Research on the Mechanisms and Pathways of Digital Economy—Driven Agricultural Green Development: Evidence from Sichuan Province, China

Abstract

This study endeavors to elucidate the mechanisms and pathways through which the digital economy shapes agricultural green development, providing theoretical underpinnings and practical guidance for the green transformation of regional agriculture. (1) Using panel data from 18 prefecture-level cities in Sichuan Province (2013–2022), a comprehensive evaluation index system for agricultural green development was formulated. Fixed-effects, mediating-effects, and threshold-effects models were employed to systematically analyze the direct effects, transmission pathways, and nonlinear characteristics of the digital economy on agricultural green development. (2) The fixed-effects model shows that the digital economy markedly propels agricultural green development in Sichuan Province. The mediating-effects model verifies two transmission pathways: “digital economy → technological progression → agricultural green development” and “digital economy → industrial structure upgrading → agricultural green development”. The threshold-effects model suggests that when the digital economy is in the low-threshold interval, it exerts a suppressive impact on agricultural green development; however, once the threshold is surpassed, its promoting effect strengthens significantly. (3) The results demonstrate the following findings: First, the digital economy exerts a significant positive effect on agricultural green development. Second, this promoting effect exhibits significant nonlinear characteristics that vary with the level of digital economy development. Third, the impact manifests remarkable regional heterogeneity, necessitating context-specific development strategies. (4) Five optimization recommendations are proposed: promote the categorized development of agricultural digital technologies and industrial upgrading; advance digital infrastructure and technology adaptation in phases; design differentiated regional policies; establish a hierarchical and classified long-term guarantee mechanism; and strengthen the “industry-university-research-application” collaborative innovation and dynamic monitoring system.

1. Introduction

Under the synergistic advancement of the “dual-carbon” strategy (aimed at carbon peaking and carbon neutrality) [1] and global climate governance, green agriculture, as an indispensable pathway to sustainable development, has become a common objective for governments and farmers worldwide. From Poland’s efforts to drive agricultural green transition by expanding the market for organic agricultural products [2] to China’s commitment to reducing agricultural carbon emission intensity while ensuring food security [3], a growing consensus on agricultural green transformation is emerging globally. In China, agricultural green development stands at the core of agricultural modernization. It not only serves as a pivotal route to achieving sustainable agricultural progress but also as an unavoidable solution to resource and environmental constraints, catering to the people’s increasing aspirations for a better life [4]. The scientific assertion that “lucid waters and lush mountains are invaluable assets” profoundly reveals the dialectical unity between environmental protection and socioeconomic development. Thus, adhering to green development and forging a modern agricultural path marked by high productivity, safe products, resource efficiency, and environmental friendliness constitutes the only viable route for China’s agricultural and rural advancement. The high-quality promotion of agricultural green development is an inherent requirement for deepening the reform of the ecological civilization system and building agricultural modernization [5], as well as a crucial pillar for comprehensively advancing agricultural green transition and consolidating the ecological foundations for high-quality development [6].

Propelled by continuous breakthroughs in cutting-edge technologies like 5G and AI, alongside accelerated development of data factor markets and an increasingly robust digital economy industrial system, China’s digital economy has experienced steady growth. According to the CAICT’s China Digital Economy Development Research Report (2024), the scale of China’s digital economy reached CNY 53.9 trillion in 2023, representing an increase of CNY 3.7 trillion over the previous year [7]. Growing from CNY 11.2 trillion in 2012 to CNY 53.9 trillion in 2023, the digital economy has expanded nearly fourfold over 11 years, establishing itself as a primary engine of socioeconomic advancement.

In the agricultural sector, the Ministry of Agriculture and Rural Affairs’ 2024 Guidelines on Accelerating Comprehensive Green Transformation of Agriculture and Promoting Rural Ecological Revitalization explicitly call for “fostering new quality productivity in agriculture through innovation”. Concurrently, Central Document No. 1 has underscored “empowering agricultural green transformation with digital technologies” for three consecutive years. Against this backdrop, the shift from traditional to green agriculture has become an inevitable trend. However, existing research has not yet sufficiently explored the transmission mechanisms linking the digital economy to agricultural green development. How does the digital economy influence agricultural green development? What roles do technological progress and industrial structure play in this process? These questions represent both cutting-edge academic topics and critical bottlenecks in practical green development. A thorough investigation of these issues, including their complex relationships and intrinsic dynamics, will enhance theoretical understanding and provide practical guidance for policy formulation and agricultural green transition, which constitutes the core objective of this study.

Current research on the digital economy and agricultural green development focuses on three domains: 1. Studies on agricultural green development: Internationally, some scholars define agricultural green development from the sustainable development goal perspective, emphasizing an agricultural model that maximizes social net benefits while meeting current and future demands for food, fiber, ecological services, and healthy living [8]; other researchers have defined it from a systems theory perspective as a composite form that encompasses green rural environments, prosperous rural populations, and sustainable agricultural practices [9], with ecological agriculture as the core carrier of green development principles [10]. Domestically, scholars have developed a localized definition system by integrating international concepts with China’s developmental-stage characteristics in agriculture. For instance, some scholars characterize it as a new agricultural paradigm centered on food systems, prioritizing “greenness” and “development” and integrating multifunctional agricultural development to achieve synergistic progress in society, economy, productivity, ecology, environment, and resources [11]; others define it from the ecological civilization perspective, as a modern agricultural pathway guided by comprehensive, coordinated, and sustainable development—aiming to enhance agricultural economic efficiency, build resource-efficient and environmentally friendly agriculture, and leverage advanced technologies, equipment, and management concepts to optimize resource utilization, forming an “ecology-oriented, efficient, and coordinated” modern agricultural model [12]. Regarding influencing factors, scholars investigate agricultural green development under the “dual-carbon” goal, covering theoretical foundations [13], goal analysis [14], evaluation systems [15,16], pathways [17], and policy design [18]. Many argue that innovation in production and technology is key to green and sustainable agriculture [19]. Factors such as labor quality, technological and economic development levels, and farmers’ income also exert varying impacts [20,21]. Policy interventions have proven effective in guiding green agricultural transformation [22]. Moderate-scale operations [23], industrial capital inflows into rural areas [24], and regional brand ecosystems [25] positively influence green development, while the synergy of technology, organization, and environment further enhances sustainability [26]. 2. Research on the digital economy: International scholars generally link the digital economy to advancements in Information and Communication Technology (ICT), defining it as encompassing e-commerce, its infrastructure, and related business activities [27,28]. Some scholars have further refined its scope, pointing out that any field involving the application of digital technology in business activities, such as products or services, falls under the category of the digital economy [29]. Chinese researchers view the digital economy as an economic paradigm emerging at the mature phase of information technology evolution [30]. The widespread adoption of the digital economy across diverse sectors in recent years has garnered significant societal focus [31]. It has not only significantly enhanced total factor productivity [32], providing new impetus for economic growth [33], but also exerted positive impacts on improving employment quality [34], optimizing and upgrading industrial structures [35], reconstructing value chains [36], and promoting corporate digital transformation and innovation capabilities [37,38]. 3. Interdisciplinary research on the digital economy and agricultural green development: Some studies have noted challenges related to environmental protection posed by the digital economy [39], while others have proposed a nonlinear, inverted U-shaped correlation between digitalization and carbon emissions [40]. In agriculture, rural digitalization and green development levels are rising, but their coupling remains suboptimal [41]. Regional disparities in digital infrastructure are identified as a root cause of coordination gaps [42]. Overall, the digital economy significantly promotes agricultural green development [43] via channels including green finance, human capital improvement, and the alleviation of resource misallocation [44].

