The Central Plains Urban Agglomeration is pivotal for accelerating the development of the central region. Understanding the diffusion of agricultural green water use technologies, particularly through urban scale borrowing, is essential for ensuring agricultural and ecological security in the region. This study examines 30 cities within the agglomeration from 2003 to 2023, utilizing a spatial econometric model to explore the impact of urban scale borrowing on agricultural green water use efficiency and spatial spillover effects. The results indicate a significant improvement in overall efficiency, with core cities showing enhanced spatial spillover effects, though regional disparities in efficiency have widened. Low-efficiency areas have demonstrated convergence, driven by policy and factor aggregation. From the ‘local-neighbor’ perspective, urban-scale borrowing notably enhances agricultural green water use efficiency and accelerates its spatial diffusion. The borrowing of urban functions and population scale positively impacts both local and neighboring cities’ efficiency, while the effect of borrowing regional innovation resources remains minimal. Robustness checks, including model adjustments and resampling, validate these findings. Additionally, the spatial spillover effect of urban scale borrowing decreases with increasing geographic distance, becoming negligible beyond 210 km. This study provides key insights and policy recommendations for advancing agricultural green development in urban agglomerations.

  • Urban agglomeration development: Urban-scale borrowing impacts agricultural green water use efficiency in both local and neighboring areas.

  • Spatial spillover effects: Urban-scale borrowing boosts efficiency in surrounding areas through spatial spillover effects.

  • Positive impact of borrowing: Urban-scale borrowing improves efficiency locally and in neighboring areas.

  • Multidimensional borrowing effect: Urban function, innovation, and population borrowing all positively affect efficiency.

  • Spillover effect decay boundary: The spillover effect of urban-scale borrowing shows a clear spatial decay with distance.

As the conflict between economic growth and environmental protection intensifies, regional spatial structures have been fundamentally reshaped. Core cities and urban agglomerations have emerged as crucial spatial entities for enhancing resource efficiency. The Central Plains Urban Agglomeration acts as both an ecological protective barrier and a new driver of economic development, playing a vital role in optimizing resource allocation efficiency. This optimization not only revitalizes the agricultural economy in the central region but also actively supports the sustainable development strategy, positioning it as the strategic core for advancing both agricultural economic growth and environmental sustainability. The ‘Central Plains Urban Agglomeration Development Plan Outline’ (2005) explicitly emphasizes establishing an urban ecological resource network centered on Zhengzhou, with Kaifeng as a key functional area. The geographical advantages of the Beijing-Guangzhou Railway and the Longhai–Lanxin Railway narrow the resource allocation gap between regions and expedite the development of green ecological demonstration zones. The ‘few agglomerations, many dispersals’ spatial pattern in the central and western regions highlights the importance of developing a network of small and medium-sized cities, which is essential for analyzing economic agglomeration and its spillover effects. Moderate urban scale expansion has significant effects on economic growth, technological innovation, and population movement, facilitating improved agricultural water use efficiency and promoting the optimization and sharing of agricultural green production water resources. However, rapid urbanization and the expansion of city sizes in the Central Plains Urban Agglomeration have caused complex ecological impacts and a multi-layered crisis concerning agricultural water resources. This expansion threatens the continuity and integrity of ecosystems, reduces biodiversity, alters land use patterns, disrupts natural groundwater replenishment and regional water cycle balance, and exacerbates pressures on agricultural green water use efficiency. This highlights the imbalance between water resource environmental constraints and urban regional development (Li et al. 2024).

Improving the green utilization efficiency of water is a crucial theoretical and practical goal for ensuring high-quality industrial development and regional sustainability, given its role as a primary natural production material and strategic resource. According to the China Water Resources Bulletin Ministry of Water Resources of the People's Republic of China (2022), agricultural water use constitutes 63.04% of the nation's total water resources, with approximately 90% of this allocated for irrigation. However, in light of this immense water demand, the current effective irrigation rate is only 0.56, significantly lower than the 0.75 rate in developed countries. To address this issue, the ‘13th Five-Year Plan’ for water conservancy reform and development proposes advancing reforms in areas such as green development, sustainable practices, and ecological priorities. Modern agricultural development urgently requires technological innovation to create new momentum for high-quality green water use efficiency, moving decisively toward an efficient, environmentally friendly, and resource-conserving agricultural production path. High-quality agricultural green development should focus not only on sustainable water resource management but also on economic efficiency, development costs in water management, and equitable distribution of evaluation methods and outcomes in decision-making (Ioris et al. 2008). Initial studies on water resource efficiency evaluation, both domestically and internationally, have primarily utilized methods such as ratio analysis (Li et al. 2008), stochastic frontier production function (Geng et al. 2014; Shah et al. 2023), and data envelopment analysis (DEA) (Sun et al. 2018; Shah et al. 2024). Because the traditional radial DEA model ignores the relaxation of the input–output variable (Charnes et al. 1978), Tone & Tsutsui (2010) introduced the super-efficiency slack-based measure (SBM) to accurately measure multiple decision-making units (DMUs). Several scholars have since employed SBM and its extended models to analyze regional ecological efficiency, focusing on green water use efficiency in the Yangtze River Economic Belt (Shang & Li 2023), the Beijing-Tianjin-Hebei region (Zhao et al. 2017), and the Yellow River Basin (Wei et al. 2021), as well as the multi-factor impact mechanisms. Additionally, exploring the spatiotemporal dynamics and differential impacts of regional green water use efficiency holds considerable theoretical and practical significance. From a spatial perspective, scholars have uncovered regional disparities in agricultural green water use efficiency in China using dynamic distribution maps and have analyzed spatial interaction effects from various angles (Yang et al. 2017; Ma et al. 2018; Shah et al. 2023). Studies have examined influencing factors at both micro and macro levels (Geng et al. 2014; Li et al. 2022). Micro-level factors include household characteristics, farmland scale, irrigation management, and technical training. Macro-level factors encompass technological progress, water conservancy facilities, and plant structures. There is currently a lack of academic research on the relationship between urban-scale borrowing and agricultural green water use efficiency. Additionally, there is no consensus on how urban-scale borrowing contributes to improving regional agricultural greenwater use efficiency. Therefore, this article will focus on the following questions: Can urban-scale borrowing enhance agricultural green water use efficiency? What are the underlying mechanisms?

Grounded in the concept of green high-quality development, the urban agglomeration economic scale borrowing system exhibits complex adaptive characteristics influenced by ongoing disturbances to resources and the environment throughout its evolution. William Alonso (1971) demonstrated that small cities can access the advanced functions of large cities by integrating into a collaborative spatial interaction network within urban agglomeration. Meijers et al. (2016) expanded on the concept of ‘scale borrowing’ through empirical research, emphasizing that this phenomenon primarily occurs through interactions within multi-dimensional urban network spaces. Small cities can leverage this mechanism to enhance their economic productivity and ecological resource allocation efficiency by sharing advanced technologies and policy experiences within the collaborative network of large cities. Simultaneously, large cities can improve their overall production capabilities by engaging in international or national urban network systems, utilizing high-quality ecological resources from global cities (Melo et al. 2009). The implementation of the urban scale borrowing mechanism is shifting urban development models toward low-carbon and energy-efficient practices. This shift strengthens the research and application of sustainable environmental technologies in cities and stimulates reforms in green innovative production dynamics (Zhang et al. 2022). Most scholars suggest that optimizing interactive connections enhances resource allocation and creates a ‘selection effect,’ which encourages resources to shift from low-efficiency-utilizing industries to high-efficiency-utilizing ones. This compels high-energy-consuming industries to reform their production methods, accelerating the diffusion of green energy utilization efficiency (Li & Gao 2016; Hao & Li 2020).

This paper makes marginal contributions compared to existing studies in the following aspects: First, most existing studies assess regional agricultural water use efficiency using a single factor or only expected outputs, often overlooking non-expected outputs, such as agricultural externalities like diffuse pollution. This oversight can introduce biases in efficiency evaluations. Second, studies on scale borrowing often analyze the contribution of spatial spillover effects from dimensions such as transportation modes, urban population scale, economic activity density, and functional borrowing. However, there is limited discussion on the ecological environment and agricultural governance. Third, many scholars explore spatial division effects based on inter-provincial data but fail to fully address the differences in urban cluster effects within provinces. This paper aims to expand on existing literature in the following ways: (1) Applying the SBM–DEA method to assess agricultural green water use efficiency in the Central Plains Urban Agglomeration, incorporating climate input factors to evaluate efficiency, and systematically exploring the impact of climate conditions on efficiency. This enriches the evaluation framework by considering the combined effects of natural and socio-economic factors. The rural social development index, calculated using the value method by integrating natural, social, ecological, and economic factors, is included as part of the expected output; (2) Innovatively exploring the inherent ‘green’ attributes of the urban scale effect and its impact mechanism on agricultural green water use efficiency to identify breakthroughs in the field; (3) Selecting cities as spatial analysis units and using the SDM model to analyze the spatial correlation between city scale borrowing and agricultural green water use efficiency. This provides a new theoretical and empirical perspective on the complex relationship between city-scale borrowing and high-efficiency agricultural green water development; (4) Innovatively analyzing the impact of urban-scale borrowing on agricultural green water use efficiency from three dimensions – urban functional borrowing, urban population scale borrowing, and urban innovation resource borrowing – while introducing a new dimension: urban economic activity density borrowing, in robustness checks. This offers a more detailed framework for analyzing regional spatial spillover effects.

In recent years, China has significantly advanced structural reforms in the agricultural supply side, prioritizing the greening and efficiency of the agricultural, forestry, animal husbandry, and fishery sectors. This effort is based on shifts in market demand and involves integrating green economic development, implementing the ‘Two Mountains' theory, and enhancing agricultural green water use efficiency. China's green development strategy seeks to transition urban development models from extensive ‘industrial civilization' to intensive, sustainable ‘ecological civilization.’ This transition relies on low-carbon, green, and circular production methods, prioritizing economic quality while considering the ecological environmental carrying capacity. Therefore, examining how urban-scale borrowing within urban agglomerations influences agricultural green water use efficiency in both local and neighboring areas – especially in the context of dynamically adjusting urban development models with spatial expansion – is both practically significant and academically valuable. The specific mechanism is illustrated in Figure 1.
Fig. 1

Analytical framework of urban scale borrowing and agricultural green water use efficiency.

Fig. 1

Analytical framework of urban scale borrowing and agricultural green water use efficiency.

Close modal

Urban scale borrowing and agricultural green water use efficiency in the local area

Integrating urban and rural scale systems will drive energy structure transformation, improve resource utilization efficiency, and promote green, high-quality regional development (Zhang & Zhong 2023). On one hand, urban-scale borrowing fosters industrial agglomeration, resulting in the gradual concentration of capital, technology, equipment, and resources, which enhances agricultural resource utilization efficiency. The positive externalities of industrial agglomeration contribute to the formation of an interconnected ‘cooperative flow’ between cities (Helena Chiu & Lee 2012), establishing a collaborative production network that accelerates agricultural specialization spillovers, facilitates knowledge sharing, and promotes interactive learning. Conversely, effective deployment of green technology can enhance competitive advantages for enterprises within a specific urban scale through isolation mechanisms and technological spillovers (Yuan & Chen 2019). This approach effectively mitigates environmental pollution and accelerates the industrial structural transformation of resource-based cities, creating a win–win situation for both the economy and ecology. Currently, some scholars qualitatively examine the impact of urban-scale borrowing on high-quality agricultural development, asserting that it promotes agricultural green development. Scholars have investigated how urban-scale borrowing enhances agricultural green water use efficiency through institutional and industrial pathways. They conclude that urban-scale borrowing empowers urban human capital, upgrades industrial structures, integrates innovative resources, allocates agricultural, forestry, water, and fishery resources, and plans urbanization, thus promoting agricultural green development (Zhan & Li 2022). Based on this, the following hypothesis is proposed:

Hypothesis 1: Urban-scale borrowing can promote the diffusion of agricultural green water use efficiency in the local area.

