How Can Science and Technology Backyards Lead Smallholder Farmers Toward Green Transformation? An Evolutionary Game Analysis of a Tripartite Interaction

Abstract

The green transition of smallholder farmers is a critical component in expanding the scale of green agricultural production in China. This research investigates how Science and Technology Backyards facilitate the ecological transformation process for small-scale agricultural producers by developing a three-party evolutionary game framework that incorporates Science and Technology Backyards (STBs), smallholder farmers, and research institutions. The main findings are as follows: (1) Under specific parameter conditions, the system converges to two stable equilibrium points: (0,0,0), where none of the three parties engage in cooperation, and (1,1,1), where full participation and collaboration among all parties are achieved. (2) Science and Technology Backyards exhibit a strong tendency to avoid bearing research costs and demonstrate high sensitivity to economic returns, indicating a clear preference for profit maximization. (3) Research institutes can effectively reduce the cost of technology trials through cooperation with Science and Technology Backyards; however, excessively high trial costs significantly weaken the willingness to collaborate. This study provides a scientific basis for decision-making by stakeholders involved in Science and Technology Backyard initiatives and offers theoretical support for advancing the green transformation of smallholder farmers through the Science and Technology Backyard.

1. Introduction

As a crucial approach to agricultural advancement, eco-friendly farming practices are essential in tackling significant issues like resource scarcity, environmental deterioration, and the degradation of farming ecosystems [1]. China has always been committed to promoting sustainable development, but sustainable development in the agricultural sector has been difficult to achieve in depth [2]. Therefore, how to promote green agricultural production is an important issue that needs to be addressed urgently for China’s sustainable development [3]. In recent years, the Chinese government has actively advanced the green development of agriculture. In 2024, the Chinese government’s Agricultural Document No. 1, titled “Opinions on Learning from and Applying the Experience of the ‘Thousand Villages Demonstration and Ten Thousand Villages Improvement’ Project to Effectively Promote the Comprehensive Revitalization of Rural Areas”, emphasized the importance of prioritizing the greening of agriculture. It called for accelerating the development of a modern rural industrial system and transforming agriculture into a modernized major industry [4]. In 2025, the Chinese government issued the “Plan for the Comprehensive Revitalization of Rural Areas (2024–2027)”, which emphasized the need to promote the deep integration of rural ecological civilization, advance the green transformation of development models, and accelerate the green and low-carbon development of agriculture [5].

In the context of China’s agricultural modernization, the ecological transition of small-scale farming households demonstrates greater significance in promoting sustainable agricultural practices compared with alternative agricultural enterprise models. On the one hand, smallholder farmers represent the fundamental unit of agricultural production in China. According to data from the third agricultural census, China has 230 million farmers, with approximately 210 million households operating less than 10 acres of arable land. The “big country, small farmers” agricultural structure is unlikely to change fundamentally in the short term, and the smallholder economy is expected to remain the dominant model of agriculture in China for the foreseeable future [6]. On the other hand, the green transformation is an essential pathway for the high-quality development of smallholder farming. The application of green production technologies can optimize the allocation of agricultural inputs, enhance the quality of agricultural products, ensure production safety, and strengthen the core competitiveness of farmers’ products [7]. Therefore, the green transformation is a crucial pathway for smallholder farmers to achieve sustainable agricultural production and progress toward long-term development [8].

To promote the green transformation of smallholder farmers, Quzhou County in Handan City, Hebei Province, China, initiated the establishment of a Science and Technology Backyard in 2009. This initiative has yielded significant results in addressing the technical challenges of green production for small-scale farmers [9]. As of 2025, more than 1800 STBs have been established across China, spanning all 31 provinces, autonomous regions, and municipalities, playing a crucial role in advancing the green development of agriculture [10,11]. In terms of operational mechanisms, the STBs, in collaboration with research institutions, have developed new technologies tailored to the specific needs of farmers, combining top–down and bottom–up measures to help farmers solve green production problems [12]. In terms of the dissemination mechanism, the STB has formed an operational situation in which agricultural green technology and science dissemination are mutually integrated, and the construction of a science dissemination scenario accelerates the innovation and diffusion of green technology through the pluralistic linkage of multiple participants [13]. There is currently no unified theoretical consensus among scholars regarding the analysis of the dynamics of the green transition in smallholder production. One perspective is that the limited availability of technology is the primary factor constraining the green transformation of smallholder farmers. Although Chinese universities and research institutes have conducted extensive research on agricultural green technology, the focus of researchers is often more on academic value than on market demand. As a result, most research findings remain at the laboratory stage, with a significant gap between their industrialization and practical application [14,15,16]. Science and Technology Backyards, primarily established in rural areas, help address the disconnect between green technology research and its practical application, as well as the misalignment between talent development and societal needs. These issues have constrained scientific and technological innovation, the transformation of research achievements, and the green development of agriculture [17]. Another perspective is that the insufficient demand for technology is a key factor hindering the widespread adoption of green agricultural practices. Smallholder farmers, as individual agricultural business entities, often face limitations in economic strength and risk resilience. They rely heavily on long-term government financial support and tend to passively accept new technologies, resulting in a relatively low level of initiative in adopting green agricultural production technologies [18]. Science and Technology Backyards promote green technologies through contract farming or outgrower schemes, aiming to increase farmers’ willingness to adopt these sustainable practices [19].

Currently, there is limited research on the green transition of smallholder farmers. Existing studies primarily focus on the green transformation of agribusiness, often examining it from the perspective of digital–physical integration and its synergistic development. From the perspective of Digital and Real Integration, Digital Intelligence Services view digital information as a key production factor [20]. By utilizing technologies such as machine learning [21], artificial intelligence [22], the Internet of Things [23], and blockchain [24], these services facilitate visual representation, digital design, and intelligent information management throughout the environmental and production–marketing processes of agricultural products, thereby promoting green transformation [25]. From the perspective of synergistic development between greening and digitalization, digital technologies, such as blockchain, empower the advancement of green practices by effectively reducing the consumption of natural resources in traditional industries [26]. These technologies drive digital transformation through the precise management of carbon emissions and other sustainability measures [27]. However, this also raises higher demands for the development and application of digital technologies [28,29]. The drivers of green transformation in agribusiness are similar to those in smallholder farming; however, they differ in terms of the internal and external dynamics that influence their respective processes. Agribusiness is primarily driven by market competition and the demand for productivity, relying predominantly on internal organizational factors [30], with the government playing a supportive role [31,32]. In contrast, smallholder farmers often lack sufficient endogenous motivation for green transformation due to their small production scale and limited economic capacity. They are heavily reliant on external resources and environmental factors, which makes it challenging for them to access green production technologies [33,34]. In summary, the differences between agribusinesses and smallholders in the drivers of green transformation suggest that while the greening strategies of agribusinesses provide valuable insights, they cannot be directly replicated by smallholders. Therefore, the pathways for green transformation in smallholder farming must be explored and developed specifically.

Existing research has offered valuable insights into achieving the green transition in agriculture; however, several gaps remain. Firstly, with regard to research methodology, few existing studies have employed evolutionary game theory to analyze the green transition of smallholder farmers. The integration of game-theoretic principles with evolutionary dynamics forms the foundation of evolutionary game theory, enabling comprehensive analysis of strategic adaptations and localized equilibrium states in group interactions across temporal dimensions. This theoretical framework demonstrates particular efficacy in investigating complex socio-technical system transformations. Secondly, current academic research has largely overlooked the significance of emerging agricultural service organizations, particularly the STB initiative. These innovative platforms effectively consolidate essential social capital from academic institutions, research organizations, corporate entities, and regional administrations. Their operational significance lies in optimizing agricultural productivity, augmenting rural income levels, and facilitating sustainable development in agricultural regions. Thirdly, with regard to research content, relatively few studies have focused on the role of Science and Technology Backyards in driving the green transformation of smallholder farmers. Existing research primarily focuses on the role of new agricultural management entities in driving the green development of smallholder farmers and the benefit linkage mechanisms between the two [35]. However, Science and Technology Backyards, as emerging platforms for agricultural technology research, development, and promotion, serve as a bridge for the communication of green technologies between research institutes and smallholder farmers, thereby facilitating the adoption of green production practices among smallholder farmers. Based on this, this study poses the following questions: (1) What is the evolutionary stable strategy (ESS) in a dynamic replication system composed of STBs, smallholder farmers, and research institutions? (2) Can the collaboration between STBs and research institutions effectively stimulate the green transformation of smallholder farmers? (3) What are the specific mechanisms of interaction among the factors influencing the behavior of STBs, smallholder farmers, and research institutions? To address these questions, this study establishes the following research propositions: (1) Methodologically, the investigation employs an evolutionary game framework, analyzing the interactive dynamics among three primary stakeholders: STBs, smallholder farmers, and academic research institutions. (2) This study focuses on Science and Technology Backyards as the primary research subject, examining how they promote the green transformation of smallholder farmers and aiming to identify an effective pathway for this transition. (3) In terms of research content, this study specifically investigates the impact of the behaviors of Science and Technology Backyards and research institutes on the green transition of smallholder farmers, analyzing their dynamic evolution through scenario analysis. The aim is to clarify the role of the technology supply and demand mechanism in this process and to establish a coordination mechanism to support the green transition of smallholder farmers.

This article is divided into several sections to systematically present the research content. The Section 2 establishes the theoretical framework and identifies key stakeholders involved in the game scenario. Moving forward, the Section 3 elaborates on the modeling process, including the assumption formulation, analytical procedures, outcome synthesis, and the examination of diverse equilibrium scenarios along with their respective stability criteria. Subsequently, the Section 4 demonstrates computational simulations and conducts parametric sensitivity assessments. The Section 5 evaluates the research outcomes through both conceptual and applied lenses. Concluding this paper, the Section 6 synthesizes key insights and proposes actionable recommendations.