Despite these contributions, the existing literature lacks a systematic integrative analysis of the relationship between the digital economy and agricultural green development, falls behind in in-depth exploration of the deep-level logic and dynamic pathways of the mechanisms, and lacks refined analysis that considers regional heterogeneities in natural endowments and digital infrastructure. To address these gaps, this study focuses on Sichuan Province and develops a holistic analytical framework to examine the nonlinear impacts of the digital economy on agricultural green development, with technological progress and industrial structure upgrading functioning as mediating channels. By introducing a threshold model and incorporating multidimensional characteristics such as natural conditions, agricultural foundations, and digital infrastructure, this study uncovers the nonlinear effects of the digital economy, providing a basis for differentiated policies. Meanwhile, this study offers empirical evidence for constructing a collaborative mechanism of “technological innovation—structural optimization—green development”.

2. Theoretical Analysis and Research Hypotheses

2.1. Direct Influence of the Digital Economy on Agricultural Green Transformation

The digital economy forms a comprehensive enabling system via a three-dimensional framework of “infrastructure support, technology application-driven, and governance model innovation”, reshaping the trajectory of agricultural green development across three dimensions: physical carriers, production paradigms, and institutional guarantees. As the underlying support, digital infrastructure lays a material foundation for sustainable production by breaking information barriers and enhancing resource allocation efficiency [45]. The universal coverage of rural fiber optic networks and 5G base stations overcomes geographical constraints, facilitating the large-scale deployment of green technologies like IoT sensors and intelligent irrigation systems. The real-time data on soil moisture, meteorology, and other factors collected by these technologies supports precision agricultural decision-making, curbing the overuse of fertilizers and pesticides at the origin. The upgrading of digital storage and logistics facilities reduces the loss rate of agricultural products by optimizing cold-chain transportation networks, thereby indirectly alleviating the environmental burden caused by resource waste and establishing dual green barriers for loss reduction at the production end and carbon reduction at the circulation end. Digital technologies centered on AI and big data further drive production paradigm reconstruction, enabling the implementation of “data-and-intelligence-driven” green production models [46]. AI algorithms integrate multidimensional data, such as historical yields and soil properties, to generate optimized crop planting structure plans, thereby avoiding ecological overload from blind production. The full-chain transparency system, built on blockchain traceability technology, compels the production end to adopt green planting and breeding standards through “transparent consumption”, forming a positive cycle in which market demand guides the innovation of production methods while accelerating the dissemination and popularization of agricultural green technologies. Digital governance tools provide institutional guarantees through a closed-loop system of “policy transmission–behavior monitoring–effect feedback”. Local governments leverage smart agricultural supervision platforms to monitor real-time metrics of agricultural non-point source pollution, including the utilization rate of livestock and poultry waste, as well as the recycling rate of agricultural mulch films, and implement data-driven, targeted regulation. Blockchain technology provides a foundation for trust in data, supporting the distribution of green subsidies, organic certification audits, and other processes. Inter-departmental data integration creates collaborative supervision networks that shift environmental policies from extensive “blanket irrigation” to precise “targeted drip irrigation”, improving regulatory efficiency while ensuring sustainable agricultural resource utilization and ecological conservation.

Hypothesis 1 (H1).

The digital economy promotes agricultural green development.

2.2. Mechanisms Underlying the Digital Economy’s Influence on Agricultural Green Transformation

Centered on data elements as its central driver and digital technologies as transformative instruments, the digital economy establishes a synergistic dual-path mechanism characterized by “technological progress and industrial structure upgrading” through reconfiguring the trajectories of technological advancement and industrial collaboration models. This mechanism exhibits strong congruence with both the new quality productive forces theory and the fundamental principles of agricultural green development. From the perspective of the new quality productive forces theory, the core of technological progress lies in the reconstruction of traditional production factors by data elements [47]. Through the logical chain of “data-element-driven—technological innovation iteration—production paradigm transformation”, it injects core momentum into agricultural green development [46]. Specifically, the agricultural production data collected in real time by IoT sensors are transformed into reusable technological innovation resources after big data analysis. This not only drives technological breakthroughs, such as intelligent early-warning models for plant diseases and pests and blockchain data certification, but also forms a closed-loop innovation ecosystem of “data collection—technological R&D—application feedback”, accelerating the iterative upgrading of agricultural green technologies. The deep integration of AI and IoT technologies gives rise to the “precision agriculture” production model. Large-scale applications of drone monitoring and smart irrigation systems enhance input efficiency, reducing non-point pollution risks at the source. Additionally, the full chain transparency system constructed with blockchain traceability technology forms a market forcing mechanism through “transparent consumption”. This encourages the production side to actively adopt green planting and breeding standards, realizing the positive transmission of “technology application—standard upgrading—ecological improvement” and promoting the transformation of agricultural production toward an environmentally friendly model.

Existing research demonstrates that industrial structure advancement driven by the digital economy enhances urban–rural income equality, urban quality development, and corporate innovation effectiveness [48,49,50]. Within agriculture, the digital economy drives the evolution of the agricultural industrial system toward “resource intensive, low-carbon and efficient” practices by reducing transaction costs in the industrial chain and optimizing the division of labor pattern. E-commerce platforms and the industrial Internet break down the information silos in the agricultural industrial chain, enabling real-time alignment between consumer-side demands and production-side supplies, guiding the adjustment of planting structures and improving the green manufacturing level of agricultural product processing. Integrated business models such as “agriculture + e-commerce” and “agriculture + cultural tourism” expand the ecological service functions of agriculture. By reducing circulation losses through livestreaming e-commerce and developing ecological value through digital cultural tourism, the transformation of agriculture from a single production function to one that creates diversified value is promoted. Digital twin technology and blockchain tools facilitate the precise management and control of energy consumption nodes in the industrial chain, optimize transportation routes, and track pollution emission data. They provide technical support for carbon footprint accounting and environmental responsibility tracing, forming a green upgrade path of “integration of the three industries—energy consumption control—ecological collaboration”.

