Impact Effects of Cooperative Participation on the Adoption Behavior of Green Production Technologies by Cotton Farmers and the Driving Mechanisms

Abstract

Nudging the adoption of agricultural green production technologies (AGPTs) by cotton farmers is a practical need to implement the national “green development” strategy. Based on the micro-survey data of 502 cotton farmers, this paper empirically analyzed the influence and driving mechanism of cotton farmers’ participation in cooperatives on their adoption of green production technology from the perspective of their inner cognition and external regulation by using the propensity score matching (PSM) model and the intermediary effect model. The study found that (1) the importance of agricultural green production technology to cotton farmers was in the order of soil testing and formula fertilization technology, green prevention and control technology, deep tillage technology, water-saving irrigation technology, new variety technology, and straw-returning technology. (2) Participation in cooperatives can significantly improve the adoption of agricultural green production technologies by cotton farmers, with an increase of about 27.16%, and the improvement effect on technology-intensive production links is pronounced. (3) By improving the inner cognition and external regulation of cotton farmers, cooperatives can enhance the green endogenous power of cotton farmers, improve environmental external constraints, and become an intermediary path to guide cotton farmers to adopt agricultural green production technology.

1. Introduction

Green development of agriculture and green production transformation of farmers are the inevitable requirements for the modernization and development of China’s agriculture, as well as the necessary requirements for realizing the strategy of rural revitalization [1]. In recent years, the large amount of agricultural production materials such as pesticides and fertilizers has kept pushing the continued growth of crop production in China, but it has brought a series of resource and environmental problems, such as arable compaction, soil nutrient loss, water eutrophication, etc., which has posed severe challenges to China’s agricultural ecological environment [2]. Therefore, steadily promoting the green development of agriculture and constructing a green production and ecological environment in agriculture are the most important things to encourage farmers to actively adopt agricultural green production technology (AGPT) in agricultural production and operation [3], bringing more economic, environmental, and social benefits to the rural areas while solving the problem of agricultural nonpoint source pollution [4,5]. At present, it is difficult for farmers’ inherent production cognition and agricultural operation means to quickly transform traditional production methods into green production. It is urgent to improve farmers’ cognition of green production and open the “black box” of cognition [6], as it can enable farmers to realize the transformation of green production. Therefore, the International Cooperative Alliance (ICA) believes that when the environmental awareness of farmers is weak [7] and policies to promote the adoption of AGPTs by farmers are not as effective as they should be, it is possible to enhance the adoption of AGPTs by farmers through the development of “education, training, and information” through cooperatives [8]. This provides an important idea for this study: to protect the agricultural ecological environment, it is not only necessary to change the agricultural production mode [9] but it is also necessary for farmers to actively adopt AGPTs [10], aim to reduce the adverse impact on the environment in the agricultural production process, improve production efficiency, and protect the quality and safety of agricultural products. At the same time, the educational guidance of cooperatives [6] makes farmers realize the importance and advantages of AGPTs and promotes the transformation of farmers’ production and operation into greener production methods [11], thus promoting the development of Chinese agriculture towards the goal of greening.

In recent years of research on cooperatives and farmers’ green production behaviors, scholars have mentioned the most important aspect: first, cooperatives can directly promote farmers’ green production behavior. Theoretically, an important condition for farmers to adopt AGPTs is that the marginal private benefit is greater than or equal to the cost and that the AGPT is a positive externality technology; therefore, the social benefit of adopting AGPTs is greater than the private benefit [12]. However, under the “rational economic man” process, small farmers are more inclined to pursue private gains, and social benefits do not provide farmers with a natural incentive to adopt green technologies [13]. The institutional arrangement of cooperatives has some advantages in addressing this problem. In China, there are a large number of smallholder cooperatives that can effectively promote the adoption of green technologies by increasing the efficiency of internalizing the external benefits of green agricultural technology adoption [14]. In particular, Huang et al. found, through empirical research, that after farmers participated in the technical training of rice, their understanding and practical ability of green production technology was significantly improved, and the use of chemical fertilizers was continuously reduced [6,7,8,14,15]. With such a driving mechanism, cooperatives become an important production organization for smallholder farmers because they provide a solid foundation for their agricultural production technologies and encourage farmers to utilize advanced technologies and share their experiences, such as the use of superior seed varieties and water-saving irrigation systems, the scientific use of fertilizers, and the adoption of environmentally friendly technologies, which broadens the scope of farmers’ choices for the implementation of AGPTs and contributes to the development of sustainable agriculture [16]. Second, cooperatives enable farmers to engage in green production through education and training [17]. In recent years, in order to realize the organic connection between farmers and the development of modern agriculture, the rapid development of cooperatives and other new agricultural management subjects has been promoted, which in turn allows cooperatives to carry out training and guidance, information exchange, and other services on the basis of multiple advantages such as industry, talent, economy, and resources [18,19,20]. It has been confirmed that government training has an important impact on farmers’ factor input and technology selection behavior [20,21], but government-led education on AGPTs is costly and ineffective, and it is difficult to form a scale effect. Therefore, agricultural technology extension institutions such as cooperatives should be encouraged to conduct education and training to promote the adoption of AGPTs by farmers [8,22]. Education, as a benchmarking guide for decision making on the adoption of green technology in agriculture, can enable farmers to raise awareness of green control in agriculture and the advantages and disadvantages of traditional control so that farmers can establish green concepts at the subjective level and effectively promote green production by farmers [23,24]. Since education cannot be separated from trust between organizations and individuals [25], mutual understanding between public relations officers and farmers [26], individual farmers’ needs [27], and ease of access to information [28] are necessary, as cooperatives are deeply rooted in rural social networks as spontaneous farmers’ organizations [8,29]. Thirdly, cooperatives influence farmers to engage in green production by changing their perceptions [30]. Relevant studies have shown that there are significant differences in the level of green cognition and production behavior of farmers in different organizational forms [31]. Farmers’ green cognition is formed through continuous acquisition and accumulation of green production in agriculture [32], and training is an important channel for them to acquire information [33]. Enhancing green production cognition is crucial for farmers to acquire knowledge about new technologies, and from cooperatives, farmers can have a deeper understanding and awareness of the value of organic fertilizers and biopesticides, which will influence farmers’ decision to adopt AGPTs [33]. However, there are fewer empirical studies on green perceptions of the adoption of AGPTs by farmers [17], and more academic attention is currently focused on the area of risk perception [20,34]. The effect of green perceptions on the adoption of AGPTs may also be different among farmers who participate or do not participate in cooperatives. Therefore, the relationship between cooperative participation and AGPT adoption behavior also requires attention to the important contextual factor of green perception [10,20].

