Efficacy of Social Networks in Promoting the Green Production Behaviors of Chinese Farmers: An Empirical Study

Abstract

The green production practices of farmers are essential for sustainable agricultural development. However, studies have mostly overlooked the social factors affecting farmers’ decisions regarding green production behaviors (GPB). Furthermore, the pathways and mechanisms through which social networks modify these behaviors have not been fully validated. Therefore, by examining 1203 farmers from China’s main grain-producing regions, this study aims to empirically investigate both the direct and indirect impacts of social networks on farmers’ GPB, thereby furthering relevant research. First, family social networks in rural areas markedly enhanced farmer engagement in GPB. After assessing the endogeneity issues associated with farmers’ self-selection behaviors using propensity score matching, this effect was found to substantially persist. Analysis of the indirect impact revealed that social networks primarily facilitated farmers’ adoption of green production through channels such as information acquisition, interactive learning, and service support. Second, heterogeneity analyses based on generational differences and crop types demonstrated a distinct, promotional impact of social networks on both “middle-generation” and “older-generation” farmers. Moreover, a comparison between grain crop farmers and cash crop farmers determined the more substantial influence of social networks on encouraging grain crop farmer GPB. Overall, this study emphasizes that rural China’s social networks, especially clan-based ones, can successfully nurture agricultural sustainability by accelerating the propagation of green technologies while offering suitable environments for elderly farmers to “learn by observing” and “learn by doing”. Relevant departments should pay attention to and make full use of farmers’ social network relations in the process of promoting farmers’ adoption of green production behavior and further promoting the green development of agriculture.

1. Introduction

The sustainable utilization of agricultural ecological resources is essential for guaranteeing food security [1]. The second objective of the 2030 Sustainable Development Goals is “to eradicate hunger, achieve food security, and promote sustainable agriculture”, aiming to deliver nutritious food for all through agriculture while reinforcing people-centered rural progress and environmental protection [2]. The surge in fertilizer input and the embracing of modern technology globally have resulted in stable growth in crop yields and enhanced food security [3]. Nevertheless, the yield-oriented agricultural production model, predominantly the excessive application of agrochemicals, such as fertilizers and pesticides, has created problems such as soil degradation and surface pollution. Although this model has efficaciously addressed food security concerns for over 1.4 billion people in China [4], the incessant overuse of agrochemicals has prompted acute resource and environmental crises, not to mention amplified greenhouse gas emissions. Thus, comprehending how to thwart the damage to agricultural production volume caused by the overuse of agricultural chemicals and encourage the green production of agriculture is critical to attaining sustainable development in Chinese agriculture [5].

As defined by the United Nations Environment Programme (UNEP) in 2011, agricultural green production is generally regarded as a sustainable development model, including diversified production techniques such as less-till or no-till technology, the application of organic fertilizers and organic pesticides, soil protection measures, and the utilization of waste resources. According to the production stage of agriculture, this paper describes the green production behavior of farmers as the adoption of green production technology, such as high-efficiency, low-toxicity pesticide use technology, organic fertilizer application technology, straw return technology, and so on. Currently, the top priority of China’s agricultural green transformation is whether farmers can “green” their production behavior, for they are the real responders and decision-makers in agricultural green production [6]. Theoretically, agricultural green production technology can augment agricultural production efficiency to some extent, which is advantageous for farmers to attain stable income growth [7]. The satisfactory ecological environment generated by green agricultural production is an archetypal public good with palpable positive externalities. Nevertheless, the profit-maximizing “rational smallholder” [8] who adopts low-carbon or green production practices is not only inhibited by unified technologies and high-cost inputs but also has trouble in procuring high economic benefits in a short period [9]. The revolution of agricultural green production has encountered severe restrictions on production factor endowments [10,11], and farmers’ green production faces the dilemma of “strong willingness but weak behavior” [4]. Hence, this research addresses the key issue of how to mitigate the abovementioned challenges and encourage farmers to execute green production behavior (GPB) decisions.

Farmers’ GPB has long been a key research topic. Some prior studies have mainly focused on the proposition of whether farmers are rational economic agents, considering the driving forces behind green production among farmers [8], their inclination and behavior toward embracing green technologies [12], and influencing factors [13]. Recent studies have broadly suggested that external incentives created by agricultural environmental policies, including technical training and economic subsidies, can alter farmers’ production methods and endorse the dissemination of green production technologies by affecting their psychological processes, including green production cognition, motivation, and willingness. Pertinent studies have demonstrated that farmers who underwent technical training were more susceptible to implementing green production [14], and the regularity of training exerts a noteworthy positive impact on the execution effect [5]. Furthermore, government-provided technical training and more effective guidance enhance farmers’ green perceptions, which, in turn, influence farmers’ GPB [15]. Conversely, from the standpoint of farmers as “economic agents”, suitable economic subsidies can successfully decrease farmers’ investment in green production. Stimulated by economic rationality, farmers will increase their incentive to embrace green production techniques [16]. Moreover, some studies have reported the impact of agricultural socialized services on farmers’ adoption of green technologies. Service providers guide small-scale farmers to engage in green production practices by providing green services, such as pest and disease prevention and soil testing for formula-based fertilization, along with deep tillage and subsoiling operations, in their agricultural production methods [6]. In practical agricultural production, however, the efficacy of these policies has been substandard. Some studies have emphasized that technical training and government subsidies exert only a limited impact on farmers’ overuse of fertilizers or pesticides. Although agricultural subsidies can decrease farmers’ production costs and augment their inclination to use environment-friendly inputs, their impact on farmers’ inclination to undergo ecological modification remains trivial [5]. Despite numerous green subsidy policies in China, the usage of fertilizers, pesticides, and agricultural films remains markedly higher than the global average [17], and farmers’ implementation of straw-returning technology lacks sustainability [18]. The large investment of subsidy funds has not successfully eased the green transformation of farmers’ production methods. Hence, the journey toward green agriculture in China is still long and challenging.

