The Impact and Mechanism of Digital Villages on Agricultural Resilience in Ecologically Fragile Ethnic Areas: Evidence from China’s Provinces

Abstract

Ecologically fragile ethnic areas constitute pivotal regions for rural revitalization and the construction of the Beautiful China initiative. The establishment of digital villages is of great significance for enhancing agricultural resilience and achieving common prosperity. Utilizing panel data from ecologically fragile ethnic areas between 2010 and 2020, this study employs a projection pursuit model to calculate scores for digital village levels and agricultural resilience. Building on this, our research employs instrumental variable methods and a mediation effect model to analyze the impact and mechanisms of digital village level on agricultural resilience in ecologically fragile ethnic areas, and heterogeneity analysis is conducted. The findings reveal that digital village level and agricultural resilience scores in ecologically fragile ethnic areas show a trend of initial increase followed by a decrease, exhibiting significant overall fluctuations and unstable growth. The promotion of digital village areas on agricultural resilience is evident, and this conclusion holds even after a series of tests including instrumental variables. Elevating the digital village level can narrow the urban–rural income gap and enhance agricultural resilience. There is significant regional heterogeneity in the impact of digital village levels on local agricultural resilience in ecologically fragile ethnic areas, with digital village development exerting a more pronounced and powerful driving force in areas with lower agricultural resilience. Therefore, leveraging the leadership of advantaged provinces, intensifying investment in digital village infrastructure, and implementing targeted strategies based on the disparities in digital village level and agricultural resilience across areas become imperative.

1. Introduction

With the rapid diffusion and deep penetration of scientific and technological advancements such as cloud computing, big data, and 5G, China’s digital economy reached RMB 50.2 trillion in 2022, accounting for 41.5% of the GDP. The digital economy has become a new driver for stable economic growth and transformative development. The continuous integration and widespread application of digitization and information technology in the agricultural sector have prompted China to proactively propose a development strategy for constructing digital village areas. The Central Document No. 1 of China in 2023 emphasizes the promotion of a resilient agricultural powerhouse with strong industrial capabilities, highlighting the crucial role of agricultural resilience in the process of high-quality agricultural development. The intensive deployment and concentrated efforts of digital village policies are increasingly intertwined with the resilience of rural areas. For ecologically fragile ethnic areas, as a key area for rural revitalization strategy and the Beautiful China initiative construction, due to deep long-term poverty, complex natural environments, and a weak economic foundation, the foundation for regional agriculture to become rich and increase income is weak, the utilization of resources is not sufficient, and agricultural resilience needs to be strengthened. Therefore, the broad proposal and implementation of digital village policies hold immense significance in solidifying the economic development foundation of ecologically fragile ethnic areas, leveraging resource endowment conditions, enhancing agricultural resilience, and promoting shared prosperity in ethnic areas.

Resilience originates from physical concepts, gradually extended to the field of economics by Gary and Huggins [1,2]. Economic resilience is the ability of an economic system to resist, reconstruct, and transform in the face of external shocks [3], that is, the ability of a regional economy to maintain its original operating state and the extent to which it can recover after external shocks [4]. The study of economic resilience can be divided into macro-level analysis, dynamically examining various factors such as the inherent endowment, institutional arrangements, and monetary cycles concerning China’s economic resilience, exploring effective development paths to enhance macroeconomic resilience [5,6]; meso-level analysis, which focuses on analyzing the spatial differentiation pattern and spatiotemporal evolution mechanism of regional economic resilience [7], as well as the influencing factors of regional economic resilience [8]; and micro-level research, which primarily concentrates on the resilience measurement of households [9] and individual economic resilience recovery [10]. Regarding agricultural resilience, there is a trend in China where agricultural resilience is continuously optimized at the provincial level while inter-provincial disparities are widening [11]. This is primarily due to regional disparities and the presence of obstacles and constraining factors within economic and social domains [12]. The development of agricultural resilience is influenced by factors such as market volatility, farmers’ consumption levels, and the rural industrial structure [13,14,15]. It is crucial to focus on the early warning and monitoring capabilities of potential natural and market risks in agriculture as well as to enhance the level of agricultural specialization to strengthen agricultural resilience [16]. The research on agricultural resilience has yielded substantial results, yet there remains a research gap in assessing agricultural resilience in ecologically vulnerable and remote areas.

The development of digital villages is essentially a tangible manifestation of the leap in rural productivity, a necessary response to the imperative of China’s unique modernization. At the current stage, there are deviations in the practice of constructing digital villages in rural areas regarding their core principles and underlying logic [17]. Issues such as low efficiency in information utilization, outdated digital infrastructure [18], a less optimistic outlook on the digital divide, and challenges in overcoming technological barriers [19] are significantly impeding the realization of agricultural industry revitalization and common prosperity [20]. In the study of digital villages, one aspect involves exploring the development paths of digital villages. This is achieved by optimizing digital village governance, enhancing farmers’ digital literacy, and integrating agricultural industries to elevate the level of digital villages [21,22,23]. Therefore, the rapid dissemination and extensive application of digital villages are of great significance in enhancing agricultural resilience and tapping into their potential in ecologically fragile ethnic areas.

