firstly analyzes the impact of China’s agricultural OFDI on agricultural imports by using panel regression model and reveals its impact mechanism, then analyzes the heterogeneity of agricultural OFDI affecting agricultural imports by grouping regression model, and finally constructs a spatial Durbin model to explore the spatial spillover effect of agricultural OFDI affecting agricultural imports. This study selects the provincial level and the period from 2014 to 2022 as the research sample, based on three main considerations. First, in terms of data availability and quality, provincial-level data strike the optimal balance between granularity and statistical standardization, thereby ensuring continuity and consistency of the sample. Second, given the substantial disparities across Chinese provinces in terms of resource endowments and economic development, provincial-level data effectively capture such heterogeneity, making the conclusions more reflective of reality. Finally, taking 2014 as the starting point allows the analysis to better capture the new characteristics of agricultural outward investment following the launch of the Belt and Road Initiative. Moreover, the data for this period are publicly available and complete, while 2022 represents the most recent year with accessible data, ensuring the timeliness of the study.

2. Material and Methods

2.1. Modeling

Based on data availability, the study select China’s provinces from 2014 to 2022 as research samples, and took the annual amount of agricultural imports and the amount of agricultural OFDI of each province as the explained variable and the explanatory variable, and build a panel regression model to explore the impact of agricultural OFDI on agricultural imports. The model is specified as follows:

$$Y_{it}={\alpha }_0+{\alpha }_1{x}_{it}+{\displaystyle \sum _{j=1}^n{\beta }_j}{x}_{jt}+{\mu }_i+{\epsilon }_{it}$$

(1)

where $Y_{it}$ denotes the amount of agricultural imports, and ${\mathrm{x}}_{\mathrm{i}\mathrm{t}}$ denotes the amount of agricultural OFDI, and ${\mathrm{x}}_{\mathrm{j}\mathrm{t}}$ denotes other factors affecting agricultural imports, and ${\mathsf{\alpha }}_0$ denotes the constant, and ${\mathsf{\mu }}_{\mathrm{i}}$ denotes individual effects, and ${\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$ denotes the random error.

In addition, with reference to the intermediary effect test method of Baron and Kenny [27], the intermediary effect model is constructed on the basis of model (1) to test the influence mechanism of agricultural OFDI on agricultural imports, as follows:

$$M_{it}={\alpha }_0+{\alpha }_1{x}_{it}+{\displaystyle \sum _{j=1}^n{\beta }_j}{x}_{jt}+{\mu }_i+{\xi }_{it}$$

(2)

$$Y_{it}={\alpha }_0+{\alpha }_1{x}_{it}+{\alpha }_2{M}_{it}+{\displaystyle \sum _{j=1}^n{\beta }_j}{x}_{jt}+{\mu }_i+{\xi }_{it}$$

(3)

where $M_i$ denotes the intermediary variable.

2.2. Variable Selection

(1)

Explained variables. The import volume of agricultural products for each province from 2014 to 2022 represents the explained variable.

(2)

Explanatory variables. The agricultural OFDI stock for each province spanning from 2014 to 2022 serves as the explanatory variable due to its ability to reflect the cumulative impact of investment levels.

(3)

Control variables. The import of agricultural products is influenced by a variety of factors, including economic development level, population size, resource endowment, and market environment. Therefore, this study utilizes control variables such as savings deposits per capita, total population, crop planting area, total water resources, agricultural product prices index, food disaster area, and investment in agricultural fixed assets.

(4)

Intermediary variables. The intermediary variables are represented by the grain output of individual provinces and the friendship cities between provinces and other countries.

The statistical features of the variables are displayed in

Table 1. Definition and descriptive statistics of variables.

Variable Name	Unit	Mean	Standard Deviation	Min	Max	Median	Mode
Explained Variables
Agricultural imports	billion dollars	53.114	74.546	0.129	321.664	15.156	15.156
Explanatory Variable
Stock of outward investment in agriculture	billion dollars	6.317	11.102	0.035	78.920	2.540	0.280
Intermediary variable
Grain production by province	ten thousand tons	2240.668	1844.398	28.800	7867.700	1527.100	1521.300
Number of friendship cities	pairs	89.188	53.125	10.000	362.000	82.000	96.000
Control variables
Agricultural exports	billion dollars	27.028	38.421	0.924	209.029	11.724	—
Savings deposits per capita	ten thousand dollars	0.884	0.522	0.293	3.991	0.730	—
Total population	ten thousand people	4789.364	2848.487	678.000	12,684.000	4104.000	2520.000
Crop planting area	thousand hectares	3979.650	3153.608	46.500	14,683.200	3150.300	6453.900
Total water resources	billion cubic meters	820.256	740.883	8.100	3237.300	539.900	168.400
Agricultural product prices index	-	104.131	6.308	89.500	123.300	102.700	99.500
Food disaster area	thousand hectares	666.705	722.491	0.000	4224.000	484.000	0.001
Investment in agricultural fixed assets	billion dollars	11.776	9.689	0.000	47.396	10.531	3.113

.