The basis of most existing studies is that farmers are “economic agents”, overlooking the social factors that affect their decision-making regarding GPB. From the standpoint of farmers as “social agents”, GPB is not only an economic behavior but also a social behavior affected by the external environment and is a type of decision-making performance, having both social and economic features [19]. In the study of urbanized communities, Jacobs [20] applied neighborhood networks as social capital and first proposed the concept of social networks. Mitchell [21] claimed that besides direct relationships between individuals, indirect relationships can also be developed by exchanging and sharing material resources in the external environment. Social networks (SN) encompass the sum of all formal and informal relationships developed between individuals [22]. China is a country with entwined blood, kinship, geography, business relationships, and intricate social networks; thus, affecting people’s behavioral orientation based on social relationships has been an integral part of China since ancient times [23]. Investigating the immersion of farmers’ GPB into relational societies aligns more with the actual state in rural China. Moreover, China is an agricultural highflyer with a long history of farming. The planting culture nurtured by farming practices and the local culture based on blood and relationships not only molds informal social norms but also affects individual behaviors through personal predilections [24]. Farmers are positioned in an explicit social environment; thus, regarding technology implementation, they often gather technology information and regulate their income prospects through mutual experience exchange and sharing and then alter their production behavior intent, leading to steady production behavior [25,26]. In fact, some studies have rigorously examined the influence of social networks on the uptake of eco-friendly technologies by farmers, including practices like conservation tillage [27] and crop cultivation techniques [28]. Importantly, vital resources—such as finance, technology, and information—are entrenched within social networks and disseminated through them, thereby markedly enabling the diffusion of green technologies in agriculture. Through a review of the above literature, we found the following problems: Can social networks play a vital role in the green transformation of agricultural production? How are the direct and indirect impacts of SN displayed in farmers’ GPB? Considering generational differences among farmers and variations across crops, what differential effects do social networks exert on their implementation of green practices? Of note, addressing the above questions will hold considerable theoretical significance and practical value in evolving the sustainable growth of Chinese agriculture.

This study furthers the existing literature and practice in three crucial ways. First, our study provides a rigorous quantitative assessment about the influence of rural social networks on farmers’ GPB—an area that warrants further investigation. Current research either considers the impact of household SN on the adoption of a single technology by farmers [27,28] or it examines the adoption of green technologies by farmers through a single indicator of social networks [29], neither of which can fully reflect the extent of farmers’ efforts in engaging in green production. Therefore, this study uses the extent to which farmers adopt such technologies as a more nuanced alternative variable. Secondly, this study advances the current literature by investigating how social networks affect farmers’ GPB. According to existing research [30,31], Agricultural Productive Services (APS) that integrate green input elements (e.g., biomass pesticides, fertilizers, and agricultural film) and green technology components (e.g., soil testing with formula fertilization techniques, eco-friendly pest control methods, and crop residue—returned farming system) have arisen as vital mechanisms for enabling the green transformation of farming methods [18]. Hence, furthering the existing literature that underpins information acquisition mechanisms and demonstration learning mechanisms, this study proposes that the impact of SN on farmers’ GPB is also arbitrated through a service support mechanism. Exploring these potential pathways can guide the Chinese government in refining pertinent environmental policies aimed at endorsing sustainable farming practices. Thirdly, this study aims to explore the impact of social networks on farmers’ green production behavior, and most of the empirical part is to verify the impact of social networks on farmers’ green production behavior and verify the impact path. The inclusion of all types of farmers in the research framework can better ensure the universality of the conclusion. Of course, we also took into account that the social networks of different types of farmers may have different impacts. Therefore, for the sake of research rigor, we further classified farmers in Section 4.5 based on the existing literature. The intergenerational theory suggests substantial disparities in values, preferences, attitudes, and behaviors across different generational cohorts owing to differences in birth years and upbringing contexts [32]. Conversely, farmers’ GPB also differ with different crops. Considering these two dimensions of difference, this study expands benchmark research by expounding how social networks clearly influence farmers’ GPB through lenses of generational diversity and crop variation.

2. Theoretical Analysis and Research Hypotheses

2.1. Direct Impact of Social Networks on Farmers’ GPB

Granovetter [33] proposed the theory of social network embeddedness, postulating that economic behavior exists within social networks and structures, and social relationships are the foremost cause of trust in economic life; thus, the trust serves as an embedded network mechanism. In addition, Granovetter [33] emphasized the significance of individual social networks on economic behavior, signifying that these relationships are integral for analyzing economic conduct. However, the problems are “under-socialization” in neoclassical economics in the analysis of economic behavior and “over-socialization” in sociological research [19]. The theory of embeddedness proficiently combines the “economic man assumption” and the “social man assumption”, founding connections with the organizational theory. Notably, the organizational theory suggests that individuals do not operate entirely independently when making decisions; rather, their entrenched social networks exert a definite impact on their behavioral choices [34,35]. In Chinese rural society, which is deeply organized around clan groups, interpersonal relationships form an intricate web mainly revolving around kinship ties [36]. Within rural China’s typical geography and kinship ties, social networks markedly influence decision-making processes associated with household production and daily life [37]. Farmers’ GPB is evident within specific village settings where their choices are intrinsically connected with their relational networks and affected by multiple embedding factors. Social networks can be measured through both quantitative and qualitative dimensions, namely, network size (weak ties) and network strength (strong ties), each affecting individuals through mechanisms like weak tie information transmission [38] and strong tie value norm guidance [39], thereby molding members’ behavioral propensities and increasing the prospects of embracing green agricultural methods.

From the network size (weak ties) perspective, the theory of embeddedness proposes that weak ties can deconstruct informational blockades by linking incongruent resources across groups—efficiently elevating individuals’ access to diverse information from other communities. Compared with strong ties, weak ties convey more nonrepetitive information and play a bigger role in information dissemination [38]. In practice, farmers can observe and emulate the behavior of other farmers or connect with other farmers through social networks, which can decrease the information asymmetry in production decision-making (e.g., decision-making on the usage of fertilizers and pesticides), help them comprehend the swift and effective transfer of knowledge on green agricultural production, and amplify the likelihood of farmers decreasing the usage of fertilizers and pesticides, along with other green farming practices [40].

From the network strength (strong ties) perspective, farmers who have been in similar living settings and cultures for a long time are more likely to embrace information from their surroundings based on trust; that is, farmers can largely decrease the cost of searching for information and act as per others’ advice or behavior. The embeddedness of strong ties and high trust in social networks exerts a robust impact on farmers’ GPB, and based on mutual trust, farmers are more cognizant of sharing resources and information. Strong ties with strong connections between farmers illustrate trust and cohesion. More importantly, farmers’ green technology adoption behavior displays remarkable convergence and consistency under the potential role of social networks, which is depicted in the collective action of “if you adopt it, I will also adopt it”, thereby prompting the “social multiplier effect” within the social networks [41].