This dual mechanism reflects how new quality productive forces undergo a revolutionary reconfiguration of conventional agricultural production functions, fostering profound interdependence between the digital economy and green agriculture via data-driven integration and industrial collaboration, providing theoretical support for the agricultural green transition.

Hypothesis 2 (H2).

The digital economy empowers agricultural green development through technological progress and industrial restructuring.

2.3. Threshold Effects of the Digital Economy on Agricultural Green Development

Existing studies have confirmed that the digital economy’s promotion of high-quality agricultural development and modernization exhibits significant nonlinear characteristics [51,52], providing a critical theoretical foundation for examining relationships between the digital economy and agriculture. Consequently, this study posits a threshold effect in the digital economy’s facilitation of agricultural green development. This nonlinearity fundamentally stems from the dynamic synergistic evolution among three dimensions—digital infrastructure, technology adoption, and governance tools—and agricultural systems. When synergistic integration is insufficient, multiple thresholds emerge. Infrastructure thresholds manifest as data flow blockages due to inadequate network coverage; technology adoption thresholds arise from farmers’ limited digital literacy, which hinders the translation of technological advantages into actual productivity gains; and institutional coordination thresholds result from misaligned digital–environmental policies that cause governance inefficiencies. Subject to these multidimensional constraints, digital-driven green transformation follows phased transition patterns. Initially, sub-threshold factor accumulation yields higher adoption costs than benefits, causing inefficient policy resource allocation. As infrastructure matures and human capital accumulates, synergistic breakthroughs at critical thresholds elevate the marginal returns of data elements. Agricultural emission reduction efficiency then exhibits increasing marginal returns as technological penetration deepens. Ultimately, a self-reinforcing cycle emerges: environmental benefits fuel technological innovation, while technological advancement further reduces environmental costs. This intrinsic logic demonstrates that the digital economy’s positive impact grows with deeper three-dimensional factor penetration, reflecting a synergistic change where “deeper penetration → stronger efficiency amplification”.

Hypothesis 3 (H3).

The digital economy’s empowerment of agricultural green development exhibits a threshold effect, i.e., its impact is nonlinear.

3. Study Design

3.1. Model Construction

3.1.1. Baseline Regression Model

Drawing on the theoretical framework outlined previously, to explore how the digital economy impacts agricultural green development in Sichuan Province, we formulate the baseline regression model as follows in Equation (1):

$${\mathrm{Agd}}_{\mathrm{it}}={\alpha }_0+{\alpha }_{1}D{\mathrm{ig}}_{\mathrm{it}}+{\alpha }_2{C}_{\mathrm{it}}+{\mu }_{\mathrm{i}}+{\delta }_{\mathrm{t}}+{\epsilon }_{\mathrm{it}}$$

(1)

where Agd represents the agricultural green development level of city i in year t. Dig denotes the digital economy development level of city i in year t. C is a vector of control variables that may affect agricultural green development. ${\mu }_{\mathrm{i}}$ and ${\delta }_{\mathrm{t}}$ stand for city and year fixed effects, respectively. ${\epsilon }_{\mathrm{it}}$ represents the stochastic error term.

3.1.2. Mediation Effect Model

To validate the transmission mechanisms through which the digital economy influences agricultural green development, we employ a mediation effect model. Specifically, in order to examine whether the digital economy promotes agricultural green development via technological progress and industrial structure upgrading, we follow the methodology of Jiang Ting [53] and construct the mediation effect models as follows:

$$M_{it}={\beta }_0+{\beta }_{1}Di{g}_{it}+{\beta }_2{C}_{it}+{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(2)

$${\mathrm{Agd}}_{\mathrm{it}}={\lambda }_0+{\lambda }_{1}D{\mathrm{ig}}_{\mathrm{it}}+{\lambda }_2{M}_{it}+{\lambda }_3{C}_{\mathrm{it}}+{\mu }_{\mathrm{i}}+{\delta }_{\mathrm{t}}+{\epsilon }_{\mathrm{it}}$$

(3)

M denotes the mediating variables (technological progress and industrial structure upgrading), and $\beta$ and $\lambda$ are coefficients to be estimated.

3.1.3. Threshold Effect Model

Considering the potential nonlinear association between the digital economy and agricultural green development, a threshold effect model is further established with the digital economy’s development level as the threshold variable:

$${\mathrm{Agd}}_{\mathrm{it}}={\eta }_0+{\eta }_{1}D{\mathrm{ig}}_{\mathrm{it}}\times I(D{\mathrm{ig}}_{\mathrm{it}}\le {\theta }_0)+{\eta }_{2}D{\mathrm{ig}}_{\mathrm{it}}\times I(Di{g}_{it}>{\theta }_0)+{\eta }_3{C}_{\mathrm{it}}+{\mu }_{\mathrm{i}}+{\delta }_{\mathrm{t}}+{\epsilon }_{\mathrm{it}}$$

(4)

The indicator function I (·) takes values 0 or 1, where ${\theta }_0$ is the threshold value. The function equals 1 if the threshold variable satisfies the condition; otherwise, it equals 0.

3.2. Variable Selection and Description

3.2.1. Dependent Variable

The dependent variable is the level of agricultural green development (Agd). Guided by the policy essence of “balancing economic, social, environmental, and ecological benefits while focusing on production environments, processes, and agricultural product greening” articulated in the 2017 Opinions on Innovating Systems and Mechanisms to Promote Agricultural Green Development, and building upon research by Wei Qi, Su Kai, Zhang Huijie et al. [54,55,56], this study constructs a comprehensive 18-indicator evaluation system across four dimensions—Resource Utilization, Environmental Protection, Green and Clean Production, and Economic Benefits. This framework aligns with new agricultural development concepts and sustainability theory, ensuring data accessibility and statistical consistency. In indicator design, Resource Utilization selects the Multiple Cropping Index and Effective Irrigation Rate to measure land/water efficiency, while incorporating Total Mechanical Power per Unit Cultivated Land Area, Cultivated Land Area per Capita, and Electricity Consumption per Hectare to capture scale and mechanization levels in agricultural production, aligning with the “innovation-driven, efficiency-first” development paradigm. Environmental Protection employs the Forest Coverage Rate to measure the ecosystem’s self-restoration capacity and uses negative indicators such as Agricultural Wastewater Discharge and Pesticide/Fertilizer/Film Usage Intensity to quantify environmental pressure from agricultural activities, upholding the “reduction and recycling” tenets. Green and Clean Production integrates the theory of human settlements through Energy Conservation and Environmental Protection Expenditures, Popularization Rates of Rural Sanitary Toilets, Rural Domestic Waste Treatment Rates, and Domestic Sewage Treatment Rates, reflecting the transformation of rural production and living modes toward low-carbon and clean styles. Economic Benefits adopts indicators such as the Per Capita Disposable Income of Rural Residents, Grain Yield, Labor Productivity, and Land Productivity to ensure the integration of agricultural green development and economic benefits, avoiding the separation of ecological protection and economic growth.