From the above, we can find that cooperatives can influence farmers’ perceptions in three areas: economic, ecological, and social [5]. According to Schultz, as rational economic beings, the important criterion for farmers’ behavioral decisions is to maximize benefits. When farmers make behavioral decisions about environmentally friendly production technologies, cooperatives enhance the green economic benefits and perceptions of the technologies through scale advantages and training and then compare the costs that farmers need to pay before making behavioral choices. When cotton farmers perceive that the benefits of AGPTs provided by cooperatives outweigh the costs, they continue to adopt AGPTs [5,35]. Ecological perceptions reflect cotton farmers’ awareness of the current state of pollution from agricultural production and influence their willingness to participate in environmental management [30]. There is an obvious dependence between agricultural production and the ecological environment, and the emergence of cooperatives can promote the adoption of AGPTs, thereby reducing agricultural nonpoint source pollution problems and improving soil quality. Some studies have pointed out that cooperatives can make farmers realize the role of green production in improving soil fertility and reducing environmental pollution and promote farmers to improve AGPT adoption behavior [33]. With the social emphasis on green development and the promotion of cooperative training, cotton farmers are increasingly paying attention to the impact of the natural environment and social development on their lives [36]. Green production can reduce environmental pollution from agricultural nonpoint sources and contribute to rural social development [20]. Through sensitization and training, the more farmers know about the AGPT policy, the higher their participation is likely to be [36,37,38]. In addition to the cognitive enhancement that contributes to the adoption of AGPTs by farmers, environmental regulation also plays an important role. On the one hand, after joining cooperatives, the cooperatives monitor members’ agricultural use and production procedures and, under the influence of restrictive regulations, organize penalties for noncompliance with production rules [38]. Restrictive regulations and policy instruments are reflected in regulatory penalties for agricultural nonpoint source pollution. Regulations increase the probability that cotton farmers will be penalized for agricultural nonpoint source pollution, which may affect the adoption of AGPTs by cotton farmers [39]. On the other hand, after joining the cooperative, assuming that the cooperative gives more incentives to members who implement green production, such as product acquisition, technology supply, and reputation enhancement, both the cooperative and cotton farmers can gain some benefits. At the same time, green production has obvious positive externalities, and giving cotton farmers certain subsidies can reduce the adoption cost of AGPTs and promote cotton farmers’ participation in green production [40]. According to the value perception theory, under the influence of government incentive regulation, cooperatives use economic incentives to make the potential value of AGPTs intuitively visible so that members can directly identify and judge the value of the behavior and then make production decisions to adopt AGPTs [11].

Therefore, based on microdata from 502 cotton farmers in Xinjiang Province, China, the main objective of this paper is to explain more comprehensively how cooperatives influence the adoption of agricultural green production technologies (AGPTs) by changing the green perceptions of cotton farmers so as to provide useful policy recommendations for solving the dilemmas of adopting AGPTs and promoting cotton farmers to realize that green production transformation is the key to effective agricultural production and management [41]. This paper mainly contributes in the following three aspects: first, based on the consideration of the heterogeneity of production segments, propensity score matching is used to assess the effect of farmers’ participation in cooperatives on cotton farmers’ AGPT adoption behavior. The previous literature mostly takes a single technology as an example, and there is a lack of further discussion on the differences in technology promotion effects of cooperatives due to the heterogeneity of production segments [15,17,26]. Considering the technology diffusion function of cooperatives, it is necessary to focus on analyzing the influence of cooperatives on the adoption behavior of AGPTs by cotton farmers. Second, from the perspective of intrinsic green cognition and extrinsic environmental regulation, we test the path of the influence effect with the help of the mediation effect model in order to contribute to the wider application of AGPTs [42]. In addition, from the perspectives of internal green perception and external environmental regulation, the mediation effect model is used to test the path of the effect so as to provide a useful reference for the wide application of AGPTs. Most studies focus on the risk perception perspective, although some scholars have pointed out that farmers’ AGPT adoption behavior is based on farmers’ cognition, which is formed by processing relevant information to implement [20,35]. Green technology training for members by cooperatives not only enhances farmers’ technological thinking and awareness of modernization but also effectively stimulates farmers’ willingness and ability to adopt new technologies. Therefore, this study introduced the three dimensions of green economic benefit cognition, green ecological efficiency cognition, and social green well-being cognition to comprehensively measure the farmers’ intrinsic cognition of AGPTs and further explored how cooperatives can influence the adoption of AGPTs by changing the farmers’ intrinsic cognition. Finally, this study is based on an important cash crop—cotton. The previous research on AGPTs has focused on food crops such as rice, wheat, and soybean [15,27,33], while less research has been conducted involving cotton. Compared with traditional food crops, cotton (Gossypium hirsutum L.) is an economically important crop in China, and global cotton production in 2023/24 is projected at 112.9 million bales, and production in China—the leading cotton producer—is forecast at 27.0 million bales. China is expected to account for 24 percent of global production this season, and its cotton production ranks first in the world [43]. However, cotton production entails an enormous environmental burden. Cotton has a higher consumption intensity of agrochemicals and water resources than other plants grown elsewhere (e.g., wheat and maize) [44]. Unsustainable water consumption accounts for 67% of the cotton production process, ranking third among unsustainable water-consuming crops [45]. China, the largest cotton producer, accounts for 25.4% of the global production [46]. This country endures severe problems with excessive fertilization in crop production. The low irrigation water productivity in China indicates a large wastewater volume [47]. The amount of nitrogen and phosphate fertilizers applied per hectare of cotton production in China is over three times that of the United States [48]. The above suggests that there is a need for research on green production technologies for cotton.

Therefore, the research hypotheses of this paper include three main aspects: (1) joining cooperatives can promote cotton farmers’ adoption of AGPTs; (2) cooperatives have a positive impact on the adoption of AGPTs by improving the internal cognition of cotton farmers; (3) cooperatives have a positive impact on the adoption of green production technology by cotton farmers through external regulation; and (4) cooperatives have a positive impact on the adoption of green production technology by cotton farmers through external regulation

The remainder of the paper is structured as follows. Section 2 contains materials and methods, Section 3 contains results, Section 4 contains a discussion, and Section 5 contains the conclusions.

2. Materials and Methods

2.1. Theoretical Framework and Research Hypotheses

According to the theory of planned behavior, between farmers’ cognition, thinking, and behavior, farmers’ cognition has a fundamental role, and farmers will deeply interpret green agricultural production technology through individual cognition, which will be further transformed into the behavior of adopting AGPTs through the operation of the thinking mode, and the overall process is the benign interaction between internal cognition and external environment, which will promote the rational and effective choice of adopting AGPTs based on the “limited rationality” of farmers. The overall process is a good interaction between internal cognition and the external environment, which promotes farmers to adopt AGPTs based on “limited rationality” to make a reasonable and effective choice. Thus, it can be seen that whether cotton farmers are willing to adopt AGPTs depends on the degree of their knowledge of green production. By joining the cooperative, the external environment will enhance the cotton farmers’ inner green production cognition and then drive the cotton farmers to adopt AGPTs so that the logical framework of this paper can be expressed as “joining the cooperative → cotton farmers’ green production inherent cognitive enhancement and the external environment to promote the regulation of the → adoption of AGPT”. The mechanism framework is shown in Figure 1.

First, participation in cooperatives can facilitate the adoption of AGPTs by cotton farmers.

To simplify the analysis, this paper assumes that the total market demand for cotton is constant and that the market has the same price so that the only factor that determines the production income of cotton farmers is income. Thus, for the adoption of green production technology, cotton farmers have the advantage of having a high degree of green awareness and compliance with good external environmental constraints so that the cotton farmers obtain a higher potential income; its disadvantage is the need to pay a sum of money for the purchase of the corresponding agricultural machinery (or mechanical services), the construction of supporting infrastructure, and so on. This paper assumes that cotton farmers only need to decide whether to adopt green production technology, for a rational economic man, when the decision to adopt several green production technologies would directly result in the maximization of cotton farmers’ profits. Therefore, only when the profit of adopting green production technology is higher than the profit of not adopting the technology will cotton farmers make the adoption decision [33]. As a result, the conditions for whether a cotton farmer adopts a green production technology f (R) can be described as follows:

$$f\left(R\right)=p\left\{\left[E\left(R\right)-TC\right]>R_0\right\}$$

(1)

Among them, $E\left(R\right)$ represents the expected benefits of cotton farmers to adopt green technology, $TC$ represents the cost of cotton farmers to adopt AGPTs, $R_0$ represents the net benefits of cotton farmers to implement ordinary production [40], and p denotes the extent to which factors, such as the cotton farmer’s internal perceptions and external environment, impact their decision to adopt green technology, as only when the expected benefits of adopting AGPTs minus the corresponding cost are greater than the net benefits of the implementation of ordinary production will the cotton farmers have the incentive to adopt AGPTs [11,49]. Cooperatives, through the play of their green production premium incentives, information transfer, and guidance, as well as the organization of the mandatory constraints of implementation to enhance the expected income of cotton farmers and reduce the cost of adopting AGPTs, inspire farmers to actively or passively adopt green production technology [49,50]. Based on this, the study is proposed, in addition to H1.

