Digital Economy and High-Quality Agricultural Development: Mechanisms of Technological Innovation and Spatial Spillover Effects

Abstract

The digital economy has become a transformative force in modern agriculture, reshaping production methods and driving global sustainable development. Using panel data from 30 provinces in China from 2012 to 2022, this paper constructs a comprehensive econometric framework using the generalized method of moments and the spatial Durbin models to reveal the mediating role of technological innovation and the existence of spatial spillover effects. The results show that (1) under different regression models, the digital economy significantly promotes the high-quality development of agriculture, which is still valid after the robustness test; (2) the digital economy improves innovation input, output, and efficiency to promote the dissemination of new agricultural technologies and the improvement of resource utilization, thus speeding up the process of high-quality agricultural development; (3) the impact of the development of the digital economy on the high-quality development of agriculture has a positive spillover effect between different provinces in China; and (4) the impact of the digital economy is more significant in areas with stronger intellectual property protection and a larger urban–rural income gap. These insights highlight the need to develop policies that strengthen digital infrastructure and foster technological innovation to maximize the benefits of digital transformation in agriculture.

1. Introduction

High-quality agricultural development is a multidimensional and dynamic process that includes increased productivity, economic efficiency, improved social welfare, and environmental sustainability. Given the increasing constraints posed by natural resource limitations, climate change, and environmental degradation, achieving this goal requires not only identifying emerging drivers of agricultural transformation but also addressing structural inefficiencies within the agricultural sector [1]. In this regard, rapid advances in science and technology have emerged as a key force in reshaping agriculture, facilitating a deeper integration of digital technologies with traditional practices. The development of technological paradigms, including digitization, automation and artificial intelligence, has dramatically improved the informatization, intelligence [2], and precision [3] of agricultural production, making agricultural systems more sustainable and efficient.

Among these advances, the digital economy has emerged as a key catalyst for the development of high-quality agriculture. With its ability to enhance data-driven decision-making, optimize resource allocation, and facilitate automation, the digital economy is fundamentally transforming agricultural production and supply chains. From a technological perspective, digitization enables precision agriculture, smart monitoring, and automation, leading to increased productivity, resource efficiency, and environmental sustainability. From a structural level, digitization reduces transaction costs, facilitates market integration, and strengthens rural–urban linkages, thereby increasing regional competitiveness and inclusive growth [4]. It also drives structural change by facilitating market connectivity, improving supply chain coordination, and increasing industrial resilience. Together, these changes contribute to increased productivity, social equity, and environmental sustainability in the agricultural sector [5].

In China, although the government has introduced policies to support the digitalization of agriculture, such as the “Digital Countryside Strategy”, challenges persist. These include low internet penetration in remote rural areas, insufficient digital literacy among farmers, and a lack of targeted training and support for adopting digital technologies [6]. Furthermore, the cost of digital tools such as precision farming equipment and data-driven platforms is also a significant barrier for smallholder farmers [7]. Internationally, farmers face similar challenges in adopting digital technologies, though the specific obstacles vary by region. In developed economies such as the United States and parts of Europe, while advanced digital tools are available, adoption rates remain uneven, particularly among smaller farms. High upfront costs for digital devices and the need for substantial investments in internet connectivity hinder widespread implementation. And, the complexity of new technologies often requires farmers to acquire additional skills that are not readily available to older farmers or those with limited access to training in rural areas [8]. In developing countries like India, the challenges are even more pronounced due to limited awareness of digitalization’s benefits, inadequate technical support, and a lack of infrastructure [9]. The high costs of digital devices, combined with low digital literacy, further exacerbate the digital divide. In the face of these obstacles, some effective measures have been taken internationally to mitigate the challenges of digital technologies in agriculture. For example, financial constraints and technological complexity have been overcome by improving digital literacy training, providing affordable digital tools, and strengthening infrastructure [10]. Mobile technologies and digital platform-based solutions have also provided farmers with easier access to market information, weather updates, and payment systems, significantly increasing their productivity and income [11].

Despite these positive developments in practice, there is still a lack of in-depth discussion in the existing literature on the mechanisms by which the digital economy drives the development of high-quality agriculture; specifically, the technological innovation approaches that drive the improvement of agricultural quality are ignored. There are also few studies that systematically examine the spatial spillover effects of the digital economy on agricultural development across regions. Filling these gaps is critical for understanding the broader impacts of digital transformation and for adopting policy measures to maximize its benefits in agriculture. To fill this gap, this study conducts a comprehensive empirical analysis of the relationship between the digital economy and high-quality agricultural development using panel data for 30 provinces in China (excluding the Tibet Autonomous Region) over the period 2012–2022. This study aims to answer the following key questions: (1) How does the digital economy affect high-quality agricultural development? (2) What mediating role does technological innovation play in this relationship? (3) Are there spatial spillovers?

This paper makes four significant contributions to the relevant literature. First, this paper provides a systematic empirical study of the impact of the digital economy on high-quality agricultural development, expanding the scope of research on the determinants of agricultural transformation. Second, it addresses the mechanistic question of how the digital economy affects agricultural quality, providing empirical validation that digitization promotes innovative inputs and efficiency and thus agricultural development. Third, it explores the spatial spillover effects of the digital economy, reveals the diffusion mechanism between regions, and enriches the geographic dimension of existing research. Finally, by examining regional heterogeneity, this study emphasizes the role of external institutional and economic factors in shaping digitalization effects, providing policy-relevant insights into regional differences and strategic interventions.

2. Literature Review

With regard to the factors of high-quality agricultural development, existing research focuses on several key areas: technological advances, environmental sustainability, and supportive policies. On the tech front, Jin and Han (2024) discovered that agricultural machinery boosts efficiency by making operations more precise, upping productivity, and cutting down on labor expenses [3]. Papadimitropoulos (2023) pointed out that open networks create synergies across different sectors by using digital resources freely, which in turn helps agriculture undergo systemic changes [12]. Mota et al. (2023) noted that when innovation and networks come together, agricultural systems become more sustainable [13]. Similarly, Räty et al. (2023) suggest that new technologies could really shake things up in traditional livestock farming and rural development [14]. However, Engås (2023) cautioned that the digital divide could prevent developing countries from getting their hands on cutting-edge tech [15]. Furthermore, Spykman (2022) emphasized that the public has to be on board with new technologies if they are going to catch on [16]. When it comes to environmental sustainability, Xu et al. (2022) stressed how vital land management and ecological transformation are for top-notch agricultural development and rural revitalization [17]. As for policy support, developing agricultural land markets runs into snags like murky ownership, technical slip-ups, and complicated transaction processes. Solid national land policies, efforts to consolidate land, and setting up land reserves are all key to making the land market work better and more efficiently [18].

In recent years, the digital economy has emerged as a critical driver of high-quality agricultural growth. People see digital technologies as essential for a sustainable future for both farming and rural communities [19]. Jin et al. (2024) points out how the digital economy can really make a difference in cutting down on agricultural carbon emissions [20]. Secondly, the digital economy and agricultural productivity go hand in hand. Hu et al. (2024) found that digital advancements in rural areas are a boon to overall agricultural productivity [21]. Costa (2023) argues that economies that lean heavily on agriculture particularly rely on the environmental and social perks of tech to optimize operations and how resources are used [22]. Lastly, the digital economy has been a major catalyst in modernizing agriculture. The importance of investment in digital tech just keeps growing, and ramping up agricultural digitization is where things are inevitably headed. Xu et al. (2022) emphasized how information and communications technology infrastructure plays a vital role in boosting carbon efficiency, scaling effectively, improving technical efficiency, and driving technological advancements [17]. So, all things considered, embracing digital transformation in agriculture is proving to be a smart move and a win–win situation for high-quality development, which benefits farmers, consumers, and the environment alike [23].