Methodologically, this study employs the entropy-weighted TOPSIS method for comprehensive evaluation. This hybrid approach utilizes entropy weighting to objectively determine indicator weights based on data dispersion (information entropy), thereby eliminating subjective bias, while integrating TOPSIS quantification to calculate the relative closeness coefficient by defining positive/negative ideal solutions. As a combined methodology, entropy-weighted TOPSIS addresses the traditional TOPSIS’s inability to emphasize indicator significance, preserves raw data integrity to minimize weighting errors, and significantly enhances evaluation accuracy and precision—making it ideal for the multidimensional, multi-indicator assessment of agricultural green development. Due to space constraints, computational details are omitted herein. Specific indicators and their corresponding weights are listed in

Table 1. Agricultural Green Development Indicator System and weights.

Primary Indicator	Secondary Indicator	Tertiary Indicator	Indicator Description	Attribute	Weight
Agricultural Green Development Level	Resource Utilization	Multiple Cropping Index	Sown Area of Crops/Cultivated Land Area	−	0.038
		Effective Irrigation Rate	Effective Irrigated Area/Cultivated Land Area (%)	+	0.070
		Total Mechanical Power per Unit Cultivated Land Area	Total Agricultural Machinery Power/Cultivated Land Area (kW/hectare)	+	0.095
		Cultivated Land Area per Capita	Cultivated Land Area/Rural Population (hectare/person)	+	0.074
		Electricity Consumption per Hectare	Rural Electricity Consumption/Cultivated Land Area (kWh/hectare)	+	0.149
	Environmental Protection	Forest Coverage Rate	%	+	0.052
		Agricultural Wastewater Discharge	10,000 m3	−	0.008
		Pesticide Application Intensity	Pesticide Usage/Sown Area of Crops (t/hectare)	−	0.018
		Fertilizer Application Intensity	Fertilizer Usage/Sown Area of Crops (t/hectare)	−	0.040
		Agricultural Film Usage Intensity	Agricultural Plastic Film Usage/Sown Area of Crops (t/hectare)	−	0.021
	Green and Clean Production	Energy Conservation and Environmental Protection Expenditure	CNY 10,000	+	0.129
		Popularization Rate of Rural Sanitary Toilets	%	+	0.059
		Rural Domestic Waste Treatment Rate	%	+	0.005
		Domestic Sewage Treatment Rate	%	+	0.010
	Economic Benefits	Per Capita Disposable Income of Rural Residents	CNY 10,000	+	0.053
		Grain Yield	kg/hectare	+	0.050
		Labor Productivity	Gross Output Value of Agriculture, Forestry, Animal Husbandry, and Fishery/Primary Industry Employees (10,000 CNY/person)	+	0.065
		Land Productivity	Gross Output Value of Agriculture, Forestry, Animal Husbandry, and Fishery/Sown Area of Crops (10,000 CNY/hectare)	+	0.064

.

3.2.2. Core Explanatory Variable

The core explanatory variable in this study is the level of digital economy development (Dig). To achieve a comprehensive and scientific assessment, this research draws upon the relevant findings of scholars such as Wang Jun and Tian Yun [57,58], references the digital economy index indicator system established by the CCID (China Center for Information Industry Development), and incorporates available urban-level data to construct an eight-indicator system across four dimensions: “Digital Infrastructure, Digital Business Scale, Digital Technology Application, and Digital Innovation Environment”. Digital Infrastructure selects Broadband Internet Users and Mobile Phone Users per 10,000 people to reflect the physical carriers and basic access capabilities for the penetration of the digital economy, serving as the underlying support for the digital transformation of agriculture. Digital Business Scale follows the development logic of “digital industrialization” and measures the industrial scale and human capital reserve of the digital economy through Total Postal and Telecommunications Revenue and Proportion of ICT Employees in information transmission, computer services, and software, reflecting the radiation-driving impact of the digital economy on agriculture. Digital Technology Application includes indicators such as the Digital Inclusive Finance Index and the Number of Taobao Villages, focusing on the actual implementation of digital technologies in agricultural scenarios. The former alleviates agricultural financing constraints, and the latter improves the circulation efficiency of agricultural products, directly linking the enabling paths of the digital economy to connect agricultural production and markets. The Digital Innovation Environment quantifies regional digital innovation capacity through R&D Expenditure and Full-Time Equivalent R&D Personnel, supporting the R&D and application of agricultural green technology. In terms of measurement methods, the entropy-weighted TOPSIS method is still used to measure the level of digital economy development. Specific indicators and their corresponding weights are listed in

Table 2. Digital Economy Development Indicator System and weights.

Primary Indicator	Secondary Indicator	Tertiary Indicator	Indicator Description	Attribute	Weight
Digital Economy Development Level	Digital Infrastructure	Broadband Internet Users per 10,000 People	Number of Broadband Internet Subscribers/Total Population (households)	+	0.037
		Mobile Phone Users per 10,000 People	Number of Mobile Phone Subscribers/Total Population (households)	+	0.015
	Digital Business Scale	Total Postal and Telecommunications Revenue	CNY 10,000	+	0.188
		Proportion of ICT Employees	Employees in Information Transmission, Computer Services, and Software/Total Employment (%)	+	0.158
	Digital Technology Application	Digital Inclusive Finance Index	-	+	0.017
		Number of Taobao Villages	unit	+	0.287
	Digital Innovation Environment	R&D Expenditure	CNY 10,000	+	0.191
		Full-Time Equivalent R&D Personnel	person-years	+	0.107

.

3.2.3. Mediating Variables

The mediating variables are technological progress (Tec) and industrial structure upgrading (Isu). Technological progress reflects the regional level of scientific and technological development and is quantified by the number of patent authorizations (logarithmized). This indicator effectively captures the scale of scientific output and innovation capabilities. Isu reflects the transformation of regional economic structures, indicating the developmental stage of industries and the efficiency of resource allocation. It is calculated using the industrial structure hierarchy coefficient, which is defined as follows:

$$Inis{u}_{it}={\displaystyle \sum _{i=1}^3{X}_{imt}}\times m,m=1,2,3$$

(5)

where $X_{imt}$ denotes the share of the m-th industry in the GDP of region i at period t. This weighted metric tracks industrial structure progression, where elevated values signal a greater transition toward high-value sectors [59].

3.2.4. Control Variables

To mitigate estimation bias and account for confounding factors, this study selects the following control variables, drawing on prior research:

The Economic Development Level (Pgdp) is quantified by the logarithm of per capita GDP. Higher regional development levels typically provide better resources, economic conditions, and technologies to support agricultural green development.