H1. 

Joining cooperatives can promote the adoption of AGPTs by cotton farmers.

Second, cooperatives promote the adoption of AGPTs by enhancing the green perceptions of cotton farmers.

Farmers’ green cognition reflects their perceptions or attitudes towards green production in the current environment, and their green cognition and behavior directly affect the adoption of AGPTs and the effectiveness of agricultural green development [51]. Through internal information transmission and communication, cooperatives enhance farmers’ green awareness and promote their adoption of AGPTs. First of all, after joining cooperatives, farmers can more obviously perceive the importance of green production and be familiar with green production technology through the green production atmosphere of cooperatives [18]. Secondly, the production demonstration and guidance training in all aspects of the cooperative before, during, and after production can play a role in promoting the importance of green production information and guiding values [19]. Finally, the social network within cooperatives, as an informal network of relationships, can play a role in information sharing, transferring technical information for farmers, providing conditions for farmers to “learn from”, and enhancing farmers’ cognition of new technologies. The essence of agricultural green development is to achieve the coordination and unification of economic benefits, ecological benefits, and social benefits [52]. In order to explore the influence of cooperatives on cotton farmers’ adoption of AGPTs by influencing their green cognition, this paper explains cotton farmers’ internal cognition of AGPTs from three dimensions: green economic benefit cognition, green ecological benefit cognition, and social green well-being cognition. In summary, hypothesis H2 is proposed.

H2. 

Cooperatives have a positive impact on the adoption of AGPTs by improving the internal cognition of cotton farmers.

H2a. 

Green economic benefit cognition has a positive effect on cotton farmers’ AGPT adoption behavior.

H2b. 

Green ecological benefit cognition has a positive effect on cotton farmers’ AGPT adoption behavior.

H2c. 

Social green well-being cognition has a positive impact on cotton farmers’ AGPT adoption behavior.

Third, the external environment promotes the adoption of AGPTs by cotton farmers through the top-level design of constraints and incentives.

As external environmental regulations involve public goods [53], they can avoid environmental problems caused by agricultural nonpoint source pollution to a certain extent and have a positive externality. Environmental regulation is mainly used to solve externalities, that is, the government’s environmental regulation policy affects the green production behavior of cotton farmers through the indirect intervention of cooperatives [4], so as to ultimately reduce environmental pollution and promote green economic development. Due to the constraints and influences of many practical factors, even if cotton farmers have a high level of cognition on green production, it is often difficult to convert it into actual behavioral willingness [16]. Therefore, it is necessary to clarify the role of government measures in influencing cotton farmers’ adoption of AGPT behavior. In summary, hypothesis H3 is proposed.

H3. 

Cooperatives have a positive impact on the adoption of green production technology by cotton farmers through external regulations.

H3a. 

Binding regulation has a positive impact on the adoption of green production technology by cotton farmers.

H3b. 

Incentive regulation has a positive impact on the adoption behavior of green production technology by cotton farmers.

2.2. Research Methodology

2.2.1. Propensity Score Matching (PSM)

PSM has become one of the effective tools to overcome selection bias in observational studies. It involves pairing treatment groups and control groups that are similar in terms of observable covariates, and then the pairing method provides an unbiased estimate of the treatment effect, such as cooperative membership [53]. Among the various matching methods, the propensity score method is recommended when there is a dimension problem because the nature of covariates (including measurement scales) makes it difficult to match covariates [40]. In fact, whether or not to join a cooperative belongs to the self-selection behavior of cotton farmers, and propensity score matching can effectively control the self-selection bias of the sample by constructing a counterfactual analytical framework so that cotton farmers who participate in cooperatives and those who do not participate in cooperatives are in a balanced and comparable state [20]; the model will then accurately assess the impact effect of participation in cooperatives on cotton farmers’ AGPT adoption behavior. The propensity score matching method (PSM) model consists of the following three steps:

First, this study focused on the effect of participation in cooperatives on cotton farmers’ AGPT adoption behavior. In this study, the explained variable is the cotton farmers’ AGPT adoption behavior, and it is a binary discrete variable, so we use a binary logit model to estimate the sample. The model is set as follows:

$$\ln \left(\frac{p}{1-p}\right)={\beta }_0+{\beta }_{1}Cooperatio{n}_i+{\displaystyle \sum _{i=1}^n{\beta }_i{X}_i}+\epsilon$$

(2)

In Equation (2), p is the probability of a cotton farmer adopting AGPTs. Cooperation_i is the core explanatory variable (whether cotton farmers participate in cooperatives). X_i is a covariate variable. ${\beta }_1$ is the regression coefficient of the core explanatory variable. ${\beta }_i$ is the regression coefficient of the control variable. ${\beta }_0$ is the intercept of the regression. $\epsilon$ is the random error term.

Second, the propensity to match score $P\left(Z_i\right)$ was used as the basis for the probability of cotton farmers’ participation in cooperatives and was estimated using a logit model with the following expression [5]:

$$P\left(Z_i\right)=P\left(Cooperation=1\left|Z_i\right.\right)=\frac{\exp (Z_i^{\prime }\beta )}{(1+\exp (Z_i^{\prime }\beta ))}$$

(3)

In Equation (3), $P\left(Cooperation=1\left|Z_i\right.\right)$ is the propensity score for cotton farmers’ participation in cooperatives. X_i is a covariate variable. $\exp (Z_i^{\prime }\beta )$ is the probability that a cotton farmer adopts green production techniques. When covariates are considered, the variable has an impact on both the cotton farmers’ participation in cooperatives and the adoption of AGPT behavior. The covariates selected for this study include farmer characteristics (age, gender, level of education, whether part-time) and family characteristics (level of income, area of cultivated land, number of laborers, whether they are village cadres, whether they have access to information on green production technologies).

Third, the appropriate matching method is selected to match the participating cooperative cotton farmers and nonparticipating cooperative cotton farmers [6], and the average treatment effect (ATT) of the sample of the matched treatment group represents the influence effect of participating cooperatives on the adoption behavior of AGPTs by cotton farmers [4]. The formula is as follows:

$$ATT=E\left(Y_1\left|Cooperation=1\right.\right)-E\left(Y_0\left|Cooperation=1\right.\right)=E\left(Y_1-Y_0\left|Cooperation=1\right.\right),$$

(4)

where Y₁ is the degree of adoption of AGPTs by cotton farmers in participating cooperatives and Y₀ is the degree of adoption of AGPTs by cotton farmers in nonparticipating cooperatives.