2.3. Data Source

The import and export value of agricultural products of each province are from the website of the Ministry of Commerce of China, and the data related to agricultural OFDI are from the Analysis Report on China’s Agricultural Outward Foreign Direct Investment and Cooperation from 2015 to 2023, with Tibet Autonomous Region and Qinghai Province excluded because of the large number of missing data. The rest of the indicators are mainly from the China Statistical Yearbook, China Rural Statistical Yearbook, provincial statistical yearbooks, the World Bank, the China Association of International Friendship Cities, etc., and the missing values are supplemented by the interpolation method. At the same time, to mitigate the influence of outliers, all variables in this study were winsorized at the 5% level.

To investigate regional heterogeneity in how agricultural outward investment affects agricultural imports, the sample provinces are categorized according to their grain supply–demand characteristics into major grain-producing areas, major grain-marketing areas, and grain supply–demand balanced areas (

Table 2. Regional classification of grain supply areas.

Name of Region	Definition of Region	Constituent Provinces
Major Grain-Producing Areas	A high-yield grain production base generating surplus beyond local consumption for regional distribution	Heilongjiang, Jilin, Liaoning, Inner Mongolia, Hebei, Henan, Shandong, Jiangsu, Anhui, Jiangxi, Hubei, Hunan, Sichuan
Major Grain-Marketing Areas	A major grain consumption region dependent on imports due to insufficient local production	Beijing, Tianjin, Shanghai, Zhejiang, Fujian, Guangdong, Hainan
Grain Supply–Demand Balanced Areas	A self-sufficient grain region maintaining supply–demand equilibrium	Shanxi, Ningxia, Gansu, Yunnan, Guizhou, Chongqing, Guangxi, Shaanxi, Xinjiang

) for comparative analysis.

3. Results

3.1. Baseline Regression Results

The data used in this paper are panel data and are usually estimated using a fixed effect model or a random effect model. However, since the fixed effect model can fix individual information and time at the same time, it is not only conducive to controlling differences in individual characteristics and eliminating individual invariance effects, thereby reducing the heteroscedasticity and correlation of error terms, and can also accurately capture the characteristics of individuals changing over time and improve the accuracy of model estimation results. So the study finally adopts the panel fixed effects model to analyze the impact of China’s agricultural OFDI on agricultural imports.

The results of model 1 in

Table 3. Effects of agricultural outward investment on agricultural imports.

Variable Name	Model 1	Model 2
Stock of outward investment in agriculture	1.010 *** (0.162)	0.344 ** (0.134)
Agricultural exports		0.765 *** (0.182)
Savings deposits per capita		64.906 *** (5.045)
Total population		0.023 ** (0.011)
Crop planting area		−0.011 *** (0.004)
Total water resources		−0.007 (0.005)
Price of agricultural products		−0.542 *** (0.190)
The area of food disaster		0.003 (0.002)
Agricultural fixed assets investment		−0.519 ** (0.255)
Constant	40.033 *** (3.925)	−28.040 (59.923)
R2	0.433	0.667
Observations	261	261

Note: **, *** indicate significant at the levels of 5% and 1%, respectively.

show that the estimated coefficient of agricultural OFDI stock is 1.010, and significant at 1% statistical level, indicating that there is a significant positive effect of agricultural OFDI stock on agricultural imports. To eliminate the influence of other factors on agricultural imports as much as possible, model 2 includes additional controls for variables such as the economic development level, population size, and resource endowment of each province. The results show that the influence coefficient of agricultural OFDI stocks decreases to 0.344 but remains statistically significant at the 5% level, which implies that disregarding additional factors could overstate the effect of agricultural OFDI on agricultural imports. Nonetheless, expanding agricultural OFDI indeed results in a boost in agricultural imports. Specifically, an increase of $100 million in the investment stock would generate $34.4 million in agricultural imports.