Ecologically fragile ethnic areas in China represent regions with a complex and diverse ecosystem, characterized by poor structural stability, relative sensitivity to environmental and biological factors, and weak ecological resilience to external impacts. These areas are not only deeply impoverished regions in China failing to achieve comprehensive well-off status but are also concentrated and contiguous areas facing particular difficulties in transitioning from absolute poverty to relative poverty. Therefore, they are crucial areas requiring exploration to achieve common prosperity for all people in the process of transitioning from absolute poverty to relative poverty in the quest for comprehensive well-off status. Current research on ecologically fragile ethnic areas is predominantly focused on two aspects. On one hand, there is a concentration on the realization of ecological product values in the regions [24], analysis of environmental protection mechanisms [25], and exploration of the evolution of ecosystems and contributing factors [26]. On the other hand, there is research and discussion on the micro-level vulnerability of farming households [27], livelihood coping strategies [28], and behavioral preferences [29].

Through literature review, it is found that existing research may have the following shortcomings: Firstly, current research rarely constructs and calculates indicators to assess the agricultural resilience of ecologically fragile ethnic areas from a macro perspective. Studies on the impact of digital villages mainly focus on regional and provincial differences, lacking attention to specific areas. Therefore, there is an urgent need to explore how to effectively evaluate agricultural resilience and digital villages in ecologically fragile ethnic areas at the macro level. Secondly, there is a lack of empirical research on the impact of digital villages on agricultural resilience, particularly an in-depth exploration of the transmission mechanisms of how the level of digital villages affects the agricultural resilience of ecologically fragile ethnic areas. This indicates the need to explore more potential transmission paths to calculate and analyze the specific impact of digital village levels on agricultural resilience in specific areas. Thirdly, previous works in the literature have not considered the widespread distribution of ecologically fragile ethnic areas in China, with significant differences in social conditions, natural resources, and economic conditions across areas. Therefore, it is necessary to deeply analyze the impact of the level of digital villages on agricultural resilience in different areas as well as the heterogeneity characteristics and trends of these impacts.

Therefore, the potential marginal contribution of this study lies in constructing a set of agricultural resilience evaluation indicators with Chinese characteristics for ecologically fragile ethnic areas in China. This system not only considers the specific ecological and economic differences in the region but also integrates a practical situation of rural development in China. In addition, this study utilizes provincial panel data from 2010 to 2020 and applies the innovative method (the projection pursuit model) for a more precise estimation of the digital village level and agricultural resilience scores in the investigated areas. This method has been rarely used in previous research, thus demonstrating significant innovation. Furthermore, this study delves into the specific pathways through which digital villages impact agricultural resilience, an area that has been relatively less explored in the existing literature. Through this analysis, at the theoretical level, this study not only enriches the theoretical foundation of the impact of digital villages on agricultural resilience but also effectively supplements and extends the existing literature in terms of methodology and theoretical innovation. On a practical level, this study provides important pathway support and practical guidance for the implementation of rural revitalization strategy and the achievement of shared prosperity goals in ecologically fragile ethnic areas of China. This study utilizes provincial panel data from ecologically fragile ethnic areas from 2010 to 2020. Through optimization calculations of the projection tracking model, we evaluate the digital village levels and agricultural resilience scores of each province and city. Furthermore, we conduct an in-depth analysis of the impact pathways of digital villages on agricultural resilience, providing theoretical support for promoting rural revitalization and achieving common prosperity in ecologically fragile ethnic areas of China.

2. Theoretical Analysis and Research Hypotheses

2.1. The Direct Effect of Digital Village Levels on Agricultural Resilience

The digital economy, through the layout of infrastructure, the application of data elements, and the diffusion of digital technologies, promotes the resource utilization efficiency of traditional industries, reduces production and transaction costs, and forms superimposed effects, diffusion effects, and penetration effects. With the development foundation and economic benefits of the digital economy, digital village areas can provide a solid foundation for the diffusion of agricultural and rural technologies. The key elements of digital village areas lie in data. By leveraging the characteristics and agility of data iteration, on the one hand, it can reduce the entry barriers for agricultural technology, accelerating diffusion and application speed; on the other hand, it can optimize the structure of agricultural factor allocation, alleviate rural information asymmetry, enhance farmers’ digital literacy [22], and promote more rational agricultural production decisions when facing natural and market risks. The development of agricultural resilience entails comprehensive improvement in rural areas and the enhancement of all elements in agriculture. It requires the establishment of rational and scientific measurement indicators, well-targeted planning of development paths, and the ability to address potential risks from internal and external shocks. This not only helps to withstand cyclical market fluctuations and impacts from seasonal changes in agriculture but also strengthens farmers’ risk prediction capabilities and elevates the level of agricultural specialization [16]. Moreover, it enhances the endogenous development momentum of agriculture, supports the digitization and green development of agriculture, drives the transformation of agricultural development, and breaks through the new growth dilemma in Chinese agriculture [30]. Therefore, the elevation of digital village levels involves the rapid penetration of digital technology into vast rural areas. It harnesses data to aggregate key agricultural production factors, unleashing economic effects from resource endowments and forming a virtuous causal loop with cumulative effects [31]. This enhances the resistance to shocks, adaptability to potential crises, and transformative nature in the face of fluctuating developments in the agricultural sector. Therefore, this study proposes the following research hypothesis:

H1. 

The enhancement of digital village levels can efficiently stimulate agricultural resilience in ecologically fragile ethnic areas.

2.2. Analysis of the Mechanisms by Which Digital Village Levels Affect Agricultural Resilience

From a market perspective, on one hand, the construction and establishment of the digital economy in rural areas have accelerated the transformation of agricultural productivity and enhanced resource allocation efficiency. This has optimized the rural labor structure, reduced agricultural labor demand, and released the rural labor force population. On the other hand, through the penetration effect, digital village areas provide accurate transaction information and professional platforms for the sale of agricultural products. This alleviates the lag in decision making and blind production caused by information barriers among farmers. The diffusion effect can rapidly drive the emergence and outward spread of new agricultural product sales models, such as e-commerce and online sales, thereby reducing production costs, improving economic efficiency, and addressing the supply–demand instability caused by market and natural risks in agriculture. From the perspective of individual farmers, the development of agricultural digitization represents a technological revolution and transformative development in market economy, production technology, ecological environment, and resource utilization. Firstly, the implementation of inclusive financial policies in digital finance reduces the time and transaction costs of financing for rural operators, enhances their production enthusiasm, and improves the efficiency of agricultural production resource allocation, thereby promoting the development of agricultural resilience. Secondly, with the continuous popularization of digital village infrastructure, the urban–rural digital divide is further narrowed, and the Matthew Effect of the internet is broken. This provides a solid foundation for the emerging development of rural digitalization and informatization, triggering a positive feedback mechanism of the digital economy and achieving explosive growth in the network value of the rural digital economy [32], thereby promoting an increase in farmers’ income. Thirdly, the transformation brought about by the digital economy, including the development of green and intensive models and technological dividends, can bring long-term foreseeable income growth for farmers. This, in turn, increases the long-term marginal consumption tendency of households, promotes rural consumption upgrading, and enhances agricultural resilience. Therefore, the following hypothesis is proposed:

H2. 

The elevation of digital village levels may diminish the urban–rural income disparity, subsequently influencing the enhancement of agricultural resilience.

3. Materials and Methods

3.1. Index Construction

3.1.1. Evaluation Index System of Agricultural Resilience in Ecologically Fragile Ethnic Areas

Different scholars have different criteria for the evaluation index system of agricultural resilience. It can be specifically divided into constructing a multidimensional index system to measure economic resilience [11] and selecting core indicator variables with high sensitivity to economic fluctuations to measure the impact of external forces on economic systems [33]. Based on the general characteristics and characteristic endowments of the evolution of agricultural resilience in ecologically fragile ethnic areas and the basis of the research of Jiang [34] and Yu [35], this study constructs an evaluation index system of agricultural resilience from three dimensions: resistance, adaptability, and transformability. Resilience refers to the ability of the regional agricultural system to reduce losses when facing destructive impacts from external forces, with sub-dimensions of intrinsic stability and production–supply robustness. Adaptability reflects the agricultural system’s ability to recover and reconstruct after encountering dual risks from the market and nature, encompassing sustainability and recoverability as sub-dimensions. Transformability characterizes the self-adjustment, transformation, and sustained development of the agricultural system after the adaptation phase following impacts, including dimensions of diverse collaboration and technological advancement. Specifics of each indicator can be found in

Table 1. Evaluation indicator system for agricultural resilience in ecologically fragile ethnic regions.