2.2.2. Intermediation Model

In order to test the driving mechanism of cooperatives on cotton farmers’ AGPT adoption behavior, the mediation model was constructed as follows, drawing on Yu’s methods and steps [54]:

$$T_i=c_{i}Cooperation+e_1$$

(5)

$$Mediato{r}_i=a_{i}Cooperation+e_2$$

(6)

$$T_i=c_{i}Cooperation+b_{i}Mediato{r}_i+e_3$$

(7)

In the above equations, $T_i$ is the ith explanatory variable; $Mediato{r}_i$ denotes the i mediating variable; and $e_1$, $e_2$, and $e_3$ are random error terms. Equation (5) analyzes the effect of cooperatives on the explanatory variables; Equation (6) analyzes the effect of cooperatives on the mediator variables; and Equation (7) puts the cooperatives and the mediator variables into the right-hand side of the model and analyzes their effect on the explanatory variables. If the coefficients $c_i$, $a_i$, and $b_i$ are significant, it proves that the Mediator_i variable plays a significant mediating role.

2.2.3. Coefficient of Variation Method

The heterogeneity of production links may affect the technology diffusion effect of cooperatives, so it is not appropriate to simply sum up the technology adoption of each link when measuring the impact effect. The coefficient of variation method, as an objective assignment method [55], can utilize the degree of data variation to determine the weight of indicators [56], which objectively reflects the relative importance of indicators for decision makers and also avoids the problem of personal preference in the subjective assignment. In order to reflect the level of cotton farmers’ adoption of AGPTs in each production segment, this paper uses the coefficient of variation method for objective assignment, which transforms the level of cotton farmers’ adoption of AGPTs into a continuous variable; the formula is as follows:

$$V_j=S_j/U_j$$

(8)

$$W_j=V_j/{\displaystyle \sum _{i=1}^n{V}_j},$$

(9)

where $S_j$ is the standard deviation of the j indicator, $U_j$ is the mean value of the jth indicator, and $V_j$ is the coefficient of variation of the indicator. As shown in Equation (8), the weight coefficient $W_j$ of the indicator can be obtained by normalizing the coefficient of variation and calculating the proportion of the coefficient of variation of a certain indicator to the sum of the coefficients of variation of all indicators.

2.3. Data Sources and Variable Selection

The data in this paper are derived from a special survey of cotton farmers in Xinjiang carried out by the research group from October to December 2022. Xinjiang Province not only ranks first in cotton production in China but also extensively applies many modern agricultural technologies such as cotton breeding, planting, and harvesting. Therefore, it is more representative to choose Xinjiang Province to conduct relevant research on cotton farmers’ adoption of AGPTs. In addition, Xinjiang, as one of the main cotton-producing areas in China, not only has unique natural conditions but also the economic aggregate between different places is not different. Therefore, choosing the two research sites can effectively control the differences in technology adoption caused by natural conditions and economic development. The research was conducted mainly in 10 cities (counties) accounting for more than 4% of the total cotton production in Xinjiang’s major cotton-producing counties: Luntai County, Bachu County, Jinghe County, Gashi County, Yuli County, Awati County, Wusu City, Kuqa County, Shawan City, and Shaya County (shown in Figure 2). Six townships were selected in each city (county) using the random sample method, and nine to ten cotton farmers in the townships (towns) were selected using the random sample method to carry out the field survey again. Ten cotton farmers carried out field surveys (without distinguishing between cotton farmers who joined or did not join the cooperative). A total of 550 paper questionnaires were distributed, and 502 valid questionnaires were returned, achieving a questionnaire effective rate of 91.27%.

In order to satisfy the recognizability of propensity score matching, based on previous studies, the AGPT adoption degree of cotton farmers was set as the explained variable. Participation in cooperatives was set as the core explanatory variable. However, considering the long agricultural production cycle, cooperatives often have a time-lag effect on the change in the green production technology adoption behavior of cotton farmers, so farmers who have been in cooperatives for less than one year were excluded. Personal characteristics of cotton farmers (age, gender, education level, and part-time employment) [57] and family characteristics (income level, cultivated land area, number of labor force, whether they are village cadres, and whether they can obtain information about green production technology) are taken as control variables. Internal cognition such as green economic benefit cognition, green ecological benefit cognition, and social green welfare cognition and external regulation such as binding regulation and incentive regulation are taken as the intermediary variables. In addition, in order to test the mechanism of the influence of cooperatives on the adoption of green production technology by cotton farmers, the mediation variables were set according to the research hypothesis proposed by theoretical analysis, and the village participation rate was taken as the instrumental variable. The details are shown in

Table 1. Definition of variables.

Variable Type	Variable Name	Variable Definition and Assignment
explanatory variable	level of adoption of AGPTs	calculated by the coefficient of variation method
core explanatory variables	participation, or not, in cooperatives	1 = participation, 0 = nonparticipation
control variables	age (Z1)	1 ≤ 20 years old, 2 = 21–30 years old, 3 = 31–40 years old, 4 = 41–50 years old, 5 ≥ 51 years old
	genders (Z2)	1 = male, 0 = female
	educational level (Z3)	1 = elementary school and below, 2 = middle school, 3 = high school/secondary school, 4 = college and above
	whether or not part-time (Z4)	1 = yes, 0 = no
	income level (Z5)	1 = CNY 20,000 and below, 2 = CNY 21,000–CNY 60,000, 3 = CNY 61,000–CNY 100,000, 4 = CNY 101,000–CNY 140,000, 5 = CNY 141,000 and above
	cotton acreage (Z6)	1 = 10 acres and under, 2 = 10.1–30 acres, 3 = 30.1–60 acres, 4 = 60.1–100 acres, 5 = 100.1 acres or more
	number of family laborers (Z7)	actual number of family laborers (persons)
	whether or not they are village cadres (Z8)	1 = yes, 0 = no
	whether information on AGPTs is accessible (Z9)	1 = yes, 0 = no
intermediary variables	perceived benefits of a green economy (Z10)	AGPTs can increase crop yields: 1 = disagree completely, 2 = do not agree very much, 3 = fairly, 4 = agree somewhat, 5 = agree completely
		AGPTs can increase household income: 1 = disagree completely, 2 = do not agree very much, 3 = fairly, 4 = agree somewhat, 5 = agree completely
	green eco-efficiency perception (Z11)	AGPTs can reduce environmental pollution: 1 = disagree completely, 2 = do not agree very much, 3 = fairly, 4 = agree somewhat, 5 = agree completely
		AGPTs improve soil quality: 1 = disagree completely, 2 = do not agree very much, 3 = fairly, 4 = agree somewhat, 5 = agree completely
	perceived green well-being of society (Z12)	AGPTs can contribute to rural eco-environmental protection: 1 = disagree completely, 2 = do not agree very much, 3 = fairly, 4 = agree somewhat, 5 = agree completely
		AGPTs can contribute to the greening of agriculture: 1 = disagree completely, 2 = do not agree very much, 3 = fairly, 4 = agree somewhat, 5 = agree completely
	binding environmental regulation (Z13)	intensity of government penalties for agricultural surface pollution: 1 = very unstrict, 2 = not too strict, 3 = average, 4 = more strict, 5 = very strict
	incentive-based environmental regulation (Z14)	strength of government subsidies for AGPTs: 1 = very unsupportive, 2 = not very supportive, 3 = average, 4 = more supportive, 5 = very supportive
instrumental variable	participation rate in cooperatives in the village (Z15)	0–20% = 1, 20.1–40% = 2, 40.1–60% = 3, 60.1–80% = 4, 80.1–100% = 5

.

