The Impact of the Integration of Digital and Real Economies on Agricultural New Quality Productive Forces: Empirical Evidence from China’s Major Grain-Producing Areas

Abstract

As the digital economy becomes increasingly integrated with the real economy, agricultural production is experiencing fundamental transformation. Digital–real integration has emerged as strategically important for cultivating agricultural new quality productive forces and safeguarding national food security. This study examines provincial panel data from 13 major grain-producing regions in China between 2012 and 2023. We develop an evaluation index system to assess both digital–real integration and agricultural new quality productive forces. Using the entropy weight method, we quantify the development levels of these two dimensions. Our empirical analysis employs fixed effects models, mediation effect models, and spatial econometric approaches to investigate how digital–real integration influences agricultural new quality productive forces in major grain-producing regions. The research findings indicate the following: (1) Digital–real integration demonstrates a robust positive correlation with agricultural new quality productive forces in major grain-producing regions. (2) Both agricultural industrial structure upgrading and agricultural green total factor productivity serve as significant mediating channels through which digital–real integration enhances agricultural new quality productive forces. (3) The impact exhibits notable heterogeneity across three dimensions: regional characteristics, industrial structure levels, and fiscal decentralization levels. (4) Digital–real integration generates substantial positive spatial spillover effects on agricultural new quality productive forces, facilitating coordinated improvements in neighboring regions. (5) A significant threshold effect exists in how digital–real integration promotes agricultural new quality productive forces. Specifically, the promotional effect intensifies once innovation level and human capital level exceed certain critical thresholds. These findings offer both theoretical insights and practical guidance for advancing high-quality development in agriculture within major grain-producing regions while strengthening the national food security strategy.

1. Introduction

The current process of agricultural modernization is accelerating, and the new technological and industrial revolutions are mutually reinforcing. These developments place higher demands on high-quality agricultural laborers, labor materials, and labor objects. Traditional agricultural productivity can no longer meet the strategic requirements for building a strong agricultural nation [1]. Therefore, there is an urgent need to explore breakthrough paths based on the formation laws of new quality productive forces. The “digital–real integration” examined in this study refers to the process of deep integration between digital technologies and the real economy, while “new quality productive forces” refers to an advanced form of productive forces driven by scientific and technological innovation. Together, they constitute the theoretical framework and practical pathway for contemporary agricultural transformation and upgrading. The former serves as the means, while the latter represents the objective. According to general principles, human practices in transforming nature are the fundamental driving force for the formation of new quality productive forces, and changes in production relations are the key opportunity for qualitative transformation of productivity. Digital–Real integration, through cultivating new development momentum and reshaping new production relations, serves as a critical driver for the formation of new quality productive forces [2]. In response to the challenges of climate change, resource constraints, and food security, major economies worldwide regard digital technologies as the core driving force for agricultural transformation. The European Union’s “Digital Agriculture Strategy,” the United States’ “Precision Agriculture Initiative,” and Japan’s “Smart Agriculture Promotion Policy” all reflect the international consensus on empowering agricultural transformation through digital technologies. In the 2025 Central Document No. 1, the concept of “agricultural new quality productive forces” is introduced for the first time, and General Secretary Xi Jinping emphasized the importance of “promoting the deep integration of digital technology with the real economy to empower the transformation and upgrading of traditional industries”; this provides clear guidance for the Chinese-style agricultural modernization path. In the agricultural field, the deep integration of digital and real economies, with technology resource sharing as the link [3], achieves efficient circulation and sharing of new production factors such as technology, data, and knowledge within the industrial chain through digital infrastructure like big data platforms, the Internet of Things, and artificial intelligence. This greatly cultivates new development momentum for agriculture, enhances the skills of laborers, innovates labor materials, and expands labor objects, thus strengthening the material foundation for new quality productive forces. On the other hand, this deep integration breaks traditional agricultural organizational methods and production modes, reshapes agricultural production relations, and gives rise to new business models such as smart farms, cloud-based cooperatives, and direct connection models for order agriculture. These developments provide important institutional foundations and environmental conditions for the “qualitative” leap in agricultural productivity. Therefore, promoting deep integration of the digital and real economies is not only an inherent requirement following the evolution laws of new quality productive forces but also a strategic measure to break through the bottlenecks of agricultural development and accelerate the construction of a strong agricultural nation.

China’s main grain-producing regions, as the core strategic areas for ensuring national food security, are currently facing deep structural contradictions such as low development quality and efficiency, and delayed transformation and upgrading. Coordinating the scientific protection and sustainable development of agricultural resources in these major grain-producing areas is not only an urgent need to address current challenges but also an inevitable choice to obey the laws of modern agricultural evolution. Cultivating and strengthening agricultural new quality productive forces has become the core driving force for achieving leapfrog development in agriculture and rural economic and social progress. In the new wave of technological revolution, how to effectively integrate high-level human resources with cutting-edge digital technologies, enhance the comprehensive guarantee capacity of major grain-producing areas, optimize the allocation of production factors, and promote the transformation of these areas from traditional development models to modern, high-quality development paradigms, has become a major theoretical and practical issue that urgently requires in-depth research and resolution [4].

Currently, the relevant research progress on digital–real integration and new quality productive forces, both domestically and internationally, is as follows [5,6,7]. Digital–real integration is generally defined as “the penetration and application of digital technology in various aspects of the real economy to achieve their coordinated development [8,9,10,11,12],” specifically covering multiple dimensions such as technology integration, factor integration, product integration, and process integration [13]. Existing studies have analyzed its spatial–temporal evolution characteristics [14] and explored its impact on enterprises, industries, and economic development [15,16,17], among other aspects. Additionally, international scholars have systematically elucidated from the perspective of digital transformation how digital technologies reshape corporate business models and value creation processes, emphasizing that the deep integration of digital technologies with traditional business operations serves as a key source of a competitive advantage for enterprises [18]. Agricultural new quality productive forces are explained as “representing the leap of agricultural productivity driven by technological innovation [19], with its core being the innovative transformation of agricultural production factors, production processes, and the organization, division of labor, and cooperation along the industrial chain [20,21].” In terms of measurement, scholars have constructed evaluation indicator systems for agricultural new quality productive forces [22,23,24] and identified the empowering roles of digital technologies and green energy transformation in their development [25,26,27]. Regarding the relationship between digital–real integration and new quality productive forces, existing studies have begun preliminary exploration at the industry and enterprise levels. From the industry perspective, digital–real integration influences new quality productive forces by promoting the rationalization and upgrading of industrial structures [28]. Through deep integration of digital technologies with physical manufacturing, the agility, precision control, and market responsiveness of production processes are enhanced. This drives the systemic digital transformation and upgrading of traditional industries, promotes the optimization and collaborative integration of resource allocation, and provides important support for the cultivation and growth of new quality productive forces [29]. From the enterprise level, digital–real integration promotes the development of new quality productive forces by reducing production costs and improving management efficiency [30]. Enterprises use digital tools to achieve precise operations, agile production, and intelligent supply chains, effectively optimizing processes, controlling costs, and improving efficiency [29]. Related international research has found that digitalization strategies can enhance firm performance by optimizing resource allocation, improving operational efficiency, and promoting innovation, thereby providing theoretical support for understanding how digital–real integration drives productivity enhancement [31].

Although the existing literature has conducted preliminary exploration on the relationship between the two based on the 30 provinces (municipalities) of China [3,32,33], which lays a good theoretical foundation for this article, the following shortcomings still exist: (1) The research objects are relatively broad, lacking focused analysis of the major grain-producing regions; (2) many studies construct evaluation indicator systems for new quality productive forces from the traditional “three labor” (labor, labor materials, labor objects) [34,35,36,37] perspective, with limited innovation; (3) there remains a research gap in conducting in-depth exploration of the mechanisms through which digital–real integration affects new quality productive forces in agriculture, specifically focusing on the 13 major grain-producing provinces as research subjects. Based on this, this paper takes 13 major grain-producing areas as the research object, with the possible marginal contributions reflected in the following. (1) Focus and deepening of the research perspective: the study can more accurately reveal the practical effects and unique patterns of digital and real integration in core agricultural regions by precisely focusing on the 13 major grain-producing areas in China. (2) Innovation and reconstruction of the evaluation indicator system: this study breaks through the limitations of the traditional “three labor” perspective by constructing a comprehensive evaluation indicator system for agricultural new quality productive forces that aligns with the three main characteristics of “high-technology investment, high-efficiency performance, and high-quality development.” This enables a more scientific and comprehensive measurement of their actual development level. (3) In-depth analysis and expansion of action mechanisms: this study takes agricultural industrial structure upgrading and agricultural green total factor productivity as mediating variables to explore the transmission pathways between digital–real integration and new quality productive forces in agriculture. Furthermore, it conducts spatial effects analysis, heterogeneity analysis, and threshold effects analysis, rendering the mechanism research more systematic and comprehensive.