2. Theoretical Basis

2.1. Application of Evolutionary Game

As a mathematical modeling approach, game theory examines the decision-making processes of rational agents in scenarios characterized by strategic interdependence, where each participant’s choices are influenced by the actions of other decision-makers [36]. By contrast, evolutionary game theory is a branch of game theory that integrates concepts of evolution and adaptation from biology to study the evolution of interactions among individuals within a population [37]. Unlike traditional game theory, evolutionary game theory challenges the assumption of complete rationality and does not require participants to be fully rational or for conditions to be entirely informative [38].

The three-party evolutionary game primarily involves a payoff matrix, three replication dynamic equations, and one or more evolutionary stable strategies. The payoff matrix represents the returns earned by different participants based on the strategies they employ. Replication dynamic equations describe the propagation and evolution of various strategies within a population, modeling how the relative frequency of these strategies changes over time [39]. An evolutionary stable strategy is one that, if dominant within a population, is resistant to being displaced by alternative strategies, even if new strategies emerge [40].

Unlike evolutionary game theory, which focuses on the evolution of individual strategies and the dynamic changes in group behavior, replicator dynamics is a methodology and mathematical modeling technique used to study the nonlinear behavior of complex systems over time [41]. It is centered on analyzing the interactions between various elements within the system and uncovering the underlying logic driving the system’s evolution [42].

The green transition of smallholders involves complex interactions and adaptation among various participants. Evolutionary game theory offers a valuable framework for studying the green transition of smallholders, enabling an understanding of how different participants make decisions in diverse social environments. Within this framework, the decision of whether a smallholder adopts a green transition can be viewed as the choice between two distinct strategies, each representing a different business model. The strategies of multiple participants interact, forming a variety of strategy combinations. Analyzing the dynamics between these strategies allows for the assessment of their adaptability to the environment, thereby aiding participants in making optimal decisions. Thus, evolutionary game theory serves as an effective analytical tool for the green transition of smallholder farmers, helping participants understand the interplay and adaptation challenges between different decisions and facilitating the development of optimal decision-making strategies.

2.2. Major Players in This Game

Agricultural green development is currently one of the most pressing topics, offering significant potential for growth and development [43]. Promoting the green transformation of smallholder farmers not only fosters the high-quality development of the agricultural economy and accelerates the modernization of China’s agriculture but also offers valuable insights for the widespread implementation of green production practices across the agricultural sector [44]. Therefore, the green transition is a key issue for the high-quality and sustainable development of agriculture [45]. This paper examines the collaborative role of STBs and research institutes in promoting the green transition of smallholder farmers.

Science and Technology Backyards: Approved by the China Rural Special Technology Association, the STB has the capacity to integrate high-quality social resources, including social organizations and research institutes, with the ultimate goal of promoting the green transformation of smallholder farmers. Under government policy guidance, STBs face a critical decision regarding the positive or negative fulfillment of their responsibilities. The positive fulfillment of responsibilities by STBs can stimulate research institutes to participate in the construction of STBs and increase farmers’ enthusiasm for applying green technologies. First, STBs cooperate with research institutes to jointly bear R&D costs and pilot test expenses, thereby reducing the economic pressure on research institutes. Second, they promote technology to farmers, empowering them to increase production and income through science and technology while also realizing the social value of the technology. Green transformation is an inevitable trend in agricultural development [46], and STBs can significantly accelerate this process. However, since these institutes have market-driven attributes rather than functioning as public institutions, their continued role in promoting green transformation depends on the synergistic mechanisms established within the STB model. Therefore, STBs must engage in strategic interactions with smallholder farmers and research institutes to identify optimal decision-making pathways.

Smallholder farmers: As the fundamental unit of agricultural production in China, the adoption of green technologies by smallholder farmers directly influences the effectiveness of the agricultural green transformation. As a for-profit social service entity at the forefront of agricultural production, the Science and Technology Backyard plays a key role in promoting the green transformation of agriculture [47]. In this process, the availability of production technology and market demand for green agricultural products are critical factors influencing smallholder farmers’ decisions to adopt green production. When the benefits of applying technology meet farmers’ expectations, smallholder farmers will adopt the technology for green production. In this case, STBs can recover their initial investments and continue to deepen cooperation with research institutes, stimulating innovation in green agricultural technology. However, due to the uncertainty of realizing the benefits of green technology applications, smallholder farmers are more inclined to seek lower technology service costs and higher agricultural product sales revenue to offset potential risks. This cost–benefit consideration directly influences their willingness to adopt green technologies.

Research institutes: As developers and providers of technical services, research institutes face the decision of whether to actively participate in the establishment of STBs. The active participation of research institutes can stimulate farmers’ enthusiasm for technology application and help STBs realize their benefits. For smallholder farmers, research institutes can develop technologies based on actual production needs and utilize their own resources to expand sales channels, ensuring the supply of technologies and the benefits of their application. For STBs, cooperation with research institutes can help them obtain technology patents and generate income from technology promotion. Although STBs can provide research institutes with laboratories and experimental fields, share R&D costs, and reduce the economic pressure of research work, their specific effectiveness is difficult to determine in the short term, which poses certain challenges for the decision-making of research institutes. Therefore, research institutes must engage in in-depth interactions with STBs and smallholder farmers to identify the optimal course of action.

Science and Technology Backyards, smallholder farmers, and research institutes make decisions based on a cost–benefit analysis. Despite sharing common interests in specific areas, their collaborative decisions stem from distinct reasoning processes, thereby revealing an underlying strategic dynamic or game-theoretic interaction between the parties involved. This game relationship is illustrated in Figure 1.

3. Calculation

3.1. Model Assumptions

In light of the ongoing ecological transformation process among smallholder farmers in China, several hypotheses have been developed concerning the behavioral patterns and motivational factors of stakeholders within the evolutionary framework.

Assumption 1. 

In conditions of incomplete information and limited rationality, Science and Technology Backyards, smallholder farmers, and research institutes each seek to maximize their own interests. These three parties, characterized by information asymmetry, engage in an evolutionary game.

Assumption 2. 

Science and Technology Backyards, smallholder farmers, and research institutes make strategic decisions based on their individual objectives and preferences. Science and Technology Backyards have the option to actively engage in their designated duties or choose not to perform them. Assume that the probability of the Science and Technology Backyard’s positive compliance is x, while the probability of negative compliance is 1 − x. One strategic option for smallholders is to engage in green production, with a probability of y, while the alternative option is to engage in traditional production, with a probability of 1 − y. One strategic option for the research institute is to cooperate positively, with a probability of z, while the alternative is to cooperate negatively, with a probability of 1 − z.

Assumption 3. 

When Science and Technology Backyards actively fulfill their responsibilities, they must establish laboratories and experimental fields to provide researchers with the necessary resources for their work. Cr1 represents the research costs incurred by research institutes, which are not fully borne by them, and a denotes the cost-bearing coefficient of STBs. Prior to the promotion of new technologies, the STBs must conduct pilot testing of research outcomes. Pt represents the cost associated with these pilot tests. Upon completion of the pilot test, the STB promotes the technology by selling cultivated seeds, leasing agricultural equipment, and engaging in other related activities. Cs1 represents the operating costs associated with the technology promotion, while Es1 denotes the operating income generated by the STB. Operating revenue is also highly correlated with the radiation range of STBs. Due to the high mobility of China’s rural population and the high turnover of rural land [48,49], this study uses the service coverage of STBs as the basis for measuring the radiation range Ra. When an STB fails to fulfill its duties, it is not responsible for the costs of scientific research, does not need to invest in pilot testing, and does not incur operating costs or generate operating income. Regardless of whether the STB actively fulfills its responsibilities, the government will allocate a certain amount of project funding to support the STB. Cg represents the government’s financial support to the STB. Since the STB is currently in its initial stage, there is no clear classification of STBs, and project funds are distributed by the government in equal amounts or applied for independently by the STB.

Assumption 4. 

Smallholder farmers can derive benefits from the STB. On the one hand, the STB can provide the agricultural resources required for green production. Cf1 represents the cost of agricultural resources needed for traditional production, while Cf2 denotes the cost of agricultural resources required for green production. On the other hand, research institutes, with their abundant resources, can expand sales channels for smallholder farmers. When the agricultural products produced by farmers meet the requirements of dealers for centralized purchasing, Ef1 represents the income from sales through traditional channels; Ef2 represents the income from sales through the sales channels of research institutes, and b denotes the coefficient of yield enhancement from green production by farmers. Due to the knowledge level and age structure of smallholder farmers in China [50], information search costs have become one of the important reasons for smallholders’ resistance to green technology adoption [51]. When the Science and Technology Backyard declines to carry out pilot testing and technology diffusion, smallholder farmers must independently search for the necessary agricultural resources for green production. This study uses the search cost Sc to represent the associated costs.

Assumption 5. 