3. Results

3.1. Variables Descriptive Statistical Analysis and Test of Difference in SAMPLE Means

This study examined the variability of cotton farmers in two different scenarios of participation and nonparticipation in cooperatives using independent sample t-tests to determine whether participation and nonparticipation in cooperatives influence the behavior of cotton farmers. The research design identifies two different groups as subgroups (i.e., core explanatory variables) while using AGPT adoption level as an explanatory variable, which in turn determines whether there is a difference between the two using the statistical method of independent sample t-tests using SPSS 17.0 software [58].

As can be seen from

Table 2. Variables’ descriptive statistical analysis and test of difference in sample means.

Variable Type	Variable Name	Average Value		Standard Deviation
		Participating in Cooperatives	Not Participating in Cooperatives
explanatory variable	level of adoption of AGPTs	2.06	1.48	0.58 ***
core explanatory variables	participation, or not, in cooperatives	1.00	0.00	—
control variables	Z1	3.11	3.18	−0.07
	Z2	0.51	0.55	0.04
	Z3	2.34	2.21	0.13
	Z4	0.34	0.59	−0.25 ***
	Z5	3.16	3.01	0.15 *
	Z6	1.65	1.55	0.10
	Z7	3.82	3.63	0.19 ***
	Z8	0.59	0.58	0.01
	Z9	0.50	0.51	−0.01
intermediary variables	Z10	1.38	1.22	0.16 ***
		0.82	0.73	−0.09 ***
	Z11	3.66	3.36	0.30 ***
		0.61	0.19	0.42 ***
	Z12	2.02	1.35	0.67 ***
		0.78	0.50	0.25 ***
	Z13	3.63	3.00	0.63 ***
	Z14	0.78	0.53	0.25 ***
instrumental variable	Z15	3.22	2.66	0.56 ***

Note: * and *** denote variables that are significant at the 10% and 1% levels, respectively.

, this paper uses independent sample t-tests to determine the effect of participation in cooperatives on the degree of adoption of AGPTs by cotton farmers, and the results show that the degree of adoption of AGPTs by cotton farmers who joined cooperatives and those who did not join cooperatives show a difference at the 0.05 level of significance, and a further comparison of their means reveals that the degree of adoption of AGPTs by cotton farmers who joined cooperatives is higher than that of those who did not, i.e., hypotheses H1, H2, and H3 passed the test.

Specifically, through the independent sample t-tests on the data of the treatment and control groups [58], it can be found that the degree of adoption of AGPTs by cotton farmers participating in cooperatives is 2.06, which is higher than that of those who do not participate in cooperatives (1.48), indicating that participation in cooperatives has the potential to promote the adoption of AGPTs by cotton farmers [33]. At the same time, the internal green cognition and external environmental regulations of cotton farmers participating in cooperatives have a significantly higher role in promoting the adoption behavior of AGPTs than cotton farmers not participating in cooperatives. Moreover, green production can improve the yield and income of cotton farmers significantly more than those who do not participate in cooperatives, indicating that AGPTs can improve the production efficiency and market premium of cotton farmers [28], which may be a mediating path for cooperatives to influence the adoption of AGPTs among cotton farmers. In addition, the village enrollment rate also differed significantly between the two groups, suggesting that there is some kind of correlation between enrollment and cotton farmers’ participation in cooperatives versus nonparticipation and that the instrumental variable may be valid.

3.2. Measurement and Analysis of the Degree of Adoption of Agricultural Green Production Technologies

To comprehensively investigate the adoption of AGPTs by cotton farmers, this paper made reference to the Technical Guidelines for Green Agricultural Development (2018–2030) [59] issued by the Ministry of Agriculture and Rural Affairs and the research of Hu, Feng et al. [60,61] and analyzed the aspects of land preparation, sowing, fertilization, irrigation, disease and pest control, and waste disposal of cotton farmers. AGPTs were selected to investigate, and (technology adoption area/total sowing area of cotton) × 100% was used as the index to measure the adoption degree of each AGPT. In this case, no adoption = 1 (adoption degree is 0), low adoption = 2 (0% < adoption degree ≤ 30%), medium adoption = 3 (30% < adoption degree ≤ 60%), and high adoption = 4 (60% < adoption degree ≤ 100%). Finally, with the help of the coefficient of variation method, the weighted average of six technologies was used to represent the adoption degree of AGPTs by cotton farmers. The specific results are shown in

Table 3. Agricultural green production technology adoption behavior indicator system and weights.

Target Level	Production Chain	Selecting Variables and Assigning Values	Coefficient of Variation	Indicator Weights	Actual Level of Adoption
					Participation in Cooperatives	Not Participating in Cooperatives
pre-production	land preparation	(adoption area of deep tillage technology/cotton planting area) × 100%	0.558	0.167	2.16	1.58
mid-production	seeding	(new variety technology adoption area/cotton planting area) × 100%	0.546	0.164	2.09	1.49
post-production	fertilization	(soil testing formula fertilization technology adoption area/cotton planting area) × 100%	0.572	0.171	1.93	1.29
	irrigation	(adoption area of water-saving irrigation technology/cotton planting area) × 100%	0.556	0.166	2.06	1.52
	pest control	(green prevention and control technology adoption area/cotton planting area) × 100%	0.571	0.171	1.89	1.36
	waste disposal	(adoption area of straw-returning technology/cotton planting area) × 100%	0.537	0.161	2.23	1.69

.

According to research data, after objective weighting using the coefficient of variation method, the weight ranking of AGPTs from high to low is soil testing formula fertilization technology, green prevention and control technology, deep tillage technology, water-saving irrigation technology, new variety technology, and straw-returning technology. The possible explanation is that the purpose of applying green agricultural technology to cotton farmers in different production stages is not consistent, and there are also certain differences in the operating points of various technologies, so the relative importance of technology will also vary. Meanwhile, based on the results in Table 2, we found that cotton farmers participating in cooperatives have a higher degree of adoption of all six green production technologies than those who are not, so there is a significant difference between cotton farmers participating in cooperatives and those who have not participated in cooperatives, indicating that participation in cooperatives promotes cotton farmers to improve the degree of adoption of AGPTs.

3.3. Assessment of the Effect of Participation in Cooperatives on Cotton Farmers’ Agricultural Green Production Technology Adoption Behavior

3.3.1. Propensity Score Estimation

In the paper, a logit model is used to estimate the probability of cotton farmers’ participation in cooperatives [49], and the regression results are shown in

Table 4. Logit estimation results for propensity score.

Target Level	Variant	Ratio	Standard Error	Z-Statistic
personal characteristics	Z1	−0.20 *	0.11	−1.81
	Z2	0.04	0.22	0.18
	Z3	0.17	0.11	1.53
	Z4	−0.99 ***	0.20	−0.54
family characteristics	Z5	0.18 *	0.11	1.65
	Z6	0.22 *	0.13	1.72
	Z7	0.25 *	0.13	1.93
	Z8	0.08	0.20	0.39
	Z9	−0.17	0.20	−0.87
constant term		−0.81	0.78	−1.04
LR chi (9)		45.32	log-likelihood	−319.83
prob > chi2		0.00	pseudo R2	0.07

Note: * and *** denote variables that are significant at the 10% and 1% levels, respectively.

: age and whether working part-time in the personal characteristics of cotton farmers and income level, cultivated land area, and number of laborers in the household characteristics have a significant impact on whether cotton farmers participate in cooperatives or not [30]. among them, the older the age and the deeper the degree of part-time work, the lower the probability of participating in cooperatives [62], and the higher the income level, the wider the area of cotton cultivation, and the higher the number of family laborers, the higher the probability of participating in cooperatives. The results of the effects of these significant variables are basically consistent with the reference, which is in line with economic theory and the reality of agricultural production.