2. Study Site Overview

China’s major grain-producing regions are key agricultural production zones designated by the government to ensure food security, occupying an irreplaceable and central position in the national food security strategy. These regions encompass 13 provinces: Inner Mongolia, Liaoning, Jilin, Heilongjiang, Jiangsu, Anhui, Jiangxi, Shandong, Henan, Hebei, Hubei, Hunan, and Sichuan. Despite accounting for less than half of the nation’s cultivated land, these areas produce nearly 80% of the country’s grain output and supply over 90% of its marketable grain, demonstrating their exceptional production concentration and supply capacity. They play a decisive role in guaranteeing absolute staple food security, stabilizing grain supply, and maintaining market equilibrium, serving as the “ballast stone” of China’s food security system. Given the strategic position of major grain-producing regions within the national food security framework, investigating the mechanisms through which digital–real integration influences new quality productive forces in agriculture holds significant theoretical value and practical implications for exploring pathways toward agricultural modernization and enhancing comprehensive grain production capacity. Therefore, focusing on major grain-producing regions as research subjects is not only representative and typical but also provides critical policy insights and practical guidance for safeguarding national food security (Figure 1).

3. Theoretical Analysis and Research Hypotheses

3.1. Direct Impact of Digital–Real Integration on Agricultural New Quality Productive Forces

Digital–real integration serves as the core driving force for developing new quality productive forces in agriculture. This process extends beyond mere technological application; it represents a profound convergence of the digital economy and the agricultural real economy at the levels of production factors, operational processes, and organizational structures. Its impact manifests across three critical dimensions. First, from the high-tech advancement perspective, digital infrastructure development and agricultural digitalization provide innovative pathways for knowledge generation and dissemination in agriculture. Technologies, such as the Internet of Things (IoT) and remote sensing monitoring, establish comprehensive data perception networks spanning the entire agricultural production cycle, enabling technological innovation to shift from experience-based accumulation to precision-driven iteration grounded in real-time data analytics. This not only directly elevates agricultural technological capabilities but, more importantly, stimulates innovation capacity across diverse business entities by lowering research and development barriers and accelerating knowledge diffusion [38]. Second, from the high-quality development dimension, digital tools enable refined management of the relationship between agricultural production and resource environments. Precision farming technologies, including variable-rate fertilization and intelligent irrigation systems, dynamically adjust inputs based on crop requirements and environmental conditions, fundamentally reducing excessive application of fertilizers and pesticides and directly mitigating non-point source pollution. Simultaneously, comprehensive data traceability and monitoring systems provide credible management tools for green business models, such as ecological and circular agriculture, driving the transformation of agricultural production toward environmental sustainability [39]. Third, from the high-efficiency dimension, the digitalization of industries and industrial digitalization reshape value creation and distribution mechanisms in agriculture. Emerging business models, such as e-commerce platforms and smart logistics systems, shorten the farm-to-table distance and enhance circulation efficiency, while technologies, like big data and blockchain, strengthen trust and coordination across all links of the industrial chain, promoting deep integration of primary, secondary, and tertiary industries in rural areas. This systematic collaborative optimization transforms agricultural growth from dependence on yield increases in isolated segments to value appreciation through efficiency improvements across the entire industrial chain [40]. In summary, digital–real integration systematically establishes the foundation for developing new quality productive forces in agriculture by infusing data elements, optimizing production processes, and restructuring industrial organization. Based on this, Hypothesis 1 is proposed:

Hypothesis 1. 

Digital–real integration can effectively promote the development of agricultural new quality productive forces.

3.2. Indirect Impact of Digital–Real Integration on Agricultural New Quality Productive Forces

In the transmission mechanism through which digital–real integration influences new quality productive forces in agriculture, this study identifies agricultural industrial structure and agricultural green total factor productivity as key mediating variables. Theoretically, the formation of new quality productive forces is inseparable from “structural optimization” and “efficiency enhancement”: industrial structural upgrading embodies profound transformations in production organization methods and resource allocation patterns, while green total factor productivity encompasses elements such as technological progress, efficiency improvement, and environmental protection. Compared to single-factor inputs or traditional productivity indicators, these two variables offer a more comprehensive depiction of the pathways through which digital–real integration drives the development of agricultural new quality productive forces, examining the process from the perspectives of “industrial structural reshaping” and “factor efficiency improvement,” respectively.

3.2.1. Digital–Real Integration, Agricultural Industrial Structure Upgrading, and Agricultural New Quality Productive Forces

Agricultural industrial structure reflects the composition and proportional relationships among various industries within agriculture and serves as an important bridge linking digital–real integration with agricultural new quality productive forces. On the one hand, digital–real integration supports the optimization and upgrading of agricultural industrial structure through promoting agricultural digital transformation. The application of digital technologies not only significantly improves the precision and efficiency of agricultural production, extends the industrial chain, and enhances value [41,42], but also gives rise to new business models such as smart agriculture, digital agriculture, and agricultural product e-commerce. These innovations reshape the traditional agricultural industrial structure and provide new pathways for its adjustment and optimization [43]. On the other hand, the optimization of agricultural industrial structure further promotes the formation and development of agricultural new quality productive forces. Upgrading agricultural industrial structure creates higher demands for production factors, guiding the aggregation and flow of high-end innovative elements, such as capital, technology, data, and talent, into the agricultural sector. This effectively stimulates agricultural innovation vitality and injects new momentum into agricultural new quality productive forces. Meanwhile, industrial structure optimization represents not only the adjustment of industrial proportions but also reflects the extension of industrial chains, enhancement of supply chain resilience, and advancement along the value chain. The new business models arising from digital–real integration deeply integrate the primary, secondary, and tertiary industries, forming a complete industrial chain of “production + processing + marketing” [44]; this strengthens industrial resilience and risk resistance, ensures that the value of agricultural products is fully realized in the market, and ultimately drives agriculture towards a high-quality development model, thereby solidifying the foundation for the development of new quality productive forces. Based on this, Hypothesis 2 is proposed:

Hypothesis 2. 

Agricultural industrial structure plays a key mediating role between digital–real integration and agricultural new quality productive forces.

3.2.2. Digital–Real Integration, Agricultural Green Total Factor Productivity, and Agricultural New Quality Productive Forces

Agricultural green total factor productivity, as a key indicator measuring agricultural growth quality and sustainable development level, is the core link driving the coordinated development of digital–real integration and agricultural new quality productive forces. On the one hand, digital–real integration creates the foundational conditions for improving agricultural green total factor productivity by deeply integrating digital technologies with physical agriculture. The wide application of technologies, such as the Internet of Things, big data, and artificial intelligence, in agriculture enables the precise and intelligent allocation of production factors. Precision agriculture technologies optimize the use of water, fertilizer, and pesticides through real-time monitoring and data analysis, reducing resource consumption and environmental pollution while stabilizing crop output [45]. Smart agricultural machinery enhances operational efficiency, cuts energy consumption, and lowers carbon emissions. Agricultural big data platforms integrate multi-source information to support production decision-making, promoting the efficient and intensive use of resources. On the other hand, the improvement of agricultural green total factor productivity provides a solid foundation for the formation of agricultural new quality productive forces. New quality productive forces essentially require efficient factor allocation and a green, low-carbon development model. The continuous growth of green total factor productivity indicates that agriculture is gradually moving away from the traditional path of high input and high consumption, and shifting towards a modern, technology-driven, and environmentally friendly development mode [46]. At the same time, the technological innovation capabilities and management innovation experience accumulated in the process of improving green total factor productivity provide endogenous driving forces for the sustainable development of agricultural new quality productive forces. Based on this, Hypothesis 3 is proposed:

Hypothesis 3. 

Agricultural green total factor productivity plays a key mediating role between the integration of the digital and real economies and agricultural new quality productive forces.