The Urbanization Level (Urbanl) is represented by the urbanization rate. Urbanization impacts agricultural green development through labor migration (promoting scale-intensive farming) and increased demand for green agricultural products.

The Proportion of Financial Support for Agriculture (Pfa) is defined as the ratio of fiscal expenditures related to agriculture, forestry, and water conservancy to the total local public budget expenditures. Higher ratios indicate stronger government support for ecological conservation, the adoption of green technology, and infrastructure development.

Regional Openness (Ro) is the total import–export trade value per capita. Openness facilitates technology and capital inflows, reshaping the factors of agricultural production and technological applications.

Rural Population Size (Ruralp) is a logarithm of rural population. This variable influences labor quantity/quality, as well as agricultural production modes.

3.3. Data Sources

Given the availability and representativeness of data, this research employs panel data covering the period from 2013 to 2022 for 18 prefecture-level cities in Sichuan Province (excluding Aba Tibetan and Qiang Autonomous Prefecture, Garze Tibetan Autonomous Prefecture, and Liangshan Yi Autonomous Prefecture). Data were retrieved from multiple sources: the Rural Statistical Yearbook of China, the Sichuan Statistical Yearbook, annual statistical compendia of individual cities, the EPS Database, and the official websites of local statistical bureaus. Specifically, the Digital Inclusive Finance Index was obtained from Peking University’s Digital Inclusive Finance Index (2011–2022), whereas the count of Taobao Villages was derived from the China Taobao Village Research Report released by Alibaba’s Research Institute. Missing data were supplemented by consulting statistical bulletins on national economic and social development, as well as relevant news reports for each city in the corresponding years; a small number of missing values were filled using linear interpolation. Descriptive statistics for all variables are presented in

Table 3. Descriptive statistics for variables.

Variable Type	Variable	Symbol	Sample Size	Mean	S. D	Min	Max
Dependent Variable	Agricultural Green Development Level	Agd	180	0.286	0.075	0.175	0.577
Explanatory Variable	Digital Economy Development Level	Dig	180	0.078	0.118	0.005	0.863
Mediating Variables	Technological Progress	Tec	180	7.333	1.152	5.075	11.397
	Industrial Structure Upgrading	Isu	180	0.821	0.042	0.721	0.971
Control Variables	Economic Development Level	Pgdp	180	10.625	0.402	9.446	11.517
	Urbanization Level	UrbanL	180	0.490	0.098	0.330	0.799
	Proportion of Financial Support for Agriculture	Pfa	180	0.140	0.036	0.008	0.269
	Regional Openness	Ro	180	5.028	1.390	2.004	9.010
	Rural Population Size	Ruralp	180	5.568	0.580	3.922	6.420

Note that the agricultural green development level and digital economy development level are calculated using the entropy-weighted TOPSIS method.

.

4. Empirical Analysis

4.1. Baseline Regression Results

To ensure the reliability of the estimation outcomes, this research first conducts a multicollinearity test. The findings reveal that the average variance inflation factor (VIF) across the variables was 3.31 (below 10), indicating no severe multicollinearity. Subsequently, the Hausman test yielded a p-value under 0.05, prompting the choice of a fixed-effects model. Incorporating year and regional dummy variables further uncovered significant fixed effects for both region and year, thus confirming the application of a two-way fixed-effects model for analysis.

Table 4. Baseline regression results.

Variables	(1)	(2)
Dig	0.380 *** (6.130)	0.129 *** (3.231)
Pgdp		−0.002 (−0.037)
Urbanl		0.798 *** (2.807)
Pfa		0.055 (0.441)
Ro		0.021 *** (5.445)
Ruralp		−0.011 (−0.668)
Constant	0.256 *** (42.584)	−0.148 (−0.416)
Region fixed	YES	YES
Time fixed	YES	YES
N	180	180
R2	0.189	0.725

Note, *** denote p < 0.01. t-statistics are reported in parentheses.

presents the baseline regression results concerning the digital economy’s impact on agricultural green development. As shown in Model (1), the digital economy index has a coefficient of 0.380, which is positive and statistically significant at the 1% level, validating the positive promotional effect of the digital economy on agricultural green development and supporting Hypothesis 1. In Model (2), after including control variables, the coefficient of the digital economy remains positive and significant, further confirming its facilitating role. Among the control variables, the urbanization level (UrbanL) and regional openness (Ro) show significantly positive effects at the 1% level. This indicates that urbanization accelerates the adoption of green agricultural practices by promoting rural labor migration to non-agricultural sectors, driving consumption upgrades, and facilitating the diffusion of technology. Regional openness improves the efficiency of green technology diffusion through spillover effects and improvements in the market mechanisms. However, Pgdp, Pfa, and Ruralp do not show statistically significant impacts.

4.2. Robustness Checks

To test the robustness of the regression results, robustness checks are conducted using three methods: replacing instrumental variables, replacing explanatory variables, and adding control variables. These tests aim to check whether the results are consistent with the abovementioned results, as specifically shown in

Table 5. Robustness check results.

Variables	(3)	(4)	(5)	(6)
	Agd	Agd	Agd	Agd
Dig	0.399 *** (5.293)	0.150 ** (2.290)	0.035 *** (3.054)	0.129 *** (3.222)
Lfd				0.003 (0.244)
Control Variables	NO	YES	YES	YES
Region fixed	YES	YES	YES	YES
Time fixed	YES	YES	YES	YES
N	162	144	180	180
R2	0.164	0.701	0.723	0.725

Note, ***, ** denote p < 0.01, p < 0.05, respectively. t-statistics are reported in parentheses.

.

Endogeneity Test: This test is an instrumental variable method. Potential endogeneity concerns between digital agriculture and agricultural green development were mitigated using the instrumental variable method. The first-order lagged value of the digital economy served as the instrumental variable. Model (3) displays the first-stage regression outcome, and Model (4) is the second-stage regression result. In Model (3), the estimated coefficient is 0.399, which is significantly positive at the 1% level, indicating that in the first-stage regression, there is a strong correlation between the instrumental variable and the endogenous explanatory variable, meeting the correlation requirement for instrumental variable selection. In Model (4), after applying the instrumental variable method in the second-stage regression, the coefficient for the digital economy is 0.150, which is significantly positive at the 5% level. This indicates that after controlling for the endogeneity problem, the digital economy still has a significant impact on the explained variable. Furthermore, the Kleibergen–Paap rk LM statistic (p-value = 0.000 < 0.05) confirms effective model identification without under-identification issues. The Cragg–Donald Wald F-statistic of 562.982 (exceeding 10) supports the appropriateness of the chosen instrument in effectively resolving endogeneity.

Substitution of the Explanatory Variable: Robustness was further tested by reconstructing the digital economy development index using the Analytic Hierarchy Process (AHP). As shown in Model (5), the digital economy coefficient remains significantly positive at 0.035, which is significantly positive at the 1% level, demonstrating a stable direction and significance across alternative measurement methods.