While existing studies have extensively explored the relationship between technological innovation and high-quality development and the relationship between the digital economy and innovation, relatively little attention has been paid to the mediating role of technological innovation between the two. First off, the digital economy is a huge engine for technical progress. Things like digital financial inclusion and digital trade have really given tech innovation a boost [24]. Secondly, tech innovation is absolutely fundamental for making agriculture better. Green tech improvements boost how efficiently we handle carbon emissions, which squares perfectly with the aims of top-notch agricultural development and protecting our regional ecosystems [25]. But, it is not just about the environment; innovation is also a big player in cutting down poverty. It shakes up how we interact, creates fresh ways to work together and helps farmers share the wealth, which is a win–win for governments and everyone involved [26]. The “digital countryside” idea is seen as a key way to push agriculture forward. Xia et al. (2019) pointed out that by using this strategy, we can really unlock agriculture’s potential by boosting scientific innovation and getting agricultural growth moving [1]. To sum it up, when tech innovation and digital transformation come together, it opens up brand new possibilities for achieving agricultural development that is both sustainable and high-quality. This highlights just how crucial they are for tackling the challenges modern agriculture faces.

Although the existing literature has extensively examined the interconnections between the digital economy, technological innovation, and high-quality agricultural development, several gaps remain. First, a comprehensive analysis from multiple perspectives is required to precisely assess the impact of the digital economy on high-quality agricultural growth. Second, further exploration of technological innovation as a mediating mechanism can provide deeper insights into the contribution of the digital economy to agricultural modernization and transformation, thereby informing more targeted policy recommendations. Based on these gaps, this study aims to construct a theoretical framework and develop an empirical model to systematically explore the intrinsic relationships between the digital economy, technological innovation, and high-quality agricultural development. Specifically, Section 3 presents a theoretical analysis of the relationship between the digital economy and high-quality agricultural development as well as the channel of technological innovation, and it formulates the corresponding research hypotheses. Section 4 provides an overview of the detailed discussion on variable selection, which includes the explained variable, the core explanatory variable, the control variables and the mechanism variable, together with the settings for both the baseline regression model and the spatial econometric model. In Section 5, the empirical results are presented and analyzed, covering the baseline regression outcomes, the robustness tests, and the mechanism tests for technological innovation. Section 6 further extends the analysis through a heterogeneity analysis and spatial effects tests. Section 7 compares the results of this study with those in the existing literature and provides specific policy recommendations.

3. Theoretical Analysis and Research Hypothesis

3.1. Defining High-Quality Agricultural Development

High-quality agricultural development extends beyond merely increasing production volumes and improving financial returns, as was traditionally emphasized. While enhancing output remains an important aspect of agricultural development [27], it also involves reimagining production methods, improving the quality of agricultural products, and modernizing the entire agricultural system [28]. At its core, high-quality agricultural development focuses on refining agricultural practices, aiming to produce superior products, enhance the efficiency of industries, and optimize production systems [29]. To achieve this, it is essential to integrate scientific advancements and reform traditional institutional frameworks [28], thereby optimizing agricultural productivity. This process not only entails improving the quality of agricultural products [30] but also promoting environmentally sustainable [31,32], resource-efficient [33], and innovative agricultural practices [34]. Ultimately, the goal is to achieve improvements on multiple fronts, including economic, social, and environmental outcomes. From an economic perspective, high-quality agricultural development seeks to enhance total factor productivity, increase farmers’ income, allocate resources more efficiently, and modernize agriculture by incorporating advanced technologies, improving management practices, and refining organizational structures. Socially, it involves enhancing the education and skill levels of farmers and strengthening the support systems for agricultural communities, thereby improving regional production efficiency [35]. Environmentally, the focus is on reducing resource consumption and mitigating the adverse effects of agricultural practices, thereby fostering more sustainable agricultural methods in the long term [27].

3.2. Digital Economy and High-Quality Agricultural Development

The digital economy has fundamentally reshaped agricultural production and management through the integration of advanced information technologies. These changes improve capital allocation, production efficiency, and market responsiveness, thereby promoting high-quality agricultural development. First, digital finance eases traditional financial constraints in agriculture by expanding access to credit; reducing information asymmetries; and facilitating targeted investments in modern equipment, land optimization, and skilled labor. These improvements not only broaden agricultural markets but also enhance farmers’ financial resilience and income stability [36,37]. Second, the adoption of digital technologies has accelerated the transformation of traditional agriculture to modern, data-driven agriculture. Digital platforms streamline supply chain coordination, reduce transaction costs, and improve the efficiency of farm-to-market logistics. By enhancing value chain integration, digitalization increases the added value of agricultural products, aligns production with changing consumer preferences, and promotes higher-quality standards across the industry. Third, big data, the Internet of Things, and artificial intelligence enable farmers to gain real-time insights into soil conditions, weather patterns, and crop health. These technologies improve communication efficiency, optimize resource use, and drive a precision-based transformation of agricultural practices, ultimately accelerating the modernization of the sector [38]. In addition to its direct impacts, the digital economy also generates spatial spillovers that enable agricultural development beyond local boundaries. According to the theory of spatial economics, digital innovations are not confined to individual regions but spread across geographic space through mechanisms such as trade links, labor mobility, knowledge diffusion, and technological externalities [39]. As digital infrastructure expands and interregional connectivity improves, neighboring regions will benefit from shared technological advances, market integration, and knowledge spillovers, thereby amplifying the regional benefits of digitally driven agricultural transformation. Based on these insights, we propose the following hypotheses:

Hypothesis 1.

The digital economy contributes to the promotion of high-quality agricultural development.

Hypothesis 2.

The impact of the digital economy on high-quality agricultural development exhibits spatial spillover effects.

3.3. Digital Economy, Technological Innovation, and High-Quality Agricultural Development

The digital economy and high-quality agricultural development are closely intertwined, with technological innovation serving as the key driving force (Figure 1). First, the digital economy stimulates technological innovation by facilitating investment. The rise of financial technologies has eased financing constraints traditionally imposed by conventional financial institutions, enabling small and medium-sized enterprises (SMEs) and startups to access capital more easily [40]. This increased financial support is directed toward research and development (R&D), advanced equipment, and market expansion, thereby fostering innovation in agriculture. Second, the platform and sharing economy model characteristic of the digital era help break down barriers to technological R&D and encourage cross-sector collaboration [41]. Through big data analytics and cloud computing, firms can identify market demands with greater precision, accelerating technological breakthroughs [29]. Moreover, the digital economy fosters an innovation ecosystem, promoting collaboration among businesses, research institutions, governments, and investors. This dynamic environment accelerates the commercialization and application of technological advancements in agricultural production. Technological innovation contributes to high-quality agricultural development in three key ways. First, it enhances production efficiency and optimizes resource allocation. Technologies such as precision agriculture and mechanized farming reduce waste, improve product quality, and increase economic returns for farmers. Second, it promotes sustainable agricultural practices. Green technologies and eco-friendly innovations—including biopesticides, soil health monitoring, and energy-efficient equipment—mitigate environmental damage while supporting ecological preservation and responsible resource management. Third, it strengthens agricultural resilience against external shocks. Advances in gene editing, improved germplasm, and climate-smart technologies enhance agriculture’s adaptability to climate change, pest outbreaks, and other uncertainties, ensuring food security and stable production. Finally, technological innovation is transforming the entire agricultural value chain. The integration of digital technologies, advanced supply chain management, and big data analytics injects intelligence into agricultural operations, facilitating the transition from traditional farming methods to modern, data-driven agricultural practices [42]. This leads to the formulation of the third research hypothesis:

Hypothesis 3.

The digital economy promotes high-quality agricultural development through technological innovation.

4. Research Design

4.1. Variable Selection

4.1.1. Explained Variable

The core focus of this study is the level of high-quality agricultural development, a complex and multidimensional concept that resists straightforward quantification. While previous research has attempted to capture this construct through multidimensional indicator systems, there is no universally accepted framework for selecting these indicators. Building upon a comprehensive analysis of high-quality agricultural development, this study identifies key indicators across five fundamental dimensions: growth catalysts, structural efficiency, sustainable practices, inclusive progress, and equitable distribution. This framework is informed by the evaluation models proposed by Cui et al. (2022) [43]. Each dimension plays a distinct role in shaping agricultural advancement. Growth catalysts drive progress in scientific and technological innovation as well as industrial upgrading, reflecting the sector’s capacity for innovation and its long-term sustainability potential. Structural efficiency highlights the necessity of coherence within agricultural subsystems and their integration with broader economic and social frameworks, emphasizing the importance of balanced and coordinated development. Sustainable practices serve as a guiding principle, necessitating a commitment to resource efficiency and environmental stewardship. Inclusive progress acknowledges the increasing globalization of agriculture, advocating for active participation in both domestic and international markets and resource networks. Equitable distribution represents the ultimate goal of high-quality agricultural development—ensuring that the benefits of growth are fairly distributed to enhance overall societal well-being. The specific indicators corresponding to these dimensions are presented in

Table 1. Evaluation index system for high-quality agricultural development.