Overall, social networks serve as conduits for information, and both strong and weak ties play crucial roles in disseminating knowledge about sustainable farming practices, which, in turn, can compensate for farmers’ lack of experience and know-how regarding eco-friendly production methods. Hence, the following hypothesis is proposed:

H1: 

Social networks are beneficial for promoting green production among farmers, and the higher the level of farmers’ social networks, the more significant the adoption of green production technologies by farmers.

2.2. Indirect Impact of Social Networks on Farmers’ GPB

Social networks display features of short pathways and high density in information dissemination, which are pivotal in the rapid and extensive spread of sustainable farming practices [26]. Thus, farmers’ decision-making about GPB tends to exhibit homogeneity and consistency through social networks. The existing literature suggests that information acquisition and social imitation are the primary mechanisms through which social networks influence farmers’ green production practices [42,43]. Moreover, considering China’s agricultural context, denoted by “big country, small farmers”, along with a collective ownership system for farmland, rural areas function under household contracts for land management. As central units of agricultural production, farmers display considerable “path dependence” toward conventional farming practices. The conversion of conventional agriculture pivots on the introduction of modern production factors to enable its transition and upgrading [8]. Practically, as agricultural production processes become increasingly divisible, agricultural productive service organizations can alleviate restraints associated with labor, technology, and capital by assimilating green input factors and technological innovations into their operations, thereby reassuring farmers to embrace sustainable practices [4]. Accordingly, this study recognizes three potential mechanisms through which social networks affect farmers’ GPB: information acquisition mechanism, demonstration learning mechanism, and service support mechanism.

First, social networks affect farmers’ GPB via the information acquisition mechanism. Information acquisition is a vital factor in facilitating farmers to accept modern agricultural technologies. Social networks are essential in enabling information dissemination and provide indispensable support for farmers seeking knowledge about green production technologies [44]. The larger the size of a farmer’s social networks, the greater the number of farmers with different production practices that can be reached, thereby augmenting their access to diverse and modern technological insights, which, in turn, nurtures environmental cognizance and polishes their skills in sustainable production practices. Conversely, previous studies have indicated that partaking in technical training certainly influenced farmers’ adoption of environment-friendly agricultural technologies [40]. As a direct avenue for comprehending cutting-edge agricultural innovations, agricultural technology training can not only successfully decrease the cost of technology learning and fulfill the demand for green production information but also accelerate the circulation of technical and market information, markedly shorten the searching process for technical information for farmers, and considerably decrease the cost of information. Thus, when farmers have robust social network connections, the knowledge spillover due to the information acquisition mechanism could result in the convergence of GPB among farmer groups. Hence, we propose the second hypothesis:

H2: 

Social networks exert a positive impact on farmers’ GPB through the information acquisition mechanism.

Second, social networks affect farmers’ GPB through the demonstration learning mechanism. The intricacy and long cycle of agricultural production regulate that the learning of green production technology is manifested by long-term, process, and dynamic attributes, and its application effect is usually ambiguous [4]. When farmers acquire more common information than private information through social networks, they tend to learn or imitate others’ green production technology adoption behaviors, encouraging continuous, homogeneous, and large-scale collective actions, that is, “behavioral convergence”. Moreover, the “free-riding” behavior of green production technology is common in rural society. Farmers facing the ambiguity of gains and losses brought by high-risk modern technology typically first observe the effects of others before deciding their own behaviors to evade related risks [37]. Indeed, with the assistance of social networks, technical exchanges among farmers can also help them accrue technical knowledge, which is favorable for the promotion and application of technology. Furthermore, social networks can offer farmers the material capital and financial support required to embrace green production technologies and hone their risk-taking aptitude [31]. Hence, we propose the third hypothesis:

H3: 

Social networks exert a positive impact on farmers’ GPB through the demonstration learning mechanism.

Third, social networks affect farmers’ GPB through the service support mechanism. To some extent, farmers’ GPB depends on household labor endowment or the cost consumption of green production technologies [4]. APS has played a key role in advancing China’s agricultural modernization [7,45], defined as the provision of the service supply of materials, capital, technology, and machinery for agricultural production, the crux of which is the division and specialization of agricultural production processes [46]. For instance, agricultural mechanization itself is one of the most efficacious technologies, particularly following the Green Revolution in developing countries [47]. In countries like China [48], Bangladesh [49], and Myanmar [50], this technological dissemination is characterized by the extensive implementation of agricultural mechanization. Depending on their situations, farmers have the option of purchasing machinery or purchasing machinery services, that is, paying for a blend of professional labor and mechanized production to service providers (e.g., machinery owners, family farms, professional cooperatives, or agribusinesses). In turn, this leads to labor substitution, increased yields, and higher farm incomes. Furthermore, agricultural production services can stimulate farmers to accept green production technologies through demonstration, scale, spillover, and feedback effects. Hence, we propose the fourth hypothesis:

H4: 

Social networks exert a positive impact on farmers’ GPB through the service support mechanism.

Based on the above-described theoretical analysis, Figure 1 shows the logical framework of this study.

3. Methodology

3.1. Data Collection

The dataset utilized in this study originates from the China Land Economy Survey (CLES) conducted in 2022 and published by Nanjing Agricultural University in 2023. This comprehensive dataset encompasses detailed data on agricultural households, their land utilization, specific land plot details, and the production and operational activities of these households. The initial survey, which served as a reference point, was conducted in 2020 and encompassed 13 cities in Jiangsu province. Only crop-producing households were surveyed in this study. According to the research needs, we used 1203 valid data (n = 1203).

Based on data from the National Bureau of Statistics of China, we plotted Figure 2 and Figure 3 to more intuitively reflect the demand for green production in the study region. In the nearly 20 years of data, some data are updated to 2021, and some data are updated to 2023. As can be seen from the figure, the total agricultural GDP of Jiangsu Province has increased rapidly and has reached nearly 500 billion yuan, which shows that the proportion of agricultural GDP of Jiangsu Province in China’s agricultural GDP is relatively high. However, after calculating the carbon emissions from agricultural production in Jiangsu Province according to the carbon emissions from agricultural production in Geng’s study (2024), we found that the carbon emissions from agricultural production in this region exceeded 45 million tons during 2005–2021, and there was no obvious downward trend. In addition, in recent years, the sown area of crops in Jiangsu Province accounted for only about 4.5% of the total sown area of crops in China, but the use of fertilizers and pesticides accounted for more than 5%. At the same time, from the trend point of view, the proportion of fertilizer and pesticide use has shown an upward trend in recent years after a slight decline. To some extent, this indicates that the Jiangsu provincial government will face great pressure in the prevention and control of agricultural non-point source pollution. In addition, Jiangsu Province, as a large agricultural province in China, is also facing the situation of having a large population, small land, a shortage of land resources, and limited agricultural development resources. Ecological and environmental pollution caused by agricultural production must be properly handled. All in all, there is an urgent need to promote green agricultural production in the region.