3.3.2. Common Support Domains and Equilibrium Tests

In order to ensure the effectiveness of propensity score matching, the matching process needs to satisfy the common support domain assumption, that is, the wider the area of the common support domain represents the less sample loss and the better the matching effect [5]. The precondition for carrying out propensity score matching is the presence of shared support or overlap between the intervention and control group’s propensity scores. In order to eliminate selectivity bias, it is imperative to administer the common support domain test on the sample, which measures the match quality between the explanatory variables of the cooperative participating and nonparticipating cotton farmer samples [35]. Statistics show that the number of samples after matching is 493, the loss is 9, and the intersection of the intervals of the propensity scores of cotton farmers participating in cooperatives and nonparticipating is taken to be [0.29, 0.83], which indicates that the test of the common support domain passes, and the matching effect is more satisfactory.

In order to ensure the validity of matching outcomes, a covariate balance test was conducted to ensure that, apart from AGPTs, there were no significant pre-existing variations in other covariates between the treatment and control groups after matching, thus rendering the matching group a suitable counterfactual [51]. In order to ensure that there is no obvious difference between the treatment group and the control group in the characteristics after matching, the paper uses the sampling with put-back and adopts four matching methods to measure the standardized gap between the two before and after matching. As shown in

Table 5. Balance test results.

Matching Method	Pseudo R2	LR Value	p Value	Mean Deviation (%)	Median Deviation (%)
before matching	0.066	45.35	0.000	15.5	13.2
nearest-neighbor matching (1 on 2)	0.005	3.93	0.916	4.0	2.6
nearest-neighbor matching (1 on 4)	0.002	1.66	0.996	2.2	1.0
radius matching (0.01)	0.001	0.81	1.000	2.1	2.3
kernel matching (0.06)	0.001	0.94	1.000	2.5	2.1

, the mean deviation and median deviation of cotton farmers participating in cooperatives and nonparticipating after matching decreased from more than 10% to less than 5%, and none of the differences were significant, indicating that the balance test was passed.

3.3.3. Estimation of Impact Effects

According to

Table 6. Estimated effects of cooperatives on cotton farmers’ agricultural green production technology adoption behavior.

Matching Method	Treatment Group Means	Control Group Means	ATT	Standard Error	T Values
before matching	2.06	1.48	0.57 ***	0.08	7.54
nearest-neighbor matching (1 on 2)	2.06	1.65	0.41 ***	0.09	4.80
nearest-neighbor matching (1 on 4)	2.06	1.61	0.45 ***	0.08	5.51
radius matching (0.01)	2.06	1.63	0.42 ***	0.08	5.29
kernel matching (0.06)	2.06	1.59	0.47 ***	0.08	6.17
average value after matching	2.06	1.62	0.44

Note: *** denote variables that are significant at the 1% levels, respectively.

, it can be seen that the ATT estimated using nearest-neighbor matching (1 vs. 2), nearest-neighbor matching (1 vs. 4), radius matching (0.01), and kernel matching (bandwidth of 0.06) with put-back sampling were 0.41, 0.45, 0.42, and 0.47, respectively, which are significant at the 1% level and have a small difference, indicating that joining cooperatives can indeed enhance cotton farmers’ adoption of AGPTs and verifies hypothesis H1; from the mean value, the adoption of AGPTs by cotton farmers participating in cooperatives increased from 1.62 to 2.06, an increase of 0.44 or 27.16%; from the comparison of ATT before and after matching, the ATT before matching was 0.57, and it decreased to 0.44 after matching, which indicates that the self-selection bias in the sample was effectively controlled.

In fact, labor characteristics and technical needs of different production links are different. In order to explore whether this heterogeneity will affect the technology promotion effect of cooperatives, the impact effects are classified and estimated. The results are shown in

Table 7. Estimates of impact effects based on different production chains.

Post-Matching	Labor-Intensive Segments			Technology-Intensive Segments
	Deep Tillage Technology	New Variety Technology	Water-Saving Irrigation Technology	Straw-Returning Technology	Soil Testing and Fertilizer Application Techniques	Green Prevention and Control Technology
treatment group means	2.16	2.10	2.06	2.23	1.94	1.89
control group means	1.77	1.67	1.73	1.85	1.35	1.39
ATT average	0.39 ***	0.45 ***	0.33 ***	0.38 ***	0.59 ***	0.50 ***
increase amplitude (%)	22.03	25.75	19.08	20.54	43.70	35.97

Note: *** denote variables that are significant at the 1% levels, respectively.

. The adoption rates of soil testing and formula fertilization, green prevention and control, new varieties, deep tillage, straw-returning, and water-saving irrigation technologies for cotton farmers increased by 43.70%, 35.97%, 25.75%, 22.03%, 20.54%, and 19.08%, respectively. The effect of fertilization and pest control was much higher than other links. The possible reason for this is that, according to Xue et al.’s division method [63], soil testing, formula fertilization, and green prevention and control technology have high requirements for labor skills and operation experience and are technology-intensive production links, while land preparation, sowing, irrigation, and straw harvesting are labor-intensive, but the technical content is low, so they are labor-intensive production links. The adoption of AGPTs in labor-intensive production processes only requires the use of agricultural machinery to implement standardized process operations, which is not difficult to operate, and the cost of capital generated by outsourcing is low, so cotton farmers are not highly dependent on cooperatives. However, the adoption of AGPTs in technology-intensive production links generally requires the technical literacy of operators and the ability to control the process of operation services. Even if outsourced to other entities, it is easy to fall into the monopoly market of “seller pricing”, resulting in high transaction costs and coordination costs, and the capital threshold is relatively high. Therefore, when cotton farmers adopt technology, they need to make up for their shortcomings with technical training, reduce the adoption cost of technology through collective purchase, and rely on cooperatives relatively more.

3.4. Examination of the Driving Mechanisms of Adoption Behavior of Agricultural Green Production Technologies for Food and Agriculture by Participating Cooperatives

The previous section has confirmed that cooperatives can increase the adoption of AGPTs by cotton farmers, but the mechanism of this promotion effect is obviously more worthy of attention. Combined with the theoretical analysis, three mediation effect models are constructed with the help of the stepwise regression method, which are used to test whether cooperatives can influence cotton farmers’ technology adoption behavior by improving their internal green cognition and external environmental regulation. Meanwhile, considering that participation in cooperatives is a self-selected behavior of cotton farmers, we first regressed the “dependent, mediating, and independent variables” without instrumental variables (under control), then regressed the “dependent, mediating, and independent variables” with instrumental variables, then the “mediating and independent variables” with “control”, then the “dependent, mediating, and independent variables” with “control”, and then the “dependent, mediating, and independent variables” with “control”. The regressions, denoted by “cooperatives iv”, were then conducted using the instrumental variables on the “dependent, mediating, and independent variables” to check the validity of the instrumental variables and to address the issue of endogeneity [8,64]. Therefore, this paper, referring to Kumar and Lin’s study, selected the village enrollment rate as an instrumental variable [65,66]; this variable, as village-level data, is very limited by the influence of individual cotton farmers, but it can influence the enrollment decision of cotton farmers through the social network and thus meets the requirements of logistic exogeneity and relevance. Moreover, the F-value of this variable is 92.72, which is much greater than 10, and it is not a weak instrumental variable. The specific results are shown in

Table 8. Analysis of the impact of cooperatives on mediating variables.