When the research institute opts for positive cooperation, it will conduct scientific research within the Science and Technology Backyard to address agricultural production challenges. In this model, (1 − a)Cr1 represents the portion of the research cost to be borne by the research institute. Research institutes will also provide assistance when Science and Technology Backyards conduct pilot tests on research outcomes. In such cases, the costs of the pilot tests will be shared by both parties. Additionally, research institutes will leverage their resources to assist smallholder farmers in developing sales channels. Cr2 represents the cost associated with developing these sales channels. Most of the research institutions cooperating with STBs are university research institutes or R&D departments of state-owned enterprises. As their operation and development mainly rely on government financial support and policy guidance, they are highly sensitive to social value [52]. When research outcomes are applied to production, their value is realized. Er2 represents the social value generated by the implementation of these research results. When research institutes engage in negative cooperation, small science and technology institutes must independently conduct their own pilot tests, and farmers are responsible for managing their own sales. Regardless of whether the research institute actively cooperates, once the Science and Technology Backyard project is approved, the research institute will receive a research grant. Er1 represents the funding allocated for the research institute’s project.

Following the aforementioned premises, Figure 2 illustrates the associated parameters. The red, green, and yellow arrows represent the impact on the returns of different decisions made by STBs, smallholder farmers, and research institutes, respectively, while the blue arrows represent the impact of different decisions on the strategy choices of the players.

Table 1. The benefit matrix of the participants for smallholder farmers green production (y).

		Science and Technology Backyards
		Positive Compliance (x)	Negative Compliance (1 – x)
Research institutes	Positive cooperation (z)	Cg − aCr1 + Ra(Es1 − Cs1), bPrEf2 − Cf2, Er1 + Er2 − (1 − a)Cr1 − Cr2.	Cg, bPrEf2 − Cf2 − Sc, Er1 + Er2 − Cr1 − Cr2 − Pt.
	Negative cooperation (1 − z)	Cg − aCr1 − Pt + Ra(Es1 − Cs1), bPrEf1 − Cf2, Er1 − (1 − a)Cr1.	Cg, bPrEf1 − Cf2 − Sc, Er1 − Cr1.

and

Table 2. The benefit matrix of the participants for smallholder farmers’ traditional production (1 − y).

		Science and Technology Backyards
		Positive Compliance (x)	Negative Compliance (1 − x)
Research institutes	Positive cooperation (z)	Cg − aCr1, PrEf1 − Cf1, Er1 − (1 − a)Cr1 − Cr2.	Cg, PrEf1 − Cf1, Er1 − Cr1 − Cr2 − Pt.
	Negative cooperation (1 − z)	Cg − aCr1 − Pt, PrEf1 − Cf1, Er1 − (1 − a)Cr1.	Cg, PrEf1 − Cf1, Er1 − Cr1.

display the payoff matrix involving the Science and Technology Backyards, smallholder farmers, and research institutions.

3.2. Expected Benefits and Replication Dynamic Equations

Utilizing the principles of evolutionary game theory and the established payoff matrix, it is possible to determine the anticipated and mean payoffs for various stakeholders, including Science and Technology Backyards, smallholder farmers, and research institutions, across different decision-making scenarios. The detailed analysis is presented below.

The expected benefits of positive and negative compliance by Science and Technology Backyards are denoted as E11 and E12, respectively, while the average benefit is represented by E1. These can be expressed by the following Equations (1)–(3):

$$\begin{gathered} \mathrm{E}11=\mathrm{y}\mathrm{z}[\mathrm{C}\mathrm{g}-\mathrm{a}\mathrm{C}\mathrm{r}\mathrm{l}+\mathrm{R}\mathrm{a}(\mathrm{E}\mathrm{s}1-\mathrm{C}\mathrm{s}1\left)\right]+\mathrm{y}(1-\mathrm{z})[\mathrm{C}\mathrm{g}-\mathrm{a}\mathrm{C}\mathrm{r}1-\mathrm{P}\mathrm{t}+\mathrm{R}\mathrm{a}(\mathrm{E}\mathrm{s}1-\mathrm{C}\mathrm{s}1\left)\right] \\ +(1-\mathrm{y})\mathrm{z}(\mathrm{C}\mathrm{g}-\mathrm{a}\mathrm{C}\mathrm{r}1)+(1-\mathrm{y})(1-\mathrm{z})(\mathrm{C}\mathrm{g}-\mathrm{a}\mathrm{C}\mathrm{r}1-\mathrm{P}\mathrm{t}) \end{gathered}$$

(1)

$$\mathrm{E}12=\mathrm{y}\mathrm{z}\mathrm{C}\mathrm{g}+\mathrm{y}(1-\mathrm{z})\mathrm{C}\mathrm{g}+(1-\mathrm{y})\mathrm{z}\mathrm{C}\mathrm{g}+(1-\mathrm{y})(1-\mathrm{z})\mathrm{C}\mathrm{g}$$

(2)

$$\mathrm{E}1=\mathrm{C}\mathrm{g}-\mathrm{x}\mathrm{P}\mathrm{t}-\mathrm{x}\mathrm{a}\mathrm{C}\mathrm{r}\mathrm{l}+\mathrm{x}\mathrm{z}\mathrm{P}\mathrm{t}-\mathrm{x}\mathrm{y}\mathrm{R}\mathrm{a}\mathrm{C}\mathrm{s}\mathrm{l}+\mathrm{x}\mathrm{y}\mathrm{R}\mathrm{a}\mathrm{E}\mathrm{s}1$$

(3)

The expected benefits of green and traditional production by smallholder farmers are denoted as E21 and E22, respectively, while the average benefit is represented by E2. These can be expressed by the following Equations (4)–(6):

$$\begin{gathered} \mathrm{E}21=\mathrm{x}\mathrm{z}(\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}2-\mathrm{C}\mathrm{f}2)+\mathrm{x}(1-\mathrm{z})(\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1-\mathrm{C}\mathrm{f}2)+(1-\mathrm{x})\mathrm{z}(\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}2-\mathrm{C}\mathrm{f}2-\mathrm{S}\mathrm{c}) \\ +(1-\mathrm{x})(1-\mathrm{z})(\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1-\mathrm{C}\mathrm{f}2-\mathrm{S}\mathrm{c}) \end{gathered}$$

(4)

$$\mathrm{E}22=\mathrm{x}\mathrm{z}(\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1-\mathrm{C}\mathrm{f}1)+\mathrm{x}(1-\mathrm{z})(\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1-\mathrm{C}\mathrm{f}1)+(1-\mathrm{x})\mathrm{z}(\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1-\mathrm{C}\mathrm{f}1)+(1-\mathrm{x})(1-\mathrm{z})(\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1-\mathrm{C}\mathrm{f}1)$$

(5)

$$\mathrm{E}2=\mathrm{y}\mathrm{C}\mathrm{f}1-\mathrm{C}\mathrm{f}1-\mathrm{y}\mathrm{C}\mathrm{f}2-\mathrm{y}\mathrm{S}\mathrm{c}+\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1-\mathrm{y}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1+\mathrm{x}\mathrm{y}\mathrm{S}\mathrm{c}+\mathrm{y}\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}\mathrm{l}-\mathrm{y}\mathrm{z}\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}\mathrm{l}+\mathrm{y}\mathrm{z}\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}2$$

(6)

The expected benefits of positive and negative cooperation of research institutes are denoted as E31 and E32, respectively, while the average benefit is represented by E3. These can be expressed by the following Equations (7)–(9):

$$\begin{gathered} \mathrm{E}31=\mathrm{x}\mathrm{y}[\mathrm{E}\mathrm{r}1+\mathrm{E}\mathrm{r}2-(1-\mathrm{a})\mathrm{C}\mathrm{r}1-\mathrm{C}\mathrm{r}2]+\mathrm{x}(1-\mathrm{y})[\mathrm{E}\mathrm{r}1-(1-\mathrm{a})\mathrm{C}\mathrm{r}1-\mathrm{C}\mathrm{r}2] \\ +(1-\mathrm{x})\mathrm{y}(\mathrm{E}\mathrm{r}1+\mathrm{E}\mathrm{r}2-\mathrm{C}\mathrm{r}1-\mathrm{C}\mathrm{r}2-\mathrm{P}\mathrm{t})+(1-\mathrm{x})(1-\mathrm{y})(\mathrm{E}\mathrm{r}1-\mathrm{C}\mathrm{r}1-\mathrm{C}\mathrm{r}2-\mathrm{P}\mathrm{t}) \end{gathered}$$

(7)

$$\mathrm{E}32=\mathrm{x}\mathrm{y}[\mathrm{E}\mathrm{r}1-(1-\mathrm{a}\left)\mathrm{C}\mathrm{r}1\right]+\mathrm{x}(1-\mathrm{y})[\mathrm{E}\mathrm{r}1-(1-\mathrm{a}\left)\mathrm{C}\mathrm{r}1\right]+(1-\mathrm{x})\mathrm{y}(\mathrm{E}\mathrm{r}1-\mathrm{C}\mathrm{r}1)$$

(8)

$$\mathrm{E}3=\mathrm{E}\mathrm{r}\mathrm{l}-\mathrm{C}\mathrm{r}\mathrm{l}-\mathrm{z}\mathrm{C}\mathrm{r}2-\mathrm{z}\mathrm{P}\mathrm{t}+\mathrm{x}\mathrm{a}\mathrm{C}\mathrm{r}1+\mathrm{y}\mathrm{z}\mathrm{E}\mathrm{r}2+\mathrm{x}\mathrm{Z}\mathrm{P}\mathrm{t}$$

(9)

Based on the principles of evolutionary game theory, the replication dynamics for Science and Technology Backyards, smallholder farmers, and research institutions can be established through their anticipated benefits, as demonstrated in Equations (10)–(12).

$$\mathrm{F}\left(\mathrm{x}\right)=\mathrm{x}(\mathrm{x}-1)(\mathrm{P}\mathrm{t}+\mathrm{a}\mathrm{C}\mathrm{r}1-\mathrm{z}\mathrm{P}\mathrm{t}+\mathrm{y}\mathrm{R}\mathrm{a}\mathrm{C}\mathrm{s}\mathrm{l}-\mathrm{y}\mathrm{R}\mathrm{a}\mathrm{E}\mathrm{s}1)$$