3.2. Variables

3.2.1. Dependent Variable

Definition and quantification of green production behaviors (GPB) of farmers. According to the definition of the concept of green production behavior in existing studies, it mainly refers to the agricultural production process in which farmers adopt production behaviors that are conducive to maintaining or enhancing agricultural production capacity and promoting sustainable development of the agroecological environment. Based on the theoretical analysis part, this study introduces the adoption of eight technical measures, namely, using good seed, soil formula fertilization, crop cultivation management, green pest control technology (such as the use of drones to spray pesticides), the use of high-efficiency low-toxic pesticides, deep ploughing of the cultivated land, water-saving irrigation, and crop residue return to the field to characterize the green production behaviors of farmers.

3.2.2. Independent Variable

Definition and quantification of social networks (SN). The social network in this study refers to the farm household family social network, and there have been measurement differences between the single-indicator method [29] and the multi-indicator comprehensive measurement method [51] in previous studies. Therefore, drawing on the research results of Lv [52] and combining the multidimensional characteristics of blood, kinship, and geography in Chinese rural social networks, we synthesized the social networks in terms of network size and network strength (

Table 1. Social Network Measurement Indicator System.

Type of Variable	Variable Definition	Variable Assignment	Property
network strength	trust in relatives (S1)	very distrustful = 1; rather distrustful = 2; generally trusting = 3; rather trusting = 4; very trusting = 5	+
	trust in neighbors (S2)		+
	trust in village cadres (S3)		+
network size	the number of cell phone contacts (S4)	numbers	+
	the number of people who can borrow 50,000 RMB when they are in trouble (S5)	numbers	+
	the number of usual participation in cultural activities (S6)	number of activities	+

Note: SN = S₁ × 0.0088 + S₂ × 0.0090 + S₃ × 0.0091 + S₄ × 0.2503 + S₅ × 0.3275 + S₆ × 0.3953. + means positive correlation

), and applied the entropy method to measure the level of social networks of farm households. Specifically, the network strength is characterized by “trust in relatives” (S₁), “trust in neighbors” (S₂), and “trust in village cadres” (S₃); the network size is characterized by “the number of cell phone contacts” (S₄), “the number of people who can borrow 50,000 RMB when they are in trouble” (S₅), and “how many daily cultural activities do you participate in?” (S₆). Then, these six indicators are dimensionlessly processed; the entropy method calculates the weights of each indicator; and finally, the comprehensive score of the social network is calculated by weighting.

3.2.3. Control Variables

The determinants of farmers’ green production behavior (GPB) are intricate, with their decisions to adopt green production practices being influenced by a multitude of factors, including the characteristics of the household head, family, land endowment, and village development. Drawing on previous studies, we considered 4 dimensions and 14 variables to reflect the above factors (

Table 2. Descriptive statistics of the variables.

Type of Variable		Variable	Variable Definition and Assignment	Mean	Std	Min	Max
Dependent variable		green production behaviors (GPB)	Number of green production technologies adopted/0~8	0.983	1.274	0	8
Independent variable		social networks (SN)	Composite score obtained by entropy method	0.049	0.053	0.000 6338	0.561 9605
Mediation variables		Technical training (TEC)	Whether any member of the household is educated or trained in agricultural technology/YES = 1; NO = 0	0.392	0.488	0	1
		Co-operative farming (COP)	whether to join cooperatives or agribusinesses /YES = 1; NO = 0	0.038	0.192	0	1
		Outsourcing service (SER)	Whether the production process is outsourced/YES = 1; NO = 0	0.420	0.494	0	1
Control variables	household head characteristics	AGE	one full year of life	62.232	11.462	17	96
		Health (HEL)	incapacitated = 1; poor = 2; fair = 3; good = 4; Excellent = 5	3.952	1.073	0	5
	farm household characteristics	Non-agricultural vocational training (NVT)	YES = 1; NO = 0	0.191	0.393	0	1
		Agricultural labor (AGL)	numbers	1.486	1.018	0	6
		Non-farm income (NIC)	Household non-farm income/total income in 2021	0.570	0.422	0	1
		Entrepreneur (EPN)	YES = 1; NO = 0	0.084	0.277	0	1
		Agricultural insurance (ISR)	YES = 1; NO = 0	0.249	0.433	0	1
	land characteristics	Scale (SCL)	Mu(unit of area equal to one-fifteenth of a hectare)	5.422	19.966	0	400
		Farmland fertility (FER)	Poor = 1; fair = 2; good = 3	2.363	0.624	1	3
		Farmland property right (FPR)	YES = 1; NO = 0	0.879	0.327	0	1
		Distance from the hardened road (DIS)	mile	0.455	1.061	0	20
		Farmland restoration (RST)	YES = 1; NO = 0	0.112	0.316	0	1
		Irrigation convenience (IRG)	YES = 1; NO = 0	0.848	0.359	0	1
	Village characteristics	Rural industry (RID)	YES = 1; NO = 0	0.176	0.381	0	1

). Specifically, they include: first, household head characteristics. The household leader plays a pivotal role in making decisions regarding sustainable farming practices. Their age (AGE) and health status (HEL) are significant indicators of their propensity to take risks when embracing green production technologies and reflect their degree of environmental consciousness [6]. The second is the characteristics of the farm household. In the rural areas of China, the family constitutes the fundamental unit for farming activities, and the non-agricultural vocational training (NVT) of family members, the workforce allocated to agricultural tasks (AGL), the percentage of non-farm income (NIC), whether they are entrepreneurs (EPN), whether they purchase agricultural insurance (ISR), and other family characteristics influence green production behavior decision-making in terms of technical information acquisition ability [4], green factor input structure [6], and economic cognition [11]. Third, land endowment. The land is a factor input carrier, and the characteristics of special land endowments directly determine the degree of demand for green technology. Farmland scale (SCL), farmland fertility (FER), farmland property right (FPR), farmland restoration (RST), irrigation convenience (IRG), and the distance from the hardened road (DIS) reflect land characteristics from the perspective of quantity and quality of cropland, respectively, and studies have shown that poorer land fertility induces farmers to buy large quantities of cheap fertilizers to improve the degree of fertility of the land, limiting the adoption of green technologies by farmers [6], and that the accessibility to irrigation and distance from hardened roads directly affect the supply of agricultural mechanization services [5]. Fourth, village characteristics. Rural industry (RID) determines the level of economic development of the village, and the information acquisition of green production technology of enterprise farmers is closely related to the level of economic development of the village where they are located, and rural enterprises and Internet agricultural information can provide technical information support for farmers [17].