Indicator	Basic Indicators	Explanation	Direction	Weight
Growth Catalysts	Number of patents granted	Direct data	+	0.1023
	Agricultural GDP yield per acre	Agricultural output value per unit of sown area	+	0.0440
	Level of farm mechanization	Agricultural machinery power per unit of sown area	+	0.0420
	Three expenditures for science	Local government spending on science and technology as a percentage of the overall local budget	+	0.0475
Structural Efficiency	Level of industrial coordination	Contribution of the primary sector to GDP	+	0.0261
	Consumption level of rural residents	Per capita rural consumption expenditure	+	0.0214
	Government subsidies for agriculture	Local government spending on agriculture, forestry, and water resources as a percentage of the overall local budget	+	0.0185
	Rural Engel coefficient	Proportion of rural residents’ consumption expenditure allocated to food	−	0.0066
	Agricultural sector restructuring index	Ratio of the output of agriculture, forestry, livestock, and fisheries to the total output of agriculture	+	0.017
Sustainable Practice	Forest coverage rate	Forest area relative to total land area	+	0.0330
	Pesticide use per unit area	Pesticide application per total sown acreage	−	0.0040
	Agricultural plastic film usage per acre	Use of agricultural plastic films per total sown acreage	−	0.0050
	Fertilizer use per unit area	Volume of agricultural fertilizer applied compared to the total area planted	−	0.0093
	Total agricultural water use	Direct data	−	0.0112
	Agricultural energy consumption	Ratio of energy consumption of agriculture, forestry, animal husbandry, and fishery to output value	−	0.0054
Inclusive Progress	Reliance on agricultural exports	Export volumes of agricultural products as a fraction of primary industry value added	+	0.1577
	Reliance on agricultural imports	Import volumes of agricultural products as a fraction of primary industry value added	+	0.3287
Equitable Distribution	Number of village clinics	Direct data	+	0.0557
	Urban–rural consumption gap	Ratio of urban to rural per capita disposable income	−	0.0042
	Urban–rural income ratio	Ratio of urban to rural per capita consumption expenditure	−	0.0102
	Per capita disposable income of residents	Direct data	+	0.0300
	Rural residents’ living standards	Rural per capita spending on culture, education, and entertainment	+	0.0206

"+" indicates that the indicator has a positive impact on high-quality agricultural development. "−" indicates that the indicator has a negative impact on high-quality agricultural development.

. Given the heterogeneity of these dimensions, the metrics used to measure them vary significantly in scale and units, making direct aggregation challenging. To address this issue, the entropy method is employed to determine the indicator weights and compute a unified composite index of high-quality agricultural development. The entropy method, grounded in information theory, objectively assigns weights based on the degree of dispersion of each indicator across regions and time periods, thereby reducing subjective bias and improving the robustness of the index construction. The specific calculation method is as follows:

First, deal with the positive and negative indicators, where max{χ_ij} is the maximum value and min{χ_ij} is the minimum value.

Calculate the positive indicator:

$${\chi }_{ij}={\displaystyle \frac{{\chi }_{ij}-\min \left\{{\chi }_{ij}\right\}}{\max \left\{{\chi }_{ij}\right\}-\min \left\{{\chi }_{ij}\right\}}}$$

(1)

Calculate the negative indicator:

$${\chi }_{ij}={\displaystyle \frac{\max \left\{{\chi }_{ij}\right\}-{\chi }_{ij}}{\max \left\{{\chi }_{ij}\right\}-\min \left\{{\chi }_{ij}\right\}}}$$

(2)

Second, calculate the share of indicator j in year i:

$${\omega }_{ij}={\displaystyle \frac{{\chi }_{ij}^{\prime }}{{\displaystyle {\sum }_{i=1}^m{\chi }_{ij}^{\prime }}}}$$

(3)

Thirdly, calculate the information entropy (e_j) and redundancy (d_j) of the indicator. The number of years to be evaluated is m.

$$e_j=-{\displaystyle \frac{1}{\ln m}}{\displaystyle {\sum }_{i=1}^m({\omega }_{ij}}\times \ln {\omega }_{ij}),0\le e\le 1$$

(4)

$$d_j=1-e_j$$

(5)

Fourth, calculate the weights of metrics v_j based on the information entropy redundancy.

$$v_j={\displaystyle \frac{d_j}{{\displaystyle {\sum }_{j=1}^m{d}_j}}}$$

(6)

Finally, calculate the target variable using the weighted method.

$$X_i={\displaystyle {\sum }_{j=1}^m{v}_j}\times {\omega }_{ij}$$

(7)

4.1.2. Core Explanatory Variable

This study adopts the evaluation framework proposed by Lyu et al. (2024) [44] to measure the level of digital economy development. The framework consists of 13 indicators spanning three key dimensions: digital infrastructure, digital industrialization, and digital financial inclusion. Digital infrastructure reflects the foundational conditions for connectivity and information flow in rural areas. Indicators such as domain count, IPv4 site count, mobile phone penetration, broadband access points, broadband access users, long-distance cable density, and cell phone base station density are included to capture the breadth and quality of regional digital access. Digital industrialization measures the integration of digital technologies into the broader economic system. It includes indicators like e-commerce revenue, number of business websites, software industry output, e-commerce penetration rate, the number of informatization enterprises, the number of 5G invention patent applications, and express volume. These indicators collectively reflect the maturity and scope of digital economic activities. Digital-inclusive finance measures the accessibility and functionality of digital financial services that support rural and agricultural development. It is divided into three indices: the digitization index covers the proportion of mobile payments, the affordability of loan interest rates, the proportion of credit-based payments, and the prevalence of QR code payments; the depth of use index comprises the frequency, amount, and user participation related to services such as payments, money market funds, credit, insurance, investment, and credit scoring; and the breadth of coverage index includes the number of Alipay accounts per 10,000 people, the proportion of Alipay users who have linked bank cards, and the average number of bank cards linked per Alipay account [45]. To ensure objectivity and avoid subjective bias in indicator weighting, this study adopts the entropy method. The specific metrics and calculation details are presented in

Table 2. Evaluation index system for digital economy development.

Primary Indicator	Secondary Indicators	Direction	Weight
Digital infrastructure	Domain count	+	0.0058
	IPv4 site count	+	0.0785
	Mobile phone adoption rate	+	0.0130
	Number of access points for broadband internet	+	0.0348
	Number of access users for broadband internet	+	0.0401
	Long-distance cable length per unit area	+	0.0750
	Cell phone base station density	+	0.1025
Digital industrialization	Sales generated from e-commerce	+	0.0848
	Websites available per 100 businesses	+	0.0065
	Software industry revenue as a share of GDP	+	0.1116
	Percentage of businesses engaged in e-commerce	+	0.0206
	Number of informatization enterprises	+	0.1141
	Number of 5G invention patent applications	+	0.1456
	Express volume	+	0.1217
Digital-inclusive finance	Digitization index	+	0.0175
	Depth of use index	+	0.0159
	Breadth of coverage index	+	0.0120

"+" indicates that the indicator has a positive impact on digital economy development.

.

4.1.3. Control Variables

To account for potential confounding factors, the following control variables are incorporated into the analysis: (1) urbanization level (Urb), in which the acceleration of urbanization is closely associated with the expansion of digital infrastructure [46], which directly affects agricultural productivity and the adoption of digital technologies in agriculture; (2) government intervention (Gov), in which policy support, financial input, tax incentives, and other methods can directly or indirectly promote farmers to improve agricultural production efficiency and quality [47]; (3) rural human capital (Edu), in which the quality of human capital in rural areas is a key determinant of high-quality agricultural development, influencing both innovation adoption and productivity enhancement [40]; (4) industrial structural adjustment (Str), in which the degree of transformation of industrial structures is critical to improving agricultural productivity and efficiency [48]; (5) openness to external markets (Ope), in which integration with external markets influences high-quality agricultural development by fostering competitive pressures and facilitating technological exchanges [49]; (6) resource endowment (Res), in which the availability of natural and economic resources establishes the foundational conditions for the development of high-quality agriculture [50]; and (7) agricultural technology training (Att), which plays a critical role in equipping farmers with the knowledge and skills necessary to adopt and effectively use modern agricultural practices and technologies [51]. The measurement methods are presented in

Table 3. Variable definitions and measurement methods.