Urban scale borrowing and agricultural green water use efficiency in surrounding areas

Urban-scale borrowing enables urban networks to exert spatial spillover effects (Huang et al. 2020), promoting the growth of urban-rural green resource efficiency through agglomerated environmental externalities (Hu & Fan 2020). Furthermore, urban agglomeration accelerates the expansion of urban scales. During this expansion, a dense communication network enhances urban agglomeration. Specifically, small and medium-sized cities can gain advantages such as human capital, scientific technology, and production factors by accessing technology from large urban areas (Wang 2010). They also benefit from the advanced functional services of these large cities, enhancing their agricultural green water use efficiency. Human capital serves as the foundation for agricultural green technology innovation and industrial structure optimization (Drucker & Feser 2012), significantly supporting the research, development, and application of agricultural green technologies. This enhances agricultural production factors and resource utilization efficiency. Innovative factors and technical talents cluster in specific areas, resulting in the emergence of more green technologies in surrounding cities. Simultaneously, influenced by the demonstration effects of innovative factors, surrounding cities can imitate local emerging technologies through cooperative exchanges. This leads to regional collaboration and the exploration of new ‘green’ growth opportunities arising from urban-scale borrowing demonstration mechanisms. Conversely, large cities can mitigate pollution concentration in agglomerated economic activities by releasing redundant functions, thereby enhancing agricultural green water use efficiency. Ecological and environmental pollution issues arising from their expansion can alert both large cities and surrounding areas, prompting them to focus on developing ecological industries. Additionally, the influence of spatial demonstration effects allows surrounding cities to explore the dynamic relationships between total factor productivity, sustainable development, and resilience, seeking improvement measures that directly impact agriculture. Based on this, the following hypothesis is proposed:

Hypothesis 2: Urban scale borrowing generates positive spatial spillover effects on the green benefits of agricultural water use.

The spatial spillover effect of urban scale borrowing on agricultural green water use efficiency exhibits a diminishing boundary

The improvement of green technological innovation efficiency during urban scale expansion results not only from the accumulation of scale effects but also from collaborative breakthrough mechanisms driven by cross-regional knowledge exchange and technological accumulation. Knowledge flow and technology transfer play a crucial role in enhancing agricultural green water use efficiency, especially under the synergistic effect of high-quality human capital and high-end functional borrowing mechanisms, which significantly amplify efficiency (Caves et al. 1982). However, this process is constrained by spatial boundaries and diffusion mechanisms. Knowledge transmission can be divided into explicit and tacit knowledge. Explicit knowledge can cross geographical barriers through literature, data, and institutional channels, while tacit knowledge relies heavily on face-to-face communication and the transfer of practical experience. This results in the ‘local,’ ‘high-density,’ and ‘spatially contact-dependent’ nature of cutting-edge green innovation knowledge. As spatial distance increases, the transmission efficiency of tacit knowledge declines sharply, and the marginal cost of knowledge exchange rises, significantly limiting the spatial spread of technology spillovers and causing a noticeable attenuation effect. Although advancements in transportation systems and information technologies somewhat alleviate this spatial constraint, innovation knowledge related to green water use efficiency still faces significant geographical diffusion resistance, further exacerbating the imbalance in regional green innovation capabilities. In this framework, the impact of urban-scale borrowing on agricultural green water use efficiency exhibits significant temporal and spatial heterogeneity. Core cities, with their scale advantages, resource aggregation capabilities, and strong innovation spillover mechanisms, effectively drive the improvement of green water use efficiency in surrounding regions, particularly with the support of high-end functional borrowing mechanisms. Through high-end service platforms, advanced technological facilities, and talent hubs, they continuously export innovation resources and support services to neighboring cities, playing a key role in enhancing regional agricultural greenwater use efficiency. This spillover effect, however, does not expand infinitely and has a clear spatial threshold. As geographical distance increases, the effectiveness of information and technology diffusion decreases, and after crossing spatial boundaries, the efficiency of knowledge transmission drops sharply. This results in a weakened capacity of peripheral regions to absorb innovation resources, exhibiting a ‘center-periphery’ gradient diffusion pattern. The expansion of population scale and aggregation of high-quality human capital are also key components of the urban scale borrowing effect. Core cities, by aggregating high-level human resources, enhance technological innovation and knowledge transfer efficiency, but the positive spillover effect diminishes with increasing spatial distance. The improvement of agricultural green water use efficiency between regions is not merely a simple scale spillover process but a complex dynamic system influenced by resource-carrying capacity, technological diffusion paths, and knowledge flow barriers. A deeper understanding of this process requires exploration from multi-dimensional perspectives, including spatial correlation, factor mobility, and institutional embedding.

Hypothesis 3: The spatial spillover effect of urban scale borrowing on agricultural green water use efficiency exhibits a diminishing boundary.

Non-radial frontiers efficient distance data envelopment model

Traditional DEA, first introduced by Charnes et al. (1978), is a systematic analytical method that evaluates relative efficiency based on various input and output factors. Because agricultural water use is vulnerable to external environmental disturbances, many decision units with efficiency values of 1 tend to overlook the issue of variable slack. Consequently, Tone & Tsutsui (2010) proposed the non-radial SBM–DEA model to address slackness in inputs and outputs. Moreover, the technical reference framework for production commons incorporates inputs and outputs from multiple periods to define the production frontier. In this context, n denotes the number of decision units, which is 30 in this study, represented as Kt (t = 1, 2,…, n). Each decision unit comprises N input vectors, M expected output vectors, and I non-expected output vectors. The following formula represents the non-expected output of the kth decision unit in the SBM–DEA. The efficiency of agricultural green water use ranges from [0,2], where denote the slackness of inputs and outputs, while represent the input–output values for the kth region at time t:
(1)

Spatial econometric model

Common spatial econometric models (Yuan & Gao 2020) include the spatial autoregressive model (SAR), which accounts for spatial lag effects where independent variables influence other areas through spatial transmission mechanisms; the spatial error model (SEM), which considers spillover effects from random shocks based on error terms; and the spatial durbin model (SDM) (LeSage & Pace 2009) (Equation (6)), which includes both dependent and independent variable spatial lags and analyzes the spatial spillover effects generated by these mechanisms. Due to the difficulty in distinguishing between endogenous and exogenous effects, the optimal choice is to exclude the influence of interference terms, thus selecting the SDM. Compared to tests using cross-sectional data, the chosen spatial panel model accounts for both temporal and spatial feature changes, accurately capturing the spatial spillover effects of agricultural green water use efficiency. Given the advantages of the SDM, this study preliminarily adopts it as the foundation for empirical analysis and assesses its spatial applicability to determine whether it is the optimal model:
(2)
The SDM simultaneously accounts for spatial lags between the dependent variables of both local and neighboring areas (spatial reference compact). It departs from the traditional regression approach that merely reflects the relationship between the two, instead referring to Elhorst (2010) and utilizing partial differential methods to decompose the spillover effects of the SDM from the perspectives of direct effects (ADI), indirect effects (AII), and total effects (ATI). A key distinction is that the difference between regression coefficients and decomposition coefficients lies in whether the spatial feedback mechanism is considered. The specific calculation where ATI denotes Total Effects, ADI denotes Direct Effects, and AII denotes Indirect Effects.
(3)

Dynamic panel model

Improving agricultural green water use efficiency requires an adjustment period, and as such, current efficiency is often influenced by the previous period's efficiency. Therefore, this paper includes the lagged agricultural green water use efficiency from the previous period in the model to control its inherent impact. Furthermore, urban-scale borrowing may influence next year's agricultural green water use efficiency through technological innovation, industrial upgrading, and policy guidance. A dynamic panel regression model is constructed for robustness testing, as follows:
(4)

The explained variable, efficiency, represents agricultural green water use efficiency, while the core explanatory variable, borrow, is a comprehensive index of urban scale borrowing derived from principal component analysis of high-level urban function borrowing, innovation resource borrowing, and population scale borrowing. Efficiencyit−1 represents the first-order lag term of agricultural green water use efficiency, and borrowit−1 is the first-order lag term of urban scale borrowing. Control variables (X) include water resource endowment, economic development level, fiscal agricultural support policies, per capita arable land area, and the climate risk physical index. The model also controls for city and time-fixed effects, with ε representing the random error term.

Variable selection and indicator description

Explained variable

Drawing on the research approach of Cui et al. (2020) and the ‘14th Five-Year Plan for National Agricultural Green Development,’ water source, land, mechanical power, and labor are selected as key agricultural input factors (Cui et al. 2020). Additionally, environmental factors not only directly influence agricultural water demand and growth conditions but also indirectly affect farmers' planting behavior and decision-making psychology. Therefore, this paper incorporates environmental factors into the input variables to more accurately assess agricultural green water use efficiency and capture their ecological and social impacts. The rural comprehensive social development index, derived from the entropy values of agricultural ‘economic-social-ecological’ dimension indicators, serves as the standard for measuring expected output (see Table 1). Currently, agricultural water resource pollution mainly results from non-point source pollution due to irrigation. Following Liu & Cao (2023), the gray water footprint is selected to measure the non-expected output variable. The specific calculation method is outlined in Equation (9), where WFagr-grey denotes the agricultural gray water footprint, L represents the number of agricultural fertilizers and other pollutants discharged, and the maximum concentration of pollutants, Cmax is determined by the nitrogen fertilizer standard limit value stipulated in the ‘Surface Water Environmental Quality Standards Basic Project Standard Limits’ for Class III water. The initial concentration of nitrogen in water bodies is denoted as Cnat, the nitrogen fertilizer leaching rate is denoted as a, and the total amount of applied nitrogen fertilizer is denoted as APPL:
(5)
Table 1

Descriptive statistical analysis explanation.

VariablesObsMeanStd. Dev.MinMaxIndicator description
Efficiency 630 .342 .259 0.03 1.919 The composition of the indicator is shown in Table 2  
Borrow1 630 .947 −1.833 5.587 Derived from principal component analysis of bden, bnet, and bfun 
Borrow2 630 1.003 −1.405 5.378 Derived from principal component analysis of bpop, bnet, and bfun 
Bpop 630 20.11 5.42 10.03 40.90 Resident population scale weighted by Euclidean distance 
Bden 630 0.44 0.35 0.01 1.76 Economic activity density weighted by Euclidean distance 
Bnet 630 21.64 24.55 0.48 128.88 Innovation resource scale weighted by Euclidean distance 
Bfun 630 0.43 0.14 0.21 1.40 High-level function scale weighted by Euclidean distance 
CPRI 630 24.862 7.231 0.816 47.56 Derived from extreme climate risk factors weighted 
Economic 630 10.181 .777 7.848 11.985 The logarithm of per capita GDP 
Plant 590 2.748 1.664 0.427 13.064 Total rural sowing area/total population 
Resource 630 0.656 0.884 0.06 10.251 The logarithm of per capita agricultural water resources 
Finance 630 0.111 .034 0.008 0.184 Agricultural and forestry expenditure/total fiscal expenditure 
VariablesObsMeanStd. Dev.MinMaxIndicator description
Efficiency 630 .342 .259 0.03 1.919 The composition of the indicator is shown in Table 2  
Borrow1 630 .947 −1.833 5.587 Derived from principal component analysis of bden, bnet, and bfun 
Borrow2 630 1.003 −1.405 5.378 Derived from principal component analysis of bpop, bnet, and bfun 
Bpop 630 20.11 5.42 10.03 40.90 Resident population scale weighted by Euclidean distance 
Bden 630 0.44 0.35 0.01 1.76 Economic activity density weighted by Euclidean distance 
Bnet 630 21.64 24.55 0.48 128.88 Innovation resource scale weighted by Euclidean distance 
Bfun 630 0.43 0.14 0.21 1.40 High-level function scale weighted by Euclidean distance 
CPRI 630 24.862 7.231 0.816 47.56 Derived from extreme climate risk factors weighted 
Economic 630 10.181 .777 7.848 11.985 The logarithm of per capita GDP 
Plant 590 2.748 1.664 0.427 13.064 Total rural sowing area/total population 
Resource 630 0.656 0.884 0.06 10.251 The logarithm of per capita agricultural water resources 
Finance 630 0.111 .034 0.008 0.184 Agricultural and forestry expenditure/total fiscal expenditure 
Table 2

Agricultural green water use efficiency.