Addition of Control Variables: Building on the previous analysis, we further incorporated the financial development level (Lfd) as an additional control variable. As shown in Model (6), after including the Lfd, the coefficient of the digital economy remains at 0.129, which is significantly positive at the 1% level. This result further validates the robustness of the regression findings.

4.3. Mediating Effect Analysis

Following the preceding theoretical framework, a mediation effect model was utilized to empirically investigate the transmission mechanisms by which the digital economy fosters agricultural green development through technological progress and industrial structure upgrading. The findings are displayed in

Table 6. Mediating effect analysis results.

Variables	(7)	(8)	(9)	(10)	(11)
	Agd	Tec	Agd	Isu	Agd
Dig	0.129 *** (3.231)	2.139 *** (6.102)	0.215 *** (3.989)	0.210 *** (7.274)	0.191 *** (3.448)
Tec			0.021 ** (1.997)
Isu					0.329 ** (2.573)
Control Variables	YES	YES	YES	YES	YES
Region Fixed	YES	YES	YES	YES	YES
Time fixed	YES	YES	YES	YES	YES
N	180	180	180	180	180
R2	0.725	0.923	0.648	0.613	0.653

Note, ***, ** denote p < 0.01, p < 0.05, respectively. t-statistics are reported in parentheses.

.

Model (7) assesses the direct influence of the digital economy on agricultural green development. The regression in Model (8) yields a coefficient of 2.139 for the digital economy’s impact on technological progress, which is significant at the 1% level. This demonstrates the digital economy’s effective role in stimulating technological advancement. Model (9) reveals that both the digital economy and technological progress display significantly positive coefficients concerning agricultural green development. Consistent with mediation effect principles, this substantiates the following pathway: “digital economy → technological progress → agricultural green development”. This suggests that the digital economy drives technological progress, integrates innovations into agricultural production, optimizes resource utilization efficiency, and ultimately drives agricultural green development. Thus, technological progress serves as a significant mediator.

In the industrial structure pathway, Model (10) indicates a coefficient of 0.210 for the digital economy’s effect on industrial structure upgrading, also significant at the 1% level. Model (11) shows significantly positive coefficients for both the digital economy and industrial structure upgrading on agricultural green development. This validates the following pathway: “digital economy → industrial structure upgrading → agricultural green development”, where industrial restructuring steers agriculture toward green and high-efficiency transformation. Collectively, the results indicate that technological progress and industrial structure upgrading serve as partial mediators in the relationship between the digital economy and agricultural green development. The digital economy thus enhances agricultural sustainability by harnessing the mediating effects of technological innovation and industrial upgrading. Hypothesis 2 is consequently supported.

4.4. Threshold Effect Analysis

4.4.1. Threshold Effect Test

This study applies Hansen’s panel threshold regression model to investigate the nonlinear characteristics and threshold points in the relationship between the digital economy and agricultural green development. During the analysis, we sequentially tested single, double, and triple thresholds using the Bootstrap method, with 500 repeated samplings, to determine the existence and number of panel thresholds. The results displayed in

Table 7. Threshold effect test.

Threshold Variable	Threshold Type	Threshold Value	p-Value	F-Statistic	Critical Values
					1%	5%	10%
Dig	Single Threshold	0.051	0.006	36.39 **	34.548	23.410	19.414
	Double Threshold	0.053	1.000	−7.72	27.243	18.676	14.555

Note: ** denote p < 0.05. The Bootstrap replications included 500 iterations.

confirm a statistically significant threshold effect. When the digital economy itself serves as the threshold variable, the single threshold specification achieves statistical significance at the 1% level, yielding an estimated threshold value of 0.051. The double threshold test failed to achieve statistical significance and was consequently excluded. Consequently, the single threshold model is adopted for regression analysis.

4.4.2. Threshold Effect Regression Results

The threshold model estimation results are presented in

Table 8. Threshold effect regression results.

Variables	Coefficient	Standard Error	t-Statistic
Dig(I ≤ 0.051)	−0.75 *	0.383	−1.960
Dig(I > 0.051)	0.134 ***	0.021	6.510
Pgdp	−0.012	0.071	−0.170
Urbanl	0.816 *	0.431	1.890
Pfa	0.052	0.059	0.880
Ro	0.016 **	0.006	2.760
Ruralp	−0.015	0.011	−1.300
Constant	0.016	0.549	0.030

Note, ***, **, * denote p < 0.01, p < 0.05, p < 0.1, respectively. t-statistics are reported in parentheses.

. When the digital economy’s development level remains at or below 0.051, the regression coefficient for its effect on agricultural green development is −0.75. While statistically significant at the 10% level, this indicates a suppressive relationship. This phenomenon may stem from the dual constraints of weak digital infrastructure and insufficient technical application capabilities in rural areas. On the one hand, it is difficult for the low-level digital economy to support the implementation of precision agriculture technology, leading to aggravated resource misallocation; on the other hand, the high sunk costs in the initial stage of digital transformation, such as the purchase of IoT devices and personnel training, squeeze the investment in green production in the short term, forming an “efficiency-cost” trade-off dilemma. When the digital economy exceeds the 0.051 threshold, the coefficient becomes 0.134, showing a 1% level of positive significance. This indicates that after the digital economy surpasses the critical threshold of 0.051, its technology diffusion effect and factor allocation optimization function are gradually released, significantly improving the level of agricultural green development. The core proposition of Hypothesis 3 is confirmed—the digital economy’s promotion of agricultural green development exhibits a significant nonlinear threshold effect. Therefore, only when digital economy development surpasses this threshold can technology–data–institution synergies be fully mobilized, facilitating the transition from traditional to green agriculture. This evidence provides an empirical basis for the policy path of “phased promotion of digital infrastructure and key breakthroughs in critical constraints”, emphasizing the need to design differentiated support strategies for different development stages.

4.5. Heterogeneity Analysis

The impact of the digital economy on agricultural green development is shaped by multiple interacting factors, manifesting significant heterogeneity. This study systematically examines the heterogeneous impacts of the digital economy on agricultural green development by integrating three dimensions: geographical location, local fiscal technology investment level, and agricultural development level. The results are summarized in

Table 9. Heterogeneity analysis results.

Variables	(12)	(13)	(14)	(15)	(16)	(17)	(18)
	Chengdu Plain Economic Zone	Southern Sichuan Economic Zone	Northeastern Sichuan Economic Zone	High Fiscal Sci-Tech Investment	Low Fiscal Sci-Tech Investment	High-Level Agricultural Development	Low-Level Agricultural Development
Dig	0.108 ** (2.207)	0.757 ** (2.167)	0.695 ** (2.584)	0.094 * (1.869)	0.138 (0.472)	0.163 *** (6.488)	−0.425 (−0.710)
Control Variables	YES	YES	YES	YES	YES	YES	YES
Region Fixed	YES	YES	YES	YES	YES	YES	YES
Time fixed	YES	YES	YES	YES	YES	YES	YES
N	90	40	50	90	90	90	90
R2	0.710	0.946	0.898	0.712	0.798	0.712	0.798

Note, ***, **, * denote p < 0.01, p < 0.05, p < 0.1, respectively. t-statistics are reported in parentheses.