Variable Type	Variable Name	Symbol	Measurement Method
Explained Variable	High-quality agricultural development	Agr	Entropy method
Core Explanatory Variable	Digital economy	Dig	Entropy method
Control Variables	Urbanization level	Urb	Urban population as a percentage of year-end total residents
	Government intervention	Gov	Ratio of general fiscal budget expenditures to gross regional product
	Rural human capital	Edu	Average duration of formal education among rural inhabitants
	Industrial structure adjustment	Str	Added value from the secondary and tertiary sectors/the added value from the primary sector
	Openness to external markets	Ope	(Total import and export of goods × USD to RMB exchange rate)/gross regional product
	Regional resource endowment	Res	Logarithm of total water resources by region
	Agricultural technology training	Att	Logarithm of the number of graduates from rural adult cultural and technical training schools

.

4.1.4. Mechanism Variable

To examine the mediating role of technological innovation in the relationship between the digital economy and high-quality agricultural development, we construct a series of mechanism variables that capture innovation input, output, and efficiency. This study adopts the framework proposed by Bilbao-Osorio and Rodríguez-Pose (2004) [52]. Three key indicators are employed: innovation input (Tec₁), measured by R&D expenditure as a percentage of GDP; innovation output (Tec₂), measured by the standardized value of agricultural science and technology patents; and innovation efficiency (Tec₃), measured using the undesirable slack-based measurement (Undesirable SBM) model combined with the Global Malmquist–Luenberger (GML) productivity index, which evaluates agricultural green total factor productivity. The indicator system is presented in

Table 4. Evaluation index system for agricultural green total factor productivity.

Indicator	Basic Indicators	Explanation
Input Factors	Labor input	Number of employees engaged in crop production
	Land input	Total sown area of crops
	Agricultural machinery input	Total power of agricultural machinery
	Fertilizer input	Quantity of chemical fertilizers used in agriculture
	Pesticide input	Amount of pesticide applied
	Agricultural film input	Amount of agricultural plastic film used
	Irrigation input	Area of effectively irrigated agricultural land
Desirable Outputs	Gross agricultural output value	Total output value of crop production
	Agricultural carbon sequestration	Total amount of carbon sequestration by crops
Undesirable Outputs	Agricultural carbon emissions	Total carbon emissions from crop production
	Agricultural non-point source pollution	Discharge of non-point source pollutants from crop production

. Among them, innovation input reflects the level of resources allocated to R&D activities at the national or regional level, serving as an indicator of a region’s commitment to agricultural technological advancement [53]. Innovation output refers to the tangible results generated from innovation activities [54]. An increase in the number of agricultural technology patents can indicate that innovative scientific discoveries are successfully translated into practical agricultural technologies. Innovation efficiency reflects the effectiveness with which innovation inputs are transformed into productivity gains. By employing the SBM-GML index, it captures both desirable and undesirable outputs, thereby providing a more comprehensive and accurate assessment of innovation performance under environmental constraints. The specific formula of the model is as follows:

$$\begin{array}{l}\rho =\min {\displaystyle \frac{1+\frac{1}{m}{\displaystyle \sum _{i=1}^m\frac{p_i^{-}}{x_{i0}}}}{1-\frac{1}{q_1+q_2}\left({\displaystyle \sum _{r=1}^{q_1}\frac{p_r^{+}}{y_{r0}}}+{\displaystyle \sum _{t=1}^{q_2}\frac{p_t^{b-}}{b_{t0}}}\right)}}\\ s.t.\left\{\begin{array}{l}{\displaystyle \sum _{j=1,j\ne j_0}^n{x}_j{\lambda }_j-p^{-}\le x_0(i=1,\cdot \cdot \cdot ,m)}\\ {\displaystyle \sum _{j=1,j\ne j_0}^n{x}_j{\lambda }_j-p^{-}\le x_0(i=1,\cdot \cdot \cdot ,m)}\\ {\displaystyle \sum _{j=1,j\ne j_0}^n{x}_j{\lambda }_j-p^{-}\le x_0(i=1,\cdot \cdot \cdot ,m)}\\ 1-{\displaystyle \frac{1}{q_1+q_2}}\left({\displaystyle \sum _{r=1}^{q_1}\frac{p_r^{+}}{y_{r0}}}+{\displaystyle \sum _{t=1}^{q_2}\frac{p_t^{-}}{b_{t0}}}\right)>0\\ {\lambda }_j,p_i^{-},p_r^{+},p_t^{b-}\ge 0(j=1,\cdot \cdot \cdot ,n,j\ne j_0)\end{array}\right.\end{array}$$

(8)

where ρ is the efficiency value; j is the number of decision units; m, q₁, and q₂ are the index numbers of input, expected output, and non-expected output, respectively; $p_i^{-}, p_r^{+}, \mathrm{a}\mathrm{n}\mathrm{d} p_t^{b-}$ are their corresponding relaxation variables, respectively.

The GML index can measure the change of agricultural green total factor productivity, and the formula is as follows:

$$GM{L}^{t,t+1}(x^t,y^{\mathrm{t}},b^{\mathrm{t}},x^{t+1},y^{\mathrm{t}+1},b^{\mathrm{t}+1})={\displaystyle \frac{1+{\overrightarrow{D}}^G(x^t,y^{\mathrm{t}},b^{\mathrm{t}})}{1+{\overrightarrow{D}}^G(x^{t+1},y^{\mathrm{t}+1},b^{\mathrm{t}+1})}}$$

(9)

The GML index is decomposed into technical efficiency change index EC and technological progress change index TC:

$$GM{L}^{t,t+1}=E{C}^{t,t+1}\times T{C}^{t,t+1}$$

(10)

$$E{C}^{t,t+1}={\displaystyle \frac{1+{\overrightarrow{D}}^t(x^t,y^{\mathrm{t}},b^{\mathrm{t}})}{1+{\overrightarrow{D}}^{t+1}(x^{t+1},y^{\mathrm{t}+1},b^{\mathrm{t}+1})}}$$

(11)

$$T{C}^{t,t+1}={\displaystyle \frac{1+{\overrightarrow{D}}^G(x^t,y^{\mathrm{t}},b^{\mathrm{t}})}{1+{\overrightarrow{D}}^t(x^t,y^{\mathrm{t}},b^{\mathrm{t}})}}\times {\displaystyle \frac{1+{\overrightarrow{D}}^{t+1}(x^{t+1},y^{\mathrm{t}+1},b^{\mathrm{t}+1})}{1+{\overrightarrow{D}}^G(x^{t+1},y^{\mathrm{t}+1},b^{\mathrm{t}+1})}}$$

(12)

Among them, GML represents agricultural green total factor productivity, and EC and TC reflect technical efficiency and technological progress, respectively. When these indicators exceed 1, it indicates that agricultural green total factor productivity has increased over the previous period.

4.2. Data Sources and Variable Descriptive Statistics

The data used in this study came from the China Statistical Yearbook, the China Water Resources Bulletin, the China Rural Statistical Yearbook, the China Information Yearbook, the China Industrial Statistical Yearbook, and provincial Statistical Yearbooks. Some missing data were filled in by linear interpolation. The descriptive statistics are presented in

Table 5. Variable descriptive statistics.

Variables	N	Mean	SD	Min	Max
Agr	330	0.190	0.0850	0.0850	0.669
Dig	330	0.125	0.0920	0.0130	0.444
Urb	330	0.607	0.117	0.363	0.896
Gov	330	0.249	0.102	0.107	0.643
Edu	330	7.831	0.607	5.848	9.884
Str	330	29.590	70.280	2.957	459.300
Ope	330	0.260	0.274	0.008	1.441
Res	330	6.067	1.435	2.092	8.082
Att	330	12.47	2.095	6.332	15.54
Tec1	330	0.0180	0.0120	0.00400	0.0680
Tec2	330	0.0310	1.001	−0.964	4.302
Tec3	330	0.701	0.234	0.251	1.201

. Specifically, the mean value of high-quality agricultural development (Agr) is 0.190, with a standard deviation of 0.085 and a range from 0.0850 to 0.669. This indicates that the variation in high-quality agricultural development across China is relatively small, with a concentrated level of development. The mean value of the digital economy (Dig) is 0.125, with a standard deviation of 0.092 and a slightly higher level of dispersion, ranging from 0.0130 to 0.444, within a narrower interval. This suggests that the regional heterogeneity in digital economy development is stronger than that of high-quality agricultural development. This difference may be attributed to variations in digital infrastructure and policy support across different regions.