3.3. Spatial Effect Analysis of Digital–Real Integration on Agricultural New Quality Productive Forces

The deep integration of digital technologies with the agricultural real economy directly enhances local agricultural production efficiency, resource utilization, and innovation capacity through applications, such as big data, IoT, and intelligent equipment, forming the core driving force behind new quality productive forces in agriculture. This impact extends beyond local boundaries, generating spillover effects through spatial linkages. On one hand, the construction of digital infrastructure and agricultural digitalization platforms exhibits network effects and connectivity functions, facilitating the cross-regional diffusion of knowledge, technologies, and management models, thereby stimulating imitative learning and technological upgrading in neighboring areas. On the other hand, agricultural industry chains transcend administrative boundaries during digital reorganization, forming cross-regional collaborative production, logistics, and marketing networks. Consequently, the effectiveness of digital–real integration in one region radiates to adjacent areas through supply chains, information flows, and service systems, manifesting spatial patterns of agglomeration or coordinated growth. These spatial effects are typically realized through mechanisms such as technology diffusion effects, factor reallocation, and regional demonstration effects [47]. Based on this, Hypothesis 4 is proposed:

Hypothesis 4. 

Digital–real integration has a spatial effect on agricultural new quality productive forces.

3.4. Threshold Effect Analysis Based on Innovation Level and Human Capital Level

Based on the theory of technological absorptive capacity and the theory of innovation ecosystems, innovation level, as a key representation of a region’s technological carrying capacity, has a decisive influence on the effective application of the integration of digital and real technologies. When the innovation level in a region is low, there is often a lack of well-established R&D infrastructure, high-end technical talent, and an innovative institutional environment, making it difficult to deeply understand and effectively utilize the complex technological systems involved in the integration of the digital and real economies. Such regions may only be in the stage of technological imitation and preliminary application, and the promotion of agricultural new quality productive forces is relatively limited. As innovation levels improve, once the innovation level crosses a certain threshold, the region will possess strong technological absorption and secondary innovation capabilities, allowing it to localize and integrate the digital and real technologies according to local agricultural resources and development needs. This deep integration of technology with actual agricultural production will significantly enhance its ability to promote agricultural new quality productive forces [48].

The integration of the digital and real economies, as a knowledge-intensive technological transformation, places higher demands on the digital literacy, learning abilities, and adaptability of workers. When the level of human capital is low, agricultural workers generally lack the necessary academic background and skills training. Faced with complex digital equipment and intelligent systems, they may have cognitive barriers and operational difficulties, which limit the actual effectiveness of the technology. When the level of human capital surpasses a certain threshold, high-quality agricultural laborers can not only master the operation of digital–real integration technologies proficiently but also identify problems in practice and propose improvements, achieving the optimal configuration of human–machine collaboration [49]. At the same time, a high level of human capital also facilitates the rapid dissemination and diffusion of knowledge, forming learning organizations and innovative teams, which further amplifies the promoting effect of the integration of the digital and real economies on agricultural new quality productive forces. Based on this, Hypothesis 5 is proposed:

Hypothesis 5. 

Digital–real integration has a threshold effect on agricultural new quality productive forces.

Accordingly, the theoretical framework and research hypotheses are shown in Figure 2.

4. Model Construction and Variable Selection

4.1. Model Construction

(1)

Benchmark Regression Model

Based on the aforementioned theory and hypotheses, this study takes the level of digital integration as the core explanatory variable and agricultural new quality productive forces as the explained variable. The following econometric model is constructed to examine the effect of digital integration on agricultural new quality productive forces.

$${\mathit{NPA}}_{\mathit{it}}={\alpha }_0+{\alpha }_1{\mathit{DRI}}_{\mathit{it}}+{\alpha }_2{\mathit{Controls}}_{it}+{\mu }_i+{\gamma }_t+{\epsilon }_{\mathit{it}}$$

(1)

In Equation (1), ${NPA}_{it}$ represents the level of agricultural new quality productive forces in province i during year $t$; ${DRI}_{it}$ represents the level of integration between digital and real economies for the corresponding year and province; ${\alpha }_0$ is the constant term; ${\alpha }_1$ represents the coefficient of the impact of the deep integration of digital and real economies on agricultural new quality productive forces; ${\alpha }_2$ represents the impact coefficients of the control variables; ${Controls}_{it}$ is the set of control variables; ${\mu }_i$ and ${\gamma }_{\mathrm{t}}$ represent the fixed effects of provinces and time, respectively; and ${\epsilon }_{it}$ is the random error term.

(2)

Mediation Effect Model

In Equation (2), to examine the mechanism through which digital–real integration impacts agricultural new quality productive forces, this study introduces a mediator variable and constructs the following mediation effect model based on the benchmark model (1):

$$M{E}_{it}={\beta }_0+{\beta }_1{\mathit{DRI}}_{it}+{\beta }_{2}Control{s}_{it}+{\mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(2)

In Equation (2), ${ME}_{it}$ represents the mediator variable in province i during year t, including agricultural industrial structure upgrading (PRO) and agricultural green total factor productivity (ATFP); β is the coefficient to be estimated; the meanings of the other variables are the same as those in Equation (1).

(3)

Spatial Effect Model

Considering potential regional spatial spillover effects, conventional panel models cannot adequately capture inter-regional spatial linkages and influences. Following diagnostic testing, this study constructs a fixed-effects spatial durbin model, as shown in Equation (3):

$$NP{A}_{it}={\alpha }_0+{\alpha }_{1}DR{I}_{it}+{\alpha }_{2}Contro{l}_{it}+W_{it}(\beta NP{A}_{it}+{\beta }_{1}DR{I}_{it}+{\beta }_{n}Contro{l}_{it})+{\mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(3)

In Equation (3), $\beta$, ${\beta }_1$, and ${\beta }_n$ represent spatial lag coefficients; W denotes the spatial weight matrix; and the remaining variables are defined consistently with Equation (1).

(4)

Threshold Effect Model

To test whether the impact of digital–real integration on the new quality productive forces of agriculture exhibits a nonlinear effect with changes in the threshold variable, this study establishes the following panel threshold model, as shown in Equation (4):

$$NP{A}_{it}={\phi }_0+{\phi }_{1}DR{I}_{it}\times I\left(T{h}_{it}\le \theta \right)+{\phi }_{2}DR{I}_{it}\times I\left(T{h}_{it}>\theta \right)+{\phi }_{3}Control{s}_{it}+{\mu }_i+{\gamma }_t+{\epsilon }_{it}$$

(4)

In Equation (4), ${\mathit{Th}}_{\mathit{it}}$ is the threshold variable; I(·) is the indicator function, which takes the value of 1 if the condition inside the brackets is met, and 0 otherwise; $\phi$ is the coefficient to be estimated; $\theta$ is the threshold value to be estimated; the meaning of the other variables remains unchanged.

4.2. Variable Selection

4.2.1. Core Explanatory Variable

The core explanatory variable is the level of digital–real integration (DRI). Currently, several policy documents, including the “13th Five-Year Plan for National Agricultural and Rural Informatization Development” and the “Digital Agriculture and Rural Development Plan (2019–2025),” have defined the core connotations of smart villages and digital villages. However, systematic research in the academic community on the measurement and assessment of the rural digital economy remains relatively limited. Digital infrastructure, by enabling the digital allocation of resource elements and the widespread coverage of networked services, provides a solid foundation for the rural digital economy. At the same time, the digital transformation of agriculture and the cultivation of digital industries together constitute the key pathways to promoting its high-quality development. This study follows the principles of objectivity, accessibility, comprehensiveness, and operability, selecting six tertiary indicators from three aspects: digital infrastructure, agricultural digitalization, and digital industrialization, to construct an indicator system for evaluating the level of the digital economy [50]. This article refers to the research by Gao He-peng on the construction of the real economy indicator system and comprehensively evaluates the agricultural economic development level in major grain-producing areas from three aspects: agricultural scale, agricultural modernization level, and agricultural development potential [14].

Drawing on the methodology of Gao Peipei et al. [51,52], this study first employs the entropy method to measure the comprehensive development levels of the digital economy and the real economy separately. Prior to applying the entropy method for calculating indicator weights, this study utilizes the Min–Max normalization method to standardize the raw data, ensuring all indicator values fall within the [0, 1] interval to satisfy the computational requirements of the entropy method. Subsequently, a coupling coordination degree model is employed to measure the degree of coordination in the coupled development of these two dimensions. The specific indicator selection is presented in

Table 1. Construction of the digital–real integration indicator system.