3.2. Robustness Test

In the study, we utilize two approaches of substituting the primary explanatory factors and eliminate a portion of the sample size to perform a robustness analysis. First, the main explanatory variable is substituted. The model for estimation replaces the investment stock with the number of agricultural enterprises engaged in outward investment in each province, as the number of investing enterprises can indicate the investment scale to a certain extent. The results of model 3 in

Table A1. Robustness test of the impact of agricultural outward investment on agricultural imports.

Variable Name	Model 3: Replacement Variables	Model 4: Exclusion Sample
Number of Enterprises Engaged in Agricultural Outward Investment	0.298 ** (0.117)
Stock of outward investment in agriculture		0.265 ** (0.116)
Constant	−110.628 *** (70.574)	−180.063 ** (71.129)
Control variable	yes	yes
R2	0.714	0.606
Observations	261	225

Note: **, *** indicate significant at the levels of 5% and 1%, respectively.

(Detailed table is available in the Appendix A.1) show that the amount of enterprises has a significant and positive influence on agricultural imports, that is, the estimated coefficient’s direction and significance align with the baseline regression’s findings. Second, some samples are excluded. As the government has given more preferential policy support to improve the import of agricultural products of municipalities directly under the Central Government, which brings errors to the estimation results of the model, in order to minimize these interference factors, this study excludes the samples of municipalities directly under the Central Government. After removing the sample, the model 4 in Table A1 shows the stock of agricultural OFDI still has a significant and positive effect on agricultural imports, which demonstrates the robustness of the baseline regression results.

3.3. Endogeneity Test

The model set up in this paper may have endogenous problems, and the two possible sources of endogenous problems are: (1) Omitted variables. Although the factors that will affect the import of agricultural products have been added to the baseline regression model as much as possible, there are still some unobservable variables left out, resulting in biased results. (2) Bidirectional causation. This study examines the impact of agricultural OFDI on agricultural imports, but it is also possible that agricultural imports influence agricultural OFDI. The reason is that agricultural OFDI aims to expand agricultural import channels and stabilize domestic supply, but at the same time, as imports increase, domestic firms will become more familiar with foreign markets, which will prompt firms to carry out further agricultural OFDI. As a result, there could be a potential reverse causality between agricultural OFDI and agricultural imports, which can result in endogeneity issues and inaccurate estimates. Therefore, this study employs the instrumental variable (IV) approach to address endogeneity. First, referring to the research of relevant scholars, this study uses agricultural outward investment and its cubic mean as an instrument variable [28]. The reason is that variable not only ensures a strong nonlinear correlation with the endogenous variable, but also, since its generation mechanism derives entirely from higher-order moments of the data, it is independent of the error term. Hence, it theoretically satisfies the exogeneity constraint of instrumental variables. Second, the first lag of agricultural outward investment is employed as an instrumental variable. The reason is that agricultural outward investment is time-series data with temporal dependence. Using the first lag allows us to capture such autocorrelation, while the lagged variable is predetermined and thus unaffected by contemporaneous shocks. In other words, the lagged agricultural outward investment is uncorrelated with the current error term, which implies that the lag variable theoretically satisfies both the relevance and exogeneity conditions of an instrumental variable. Subsequently, the two instrumental variables are incorporated separately into the model, and estimation is conducted using two-stage least squares (2SLS) (see

Table A2. Endogeneity test of the impact of agricultural outward investment on agricultural imports.

Variable Name	Model 5: Agricultural Outward Investment and Its Cubic Mean		Model 6: Lagged Agricultural Outward Investment
	First Stage	Second Stage	First Stage	Second Stage
Agricultural Outward Investment and Its Cubic Mean	0.002 *** (0.000)
Lagged Agricultural Outward Investment			0.437 *** (0.076)
Agricultural Outward Investment Stock		1.047 *** (0.166)		1.475 *** (0.371)
Constant	12.058 (17.124) –31.709 (88.562)		25.629 (46.365)–148.634 (101.241)
Control Variables	Yes	Yes	Yes	Yes
F-statistic	72.65 ***		4.21 ***
Hausman Test	12.08 **		20.75 **

Note: **, *** indicate significant at the levels of 5% and 1%, respectively.

in Appendix A.2).