4.3. Model Design

4.3.1. Baseline Regression Model

Building on the previous theoretical analysis, this study seeks to empirically assess the impact of the digital economy on the high-quality development of agriculture. Drawing from the work of Belhadi et al. (2023) [55], the following model is formulated:

$$Ag{r}_{it}={\beta }_0+{\beta }_{1}Di{g}_{it}+{\beta }_{c}Contro{l}_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(13)

where i is the province and t denotes time; Agr indicates high-quality agricultural development; Dig represents digital economy; Control refers to a set of provincial-level control variables; and ε_it is a random error term.

In addition, the possible autocorrelation of high-quality agricultural development and its five dimensions in the time series, that is, the current level, is significantly affected by the previous level. At the same time, both the core explanatory variables and the explained variables are compound indicators. This may cause endogeneity problems in the model. In order to overcome the above problems, in addition to the fixed effect model for regression analysis, this study further introduces the differential generalized moment estimation (DIF-GMM) model and the system generalized moment estimation (SYS-GMM) model for robustness tests and compares and analyzes the estimation results with the fixed effect model to ensure the reliability of the research results. Specifically, the DIF-GMM model uses the lag level of the endogenous variable as a tool to convert the original equation into a first-order difference to eliminate unobtrusively individual effects. However, when regressions exhibit high persistence, the DIF-GMM may suffer from weak tools, potentially undermining the estimation efficiency. To overcome this limitation, the SYS-GMM model is used in this paper. SYS-GMM combines difference equations with horizontal equations, using lag difference as an instrument for horizontal equations in addition to the standard DIF-GMM instrument. This model improves the estimation efficiency and reduces the bias. To further verify the validity of the GMM models, diagnostic tests are conducted. The Arellano–Bond test (A-B test) is used to detect autocorrelation in the first-differenced residuals; valid specification requires first-order (AR (1)) but not second-order (AR (2)) autocorrelation. Moreover, both the Hansen test and the Sargan test of overidentifying restrictions are applied to evaluate the overall validity of the instruments. While the Hansen test is robust to heteroskedasticity, the Sargan test assumes homoskedasticity. If the test results do not reject the null hypothesis, this indicates that the instrument set is valid and the model is correctly specified. The results of the A-B test, Sargan test, and Hansen test are listed in the subsequent regression results to verify the robustness of the model specification and the reliability of the dynamic estimation.

4.3.2. Spatial Econometric Model

The digital economy exerts a ripple effect on the technological innovation capabilities of adjacent regions, which in turn shapes agricultural progress in those areas, so a spatial econometric analysis is necessary. Prior to the analysis, a series of diagnostic tests are required to determine the most appropriate model. First, the Hausman test is used to determine whether to apply a fixed effects or random effects model. Second, a Likelihood Ratio (LR) test is conducted, where the Spatial Durbin Model (SDM) is assumed initially, and then, comparisons are made to determine if it deteriorates into a Spatial Error Model (SEM) or Spatial Lag Model (SLM). Third, a Wald test is performed, also to assess whether the SDM degrades into the SEM or SLM. Finally, the results of these tests are compared to determine the most suitable spatial econometric model. The test results are shown in

Table 6. Hausman test, LR test, and Wald test results.

Method	Results
	0–1 Neighborhood Matrix	Geographic Distance Matrix	Economic Geography Matrix
LR spatial lag	81.137 ***	85.119 ***	127.837 ***
LR spatial error	93.708 ***	95.342 ***	58.785 ***
Wald spatial lag	104.63 ***	45.32 ***	48.84 ***
Wald spatial error	90.67 ***	27.23 ***	22.67 ***
Hausman Test	21.81 ***	22.98 ***	30.92 ***

*** p < 0.01.

. The Hausman test results indicate that the null hypothesis can be rejected, suggesting the use of a fixed effects model. Furthermore, both the LR and Wald test results support the selection of SDM. Therefore, this study chooses SDM for the empirical analysis. The model is as follows:

$$Ag{r}_{it}={\beta }_0+\rho {\displaystyle {\sum }_{j=1}^n{W}_{ij}}Ag{r}_{it}+\alpha Di{g}_{it}+{\theta }_1{\displaystyle {\sum }_{j=1}^n{W}_{ij}}Di{g}_{it}+\beta Control{s}_{it}+{\theta }_2{\displaystyle {\sum }_{j=1}^n{W}_{ij}}Control{s}_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}$$

(14)

Here, ρ represents the spatial autocorrelation coefficient, capturing the impact of surrounding regions on agricultural development (Agr). The coefficient θ₁ denotes the spatial lag term, reflecting the influence of neighboring provinces. The element W_ij is derived from the spatial weight matrix W. This study utilizes three distinct spatial weight matrices: the neighboring spatial weight matrix, the geographic distance spatial weight matrix, and the economic–geographic nested matrix. The neighboring spatial weight matrix is employed due to the significant spatial spillover effects of the digital economy, where neighboring regions influence each other’s agricultural development through mechanisms such as trade, labor mobility, and knowledge diffusion. The geographic distance spatial weight matrix is selected to capture the effect of geographical proximity, as regions that are closer in physical distance tend to have stronger interactions in areas such as resource sharing and technological transfer, which are crucial for agricultural development. The economic–geographic nested matrix combines both economic and geographical factors, with the element W_ij computed as (1/GDP per capita) × 0.5 + (1/geographic distance) × 0.5. This matrix is chosen to reflect the dual influence of economic capacity and spatial proximity, providing a more comprehensive representation of the spatial dynamics shaping agricultural growth.

5. Results and Analysis

5.1. Baseline Estimate

The results are listed in

Table 7. Baseline regression results.

Variables	FE	DIF-GMM	SYS-GMM
	(1)	(2)	(3)
Dig	0.0384 ***	0.0856 ***	0.0952 ***
	(0.0146)	(0.0094)	(0.0347)
Urb	0.3548 ***	−0.2443	−0.1105 **
	(0.0246)	(0.1909)	(0.0451)
Gov	−0.0129	−0.0026	−0.1543 ***
	(0.0277)	(0.0806)	(0.0537)
Edu	0.0101 **	−0.0074	−0.0047
	(0.0046)	(0.0138)	(0.0037)
Str	0.0012 ***	0.0009 ***	0.0007 ***
	(0.0000)	(0.0002)	(0.0001)
Ope	−0.0033	0.0597 *	0.0312 **
	(0.0127)	(0.0336)	(0.0127)
Res	−0.0029	−0.0031	0.0057 *
	(0.0025)	(0.0056)	(0.0029)
Att	0.0008	0.0065 ***	−0.0015
	(0.0010)	(0.0025)	(0.0012)
_cons	−0.1336 ***	−0.5458 ***	0.1865 ***
	(0.0427)	(0.0698)	(0.0671)
AR (2)		0.230	0.107
Sargan		0.896	0.796
Hansen		1.000	1.000
N	330	270	300
R2	0.8915

Values in parentheses are robust standard errors, * p < 0.1, ** p < 0.05, *** p < 0.01.

. The results of fixed effect estimation show that the digital economy has a positive effect on promoting high-quality agricultural development and passes the 1% significance level test. In the GMM models, the AR (2) test results indicate that the residuals of the regression estimates do not exhibit second-order autocorrelation. The statistics of the Sargan and Hansen tests are not significant, suggesting that the null hypothesis of instrument validity cannot be rejected. Therefore, the model is valid. And, the coefficient of Dig was significantly positive at, at least, 5%, that is, the digital economy can promote the high-quality development of agriculture, which verified research hypothesis 1.

Among the control variables, industrial structure adjustment (Str) shows a consistently positive and significant effect across all models, suggesting that economic restructuring toward advanced industries contributes to improved agricultural quality, possibly through technology diffusion or improved resource allocation. Similarly, rural human capital (Edu) is positively associated with agricultural development, although the significance weakens in the GMM models, indicating its potential long-term rather than immediate effect. Other variables exhibit inconsistent coefficient signs and varying levels of significance across different models. This may be attributed to differences in implementation, adoption, and local context, or the disparity between short-term and long-term effects.