First-Level Indicators	Second-Level Indicators	Third-Level Indicators	Indicator Description	Attributes
Digital Economy	Digital Infrastructure	Rural Internet Penetration Rate (%)	Rural Broadband Access Users	+
		Agricultural Meteorological Observation Stations (Number)	Agricultural Meteorological Observation Services	+
	Agricultural Digitalization	Digital Trading of Agricultural Products (Hundred Million RMB)	E-commerce Sales of Agricultural Products	+
		Investment Intensity in Agricultural Production (%)	Fixed Asset Investment in Agriculture, Forestry, Animal Husbandry, and Fishery/Total Social Fixed Asset Investment	+
	Digital Industrialization	Rural Information Technology Application (Number of People)	Average Population Served per Postal Service Outlet	−
		Agricultural and Rural Entrepreneurship and Innovation Bases (Number)	Number of Taobao Villages	+
Real Economy	Agricultural Scale	Total Output Value of Agriculture, Forestry, Animal Husbandry, and Fishery (Hundred Million RMB)	Direct Access	+
		Total Import and Export Value of Agricultural Products (Hundred Million RMB)	Direct Access	+
	Agricultural Development Potential	Added Value of Agriculture, Forestry, Animal Husbandry, and Fishery (Hundred Million RMB)	Direct Access	+
	Level of Agricultural Modernization	Total Power of Agricultural Machinery (Ten Thousand kW)	Direct Access	+
		Irrigation Conditions of Farmland (Hectare-Meters, hm2)	Effective Irrigated Area	+

.

4.2.2. Explained Variable

The explained variable is the level of agricultural new quality productive forces (NPA). According to the law of the evolution of the forms of productivity in Marxist political economy, the generation of agricultural new quality productive forces is essentially a historical leap in agricultural production methods. Its inherent regularity is reflected in the dialectical unity among the structural improvement of labor productivity, the deep integration of technological innovation elements, and the maintenance of the sustainability of the ecological reproduction system. This highlights the revolutionary transformation of the modern agricultural development paradigm [53].

Starting from General Secretary Xi Jinping’s assertion that new-type productivity features “high technology, high quality, and high efficiency,” this paper constructs an indicator system to measure the level of new-type productivity. High technology reflects innovation and upgrading in agricultural production on the scientific, technological, and management levels; high quality reflects the improvement of agricultural product quality and industrial development, as well as the promotion of a green and sustainable agricultural development model; while high efficiency emphasizes the breaking down and integration of various links in agricultural production and the boundaries of industries, enhancing overall efficiency, including the comprehensive improvement of labor productivity, land productivity, and agricultural value-added rate.

Based on the existing literature [54], this paper, grounded in the basic connotation and characteristics of agricultural new quality productive forces, and adhering to the basic principles of scientific validity, rationality, and data availability, constructs a comprehensive evaluation system consisting of 3 first-level indicators, 7 second-level indicators, and 16 third-level indicators, as shown in

Table 2. Evaluation index system for agricultural new quality productive forces.

First-Level Indicators	Second-Level Indicators	Third-Level Indicators	Indicator Description	Attribute
High Technology Investment	Innovative Capability	Agricultural Science and Technology Practitioners (Persons)	R&D Personnel × (Total Agricultural, Forestry, Animal Husbandry, and Fishery Output/Regional GDP)	+
		Agricultural Science and Technology Activity Funds (Ten Thousand RMB)	R&D Activity Funds × (Total Agricultural, Forestry, Animal Husbandry, and Fishery Output/Regional GDP)	+
	Technological Level	Degree of Agricultural Mechanization (%)	Area of Mechanized Tillage/Arable Land Area	+
		Advanced Technology Support (Items)	Number of Patents Granted to Digital Agriculture Enterprises	+
High-Quality Development	Green Environmental Protection	Forest Coverage Rate (%)	Direct Acquisition	+
		Fiscal Environmental Expenditure (%)	Environmental Protection Fiscal Expenditure/Government Public Fiscal Expenditure	+
	Environmental Pollution	Agricultural COD Emission Intensity (Tons/Hundred Million RMB)	Agricultural COD Emissions/Total Agricultural, Forestry, Animal Husbandry, and Fishery Output	−
		Agricultural Ammonia Nitrogen Emission Intensity (Tons/Hundred Million RMB)	Agricultural Ammonia Nitrogen Emissions/Total Agricultural, Forestry, Animal Husbandry, and Fishery Output	−
	Green Management	Green Operations (Items)	Number of Green Agricultural Cooperatives	+
		Green Sales (Items)	Number of Certified Green Foods	+
High Efficiency Performance	Industry Integration	Primary and Secondary Industry Integration (%)	Output of Agricultural, Forestry, Animal Husbandry, and Fishery Processing Industry/Total Agricultural, Forestry, Animal Husbandry, and Fishery Output	+
		Primary and Tertiary Industry Integration (%)	Output of Agricultural, Forestry, Animal Husbandry, and Fishery Service Industry/Total Agricultural, Forestry, Animal Husbandry, and Fishery Output	+
		Agriculture-Tourism Integration (%)	Leisure Agriculture Revenue/Total Agricultural, Forestry, Animal Husbandry, and Fishery Output	+
	Production Efficiency	Labor Productivity (%)	Output of Primary Industry/Primary Industry Workforce	+
		Land Productivity (%)	Total Agricultural Output/Arable Land Area	+
		Agricultural Value-Added Rate (%)	Value Added of Agricultural, Forestry, Animal Husbandry, and Fishery Industry/Total Agricultural, Forestry, Animal Husbandry, and Fishery Output	+

. The entropy method is used to calculate the level of agricultural new quality productive forces.

4.2.3. Mediating Variable

Agricultural Industry Structure Upgrading (PRO) refers to the process in which the agricultural industry shifts from a dominance of the primary and secondary industries to a dominance of the tertiary industry. It reflects the extent to which agricultural production extends from the basic planting and breeding to higher-end links in the value chain, such as processing and services. Based on existing research [55], this paper uses the industrial structure hierarchy coefficient to measure the level of agricultural industry structure upgrading.

Agricultural Green Total Factor Productivity (ATFP) is an important indicator to measure the level of agricultural green development. It reflects the efficiency and sustainability of agricultural production. Based on previous studies [56], this paper adopts the SBM-GML index method to select three types of indicators—input, expected output, and non-expected output—for measurement. This calculation process is implemented through MATLAB R2022a programming. The GML index expression for the analytical model of factors influencing agricultural green total factor productivity is as follows:

$$GM{L}^{t,t+1}\left(x^t,y^t,b^t,x^{t+1},y^{t+1},b^{t+1}\right)=\left[1+D_G^T\left(x^t,y^t,b^t\right)\right]/\left[1+D_G^T\left(x^{t+1},y^{t+1},b^{t+1}\right)\right]$$

(5)

where ${GML}^{t,t+1}$ represents the global index; ($x^t$, $y^t$, $b^t$) denote input factors, expected output, and non-expected output in period t, respectively; ($x^{t+1}$, $y^{t+1}$, $b^{t+1}$) denote input factors, expected output, and non-expected output in period t + 1, respectively; $D_G^T(x^t, y^t, b^t)$ represents the directional distance function in period t; and $D_G^T(x^{t+1}, y^{t+1}, b^{t+1})$ represents the directional distance function in period t + 1.

4.2.4. Control Variable

To avoid potential endogeneity issues arising from omitted variables, this study includes a series of control variables related to in the regression model, in order to improve the validity and robustness of the estimates. The specific variables are as follows: economic development level (pgdp), represented by per capita GDP; government fiscal strength (gov), represented by the ratio of local government general budget expenditure to regional GDP; urbanization level (urb), represented by the ratio of urban population to total population; degree of openness (open), represented by the ratio of total import and export value to regional GDP; and agricultural disaster rate (ad), represented by the ratio of disaster-affected agricultural area to total sown area of crops.