The results of Models 5 and 6 indicate that the Hausman tests in both models pass the significance test, suggesting that agricultural outward investment is indeed an exogenous variable and confirming the appropriateness of the instrumental variable approach. In addition, the F-statistics from the first stage are all significantly greater than the baseline value of 10, implying that both instrumental variables are not weak instruments. After addressing endogeneity through the instrumental variable method, the stock of agricultural OFDI continues to exert a significant positive impact on agricultural imports in both models, which is consistent with the results of baseline regression.

3.4. Influence Mechanism

(1) China’s agricultural OFDI reduces agricultural imports through reverse technology spillover effects. The purpose of the Chinese government to encourage agriculture to “go out” is also to strengthen communication and exchange with foreign agricultural production, bring out superior technologies, and at the same time make use of foreign resources to develop new technologies and introduce foreign advanced technologies, so as to promote the construction of an agricultural power. As a result, China’s enterprises carry out agricultural OFDI to learn and absorb foreign advanced technology and apply it to the country, which can improve domestic agricultural productivity and thus have an impact on agricultural imports, which is one of the ways for agricultural OFDI to affect the import of agricultural products. Based on available data, the study utilizes the grain yield of each province as an indicator of agricultural productivity.

The results of model 7 in

Table 4. Mechanism test of the impact of agricultural outward investment on agricultural imports (1).

Variable Name	Model 7: Grain Production	Model 8: Agricultural Imports
Stock of outward investment in agriculture	1.243 * (0.652)	0.205 (0.135)
Grain production		−0.036 *** (0.013)
Constant	−89.211 (80.075)	−138.284 (70.238)
Control variable	yes	yes
R2	0.830	0.719
Observations	261	261

Note: *, *** indicate significant at the levels of 10% and 1%, respectively.

show that agricultural OFDI has a significant positive impact on grain production, suggesting that it can result in reverse technology spillovers and subsequently enhance domestic agricultural production. The results of model 8 show that increased grain production significantly reduces agricultural imports, the reason is high production means that the domestic agricultural self-sufficiency rate has been improved, which can reduce external dependence and reduce imports.

(2) China’s agricultural OFDI can promote agricultural imports by strengthening international relations. Specifically, it helps enhance political ties between China and host countries, thereby laying a solid foundation for the expansion of import trade. Most investment areas are focused on developing countries in Asia and Africa, where agricultural technology is less advanced. During the investment process, China’s enterprises provide the host country with capital and advanced production technology to help raise the level of local agricultural production and improve the supply of agricultural products, which helps to enhance the host country’s recognition of China, establish friendly international relations, reduce trade frictions and reduce the impact of trade protectionism, so as to promote China’s import of agricultural products. Therefore, the study utilizes the number of friendship cities signed by each province to indicate the level of international relations friendship.

The results of model 9 in

Table 5. Mechanism test of the impact of agricultural outward investment on agricultural imports (2).

Variable Name	Model 9: Number of Friendship Cities	Model 10: Agricultural Imports
Stock of outward investment in agriculture	0.090 * (0.053)	0.173 (0.135)
Number of friendship cities		0.500 *** (0.174)
Constant	77.219 *** (3.593)	−144.912 ** (70.315)
Control variable	yes	yes
R2	0.707	0.719
Observations	261	261

Note: *, **, *** indicate significant at the levels of 10%, 5%, and 1%, respectively.

indicate that agricultural OFDI has a significantly increases the number of friendship cities, pointing to a potential enhancement of international relations between China and the host country through investment expansion. The results of model 10 show that the number of friendship cities exerts a significant and positive effect on agricultural imports, with results reaching a 1% statistical significance level, indicating that the improvement of international relations is an important factor in increasing trade between the two countries. To sum up, agricultural OFDI can boost international relations, which can then support an increase in agricultural imports.

3.5. Heterogeneity Analysis

There may be variations in economic development, resource allocation, geography, and national policies among provinces, leading to heterogeneous effects of agricultural OFDI on agricultural imports. In addition, the Belt and Road Initiative creates opportunities for China’s agricultural trade and OFDI. However, differences in policy environments between provinces along the route and those outside it may lead to heterogeneous impacts of agricultural OFDI on agricultural trade. Therefore, this study examines the diverse effects of agricultural OFDI on agricultural imports in various regions, including major grain-producing areas, major grain-marketing areas, grain supply–demand balanced areas, as well as provinces along and not alone the “Belt and Road”.

(1)

Three functional areas for grain production. Model 11 to 13 in

Table 6. Heterogeneity analysis of the impact on agricultural imports (1).