(10)

$$\mathrm{F}\left(\mathrm{y}\right)=\mathrm{y}(1-\mathrm{y})(\mathrm{C}\mathrm{f}\mathrm{l}-\mathrm{C}\mathrm{f}2-\mathrm{S}\mathrm{c}+\mathrm{x}\mathrm{S}\mathrm{c}-\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}\mathrm{l}+\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}\mathrm{l}-\mathrm{z}\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}\mathrm{l}+\mathrm{z}\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}2)$$

(11)

$$\mathrm{F}\left(\mathrm{z}\right)=\mathrm{z}(\mathrm{z}-1)(\mathrm{C}\mathrm{r}2+\mathrm{P}\mathrm{t}-\mathrm{y}\mathrm{E}\mathrm{r}2-\mathrm{x}\mathrm{P}\mathrm{t})$$

(12)

The replicator dynamics framework elucidates how participants in a game can emulate prevailing strategies, facilitating the decomposition of systemic behaviors and mechanisms into discrete elements. This approach provides a clear depiction of the interconnections among various system components. Consequently, a replicator dynamics flow diagram is employed to explain the mechanisms through which Science and Technology Backyards and research institutes influence smallholder farmers to adopt the green production strategy, as shown in Figure 3. The active cooperation between STBs and research institutes has reduced the technology search costs of smallholder farmers, increased their technological gains, and, thereby, changed their enthusiasm for green production. However, the external uncertainties that follow are also important factors affecting the decision-making of smallholder farmers.

3.3. Analysis of Participants’ Local Stable Strategies

At the point where the replication dynamic equation reaches 0, the variables x, y, and z become time-invariant, indicating that all participants have converged to their optimal strategies. Based on the stability theory of differential equations, a system achieves equilibrium when both the equation itself equals 0 and its first-order derivative is negative. Therefore, the stability of Science and Technology Backyards, smallholder farmers, and research institutes is analyzed as follows.

Based on the replication dynamic equation f(x), the following conclusions can be drawn for Science and Technology Backyards:

When $\mathrm{y}\mathrm{R}\mathrm{a}(\mathrm{E}\mathrm{s}1-\mathrm{C}\mathrm{s}1)>\mathrm{a}\mathrm{C}\mathrm{r}1+(1-\mathrm{z})\mathrm{P}\mathrm{t}$, ${{\displaystyle \frac{\mathrm{d}\left(\mathrm{f}\right(\mathrm{x}\left)\right)}{\mathrm{d}\mathrm{x}}}|}_{\mathrm{x}=1}<0$, ${{\displaystyle \frac{\mathrm{d}\left(\mathrm{f}\right(\mathrm{x}\left)\right)}{\mathrm{d}\mathrm{x}}}|}_{\mathrm{x}=0}>0$. Therefore, x = 1 represents an evolutionary stable point for Science and Technology Backyards, indicating that they choose to comply positively.

When $\mathrm{y}\mathrm{R}\mathrm{a}(\mathrm{E}\mathrm{s}1-\mathrm{C}\mathrm{s}\mathrm{l})=\mathrm{a}\mathrm{C}\mathrm{r}\mathrm{l}+(1-\mathrm{z})\mathrm{P}\mathrm{t}$, $\mathrm{F}\left(\mathrm{x}\right)=0$. At this point, the benefits of positive and negative compliance by Science and Technology Backyards are equal, and all values are at an evolutionary steady state.

When $\mathrm{y}\mathrm{R}\mathrm{a}(\mathrm{E}\mathrm{s}1-\mathrm{C}\mathrm{s}1)<\mathrm{a}\mathrm{C}\mathrm{r}1+(1-\mathrm{z})\mathrm{P}\mathrm{t}$, ${{\displaystyle \frac{\mathrm{d}\left(\mathrm{f}\right(\mathrm{x}\left)\right)}{\mathrm{d}\mathrm{x}}}|}_{\mathrm{x}=1}>0$, ${{\displaystyle \frac{\mathrm{d}\left(\mathrm{f}\right(\mathrm{x}\left)\right)}{\mathrm{d}\mathrm{x}}}|}_{\mathrm{x}=0}<0$. Therefore, x = 0 represents an evolutionary stable point for Science and Technology Backyards, indicating that they choose to comply negatively.

Based on the replication dynamic equation f(y), the following conclusions can be drawn for smallholder farmers:

When $(1-\mathrm{z})\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1+\mathrm{z}\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}2-\mathrm{C}\mathrm{f}2-(1-\mathrm{x})\mathrm{S}\mathrm{c}>\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}\mathrm{l}-\mathrm{C}\mathrm{f}\mathrm{l}$, ${{\displaystyle \frac{\mathrm{d}\left(\mathrm{f}\right(\mathrm{y}\left)\right)}{\mathrm{d}\mathrm{y}}}|}_{\mathrm{y}=1}<0$, ${{\displaystyle \frac{\mathrm{d}\left(\mathrm{f}\right(\mathrm{y}\left)\right)}{\mathrm{d}\mathrm{y}}}|}_{\mathrm{y}=0}>0$. Therefore, x = 1 represents an evolutionary stable point for smallholder farmers, indicating that they choose green production.

When $(1-\mathrm{z})\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1+\mathrm{z}\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}2-\mathrm{C}\mathrm{f}2-(1-\mathrm{x})\mathrm{S}\mathrm{c}=\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1-\mathrm{C}\mathrm{f}1$, $\mathrm{F}\left(\mathrm{y}\right)=0$. At this point, the benefits of choosing green and traditional production by smallholder farmers are equal, and all values are at an evolutionary steady state.

When $(1-\mathrm{z})\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1+\mathrm{z}\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}2-\mathrm{C}\mathrm{f}2-(1-\mathrm{x})\mathrm{S}\mathrm{c}<\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1-\mathrm{C}\mathrm{f}1$, ${{\displaystyle \frac{\mathrm{d}\left(\mathrm{f}\right(\mathrm{y}\left)\right)}{\mathrm{d}\mathrm{y}}}|}_{\mathrm{y}=1}>0$, ${{\displaystyle \frac{\mathrm{d}\left(\mathrm{f}\right(\mathrm{y}\left)\right)}{\mathrm{d}\mathrm{y}}}|}_{\mathrm{y}=0}<0$. Therefore, x = 0 represents an evolutionary stable point for smallholder farmers, indicating that they choose traditional production.

Based on the replication dynamic equation f(z), the following conclusions can be drawn for research institutes:

When $\mathrm{C}\mathrm{r}2+(1-\mathrm{x})\mathrm{P}\mathrm{t}<\mathrm{y}\mathrm{E}\mathrm{r}2$, ${{\displaystyle \frac{\mathrm{d}\left(\mathrm{f}\right(\mathrm{z}\left)\right)}{\mathrm{d}\mathrm{z}}}|}_{\mathrm{z}=1}<0$, ${{\displaystyle \frac{\mathrm{d}\left(\mathrm{f}\right(\mathrm{z}\left)\right)}{\mathrm{d}\mathrm{z}}}|}_{\mathrm{z}=0}>0$. Therefore, x = 1 represents an evolutionary stable point for the research institutes, indicating that they choose to cooperate positively.

When $\mathrm{C}\mathrm{r}2+(1-\mathrm{x})\mathrm{P}\mathrm{t}=\mathrm{y}\mathrm{E}\mathrm{r}2$, $\mathrm{F}\left(\mathrm{z}\right)=0$. At this point, the benefits of positive and negative cooperation for the research institutes are equal, and all values are at an evolutionary steady state.

When $\mathrm{C}\mathrm{r}2+(1-\mathrm{x})\mathrm{P}\mathrm{t}>\mathrm{y}\mathrm{E}\mathrm{r}2$, ${{\displaystyle \frac{\mathrm{d}\left(\mathrm{f}\right(\mathrm{z}\left)\right)}{\mathrm{d}\mathrm{z}}}|}_{\mathrm{z}=1}>0$, ${{\displaystyle \frac{\mathrm{d}\left(\mathrm{f}\right(\mathrm{z}\left)\right)}{\mathrm{d}\mathrm{z}}}|}_{\mathrm{z}=0}<0$. Therefore, x = 0 represents an evolutionary stable point for the research institutes, indicating that they choose to cooperate negatively.