First-Level Indicators	Second-Level Indicators	Third-Level Indicators	Indicator Explanation	Indicator Direction
Resilience	Intrinsic Stability	Arable Land Area (thousand hectares)	Direct data	+
		Effective Irrigation Rate (%)	Effective irrigation area/arable land area	+
		Rural Primary Industry Employment (ten thousand people)	Direct data	+
		Rural Residents’ Engel Coefficient (%)	Direct data	−
	Production–Supply Robustness	Grain Output (ten thousand tons)	Direct data	+
		Proportion of Town and Village Retail Sales of Consumer Goods to Total Social Retail Sales (%)	Direct data	+
		Agricultural Production Material Price Index (%)	Direct data	−
		Grain Yield per Hectare (kg/hectare)	Grain output/seeded area	+
Adaptability	Sustainability	Intensity of Fertilizer Use (kg/hectare)	Pure quantity of agricultural fertilizer/reseeded area	−
		Forest Coverage (%)	Rural forest area/land area	+
		Disaster Rate (%)	Crop disaster area/seeded area	−
		Agricultural Plastic Film Usage (ten thousand tons)	Direct data	−
	Recoverability	Replanting Index (%)	Total area of crops planted throughout the year/total cultivated land area	+
		Rural Household Per Capita Disposable Income (RMB/person)	Direct data	+
		Rural Residents’ Per Capita Consumption Expenditure (RMB/person)	Direct data	+
		Agricultural Value-Added Growth Rate (%)	Direct data	+
Transformability	Diverse Collaboration	Rural Electricity Consumption (billion kWh)	Direct data	+
		Total Power of Agricultural Machinery (ten thousand kW)	Direct data	+
		Agricultural Insurance Amount (billion RMB)	Direct data	+
		Value Added of Agriculture, Forestry, Animal Husbandry, and Fishery Services (billion RMB)	Direct data	+
	Technological Advancement	Agricultural Research Expenditure (billion RMB)	Estimated by formula	+
		Agricultural Fixed Investment (billion RMB)	Direct data	+
		Financial Expenditure on Agriculture, Forestry, and Water (billion RMB)	Direct data	+
		Agricultural Technical Personnel in Public Economic Entities (ten thousand people)	Direct data	+

.

3.1.2. Digital Village Level Evaluation Indicator System

Scholars have different approaches to constructing evaluation indicators for the development of digital villages. Some focus on six aspects, developmental environment, information infrastructure, human resources, technical support, green development, and industrial benefits, to build a digital agriculture index system [36]. Others consider the essence of the rural digital economy and begin by focusing on the three basic elements of digital villages to establish an index system for the agricultural and rural digital economy [37]. Furthermore, some construct a comprehensive evaluation index for digital villages from the perspectives of infrastructure, economy, governance, and lifestyle [38]. This study, referencing the research of Wang [39], Jin and Ren [40], and Sun et al. [41], utilizes the projection pursuit model to measure the level of digital village development from four dimensions: inclusive finance in digital village areas, digital trade and services in rural areas, investment in digital services and IoT technology applications in rural areas, and the construction of digital information infrastructure in rural areas (

Table 2. Digital village level evaluation indicator system.

First-Level Indicators	Second-Level Indicators	Third-Level Indicators	Indicator Explanation
Rural Digitization Inclusive Finance	Rural Inclusive Digital Finance Index	The Peking University Digital Financial Inclusion Index of China	+
	Rural Inclusive Digital Finance Digitalization Level	Digitalization Level Index	+
Rural Digitization Trade and Services	E-commerce Sales Volume (billion RMB)	Direct Data	+
	E-commerce Procurement Total (billion RMB)	Direct Data	+
	Rural Digital Bases (count)	Taobao Village Quantity	+
Rural Digitization Services and IoT Technology Application Investment	Rural Transportation, Warehousing, and Postal Industry Fixed-Asset Investment (billion RMB)	Direct Data	+
	Rural Residents’ Per Capita Expenditure on Transportation and Communication (RMB/person)	Direct Data	+
	Rural Postal Route Length (ten thousand kilometers)	Direct Data	+
Rural Digital Information Infrastructure Construction	Average Mobile Phones per Hundred Households in Rural Residents’ Families (units/hundred households)	Direct Data	+
	Average Computers per Hundred Households in Rural Residents’ Families (units/hundred households)	Direct Data	+
	Average Televisions per Hundred Households in Rural Residents’ Families (units/hundred households)	Direct Data	+
	Rural Internet Broadband Access Households (ten thousand households)	Direct Data	+

). Among them, inclusive finance in digital village areas mainly reflects the application degree of inclusive digital financial infrastructure in rural areas as well as the in-depth development of new digital financial service models in rural areas. Rural digitization trade and services mainly reflect the penetration and popularization of digital technology in the agricultural industry as well as the infrastructure and total sales and purchases of agricultural products in the e-commerce sector. Rural digitization services and IoT (Internet of Things) technology application investment mainly reflect the level of upgrading in communication, express delivery, transportation, and other aspects in the development process of digital villages as well as rural areas and farmers. Rural digital information infrastructure construction mainly represents the material carrier for achieving rural digitization. Through communication equipment, it reflects the level of digital infrastructure, the utilization of digital resources, and the degree of popularization of digital technology in rural areas.