Variable Name	Model 11: Major Grain-Producing Areas	Model 12: Major Grain-Marketing Areas	Model 13: Grain Supply–Demand Balanced Areas
Stock of outward investment in agriculture	0.284 (0.709)	0.652 ** (0.267)	0.681 * (0.415)
Constant	177.317 (196.636)	−82.460 (146.561)	−43.330 (44.920)
Control variable	yes	yes	yes
R2	0.433	0.807	0.656
Observations	117	63	81

Note: *, ** indicate significant at the levels of 10% and 5% respectively. Coefficient difference test between major grain-producing areas and grain supply–demand balanced areas = 0.446 (p = 0.199).

results show that the impact of agricultural OFDI on agricultural imports in different regions is different. Specifically, agricultural OFDI has a positive but not significant impact on agricultural imports in major grain-producing areas, in contrast, agricultural OFDI has a significant positive impact on agricultural imports in major grain-marketing areas and grain supply–demand balanced areas at the 5% and 10% statistical levels, respectively. The primary reason for this disparity lies in the dominant role of grain production in the major producing regions, where strong self-sufficiency and low external dependence render the marginal effect of agricultural outward investment on imports relatively limited. On the other hand, although the sample sizes across all regions are relatively small—somewhat weakening the power of statistical inference—major grain-marketing areas and the grain supply–demand balanced areas still exhibit significant effects, which are closely tied to their own characteristics. The major grain-marketing areas are mostly located in the eastern coastal regions, characterized by developed economies but scarce agricultural resources. These regions are net grain inflow areas with high external dependence. Given their limited local production capacity, agricultural OFDI plays a particularly prominent role in enhancing access to overseas grain supplies. In contrast, grain supply–demand balanced areas are concentrated in the western border provinces, where economic development and production technology lag behind. However, their proximity to national borders provides geographic advantages that facilitate agricultural investment and cross-border trade, thereby amplifying the import-promoting effects of outward investment. It is noteworthy that although the estimated coefficient for the balanced areas is slightly higher than that of the marketing areas, coefficient difference tests reveal that the two are not statistically significant. This indicates that the positive effect of agricultural outward investment on agricultural imports does not differ significantly between major grain-marketing areas and grain supply–demand balanced areas (see Table 6).

(2)

Provinces along the “Belt and Road” and provinces not along the route. The results of model 14 and model 15 in

Table 7. Heterogeneity analysis of the impact on agricultural imports (2).

Variable Name	Model 14: Provinces Along the “Belt and Road”	Model 15: Provinces Not Along the “Belt and Road”
Stock of outward investment in agriculture	0.316 ** (0.139)	0.432 (0.326)
Constant	−131.568 ** (57.317)	203.224 (243.267)
Control variable	yes	yes
R2	0.815	0.682
Observations	144	117

Note: ** indicate significant at the levels of 5%.

show that agricultural OFDI has a significant positive impact on agricultural imports in provinces along the “Belt and Road”, and a positive but statistically insignificant impact on imports from non-route provinces. The reason is that the “Belt and Road” initiative provides an important opportunity for the trade development of provinces along the route, and infrastructure, policy communication and personnel exchanges are the key cooperation elements emphasized by the initiative. The improvement of the infrastructure is conducive to the improvement of road accessibility and the level of transportation facilitation between the provinces along the route and the host countries. Effective policy communication can encourage trade agreements between parties, lower trade tariffs, and facilitate a favorable business environment. Personnel exchanges help to improve the transparency of information and the matching degree of cooperation between the two sides, and then improve the level of trade cooperation. Therefore, the provinces along the route have geographical advantages, convenient transportation conditions, a favorable communication environment, and corresponding trade policy support under the “Belt and Road” initiative in comparison to those not along the route, which has resulted in better conditions for expanding the trade scale and is more conducive to fully utilizing the import effect of agricultural OFDI.

3.6. Spatial Spillover Effect

Spatial spillover effect refers to the impact of a region’s economic activities on neighboring regions. In this study, when a province carries out agricultural OFDI, it may have an impact on the agricultural imports of neighboring provinces. Therefore, the study constructs a spatial econometric model to test whether there is a spatial spillover effect of agricultural OFDI on agricultural imports. The study calculates the Moran index based on the 0–1 adjacency matrix to test whether the imports of agricultural products have spatial autocorrelation, and the results in

Table 8. Global Moran’s index test results for agricultural imports by province in China.