3.4. Stability Analysis of Evolutionary Systems

Within scenarios characterized by information asymmetry, the evolutionary stable strategies that emerge in asymmetric game theory contexts invariably constitute pure strategy Nash equilibria. The system yields eight distinct pure strategy Nash equilibria when Equations (10)–(12) are simultaneously set to 0: E1 = (0,0,0), E2 = (0,1,0), E3 = (0,0,1), E4 = (0,1,1), E5 = (1,0,0), E6 = (1,1,0), E7 = (1,0,1), E8 = (1,1,1). However, it is challenging to directly determine which of the equilibria are in a steady state. In this research, the local stability of equilibrium points is examined through the application of the Lyapunov indirect approach, with the Jacobian matrix serving as the primary analytical tool. An equilibrium point is considered asymptotically stable only if all eigenvalues of the Jacobian matrix are negative. Building on this, we calculate the Jacobian matrix and its eigenvalues by taking the first-order partial derivatives of F(x), F(y), and F(z) with respect to x, y, and z. This allows us to analyze the trend of evolutionary stabilization among Science and Technology Backyards, smallholder farmers, and research institutes, as illustrated in Equation (13).

$$\begin{gathered} \mathrm{J}=\left(\begin{array}{ccc}{\mathrm{J}}_{11}& {\mathrm{J}}_{12}& {\mathrm{J}}_{13}\\ {\mathrm{J}}_{21}& {\mathrm{J}}_{22}& {\mathrm{J}}_{23}\\ {\mathrm{J}}_{31}& {\mathrm{J}}_{32}& {\mathrm{J}}_{33}\end{array}\right)=\left(\begin{array}{ccc}{\displaystyle \frac{\partial \mathrm{F}\left(\mathrm{x}\right)}{\partial \mathrm{x}}}& {\displaystyle \frac{\partial \mathrm{F}\left(\mathrm{x}\right)}{\partial \mathrm{y}}}& {\displaystyle \frac{\partial \mathrm{F}\left(\mathrm{x}\right)}{\partial \mathrm{z}}}\\ {\displaystyle \frac{\partial \mathrm{F}\left(\mathrm{y}\right)}{\partial \mathrm{x}}}& {\displaystyle \frac{\partial \mathrm{F}\left(\mathrm{y}\right)}{\partial \mathrm{y}}}& {\displaystyle \frac{\partial \mathrm{F}\left(\mathrm{y}\right)}{\partial \mathrm{z}}}\\ {\displaystyle \frac{\partial \mathrm{F}\left(\mathrm{z}\right)}{\partial \mathrm{x}}}& {\displaystyle \frac{\partial \mathrm{F}\left(\mathrm{z}\right)}{\partial \mathrm{y}}}& {\displaystyle \frac{\partial \mathrm{F}\left(\mathrm{z}\right)}{\partial \mathrm{z}}}\end{array}\right) \\ {\mathrm{J}}_{11}=(2\mathrm{x}-1)(\mathrm{P}\mathrm{t}+\mathrm{a}\mathrm{C}\mathrm{r}1-\mathrm{z}\mathrm{P}\mathrm{t}+\mathrm{y}\mathrm{R}\mathrm{a}\mathrm{C}\mathrm{s}1-\mathrm{y}\mathrm{R}\mathrm{a}\mathrm{E}\mathrm{s}1) \\ {\mathrm{J}}_{12}=\mathrm{x}(1-\mathrm{x})(\mathrm{R}\mathrm{a}\mathrm{E}\mathrm{s}1-\mathrm{R}\mathrm{a}\mathrm{C}\mathrm{s}1) \\ {\mathrm{J}}_{13}=\mathrm{x}(1-\mathrm{x})\mathrm{P}\mathrm{t} \\ {\mathrm{J}}_{21}=\mathrm{y}(1-\mathrm{y})\mathrm{S}\mathrm{c} \\ {\mathrm{J}}_{22}=(1-2\mathrm{y})(\mathrm{C}\mathrm{f}1-\mathrm{C}\mathrm{f}2-\mathrm{S}\mathrm{c}+\mathrm{x}\mathrm{S}\mathrm{c}-\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1+\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1-\mathrm{z}\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1+\mathrm{z}\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}2) \\ {\mathrm{J}}_{23}=\mathrm{y}(1-\mathrm{y})(\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}2-\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1) \\ {\mathrm{J}}_{31}=\mathrm{z}(1-\mathrm{z})\mathrm{P}\mathrm{t} \\ {\mathrm{J}}_{32}=\mathrm{z}\left(1-\mathrm{z}\right)\mathrm{E}\mathrm{r}2 \\ {\mathrm{J}}_{33}=(2\mathrm{z}-1)(\mathrm{C}\mathrm{r}2+\mathrm{P}\mathrm{t}-\mathrm{y}\mathrm{E}\mathrm{r}2-\mathrm{x}\mathrm{P}\mathrm{t}) \end{gathered}$$

(13)

In accordance with the Lyapunov indirect approach, the stability characteristics of dynamic systems can be assessed by examining the eigenvalues derived from the Jacobian matrix [53]. When analyzing the stability of a system at its equilibrium position, the system exhibits instability if every eigenvalue associated with the Jacobian matrix possesses a positive value; if all eigenvalues are less than 0, the point is evolutionary stable; and if one or more eigenvalues are greater than 0, the point is a saddle point. The eight equilibrium points are substituted into the Jacobian matrix to obtain the corresponding eigenvalues and their respective signs, as shown in

Table 3. Stability of equilibrium point.

Equilibrium Points	Jacobian Eigenvalues	Real Symbol	Stability Conclusion
	λ1, λ2, λ3
E1(0,0,0)	−Pt − aCr1, Cf1 − Cf2 − Sc − PrEf1 + bPrEf1, −Cr2 − Pt.	(−,s,−)	saddle point or stable point
E2(0,1,0)	RaEs1 − aCr1 − RaCs1 − Pt, Cf2 − Cf1 + Sc + PrEf1 − bPrEf1, Er2 − Cr2 − Pt.	(s,s,+)	saddle point
E3(0,0,1)	−aCr1, Cf1 − Cf2 − Sc − PrEf1 + bPrEf2, Cr2 + Pt.	(−,−,+)	saddle point
E4(0,1,1)	RaEs1 − RaCs1 − aCr1, Cf2 − Cf1 + Sc + PrEf1 − bPrEf2, Cr2 − Er2 + Pt.	(s,+,−)	saddle point
E5(1,0,0)	Pt + aCr1, Cf1 − Cf2 − PrEf1 + bPrEf1, −Cr2.	(+,s,−)	saddle point
E6(1,1,0)	Pt + aCr1 + RaCs1 − RaEs1, Cf2 − Cf1 + PrEf1 − bPrEf1, Er2 − Cr2.	(s,s,+)	saddle point
E7(1,0,1)	aCr1, Cf1 − Cf2 − PrEf1 + bPrEf2, Cr2.	(+,s,+)	saddle point
E8(1,1,1)	aCr1 + RaCs1 − RaEs1, Cf2 − Cf1 + PrEf1 − bPrEf2, Cr2 − Er2.	(s,s,−)	saddle point or stable point

. The symbols “+”, “−”, and “s” represent eigenvalues greater than 0, less than 0, and those with indeterminate signs, respectively.

According to the Lyapunov indirect method, it can be concluded that E2 = (0,1,0), E3 = (0,0,1), E4 = (0,1,1), E5 = (1,0,0), E6 = (1,1,0), and E7 = (1,0,1) cannot be evolutionary stable points. This implies that the interests of one party are maximized while those of the other two parties are not or that the interests of at least one party remain unmaximized. Consequently, the six aforementioned points will not be elaborated upon further. Among the potential evolutionary stable points, E1 = (0,0,0) and E8 = (1,1,1) represent scenarios where either one participant has optimized its gains, or all parties have attained their peak benefits. Based on the value proposition of this study, smallholder farmers actively participate in green production within the framework of Science and Technology Backyards.

3.5. Scenario Analysis of Evolutionary Systems

3.5.1. Data Sources

The maturity of relevant green production technologies is a critical prerequisite for the adoption of green production practices in agriculture. Grapes, loquats, and other cash crops often face unfavorable conditions for the application of technology. The development of green production technologies for these crops is insufficient, and in practice, green technologies frequently fail to meet the production needs of such crops. Therefore, this paper focuses on food crops like rice and wheat as examples to examine the role of STBs in promoting the green transformation of smallholder farmers. On the one hand, rice, wheat, and other major food crops hold significant strategic importance and social value [54]. Food is not only the foundation of human survival but also a crucial pillar of economic, social, cultural, and ecological stability [55]. Ensuring food security is essential for long-term national security, social harmony, and stability and serves as a core component in promoting the sustainable development of human society [56]. On the other hand, while China has an extensive grain cultivation area, its production methods remain relatively inefficient. In 2024, the country’s grain sown area reached 119,319 thousand hectares, reflecting an increase of 351 thousand hectares, or 0.3%, compared to 2023 [57]. However, at present, China’s grain production remains relatively rudimentary, with low levels of mechanization and standardization [58]. Therefore, there is an urgent need for China to advance green production technologies and enhance the level of sustainable food production.

This study focuses on Ningbo City, Zhejiang Province, as the study area. Firstly, Zhejiang Province has developed a robust network of Science and Technology Backyards, making it a key reference for China in promoting the model of Science and Technology Backyards. In the national development and promotion of Science and Technology Backyards, Zhejiang Province stands out for its exceptional performance. To date, Zhejiang has led the country in the number of Science and Technology Backyards approved by the China Rural Special Technology Association, with a total of 119. These Science and Technology Backyards are distributed across 10 prefecture-level cities in Zhejiang and have established partnerships with nine higher education institutions. Secondly, Ningbo Science and Technology Backyards places significant emphasis on food production, with the establishment of the Yuyao Rice and Wheat Science and Technology Backyard already in place. The Yuyao Rice and Wheat Science and Technology Yard is a family farm integrating rice and wheat planting, processing, sales, and agricultural machinery services, with a contracted area of more than 1200 acres, and is a provincial-level demonstration base for new rice varieties. Thirdly, due to the accessibility of data, the research team conducted a long-term study tracking the planting, growth, and marketing of food crops in Ningbo, thereby obtaining a substantial amount of firsthand data.