3.2.4. Mediation Variables

As described in the theoretical framework section above, the social networks influence farmers’ GPB mainly through the information acquisition mechanism, the demonstration learning mechanism, and the service support mechanism. In addition, we distinguish these intermediate variables to obtain detailed test results. First, it has been shown that technical training and technical guidance provided by the government tend to be more likely to enhance farmers’ green cognitive ability, which in turn affects farmers’ green production behavior [15]. Engagement in technical training by farmers can reduce the distance over which information about green production technologies is disseminated; therefore, this paper constructs the “whether any member of the household is educated or trained in agricultural technology” (TEC) variable as a proxy variable for the information acquisition mechanism. Second, in rural China, professional cooperatives are organizations formed by farmers on their own initiative, and they are a bridge between farmers and the market. It (or agribusinesses) can provide farmers with the material capital and financial support needed to adopt green production technologies, improve their risk-taking ability, etc. [31], and at the same time, it can provide a place for members within the cooperative to interact and learn. Farmers’ green production behaviors have strong convergence and consistency in close-knit social groups due to the normative pressure within cooperatives (or agribusinesses) [37]. Therefore, in this paper, “whether to join cooperatives or agribusinesses” (COP) is selected as a proxy variable for the demonstration learning mechanism. Finally, there is an emerging literature that links the role of agricultural production services in green agricultural production and promotes green production by changing the factor input structure of capital, labor, and green technology [31], and selects “Whether the production process is outsourced” (SER) as a proxy variable for the service support mechanism. The specific values are shown in Table 2.

3.3. Model

3.3.1. The Direct Impact Regression Model

The ordered probit model was employed in this study. The dependent variable, green production behavior (GPB), was quantified as a categorical ordinal outcome, reflecting the count of green production techniques implemented by farmers, which ranges from 0 to 8. This variable is ordinal with a fixed sequence. Following the methodology used by Yang [53], the Ordered Probit Model was deemed appropriate for the empirical examination and was formulated as described below.

$${\mathrm{GPB}}_{\mathrm{i}}^{*}={\alpha }_{\mathrm{i}}+\alpha {\mathrm{SN}}_{\mathrm{i}}+\delta {\mathrm{C}}_{\mathrm{i}}+{\epsilon }_{\mathrm{i}}$$

(1)

$${\mathrm{GPB}}_{\mathrm{i}}=\left\{\begin{array}{c}0, \mathrm{i}\mathrm{f} {\mathrm{GPB}}_{\mathrm{i}}^{*}\le {\mathrm{r}}_0\hfill \\ 1, \mathrm{i}\mathrm{f} {\mathrm{r}}_0 {<\mathrm{GPB}}_{\mathrm{i}}^{*}\le { \mathrm{r}}_1\hfill \\ 2, \mathrm{i}\mathrm{f} {\mathrm{r}}_1<{\mathrm{GPB}}_{\mathrm{i}}^{*}\le { \mathrm{r}}_2\hfill \\ 3, \mathrm{i}\mathrm{f} {\mathrm{r}}_2<{\mathrm{GPB}}_{\mathrm{i}}^{*}\le {\mathrm{r}}_3\hfill \\ 4, \mathrm{i}\mathrm{f} {\mathrm{r}}_3<{\mathrm{GPB}}_{\mathrm{i}}^{*}\le {\mathrm{r}}_4\hfill \\ 5, \mathrm{i}\mathrm{f} {\mathrm{r}}_4<{\mathrm{GPB}}_{\mathrm{i}}^{*}\le {\mathrm{r}}_5\hfill \\ 6, \mathrm{i}\mathrm{f} {\mathrm{r}}_5<{\mathrm{GPB}}_{\mathrm{i}}^{*}\le {\mathrm{r}}_6\hfill \\ 7, \mathrm{i}\mathrm{f} {\mathrm{r}}_6<{\mathrm{GPB}}_{\mathrm{i}}^{*}\le {\mathrm{r}}_7\hfill \\ 8, \mathrm{i}\mathrm{f} {\mathrm{GPB}}_{\mathrm{i}}^{*}>{\mathrm{r}}_7\hfill \end{array}\right.$$

(2)

In Equation (1), ${\mathrm{GPB}}_{\mathrm{i}}^{*}$ is the latent variable of green production behavior of farmer i, SN represents the social networks, C represents the control variables, $\alpha$ and $\delta$ are the coefficients to be estimated, and $\epsilon$ is the random perturbation term, i. In Equation (2), ${\mathrm{r}}_0$–${\mathrm{r}}_7$ are the unknown split points of the quantity of green production technology adoption by farmers, and ${\mathrm{r}}_0$<${\mathrm{r}}_1$<……< ${\mathrm{r}}_7$.

3.3.2. The Indirect Impact Regression Model

To delve deeper into the causal connections and mechanisms through which social networks and intermediary variables impact farmers’ engagement in green production practices, this study adopts Taylor’s sequential mediation model for regression analysis [54]. Theoretical insights suggest that social networks shape farmers’ green production actions via several pathways: the mechanism of information gathering, the process of observational learning, and the provision of supportive services. The mediation model is structured as follows:

$${\mathrm{GPB}}_{\mathrm{i}}^{*}={\alpha }_{\mathrm{i}}{+\alpha }_1{\mathrm{SN}}_{\mathrm{i}}+{\delta }_1{\mathrm{C}}_{\mathrm{i}}+{\epsilon }_{\mathrm{i}}$$

(3)

$${\mathrm{M}}_{\mathrm{i}}={\gamma }_{\mathrm{i}}+ϵ{\mathrm{SN}}_{\mathrm{i}}+{\delta }_2{\mathrm{C}}_{\mathrm{i}}+{\epsilon }_{\mathrm{i}}$$

(4)

$${\mathrm{GPB}}_{\mathrm{i}}^{*}={\alpha }_{\mathrm{i}}{+\alpha }_2{\mathrm{SN}}_{\mathrm{i}}+\beta {\mathrm{M}}_{\mathrm{i}}+{\delta }_3{\mathrm{C}}_{\mathrm{i}}+{\epsilon }_{\mathrm{i}}$$

(5)

In the above formula, ${\mathrm{M}}_{\mathrm{i}}$ denotes the Intermediate variables, and, in this study, TEC, COP, and SER were used as the Intermediate variables, respectively. Other variables are explained as in Equation (1). When the coefficients ${\alpha }_1$, $\epsilon$, and $\beta$ are significant, there is a mediation effect. If ${\alpha }_2$ is also significant, the mediation effect is partial; otherwise, it is a full mediation effect.