Variable	Model I (Green Economic Benefit Cognition)		Model II (Green Ecological Benefit Cognition)		Model III (Social Green Well-Being Cognition)		Model IV (External Regulations)
	Coefficient		Coefficient		Coefficient		Coefficient
cooperatives	0.152 ***	0.086 ***	0.211 **	0.403 ***	0.622 ***	0.271 ***	0.512 ***	0.222 ***
cooperatives iv	0.113 **	0.106 **	0.220	0.326 **	1.307 ***	0.251 **	0.614 **	0.172
participation rate in cooperatives in the village	control	control	control	control	control	control	control	control
endogeneity test F/chi 2 value	0.59	0.17	0.01	0.42	12.87 ***	0.03	0.20	0.18

Note: ** and *** denote variables that are significant at the 5% and 1% levels, respectively.

and

Table 9. Analysis of the impact of cooperatives and mediating variables on the dependent variable.

Variable	Model I (Level of Adoption of AGPT)		Model II (Level of Adoption of AGPT)		Model III (Level of Adoption of AGPT)		Model IV (Level of Adoption of AGPT)
	Coefficient	Coefficient	Coefficient	Coefficient	Coefficient	Coefficient	Coefficient	Coefficient
cooperatives	0.375 ***	0.443 ***	0.149 ***	0.117 ***	0.134 ***	−0.053 **	−0.076 ***	−0.065 ***
Z10	0.685 ***
		0.246 **
Z11			0.013
				0.088 ***
Z12					0.028 **
						−0.120 ***
Z13							0.020 *
Z14								0.091 ***
Z15	control	control	control	control	control	control	control	control

Note: *, **, and *** denote variables that are significant at the 10%, 5%, and 1% levels, respectively.

.

In terms of model I in Table 8, firstly, the result of the propensity score matching confirms that joining cooperatives can lead cotton farmers to adopt AGPTs, i.e., Equation (4) holds, suggesting that the mediating roles of intrinsic perceptions and external regulation may exist. Second, the results of the endogeneity test in Table 8 show that the endogeneity problem of cooperatives on the path of influence on intrinsic cognition and external regulation is not serious, so the regression results of OLS should be used, i.e., participation in cooperatives positively influences both intrinsic cognition and external regulation of cotton farmers, indicating that Equation (5) holds. Finally, the regression results in Table 9 indicate that cooperatives have an impact on cotton farmers’ AGPT adoption through both intrinsic cognition and external regulation, i.e., Equation (6) holds.

Specifically, the results of model I in Table 9 show that cooperatives promote the adoption of AGPTs by cotton farmers by improving their yields and incomes in the process of green production, which gradually improves their knowledge of green economic benefits and verifies the research hypothesis H2a. The results of model II show that the cooperative can improve the cotton farmers’ concept of ecological environmental protection through long-term environmental protection activities and then promote the cotton farmers to adopt AGPTs, which verifies the research hypothesis H2b. It is worth noting that the cooperative cannot improve the cotton farmers’ cognition that green production can reduce environmental pollution so that the cotton farmers can obtain higher cognition of the green ecological benefits, but rather they expect that improving the quality of the soil has a significant effect. It is possible that the cotton farmers’ ability to identify with the “rational economic man” situation, rather than recognizing that relying solely on green production can reduce environmental pollution, prevents cotton farmers from achieving the maximum benefits. The results of model III show that as the cooperatives pay more and more attention to rural ecological protection and green agricultural development through training and publicity, the cotton farmers will be more and more concerned about rural ecological protection and green agricultural development, which will have a significant positive impact on the cotton farmers’ adoption of AGPTs, thus verifying research hypothesis H2c.

The results of the test of model IV in Table 9 showed that cooperatives can promote cotton farmers’ adoption of AGPTs through constraining environmental regulations and incentive types. On the one hand, cooperatives use constraint-based regulation to monitor agricultural surface pollution through the role of organizational norms, which has a significant positive effect on cotton farmers’ adoption of AGPTs, verifying the study plus H3a. On the other hand, incentive-based regulation is an effective way to promote green production in Mi’ahmu, and cooperatives enable cotton farmers to obtain directly recognizable environmental regulations for both benefits and costs, indicating that the pathway is valid. The cooperative enables cotton farmers to obtain values that can be directly identified and judged in terms of both benefits and costs, converting incentive-based environmental regulation into actual benefits, and significantly positively influencing cotton farmers’ adoption of AGPT behavior, validating the study plus H3b.

4. Discussion

4.1. Similarities and Differences with Existing Studies

Green development is an inevitable requirement for the construction of ecological civilization [67], and it should not be overlooked that agricultural pollution has become one of the most serious environmental pollution problems in China [35], and green development in agriculture is imminent. In China, farmers are the front-line main body of agricultural production, as well as the direct adopters and beneficiaries of agricultural green production methods [37]. Therefore, it is important to pay special attention to the main position of farmers in the green development of agriculture [68]. In addition, in the process of agricultural production-related policy practice, the tool used in multilevel governments to realize the organization of farmers’ production is mainly cooperatives; the government tends to provide production services and policy guidance to farmers through cooperatives. Many studies have focused on the impact of farmers’ membership in cooperatives on green agricultural production. As Ncube et al. found, farmers’ professional cooperatives have important functions in providing technical support, forming external economies of scale, and releasing price incentive signals, which help farmers and cooperators to observe and communicate with each other and promote the generation of social learning [69,70,71]. Cooperatives are involved in agricultural technology concept promotion [72] and fertilizer management for farmers, and they become the source of production and management training, standardized production services [73], and ecological concept dissemination [74,75,76,77].

Much of the literature shows that cooperatives can promote farmers’ adoption of AGPTs; for example, Luo et al. demonstrated that cooperatives can enhance farmers’ green production willingness [8], but these are only unilateral promotional effects. Where possible, mediating pathways need to be explored, as academics have focused more on the area of farmers’ risk perceptions [20,34], and there are relatively few empirical studies that have been conducted on intrinsic green perceptions and external environmental regulations affecting farmers’ AGPT adoption behavior [17]. Similar to existing findings [78,79,80,81], we also found a significant positive impact relationship between participation in cooperatives and farmers’ green production behavior. However, what is more meaningful in this paper is that we explored the differences in the promotion effect of AGPTs from the heterogeneity of the agricultural production chain and introduced the two aspects of internal cognition (green production) and external regulation (environmental regulation) to identify the realization path of cotton farmers towards green production—“cognition → willingness → activity”. The results show that cooperatives can increase cotton farmers’ green endogenous motivation and environmental external constraint enhancement by enhancing their internal cognition and external regulation, and they can form an intermediary path to guide cotton farmers to adopt AGPTs in agriculture.

This paper also differs significantly from the previous literature [31] as it suggests that farmers’ perceptions of green production may influence the likelihood and intensity of their adoption of green production techniques in the agricultural production process, whereas some studies will fail to notice the possible mediating role of green perceptions in farmers’ green production. For example, Li et al. found that farmers’ willingness toward green production is increasing in the context of increasingly tight resource and environmental constraints and that farmers’ perceived benefits are a key factor in their green production [31]. Li et al. found that farmers’ sustainable behaviors may be influenced by many factors, among which risk perception can provide effective information and reduce uncertainty, which can motivate the subject to make reasonable decisions, and is an important factor influencing farmers to engage in green production [82]. Unlike them, we found that cotton farmers’ intrinsic perceptions of green production and external regulations are mediating variables that affect their adoption of AGPTs.