DimensionIndicatorIndicator descriptionRural social development index composition
Input Water resource input Agricultural water footprint Economic Rural per capita net income Yuan 
Land input Total agricultural sown area Rural per capita income growth rate 
Power input Total agricultural mechanical power Rural per capita living expense Yuan 
Labor input Primary industry employment Social Urban and rural residents' disposable income ratio 
Natural conditions Precipitation Per capita electricity consumption kWh·person−1 
Natural conditions Temperature Population natural growth rate 
Output Expected output Total agricultural output value Water usage per unit of agricultural output value m3·thousand yuan−1 
Rural social development index (see right) Ecology Effective irrigation rate 
Non-expected output Grey water footprint Average afforestation area Hectares/10,000 people 
DimensionIndicatorIndicator descriptionRural social development index composition
Input Water resource input Agricultural water footprint Economic Rural per capita net income Yuan 
Land input Total agricultural sown area Rural per capita income growth rate 
Power input Total agricultural mechanical power Rural per capita living expense Yuan 
Labor input Primary industry employment Social Urban and rural residents' disposable income ratio 
Natural conditions Precipitation Per capita electricity consumption kWh·person−1 
Natural conditions Temperature Population natural growth rate 
Output Expected output Total agricultural output value Water usage per unit of agricultural output value m3·thousand yuan−1 
Rural social development index (see right) Ecology Effective irrigation rate 
Non-expected output Grey water footprint Average afforestation area Hectares/10,000 people 

Core explanatory variable

Meijers and burger classified scale (size) in borrowed-size into functional scale (function) and economic activity scale (performance). In contrast, Gu & Le (2023) focus on the flow of innovative factors and knowledge spillovers between cities, defining innovation resource borrowing variables (bneti) to analyze the impact of urban scale borrowing – including population, function, and innovation factor flows – on agricultural green water use efficiency. Building on existing studies, this paper innovatively divides scale borrowing into four dimensions: high-level function borrowing (bfun), population-scale borrowing (bpop), economic activity borrowing (bden), and innovation resource borrowing (bnet). Using the method of Camagni et al. (2016), the paper calculates sub-indicators of scale borrowing using a city spatial weighting distance matrix (including geographical and time distances) and applies principal component analysis to derive the overall borrowing scale (borrow):
(6)
(7)
(8)
where wij represents the geographical distance between city i and city j in year t (measured by Euclidean distance), bfuni represents urban high-level function borrowing, measured by the proportion of the labor population engaged in high-end industries (e.g., transportation, postal services, IT services, financial services, real estate, leasing, and scientific research); bpopi represents population scale borrowing, measured by the logarithm of city population size; bneti represents innovation resource borrowing, measured by the number of green patents authorized at the prefecture-level; bdeni represents economic activity density borrowing, measured by the daily density of urban public transport volume per built-up area, considering the endogeneity between employment density and labor productivity (Su & Wei 2013; Su et al. 2014; Zhao et al. 2021).

Control variables

The following control variables are selected to account for the impact of uncontrollable factors on the empirical results, based on existing research (Ma et al. 2018; Du et al. 2021; Yu et al. 2022). These include water resource endowment, economic development level, fiscal agricultural support policies, and per capita arable land area. Among these, per capita agricultural water resources are represented by the logarithm of per capita agricultural water resources (resource); the logarithm of per capita gross domestic product (GDP) in urban areas represents economic development level (economic); the ratio of agricultural and forestry expenditure to total fiscal expenditure represents fiscal agricultural support policy (finance); and per capita arable land area is calculated by dividing the agricultural sowing area by the rural population (plant). With the advancement of agricultural modernization, the long-standing strategy of increasing income and production has led to widespread resource consumption and environmental pollution. Therefore, this paper includes the climate risk physical index (CPRI), calculated by weighting extreme climate factors, as a control variable to account for the impact of extreme external environmental factors on agricultural production efficiency.

Data sources and descriptive statistics

This study utilizes balanced panel data from 30 prefecture-level cities in the Central Plains urban agglomeration to examine the relationship between digital rural construction and the green allocation efficiency of agricultural water resources. The data are sourced from the ‘China Statistical Yearbook,’ the ‘China Urban Yearbook,’ the ‘China Rural Statistical Yearbook,’ and the statistical yearbooks of various provinces and municipalities. Considering data availability and completeness, this study analyzes a cross-sectional period of 21 years from 2003 to 2023. Table 1 presents the descriptive statistical analysis of the variables. The results indicate that the maximum and minimum values of agricultural green water efficiency are 1.919 and 0.03, respectively. Similarly, the urban function borrowing index has maximum and minimum values of 5.378 and −1.405, respectively. When conducting baseline regression, it is essential to control for both time and individual effects to mitigate the impact of spatial and temporal differences caused by confounding variables. The correlation coefficients among the variables range from −0.0014 to 0.7090, and all variant inflation factor (VIF) values are below 10, indicating that multicollinearity is not an issue.

Analysis of agricultural green water use efficiency results

Using the super-efficiency SBM–DEA method to measure agricultural green water use efficiency, Table 3 presents the annual average efficiency of prefecture-level cities in the five provinces of the Central Plains Urban Agglomeration. The results of Figure 2 shows that prefecture-level cities in Henan, Hebei, and Shandong consistently ranked higher than the regional average in the Central Plains Urban Agglomeration. From 2003 to 2023, the cities in Henan had the highest water use efficiency (0.38), followed by those in Shanxi (0.36) and Hebei (0.34), with Shandong and Anhui tied for second place (0.23). This is due to Henan being the core region of the Central Plains Urban Agglomeration, where advanced green technology innovations are more easily introduced and exchanged. This leads to the coordinated advancement of agricultural ecological resources and the environment, demonstrating a ‘dual-driven’ growth model that boosts efficiency in both the core and neighboring areas. (1) From a temporal perspective, from 2003 to 2018, the average efficiency of the Central Plains Urban Agglomeration remained below 0.4, fluctuating in a ‘zigzag’ pattern. From 2019 to 2023, water use efficiency in some provinces increased sharply. This growth can be attributed to the introduction of the ‘National Water Saving Action Plan’ in 2019, which significantly improved agricultural green water use efficiency across the region. However, this growth was uneven, and regional disparities widened. Specifically, provinces like Henan, Hebei, and Shanxi achieved significant improvements in water use efficiency, while provinces like Shandong and Anhui lagged behind due to technological delays and uneven resource allocation. (2) In terms of inter-provincial differences, the agricultural green water use efficiency in Henan cities rose from 0.53 in 2019 to 0.89 in 2022, with an average annual increase of 17%. In 2023, the efficiency value dropped significantly but remained the highest among the five provinces. Cities in Hebei showed the greatest fluctuation, with a sharp increase from 2018 to 2023, averaging 21% annually, becoming a key driver of rapid growth in agricultural water use efficiency in the Central Plains Urban Agglomeration. Henan, as China's ‘Central Granary,’ continues to promote the development philosophy of ‘storing grain in the land, storing grain in technology,’ with targeted irrigation and water-saving technology significantly improving water use efficiency.
Table 3

Average agricultural green water use efficiency measurement in the central plains urban agglomeration.

Area20032004200520062007200820092010201120122013
Henan 0.27 0.29 0.25 0.25 0.27 0.25 0.26 0.26 0.24 0.26 0.28 
Hebei 0.25 0.23 0.33 0.34 0.37 0.37 0.36 0.24 0.27 0.17 0.19 
Shanxi 0.28 0.29 0.21 0.22 0.24 0.26 0.25 0.25 0.22 0.21 0.16 
Shandong 0.10 0.09 0.14 0.13 0.10 0.06 0.36 0.37 0.36 0.35 0.33 
Anhui 0.27 0.27 0.15 0.16 0.19 0.21 0.32 0.33 0.31 0.28 0.18 
Agglomeration 0.23 0.23 0.22 0.22 0.23 0.23 0.31 0.29 0.28 0.25 0.23 
 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 Average 
Henan 0.26 0.26 0.26 0.33 0.35 0.53 0.68 0.88 0.89 0.74 0.38 
Hebei 0.16 0.14 0.19 0.17 0.16 0.38 0.39 0.50 0.61 1.22 0.34 
Shanxi 0.12 0.14 0.13 0.28 0.38 0.51 0.66 0.91 0.76 1.11 0.36 
Shandong 0.40 0.38 0.36 0.30 0.26 0.18 0.11 0.03 0.14 0.19 0.23 
Anhui 0.22 0.23 0.25 0.21 0.24 0.21 0.13 0.12 0.21 0.28 0.23 
Agglomeration 0.23 0.23 0.24 0.26 0.28 0.36 0.40 0.49 0.52 0.71 0.31 
Area20032004200520062007200820092010201120122013
Henan 0.27 0.29 0.25 0.25 0.27 0.25 0.26 0.26 0.24 0.26 0.28 
Hebei 0.25 0.23 0.33 0.34 0.37 0.37 0.36 0.24 0.27 0.17 0.19 
Shanxi 0.28 0.29 0.21 0.22 0.24 0.26 0.25 0.25 0.22 0.21 0.16 
Shandong 0.10 0.09 0.14 0.13 0.10 0.06 0.36 0.37 0.36 0.35 0.33 
Anhui 0.27 0.27 0.15 0.16 0.19 0.21 0.32 0.33 0.31 0.28 0.18 
Agglomeration 0.23 0.23 0.22 0.22 0.23 0.23 0.31 0.29 0.28 0.25 0.23 
 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 Average 
Henan 0.26 0.26 0.26 0.33 0.35 0.53 0.68 0.88 0.89 0.74 0.38 
Hebei 0.16 0.14 0.19 0.17 0.16 0.38 0.39 0.50 0.61 1.22 0.34 
Shanxi 0.12 0.14 0.13 0.28 0.38 0.51 0.66 0.91 0.76 1.11 0.36 
Shandong 0.40 0.38 0.36 0.30 0.26 0.18 0.11 0.03 0.14 0.19 0.23 
Anhui 0.22 0.23 0.25 0.21 0.24 0.21 0.13 0.12 0.21 0.28 0.23 
Agglomeration 0.23 0.23 0.24 0.26 0.28 0.36 0.40 0.49 0.52 0.71 0.31 
Fig. 2

Mean efficiency of the central plains urban agglomeration.

Fig. 2

Mean efficiency of the central plains urban agglomeration.