3.1.3. Calculation Methodology

This study employs a projection pursuit model to assign weights, uses MATLAB R2016 to estimate, and calculate the evaluation indicator system for agricultural resilience in ecologically fragile ethnic areas and the digital village level assessment indicator system. Proposed by Kruskal in 1972, this model is a novel mathematical and statistical approach for handling non-linear, non-normal high-dimensional data. The fundamental idea involves projecting high-dimensional data into a lower-dimensional subspace. Subsequently, the analysis of the high-dimensional data’s structure in the lower-dimensional space is conducted based on the projected values that reflect the structural characteristics of the high-dimensional data. This approach facilitates the research objectives concerning the original high-dimensional data. In practical applications, the algorithm for finding the optimal projection direction becomes crucial regarding the effectiveness of the projection pursuit model. The use of genetic algorithm-based projection pursuit models is widespread, representing a method well suited for multidimensional global optimization. The genetic algorithm simulates the process of biological evolution, preventing traditional optimization methods from getting stuck in local optima, premature convergence, or premature maturation. The algorithm can directly find the optimal solution for the projection index function. Therefore, this study utilizes a projection pursuit model based on genetic algorithms for calculation. The basic steps are as follows:

Standardization of indicators. Consider a sample set $\left\{x^{\ast }(i,j)\right.|i=1,2,\cdot \cdot \cdot ,n;j=1,2,\cdot \cdot \cdot ,\left.p\right\}$ for the indicator system, where $x^{\ast }(i,j)$ is the $j$-th indicator value of the $i$-th sample. The values of $n,p$ represent the number of samples and the number of indicators, respectively. To eliminate dimensional impact, Formula (1) is used to normalize the data for each indicator.

$$\left\{\begin{array}{l}{x}^{\ast }(i,j)=\frac{x^{\ast }(i,j)-x_{\min }(j)}{x_{\max }(j)-x_{\min }(j)} \mathrm{For} \mathrm{positive} \mathrm{indicators}\\ x^{\ast }(i,j)=\frac{x_{\max }(j)-x^{\ast }(i,j)}{x_{\max }(j)-x_{\min }(j)} \mathrm{For} \mathrm{negative} \mathrm{indicators}\end{array}\right.$$

(1)

Construct the objective function $Q(a)$ for projection. Aggregate $\left\{x^{\ast }(i,j)\right.|i=1,2,\cdot \cdot \cdot ,n;j=1,2,\cdot \cdot \cdot ,\left.p\right\}$ dimensions of $p$ into a one-dimensional projection value $z(i)$ along the direction of $a=\left\{a(1),a(2),a(3),\cdot \cdot \cdot ,\left.a(p)\right\}\right.$.

$$z(i)={\displaystyle \sum _{j=1}^{p}a(j)x(i,j)} (i=1,2,\cdot \cdot \cdot ,n)$$

(2)

$$Q(a)=S_z{D}_z$$

(3)

where $S_Z$ is the standard deviation of the projection values $z(i)$, and $D_Z$ is the local density of the projection values $z(i)$, namely:

$$S_z=\sqrt{\frac{{\displaystyle \sum _{i=1}^n{\left[z(i)-E(z)\right]}^2}}{n-1}}$$

(4)

$$D_z={\displaystyle \sum _{i=1}^n{\displaystyle \sum _{j=1}^p\left[R-r(i,j)\right]}u\left[R-r(i,j)\right]}$$

(5)

where $E(z)$ is the mean value of sequence $\left\{z(i)|i=1,2,\cdot \cdot \cdot ,\left.n\right\}\right.$; $R$ is the window radius of local density, typically taking the value $0.1S_z$; $r(i,j)$ represents the distance between samples; $u(t)$ is a unit step function with a value of 0 when $t<0$ and 1 otherwise.

Optimize the projection objective function. When the set of values for each indicator is given, the projection objective function $Q(a)$ changes with the variation in the projection direction $a$, aiming to find the optimal projection direction. In this study, to address the high-dimensional global optimization problem, an accelerated genetic algorithm based on real number encoding is employed, incorporating simulated biological survival of the fittest and intra-population chromosome information exchange mechanisms. The formula for solving the optimal projection direction is:

$$\mathrm{Max}:{Q\left(a\right)=S}_z{D}_z$$

(6)

$$s.t:{\displaystyle \sum _{j=1}^p{a}^2(j)=1}$$

(7)

Classification and evaluation: To calculate the projection values $z^{\ast }(i)$ for each sample point, it is necessary to substitute the optimal projection direction $a^{\ast }$ into Equation (2). Finally, by arranging $z^{\ast }(i)$ in descending order, the ranking of sample quality can be obtained.