4.3. Data Sources and Descriptive Statistics

This paper conducts an empirical study using panel data from 13 provincial-level food production regions from 2012 to 2023, with all data characterized by explicit public accessibility and availability. The data mainly come from the following sources: China Statistical Yearbook, China Rural Statistical Yearbook, China Energy Statistical Yearbook, China Meteorological Yearbook, China Environmental Statistical Yearbook, China Brand Agricultural Yearbook, China Financial Yearbook, China Fixed Asset Investment Statistical Yearbook, China Leisure Agriculture Statistical Yearbook, China Tertiary Industry Statistical Yearbook, China Green Food Bulletin, provincial statistical yearbooks, the Taobao Village Development Report from the Alibaba Research Institute, the Qiyan·Social Science Big Data Platform, and announcements on local economic and social development. During the process of integrating these multi-source data to construct a unified panel dataset, indicator definitions and statistical calibers vary across different statistical sources, necessitating careful comparison and standardization adjustments. For certain indicators with missing values in specific years, linear interpolation methods were employed to supplement the data gaps. Descriptive statistics of the variables are presented in

Table 3. Descriptive statistics of each variable.

Variable	Sample Size	Mean	Standard Deviation	Minimum	Maximum
NPA	156	0.234	0.115	0.089	0.619
DRI	156	0.446	0.152	0.176	0.935
PRO	156	1.306	0.076	1.061	1.386
ATFP	156	0.410	0.196	0.190	1.030
pgdp	156	1.024	0.293	0.671	1.929
gov	156	0.216	0.058	0.118	0.398
urb	156	0.593	0.074	0.420	0.750
open	156	0.179	0.118	0.056	0.644
ad	156	0.132	0.107	0.001	0.590

.

It can be seen that the average level of agricultural new quality productive forces is 0.234, indicating that the overall level is not high; the average level of digital–real integration is 0.446, but the maximum value is 0.935 and the minimum value is 0.176, showing significant regional differences. The maximum value of agricultural green total factor productivity is 1.030, and the minimum value is 0.190, with obvious regional disparities. Therefore, it is necessary to fully consider the heterogeneity situation.

4.4. Multicollinearity Test

To avoid multicollinearity, variance inflation factor (VIF) tests are conducted on the explanatory variables and control variables. The results show that all VIF values are less than 10, indicating that multicollinearity does not exist (

Table 4. Collinearity test.

Variable	VIF	1/VIF
DRI	2.100	0.476
pgdp	6.650	0.150
gov	3.660	0.273
urb	2.710	0.369
open	3.870	0.259
ad	1.380	0.726
Mean VIF	3.390

).

5. Results

5.1. Baseline Regression Results

Based on the Hausman test, this study ultimately selects the fixed effects model.

Table 5. Benchmark regression results.

Variable	(1)	(2)	(3)
DRI	0.504 ***	0.253 ***	0.573 ***
	(0.046)	(0.060)	(0.068)
pgdp		−0.168 ***	0.104 **
		(0.055)	(0.041)
gov		−1.086 ***	−0.186
		(0.207)	(0.124)
urb		0.547 ***	0.141
		(0.140)	(0.366)
open		0.159	0.299 ***
		(0.105)	(0.109)
ad		−0.097	0.033
		(0.069)	(0.031)
_cons	0.009	0.188 ***	−0.229
	(0.021)	(0.067)	(0.182)
N	156.000	156.000	156.000
Controls	YES	YES	YES
ID	NO	NO	YES
YEAR	NO	NO	YES
R2	0.438	0.534	0.936

Note: **, and *** represent significance at the 5%, and 1% levels, respectively; the numbers in parentheses are standard errors.

reports the baseline regression results with both individual and time fixed effects. Column (1) reports the regression results without controlling for fixed effects or introducing control variables; column (2) reports the regression results after introducing various control variables based on column (1); column (3) reports the regression results after introducing control variables and controlling for time and province fixed effects. The results show that the coefficient estimates of digital–real integration are all positive and pass the 1% significance level test, indicating that digital–real integration in major grain-producing areas can significantly promote the development of agricultural new quality productive forces, thus verifying Hypothesis 1.

Regarding control variables, several patterns emerge from the analysis. The economic development level (pgdp) shows a coefficient of 0.104, significant at the 5% level. This positive relationship suggests that higher per capita GDP facilitates agricultural new quality productive forces through multiple channels: enhanced fiscal capacity, improved market mechanisms, and increased technology spillover. These factors collectively support agricultural infrastructure construction and advanced technology adoption. Government fiscal strength (gov) presents a coefficient of −0.186 without statistical significance. This result indicates that expanding the proportion of local government general budget expenditure does not automatically advance agricultural new quality productive forces. Rather, what matters is optimizing expenditure allocation, enhancing fund utilization efficiency, and targeting investments toward digital technology applications. The urbanization level (urb) coefficient stands at 0.141 but lacks statistical significance. After controlling for inter-regional differences and temporal trends, the urban-to-total population ratio shows limited direct influence on agricultural new quality productive forces. Urbanization appears to matter more through cross-regional structural variations than through time-series dynamics. Degree of openness (open) exhibits a coefficient of 0.299, significant at the 1% level. This strong positive effect demonstrates that international engagement substantially promotes agricultural new quality productive forces. The mechanisms include expanded agricultural export markets, access to international advanced technologies and management practices, and deeper integration into global agricultural value chains. Agricultural disaster rate (ad) shows a coefficient of 0.033 without reaching significance. This finding suggests that improvements in disaster prevention infrastructure, agricultural insurance coverage, and disaster-resistant technologies have effectively buffered the negative impacts of natural disasters on agricultural production. Consequently, disaster occurrences exert minimal influence on agricultural new quality productive forces.

5.2. Robustness and Endogeneity Tests

5.2.1. Robustness Test

(1)

Shortening the Sample Period

The year 2012 coincided with the pilot launch of digital–real integration policies and the transitional period of agricultural statistical caliber adjustments. The data for that year may be subject to atypical fluctuations or lack comparability due to factors such as the instability of policy implementation, the initial stage of technological application, or changes in statistical indicators. Additionally, technologies such as the agricultural Internet of Things and big data platforms only began to achieve large-scale application starting in 2013. Therefore, to reduce the interference of the policy adaptation period, statistical inconsistencies, or early-stage technical noise on the results, data from this year were excluded to ensure that the remaining sample period (2013–2023) is more consistent in terms of policy continuity, technological maturity, and data comparability. As shown in column (1) of

Table 6. Robustness and endogeneity tests.

Variable	(1)	(2)	(3)	(4)
	NPA	NPA	Replacing NPA	NPA
DRI	0.587 ***	0.562 ***	0.445 ***	0.711 ***
	(0.073)	(0.067)	(0.060)	(0.070)
_cons	−0.160	−0.248	−0.223	−0.021
	(0.207)	(0.178)	(0.159)	(0.181)
N	143.000	156.000	156.000	143.000
Controls	YES	YES	YES	YES
ID	YES	YES	YES	YES
YEAR	YES	YES	YES	YES
R2	0.951	0.950	0.956	0.951

Note: *** represent significance at the 1% levels; the numbers in parentheses are standard errors.

, after adjusting the sample period, the coefficient of digital–real integration remains significantly positive, indicating that it still has a stable promotional effect on agricultural new quality productive forces, further verifying the reliability of the research conclusion.

(2)

Trimming Outliers

To eliminate the interference of outliers on the estimation results, this study conducted a two-tailed trimming process at the upper and lower 1% for all continuous variables except for the core explanatory variable in the baseline regression and then re-ran the regression test. The results, as shown in column (2) of Table 6, demonstrate that after excluding the outliers, the coefficient of digital–real integration remains significantly positive, indicating that the conclusion is robust.

(3)

Replacing the Explained Variable

To ensure the robustness of the conclusions, this study constructed an indicator system for the level of agricultural new quality productive forces from the perspective of “three labor forces” and replaced the original explained variable by calculating it using the entropy method. The results are shown in column (3) of Table 6. The coefficient of digital–real integration remains significantly positive, indicating that the conclusion is still robust after replacing the explained variable.