4. Empirical Estimation Results

4.1. The Direct Impact of Social Networks on Farmers’ GPB

In this section, we used Stata 17.0 to analyze the impact of SN on farmers’ GPB using the Ordered Probit Model. The results of the model estimation are presented in

Table 3. The results of the impact of social networks on farmers’ green production behavior.

Variable	GPB
	Ordered Probit		Average Marginal Effect
	(1)	(2)	SN_Predict	dy/dx	Delta-Method Std. Err
SN	1.8381 *** (0.6599)	1.9287 *** (0.6618)	0	−0.7261 ***	0.2472
AGE	—	−0.0010 (0.0032)	1	0.1262 ***	0.0451
HEL	—	0.0184 (0.0315)	2	0.3139 ***	0.0975
NVT	—	0.0125 (0.0869)	3	0.1194 ***	0.0439
AGL	—	0.2553 *** (0.0333)	4	0.0617 ***	0.0246
NIC	—	0.0415 (0.0818)	5	0.0365 **	0.0162
EPN	—	0.0416 (0.1267)	6	0.0424 **	0.0193
ISR	—	−0.1469 * (0.0783)	7	0.0126 *	0.0076
SCL	—	0.0067 *** (0.0019)	8	0.0133 *	0.0078
FER	—	0.0322 (0.0522)	—	—	—
FPR	—	0.2052 ** (0.1033)	—	—	—
DIS	—	0.0534 ** (0.0256)	—	—	—
RST	—	0.0327 (0.0978)	—	—	—
IRG	—	0.0018 (0.0934)	—	—	—
RID	—	−0.2556 *** (0.0885)	—	—	—
Pseudo R²	0.0030	0.0375	—	—	—

Note: * p < 0.10, ** p < 0.05, *** p < 0.01; Values in parentheses are robust std.err.

. Notably, the Pseudo R² value for model (2), which includes additional control variables, is considerably higher than that of model (1), suggesting a more robust explanatory power. Consequently, the analysis primarily concentrates on the outcomes of model (2).

This paper focuses on exploring the link between farmers’ SN and their GBP. The findings from model (2) in Table 3 demonstrate that SN positively influence GPB, with the result being statistically significant at the 1% level. This suggests that stronger social networks significantly encourage farmers to adopt green production methods, and the more extensive the social networks, the more pronounced the encouragement for their green production practices, thus confirming our Hypothesis 1. Given that GPB is a categorical ordinal variable, we followed [55] to calculate the marginal effects of each independent variable. The average marginal effects are detailed in Table 3. For SN’s impact on GPB, the average marginal effect is −0.7261 for GPB = 0, which is significant at the 1% level. For GPB = 1, it is 0.1262; for GPB = 2, it is 0.3139; for GPB = 3, it is 0.1194; for GPB = 4, it is 0.0617; for GPB = 5, it is 0.0365; for GPB = 6, it is 0.0424; for GPB = 7, it is 0.0126; and for GPB = 8, it is 0.0133. All these effects are positive and statistically significant. This indicates that social networks significantly discourage farmers from not adopting any green production technology and significantly encourage them to adopt at least one green production technology, reaffirming Hypothesis 1.

Regarding the control variables, the outcomes are generally consistent with prior scholarly research. From model (2) in Table 3, it can be seen that AGL and SCL are positively related to GPB, aligning with Geng et al.’s findings [4]. In terms of land characteristics, a more secure land property right (FPR) fosters an environment conducive to green production practices, and a shorter distance to markets (DIS) facilitates mechanized operations, echoing the conclusions drawn by Deng et al. [7]. The negative association of irrigation sources reliability (ISR) and the remoteness of irrigation facilities (RID) with GPB could be attributed to the upfront costs required for adopting green production technologies. The ‘crowding out effect’ stemming from the need to invest in planting insurance as a ‘bottom-up’ mechanism may deter farmers from undertaking GPB, especially when the costs are high.

4.2. Robustness Testing

To affirm the reliability of the empirical findings of the previous study, this study carried out a suite of robustness checks, including the following three aspects. These checks encompass three main areas: altering the regression approach, redefining the key independent variables, and incorporating additional control variables into the analysis.

First, we changed the regression method. In addition to the ordered probit model used in the baseline regression, this paper employed an OLS model to confirm the consistency of our results. These findings are presented in column (3) of

Table 4. The results of the robustness testing.

Variable	GPB
	Chang Method	Reassign Core Explanatory Variables			Add Control Variables
	(3)	(4)	(5)	(6)	(7)
SN	2.378 *** (0.8913)	—	—	—	1.8667 *** (0.6643)
CADRE	—	0.2793 *** (0.0876)		—	—
ln_CON	—	—	0.1129 *** (0.0257)	—	—
ACT	—	—	—	0.1638 *** (0.0585)	—
RISK	—	—	—	—	−0.1285 ** (0.0612)
TIME	—	—	—	—	0.0865 * (0.0488)
Control variables	YES
R2	0.0976	—	—	—	—
Pseudo R²	—	0.0379	0.0412	0.0372	0.0399

Note: * p < 0.10, ** p < 0.05, *** p < 0.01; Values in parentheses are robust std.err.

, where the OLS model yields significantly positive coefficients for social networks, aligning with our initial outcomes. Second, acknowledging the limitations of the single-indicator approach to measuring social networks (SN) as noted by [29], we selected alternative variables that could represent SN as the key independent variable for our regression analysis. Given the cultural context of rural China, where social standing and network strength are crucial for leadership roles, we substitute the core explanatory variable with an indicator of whether a household member has served as a village official (CADRE), coded as “1” for yes and “0” for no. In addition, we replaced new core explanatory variables, such as “number of mobile phone contacts” (CON) and “number of social activities” (ACT), which can reflect the SN to some extent. The results, displayed in columns (4), (5), and (6) of Table 4, are in line with the main findings from column (2) of Table 3, showing significant positive effects. Third, considering that agricultural green production involves high initial costs and deferred benefits, we recognize that farmers’ attitudes towards risk and time horizons could influence their GPB. Therefore, in this paper, two control variables have been added, which are the risk preference (RISK) (which kind of investment is preferred: 1 = risky investment, high risk, high return, and high loss; 2 = medium risk investment, medium risk, medium return, and medium loss; 3 = small risky investment, small risk, small return, and small loss) and time preference (TIME) (1 = I focus on the current return regardless of what happens in the future; 2 = I focus on both current and future return; 3 = I focus on both current and future return, not on the present), and column (7) of Table 4 provides further evidence of the robustness of the previous findings. Therefore, H1 of this paper is not only valid but also reliable.