3.2. Model Specification

3.2.1. Benchmark Regression Model

To analyze the factors influencing agricultural resilience in ecologically fragile ethnic areas by the level of digital village development, this study establishes the following benchmark regression model:

$$AgriRes{i}_{i,t}={\beta }_0+{\beta }_{1}AgriDig{i}_{i,t}+{\beta }_{2}Control{s}_{i,t}+{\displaystyle \sum Pro}+{\displaystyle \sum Year+{\epsilon }_{i,t}}$$

(8)

3.2.2. Mediation Model

This study refers to the research findings of Wen et al. [42] to construct the following mediation model, aiming to examine the direct and indirect effects of the level of digital village development on agricultural resilience in ecologically fragile ethnic areas.

$$Med{i}_{i,t}={\alpha }_0+{\alpha }_{1}AgriDig{i}_{i,t}+{\alpha }_{2}Control{s}_{i,t}+{\displaystyle \sum Pro}+{\displaystyle \sum Year+{\epsilon }_{i,t}}$$

(9)

$$\begin{array}{l}AgriRes{i}_{i,t}={\delta }_0+{\delta }_{1}Med{i}_{i,t}+{\delta }_{2}AgriDig{i}_{i,t}+\\ {\delta }_{3}Control{s}_{i,t}+{\displaystyle \sum Pro}+{\displaystyle \sum Year+{\epsilon }_{i,t}}\end{array}$$

(10)

where $AgriRes{i}_{i,t}$ represents the agricultural resilience of Province $i$ in the $t$-th year, $AgriDigi$ represents the score of digital village development, $\epsilon$ is a random disturbance term, and the model controls for fixed effects of province and year. $Controls$ represents various control variables, and in this study, based on references from the relevant literature [43,44,45], the following variables are controlled for. Among these, the degree of rural aging is measured using the elderly dependency ratio in rural areas; crop planting diversity is assessed using the proportion of the total output of three major grain crops in each province and city, calculated using the Shannon–Wiener diversity index (the formula for the Shannon–Wiener diversity index is given by $H^{\prime }=-{\displaystyle \sum \left[{(P_i)}^{\ast }(\ln P_i)\right]}$, where $P_i$ represents the proportion of the three major grain crops, namely corn, wheat, and rice, in the total grain production of each province and city in the current year); urban–rural income gap is measured using the Theil index; rural human capital stock is measured based on the actual level of rural human capital; the level of industrialization is represented by the proportion of regional industrial value added to the regional gross domestic product in the current year; and the level of rural renewable resource utilization is measured using the total gas production from rural biogas tanks. Due to the relatively large values of actual rural human capital and biogas tank gas production, a natural logarithm transformation is applied to alleviate heteroscedasticity issues. This study uses StataMP 17 software to estimate the benchmark regression model and the mediation model.

3.3. Data Source

This study selects ethnic regions including Xinjiang, Tibet, Ningxia, Inner Mongolia, Guangxi, Qinghai, Yunnan, Guizhou, Gansu, Sichuan, and Chongqing as research samples based on the overlapping areas of key ecologically fragile regions [46] and the narrow concept of ethnic regions [47]. The sample period is set from 2010 to 2020. The aforementioned region is widely populated by ethnic minority communities, and there are provincial and municipal-level administrative departments with ethnic autonomy rights. The ethnic minority population in this region accounts for 68.95% of China’s total ethnic minority population. The area encompasses various ecologically vulnerable terrains in China, including the northern sandy area, the Qinghai–Tibet cold, high-altitude area, the arid desert area, the Loess Plateau, and the karst area. It is a typical ecological key protection and governance area in China. Seven of China’s fourteen concentrated contiguous and deep-poverty areas are located in the region. In 2010, the region accounted for nearly 40% of the poor population, and in 2020, the contribution of the poverty-stricken population reached 50%. It can be found that the selected area, as a typical ecologically fragile ethnic area, not only has a fragile ecological environment but also has a huge labor force for poverty alleviation. Therefore, it is a typical area in which to study the impact of digital village construction on ecologically fragile ethnic areas, preventing poverty return and enhancing the resilience of the agricultural industry.

The data for sample variables are sourced from the “China Statistical Yearbook”, the “China Rural Statistical Yearbook”, annual statistical yearbooks of various provinces, the EPS database, “The Peking University Digital Financial Inclusion Index of China” [48], and “Human Capital in China 2022” (published by the China Center for Human Capital and Labor Market Research (CHLR), http://humancapital.cufe.edu.cn/rlzbzsxm.htm, accessed on 2 January 2024). Some variables have a small amount of missing data; therefore, interpolation methods and annual average growth rates are employed for data imputation. For the variable of agricultural research expenditure, data are not directly available for each province. Hence, this study refers to the estimation method proposed by Hao and Tan [49] to estimate the data for this variable (

Table 3. Variable setting and descriptive statistics.

Variable	Mean	Standard Deviation	Minimum	Median	Maximum
Agricultural Resilience	0.457	0.152	0.209	0.461	0.801
Digital Village Level	0.263	0.165	0.026	0.219	0.865
Urban–Rural Income Disparity	0.129	0.036	0.063	0.125	0.210
Rural Renewable Resource Utilization Level	10.061	1.567	6.533	10.482	12.518
Degree of Rural Population Aging	0.157	0.074	0.071	0.143	0.446
Rural Human Capital Stock	11.844	0.344	11.166	11.891	12.538
Industrialization Level	0.320	0.101	0.068	0.327	0.495
Crop Planting Diversity	0.731	0.170	0.404	0.806	0.996

).