5.2.2. Endogeneity Test

To mitigate potential endogeneity bias (such as reverse causality and omitted variables), this study adopts the instrumental variable method for estimation. The one-period lagged level of digital–real integration is selected as the instrumental variable. This choice is based on two core justifications: First, relevance: the one-period lagged value is highly correlated with the current period’s digital–real integration level in time series, satisfying the basic requirement for instrumental variables. Second, exogeneity: the lagged variable temporally precedes current-period agricultural new quality productive forces, effectively severing the causal loop whereby current-period productive forces reversely influence current-period integration levels. More importantly, from a theoretical logic perspective, the digital–real integration level in the past period primarily affects the current-period level through its continuity and path dependence effects and is unlikely to directly influence current-period agricultural new quality productive forces through other omitted channels besides the current-period integration level. This satisfies the basic logic of the exclusion restriction. Accordingly, we employ two-stage least squares (2SLS) for regression analysis. According to the results shown in Column (4) of Table 6, the correlation coefficient is 0.711, indicating no weak instrument problem. These results demonstrate that after controlling for endogeneity bias, digital–real integration maintains a robust positive driving effect on agricultural new quality productive forces, thereby further supporting Hypothesis 1.

5.3. Test of the Mechanism

Given that the third step in the traditional three-step mediation effect test may suffer from significant endogeneity problems, this study draws on the methodology proposed by Jiang Ting [57]. Specifically, we empirically examine only the relationships between digital–real integration, agricultural industrial structure upgrading, and agricultural green total factor productivity. A two-way fixed effects model controlling for both time and individual dimensions is employed. The empirical results are presented in

Table 7. Regression results of mechanism analysis.

Variable	(1)	(2)	(3)
	NPA	PRO	ATFP
DRI	0.573 ***	0.243 ***	0.904 ***
	(0.068)	(0.064)	(0.158)
_cons	−0.229	0.779 ***	−2.128 ***
	(0.182)	(0.135)	(0.336)
N	156.000	156.000	156.000
Controls	YES	YES	YES
ID	YES	YES	YES
YEAR	YES	YES	YES
R2	0.936	0.897	0.906

Note: *** represent significance at the 1% levels; the numbers in parentheses are standard errors.

.

5.3.1. Mediating Effect Test of Agricultural Industrial Structure Upgrading

As shown in column (2) of Table 7, the estimated coefficient is 0.243 and statistically significant at the 1% level. This result suggests that digital–real integration exerts a substantial positive influence on agricultural industrial structure upgrading. The underlying mechanisms operate through multiple pathways. First, technological advances accelerate the shift from traditional to modern agricultural practices, driving the agricultural value chain toward higher value-added segments. This process facilitates deeper integration across primary, secondary, and tertiary industries while fostering the emergence of new business models [58]. Second, digital tools enable more efficient allocation of production factors, including land, labor, and capital, thereby improving overall resource utilization. Several supporting conditions reinforce this transformation process. Digital infrastructure provides essential foundational support. Consumption upgrading stimulates demand for higher-quality agricultural products. Expanded digital channels enhance value realization efficiency. Meanwhile, continuous improvements in the policy environment create favorable conditions for systematic upgrading of the agricultural industrial structure. These combined forces establish a solid foundation for developing agricultural new quality productive forces. Based on this analysis, agricultural industrial structure upgrading is confirmed as a significant mediating channel linking digital–real integration to agricultural new quality productive forces. Hypothesis 2 is, therefore, supported.

5.3.2. Mediating Effect Test of Agricultural Green Total Factor Productivity

Column (3) of Table 7 reports an estimated coefficient of 0.904, which is positive and statistically significant at the 1% level. This finding indicates that digital–real integration promotes agricultural new quality productive forces by enhancing agricultural green total factor productivity. Digital technologies improve production efficiency while reducing resource waste. They also facilitate the adoption of green technologies such as precision agriculture and smart farming systems. Through data-driven optimization, these technologies achieve simultaneous emission reduction and efficiency gains, thereby strengthening sustainable development capabilities and stimulating agricultural technological innovation. The significance of this mechanism can be attributed to several factors. In precision agriculture, technologies such as GPS positioning, remote sensing, and big data analytics have substantially improved input accuracy and resource utilization efficiency. Smart monitoring systems enable real-time tracking of environmental variables including soil conditions and climate patterns, thereby optimizing production decisions. Additionally, increasingly stringent environmental regulations have accelerated the green transformation of agricultural production methods. Growing societal awareness of sustainable agriculture has further reinforced this trend. Digital technologies also contribute by optimizing agricultural production processes and enabling efficient utilization of agricultural waste [59]. Collectively, these factors drive the development of agricultural new quality productive forces. Based on this analysis, agricultural green total factor productivity is confirmed as a significant mediating channel linking digital–real integration to agricultural new quality productive forces. Hypothesis 3 is, therefore, supported.

5.4. Spatial Effects Analysis

Before measuring the spatial spillover effects, a global Moran’s I test was conducted on agricultural new quality productive forces in major grain-producing regions. This study employs a Spatial Contiguity Matrix for the analysis, with results presented in

Table 8. Global Moran’s I index of agricultural new quality productive forces.

Year	Global Moran’s I	Z	p-Value
2012	0.368	2.865	0.002
2013	0.514	3.528	0.000
2014	0.488	3.421	0.000
2015	0.525	3.544	0.000
2016	0.629	3.905	0.000
2017	0.640	3.857	0.000
2018	0.663	3.977	0.000
2019	0.631	3.838	0.000
2020	0.618	3.725	0.000
2021	0.634	3.809	0.000
2022	0.529	3.275	0.001
2023	0.791	4.590	0.000

. The findings reveal that the global Moran’s I indices for major grain-producing regions during 2012–2023 are all significantly positive, indicating substantial spatial positive correlation among regional agricultural new quality productive forces. Following sequential LM tests, Wald tests, and Hausman tests, the Spatial Durbin Model (SDM) with dual fixed effects was identified as the optimal specification.

Table 9. Estimation results of the Spatial Durbin Model (SDM).

Variable	Spatial Contiguity Matrix
	Direct Regression Coefficients	Direct Effects	Indirect Effects	Total Effects
DRI	0.383 ***	0.436 ***	0.545 ***	0.982 ***
	(6.23)	(7.76)	(3.10)	(5.58)
W × DRI	0.259 **
	(2.02)
ρ	0.382 *** (4.27), p < 0.001
N	156.000
Controls	YES
ID	YES
YEAR	YES
R2	0.159

Note: **, and *** represent significance at the 5%, and 1% levels, respectively; the numbers in parentheses are standard errors.

presents the spatial econometric results. The direct regression coefficient of digital–real integration equals 0.383, significant at the 1% level. When decomposing the spatial effects, we observe direct effects of 0.436, indirect effects of 0.545, and total effects of 0.982, all statistically significant. These findings reveal a dual mechanism: digital–real integration strengthens local agricultural new quality productive forces while simultaneously generating substantial positive spatial spillover effects on neighboring regions. The spatial lag coefficient for digital–real integration stands at 0.259, significant at the 5% level. This coefficient provides additional evidence of spatial dependence in how digital–real integration operates across regions. Based on these results, Hypothesis 4 (H4) receives empirical support. The underlying mechanism stems from the inherent characteristics of digital technology integration with the real economy. Network externalities and technology diffusion properties enable digital–real integration to transcend geographical boundaries when promoting agricultural new quality productive forces. These effects propagate spatially through three primary channels: factor mobility across regions, knowledge spillover among production units, and coordination along industrial chains. Together, these transmission mechanisms amplify the impact beyond individual localities.

5.5. Further Analysis

5.5.1. Heterogeneity Analysis

(1)

Regional Heterogeneity Analysis

Substantial differences exist across Chinese regions in terms of resource endowments, policy environments, and stages of economic development. Such imbalances generate noticeable regional heterogeneity in the development levels of both digital–real integration and agricultural new quality productive forces. This study divides the major grain-producing regions into sub-samples based on the north-south geographical boundary (the Qinling Mountains-Huaihe River Line) and conducts separate regression analyses. The group regression results presented in columns (1) and (2) of

Table 10. Heterogeneity analysis.