Year	Moran’s I	p	Z
2014	0.085	0.122	1.163
2015	0.106	0.102	1.268
2016	0.161 **	0.039	1.758
2017	0.165 **	0.036	1.798
2018	0.176 **	0.029	1.901
2019	0.108 *	0.098	1.293
2020	0.123 *	0.078	1.420
2021	0.076	0.160	0.993
2022	0.098	0.118	1.185

Note: *, ** indicate significant at the levels of 10% and 5% respectively.

show that the Moran index of the imports of agricultural products in most years is significantly positive, indicating that the imports of agricultural products in all the provinces show a spatial positive correlation, which confirms the reasonableness of the use of spatial econometric model.

(1)

Determination of spatial econometric model. The Moran index analysis provided an initial test for spatial effects, and to further determine the specific form of the regression model, the LM test, the LR test, and the Hausman test were performed. The results show that the three tests significantly reject the original hypothesis, indicating that it is appropriate to choose the spatial Durbin model under the fixed effect to analyze the spatial spillover effects of agricultural OFDI on agricultural imports.

(2)

The regression results of the Spatial Durbin Model (SDM) in Model 16 (

Table 9. Results of spatial Durbin model (SDM) estimation of the impact on agricultural imports.

Variable Name	Model 20: SDM
	Coefficient	Standard Error
Stock of outward investment in agriculture	2.080 ***	0.325
W × Stock of outward investment in agriculture	1.222 **	0.604
ρ	0.196 **	0.083
Control variable	yes
R2	0.265
Observations	261

Note: **, *** indicate significant at the levels of 5% and 1%, respectively.

) show that the spatial lag coefficient (ρ) is positive and statistically significant at the 5% level. This indicates a significant positive spatial externality of agricultural OFDI. After accounting for spatial factors, agricultural OFDI continues to have a significant positive impact on agricultural imports. Furthermore, the spatial effect of agricultural OFDI on imports remains significantly positive, suggesting that OFDI in neighboring provinces promotes increased agricultural imports in the local area. The possible reason is that, on the one hand, agricultural investment by neighboring provinces abroad may increase the supply of agricultural products in the region, but due to the limited neighboring provinces’ demand, some agricultural products may be exported to local provinces. On the other hand, due to differences in transportation conditions among provinces, in order to save transportation costs, provinces with outward investment in agriculture may transport agricultural products to neighboring provinces with convenient transportation.

(3)

Decomposition of spatial spillover effect. The spillover effect can be divided into direct effect and indirect effect, the direct effect refers to the influence of the province’s agricultural OFDI on the province’s agricultural import, and the indirect effect refers to the influence of the neighboring province’s agricultural OFDI on the province’s agricultural import. The results of

Table 10. Decomposition of effects in the spatial Durbin model of the impact on Agricultural imports.

Variable Name	Total Effect		Direct Effect		Indirect Effect
	Coefficient	Standard Error	Coefficient	Standard Error	Coefficient	Standard Error
Stock of outward investment in agriculture	4.154 ***	0.833	2.174 ***	0.338	1.980 ***	0.688
Control variable		yes
Observations		261

Note: *** indicate significant at the levels of 1%.

show that the total, direct and indirect effects of agricultural OFDI on agricultural import are significantly positive, indicating that agricultural OFDI in neighboring provinces does have a promoting effect on the import of agricultural products in the province. Which also inspires that all regions should establish a complete cooperation mechanism to give full play to the spillover effects of agricultural OFDI.

4. Discussion

Based on the above analysis results, this study discusses the following:

Firstly, agricultural OFDI has a significant positive impact on China’s agricultural imports, indicating that agricultural OFDI is conducive to increasing China’s agricultural imports. China’s agricultural OFDI is often carried out by transferring funds and technology to countries with insufficient investment and low agricultural production efficiency, making use of agricultural resources such as cultivated land, water, labor and other agricultural resources of the host country to carry out agricultural production, and increase the output of agricultural products of the host country by improving agricultural production efficiency. Whether these new agricultural products flow into the domestic market or the international market, they will increase the supply of agricultural products in the entire international market, which in turn will have a beneficial impact on China’s increased imports [29].