O’Connor and Assakerd et al. empirically verified the mechanisms through which farmers’ green perceptions influence their adoption of AGPTs [83,84,85]. Similarly, this paper confirms the effect of green perceptions on green production technology adoption behavior and its internal mechanisms. The difference, however, is that the former uses the double-hurdle model (DHM) and Poisson model, respectively, while we use a mediated effects model, which provides better control of the endogeneity problem through the use of instrumental variables and more reliable conclusions. In addition, in a significant departure from the existing studies that mainly consist of green perceptions or are specific to agriculture, we use the theory of farmer behavior to provide a more comprehensive and accurate explanation of green production perceptions, confirming the driving mechanisms of cotton farmers’ adoption of AGPTs in terms of both intrinsic perceptions and external regulation.

4.2. Limitations and Future Recommendations

The limitations of this paper are the wide scope of greening agriculture and the fact that the green production transition of farmers is not only about the use of AGPTs. In addition, the number of farmers who should be greening their production is difficult to estimate. The results would be more scientific and generalizable if a large amount of data from different countries, regions, and crop types was available to provide empirical evidence. In addition, we only focused on the current state of adoption of AGPTs among cotton farmers, while ignoring the greening of the agricultural market, which is more likely to ultimately contribute to farmers’ green production transition. These limitations are the result of unexpected challenges that arose when we initially chose how to design the study, and it would be of great research interest if they could be overcome.

Currently, the food safety issue clearly exposes all the vulnerabilities and shortcomings caused by surface pollution of agricultural origin, while at the same time, it highlights our mistakes and ignorance of the catastrophes caused by environmental pollution and the consequences of unsustainable human activities, increasing the likelihood of the transmission of viruses to human beings [77]. In order to feed a larger population on a very small amount of arable land, to supply safer food for human beings, and to solve the problem of a lack of food and clothing in poor areas, research in the field of green development and sustainable development of agriculture should be deepened. On this basis, future research can be expanded in the following two aspects: First, the research content can be expanded to the green production transition of farmers, acknowledging that different AGPTs have different attributes, but in the context of the aging labor force, the tendency of farmers to adopt AGPTs may be different. Secondly, there is a fundamental difference in the production motivation of the two types of farmers, namely, those who “farm for a living” and those who “farm for a profession”, and the question of whether different production goals affect the adoption of AGPTs by large-scale farmers and small-scale farmers should be investigated. Further research on these issues is needed.

5. Conclusions

5.1. Conclusions

In recent years, people’s demand for agricultural products has gradually changed from “quantity” to “quality”, and the quality and safety of agricultural products have increasingly become a key factor in promoting the green development of agriculture. To steadily improve the quality and safety of agricultural products and promote green agricultural production and ecological environmental protection, the most important thing is to rationally use AGPTs in agricultural production and management in order to solve the problem of large-scale agricultural pollution while at the same time promoting the sustainable development of agriculture, otherwise, people’s health will be greatly threatened. The relevant literature in recent years has pointed out that the advocacy of green and sustainable development in agriculture has enhanced farmers’ perceptions of green production, and the impact of this initiative has been greatly increased, especially through the educational guidance of cooperatives. At the same time, food safety issues have clearly exposed all the vulnerabilities and shortcomings caused by agricultural surface pollution, encouraging both governments and farmers to move away from traditional agricultural production methods that cause greater pollution. Therefore, based on microdata from 502 cotton farmers in Xinjiang, China, this paper combines some of the indicators adopted in previous related studies to explain more comprehensively how cooperatives can influence the adoption of agricultural AGPTs by changing the green perceptions of cotton farmers, to provide useful policy recommendations for tackling the dilemmas of adopting AGPTs, and to facilitate the transition of cotton farmers to green production and the realization that it is a key aspect of agricultural production and management. Through empirical analysis, using models such as the propensity score matching, mediation effect, and coefficient of variation methods, the results showed the following:

(1)

Guiding cotton farmers to participate in cooperatives is the right choice to enhance their adoption of AGPTs. These technologies vary in importance to cotton farmers in the order of soil testing and fertilizer application technology, green prevention and control technology, deep pine tillage technology, water-saving irrigation technology, new variety technology, and straw-returning technology, but the magnitude of the enhancement is affected by the heterogeneity of the production chain.

(2)

The cooperatives significantly enhanced cotton farmers’ knowledge of green production through their education and training functions and, combined with the government’s environmental regulations, strengthened cotton farmers’ behavior of adopting AGPTs. Therefore, green production cognition and external environmental regulations are important mediators of the effect of cotton farmers’ adoption of green production technology.

5.2. Policy Recommendations

In the dual context of green and sustainable development in agriculture, agricultural policies around the world must respond positively in this regard while combatting some of the real dilemmas. For example, governments have not undertaken sufficient action to raise farmers’ awareness of the need to protect the ecological environment; some farmers do not have a good understanding of the intrinsic relationship between green production and the ecological environment, and there is no targeted education or programs for farmers on green production. Secondly, because farmers are also vulnerable, when the government puts farmers as the main consideration of policy design, they often ignore whether farmers have strong executive power; this is especially important for poor farmers who are far away from the city and highlights the need for cooperatives to serve as an educational guide. Finally, it has been proven that cooperatives are beneficial to green development, but small farmers adhering to the perspective of the “rational economic man” have a strong sense of risk and need to be incentivized and constrained by the double guarantee of the policy in order to transition. Based on the above phenomenon, this paper puts forward the following policy recommendations on the basis of the literature review and empirical analysis:

(1)

In the process of promoting the green transformation of agriculture, we should fully recognize the role of cotton farmers in promoting green production awareness. Specifically, the cooperative, in order to improve cotton production, should promote cotton farmers’ environmental awareness to endorse cotton farmers’ green production, as well as strengthen green production in rural areas and promote education to improve cotton farmers’ ecological civilization literacy. In addition, support should be given to the implementation of AGPTs to encourage science and technology investment in the green development model, as well as focusing on the top-level design and construction of green development in the countryside as a mechanism to enhance rural green development. In order to increase the income of cotton farmers, constraint-based and incentive-based environmental regulations should be offered to enhance the cotton farmers’ green production, while actively applying for more policy subsidies and incentives to support them.

(2)

Farmers should be actively encouraged to join cooperatives, giving full play to the role of farmers’ professional cooperatives as a model. Farmers’ professional cooperatives can enhance the market premium and avoid market risks by using AGPTs to effectively reduce input costs and equalizing the effect of risk—for example, making tea farmers, as a collective, bear the risk is far less dangerous than a single tea farmer having to face the risks of the market alone. Tea farmers should be actively encouraged to participate in farmers’ cooperatives; through the cooperative training guide, cooperatives should actively promote the adoption of AGPTs and encourage others to try new technologies, allowing other activities to be implemented more smoothly, improving the participation of farmers in the cooperative, playing a leading role in green production, and promoting the transformation of farmers toward green production.

(3)

Support for professional farmers’ cooperatives should be increased to promote the organized implementation of AGPTs. The development of farmers’ “professional” cooperatives must be in accordance with the requirements of modern agricultural development and further improve the use of cooperatives, implementing standardized operating conditions to support the standardization of the process. Therefore, there is a need to increase the financial support for farmers’ professional cooperatives and to ease the financial pressure faced by farmers’ professional cooperatives in the process of providing diversified services; this also requires support for the establishment of new types of rural financial organizations, such as village and town banks and rural capital mutual aid societies, in order to increase the financial support for cooperatives. And a market-oriented approach to the establishment of a rural revitalization fund should be explored, focusing on supporting the development of agricultural modernization and development, to attract more tea farmers to join the farmers’ professional cooperatives so that the cooperatives can be organized to start to promote AGPTs and enhance the effect of the transformation of farmers toward green production.