Variable	Regional Group Regression		Industrial Structure Upgrading Level Group Regression		Government Fiscal Decentralization Level Group Regression
	(1)	(2)	(3)	(4)	(5)	(6)
	Southern Region	Northern Region	Low Level	High Level	Low Level	High Level
DRI	0.254 **	0.663 ***	−0.029	0.599 ***	0.433	0.648 ***
	(0.106)	(0.087)	(0.107)	(0.107)	(0.271)	(0.089)
_cons	0.677 *	−0.038	−0.488 ***	−0.699 **	−0.905 ***	0.350
	(0.359)	(0.177)	(0.167)	(0.262)	(0.307)	(0.214)
N	72.000	84.000	78.000	78.000	78.000	78.000
Controls	YES	YES	YES	YES	YES	YES
ID	YES	YES	YES	YES	YES	YES
YEAR	YES	YES	YES	YES	YES	YES
R2	0.982	0.930	0.953	0.970	0.954	0.970

Note: *, **, and *** represent significance at the 10%, 5%, and 1% levels, respectively; the numbers in parentheses are standard errors.

reveal that digital–real integration exerts a significantly positive effect in both southern and northern regions. However, the magnitude of this effect differs considerably between the two areas. The estimated coefficient for the northern region is 0.663, substantially higher than the 0.254 observed in the southern region. Both coefficients are statistically significant at the 1% or 5% level, respectively. These findings indicate that digital–real integration generates a stronger promotional effect on agricultural new quality productive forces in northern areas. Several factors may account for this north–south disparity. The Northern plains feature contiguous farmland and a high degree of large-scale cultivation, conditions that are well-suited for the standardized deployment of large agricultural machinery and digital equipment. In contrast, the southern regions are characterized by complex hilly terrain and fragmented farmland distribution. These geographical constraints make it difficult to achieve economies of scale in agricultural production, even when advanced digital infrastructure is available. Furthermore, northern regions primarily cultivate bulk grain crops such as wheat and corn, whose production processes are relatively standardized and therefore more amenable to digital management. Southern regions, by comparison, grow a wider variety of agricultural products with more complex and variable production models, which reduces the adaptability of digital technologies.

(2)

Industrial Structure Level Heterogeneity Analysis

Recognizing significant variations in agricultural industrial structure within the major grain-producing regions, this study draws on the existing literature to divide the full sample into two sub-groups based on the level of agricultural industrial structure: a high-level group and a low-level group. Specifically, regions with values above the median are classified into the high-level group, while those below the median are assigned to the low-level group. The level of agricultural industrial structure is measured by the proportion of the combined output value of secondary and tertiary industries relative to GDP. The group regression results in Table 10 demonstrate significant heterogeneity in the effect of digital–real integration on agricultural new quality productive forces across regions with different levels of agricultural industrial structure. As shown in columns (3) and (4), the coefficient of digital–real integration in regions with high-level agricultural industrial structure is 0.599, statistically significant at the 1% level. This indicates that a one-unit increase in digital–real integration is associated with approximately 0.599 units of improvement in agricultural new quality productive forces. In contrast, the coefficient for regions with low-level agricultural industrial structure is −0.029 and fails to achieve statistical significance. This substantial disparity may be attributed to several underlying factors. Regions with higher-level agricultural industrial structure typically possess a more comprehensive foundation for digital technology application, more efficient resource integration capabilities, and stronger technological absorption capacities. These advantages enable digital–real integration to be more effectively converted into productivity gains. Conversely, regions with lower-level agricultural industrial structure may face constraints such as shorter industrial chains, inadequate technological adaptability, or insufficient supporting policies. These limitations make it difficult to realize the marginal benefits of digital–real integration.

(3)

Government Fiscal Decentralization Level Heterogeneity Analysis

Given the significant differences in fiscal decentralization levels across governments in the major grain-producing regions, this study further divides these regions into high and low sub-samples. Regions with fiscal decentralization levels above the median are categorized as the high-level group, while those below the median constitute the low-level group. Fiscal decentralization is measured by the ratio of local fiscal revenue to local fiscal expenditure. The group regression results in Table 10 reveal significant heterogeneity in the impact of digital–real integration on agricultural new quality productive forces across regions with different levels of government fiscal decentralization. As presented in columns (5) and (6), the coefficient of digital–real integration in regions with high fiscal decentralization is 0.648, which is highly significant at the 1% level. This result highlights the technology-enabling effect in these areas. In contrast, the coefficient in regions with low fiscal decentralization is 0.433, which does not achieve statistical significance. This difference may stem from several factors. Regions with high fiscal decentralization generally possess stronger fiscal autonomy and greater resource allocation efficiency. Their digital infrastructure tends to be relatively well-developed, and their application of digital technologies is more mature and stable. Consequently, the promotional effect is more pronounced, although the marginal effect may be relatively smaller. In contrast, regions with low fiscal decentralization rely more heavily on central government finances. Limited funding leads to inadequate supporting infrastructure and immature technology applications. As a result, the stability and significance of the effects are weaker in these areas.

5.5.2. Threshold Effect Analysis

Provincial variations in innovation level and human capital level across the major grain-producing regions may lead to differential impacts of digital–real integration on agricultural new quality productive forces. To identify critical points in the influencing mechanism, this study introduces innovation level (PAT) and human capital level (HR) as threshold variables. This approach enables a more precise analysis of the empowering effect of digital–real integration on agricultural new quality productive forces, thereby providing more targeted theoretical and empirical support for policy-making under different development conditions. Innovation level is measured by the natural logarithm of the number of domestic invention patents granted. Human capital level is measured by the proportion of university students relative to the total population.

Before conducting threshold regression, it is necessary to first test whether the threshold effect is significant. This study uses the bootstrap method, with 500 resamples, to estimate and validate the existence and specific values of the threshold. According to the statistical results in

Table 11. Threshold regression test.

Threshold Variable	Number of Thresholds	F Statistic	p-Value	10% Critical Value	5% Critical Value	1% Critical Value	Threshold Value	Confidence Interval
Innovation Level	Single Threshold	30.10	0.002	14.034	16.291	30.10	11.1950	[10.4617~11.2175]
Human Capital Level	Single Threshold	21.10	0.016	13.227	14.990	22.070	0.0229	[0.0226~0.0229]

, the regression results for innovation level and human capital level as threshold variables indicate a single threshold effect, with threshold values of 11.195 and 0.0229, respectively. However, neither passes the double-threshold or triple-threshold tests. Meanwhile, Figure 3a,b shows the threshold values and confidence intervals corresponding to innovation level and human capital level. In Figure 3b, the confidence interval for the single threshold of human capital level shows multiple intersections, mainly due to the interaction between data distribution characteristics and statistical methods: as a proportion variable, the actual data for human capital level may experience local clustering or discrete jumps near the threshold value of 0.0229 (e.g., multiple provinces’ HR values are densely distributed in the 0.02–0.03 range), causing the LR function to fluctuate non-monotonically during computation. Additionally, the limited sample size of panel data and discrete values cause the LR statistic to cross the critical value line repeatedly, creating multiple intersections. Furthermore, the random variation in the bootstrap sampling (500 times) may amplify data noise, especially when the HR distribution is multimodal or skewed, further fragmenting the confidence interval boundaries. This phenomenon does not negate the significance of the single threshold but indicates that the sensitivity and uncertainty of the threshold effect near the critical point are higher. Therefore, research Hypothesis 5 is validated.

Table 12. Regression results of the threshold effect model.

Variable	Model 1	Model 2
PAT ≤ 11.1950	0.397 **
	(0.151)
PAT > 11.1950	0.499 ***
	(0.122)
HR ≤ 0.0229		0.497 ***
		(0.107)
HR > 0.0229		0.562 ***
		(0.098)
_cons	−0.388	−0.303
	(0.296)	(0.267)
N	156.000	156.000
Controls	YES	YES
ID	YES	YES
YEAR	YES	YES
R2	0.890	0.884

Note: **, and *** represent significance at the 5%, and 1% levels, respectively; the numbers in parentheses are standard errors.

presents the results of the threshold effect analysis, examining how innovation level and human capital level influence both the pathway and intensity of the impact of digital–real integration on agricultural new quality productive forces in the major grain-producing regions. Model 1 employs innovation level as the threshold variable. The results indicate that when innovation level falls below or equals the threshold value (Th₁ ≤ 11.1950), the effect of digital–real integration on agricultural new quality productive forces is 0.397, which is statistically significant at the 5% level. When the innovation level exceeds this threshold (Th₁ > 11.1950), the effect increases substantially to 0.499 and achieves significance at the 1% level. These findings suggest that higher innovation levels amplify the promotional effect of digital–real integration on agricultural new quality productive forces. Regions with elevated innovation levels possess stronger capacities for technology absorption and application. This enhanced capacity enables more effective utilization of digital–real integration technologies, thereby generating greater improvements in agricultural new quality productive forces. Model 2 employs human capital level as the threshold variable. The results show that when human capital level falls below or equals the threshold value (Th₂ ≤ 0.0229), the effect of digital–real integration on agricultural new quality productive forces is 0.497, which is statistically significant at the 1% level. When human capital level exceeds this threshold (Th₂ > 0.0229), the effect rises to 0.562. These findings demonstrate that improvements in human capital level can significantly strengthen the promotional effect of digital–real integration on agricultural new quality productive forces. This result underscores the critical role of highly skilled workers in driving the development of agricultural new quality productive forces.