.

4.5.1. Regional Heterogeneity Analysis

Based on the territorial spatial planning of Sichuan Province, the 18 prefecture-level cities are divided into three economic zones: the Chengdu Plain Economic Zone, Southern Sichuan Economic Zone, and Northeastern Sichuan Economic Zone. Variations in geographical conditions, resource endowments, and digital economy maturity drive distinct spatial heterogeneity in the digital economy’s enabling effects on agricultural green development. As the core region for the province’s economy and digital infrastructure, the Chengdu Plain Economic Zone demonstrates a digital economy coefficient of 0.108 in Model (12), which is statistically significant at the 5% confidence level. This indicates that the digital economy in this region has established a stable mechanism to drive agricultural green development. However, due to technological penetration reaching a plateau, the marginal effect exhibits a “high base, slow growth” pattern. By contrast, in the Southern Sichuan Economic Zone, Model (13) reveals the highest digital economy coefficient (0.757) among the three zones. This may stem from its strategic location as a hub of the Yangtze River Economic Belt, where the deep integration of “digital + energy” innovations significantly enhances the synergy between agricultural productivity and ecological benefits. Compared with the former two, in the Northeast Sichuan Economic Zone, Model (14) shows a digital economy coefficient of 0.695, demonstrating that optimizing agricultural production factor allocation through digitalization effectively drives green transformation.

4.5.2. Heterogeneity in Fiscal Investment in Science and Technology

Fiscal investment in S&T can influence the effectiveness of the digital economy through two paths: on the one hand, by directly improving agricultural digital infrastructure and enhancing technical adaptability; on the other hand, by incentivizing agricultural green technology innovation through policies such as R&D subsidies and tax incentives, forming a transmission mechanism of “fiscal investment—technology innovation—digital empowerment”. Based on this, the ratio of fiscal S&T expenditure to GDP is adopted to measure the level of fiscal S&T investment in prefecture-level cities. Empirical findings reveal that in regions with high fiscal tech investment, Model (15) yields a digital economy coefficient of 0.094, which is significant at the 10% level. This indicates that ample fiscal resources are being invested in fueling digital technology R&D and agricultural digital transformation, effectively unlocking the digital economy’s promotion of agricultural green development. Conversely, in regions with low fiscal tech investment, Model (16) exhibits no statistical significance—likely because scarce fiscal resources constrain the application and innovation of digital technologies in agriculture, preventing the full manifestation of their green development promotion effect. This divergence underscores the pivotal role of fiscal tech investment in enabling the digital economy to drive green agricultural development. In regions with robust economies and abundant fiscal tech investment, the integration of the digital economy and agricultural green development runs deeper, significantly amplifying the promotion effect, whereas insufficient resources and investment hinder the digital economy from fully unleashing its efficiency.

4.5.3. Heterogeneity Analysis of Agricultural Development Levels

Disparities in agricultural development levels directly affect a region’s capacity to absorb and transform the digital economy. Regions with high development levels, due to their robust technological foundations, can amplify the enabling effects of the digital economy; conversely, regions with low development levels are constrained by weak infrastructure, hindering the effective realization of the digital economy’s potential. Using the gross agricultural output value as the benchmark, the sample was median-split into high-/low-development cohorts. Empirical analysis confirms significant heterogeneity in digital economy impacts on agricultural green development. Model (17) indicates that in regions with advanced agricultural development, the digital economy coefficient stands at 0.163, which is statistically significant at the 1% level. This indicates that these regions leverage mature digital infrastructure and robust technological application capacities to intensify the integration of digital technologies into agricultural production, fostering both efficient resource utilization and ecological preservation. Conversely, Model (18) reports a statistically non-significant coefficient (−0.425) for low-development regions. This outcome implies that inadequate agricultural infrastructure, limited technological application capacity, and incomplete support systems prevent effective digitally driven green agricultural development. In certain cases, improper technology deployment may even exacerbate resource depletion or environmental harm. This outcome further confirms the presence of a threshold effect in the digital economy, corroborating the argument that “a robust agricultural development base is indispensable for the digital economy to fuel growth”—low-level regions must prioritize infrastructure development and digital capacity building to unlock the digital economy’s green development potential.

5. Conclusions and Recommendations

Leveraging panel data from 18 prefecture-level cities in Sichuan Province, this research systematically investigated the multidimensional influences of digital economy development on green agricultural development by constructing two-way fixed-effects models, mediating effects frameworks, and threshold-effect models. The key findings are outlined below:

First, the digital economy exerts a significant positive effect on green agricultural development. This effect is achieved through two paths: one is the technology-enabling path, i.e., using digital technologies such as IoT big data applications to improve factor allocation efficiency, which reduces resource waste and pollution emissions; the other is the industrial restructuring path, that is, digital extension across the agricultural value chain (“production-processing-sales”) refines labor division, disseminates green technologies, and accelerates the transition toward low-carbon circular models.

Second, the promotional effect of the digital economy on green agricultural development shows significant non-linear characteristics. In the initial development stages, infrastructure deficiencies and poor technical adaptation constrain positive impacts, potentially causing negative effects through resource misallocation. When digital economy advancement surpasses the critical threshold, technological penetration and optimized resource allocation substantially accelerate agriculture’s transition to environmentally sustainable, low-carbon systems.

Third, the impact of the digital economy on green agricultural development exhibits significant regional variations. Economically advanced regions exhibit stable digital economy-driven green development mechanisms with “high-base, slow-growth” marginal effects. Regions with geographic advantages that enable deep digital–industry integration experience more pronounced digital economy-driven outcomes. Crucially, fiscal technology investment and agricultural development levels jointly intensify regional disparities. Where substantial investment and robust agricultural foundations coexist, digital economy contributions to green development prove to be the most impactful. Conversely, infrastructure gaps and technological adaptation limitations constrain digital empowerment elsewhere.

Based on these findings, the following policy implications arise:

First, promote the categorized development of agricultural digital technologies and industrial upgrading. Deploy technology research and application promotion differentially according to the needs of regions at different development levels. For high-digital regions, such as the Chengdu Plain Economic Zone, prioritize breakthroughs in cutting-edge technologies, including AI algorithms, agricultural IoT, and carbon footprint accounting; establish provincial-level agricultural digital technology innovation funds; support the construction of national key laboratories for digital agriculture; promote the tertiary sector integration of “digital + agriculture + cultural tourism”; and build regionally influential digital agriculture industrial clusters. For regions with development potential, such as the Southern and Northeastern Sichuan Economic Zones, focus on adaptive technology R&D; support the transformation of technologies like lightweight smart agricultural machinery and IoT monitoring equipment for hilly areas; promote mature digital technologies through collaborative “enterprise + cooperative + farmer” models; lower technical application barriers; and achieve the synergistic development of technological innovation and industries.