5.2. Robustness Test

5.2.1. Replacement of Core Variables

Building on the framework established by Li and Yang (2024) [56], this study redefines the assessment of the digital economy. Utilizing a principal component analysis (PCA), a variety of indicators are synthesized, including year-end permanent resident population, mobile phone and internet penetration rates, broadband and mobile user data, per capita and total telecommunications revenue, employment figures in the IT and software sectors, and the digital financial inclusion index. These consolidated indicators are then subjected to a regression analysis, with the results presented in

Table 8. Robustness tests (1).

Variables	Replacement of Core Explanatory Variables			Replacement of Explained Variable
	FE	DIF-GMM	SYS-GMM	FE	DIF-GMM	SYS-GMM
	(1)	(2)	(3)	(4)	(5)	(6)
Dig_PCA	0.1194 ***	0.1209 ***	0.1155 ***	1.4741 ***	0.5512 ***	0.6734 ***
	(0.0325)	(0.0141)	(0.0142)	(0.4913)	(0.1448)	(0.1070)
_cons	0.1096	−0.2116 ***	−0.0701 ***	−1.7463	−2.6853 ***	−0.3915 **
	(0.0743)	(0.0287)	(0.0214)	(1.0897)	(0.3239)	(0.1751)
Control	Yes	Yes	Yes	Yes	Yes	Yes
AR (2)		0.429	0.494		0.107	0.114
Sargan		0.996	0.505		0.828	0.247
Hansen		0.999	0.867		1.000	1.000
N	330	270	300	330	270	300
R2	0.9841			0.9866

Values in parentheses are robust standard errors, ** p < 0.05, *** p < 0.01.

columns (1)–(3). Notably, the core explanatory variable continues to show a positive coefficient. To strengthen the robustness of the findings, this study further analyzes the dependent variable using PCA, and the results in Table 8 columns (4)–(6) remain consistent with the initial regression outcomes.

5.2.2. Removal of Anomalous Data

Recognizing that Beijing, Shanghai, Guangdong, and Zhejiang are leaders in digital economy development and benefit from substantial national policy support, their data points significantly diverge from those of other regions. To minimize potential skewing effects, these four regions are excluded, and the regression is re-estimated using the remaining sample. The results in

Table 9. Robustness tests (2).

Variables	Removal of Anomalous Data			1% Bilateral Shrinkage			Considering External Shocks
	FE	DIF-GMM	SYS-GMM	FE	DIF-GMM	SYS-GMM	FE	DIF-GMM	SYS-GMM
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Dig	0.0186 ***	0.0135 ***	0.0130 ***	0.0350 **	0.0266 ***	0.0363 ***	0.0362 ***	0.1570 **	0.0159 **
	(0.0075)	(0.0013)	(0.0052)	(0.0157)	(0.0101)	(0.0107)	(0.0103)	(0.0732)	(0.0074)
_cons	0.1420 ***	−0.0910	0.0575	0.2249 ***	−0.0051	0.1337 **	0.1592 **	0.3182	−0.0705 *
	(0.0426)	(0.1056)	(0.0502)	(0.0781)	(0.0656)	(0.0647)	(0.0735)	(0.4895)	(0.0401)
Control	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
AR (2)		0.846	0.459		0.357	0.430		0.751	0.159
Sargan		0.309	0.577		0.997	0.697		0.143	0.095
Hansen		0.998	0.989		1.000	0.792		0.263	0.267
N	286	234	260	330	270	300	270	120	180
R2	0.972			0.9813			0.9824

Values in parentheses are robust standard errors, * p < 0.1, ** p < 0.05, *** p < 0.01.

confirm that the Dig coefficient remains significantly positive, underscoring the pivotal role of the digital economy.

5.2.3. 1% Bilateral Shrinkage

To address potential bias in the estimates due to outliers and extremes, this study reruns the regression analysis using a 1% bilateral shrinkage for all variables. The results in Table 9 are consistent with those of the baseline model.

5.2.4. Considering External Shocks

In 2018, the Ministry of Agriculture of China designated the year as the “Year of Agricultural Quality”, with a focus on advancing high-quality agricultural development through the standardization of agricultural production, monitoring of agricultural product quality, purification of the production environment, and enhancement of agricultural product branding. Furthermore, the outbreak of the COVID-19 pandemic in 2020 significantly impacted agricultural activities. Therefore, this study excludes these two years and re-conducts the regression analysis. The results remain consistent with the baseline regression, confirming the robustness of the model.

5.3. Mechanism Test

In order to examine whether technological innovation serves as a mediating channel in the relationship between digital economy and high-quality agricultural development, this section conducts mechanism tests. Considering that it takes a certain amount of time for technological innovation to emerge and be applied in agricultural practice, DIF-GMM and SYS-GMM are also used for regression by reference to baseline regression, and the results are shown in

Table 10. Mechanism tests.

Variables	Innovation Input			Innovation Output			Innovation Efficiency
	FE	DIF-GMM	SYS-GMM	FE	DIF-GMM	SYS-GMM	FE	DIF-GMM	SYS-GMM
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Dig	0.0041 **	0.0060 **	0.0030 **	2.1646 ***	1.6839 ***	0.6870 ***	0.0465 ***	0.3640 ***	0.2736 ***
	(0.0016)	(0.0030)	(0.0015)	(0.5824)	(0.6283)	(0.1773)	(0.0148)	(0.0998)	(0.0342)
_cons	0.0089	−0.0041	−0.0018	−0.1026	0.8201	−4.4712 ***	0.9013	0.1709	0.2710 **
	(0.0082)	(0.0104)	(0.0034)	(2.0048)	(4.3353)	(0.6165)	(0.5697)	(0.6635)	(0.1112)
Control	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
AR (2)		0.273	0.257		0.076	0.082		0.968	0.589
Sargan		0.943	0.407		0.981	0.231		0.250	0.651
Hansen		1.000	1.000		1.000	0.580		0.993	0.757
N	330	270	300	330	270	300	330	270	300
R2	0.9847			0.8391			0.8364

Values in parentheses are robust standard errors, ** p < 0.05, *** p < 0.01.

. Regarding innovation inputs, columns (1)–(3) indicate that the Dig coefficient is positive, suggesting that the digital economy increases innovation inputs in agriculture. This effect is attributable to the extensive application of information technology and the development of digital infrastructure, which effectively reduces innovation costs. Second, regarding innovation output, columns (4)–(6) indicate that Dig is positive, suggesting that the digital economy positively influences the output of agricultural technology innovations. Specifically, the digital economy accelerates the research, development, and diffusion of new agricultural technologies and products through information sharing, network synergy, and technology diffusion. Finally, in terms of innovation efficiency, columns (7)–(9) demonstrate that the Dig coefficient remains positive, suggesting that the digital economy improves the overall efficiency of agricultural innovation. This improvement is attributed to the promotion of data-driven decision-making, which effectively enhances resource utilization rates and the conversion of technological achievements during the innovation process. This transformation allows for agricultural innovation to evolve from mere inputs to high-quality outputs. Thus, hypothesis 3 is confirmed.

6. Further Analysis

To explore the intricate role of the digital economy across different economic conditions and policy frameworks, this paper delves deeper into several key factors. First, shifts in the broader economic climate, particularly related to intellectual property protection and the income divide, are pivotal in assessing whether the digital economy can foster technological innovation and drive agricultural growth. Second, recognizing that the digital economy’s impact on agriculture may extend beyond regional borders, a spatial analysis is conducted to evaluate the spillover effects.

6.1. Heterogeneity Analysis

6.1.1. Degree of Intellectual Property Protection

The integration of digital technologies into agriculture inevitably raises issues concerning intellectual property rights. Therefore, understanding how varying levels of intellectual property rights (IPR) protection influence outcomes is critical. This study measures the strength of regional IPR protection by examining the volume of settled patent infringement disputes within a province [57]. Provinces were divided based on the median level of IPR protection into two categories: those with robust protection and those with weak protection. As shown in

Table 11. Heterogeneity of degree of intellectual property protection.