In this research, the starting values for the parameters were established by employing two distinct approaches. (1) On-site investigations. The research team visited and conducted a study at the Yuyao Rice and Wheat Science and Technology Backyard, where they collected relevant data. The Yuyao Rice and Wheat Science and Technology Backyard, affiliated with Zhejiang University, covers an area of 1200 mu and influences approximately 280,000 mu of rice and wheat cultivation in the county. It provides local farmers with essential agricultural services, including seedlings, fertilizers, and other supplies. However, the Science and Technology Backyard lacks government financial support and does not receive funding for research. Instead, the research and development efforts are fully financed by Zhejiang University. Upon completion of the research and development phase, pilot testing of the results was conducted over a three-year period at the Science and Technology Backyard. Based on comprehensive assessments, the cost of the pilot, including land rental for the experimental fields, was estimated at approximately CNY 400,000. In traditional rice and wheat cultivation, each acre requires around 20 catties of seed, with seed prices ranging from CNY 5 to CNY 10 per catty. Additionally, costs for pesticides, fertilizers, and other inputs amount to about CNY 270 per mu. The yield is approximately 750 catties per mu, with a stable market price of about CNY 1.2 per catty. In green production, the additional costs primarily stem from increased seedling expenses and the standardization of production processes. The total cost for green production is approximately CNY 440 per mu. The yield is around 840 catties per mu, and Zhejiang University collaborates with food processing enterprises to establish unified acquisition and sales channels. The cost for these activities is about CNY 100,000, with the purchase price of rice and wheat reaching CNY 1.4 per catty. Before the establishment of the STB, smallholder farmers had to independently source the necessary agricultural materials for green production, incurring additional costs of about CNY 1000. (2) In the domain of agricultural economics, specialists have provided their assessments regarding the significance of disseminating research outcomes for the research institution. The specific parameter configurations, as detailed earlier, are illustrated in

Table 4. The initial values of all variables.

Variables	Parameters	Value	Unit of Measure
Cr1	Research costs	380	CNY 10,000
a	Research cost-sharing factor	0	%
Pt	Costs of the pilot test	40	CNY 10,000
Cs1	Operating Costs of the Science and Technology Backyard	100	CNY per mu
Es1	Operating income of the Science and Technology Backyard	120	CNY per mu
Ra	Radiation area	28	Ten thousand mu
Cg	Government financial support for the Science and Technology Backyard	100	CNY 10,000
Cf1	Production costs associated with traditional production methods	370	CNY per mu
Ef1	Prices for sales via traditional distribution channels	1.2	CNY per catty
Cf2	Production costs associated with green production methods	440	CNY per mu
Ef2	Prices for sales via the research institute’s distribution channels	1.4	CNY per catty
Sc	Search costs associated with sourcing green production agricultural products	1000	CNY
Pr	Yield from traditional production methods	750	Catty
b	Factor for yield enhancement in green production	112	%
Er1	Funding allocated for the research institute’s project	400	CNY 10,000
Er2	Social value generated by the implementation of these research results	450	CNY 10,000
Cr2	Cost associated with developing these sales channels	10	CNY 10,000

.

3.5.2. Scenario Analysis

Scenario 1: The point (0,0,0) represents the evolutionarily stable state. According to the data presented in Table 3, a system of inequalities (14) can be formulated, demonstrating that Science and Technology Backyards fail to gain advantages from technological dissemination, consequently leading to their negative performance of responsibilities. The profitability of green production for smallholder farmers is lower than that of traditional production, leading them to opt for traditional production. The benefits of cooperation for research institutes are outweighed by the investment required, leading them to opt for negative cooperation.

$$\begin{gathered} \mathrm{P}\mathrm{t}+\mathrm{a}\mathrm{C}\mathrm{r}\mathrm{l}>0 \\ \mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1-\mathrm{C}\mathrm{f}2<\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}1-\mathrm{C}\mathrm{f}1 \\ \mathrm{C}\mathrm{r}2+\mathrm{P}\mathrm{t}>0 \end{gathered}$$

(14)

Scenario 1 illustrates that, in the case of negative cooperation between Science and Technology Backyards and research institutes, the costs of pilot testing scientific research results and sourcing green technologies are difficult to eliminate. As a result, smallholder farmers struggle to cover the expenses associated with green production through their earnings, leading them to reject the green transition. Upon satisfying the constraints specified in Scenario 1, the evolutionary system attains equilibrium. To validate this finding, we conducted strategy evolution simulations for Science and Technology Backyards, smallholder farmers, and research institutions utilizing MATLAB R2024a. The initial parameter configuration failed to meet the requirements of the inequality group (14). To elucidate the evolutionary outcomes and accommodate regional policy variations, we modified the search cost parameters associated with smallholder farmers’ acquisition of green agricultural technologies. We specifically assigned Pt a value of 5000, maintaining other parameters unchanged. The simulation outcomes are presented in Figure 4. To elucidate the evolutionary outcomes and accommodate regional policy variations, we modified the search cost parameters associated with smallholder farmers’ acquisition of green agricultural technologies. We specifically assigned Pt a value of 5000, maintaining other parameters unchanged. The simulation outcomes are presented in Figure 4. Figure 4a demonstrates that Science and Technology Backyards invariably adopt the negative performance strategy, irrespective of their initial conditions; Figure 4b reveals that smallholder farmers consistently select the conventional production strategy, regardless of their starting point; Figure 4c illustrates that research institutes invariably implement the negative cooperation strategy, independent of their initial state; Figure 4d displays a three-dimensional representation of the evolutionary strategies for Science and Technology Backyards, smallholder farmers, and research institutes, indicating that the system’s dynamic evolution ultimately converges to the stable point (0,0,0).

Scenario 2: The point (1,1,1) represents the evolutionary stable state. As demonstrated in Table 3, a system of inequalities (15) can be formulated, revealing that Science and Technology Backyards gain advantages from technological dissemination, consequently motivating them to positively perform their responsibilities. The profit from green production for smallholder farmers exceeds that of traditional production, leading them to opt for green production. Similarly, the benefits of cooperation for research institutes outweigh the associated costs, prompting them to choose positive cooperation.

$$\begin{gathered} \mathrm{a}\mathrm{C}\mathrm{r}1<\mathrm{R}\mathrm{a}(\mathrm{E}\mathrm{s}1-\mathrm{C}\mathrm{s}1) \\ \mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}\mathrm{I}-\mathrm{C}\mathrm{f}\mathrm{I}<\mathrm{b}\mathrm{P}\mathrm{r}\mathrm{E}\mathrm{f}2-\mathrm{C}\mathrm{f}2 \\ \mathrm{C}\mathrm{r}2<\mathrm{E}\mathrm{r}2 \end{gathered}$$

(15)

Scenario 2 demonstrates that with a high degree of cooperation between Science and Technology Backyards and research institutes, the costs of pilot testing research results and sourcing green technologies are eliminated. As a result, Science and Technology Backyards can generate profit by selling green production technologies, smallholder farmers can increase their earnings through green production, and research institutes can realize their value by promoting these technologies. This collaboration leads to mutual benefits for all three parties. Upon satisfying the constraints specified in Scenario 1, the evolutionary system attains equilibrium. To validate this finding, we conducted strategy evolution simulations for Science and Technology Backyards, smallholder farmers, and research institutions utilizing MATLAB R2024a. The initial parameter configuration failed to meet the requirements of the inequality group (15). To elucidate the evolutionary outcomes and accommodate policy variations across regions, we modified the parameter values associated with search costs for smallholder farmers in acquiring green agricultural technologies. Specifically, we assigned Ef2 a value of 14 while maintaining other parameters unchanged. The simulation outcomes are depicted in Figure 5. Figure 5a demonstrates that Science and Technology Backyards invariably adopt the positive performance strategy, irrespective of their initial conditions; Figure 5b reveals that smallholder farmers consistently select the green production strategy, regardless of their starting point; Figure 5c illustrates that research institutes invariably implement the positive cooperation strategy, independent of their initial state; Figure 5d displays a three-dimensional representation of the evolutionary strategies for Science and Technology Backyards, smallholder farmers, and research institutes, indicating that the system’s dynamic evolution ultimately converges to the stable point (1,1,1).

4. Results

The scenario analysis investigated two distinct evolutionary steady states within the model framework: the equilibrium points (0,0,0) and (1,1,1). Based on the initial parameter configurations, the system converges to the evolutionary steady state (1,1,1), which accurately reflects the current operational conditions of the Yuyao STB. However, most STBs are still in the (0,0,0) state. To explore how to effectively leverage the radiation-driven role of STBs in promoting the green development of Chinese agriculture, this study conducts a sensitivity analysis of the assigned parameters. The evolutionary stable state of (1,1,1) is taken as the desired outcome—wherein Science and Technology Backyards positively fulfill their responsibilities, smallholder farmers adopt green production practices, and research institutes engage in active collaboration. Based on this, a logical framework is constructed to illustrate how Science and Technology Backyards can promote the green transformation of smallholder farmers, as shown in Figure 6. Firstly, Science and Technology Backyards and research institutes collaborate to develop scientific research projects and conduct pilot tests on the research outcomes. Once the technology has matured, the STBs will promote it to help smallholder farmers improve the yield and quality of agricultural products, thereby increasing their production income. In this process, Science and Technology Backyards can also generate profit, while research institutes can realize the social value of their scientific research results through their dissemination. This collaborative effort leads to value creation for all three parties, stabilizing them in this mutually beneficial state [59]. This study conducts a sensitivity analysis of key parameters, including the cost-bearing coefficient of the Science and Technology Backyard, the cost of pilot testing research results, the selling price of green products, the yield enhancement coefficient, and the operating income of the Science and Technology Backyard, using real-world data as a reference. This analysis aims to identify the conditions under which all three parties actively participate, thereby achieving Pareto optimization.