Close modal
Using ArcGIS software, maps of agricultural green water use efficiency for 2003, 2009, 2019, and 2023 were generated, enabling spatio-temporal visualization of the results. The isopoint method classified the values into four levels: excellent, good, average, and poor (see Figure 3). The results show that in 2003, most cities in the Central Plains Urban Agglomeration had agricultural green water use efficiency below or at average levels, with only a few cities achieving excellent or good ratings. By 2009, efficiency levels in most cities had improved, particularly in central cities like Zhengzhou, which showed significantly higher efficiency than surrounding areas, reflecting the growing regional influence of core cities. By 2019, overall efficiency in the urban agglomeration had improved further, with a significant increase in the number of cities receiving excellent or good ratings. By 2009, efficiency levels in most cities had improved, particularly in central cities like Zhengzhou, which showed significantly higher efficiency than surrounding areas, reflecting the growing regional influence of core cities. By 2019, overall efficiency in the urban agglomeration had improved further, with a significant increase in the number of cities receiving excellent or good ratings. By 2023, most cities in the urban agglomeration had high agricultural green water use efficiency, with previously inefficient regions showing significant catch-up effects and narrowing spatial efficiency disparities. Due to factors like geographical location, regional resources, national development strategies, and local government planning, efficiency values across cities have been unevenly distributed over the years. The overall situation reflects both ‘neighborly’ and ‘neighbor-exclusion’ phenomena, making coordinated development of green water use efficiency difficult across the region. Moreover, the spatial spillover effect from central cities has become more pronounced, with smaller cities enhancing their comprehensive strength by borrowing high-level functions and innovation resources from core cities. This has led to improved agricultural green water use efficiency through expanded agricultural infrastructure and upgraded planting technology, fostering coordinated ‘social-economic-ecological’ development. Overall, compared to higher-efficiency areas, the low-efficiency regions in the Central Plains Urban Agglomeration have shown faster growth rates, especially due to spatial spillover effects. The urbanization process in these low-efficiency areas often reflects a lack of water resource management and technological applications, urging policymakers and urban planners to adopt focused strategies for spatial layout adjustments. Thus, cities with low water use efficiency can increase investment and upgrade water resource management infrastructure, improving efficiency and enabling sustainable water utilization, often creating a ‘low-efficiency club’ phenomenon.
Fig. 3

Temporal and spatial evolution of agricultural green water use efficiency in 2003, 2009, 2019, and 2023.

Fig. 3

Temporal and spatial evolution of agricultural green water use efficiency in 2003, 2009, 2019, and 2023.

Close modal

Benchmark regression of urban scale borrowing and agricultural green water use efficiency

The Hausman test results indicate the selection of the fixed effects model, rejecting the initially assumed random effects model. First, an ordinary least squares (OLS) fixed effects model including control variables was used for the baseline regression. Columns (1) through (5) sequentially add control variables in the regression. The results of Table 4 shows that, regardless of the inclusion of control variables, the coefficient for agricultural green water use efficiency is significantly positive at the 1% level. This suggests that urban-scale borrowing accelerates the diffusion of agricultural green water use efficiency, confirming Hypothesis 1. As the urban scale expands, the interaction between urban and rural areas strengthens, facilitating the widespread dissemination of advanced water resource management technologies and sustainable agricultural practices. This process enhances the green water use efficiency of surrounding agricultural regions, optimally allocating resources and aligning with environmental protection goals. Columns (6)–(8) display the grouped regressions for the three sub-indicators of urban scale borrowing. The results indicate that urban function borrowing positively promotes agricultural green water use efficiency at the 10% significance level. Previous studies confirm that urban open space systems, with ecological regulation, economic culture, and social systems, significantly advance urban ecological planning and sustainability processes (Wang & Qiu 2015). Urban innovation resource scale borrowing significantly reduces agricultural water resource green efficiency. Specifically, a 1% increase in urban innovation resource borrowing leads to a 0.005% increase in agricultural green water use efficiency. This is because the increase in urban innovation resources significantly enhances agricultural water resource efficiency through technological innovation, financial support, and optimized management models. Specifically, increased urban innovation resources provide advanced technological applications and management tools for agriculture, such as smart irrigation systems and efficient water management platforms. These innovations promote the conservation and rational use of agricultural water resources, thus improving green efficiency. Therefore, as the scale of urban innovation resources expands, the green water use efficiency of agricultural water resources improves. Urban population scale borrowing negatively affects agricultural green water use efficiency at the 5% significance level, though its impact is significantly smaller than the other three factors. The expansion of the Central Plains Urban Agglomeration has promoted the coordinated development of regional economic and social clusters and, through ‘soft power’ – the borrowing and synergistic effects of urban functions – facilitated the efficient flow of innovative elements. Meanwhile, the ‘hard power’ of the agglomeration – specifically, the concentration of innovation resources – has strongly supported the green allocation of agricultural water resources through efficient technological and capital investments, enhancing resource utilization and promoting sustainable development.

Table 4

Stepwise empirical regression results of the baseline model

(1)(2)(3)(4)(5)(6)(7)(8)
VARIABLESyyyyyyyy
         
Borrow2 0.079*** 0.052*** 0.038*** 0.043*** 0.038***    
 (0.008) (0.007) (0.009) (0.009) (0.009)    
Bfun      0.136*   
      (0.08)   
Bpop       −0.005**  
       (0.002)  
Bcreative        0.006*** 
        (0.001) 
CPRI 0.001 0.001 0.000 0.003** 0.003** 0.003** 0.005*** 0.004*** 
 (0.0001) (0.001) (0.001) (0.013) (0.013) (0.001) (0.002) (0.001) 
Resource  0.106*** 0.100*** 0.098*** 0.096*** 0.096*** 0.053*** 0.054*** 
  (0.013) (0.013) (0.013) (0.013) (0.01) (0.013) (0.010) 
Economic   0.036*** 0.056*** 0.052*** 0.077*** 0.100*** −0.028* 
   (0.012) (0.012) (0.012) (0.01) (0.015) (0.016) 
Finance    −0.568*** −0.608*** −0.627*** −0.520*** −0.383** 
    (0.139) (0.138) (0.01) (0.163) (0.152) 
Plant     0.012 0.020** 0.047*** 0.026*** 
     (0.008) (0.01) (0.006) (0.007) 
Constant 0.195*** 0.129*** −0.232* −0.462*** −0.449*** −0.747*** −0.800*** 0.340** 
 (0.028) (0.029) (0.124) (0.125) (0.125) (0.113) (0.139) (0.154) 
Observations 630 630 630 630 630 567 567 567 
40.16 66.06 52.71 50.42 42.23 36.30 35.36 59.10 
ID YES YES YES YES YES YES YES YES 
YEAR YES YES YEAR YES YES YES YES YES 
(1)(2)(3)(4)(5)(6)(7)(8)
VARIABLESyyyyyyyy
         
Borrow2 0.079*** 0.052*** 0.038*** 0.043*** 0.038***    
 (0.008) (0.007) (0.009) (0.009) (0.009)    
Bfun      0.136*   
      (0.08)   
Bpop       −0.005**  
       (0.002)  
Bcreative        0.006*** 
        (0.001) 
CPRI 0.001 0.001 0.000 0.003** 0.003** 0.003** 0.005*** 0.004*** 
 (0.0001) (0.001) (0.001) (0.013) (0.013) (0.001) (0.002) (0.001) 
Resource  0.106*** 0.100*** 0.098*** 0.096*** 0.096*** 0.053*** 0.054*** 
  (0.013) (0.013) (0.013) (0.013) (0.01) (0.013) (0.010) 
Economic   0.036*** 0.056*** 0.052*** 0.077*** 0.100*** −0.028* 
   (0.012) (0.012) (0.012) (0.01) (0.015) (0.016) 
Finance    −0.568*** −0.608*** −0.627*** −0.520*** −0.383** 
    (0.139) (0.138) (0.01) (0.163) (0.152) 
Plant     0.012 0.020** 0.047*** 0.026*** 
     (0.008) (0.01) (0.006) (0.007) 
Constant 0.195*** 0.129*** −0.232* −0.462*** −0.449*** −0.747*** −0.800*** 0.340** 
 (0.028) (0.029) (0.124) (0.125) (0.125) (0.113) (0.139) (0.154) 
Observations 630 630 630 630 630 567 567 567 
40.16 66.06 52.71 50.42 42.23 36.30 35.36 59.10 
ID YES YES YES YES YES YES YES YES 
YEAR YES YES YEAR YES YES YES YES YES 

Note: ***, **, * indicate significance at the 1, 5, and 10% levels, respectively. Standard errors are in parentheses.

Spatial autocorrelation test

According to Tobler's first law of geography, green innovation activities between regions are interconnected and involve spatial impacts generated by the flow of innovation. To determine whether economic variables exhibit spatial characteristics, exploratory spatial data analysis (ESDA) methods are widely employed. ESDA aims to analyze both the randomness and non-randomness of spatial distributions, including global and local spatial correlation analysis. From a global perspective, inertia analysis is typically conducted using Moran's I method, whose formula is:
(9)
where xi and xj represent the observation values for cities i and j (including urban scale borrowing and agricultural green water use efficiency), n is the total number of cities, W is the standardized spatial weight matrix, and S2 is the variance of city observation values.

Under water resource constraints (Fang & Bu 2004), urban-scale construction is marked by high consumption and low efficiency, relying on the development of total resources. This suggests an inseparable correlation between the two. Li & Luo (2016) noted that inter-city location factors influence the ecological and economic environment of both the city and its neighboring areas, implying that the ecological environment of neighboring areas is likely to experience spatial spillover effects. Additionally, Moran's I index was used to check for significant spatial autocorrelation between urban-scale borrowing and agricultural green water use efficiency. The closer the Moran's I value is to 1, the stronger the spatial positive correlation, ensuring the rigor and scientific validity of the hypothesis. The results in Table 5 show the global Moran's index measurement, with the Moran's index of variables consistently above 0, indicating a strong positive correlation. This suggests an interdependent relationship between the explanatory and dependent variables, accompanied by clustering effects in spatial distribution. The Moran's I index fluctuated minimally from 2015 to 2023, indicating stable spatial aggregation and dependence of the variables during this period. Therefore, using spatial econometric models for subsequent empirical analysis is both reasonable and feasible.

Table 5

Global moran index of dependent and independent variables.

VariablesIp-value*VariablesIp-value*
efficiency2003 −0.062 0.062 borrow2003 0.105 0.102 
efficiency2004 −0.019 0.423 borrow2004 0.102 0.106 
efficiency2005 0.095 0.051 borrow2005 0.111 0.092 
efficiency2006 0.089 0.057 borrow2006 0.122 0.077 
efficiency2007 0.069 0.094 borrow2007 0.116 0.081 
efficiency2008 −0.155 0.060 borrow2008 0.123 0.075 
efficiency2009 −0.013 0.340 borrow2009 0.110 0.094 
efficiency2010 −0.141 0.088 borrow2010 0.087 0.033 
efficiency2011 −0.112 0.062 borrow2011 0.086 0.035 
efficiency2012 −0.041 0.467 borrow2012 0.082 0.144 
efficiency2013 0.005 0.299 borrow2013 0.107 0.098 
efficiency2014 −0.153 0.055 borrow2014 0.107 0.098 
efficiency2015 −0.019 0.416 borrow2015 0.088 0.031 
efficiency2016 −0.022 0.431 borrow2016 0.094 0.020 
efficiency2017 0.023 0.030 borrow2017 0.082 0.040 
efficiency2018 −0.087 0.251 borrow2018 0.063 0.181 
efficiency2019 0.076 0.079 borrow2019 0.051 0.012 
efficiency2020 −0.063 0.349 borrow2020 0.031 0.271 
efficiency2021 −0.076 0.297 borrow2021 0.061 0.183 
efficiency2022 −0.027 0.462 borrow2022 0.082 0.032 
efficiency2023 −0.090 0.224 borrow2023 0.101 0.096 
VariablesIp-value*VariablesIp-value*
efficiency2003 −0.062 0.062 borrow2003 0.105 0.102 
efficiency2004 −0.019 0.423 borrow2004 0.102 0.106 
efficiency2005 0.095 0.051 borrow2005 0.111 0.092 
efficiency2006 0.089 0.057 borrow2006 0.122 0.077 
efficiency2007 0.069 0.094 borrow2007 0.116 0.081 
efficiency2008 −0.155 0.060 borrow2008 0.123 0.075 
efficiency2009 −0.013 0.340 borrow2009 0.110 0.094 
efficiency2010 −0.141 0.088 borrow2010 0.087 0.033 
efficiency2011 −0.112 0.062 borrow2011 0.086 0.035 
efficiency2012 −0.041 0.467 borrow2012 0.082 0.144 
efficiency2013 0.005 0.299 borrow2013 0.107 0.098 
efficiency2014 −0.153 0.055 borrow2014 0.107 0.098 
efficiency2015 −0.019 0.416 borrow2015 0.088 0.031 
efficiency2016 −0.022 0.431 borrow2016 0.094 0.020 
efficiency2017 0.023 0.030 borrow2017 0.082 0.040 
efficiency2018 −0.087 0.251 borrow2018 0.063 0.181 
efficiency2019 0.076 0.079 borrow2019 0.051 0.012 
efficiency2020 −0.063 0.349 borrow2020 0.031 0.271 
efficiency2021 −0.076 0.297 borrow2021 0.061 0.183 
efficiency2022 −0.027 0.462 borrow2022 0.082 0.032 
efficiency2023 −0.090 0.224 borrow2023 0.101 0.096 