4.3. Endogenous Discussion

Considering that whether farmers construct, develop, and maintain social networks is a kind of “self-selection” behavior, which is affected by a combination of factors, such as personal characteristics, family characteristics, risk preferences, etc., which affects the level of farmers’ social networks; the above estimation results may lead to the endogeneity of the model due to the self-selection bias of the samples. Therefore, the propensity score matching method (PSM) was used to divide the selected samples into farmers whose SN is greater than the mean as the experimental group (assigned a value of 1) and those whose SN is less than the mean as the control group (assigned a value of 0), and a counterfactual analysis framework was constructed, thus effectively eliminating the biased estimation of non-random distributions of the samples, as well as attempting to solve the problem of endogeneity between SN and GPB. Estimation was carried out by using four matching methods: k-nearest neighbor matching (KN), k-nearest neighbor matching with caliper (KNC), kernel matching (KE), and radius matching (RA).

Table 5. The results of different PSMs.

Method	Treated	Controls	ATT	S.E.	T-Stat
KN (n = 1)	1.24	0.9890	0.2509	0.1281	1.96 **
KN (n = 4)	1.24	0.9964	0.2436	0.1055	2.31 ***
KNC	1.24	0.9727	0.2673	0.1057	2.53 ***
KE	1.24	0.9495	0.2905	0.0965	3.01 ***
RA	1.24	0.9701	0.2699	0.0979	2.76 ***

Note: Caliper range selection, C = 0.01. ** p < 0.05, *** p < 0.01.

presents the Average Treatment Effect on the Treated (ATT) outcomes for each of these methods, indicating that all ATT estimates are statistically significant. It is observed that the group with a higher level of social networks (the treatment group) is approximately 26% more likely to engage in green production practices than the group with a lower level (the control group), which aligns with the findings of the previous study. In addition, the result of k-nearest neighbor matching with the caliper (n = 4) was tested using a common support domain test and a balance test. Figure 4 illustrates the propensity score density functions before and after matching. There is substantial overlap in the propensity score ranges between the treatment and control groups post matching, with the majority of observations falling within the shared value range. This confirms the reliability of the propensity score matching (PSM) estimation results and further substantiates the robustness of the findings presented in this paper. In conclusion, Hypothesis 1 is comprehensively validated.

4.4. Mechanism Testing

Taylor’s mediated effects model was selected for stepwise regression to validate the effect of social networks on farmers’ GPB [54]. Notably, for the mediation model in which the explanatory and mediator variables are categorical or hierarchical variables and the core explanatory variables are continuous variables, we cannot just implement the method of addressing the mediation effect of continuous variables and directly multiply the regression coefficients $ϵ$ and $\beta$ in Equations (4) and (5) to obtain the size of the mediation effect. Hence, we drew on the test method of Sun [56] to evaluate the significance of the mediation model test steps separately. The Sobel test p < 0.05 signifies that the mediation effect is established.

Table 6. Mediation effect test.

Variable	Information Acquisition		Demonstration Learning		Service Support
	TEC (8)	GPB (9)	COP (10)	GPB (11)	SER (12)	GPB (13)
SN	4.0306 *** (0.7952)	1.3630 ** (0.6685)	2.2638 ** (1.0055)	1.7955 *** (0.6405)	1.8882 *** (0.7321)	1.3823 * (0.7605)
TEC	—	0.4285 *** (0.0664)	—	—	—	—
COP	—	—	—	0.3903 ** (0.1689)	—	—
SER	—	—	—	—	—	1.4669 *** (0.0794)
Control variables	YES
Sobel	p = 0.000		p = 0.039		p = 0.014
Pseudo R2	0.0509	0.0496	0.1218	0.0392	0.0451	0.1668

COP, cooperatives or agribusinesses; GPB, green production behaviors; SER, whether the production process is outsourced; SN, social networks; TEC, trained in agricultural technology. * p < 0.10; ** p < 0.05; *** p < 0.01.

presents the results of the mechanism testing.

First, the information acquisition mechanism was validated. Table 6, column (8), presents the regression results with TEC, a proxy variable for the information acquisition mechanism, as the explanatory variable and SN as the core explanatory variable. Of note, SN is significantly positive at the 1% level, implying that farmers’ social networks can play a vital role in information dissemination and offer crucial support for farmers’ access to green production technology information [44], thereby promoting the participation of farmers in agricultural technology education or training. Meanwhile, Table 6, column (9), shows the results of regressing SN and TEC as explanatory variables, where SN remains significant at the 5% level, while TEC also passes the 1% significance level, with positive coefficients, and simultaneously passes the Sobel test for mediating effects at the 1% level, signifying a partial mediating role of receiving agricultural technical education and training. Hence, H2 is established.

Second, the demonstration learning mechanism was validated. Table 6, column (10), presents the regression results. The coefficient of COP, a proxy variable for the demonstration learning mechanism, is significantly positive, suggesting that social networks motivate farmers to join cooperatives to learn or emulate others’ green production technology adoption behaviors, as well as induce continuous, homogenized, and large-scale collective actions, that is, “behavioral convergence”. Table 6, column (11), uses SN and COP as explanatory variables in the regression; the coefficients of both exert a positive impact on GPB at the 1% significance level and simultaneously pass the Sobel test of mediation effect at the 1% level, suggesting that social networks can indirectly promote farmers’ GPB through the interactive learning mechanism. Hence, H3 is established.

Finally, the service support mechanism was validated. The estimation results in Table 6, column (12), illustrate that SN exerts a positive impact on SER and is significant at the 1% level, signifying that under the premise of China’s urbanization, ever more rural laborers are migrating to the cities to work, and those who stay in the countryside, in the absence of laborers, purchase services to mitigate the limitations of factors of production, such as labor, technology, capital, and so on. Table 6, column (13), demonstrates that after adding the mediating variable SER, SN and SER are significant at the 10% and 1% levels, respectively, and the coefficients are positive while passing the Sobel test. Hence, SER exerts a partial mediating effect in the promotion of farmers’ GPB by social networks, and endorsing the growth of APS is one of the mediating paths of social networks to promote farmers’ GPB.