4.1. Impact of the Cost-Bearing Coefficient of the STB

The domain of the cost-bearing coefficient for the STB is defined as [0, 1]. To explore the conditions under which the cost-bearing coefficients optimize the system, this study assigns values to the coefficient within this range, using a step size of 0.25. Specifically, a = 0 indicates that the Science and Technology Backyard bears no cost for this research, while a = 1 means that the Science and Technology Backyard fully assumes the research costs. The influence of cost-bearing coefficients on the decision-making approaches of various stakeholders is demonstrated in Figure 7 while maintaining consistent values for all other variables. Specifically, Figure 7a displays the effects on Science and Technology Backyards; Figure 7b reveals the consequences for smallholder farmers, and Figure 7c presents the outcomes for research institutes. The findings from the simulation demonstrate that Science and Technology Backyards exhibit limited capacity in sharing research expenditure burdens. When the Science and Technology Backyard does not bear the research costs, both the Science and Technology Backyard and the research institute tend to reach a 1-stable state. In this scenario, full cooperation between the two parties eliminates the barrier of technology supply, and smallholder farmers are not required to bear the costs of searching for green production technologies, allowing them to rapidly stabilize at a 1-stable state. When the cost-bearing coefficient a is set at 0.25, the Science and Technology Backyards, motivated by the potential profits from promoting research results, initially show a tendency to actively fulfill their responsibilities. However, due to the relatively small scale of the STB and its inability to bear the research costs, coupled with insufficient short-term operating income to cover the research investments made by the institute, the system ultimately tends to stabilize at the point (0,0,0). As the cost-bearing coefficient increases further, particularly when the STB assumes the full burden of research costs, the system rapidly converges to a 0-stable state. For research institutes, having the Science and Technology Backyard bear part of the scientific research costs can alleviate some of their financial pressure. However, if the Science and Technology Backyard solely assumes responsibility for promoting the research results, this arrangement yields greater benefits for the research institutes. This, in turn, creates stronger incentives for all three parties, enhancing the stability of the system and driving it toward a 1-stable state.

4.2. Impact of the Cost of Pilot Testing Research Results

Pilot testing of scientific research results is a crucial step in bridging small-scale experiments conducted in research laboratories and large-scale industrial production. It provides a reliable technical foundation for the successful transition to industrialized production. This research investigates the optimal conditions of the system by adjusting the pilot testing cost of research outcomes within a [−100%, 100%] range relative to their baseline values, with specific values set at 0, 20, 40, 60, and 80. Maintaining other parameters unchanged, the simulation outcomes are displayed in Figure 8. Figure 8a–c sequentially depicts how the pilot testing cost influences the strategic choices of Science and Technology Backyards, smallholder farmers, and research institutions. The findings reveal that a reduction in the pilot testing cost leads to a stronger inclination of the three stakeholders toward a stable equilibrium, with the evolutionary trajectories of Science and Technology Backyards and research institutions showing convergence. The game system converges to the equilibrium point (1,1,1) when the pilot testing cost is set at 0, 20, 40, and 60. However, different behaviors were observed at varying costs of the pilot test. When the pilot testing cost is 0 or 20, the evolutionary paths of the STBs and research institutes exhibit an outward concave shape. This suggests that the change in decision-making willingness of each party is not influenced by the other’s behavior; rather, it is driven by the profit incentives associated with the promotion of scientific research and technology. When the pilot testing cost is 40 or 60, the evolutionary paths of the STBs and research institutes display an inward concave shape. This indicates that the willingness of one party is influenced by the willingness of the other party. The cooperation between Science and Technology Backyards and research institutes effectively mitigates the economic burden of the pilot testing cost, driving the decision-making willingness of both parties to steadily evolve toward a 1-stable state. When the pilot testing cost is 80, the STBs and research institutes are unable to cooperate efficiently. As a result, the impact of the pilot testing cost cannot be mitigated in a short period, and the substantial economic burden prevents both parties from gaining the necessary incentives to evolve towards a 1-stable state. Consequently, the system ultimately tends toward the (0,0,0) stable state. The decision-making willingness of Science and Technology Backyards and research institutes influences the channels for technology supply. When these institutions cooperate efficiently, smallholders are not required to bear the costs of technology searching, making the adoption of green production methods profitable. This results in a rapid stabilization at a 1-stable state. In contrast, when cooperation is lacking, smallholders reject the green transition and continue with traditional production methods.

4.3. Impact of the Selling Price of Green Products

An increase in the selling price of green products will attract greater investment in resources such as capital, technology, and talent, thereby fostering the development and expansion of the green industry and driving the optimization and upgrading of the industrial structure. In order to investigate the optimal conditions for the green product market system, this research conducted a scaling analysis of utility values. This study established a baseline for green product pricing and examined utility variations within a symmetrical range of [−100%, 100%]. Specifically, the analysis incorporated five distinct scaling factors: 1.2, 1.3, 1.4, 1.5, and 1.6, representing incremental adjustments to the baseline value. A price of 1.2 represents the current selling price of conventional agricultural products, indicating no market demand for green agricultural products. Conversely, a price of 1.6 reflects a situation where the price difference between green agricultural products and conventional agricultural products is twice the normal price differential between the two. The simulation results, with all other parameters held constant, are presented in Figure 9. Figure 9a–c respectively illustrates the impact of green product sales prices on the strategic decisions of STBs, smallholder farmers, and research institutes. The simulation results demonstrate that the selling price of green products can effectively promote the green transformation of smallholder farmers while also indirectly stimulating the role of STBs and research institutes. When the market selling price is too low, at values of 1.2 or 1.3, smallholder farmers are unable to cover the costs associated with green production through the benefits of producing green agricultural products. As a result, their decision-making willingness quickly stabilizes at zero. When the market selling price is set at 1.4, 1.5, or 1.6, the decision-making willingness of smallholders follows a “U” shape, with the magnitude of the “U” shape decreasing as the price increases. This suggests that market demand can incentivize smallholder farmers to adopt green production practices and enhance their demand for green production technologies. As smallholder farmers’ willingness to adopt green production increases, Science and Technology Backyards are able to profit from promoting technology, thereby realizing the value of scientific research results, which leads to stabilization at a 1-stable state.

4.4. Impact of the Yield Enhancement Coefficient

The yield enhancement coefficient can effectively stimulate production incentives, thereby increasing the demand for production technologies among business entities. This research investigates the optimal conditions for maximizing system performance by varying the yield enhancement coefficient. The coefficient is scaled from −100% to +100% of its baseline value, with specific values set at 1, 1.06, 1.12, 1.18, and 1.24. The baseline value of 1 corresponds to conventional agricultural production levels. The value of 1.24 signifies a substantial improvement in green production output, specifically representing double the yield difference between green and traditional farming methods. The simulation outcomes obtained while maintaining constant parameters are displayed in Figure 10. Figure 10a–c demonstrates how the yield enhancement coefficient influences the decision-making processes of Science and Technology Backyards, smallholder farmers, and research institutions, respectively. The simulation results demonstrate that a higher yield enhancement coefficient in technology application can effectively increase smallholders’ demand for green production technology. When the yield enhancement coefficient is too low (1, 1.06), the benefits from producing green agricultural products are insufficient to cover the costs incurred in green production, leading smallholder farmers to reject the green transition. As a result, their decision-making willingness quickly stabilizes at 0. However, when the yield enhancement coefficient is set to 1.12, 1.18, or 1.24, the decision-making willingness of small farmers exhibits a “U” shape, with the magnitude of the “U” shape decreasing as the price rises. This suggests that an increase in the yield enhancement coefficient can drive small farmers toward green production and heighten their demand for green production technology. As small farmers’ willingness to adopt green production grows, the Science and Technology Backyard benefits from technology promotion, allowing for the realization of the value of scientific research results, ultimately stabilizing the system at a 1-stable state.

4.5. Impact of the Operating Income of the STB

The revenue generated by the operation of the STB is derived from the sale of agricultural inputs for green production to local farmers. Factors such as the selling price of these agricultural inputs directly influence the production costs incurred by farmers when adopting green production methods. In order to investigate the optimal operational conditions of the system, this research modifies the STB’s revenue performance across a spectrum spanning from −100% to +100% relative to its standard level. The experimental parameters were established at five distinct points: 100, 110, 120, 130, and 140. A value of 100 represents the cost price; 120 represents the current price, and 140 indicates that the profit generated by the STB is double the operating profit at the current price. Correspondingly, the cost of green production for smallholders was adjusted to 420, 430, 440, 450, and 460, respectively. The outcomes of the simulation, under constant parameter conditions, are depicted in Figure 11. Figure 11a–c sequentially demonstrates how the operational revenue of the Science and Technology Backyard influences the strategic choices of STBs, smallholder farmers, and research institutions. These findings reveal that the financial performance of the STB substantially boosts its motivation to proactively execute its designated functions. This, in turn, eliminates the costs associated with product trials and search costs for smallholder farmers, promoting the system’s evolution toward the equilibrium point (1,1,1). When the operating income of the STB is 100, agricultural materials are sold at cost price, resulting in no profit for the STB, which, thus, chooses to fulfill its responsibilities passively. When the operating income is 110, the evolution path follows an inverted “U” shape, briefly increasing before quickly converging to 0 and stabilizing. This occurs because the willingness of the Science and Technology Backyard to increase its contribution is slow, and within a short period, it is difficult to eliminate the search costs of green production, which hinders the decision-making willingness of smallholder farmers. However, as the operating income increases to 120, 130, and 140, the Science and Technology Backyard rapidly converges to a stable state at 1. At these levels, the costs of technology pilot testing and green production search are effectively eliminated, leading to the convergence of decision-making willingness among small farmers and research institutes to a stable state at 1.