Second, advance digital infrastructure and technology adaptation in phases. In low-digital, weak-agricultural foundation regions, implement digital infrastructure weakness-targeted actions to achieve full 5G network coverage in townships and county-wide coverage of agricultural data platforms; conduct digital skills training for new-generation professional farmers; provide state subsidies for farmers purchasing smart equipment; and prioritize solving the issues of “inability to use and unaffordability”. In advanced regions, deepen the integration of digital technologies with green agricultural production; build “unmanned farm” demonstration bases in major grain-producing areas; promote “IoT + precision farming” systems in specialty production zones; and accelerate the transformation of technological achievements.

Third, design differentiated regional policies. In high-investment, developed regions, establish a “market-led + government-guided” resource allocation mechanism; implement super-deduction policies for corporate R&D expenses; support the formation of digital agriculture industry innovation alliances; promote the commercialization of technological achievements; and create “technology-exporting” digital agriculture benchmarks. In regions with distinct geographical advantages but low fiscal technology investment, increase provincial fiscal transfer payments; establish special agricultural digitization subsidies; build distinctive digital agriculture demonstration zones anchored in major regional strategies like the Yangtze River Economic Belt and the Chengdu–Chongqing Economic Circle; and achieve complementary advantages and differentiated development.

Fourth, establish a hierarchical and classified long-term guarantee mechanism. Refine evaluation systems by incorporating digital technology applications into rural revitalization assessments, adding “technological innovation contribution” indicators for high-level regions and emphasizing “infrastructure compliance rate” and “training coverage rate” assessments for low-level regions. Introduce specialized loans for low-level regions, with government risk compensation funds sharing partial credit risks; pilot agricultural data asset collateralized financing in high-level regions to broaden financing channels for new types of agribusinesses. For talent development, implement the “Digital Agriculture Talent Recruitment Program” to cultivate interdisciplinary professionals proficient in both agricultural technology and digital literacy based on regional development levels.

Fifth, strengthen the “industry-university-research-application” collaborative innovation and dynamic monitoring system. Support universities, research institutions, and enterprises in co-building “Digital Agriculture Joint Laboratories” to tackle key common technologies for green agricultural development, promoting cross-domain technology integration and scenario-based applications. Establish a digital agriculture dynamic monitoring platform that integrates multi-dimensional data (production, resources, environment, and technology) to track the effects of regional digital economy empowerment in real time, providing data support for precise policy adjustments. Break through technological innovation bottlenecks via scientific collaboration, achieve “targeted policy irrigation” through dynamic monitoring, and form a closed-loop “R&D → Application → Evaluation → Optimization” management system to provide sustained momentum for green agricultural development.

6. Discussion

6.1. Research Innovations and Academic Contributions

Compared with the existing literature, this study achieved significant innovations in theoretical construction, empirical findings, and policy applications. First, it achieves multidimensional expansion in mechanism analysis by transcending the limitations of conventional single-path analytical approaches and constructing a synergistic dual-path framework of “technological progress—industrial structure upgrading”. This systematically explains the composite enabling mechanism through which digital technologies optimize factor allocation efficiency and drive the end-to-end digital integration of industrial chains, effectively addressing the limitations of current studies that either focus solely on technology application or emphasize unidimensional industrial evolution. Furthermore, in heterogeneity analysis, it incorporates regional agricultural development levels into the differential analysis framework, revealing their significant interactive moderating effects with fiscal technology investment. Empirical evidence shows that regions with high agricultural development levels leverage their mature production systems to more effectively transform fiscal inputs into digital economy efficacy, enriching the theoretical understanding that “digital empowerment requires synergistic interplay among technological foundations, industrial bases, and policy support”. Ultimately, it achieves precision-targeted innovation in policy design. By closely aligning with Sichuan’s spatial economic pattern of “one-core, multi-branches” and building on the heterogeneity findings, this study proposes a “phased advancement” strategy—prioritizing digital infrastructure enhancement in low-level regions while deepening technology–industry integration in high-level regions. Compared to universal policy solutions, this approach enhances alignment with regional resource endowments and developmental gradients, providing an actionable differentiated policy paradigm for western China and similar complex geoeconomic regions.

6.2. Limitations and Future Prospects

Although this research elucidates multidimensional mechanisms linking the digital economy to agricultural green development, several limitations merit attention. Data Limitations: The analysis relies solely on prefecture-level panel data from Sichuan, lacking cross-regional comparisons with eastern grain-producing areas or northwestern arid zones. The absence of micro-level operator data impedes the precise identification of technology adoption disparities between smallholders and leading enterprises, as well as nuances related to the “digital divide”. Owing to data accessibility issues, granular indicators, such as smart machinery penetration and agricultural IoT coverage, remain unincorporated, potentially affecting comprehensiveness. Mechanism Depth: While the dual paths of “technological progress” and “industrial structure upgrading” are verified, micro-level mechanistic analyses, such as how digital technologies specifically affect the precise application of fertilizers/pesticides or intelligent processing efficiency, require deeper exploration. Methodological Constraints: Despite applying bidirectional fixed-effects, mediating-effect, and threshold-effect models, unresolved endogeneity concerns persist—notably bidirectional causality between the digital economy and agricultural green development—necessitating further investigation.

Future research should prioritize the following ideas: First, data and mechanism analysis should be deepened by combining county-level data and micro-entity surveys to systematically analyze the cost–benefit structures of technology adoption across different geographic–climatic zones (eastern plains, northwestern arid regions, and southwestern hills) and new types of agricultural operating entities (smallholders vs. leading enterprises), comparing technology adaptation differences caused by terrain, climate, and industrial foundations while incorporating more granular indicators to capture impacts of digitized sub-processes. Second, methodological innovation and the exploration of complex relationships should be strengthened by comprehensively applying multi-period DID models to evaluate the long-term dynamic effects of policies like “Digital Villages”. This will reveal the spatial spillover effects of the digital economy on green agricultural development through spatial econometric analysis, qualitative field studies on micro-mechanisms, and the use of text analysis tools to provide a holistic understanding of digital–green interactions. Third, global perspectives and localized pathways should be expanded by drawing on international smart agriculture best practices to explore effective models for integrating digital technologies with China’s smallholder economy, constructing localized “small-scale intelligence” green development solutions. Fourth, interdisciplinary integration should be deepened by synthesizing theories and methods from agricultural economics, environmental science, and computer science to support the development of more systematic and universally applicable theoretical frameworks and practical guidance.