Secondly, agricultural OFDI can increase domestic agricultural production through reverse technology spillovers, thereby reducing domestic agricultural import demand. At present, the vast majority of scholars agree that there is a technological spillover effect in foreign investment, that is, carrying out foreign investment is conducive to improving the technological level of the host country [30], but only a small number of scholars believe that there is a reverse technology spillover effect in foreign investment, which can improve the technological level of the investing country. Kogut and Chang [31] first proposed the concept of the reverse technology spillover effect in 1991 when they examined the impact of Japan’s foreign investment in the United States. They argued that investing in foreign countries allows firms to acquire advanced technological knowledge and innovative practices from their overseas counterparts, which can then be transferred back to the investing home country, leading to improvement in the overall technological level of the investing country. Later, the reverse technology spillover effect was confirmed by some scholars [32], and this study also confirms the view that China’s agricultural OFDI contributes to enterprises’ learning, absorption of foreign production technology, and subsequent feedback to China, which improves domestic agricultural production levels through reverse technology spillover effects, further increasing the output of agricultural products, and helping to reduce domestic dependence on agricultural imports. This mechanism appears to contradict the baseline regression result that “agricultural outward investment exerts a significant positive effect on China’s agricultural imports.” In fact, the two are not in conflict and can be explained from several perspectives. First, the reverse technology spillover effect exhibits time lags. The absorption, application, and diffusion of agricultural technologies require relatively long cycles, making it difficult to generate effective import substitution in the short term. By contrast, agricultural OFDI often directly drives the import of related agricultural products and raw materials, which is reflected in the data as an immediate positive effect. Second, at the macro level, the demand for agricultural products continues to expand. Even if reverse technology spillovers contribute to increased domestic production, concurrent growth in domestic consumption, as well as dietary upgrading, may further amplify import demand. In other words, while investment promotes supply growth, demand expands even faster, ultimately resulting in import expansion. Third, there exists a direct linkage between firms’ outward investment behavior and imports. When enterprises invest abroad in the cultivation or processing of agricultural products, they often transport the output directly back to the domestic market. Statistically, this cross-border flow is recorded as agricultural imports. In sum, the restraining effect of reverse technology spillovers on agricultural imports is long-term and structural, whereas the promoting effect of agricultural outward investment on imports is multi-path, immediate, and significant. These mechanisms coexist and operate at different levels, with the macro-level net effect manifesting as an increase in imports, consistent with the baseline regression findings. This study not only confirms the applicability of reverse technology spillovers in the agricultural sector but also provides a more systematic explanation for the complex relationship between agricultural OFDI and imports.

Thirdly, this study finds that agricultural outward investment across Chinese provinces not only directly promotes agricultural imports within the investing province, but also generates significant positive spatial spillover effects on neighboring provinces. This conclusion aligns with the perspective of related literature, which emphasizes that agricultural outward investment and trade flows are highly susceptible to systemic influences from the global economic and political environment [33]. The interplay of shifting economic conditions, geopolitical tensions (e.g., the Sino–U.S. trade frictions and the Russia–Ukraine conflict), and other policy-driven factors may trigger chain reactions within international trade networks, thereby exerting profound impacts on agricultural trade flows at both regional and global scales [34,35]. Consistent with this view, the study by Georgios and Nikos (2023) similarly demonstrates that economic shocks in China not only reshape the country’s own trade networks but also generate significant spillover effects on its major trading partners, including the United States, Japan, and South Korea [36]. This finding provides further support for the presence of spatial externalities identified in the present study. Moreover, when analyzing the heterogeneity of the impact of agricultural OFDI on agricultural imports in different provinces, this study also finds that agricultural OFDI plays a more significant role in promoting agricultural imports in provinces along the “Belt and Road”. This aligns with the research findings of relevant scholars [37], who believe that the “Belt and Road” initiative provides an important opportunity for the trade development of relevant provinces in China, the provinces and countries along the route can gain rich and mutually beneficial economic benefits from the “Belt and Road” initiative, which is conducive to the establishment of extensive economic and trade cooperation between the two sides [38], and is more conducive to giving full play to the import effect of agricultural OFDI.