Although the global Moran's I value reflects spatial correlation, Arellano & Bover (1995) argue that it may overlook the atypical characteristics of local spatial samples. Therefore, local Moran's I was applied to investigate the local spatial distribution of the variables. The formula is:
(10)
The results of the local Moran's I test are shown in scatter plots (see Figure 4). This study selected 2009 and 2019 as key years for local Moran's I analysis, as these two years marked important policy turning points for the Central Plains Urban Agglomeration. 2009 marked the year the concept of the Central Plains Urban Agglomeration was proposed, leading to a regional strategy adjustment. In 2019, the National Water Saving Action Plan was implemented, enhancing agricultural water use efficiency. By analyzing agricultural green water use efficiency and urban scale borrowing data for these two years, we can uncover the spatial effects and diffusion mechanisms between these variables under policy changes. During the economic growth and new urbanization of the Central Plains Urban Agglomeration, the spatial distribution in most regions exhibited clear high–high and low–low clustering characteristics. Specifically, both urban-scale borrowing and agricultural green water use efficiency demonstrated this spatial clustering pattern. This spatial pattern indicates that improvements in urban-scale borrowing are positively correlated with growth in agricultural green water use efficiency. This effect has significant spillover effects between regions, further driving spatial clustering in neighboring areas. Combining the benchmark regression results, urban scale borrowing – through borrowing urban functions and innovation resources – facilitates the diffusion of agricultural green water use efficiency and strengthens efficiency improvements in surrounding areas through spatial spillover effects. Therefore, the local Moran's I analysis confirms the positive spatial correlation between urban scale borrowing and agricultural green water use efficiency, revealing the synergistic development and mutual influence mechanism within the region.
Fig. 4

Scatter plot of the local Moran's I index.

Fig. 4

Scatter plot of the local Moran's I index.

Close modal

Spatial effects

Following the approach of Elhorst (2014), several methods were used for the identification analysis of spatial econometric models, including the Lagrange multiplier (LM) test, Hausman test, and simplification tests such as the Wald and likelihood ratio (LR) tests. To comprehensively test spatial effects, this study employed five different spatial weight matrices: adjacency matrix, economic distance matrix, first-order inverse distance matrix, second-order inverse distance matrix, and economic nested distance matrix. Each matrix passed the significance test, confirming that spatial effects consistently show significance in the selected models. These results further validate the universal presence of spatial effects and show that the selected models can effectively capture various types of spatial dependencies. The results in Table 6 confirm the spatial econometric models using standardized spatial distance matrices and the LM test to determine the optimality of the SAR, SEM, and SDM models, followed by a SEM test. The LM test results show that the P-values for both the SAR and SEM models are significantly below 10%, indicating the simultaneous presence of spatial error and lag effects. In the LM test, the spatial econometric models using the spatial distance weight matrix reject the hypotheses ‘H0:θ = 0’ and ‘H0:θ + βp = 0,’ suggesting that the SDM outperforms the spatial lag and error models. The Hausman test results subsequently reject random effects, leading to the selection of the spatial econometric model under fixed effects. The Wald or LR test results confirm that the SDM model does not degrade into a spatial lag model (SLM) or SEM model, and related studies have shown that the SDM mitigates endogeneity problems to some extent (Vega-Carrillo et al. 2015).

Table 6

Spatial econometric model test results.

Test statisticAdjacencyEconomic distanceFirst-order inverse distanceSecond-order inverse distanceEconomic nesting
LM-lag 22.368*** 15.502*** 22.058*** 7.489*** 20.082*** 
Robust LM-lag 7.671*** 8.679*** 17.407*** 18.825*** 17.479*** 
LM-error 44.328*** 38.613*** 90.436*** 34.211*** 81.739*** 
Robust LM-error 29.630*** 31.790*** 85.784*** 45.548*** 79.136*** 
Hausman test 43.67*** 21.14* 39.39*** 22.39** 40.16*** 
LR lag 15.50** 27.03*** 14.76** 49.63*** 24.03*** 
LR error 15.70** 28.29*** 18.79** 49.28*** 26.57*** 
Wald lag 16.83*** 10.24* 25.03*** 16.11*** 31.30*** 
Wald error 23.34*** 11.39* 28.14*** 23.63*** 34.30*** 
Test statisticAdjacencyEconomic distanceFirst-order inverse distanceSecond-order inverse distanceEconomic nesting
LM-lag 22.368*** 15.502*** 22.058*** 7.489*** 20.082*** 
Robust LM-lag 7.671*** 8.679*** 17.407*** 18.825*** 17.479*** 
LM-error 44.328*** 38.613*** 90.436*** 34.211*** 81.739*** 
Robust LM-error 29.630*** 31.790*** 85.784*** 45.548*** 79.136*** 
Hausman test 43.67*** 21.14* 39.39*** 22.39** 40.16*** 
LR lag 15.50** 27.03*** 14.76** 49.63*** 24.03*** 
LR error 15.70** 28.29*** 18.79** 49.28*** 26.57*** 
Wald lag 16.83*** 10.24* 25.03*** 16.11*** 31.30*** 
Wald error 23.34*** 11.39* 28.14*** 23.63*** 34.30*** 

Note: ***, **, * indicate significance at the 1, 5, and 10% levels, respectively. Standard errors are in parentheses.

The empirical regression results for the SDM are presented in Table 7. Columns (1) and (3) represent the main effect regressions of the comprehensive urban scale borrowing index and the three sub-dimensions on agricultural green water use efficiency, respectively. Columns (2) and (4) represent the spatial effect regressions for the total urban scale borrowing indicator and the three sub-dimension indicators on agricultural green water use efficiency. The model results provide support for the theoretical analysis and prior assumptions. The spatial correlation coefficients in both the comprehensive urban scale borrowing regression and the three sub-dimension regressions are significantly negative, which can be explained in several ways: On one hand, as distance increases, peripheral cities' ability to absorb central cities' scale borrowing weakens, particularly in terms of technology transfer, knowledge dissemination, and policy diffusion. Spatial barriers and information asymmetry may diminish the spillover effect. On the other hand, the expansion of central cities may lead to excessive competition and resource misallocation, causing peripheral cities to face funding, technology, and other limitations when implementing green water use technologies and policies, thereby hindering efficiency improvements. The LR test results for both regression types support using fixed effects to control unobserved time differences, enhancing the reliability of the model's regression coefficients as ‘net effects.’ In column (1), the coefficient for urban scale borrowing (Borrow) shows that a 1% increase in urban function borrowing raises local agricultural green water use efficiency by 0.0985%. Hypothesis 1 is confirmed. This suggests that urban-scale borrowing, through urban spatial network accumulation, drives local economic growth, which significantly boosts agricultural green water use efficiency. The spatial coefficient regression shows that urban scale borrowing is positively significant at the 10% level, with an increasing coefficient value. This suggests that after accounting for spatial dependence, the borrowing effect between cities becomes more pronounced. This phenomenon can be explained by the role of large cities in technology innovation, experience accumulation, policy dissemination, and other multidimensional spillover effects. The advanced practices and resource advantages of large cities effectively promote agricultural green water use efficiency in neighboring areas, advancing regional green development. Column (3) results show that urban function borrowing (bfun), innovation resource borrowing (creative), and population scale borrowing (people) all positively affect local green water use efficiency. Specifically, a 1% increase in urban functions raises local agricultural green water use efficiency by 0.334%; a 1% increase in urban innovation resources raises efficiency by 0.00119%; and a 1% increase in population scale raises efficiency by 0.0120%. The synergy between urban function borrowing, innovation resource borrowing, and population scale borrowing enhances the green efficiency of urban scale borrowing. This indicates that communication between cities promotes industrial economic growth, technological innovation, and the implementation of functional division, strengthening the externalities of urban networks. Among these, diversified industrial functions and economic integration levels are key foundations for the diffusion and improvement of agricultural green water use efficiency (Marshall 1890).

Table 7

Spatial econometric model regression results

(1)(2)(3)(4)
Main CoefficientSpatial CoefficientMain CoefficientSpatial Coefficient
Borrow2 0.0985*** 0.112*   
 (0.0244) (0.0476)   
Bfun   0.334* 0.737* 
   (0.1486) (0.409) 
Bpop   0.0120* −0.0110 
   (0.00667) (0.0108) 
Bcreative   0.00119 0.00354 
   (0.00102) (0.00206) 
CPRI 0.00124 −0.000642 0.00124 −0.000447 
 (0.00155) (0.00392) (0.00156) (0.00396) 
Resource 0.127*** 0.116 0.131*** 0.0822 
 (0.0185) (0.0719) (0.0188) (0.0753) 
Economic −0.154*** 0.0560 −0.147*** 0.0781 
 (0.0343) (0.0765) (0.0352) (0.0777) 
Finance −0.375* −0.0545 −0.408* −0.121 
 (0.178) (0.540) (0.228) (0.542) 
Plant 0.00702 −0.0327* 0.00850 −0.0379** 
 (0.00602) (0.0134) (0.00608) (0.0141) 
Spatial −0.1318* −0.1318* −0.1477* −0.1477* 
rho (0.0622) (0.0622) (0.0624) (0.0624) 
Variance 0.0176*** 0.0176*** 0.0175*** 0.0175*** 
sigma2 e (0.000996) (0.000996) (0.000987) (0.000987) 
630 630 630 630 
ID YES YES YES YES 
YEAR YES YES YEAR YES 
(1)(2)(3)(4)
Main CoefficientSpatial CoefficientMain CoefficientSpatial Coefficient
Borrow2 0.0985*** 0.112*   
 (0.0244) (0.0476)   
Bfun   0.334* 0.737* 
   (0.1486) (0.409) 
Bpop   0.0120* −0.0110 
   (0.00667) (0.0108) 
Bcreative   0.00119 0.00354 
   (0.00102) (0.00206) 
CPRI 0.00124 −0.000642 0.00124 −0.000447 
 (0.00155) (0.00392) (0.00156) (0.00396) 
Resource 0.127*** 0.116 0.131*** 0.0822 
 (0.0185) (0.0719) (0.0188) (0.0753) 
Economic −0.154*** 0.0560 −0.147*** 0.0781 
 (0.0343) (0.0765) (0.0352) (0.0777) 
Finance −0.375* −0.0545 −0.408* −0.121 
 (0.178) (0.540) (0.228) (0.542) 
Plant 0.00702 −0.0327* 0.00850 −0.0379** 
 (0.00602) (0.0134) (0.00608) (0.0141) 
Spatial −0.1318* −0.1318* −0.1477* −0.1477* 
rho (0.0622) (0.0622) (0.0624) (0.0624) 
Variance 0.0176*** 0.0176*** 0.0175*** 0.0175*** 
sigma2 e (0.000996) (0.000996) (0.000987) (0.000987) 
630 630 630 630 
ID YES YES YES YES 
YEAR YES YES YEAR YES 

Note: ***, **, * indicate significance at the 1, 5, and 10% levels, respectively. Standard errors are in parentheses.