Variables	Higher Degree of Intellectual Property Protection			Lower Degree of Intellectual Property Protection
	FE	DIF-GMM	SYS-GMM	FE	DIF-GMM	SYS-GMM
	(1)	(2)	(3)	(4)	(5)	(6)
Dig	0.0335 ***	0.0351 ***	0.0200 ***	0.0229	0.0021	0.0139
	(0.0054)	(0.0023)	(0.0012)	(0.0171)	(0.0077)	(0.0101)
_cons	0.5930 ***	0.2622 **	0.4756 ***	0.0904	0.0036	0.0201
	(0.1148)	(0.1281)	(0.1700)	(0.0592)	(0.0035)	(0.0473)
Control	Yes	Yes	Yes	Yes	Yes	Yes
AR (2)		0.212	0.608		0.374	0.354
Sargan		0.997	0.351		0.229	0.934
Hansen		1.000	1.000		1.000	1.000
N	174	162	169	156	108	131
R2	0.9137			0.9808

Values in parentheses are robust standard errors, ** p < 0.05, *** p < 0.01.

, the coefficients in columns (1)–(3) remain positive, suggesting that the digital economy fosters high-quality agricultural development in provinces with strong IPR protection. In contrast, the coefficients in columns (4)–(6) are statistically insignificant, indicating that in provinces with weak IPR protection, the digital economy does not have a significant impact. This disparity is likely due to the stronger legal frameworks and enforcement mechanisms in regions with robust IPR protection, which incentivize agricultural businesses and farmers to invest in and adopt advanced technologies. Conversely, in regions where IPR enforcement is weak, innovation is often stifled due to the lack of legal safeguards, deterring businesses and farmers from investing in digital technologies. The fear of intellectual property theft diminishes their willingness to engage with technological advancements, ultimately hindering the potential of the digital economy to drive agricultural innovation.

6.1.2. Urban–Rural Income Gap

The urban–rural income gap is a crucial indicator of the unequal distribution of resources and the economic divide between urban and rural areas [58]. In regions with a significant income gap, farmers are often more motivated to increase their income, which makes them more inclined to adopt and advance digital technologies. This study uses the ratio of per capita disposable income between urban and rural households to quantify this divide [59]. The sample was divided into two groups: those with a large income gap and those with a smaller one, and regression analyses were performed for each. As shown in columns (1)–(3) of

Table 12. Heterogeneity of urban–rural income gap.

Variables	Greater Urban–Rural Income Gap			Smaller Urban–Rural Income Gap
	FE	DIF-GMM	SYS-GMM	FE	DIF-GMM	SYS-GMM
	(1)	(2)	(3)	(4)	(5)	(6)
Dig	0.1027 ***	0.1257 ***	0.0631 **	0.0050	0.0701 *	0.0200
	(0.0328)	(0.0195)	(0.0272)	(0.0106)	(0.0419)	(0.0612)
_cons	0.3765 ***	0.0186	−0.1111	0.0587	0.0455	−0.0270
	(0.1430)	(0.0724)	(0.1150)	(0.0766)	(0.1479)	(0.1462)
Control	Yes	Yes	Yes	Yes	Yes	Yes
AR (2)		0.364	0.332		0.143	0.106
Sargan		0.990	0.522		0.998	0.817
Hansen		1.000	1.000		1.000	0.950
N	174	162	169	156	108	131
R2	0.9763			0.9948

Values in parentheses are robust standard errors, * p < 0.1, ** p < 0.05, *** p < 0.01.

, a positive correlation between the digital economy and income is observed in the high-disparity group, while columns (4)–(6) indicate no statistically significant impact in the low-disparity group. This disparity may be due to the fact that in regions with a large urban–rural income gap, rural residents face a more urgent need to improve their living standards and agricultural productivity. The digital economy helps address this need by optimizing resource allocation and creating new market opportunities and jobs. In contrast, in areas where the income gap is smaller, the demand-driven effect is relatively limited, and the marginal benefits of the digital economy are less pronounced.

6.2. Spatial Effects Test

6.2.1. Spatial Correlation Analysis

Before building a spatial measurement model, it is essential to determine if the data show any spatial dependence. The Global Moran Index formula is as follows:

$$I={\displaystyle \frac{{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^{n}w{}_{ij}(x_i-\overline{x})}}(x_j-\overline{x})}{s^2{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n{w}_{ij}}}}}$$

(15)

where w_ij is the spatial weight matrix. Global Moran’s I ranges from −1 to 1, with I > 0 signifying positive spatial correlation, I < 0 indicating negative correlation, and I ≈ 0 reflecting spatial independence.

Table 13. Global Moran’s I for high-quality agricultural development.

Year	0–1 Neighborhood Matrix		Geographic Distance Matrix		Economic Geography Matrix
	Moran’s I	Z Value	Moran’s I	Z Value	Moran’s I	Z Value
2012	0.312 ***	0.000	0.311 ***	0.000	0.106 ***	0.000
2013	0.343 ***	0.000	0.345 ***	0.000	0.120 ***	0.000
2014	0.321 ***	0.000	0.338 ***	0.000	0.115 ***	0.000
2015	0.288 ***	0.001	0.299 ***	0.000	0.098 ***	0.000
2016	0.281 ***	0.001	0.270 ***	0.000	0.090 ***	0.000
2017	0.227 ***	0.005	0.208 ***	0.002	0.065 ***	0.001
2018	0.219 ***	0.006	0.193 ***	0.004	0.058 ***	0.001
2019	0.195 ***	0.010	0.166 ***	0.009	0.046 ***	0.004
2020	0.223 ***	0.005	0.201 ***	0.003	0.059 ***	0.001
2021	0.174 **	0.015	0.132 **	0.021	0.035 ***	0.009
2022	0.175 **	0.015	0.127 **	0.025	0.033 **	0.011

** p < 0.05, *** p < 0.01.

presents the Global Moran’s I results across three weight matrices, with all values being significantly positive and passing the Z-test, indicating strong spatial dependence in agricultural high-quality development.

To further identify local aggregations and outliers in spatial data, this paper also uses the local Moran index; the formula is as follows:

$$I_i=z_i{\displaystyle {\sum }_{i=1}^n{w}_{ij}{z}_{ij}}$$

(16)

where z_i = x_i − $\overline{x}$, z_j = x_j − $\overline{x}$. I_i can be described by a Moran scatter plot.

Figure 2 shows from top to bottom the creation of local Moran’s I scatter plots in 30 provinces of China in 2012 and 2022 using an adjacency spatial weight matrix, a geographical distance spatial weight matrix, and an economic geography nested matrix. The local Moran’s I of most provinces is in the category of High–High regions, characterized by high-value clusters, and Low–Low regions, characterized by low-value clusters, indicating the existence of positive spatial autocorrelation. Therefore, the above analysis shows that spatial econometric model estimation is necessary.

6.2.2. Space–Time Evolution Characteristics Analysis

To further investigate the spatiotemporal evolution of high-quality agricultural development and the digital economy in China, this study utilizes ArcGIS to analyze the temporal and spatial characteristics over six selected years: 2012, 2014, 2016, 2018, 2020, and 2022.

From a temporal perspective, as illustrated in Figure 3, the overall level of high-quality agricultural development in China has shown a steady improvement. The variations in the upper and lower limits of the values across the years suggest an increasing dispersion at the national level, indicating that the development trajectories of provinces have not been fully synchronized. Some provinces have experienced rapid progress during certain periods, while others have lagged behind. From a spatial perspective, high-quality agricultural development reveals significant regional disparities. The central and certain eastern regions, such as Shandong and Zhejiang, are characterized by darker shades, indicating higher levels of development. This can be attributed to these regions’ strong agricultural foundations, abundant resources, and relatively advanced agricultural technologies and industrial systems. In contrast, the western and northeastern regions, including Xinjiang, Qinghai, and Neimenggu, are represented by lighter shades, indicating lower levels of development. This disparity is likely influenced by factors such as natural conditions, geographical location, and relatively insufficient investment in agricultural modernization. By integrating both temporal and spatial dimensions, two key observations emerge. First, the relative positioning of high-value and low-value regions remains largely stable over time, reflecting the path-dependent nature of agricultural development and the significant role of regional endowments. Second, the annual fluctuations in specific provincial values suggest that various regions are influenced by the combined effects of policies, market dynamics, and technological innovations at different times, leading to continuous variations in the level of high-quality agricultural development.