6. Conclusions and Recommendations

This study develops a comprehensive evaluation index system for digital–real integration and agricultural new quality productive forces. Drawing on panel data from 13 major grain-producing regions in China spanning 2012 to 2023, we investigate the mechanisms through which digital–real integration influences agricultural new quality productive forces. The main findings are as follows:

• Digital–real integration in major grain-producing regions significantly promotes agricultural new quality productive forces. Specifically, a one-unit increase in digital–real integration corresponds to an average improvement of 0.573 units in agricultural new quality productive forces. This finding remains robust across multiple sensitivity tests, including endogeneity corrections, shortened sample periods, winsorization, and alternative dependent variable specifications.
• Both agricultural industrial structure upgrading and agricultural green total factor productivity serve as significant transmission channels through which digital–real integration enhances agricultural new quality productive forces. These two pathways constitute the core mediation effect linking digital–real integration to productivity gains.
• The relationship between digital–real integration and agricultural new quality productive forces displays notable heterogeneity across three dimensions. Regional heterogeneity follows a “stronger in the north, weaker in the south” pattern. Industrial structure heterogeneity exhibits a “high-level reinforcement, low-level inhibition” characteristic, reflecting threshold effects. Fiscal decentralization heterogeneity demonstrates a “high decentralization strengthens, low decentralization weakens” tendency.
• The promotional effect of digital–real integration on agricultural new quality productive forces generates significant positive spatial spillover effects, facilitating coordinated productivity improvements in neighboring regions.
• Digital–real integration in major grain-producing regions exhibits a notable threshold effect on agricultural new quality productive forces. When innovation level and human capital level exceed specific thresholds, their promotional impact on developing agricultural new quality productive forces intensifies significantly.

Based on the above conclusions, the following recommendations are proposed:

• Implement a differentiated digital infrastructure construction strategy to narrow the development gap between the northern and southern regions in terms of the integration of digital and real elements. Given the regional heterogeneity of “weaker in the south, stronger in the north,” for northern grain-producing regions, more investment should be made in new digital infrastructure such as 5G networks, the Internet of Things (IoT), and cloud computing. This should focus on bridging the digital technology gap and developing more adaptable digital agricultural technologies, such as cold-resistant smart agricultural machinery and drought warning systems, to ensure that digital technology aligns with local agricultural production needs. At the same time, establish a digital technology collaboration mechanism between the North and South, encouraging digital technology enterprises from the developed southern regions to expand into the northern regions. Through technology transfer and talent mobility, regional digital technology development can be balanced.
• Create a layered and categorized industrial structure upgrading guidance system to release the potential of the integration of digital and real elements in regions at different industrial levels. Based on the industrial structure heterogeneity of “high level reinforces, low level fails,” for regions with a high level of industrial structure, efforts should focus on driving their ascent to the high-end of the value chain, developing technology-intensive industries such as smart agriculture, precision agriculture, and digital agriculture, and strengthening the productivity transformation effect of digital–real integration. For regions with a low level of industrial structure, priority should be given to improving agricultural infrastructure, enhancing mechanization, standardization, and scale, solidifying the industrial foundation for the integration of digital and real elements, and gradually improving the efficiency of transforming digital–real integration into productivity, avoiding low-level regions from falling into development traps.
• Optimize the fiscal decentralization system to improve resource allocation efficiency and enhance local support for the integration of digital and real elements. In light of the heterogeneity of fiscal decentralization, which shows the characteristic of “high decentralization strengthens, low decentralization weakens,” it is necessary to establish and improve the division of fiscal responsibilities and expenditure between central and local governments, appropriately increasing local governments’ fiscal autonomy in the development of digital–real integration. At the same time, the transfer payment system should be improved, increasing support for regions with lower fiscal decentralization. Establish a special fund for the development of digital–real integration, using fiscal subsidies, tax incentives, and government procurement to guide and leverage social capital to form a diversified investment pattern.
• Strengthen the intermediary guiding role of agricultural industrial structure upgrading and the improvement of green total factor productivity. In terms of industrial structure upgrading, efforts should be made to extend agriculture to the high-end of the value chain, developing new industries such as deep processing of agricultural products and agricultural services. Digital technologies should be used to achieve the digital transformation of the industrial chain. In terms of improving green total factor productivity, green production technologies should be promoted, an agricultural production environmental monitoring system should be established, and digital technologies should be used for precise fertilization, water-saving irrigation, and other green production methods. At the same time, an evaluation and assessment system for industrial structure upgrading and green development should be established and incorporated into local government performance evaluations to ensure the effective functioning of intermediary roles.
• Improve regional coordination and benefit compensation mechanisms to effectively transform spatial spillover effects into collaborative development momentum. In response to the significant spatial dependence of digital–real integration impacts, administrative boundaries should be broken down to promote the establishment of a normalized digital agriculture collaborative development alliance between major grain-producing regions and surrounding areas. Through systematically promoting the beneficial flow of knowledge, technology, and production factors via means such as jointly building regional agricultural data-sharing platforms, collaboratively conducting smart agriculture technology demonstration and promotion, and creating cross-regional digital agricultural industry chains, positive spillover effects can be fostered. Simultaneously, explore the establishment of horizontal ecological compensation or benefit-sharing mechanisms based on spillover effects, providing policy preferences or financial compensation to core regions that generate significant positive spillovers, thereby incentivizing their enthusiasm to play a radiating and driving role. This approach internalizes spatial externalities and achieves coordinated enhancement of regional agricultural new quality productive forces as a whole.
• Build a gradient development mechanism to break through the constraints of development thresholds. In terms of improving innovation capabilities, agricultural R&D investment should be increased, high-level agricultural research institutes and innovation platforms should be established, and efforts should be made to overcome key agricultural technologies and accelerate the transformation of research outcomes. A sound incentive mechanism for agricultural technological innovation should be established, intellectual property protection systems should be improved, and a good innovation ecosystem should be created. In terms of enhancing human capital, a multi-level and multi-form farmer education and training system should be established, cultivating a new type of professional farmer. At the same time, talent introduction and training mechanisms should be improved. By offering attractive benefits, entrepreneurial support, and other measures, more high-quality talent can be attracted, providing strong intellectual support and talent assurance for breaking through threshold limits and fully realizing the effects of the integration of digital and real elements.

7. Discussion

This study makes three important contributions to the existing literature on how digital–real integration influences agricultural new quality productive forces. First, it narrows the research focus from the national scale to major grain-producing regions spanning 13 provinces. This approach uncovers distinctive operational patterns of digital–real integration within core agricultural zones, addressing a notable gap in geographically targeted research. The findings offer a more precise theoretical framework for understanding development drivers in areas critical to national food security. Second, the study moves beyond conventional “three labors” frameworks by developing a comprehensive evaluation system for agricultural new quality productive forces. This system incorporates three defining characteristics: high-technology investment, high-efficiency performance, and high-quality development. The innovative indicators enhance both measurement rigor and contemporary relevance, offering valuable tools and theoretical perspectives for future investigations. Third, the research systematically examines how agricultural industrial structure upgrading and agricultural green total factor productivity serve as mediating pathways. It further explores spatial spillover effects, heterogeneity patterns, and nonlinear threshold characteristics. These analyses transform our understanding from simple correlations to a sophisticated framework that captures multi-path transmission and contextual moderation. This advancement significantly deepens theoretical comprehension in the field.

Despite these contributions, several limitations point toward future research directions. Regarding research scale, the empirical analysis relies primarily on provincial-level panel data. Future studies could examine municipal or county levels to capture finer geographical variations and better understand intra-regional differences. Concerning methodology, the current indicator system draws mainly from publicly available statistics and has not fully integrated micro-level behaviors and perceptions. Subsequent research could incorporate field surveys and in-depth interviews to gather primary data. Such approaches would refine measurement methods and strengthen the real-world applicability of findings. Moving forward, researchers should expand data sources and diversify methodological approaches. These efforts will strengthen the empirical foundation of this field and advance theoretical refinement.