To compare the direct and spillover effects of urban-scale borrowing on agricultural green water use efficiency more effectively, a SDM decomposition was performed. The direct effect measures how changes in a city's own explanatory variables, along with the spillover of neighboring cities' variables, influence the dependent variable through a ‘feedback mechanism.’ The indirect effect measures the impact of explanatory variable changes in other regions on the dependent variable in the local area. This study employed two weight matrices to decompose and test the effects, with the results presented in Table 8. Both the direct and total effects of urban scale borrowing are significantly positive under both weight matrices, indicating that it not only improves local agricultural green water use efficiency but also promotes efficiency improvements in neighboring areas through spatial spillover effects. Specifically, with the economic distance matrix weighting, the direct effect (0.0978) and total effect (0.196) are positive at the 1% significance level, while the adjacency matrix shows significance at least at the 10% level. This reflects that large cities' green development advantages, through mechanisms like technological innovation, resource sharing, and policy transmission, can radiate to surrounding regions via spatial spillover effects. However, the indirect effects are less significant under both spatial weight matrices, with some even being insignificant. This suggests that the transmission of borrowing effects across regions is delayed or heterogeneous, particularly due to differences in economic, technological, and policy responses. Notably, the positive relationship of borrowing effects aligns with the benchmark regression results, further confirming that large cities, through their comprehensive advantages, enhance agricultural green water use efficiency in neighboring cities. The positive impact of indirect spillover effects operates through the following mechanism: with the borrowing of advanced urban functions, large cities absorb capital, technologies, and implicit knowledge, while simultaneously shifting emerging productive services to neighboring regions, thus promoting a new round of agricultural green water use efficiency growth. The combined positive impact of direct and indirect spillover effects, reflecting a ‘win–win’ spatial spillover from healthy competition and positive demonstration effects, benefits rural agriculture by generating new growth points in ‘green’ resource production efficiency, confirming Hypothesis 2. In summary, although urban scale borrowing significantly contributes to regional green development through direct and indirect spillover effects, the weakening of indirect effects suggests regional differences in spatial policy transmission, highlighting the need for further optimization of cross-regional policy design.

Table 8

Decomposition of total, direct, and indirect effects.

VariablesEconomic geography matrixAdjacency matrix
LR directLR indirectLR totalLR directLR indirectLR total
Borrow2 0.0978*** 0.0983* 0.196*** 0.105** 0.0463 0.151* 
 (0.0253) (0.0464) (0.0444) (0.0357) (0.0769) (0.0680) 
CPRI 0.00118 −0.000822 0.000360 −0.000507 −0.00184 −0.00235 
 (0.00150) (0.00349) (0.00386) (0.00190) (0.00403) (0.00416) 
Resource 0.127*** 0.103 0.230*** 0.0630** 0.168** 0.231*** 
 (0.0181) (0.0656) (0.0611) (0.0212) (0.0598) (0.0664) 
Economic −0.156*** 0.0668 −0.0890 −0.0744 −0.266** −0.340*** 
 (0.0342) (0.0756) (0.0679) (0.0578) (0.0996) (0.0775) 
Finance −0.373* −0.0363 −0.410 −0.174 1.260 1.086 
 (0.168) (0.511) (0.595) (0.208) (0.661) (0.777) 
Plant 0.00767 −0.0311* −0.0235 0.00735 0.0146 0.0220 
 (0.00582) (0.0128) (0.0126) (0.00760) (0.0187) (0.0191) 
VariablesEconomic geography matrixAdjacency matrix
LR directLR indirectLR totalLR directLR indirectLR total
Borrow2 0.0978*** 0.0983* 0.196*** 0.105** 0.0463 0.151* 
 (0.0253) (0.0464) (0.0444) (0.0357) (0.0769) (0.0680) 
CPRI 0.00118 −0.000822 0.000360 −0.000507 −0.00184 −0.00235 
 (0.00150) (0.00349) (0.00386) (0.00190) (0.00403) (0.00416) 
Resource 0.127*** 0.103 0.230*** 0.0630** 0.168** 0.231*** 
 (0.0181) (0.0656) (0.0611) (0.0212) (0.0598) (0.0664) 
Economic −0.156*** 0.0668 −0.0890 −0.0744 −0.266** −0.340*** 
 (0.0342) (0.0756) (0.0679) (0.0578) (0.0996) (0.0775) 
Finance −0.373* −0.0363 −0.410 −0.174 1.260 1.086 
 (0.168) (0.511) (0.595) (0.208) (0.661) (0.777) 
Plant 0.00767 −0.0311* −0.0235 0.00735 0.0146 0.0220 
 (0.00582) (0.0128) (0.0126) (0.00760) (0.0187) (0.0191) 

Note: ***, **, * indicate significance at the 1, 5, and 10% levels, respectively. Standard errors are in parentheses.

Modifications to the empirical model

Considering the lag effects in the relationship between urban scale borrowing and agricultural green water use efficiency, this study introduces a dynamic panel model to analyze the impact paths of urban scale borrowing across different time periods and its dynamic effect mechanism on agricultural green water use efficiency (see Table 9). The system GMM (sGMM) and difference GMM (dGMM) methods are employed to address endogeneity and estimation bias in the static panel model. In model (2), the lagged agricultural green water use efficiency from the previous period is included as an explanatory variable. The regression results indicate that the lagged dependent variable (L.y) is significantly positive in both the sGMM and dGMM models, with a strong inertia effect at the 1% significance level. Robustness tests were conducted to verify the model's specification and the validity of the instrument variables. The results show that the disturbance term exhibits first-order autocorrelation but no second-order autocorrelation, with a Hansen test value greater than 0.1, confirming no over-identification problem and validating the instrument variables. The coefficients for the urban scale borrowing variable are 0.794 (sGMM) and 0.674 (dGMM) in both models, both significantly positive, confirming the positive impact of urban scale borrowing on agricultural green water use efficiency. Overall, the robustness test results support the stability of the regression results, showing that urban scale borrowing significantly promotes agricultural green water use efficiency and that the lag effect has a significant inertia impact on improving it.

Table 9

Dynamic panel regression results.

sGMMdGMM
Variablesyy
L.y 0.794*** 0.674*** 
 (0.01) (0.01) 
L2.y −0.036 −0.081 
 (0.74) (0.43) 
Borrow2 0.153*** 0.206*** 
 (0.01) (0.01) 
L.Borrow2 −0.160*** −0.115 
 (0.01) (0.10) 
Finance −0.850 −0.511 
 (0.88) (0.61) 
L.Finance 0.977 0.344 
 (0.98) (0.33) 
L2.Finance −0.295 0.155 
 (0.41) (0.70) 
Plant −0.000 0.003 
 (0.99) (0.86) 
L.Plant 0.033 0.021* 
 (0.04) (0.07) 
L2.Plant −0.024 −0.051** 
 (0.11) (0.02) 
Constant 0.477*** 0.562*** 
 (0.00) (0.001) 
AR1 0.001 0.151 
AR2 0.235 0.689 
Hansen 0.762 0.689 
sGMMdGMM
Variablesyy
L.y 0.794*** 0.674*** 
 (0.01) (0.01) 
L2.y −0.036 −0.081 
 (0.74) (0.43) 
Borrow2 0.153*** 0.206*** 
 (0.01) (0.01) 
L.Borrow2 −0.160*** −0.115 
 (0.01) (0.10) 
Finance −0.850 −0.511 
 (0.88) (0.61) 
L.Finance 0.977 0.344 
 (0.98) (0.33) 
L2.Finance −0.295 0.155 
 (0.41) (0.70) 
Plant −0.000 0.003 
 (0.99) (0.86) 
L.Plant 0.033 0.021* 
 (0.04) (0.07) 
L2.Plant −0.024 −0.051** 
 (0.11) (0.02) 
Constant 0.477*** 0.562*** 
 (0.00) (0.001) 
AR1 0.001 0.151 
AR2 0.235 0.689 
Hansen 0.762 0.689 

Note: ***, **, * indicate significance at the 1, 5, and 10% levels, respectively. Standard errors are in parentheses.

Modifications to the data sample

To ensure the scientific rigor of the empirical study, data samples were selected for model testing to verify robustness, as shown in Table 10. Since Zhengzhou, the capital of Henan province, and Jiyuan, a sub-provincial city, have notably developed in services, high-tech industries, and financial services, their strong economic influence in both scale and overall development exceeds that of other prefecture-level cities. To avoid the influence of these cities on the empirical results, column (1) presents the regression results after excluding the provincial capital Zhengzhou and the sub-provincial city Jiyuan from the sample. Column (2) presents the regression results after trimming the top and bottom 1% of the sample to mitigate endogeneity effects. Column (3) presents the results with the dependent variable measured using SBM–maximum likelihood (ML) rather than SBM–DEA. Column (4) shows the new explanatory variables weighted by principal component analysis of urban function borrowing (bfun), urban economic activity density borrowing (bden), and urban innovation resource borrowing (bnet). The robustness test results are consistent with the OLS and SDM model regression coefficients, confirming the positive effect of urban-scale borrowing on agricultural green water use efficiency.

Table 10

Spatial regression robustness test.

(1)(2)(3)(4)
Exclusion of samples
Trimming 1%
Replacing Variable 1 (Dependent Variable)
Replacing Variable 2 (Independent Variable)
Main CoefficientSpatial CoefficientMain CoefficientSpatial CoefficientMain CoefficientSpatial CoefficientMain CoefficientSpatial Coefficient
Borrow2 0.116*** 0.155*** 0.119*** 0.163***   0.134*** 0.183*** 
 (0.0246) (0.0470) (0.0253) (0.0485)   (0.0264) (0.0547) 
Borrow1     0.0103** −0.0128   
     (0.00373) (0.00726)   
Spatial −0.1213* −0.1213* −0.1220* −0.1220* −0.1303* −0.1303* −0.1321* −0.1321* 
rho (0.0624) (0.0624) (0.0627) (0.0627) (0.0724) (0.0724) (0.0734) (0.0734) 
Variance 0.0181*** 0.0181*** 0.0193*** 0.0193*** 0.000414*** 0.000414*** 0.0278*** 0.0278*** 
sigma2 e (0.00105) (0.00105) (0.00109) (0.00109) (0.0000233) (0.0000233) (0.00157) (0.00157) 
630 630 630 630 630 630 630 630 
ID YES YES YES YES YES YES YES YES 
YEAR YES YES YEAR YES YES YES YES YES 
(1)(2)(3)(4)
Exclusion of samples
Trimming 1%
Replacing Variable 1 (Dependent Variable)
Replacing Variable 2 (Independent Variable)
Main CoefficientSpatial CoefficientMain CoefficientSpatial CoefficientMain CoefficientSpatial CoefficientMain CoefficientSpatial Coefficient
Borrow2 0.116*** 0.155*** 0.119*** 0.163***   0.134*** 0.183*** 
 (0.0246) (0.0470) (0.0253) (0.0485)   (0.0264) (0.0547) 
Borrow1     0.0103** −0.0128   
     (0.00373) (0.00726)   
Spatial −0.1213* −0.1213* −0.1220* −0.1220* −0.1303* −0.1303* −0.1321* −0.1321* 
rho (0.0624) (0.0624) (0.0627) (0.0627) (0.0724) (0.0724) (0.0734) (0.0734) 
Variance 0.0181*** 0.0181*** 0.0193*** 0.0193*** 0.000414*** 0.000414*** 0.0278*** 0.0278*** 
sigma2 e (0.00105) (0.00105) (0.00109) (0.00109) (0.0000233) (0.0000233) (0.00157) (0.00157) 
630 630 630 630 630 630 630 630 
ID YES YES YES YES YES YES YES YES 
YEAR YES YES YEAR YES YES YES YES YES 

Note: ***, **, * indicate significance at the 1, 5, and 10% levels, respectively. Standard errors are in parentheses.