4. Results

4.1. The Spatiotemporal Evolution Characteristics of Agricultural Resilience and Digital Village Level

Using the projection pursuit model, this study calculates the agricultural resilience and digitalization level of rural areas in ecologically fragile ethnic areas between 2010 and 2020 and examines their evolutionary traits and development patterns from both temporal and spatial perspectives.

4.1.1. The Spatiotemporal Evolution Characteristics of Agricultural Resilience

As shown in Figure 1, the agricultural resilience of ecologically fragile ethnic areas generally shows a trend of rising first and then falling. The average score of agricultural resilience increased from 0.430 in 2010 to 0.482 in 2015 and then decreased to 0.440 in 2020, with an overall growth of 2.33%. From the perspective of the level of development between regions, the provinces with an average score of agricultural resilience above 0.5 are mainly concentrated in the southern region, accounting for 75%; the score of agricultural resilience in the northern region is generally lower than that in the southern region. The highest average score of agricultural resilience in the northern ecologically fragile ethnic areas is for Inner Mongolia, reaching 0.594, while the highest score in the south is for Sichuan, with an average score of agricultural resilience of 0.768. In regard to provincial breakdown, the top three provinces in terms of average agricultural resilience scores are Sichuan, Inner Mongolia, and Guangxi. In 2010, the top three provinces regarding agricultural resilience scores were Sichuan, Inner Mongolia, and Yunnan, while in 2020, they changed to Sichuan, Guangxi, and Yunnan. It is worth noting that Ningxia, Inner Mongolia, and Qinghai show a slightly decreasing trend in agricultural resilience scores. Additionally, the average agricultural resilience scores of Tibet, Ningxia, Qinghai, Gansu, and Chongqing are below the average level of agricultural resilience scores in ecologically fragile ethnic areas.

4.1.2. The Spatiotemporal Evolution Characteristics of Digital Village Level

Taking a holistic view (Figure 2), the digital village level and agricultural resilience in ecologically fragile ethnic areas showed a similar trend of rising first and then falling, increasing from 0.369 in 2010 to 0.424 in 2014, reaching a peak and then falling to 0.358 in 2020. From the perspective of regional digital village level scores, average scores above 0.5 are mainly concentrated in the southern regions. The digital village level scores in northern ecologically fragile ethnic areas are lower than those in southern regions, indicating a certain gap in the digital village development level between the north and south. Among the northern provinces, Inner Mongolia has the highest average score, reaching 0.511, while Tibet has the lowest average score at 0.155. In the southern region, Sichuan has the highest average score, reaching 0.717, and Guizhou has the lowest average score at 0.278. At the provincial level, in 2010, the top three provinces in digital village level scores were Sichuan, Yunnan, and Guangxi. By 2020, it had changed to Sichuan, Chongqing, and Guangxi. However, the average digital village level scores in Tibet, Ningxia, Qinghai, Guizhou, and Gansu are below the overall average. With the exception of Tibet, the other four provinces demonstrate a decreasing trend in digital village level scores.

4.2. Benchmark Regression

Before estimating the model, a multicollinearity test was conducted on the core explanatory variables and control variables. The results revealed VIF values ranging from 1.16 to 2.36, indicating the absence of significant multicollinearity issues among the variables.

As shown in

Table 4. Benchmark regression results.

Variable	Agricultural Resilience
	Ordinary OLS	FE	Panel Tobit
Digital Village Level	0.6941 ***	0.0994 ***	0.5563 ***
	(12.488)	(2.930)	(7.294)
Rural Renewable Resource Utilization Level		0.0276 ***	0.0529 ***
		(3.846)	(8.465)
Degree of Rural Population Aging		−0.1667	−0.5080 *
		(−1.414)	(−1.742)
Rural Human Capital Stock		0.0346	0.0671
		(0.605)	(1.421)
Industrialization Level		0.3978 ***	0.2266 **
		(3.910)	(2.704)
Crop Planting Diversity		0.2441 **	−0.1600 ***
		(3.014)	(−6.111)
Provincial/Year Fixed Effects	NO	YES	YES
Observations	121	121	121
R2	0.567	0.972	—
Adj. R2	0.564	0.963	—

Note: Robust standard errors are in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1.