4.5. Heterogeneity Analysis

In the 1950s, German sociologist Mannheim [32] proposed the intergenerational differences theory, theorizing that differences in birth age and developmental backgrounds inevitably cause differences among generations regarding values, behavioral preferences, and social network relationships. This intergenerational differentiation has become increasingly pronounced with the rapid growth of China’s society and economy, leading to heterogeneous impacts of social network relationships on the choice of agricultural production methods across different generational cohorts of farmers. Per the World Health Organization’s (2013) classification—designating people aged ≤44 years as “young”, those aged 45–59 years as “middle-aged”, and those aged ≥60 years as “older”—we categorized farmers into three groups: “new generation” (NEW), “middle generation” (MID), and “old generation” (OLD). Further analysis results, presented in

Table 7. Heterogeneity test results.

Variable	GPB
	Intergenerational Differences			Crop Differences
	NEW	MID	OLD	GRAIN	CASH
	(14)	(15)	(16)	(17)	(18)
SN	1.3636 (1.6774)	1.8243 ** (0.9211)	2.4681 *** (1.0444)	2.1650 *** (0.7645)	2.2192 * (1.1427)
Control variables	YES
n	96	343	764	881	421
Pseudo R²	0.1366	0.0712	0.0597	0.0508	0.0735

GPB, green production behaviors; SN, social networks. Note: The sum of n in (17) and (18) exceeds the total because some farmers grow both food crops and cash crops. * p < 0.10; ** p < 0.05; *** p < 0.01.

, columns (14)–(16), indicated that the promotional effect of SN on GPB was significant only within the MID and OLD cohorts. This phenomenon can be ascribed to multiple factors: first, with China’s economic advancement and urban–rural integration, conventional rural “social networks” manifested by interpersonal relationships are increasingly diluted by modern societal changes; therefore, values held by older generations are frequently dismissed by younger counterparts. In addition, from the mechanistic viewpoint discussed earlier, NEW farmers tend to be skillful at utilizing digital technologies, such as smartphones; therefore, compared with their MID or OLD peers, they rely less on information acquisition through social networks. Moreover, they can access timely updates about green agricultural technologies, as well as information about agricultural loans independently, which markedly decreases their dependence on social capital for these resources. Notably, >60% of our sample comprised OLD farmers, while <10% represented NEW farmers—a reflection of an intensifying aging trend in rural China where MID and OLD farmers constitute a large portion of agricultural producers who must leverage social networks to implement green production technologies successfully; this transition will facilitate China’s transition toward sustainable agriculture.

Considering the differences in crops, social networks exert different effects on the GPB of farmers growing different crops, which we further categorized into grain crops (GRAIN) and cash crops (CASH). Table 7, columns (17) and (18), demonstrates that for farmers growing GRAIN, the impact of SN on GBP is more significant, and its coefficient value of 2.1650 is significant at the 1% level, which also closely correlates with the “food security” policy promoted by China in recent years. For CASH farmers, the positive impact of SN on GPB also passed the 10% significance test, further indicating the positive impact of SN on promoting green production among farmers.

Table 8. Detailed types of food crops and cash crops.

GRAIN CROPS		CASH CROPS
Crop Name	Number	Crop Name	Number
Wheat	354	Soya bean	99
Corn	131	Potato	9
Rice	469	Cotton	5
		Oilseed rape	81
		Peanut	43
		Sesame	4
		Vegetable	120
		Pear	2
		Peach	10
		Grape	6
		Others	61
Total	954	Total	373

Note: The sum of grain crops and cash crops is different from (17) and (18) in Table 7 because some farmers grow multiple crops at the same time.

shows the types of grain crops and cash crops and the number of farming households.

5. Conclusions and Policy Implications

Social networks rooted in China’s rural culture exercise a remarkable impact on the decision-making processes related to farmers’ GPB. Nevertheless, limited research has addressed whether and how these social networks influence Chinese farmers’ GPB. Hence, this study empirically investigated both the direct and indirect impacts of social networks on farmers’ GPB using data from Jiangsu province, China, thereby expanding the existing literature. Social networks in rural areas markedly augment the GPB of farming households. The mechanism analysis suggests that three primary pathways—information acquisition, interactive learning, and service support—enable this promotion by social networks within rural communities. Furthermore, various robustness checks validate the reliability and validity of our results. Notably, PSM successfully addresses potential endogeneity issues originating from farmers’ self-selection biases. Next, heterogeneity analyses based on the intergenerational differences among farmers and crop differences illustrate that social networks markedly endorse green production, especially among the “middle generation” and “old generation”. Regarding crop categories, social networks exert a more pronounced impact on grain crop farmers than on cash crop farmers.

Considering the aforementioned results and the prevailing landscape of sustainable agricultural growth in China, this study proposes the following recommendations. First, it is imperative to determine the role of social networks in endorsing green production technologies. Predominantly during the initial stages of technology dissemination, leveraging social networks—specifically clan-based connections—can fast-track the acceptance rate of these technologies. This approach provides opportunities for an aging labor force to participate in “learn by observing” and “learn by doing”, thereby enhancing the efficiency of green production technology promotion. Second, it is crucial to consider the aging workforce as we advance toward a greener transformation in agricultural production; amplifying their participation rates in technical training through innovative methods will increase their human capital and create promising conditions for embracing green production technologies among older farmers. Third, assimilating modern machinery, financial resources, and green technological elements into agricultural practices through specialized, standardized, and intensive-scale services can successfully endorse sustainable farming [57]. Thus, government initiatives should prioritize fortifying multiple stakeholders within the agricultural production service market by nurturing a favorable institutional environment and offering financial support that drops entry barriers. This policy will facilitate smallholder farmers to fulfill their service requirements fully while ensuring that all types of farmers—especially smallholders—benefit from scale efficiencies and technological developments offered by agricultural services.

Furthermore, despite the aforementioned findings, this study has certain limitations. In addition, despite these findings, there are certain limitations to this study. This study is based on the survey data of Jiangsu Province, China. However, there are large interprovincial differences in China’s agricultural practices, so the social network is a comparative analysis of green production of farmers in different provinces. In the future, it is necessary to further expand the research area, obtain data at the national level for empirical comparative analysis, and provide a more comprehensive realistic basis for the green development of China’s agriculture.