Fourthly, while this study provides an empirical analysis of the impact of agricultural outward investment on agricultural imports, certain limitations remain, offering promising directions for future research. First, there is scope to expand the selection of control variables. Although this paper has accounted for multiple factors such as economic development, population, and resource endowment, several theoretically important variables could not be included due to data availability constraints. For example: (1) Agricultural technology level—regional innovation and technological advancement are crucial determinants of domestic agricultural supply capacity [39]. Future studies may incorporate more detailed data on provincial agricultural R&D expenditure or agricultural technology market transactions to better disentangle the effects of technological progress. (2) Policy and institutional factors—policies such as the establishment of “Belt and Road” hub cities or pilot zones for agricultural opening-up and cooperation are likely to exert profound influences on agricultural investment and trade. Future work could adopt causal identification strategies, such as difference-in-differences (DID), to rigorously evaluate the net effects of these policies. (3) International logistics and trade costs—indicators reflecting trade facilitation, such as port throughput capacity or the number of international railway services, directly affect import costs. Subsequent research could attempt to construct a more comprehensive trade facilitation index to better capture this dimension. Second, this study primarily focuses on the macro-level analysis at the provincial scale. Future research may refine the analysis to the micro-level of enterprises by utilizing firm-level outward investment and trade data, thereby revealing the behavioral mechanisms of microeconomic actors and mitigating the aggregation bias inherent in macro data. Despite these limitations, the core conclusions of this paper remain robust under the current set of control variables and model specifications. Acknowledging such limitations does not undermine the value of the study; rather, it aims to provide valuable guidance for subsequent research that is more detailed and in-depth.

5. Conclusions

Based on provincial panel data for China from 2014 to 2022, this study employs a combination of fixed-effects models, mediation models, and spatial Durbin models to systematically examine the impact, mechanisms, and regional heterogeneity of agricultural outward investment on agricultural imports. The empirical findings are as follows. First, agricultural outward investment significantly promotes agricultural imports overall. Second, this impact operates through two mechanisms—“reverse technology spillovers” and “improvement of international relations.” While the former partly suppresses imports by enhancing domestic production capacity, the latter facilitates trade flows, with the net effect being an expansion of imports. Third, the effects of agricultural outward investment on agricultural imports exhibit marked regional heterogeneity and spatial spillover effects. Compared with the major grain-producing regions, the positive impact is more pronounced in grain marketing areas and supply–demand balanced areas. Moreover, the impact is stronger for provinces along the Belt and Road compared to non-Belt and Road provinces. Fourth, agricultural imports across provinces display significant positive spatial correlation, and agricultural outward investment not only directly promotes imports within the investing province but also generates substantial positive spillover effects, whereby investment activities in neighboring provinces contribute to higher imports locally.

This study makes several contributions at both the theoretical and policy levels. Theoretically, it enriches empirical research on the nexus between agricultural outward investment and international trade. First, by identifying the seemingly contradictory but coexisting mechanisms of “reverse technology spillovers” and “improvement of international relations,” it uncovers the complex pathways through which agricultural outward investment affects imports, providing a more nuanced theoretical explanation of the investment–trade linkage. Second, by incorporating spatial spillover effects into the analytical framework, it demonstrates that agricultural outward investment influences not only the investing region but also neighboring provinces, thereby extending the application of spatial econometric methods in agricultural economics. Third, adopting the perspective of food security, it reveals heterogeneous effects across major producing, marketing, and supply–demand-balanced areas, offering theoretical foundations for differentiated policy design.

At the policy and practice level, the findings provide critical insights for optimizing China’s agricultural outward investment layout and safeguarding food security. First, the results reinforce the policy legitimacy of adhering to the “going global” strategy in agriculture. The government should continue to improve support systems for outward investment and incentivize enterprises through fiscal and financial measures, thereby strengthening market mechanisms and building stable and sustainable import supply chains. Second, a “dual-pathway” strategy should be pursued. On one hand, China should enhance the absorption of reverse technology spillovers by increasing investment in agricultural science and technology and attracting talent, thus fostering technological internalization and capacity upgrading. On the other hand, international relations should be integrated into the policy objectives of agricultural outward investment, with greater emphasis on regulatory compliance, corporate social responsibility, and trust-building with host countries. Third, differentiated regional investment guidance policies should be implemented: encouraging major grain-producing areas to focus on technology-seeking investments, while supporting grain marketing and supply–demand-balanced regions in resource-seeking investments to diversify import channels; particularly, Belt and Road provinces should leverage their locational advantages to transform them into supply chain advantages. Finally, the establishment of interprovincial agricultural investment–trade coordination mechanisms is recommended, fostering information sharing, policy coordination, and enterprise cooperation across regions, thereby reinforcing positive spillovers, avoiding negative competition, and achieving more optimal allocation of the benefits of agricultural outward investment on a broader scale.