For the digital economy, Figure 4 illustrates that from 2012 to 2022, the highest levels of digital economy development have consistently increased, reflecting the rapid growth of the digital economy across China. Furthermore, changes in the value range over the years suggest a widening gap in the levels of digital economy development among provinces. From a spatial perspective, the development of the digital economy exhibits a clear east–west gradient disparity, with stronger development in the eastern regions and weaker development in the western regions. The eastern coastal provinces, such as Guangdong, Zhejiang, and Jiangsu, are characterized by darker shades, highlighting them as the leading regions in digital economy development. These provinces benefit from well-established information technology industries, abundant human capital, and comprehensive digital infrastructure. The central region shows moderate development, although still trailing behind the eastern coastal areas. In contrast, the western and northeastern regions, including provinces such as Xinjiang, Qinghai, and Neimenggu, are represented by lighter shades, indicating slower digital economy development. This lag can be attributed to factors such as lower overall economic development, insufficient technological investment, and talent outflows. From a spatiotemporal perspective, in the early stages, high-value clusters of digital economy development were primarily concentrated in a few eastern and central provinces. Over time, however, there has been a noticeable diffusion of high-value regions toward certain central provinces, demonstrating the spillover effects of digital economy development. These spillover effects may result from factors such as technology diffusion, industrial relocation, and other related processes.

The observed spatiotemporal evolution characteristics reveal that high-quality agricultural development and digital economy development are interconnected across regions. The level of digital economy development and its diffusion across different areas can significantly influence the process and patterns of agricultural–industrial integration. For example, in regions where the digital economy is well established, emerging agricultural business models, such as agricultural e-commerce and smart agriculture, are likely to evolve more rapidly. The experiences and models developed in these regions can then be disseminated to others through spatial spillover effects. Investigating these spillover effects offers valuable insights into how the diffusion of the digital economy across regions facilitates agricultural–industrial integration, as well as the opportunities and challenges faced by different areas in this process. And, this understanding can promote a more comprehensive and widespread integration of the digital economy with agriculture, contributing to high-quality agricultural development.

6.2.3. Decomposition of Spatial Effects

The direct, indirect, and overall impacts of the growth of the digital economy on the promotion of high-quality agriculture are significantly positive, as shown in

Table 14. Decomposition of spatial effects of spatial Durbin models.

Variables	0–1 Neighborhood Matrix			Geographic Distance Matrix			Economic Geography Matrix
	Direct	Indirect	Total	Direct	Indirect	Total	Direct	Indirect	Total
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)
Dig	0.018 ***	0.100 ***	0.118 ***	0.039 ***	0.182 ***	0.221 ***	0.044 ***	0.467 ***	0.511 ***
	(0.006)	(0.029)	(0.033)	(0.012)	(0.034)	(0.038)	(0.012)	(0.103)	(0.107)
Control	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
ρ		0.177 **			0.129			−0.077
		(0.069)			(0.092)			(0.193)
sigma2_e		0.000 ***			0.000 ***			0.000 ***
		(0.00)			(0.00)			(0.00)
N	330	330	330	330	330	330	330	330	330
R2	0.463	0.463	0.463	0.635	0.635	0.635	0.514	0.514	0.514

Values in parentheses are robust standard errors, ** p < 0.05, *** p < 0.01.

, which is positive for all three weighting methods. This shows that the development of the digital economy not only promotes the development of high-quality agriculture within the region but also plays a crucial role in improving agricultural practices in the surrounding areas. It has stimulated technological innovation, optimized resource allocation, and directly promoted the development of regional agriculture to a higher quality. In addition, the digital economy has generated technology spillover effects, expanding market access and increasing cooperation opportunities. These factors provide valuable learning experiences and external incentives that greatly promote the agricultural development of the surrounding area. Therefore, research hypothesis 2 is confirmed.

7. Conclusions and Implications

7.1. Discussion and Conclusions

This paper empirically examines the impact of the digital economy on high-quality agricultural development, as well as the mediating role of technological innovation, using panel data from 30 Chinese provinces (excluding the Tibet Autonomous Region) from 2012 to 2022. The results show that the digital economy significantly promotes high-quality agricultural development. This finding is consistent with that in the existing literature [1,60,61]. The mechanism analysis reveals that, first, the digital economy reduces innovation costs and increases innovation input. Second, it facilitates the application and diffusion of new agricultural technologies, thereby enhancing innovation output. Third, it improves innovation efficiency through data-driven optimization of resource allocation and decision-making processes, thus accelerating high-quality agricultural development. Gao and Lyu (2023) [61], and Zhang et al. (2024) [62] have also reached similar conclusions, namely that technological innovation is a core channel through which the digital economy promotes high-quality agricultural development.

In addition, this study highlights the regional heterogeneity of the digital economy’s effects, indicating that its positive impact is more pronounced in regions with stronger intellectual property protection and larger urban–rural income gaps. This suggests that a more robust intellectual property regime encourages enterprises to pursue digital technological innovation with greater confidence [63], while a wider urban–rural income gap drives farmers to adopt digital tools and practices more urgently [64]. In contrast, in regions with weaker institutional environments or more balanced income distribution, the motivation to embrace digital transformation may be less evident, which could limit its impact on agricultural advancement.

Moreover, this paper reveals that the digital economy has significant positive spatial spillover effects on high-quality agricultural development, promoting the interprovincial flow of innovative resources and enhancing agricultural quality across regions. This is consistent with previous research findings. Specifically, Hua et al. (2024) argue that the digital economy can promote the coordinated development of high-quality agriculture in neighboring regions [60]. Gao and Lyu (2023) point out that due to its characteristics of shareability and penetration, the digital economy can overcome spatial limitations and constraints, enabling the flow of information, production factors, and technology. Neighboring regions can enhance the breadth and depth of their agricultural activities by absorbing advanced agricultural technologies, experiences, and outstanding achievements from more developed regions, thereby improving the quality of agricultural development [61].

7.2. Policy Implications

Enhance digital infrastructure and reduce the digital divide: Governments should prioritize investments in rural digital infrastructure, focusing on expanding broadband access and enhancing mobile connectivity in remote areas. Programs aimed at increasing the rural penetration of digital technologies, such as subsidies for broadband installation, training for farmers, and setting up low-cost digital centers, are crucial. Special attention should be given to addressing the digital divide between urban and rural areas to ensure equal access to technological advancements, thereby enhancing productivity and agricultural quality.

Promote the adoption and technological innovation of smart agriculture: Policies should create incentives to encourage the widespread adoption of smart agricultural technologies in the sector, including precision irrigation, automated pest monitoring, and remote sensing tools. Public–private partnerships between technology providers, agribusinesses, and research institutions should also be strengthened to accelerate the commercialization of agricultural innovations through synergies, while fostering an environment conducive to innovation. This will ensure that technological advances effectively translate into productivity gains. Additionally, the establishment of an agricultural innovation fund could support research and development efforts, fostering the development of new agricultural technologies and enhancing the overall competitiveness of the sector.

Address regional disparities through tailored policy approaches: In regions with weak intellectual property protection, policy interventions should prioritize strengthening IP enforcement in the agricultural sector to ensure that innovators are adequately incentivized to commercialize their research and development outputs. In regions where income inequality is evident, targeted investments such as broadband infrastructure expansion and subsidies for digital adoption are essential to break barriers to technology adoption. Furthermore, governments should adopt region-specific strategies that account for local economic conditions. This includes offering financial incentives and ensuring access to affordable digital tools, thereby lowering the costs of digital technology adoption for smallholder farmers and enhancing the effectiveness and contextual relevance of policy measures.

7.3. Limitations and Future Research

Firstly, the sample used in this paper is limited to regions within China. This geographical concentration could limit the generalizability of the findings to other regions or sectors. Future research should expand the analytical scope by incorporating data from other countries to validate and extend the conclusions derived from this study.

Secondly, this paper relies on provincial panel data, which provide a macroscopic view of the digital economy’s impact on high-quality agricultural development. Consequently, the effects of digital transformation at the city or firm level remain unobserved. Future investigations could utilize more granular data to capture the influence of the digital economy on high-quality agricultural development at the micro-level, such as individual cities or enterprises.

Lastly, this paper focuses on the mechanism of technological innovation. However, other relevant factors such as policy support for digital agriculture and farmers’ digital literacy may also significantly influence the relationship between the digital economy and agricultural development. Future research could examine the impacts of these factors to provide a more nuanced understanding of how digital economy contributes to high-quality agricultural development.