Modifications to the weight matrix

To ensure the robustness of the empirical analysis, this study further modified the spatial weight matrix settings, selecting the second-order inverse distance matrix, first-order inverse distance matrix, and economic nested weight matrix for reanalysis. The regression results are shown in Table 11. The estimation results for the direct effect show that the core explanatory variable ‘borrow’ maintains a stable, positive, significant effect under all three weight matrix settings, with significance levels above 10%, confirming the robustness of the direct effect despite changes in the spatial weight matrix. Regarding the spatial effect, the ‘borrow’ variable's performance in the spatial spillover effect coefficient varies slightly. Under the second-order inverse distance matrix, the spatial effect coefficient is 0.131, significant only at the 10% level, while the first-order inverse distance and economic nested weight matrices show coefficients of 0.408 and 0.448, with significance increasing to 5 and 1%, respectively. This suggests that as spatial distance decreases and economic ties strengthen, inter-regional borrowing behavior is more likely to generate significant spatial spillover effects. Additionally, the spatial autocorrelation parameter rho shows positive clustering under all three spatial weight matrix settings, confirming the robustness and effectiveness of the model's spatial correlation structure. In conclusion, the robustness test results for the changed spatial weight matrices show that the direct effect of the core variable ‘borrow’ remains stable, and the spatial spillover effect becomes more pronounced with strengthened spatial economic ties, supporting the analysis conclusions. This also highlights the importance of the spatial transmission mechanism in fiscal behavior.

Table 11

Spatial regression robustness test.

VariablesSecond-Order Inverse Distance
First-Order Inverse Distance
Economic Nested Main Coefficient
Main CoefficientSpatial CoefficientMain CoefficientSpatial CoefficientMain CoefficientSpatial Coefficient
Borrow2 0.0682* 0.131* 0.0684* 0.408** 0.0692* 0.448*** 
 (0.0379) (0.0729) (0.0360) (0.185) (0.0384) (0.127) 
Spatial 0.244** 0.244** 0.0477 0.0477 0.0412 0.0412 
rho (0.1173) (0.1173) (0.152) (0.152) (0.152) (0.152) 
Variance 0.0270*** 0.0270*** 0.0273*** 0.0273*** 0.0270*** 0.0270*** 
sigma2 e (0.00153) (0.00153) (0.00154) (0.00154) (0.00152) (0.00152) 
630 630 630 630 630 630 
ID YES YES YES YES YES YES 
YEAR YES YES YES YES YES YES 
VariablesSecond-Order Inverse Distance
First-Order Inverse Distance
Economic Nested Main Coefficient
Main CoefficientSpatial CoefficientMain CoefficientSpatial CoefficientMain CoefficientSpatial Coefficient
Borrow2 0.0682* 0.131* 0.0684* 0.408** 0.0692* 0.448*** 
 (0.0379) (0.0729) (0.0360) (0.185) (0.0384) (0.127) 
Spatial 0.244** 0.244** 0.0477 0.0477 0.0412 0.0412 
rho (0.1173) (0.1173) (0.152) (0.152) (0.152) (0.152) 
Variance 0.0270*** 0.0270*** 0.0273*** 0.0273*** 0.0270*** 0.0270*** 
sigma2 e (0.00153) (0.00153) (0.00154) (0.00154) (0.00152) (0.00152) 
630 630 630 630 630 630 
ID YES YES YES YES YES YES 
YEAR YES YES YES YES YES YES 

Note: ***, **, * indicate significance at the 1, 5, and 10% levels, respectively. Standard errors are in parentheses.

Examination of the spatial spillover effect decay boundary

According to Tobler's first law of geography, spatial correlation within cities decreases as the geographic distance between them increases. Thus, does the spatial spillover effect of urban scale borrowing on agricultural green water use efficiency also follow a decay pattern as geographic distance increases? To explore this, the study adopts the method of Sun et al. (2022), reconstructing the spatial weight matrix with a threshold, as follows:
(11)

The matrix in the formula is a thresholded spatial weight matrix, and . where d represents the distance threshold, dmin is the shortest distance between prefecture-level cities, and dmax is the longest distance. The distance increment from dmin to dmax is set to 30 km in this study.

Based on this setting, the shortest distance between prefecture-level cities is set to 30 km, and the SDM is tested. The trend of spatial spillover effect coefficients with geographic distance is displayed in Figure 5. The figure clearly shows that the effect of urban-scale borrowing on improving agricultural green water use efficiency exhibits a significant spatial decay boundary. This effect diminishes and can even become negative as geographic distance increases. Cities closer to each other can access more innovation resources and technological support through scale borrowing. However, as the distance between cities grows, the interaction effect weakens, ultimately limiting or even reversing the improvement in green water use efficiency. The significance of the spatial spillover effect coefficients shows that the spatial spillover effect of urban scale borrowing on agricultural green water use efficiency is only significant within 210 km. Beyond this distance, the effect becomes insignificant, indicating that the spatial spillover boundary for urban scale borrowing accelerating the diffusion of agricultural green water use efficiency is 210 km. This phenomenon underscores the critical role of geographic distance in green technology innovation and resource sharing. It also offers empirical evidence for regional policymakers to optimize spatial layouts, allocate resources efficiently, and promote the coordinated improvement of green water use efficiency. It conforms to the assumption 3 setting.
Fig. 5

Spatial spillover effect decay boundary chart.

Fig. 5

Spatial spillover effect decay boundary chart.

Close modal

The previous results indicate that agricultural green water use efficiency exhibits significant temporal heterogeneity, making a temporal attribute analysis necessary. Specifically, it is crucial to examine whether the impact of urban-scale borrowing on agricultural green water use efficiency exhibits phase characteristics over time. This study uses 2009 as the reference year (the year the concept of the Central Plains Urban Agglomeration was introduced) to explore the phased characteristics of urban scale borrowing's impact on agricultural green water use efficiency. According to the regression results in Table 12, after 2009, the significance of the impact of urban-scale borrowing on agricultural green water use efficiency increased. This change reflects that with the advancement of the regional integration strategy, spatial effects and policy synergies within the region became more evident, driving cross-regional improvements in agricultural green water use efficiency. Notably, the increase in the significance of both spatial and primary coefficients indicates that the spillover effects of policy and technology have become more efficient in inter-regional transmission, further enhancing agricultural green water use efficiency. Additionally, considering that urban-scale borrowing effects are more pronounced in highly urbanized areas, this further validates the advantages of these areas in resource allocation, technological innovation, and policy support. The infrastructure and economic activity density in these regions enables them to better absorb and utilize the green water benefits brought by urban-scale borrowing, thereby improving agricultural green water use efficiency. High urbanization enhances regional economic vitality and drives the application and innovation of green technologies, further optimizing agricultural water management.

Table 12

Spatial regression temporal attribute heterogeneity analysis.

Variables2003-2008
2009-2023
Main coefficientSpatial coefficientMain coefficientSpatial coefficient
Borrow −0.163 0.208 0.143*** 0.213*** 
 (0.162) (0.338) (0.0270) (0.0542) 
Controls Controlled Controlled Controlled Controlled 
Spatial rho −0.207** −0.207** 0.161 0.161 
 (0.0942) (0.0942) (0.0846) (0.0846) 
Variance 0.00694*** 0.00694*** 0.0161*** 0.0161*** 
sigma2 e (0.000733) (0.000733) (0.00108) (0.00108) 
180 180 450 450 
Low urbanization
High urbanization
Main coefficientSpatial coefficientMain coefficientSpatial coefficient
Borrow 0.0685 0.239* 0.259*** 0.390* 
 (0.0512) (0.134) (0.0483) (0.217) 
Controls Controlled Controlled Controlled Controlled 
Spatial rho −0.0120 −0.0120 −0.0765 −0.0765 
 (0.0946) (0.0946) (0.0902) (0.0902) 
Variance 0.0263*** 0.0263*** 0.0253*** 0.0253*** 
sigma2 e (0.00197) (0.00197) (0.00217) (0.00217) 
357 357 273 273 
ID YES YES YES YES 
YEAR YES YES YEAR YES 
Variables2003-2008
2009-2023
Main coefficientSpatial coefficientMain coefficientSpatial coefficient
Borrow −0.163 0.208 0.143*** 0.213*** 
 (0.162) (0.338) (0.0270) (0.0542) 
Controls Controlled Controlled Controlled Controlled 
Spatial rho −0.207** −0.207** 0.161 0.161 
 (0.0942) (0.0942) (0.0846) (0.0846) 
Variance 0.00694*** 0.00694*** 0.0161*** 0.0161*** 
sigma2 e (0.000733) (0.000733) (0.00108) (0.00108) 
180 180 450 450 
Low urbanization
High urbanization
Main coefficientSpatial coefficientMain coefficientSpatial coefficient
Borrow 0.0685 0.239* 0.259*** 0.390* 
 (0.0512) (0.134) (0.0483) (0.217) 
Controls Controlled Controlled Controlled Controlled 
Spatial rho −0.0120 −0.0120 −0.0765 −0.0765 
 (0.0946) (0.0946) (0.0902) (0.0902) 
Variance 0.0263*** 0.0263*** 0.0253*** 0.0253*** 
sigma2 e (0.00197) (0.00197) (0.00217) (0.00217) 
357 357 273 273 
ID YES YES YES YES 
YEAR YES YES YEAR YES 

Note: ***, **, * indicate significance at the 1, 5, and 10% levels, respectively. Standard errors are in parentheses.

Research conclusions

Focusing on the Central Plains Urban Agglomeration, this study builds an input–output system to measure agricultural green water use efficiency from 2003 to 2023 using indicators such as agricultural water footprint, sowing area, total agricultural output, and rural social development index, applying the SBM–DEA model. This study examines the spatial spillover effects of urban function borrowing on agricultural green water use efficiency through three sub-dimensions: urban high-level function borrowing, urban innovation resource borrowing, and urban population scale borrowing. The results show that: (1) Agricultural green water use efficiency in the Central Plains Urban Agglomeration significantly increased from 2003 to 2023, but regional disparities widened, with spatial differentiation becoming more pronounced. The spillover effect from core cities gradually became more evident, and low-efficiency regions experienced faster catch-up growth driven by policy direction and agglomeration effects, resulting in a notable spatial division. (2) The direct impact coefficient of urban scale borrowing and the spatial weighting coefficient are significantly positive, indicating that urban scale borrowing improves agricultural green water use efficiency in both local and neighboring cities. (3) The spatial spillover results show that urban scale borrowing in the Central Plains Urban Agglomeration has a significant positive spillover effect on agricultural green water use efficiency, with the spillover effect being stronger than the rebound effect during the sample period. Furthermore, after conducting robustness checks, including changing the measurement model, replacing spatial matrices, and selecting samples, the conclusions remain valid. The spatial spillover effect of urban scale borrowing on agricultural green water use efficiency shows a decaying characteristic as geographic distance increases. When the distance between cities exceeds 210 km, the spatial spillover effect ceases to be significant. (4) The heterogeneity results show that with the introduction of the Central Plains Urban Agglomeration concept and the ongoing regional integration strategy, the spatial effects and policy synergies of urban scale borrowing on agricultural green water use efficiency have become more pronounced, particularly in highly urbanized areas.

Policy implications

To promote the diffusion and coordination of agricultural green water use efficiency in the Central Plains Urban Agglomeration, this study proposes the following measures:

First, promote the research and development of agricultural green water-saving technologies by increasing investment in personnel and funding from industry, academia, research, government, and enterprises, and fully utilize multi-channel and multi-platform research efforts.