5. Discussion

As one of the world’s largest agricultural countries, China confronts substantial obstacles stemming from resource depletion and environmental degradation, which pose serious impediments to its economic and social advancement [60]. Promoting the green transformation of smallholder farmers—using STBs as a platform—can play a crucial role in addressing these issues. By reducing the use of chemical fertilizers and pesticides and minimizing soil and environmental pollution through the adoption of green agricultural technologies [61,62], it is possible to advance both the high-quality and sustainable development of Chinese agriculture [63]. Therefore, expanding the radiation-driven role of STBs and facilitating the green transformation of smallholder farmers are critical issues in China’s pursuit of ecological civilization and the construction of a “Green China”.

International research institutions have long been engaged in studying strategies to promote the green transition of smallholder farmers. The International Fund for Agricultural Development (IFAD), a specialized agency of the United Nations, focuses on supporting small-scale producers by helping them increase yields, enhance resilience, and gradually adopt environmentally sustainable practices [64]. The U.S. Regenerative Organic Coalition is dedicated to advancing regenerative organic agriculture by establishing rigorous standards through its certification program (ROC™), which encompasses soil health, animal welfare, and social equity [65]. EiT Food, an organization co-funded by the European Union, fosters innovation in the food system through investment projects that focus on food security and environmental sustainability [66]. Similarly, scientific research institutions such as the China Agricultural Green Development Research Society and the National Academy for Agriculture Green Development have played a key role in popularizing agroecology and environmental protection knowledge, thereby promoting the sustainable development of agriculture [67]. Meanwhile, the Science and Technology Backyard, as an emerging scientific research platform, holds a distinct advantage in facilitating the green transformation of smallholder farmers [68]. Based on the outcomes derived from the computational modeling conducted in this research, Science and Technology Backyards can promote green agricultural production technologies and drive the green transformation of agriculture by reducing the costs associated with pilot testing scientific research results and increasing the availability of green production technologies.

Enhancing farmers’ motivation to adopt green production technologies is a critical factor in achieving the green transformation of agriculture. Market demand for green agricultural products can serve as a strong incentive for farmers to shift toward green production methods [69]. However, Chinese consumers generally adhere to traditional consumption habits and lack the routine of purchasing green agricultural products, resulting in low economic returns for farmers engaged in green production [70,71,72]. From an economic rationality standpoint, farmers assess eco-friendly innovations through a comprehensive evaluation of expenditures versus projected gains, with adoption decisions contingent upon the expected financial advantages surpassing the required investment thresholds. To support this transition, research institutes should focus on developing technologies that are tailored to the specific needs of smallholder farmers, aiming to maximize crop yields within the framework of green agricultural practices. Simultaneously, Science and Technology Backyards can help safeguard the economic viability of green production by organizing unified purchasing schemes or partnering with agricultural organizations to facilitate the acquisition of green products. These mechanisms contribute to securing farmers’ incomes and increasing their willingness to adopt green technologies. This framework aligns with the simulation results of this study, which indicate that both the sales price of green products and the yield enhancement coefficient associated with technology adoption significantly and positively influence farmers’ willingness to engage in green transformation. In contrast, smallholders’ sensitivity to production costs appears to be comparatively low.

Overall, addressing the challenges of technology supply and demand is essential to fully leverage the radiation-driven role of Science and Technology Backyards. Identifying a system-level optimal solution for matching technology supply with demand can significantly promote the widespread adoption of green agricultural production technologies among smallholder farmers. The results of this study highlight several key factors. On the supply side, both the cost-bearing coefficient of STBs and the pilot testing costs of scientific research outcomes are significant determinants. STBs are generally not well-suited to bear the burden of research costs due to the limited scale and financial capacity of their supporting institutions. Imposing such costs can create substantial economic pressure [73]. The strength of Science and Technology Backyards lies in its collaborative structure—particularly in partnerships with research institutes—which helps to reduce pilot testing costs, streamline technology transfer, and enable both parties to realize financial and social value through the dissemination of green technologies. This collaboration fosters co-created value among all three stakeholders: Science and Technology Backyards, smallholder farmers, and research institutions [74]. On the demand side, the selling price of green products and the yield enhancement coefficient are key drivers of farmers’ willingness to adopt green technologies. By contrast, the production costs borne by smallholders—represented by the operating income of Science and Technology Backyards—do not significantly influence their adoption decisions. As economically vulnerable individual producers, smallholder farmers exhibit high levels of risk aversion, with their sensitivity to perceived losses and gains closely tied to their willingness to adopt new practices [75]. Therefore, during the early stages of the green transition, efforts by research institutes to help farmers establish reliable sales channels and ensure the profitability of green technologies can substantially enhance technology adoption [76]. Similarly, the potential of green technologies to increase yields exerts a positive influence on farmers’ decisions. Interestingly, farmers show relatively low sensitivity to the costs associated with applying green technologies. This may be attributed to two primary factors: first, unlike enterprises, smallholder farmers rely almost exclusively on agricultural production as their primary source of income, and increasing returns on limited land is aligned with their fundamental interests [77]; second, higher production costs often translate into increased profits for small science and technology institutes, which, in turn, strengthens their incentive to actively fulfill their service responsibilities—such as reducing farmers’ search costs for green technologies. In conclusion, stimulating smallholder farmers’ motivation for green agricultural transformation requires a coordinated approach to aligning technology supply and demand. This includes strengthening supply channels, ensuring demand-side incentives, and fostering collaborative value creation among Science and Technology Backyards, smallholder farmers, and research institutions.

6. Conclusions

Based on evolutionary game theory and practical research, this study constructs an evolutionary game model to examine how Science and Technology Backyards can lead smallholders toward green transformation. This model reveals the dynamic game relationships among Science and Technology Backyards, smallholder farmers, and research institutes in the context of benefit-sharing. The primary findings can be summarized as follows: First, given specific circumstances, the three-party framework encompassing Science and Technology Backyards, individual agricultural producers, and academic research institutions may develop into two equilibrium configurations: (0,0,0) and (1,1,1). The state (0,0,0) represents a scenario in which none of the three parties participate in the process, whereas (1,1,1) reflects full engagement by all parties, leading to a successful green transformation among farmers. Second, there is a pronounced tendency among Science and Technology Backyards to avoid bearing research costs. Only when the cost-bearing coefficient is reduced from 25% to 0%, Science and Technology Backyards begin to actively engage in their intended role. Additionally, Science and Technology Backyards are highly sensitive to economic incentives; their level of engagement increases with profit margins. When profits fall below 20%, Science and Technology Backyards are effectively inactive and refuse to participate. Third, while cooperation with Science and Technology Backyards enables research institutes to reduce the costs associated with pilot testing their research outcomes, the likelihood of establishing such cooperation decreases as pilot costs rise. When these costs exceed the difference between the social value of the research and the costs borne by it or when the profit margin for the Science and Technology Backyard falls below the 20% threshold, forming a cooperative relationship becomes increasingly difficult. Finally, farmers expect to achieve higher income through increased yields or improved product pricing as a result of green transformation. When the sales profit exceeds 12%, farmers demonstrate a clear willingness to adopt green production practices. Conversely, the direct impact of increased production costs on their willingness to transition is relatively low. Instead, higher production costs tend to encourage Science and Technology Backyards to release green agricultural inputs, which indirectly accelerates the green transformation process among smallholder farmers.

Based on the findings of this study, the following policy recommendations are proposed:

Firstly, targeting the tendency of STBs to avoid R&D costs, a “ladder cost allocation model” should be established to alleviate the economic pressure on STBs while maintaining their enthusiasm for technological research and development. For example, STBs can be classified according to their profit levels and then bear different shares of R&D costs based on this classification, thereby achieving a dynamic allocation of R&D risks among all participants.

Secondly, regarding the impact of pilot test costs on the willingness of STBs to cooperate with research institutes, a mixed funding system of “government basic investment + market supplementary financing” can be established to form a sustainable mechanism for absorbing pilot test costs. On the one hand, the government can set up special technical verification funds to provide long-term and stable financial support to STBs and research institutes. On the other hand, STBs can introduce green financial tools, such as issuing green bonds and applying for green loans, to improve their ability to bear pilot test costs.

Thirdly, regarding the income sensitivity of smallholder farmers’ decision-making, it is essential to place greater emphasis on “visual revenue transmission” while establishing sales channels for their green products. On the one hand, research institutes should leverage their strengths to collaborate with large agricultural enterprises or food processing companies on projects to address the sales challenges of smallholder green agricultural products. On the other hand, STBs should establish a county-level demonstration farm network, utilize emerging technologies such as blockchain to enable traceability and value addition of green agricultural products, and reinforce farmers’ income expectations through empirical cases, thereby instilling confidence in both the technology and its benefits.

It is important to acknowledge two key limitations of this study. First, the analysis focuses primarily on the role of STBs in the supply and demand of agricultural technologies without fully considering their interactions with the marketing aspects of agricultural products. Future research could expand the scope by exploring how STBs influence the supply and demand dynamics of agricultural products, particularly in the context of market integration and value chain development. Second, the empirical data used in this study were collected from the Yuyao Rice and Wheat Science and Technology Backyard in Ningbo, which primarily engages in the cultivation and research of staple food crops. As a result, the conclusions drawn may not be directly generalizable to other STBs focused on cash crops. Further research is needed to validate the findings across diverse agricultural contexts and crop types. Third, this study is based solely on existing static models and does not introduce time-dependent variables. Future research could consider incorporating the life cycle characteristics of technology diffusion and constructing dynamic game models for further investigation.