Second, deepen the exchange and cooperation on green water-saving technologies within the urban agglomeration by establishing an agricultural technology promotion alliance, strengthening regular exchanges between agricultural technicians, launching pilot programs for green water-saving, and setting up specialized technology collaboration alliances at the urban agglomeration level.

Third, promote the free flow of green innovation elements within the urban agglomeration, optimize the allocation of green innovation resources on a larger scale, and guide the concentration of these resources toward major grain-growing areas to build water-saving irrigation agricultural zones.

Fourth, strengthen the assessment of climate risks in different regions, and develop differentiated water resource management and agricultural policies based on the region's climate vulnerability and environmental compatibility. For example, regions more severely affected by climate change should be prioritized for technical support and financial investment to enhance their disaster resilience and the sustainable use of agricultural water resources.

Fifth, promote the advanced construction and application of farmland irrigation infrastructure. Fully support the research, development, and deployment of agricultural water-saving irrigation equipment, and strengthen the construction and management of small-scale irrigation facilities.

Sixth, enhance policy support for green water-saving technologies and equipment by offering tax reductions, financial subsidies, or low-interest loans to foster agricultural green water-saving development. Encourage agricultural enterprises and farmers to adopt these technologies.

Seventh, establish a regional development model that integrates urbanization levels with improvements in agricultural green water use efficiency. Promote ongoing efforts in agricultural water resource management, technological innovation, and policy synergy in highly urbanized areas. Explore locally tailored green development paths and drive the synchronized development of surrounding less urbanized areas.

Limitations and future studies

This study on agricultural green water use efficiency in the Central Plains Urban Agglomeration has several limitations. First, the analysis focuses on the period from 2003 to 2023, which may limit the applicability of the findings to other time frames or regions. Future research should consider extending the temporal scope to capture long-term trends. Second, the input–output system is based on selected indicators, which may overlook other critical factors affecting agricultural sustainability. Future studies could incorporate additional indicators, such as climate change and land use, to provide a more comprehensive assessment.

Additionally, while this study identifies spatial spillover effects, it does not fully explore the mechanisms behind these effects. Future research could examine the interactions between urban and rural areas to better understand how urban functions influence agricultural water efficiency. Lastly, the policy recommendations are context-specific and may require further evaluation. Future studies could assess the effectiveness of these policies through case studies in different regions to inform more effective strategies. Addressing these limitations will enhance our understanding of agricultural water use efficiency and improve policy development.

This study was supported by the National Natural Science Foundation of China [72204099] and the Jiangsu University Philosophy and Social Science Outstanding Innovative Team Construction Project [SJSZ2020-20].

Data cannot be made publicly available; readers should contact the corresponding author for details.

The authors declare there is no conflict.

Alonso
W.
(
1971
)
The economics of urban size
,
Papers of the Regional Science Association
, 26,
67
83
.
Arellano
M.
&
Bover
O.
(
1995
)
Another look at the instrumental variable estimation of error-components models
,
Journal of Econometrics
,
68
,
29
51
.
Camagni
R.
,
Capello
R.
&
Caragliu
A.
(
2016
)
Static vs. dynamic agglomeration economies. Spatial context and structural evolution behind urban growth
,
Papers in Regional Science
,
95
,
133
159
.
Caves
D. W.
,
Christensen
L. R.
&
Diewert
W. E.
(
1982
)
The economic theory of index numbers and the measurement of input, output, and productivity
,
Econometrica: Journal of the Econometric Society
,
50
,
1393
1414
.
Charnes
A.
,
Cooper
W. W.
&
Rhodes
E.
(
1978
)
Measuring the efficiency of decision making units
,
European Journal of Operational Research
,
2
,
429
444
.
Cui
Y.
,
Han
Y.
&
Lv
N.
(
2020
)
Measurement of agricultural ecological efficiency based on the super-efficiency SBM model
,
Statistical Decision
,
36
,
87
90
.
Du
S.
,
Liang
Y.
,
Zhang
M.
,
An
X.
&
Wang
H.
(
2021
)
Spatial-temporal differentiation characteristics and influencing factors of agricultural water resource use efficiency in Hebei Province
,
Water-saving Irrigation
,
74
80
.
Elhorst
J. P.
(
2010
)
Applied spatial econometrics: raising the bar
,
Spatial Economic Analysis
,
5
,
9
28
.
Elhorst
J. P.
(
2014
)
Matlab software for spatial panels
,
International Regional Science Review
,
37
,
389
405
.
Fang
C.
&
Bu
B.
(
2004
)
Study on the urban competitiveness and expansion in the Hexi Corridor under water resource constraints
,
Geographical Science
,
24
,
513
521
.
Geng
X.
,
Zhang
X.
&
Song
Y.
(
2014
)
Empirical analysis of agricultural irrigation water efficiency and its influencing factors – based on the stochastic frontier production function and survey data of cotton farmers in Xinjiang
,
Journal of Natural Resources
,
29
,
934
943
.
Gu
R.
&
Le
H.
(
2023
)
The impact of borrowing scale of innovative elements on urban innovation capacity – A case study of the Yangtze River Delta urban agglomeration
,
Journal of Jiangsu University (Social Science Edition)
,
25
,
74
89
.
Hao
L.
&
Li
X.
(
2020
)
The impact of urban sprawl on regional productivity growth – based on the perspective of industrial dynamic agglomeration
,
Journal of Central South University (Social Science Edition)
,
26
,
21
31
.
Ioris
A. A.
,
Hunter
C.
&
Walker
S.
(
2008
)
The development and application of water management sustainability indicators in Brazil and Scotland
,
Journal of Environmental Management
,
88
,
1190
1201
.
Lesage
J.
&
Pace
R. K.
(
2009
)
Introduction to Spatial Econometrics
.
Boca Raton, FL
:
Chapman and Hall/CRC
.
Li
Q.
&
Gao
N.
(
2016
)
The ecological environmental effect of urban sprawl – based on the analysis of panel data from 34 large and medium-sized cities
,
China Population Science
,
2
,
58
67
.
Li
J.
&
Luo
N.
(
2016
)
The impact of urban scale on ecological efficiency and regional differences analysis
,
China Population, Resources and Environment
,
26
,
129
136
.
Li
S.
,
Cheng
J.
&
Wu
Q.
(
2008
)
Regional differences in China's water resource utilization efficiency
,
China Population, Resources and Environment
,
18
,
215
220
.
Li
Q.
,
Zhang
F.
,
Su
W.
,
Yang
Q.
,
Sun
C.
&
Wei
Z.
(
2022
)
Measurement of agricultural water use green efficiency and influencing factors in the Yangtze River Economic Belt – based on the super-efficiency EBM-Geodetector model
,
China Agricultural Resources and Regional Planning
,
43
,
40
52
.
Li
R.
,
Chen
Q.
,
Gao
X.
,
Meng
S.
,
Wei
G.
&
Liu
Y.
(
2024
)
The gradient conversion rule and influencing factors of urban green economic efficiency in the Yangtze River Economic Belt
,
Journal of Natural Resources
,
39
,
125
139
.
Liu
S.
&
Cao
J.
(
2023
)
Spatio-temporal differentiation and driving factors of agricultural water green efficiency in China
,
Bulletin of Soil and Water Conservation
,
43
,
346
357
.
Ma
J.
,
Tong
J.
,
Wang
H.
&
Wang
S.
(
2018
)
Spatial effect study of overall technical efficiency of agricultural water use in the Yangtze River Economic Belt
,
Resources and Environment in the Yangtze Basin
,
27
,
2757
2765
.
Meijers
E. J.
,
Burger
M. J.
&
Hoogerbrugge
M. M.
(
2016
)
Borrowing size in networks of cities: city size, network connectivity and metropolitan functions in Europe
,
Papers in regional science
,
95
,
181
199
.
Melo
P. C.
,
Graham
D. J.
&
Noland
R. B.
(
2009
)
A meta-analysis of estimates of urban agglomeration economies
,
Regional Science and Urban Economics
,
39
,
332
342
.
Ministry of Water Resources of the People's Republic of China
(
2022
)
China Water Resources Bulletin
. .
Shang
J.
&
Li
Q.
(
2023
)
Research on agricultural production efficiency under carbon emission constraints in the Yangtze River Economic Belt
,
Journal of Henan Agricultural University
,
57
,
1062
1074
.
Su
H.
&
Wei
H.
(
2013
)
Density effect, optimal urban population density, and intensive urbanization
,
China Industrial Economics
,
10
,
5
17
.
Su
H.
,
Wei
H.
&
Deng
M.
(
2014
)
The multidimensionality of urban agglomeration economies and its empirical verification
,
Finance and Trade Economics
,
35
,
115
126
.
Sun
D.
,
Yuan
Y.
&
Guo
R.
(
2022
)
The impact of multi-agent collaborative innovation on regional industrial upgrading – Based on the perspective of spatial spillover
,
Science Research Management
,
43
,
154
163
.
Tone
K.
&
Tsutsui
M.
(
2010
)
Dynamic DEA: a slacks-based measure approach
,
Omega
,
38
,
145
156
.
Vega-Carrillo
H. R.
,
Esparza-Garcia
I. R.
&
Sanchez
A.
(
2015
)
Features of a subcritical nuclear reactor
,
Annals of Nuclear Energy
,
75
,
101
106
.
Wang
X.
(
2010
)
Economic analysis of urbanization paths and urban scale in China
,
Economic Research
,
45
,
20
32
.
Wang
F.
&
Qiu
L.
(
2015
)
Study on the functional cognition of urban green open space system – a case study of Lianyungang city
,
Geographical Science
,
35
,
583
592
.
Wei
J.
,
Lei
Y.
,
Yao
H.
,
Ge
J.
,
Wu
S.
&
Liu
L.
(
2021
)
Estimation and influencing factors of agricultural water efficiency in the Yellow River basin, China
,
Journal of Cleaner Production
,
308
,
127249
.
Yang
Q.
,
Wu
R.
&
Wang
H.
(
2017
)
Distribution pattern and spatial interaction of agricultural water use efficiency in China: 1998–2013
,
Quantitative Economic and Technical Economic Research
,
34
,
72
88
.
Yu
F.
,
Wang
G.
&
Lin
S.
(
2022
)
Key issues and path choices for agricultural green development in the main grain-producing areas
,
Chongqing Social Sciences
,
34
,
6
18
.
Yuan
Y.
&
Chen
Z.
(
2019
)
Environmental regulation, green technological innovation, and the transformation and upgrading of China's manufacturing industry
,
Studies in Science of Science
,
37
,
1902
1911
.
Yuan
Y.
&
Gao
K.
(
2020
)
Industrial collaborative agglomeration, spatial knowledge spillover, and regional innovation efficiency
,
Studies in Science of Science
,
38
,
1966
1975
.
+ 2007
.
Zhan
Y.
&
Li
S.
(
2022
)
Smart city construction, entrepreneurial vitality, and high-quality economic development – analysis from the perspective of green total factor productivity
,
Financial and Economic Research
,
48
,
4
18
.
Zhang
Z.
&
Zhong
R.
(
2023
)
Digital economy, green technological innovation, and urban low-carbon transformation
,
China Circulation Economics
,
37
,
60
70
.
Zhao
J.
,
Meng
H.
&
Gong
J.
(
2017
)
Analysis of agricultural total factor water use efficiency and influencing factors in the Beijing-Tianjin-Hebei region
,
Journal of China Agricultural University
,
22
,
76
84
.
Zhao
X.
,
Chen
L.
&
Liu
C.
(
2021
)
Can informal environmental regulations induce green innovation?—Verification based on the perspective of ENGOs
,
China Population, Resources and Environment
,
31
,
87
95
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).