, this study initially conducted a preliminary regression test on digital village levels and agricultural resilience using a simple ordinary least squares (OLS) model, without incorporating control variables and considering province- and time-fixed effects. It can be observed that the digital village level passes the significance test at the 1% level and has a significant positive impact on agricultural resilience. Subsequently, based on the results of the Hausman test, this study employed a fixed-effects model for a reexamination of the relationship between digital village level and agricultural resilience. The results indicate that the direction of the impact of digital villages on agricultural resilience did not undergo significant changes. Finally, since the agricultural resilience score is between 0 and 1, which indicates truncated data, the panel Tobit model can be used; however, the likelihood method has great difficulties in estimating the fixed effect in the panel Tobit model, and the semi-parametric estimation method [50] can effectively estimate the Tobit model of panel fixed effects through the Stata command “pantob” without assuming the specific setting of the residual. In the subsequent analysis, incorporating control variables and fixing province and time effects, this study employed a panel Tobit model. The results show that the significance of rural digitalization remains evident at the 1% level, and the direction of its impact has not changed. Regarding control variables, in the fixed-effects model, rural renewable resource utilization, industrialization level, and crop diversity in planting are significantly positive. In the panel Tobit model, rural renewable resource utilization and industrialization level are significantly positive, while the degree of rural population aging and crop planting diversity are significantly negative. However, these variables do not affect the significance of the impact of digital villages on agricultural resilience and the sign of the estimated coefficients. Therefore, digital village level has a direct effect on agricultural resilience, confirming the expected hypothesis H1.

4.3. Mediation Effect

Table 5. Mediation effect test results.

Variable	Benchmark Model	Mediation Model
	Agricultural Resilience	Urban–Rural Income Disparity	Agricultural Resilience
Digital Village Level	0.5523 *** (0.093)	−0.0651 ** (0.027)	0.1691 *** (0.042)
Urban–Rural Income Disparity			−1.2199 ** (0.513)
Control variable	YES	YES	YES
Province/Year Fixed Effects	YES	YES	YES
Observations	121	121	121
F-value	17.48	17.44	29.22
R2	0.964	0.866	0.967
Adj R2	0.959	0.846	0.961
Bootstrap test r(ind_eff)	−1.429 *** (0.309)
Bootstrap test r(dir_eff)	−0.3930 (0.392)

Note: Robust standard errors are in parentheses; *** p < 0.01, ** p < 0.05.

reports the results of the regression test for the mediating effects. The total effect of digital village levels on agricultural resilience is 0.5523, passing a significance test at the 1% level. The direct effect is 0.1691, and the result is significant at the 1% level. The indirect effect of digital village levels on the development of agricultural resilience through the urban–rural income gap is 0.0794, constituting 14.38% of the total effect. The results indicate that enhancing digital village levels in rural areas can help narrow the urban–rural income gap. Continuously reducing this gap can significantly and effectively promote the sustainable development of agricultural resilience in ecologically fragile ethnic areas. This supports hypothesis H2 as expected. Furthermore, to further validate the robustness of the mediating effect, this study employs bootstrap testing to robustly examine the mediating effect of the urban–rural income gap. The test results indicate that the indirect effect passes the significance test at the 1% level, and the confidence interval does not include 0. The mediating effect results are robust, further confirming hypothesis H2.

4.4. Endogenous Test

Due to data limitations and the potential for reverse causality between digital village level and agricultural resilience, efforts have been made to identify the net effects of digital villages on agricultural resilience in ecologically fragile ethnic areas. This aims to mitigate potential endogeneity issues in the model. In this study, the instrumental variables selected are the two-stage lag of digital village level and the per capita cultivated land area of agricultural, forestry, animal husbandry, and fishery employees. The estimation is conducted using the two-stage least squares method. From the perspective of the two-stage lag of digital village level, there is a policy lag in the effect of digital construction to meet the relevant requirements. The current agricultural resilience is not significantly affected by digital village construction in the first two periods, which meets the exogenous requirements. Regarding the per capita arable land area of agricultural, forestry, animal husbandry, and fishery employees, the construction of digital villages contributes to the improvement of agricultural infrastructure and the level of scale operation. The intensive cultivation of arable land per capita also reflects the level of digital informatization, meeting the relevant requirements. Moreover, the association between the per capita arable land area of agricultural, forestry, animal husbandry, and fishery workers and agricultural resilience is relatively weak, satisfying the exogeneity requirement. The results are shown in

Table 6. Endogenous test results.

Variable	The First Stage	The Second Stage
	Agricultural Resilience	Agricultural Resilience
Digital Village Level	0.0994 **	0.0882 **
	(0.0383)	(0.037)
Control variable	YES	YES
Province/Year Fixed Effects	YES	YES
Kleibergen–Paap rk Wald LM statistic		24.454 ***
Cragg–Donald Wald F statistic		144.795 ***
Hansen J statistic		0.685
Observations	121	121
R2	0.972	0.975

Note: Robust standard errors are in parentheses; *** p < 0.01, ** p < 0.05.

. In the unrecognized test, the KP-LM statistic is 24.454, which significantly rejects the null hypothesis through the significance test at the 1% statistical level. Additionally, the C-D Wald F statistic in the weak instrumental variable test is 144.795, far exceeding the critical value of 19.93 at the 10% significance level, indicating the absence of a weak instrumental variable problem. The p-value of the Hansen J statistic in the over-identification test is 0.685, indicating that the selected instrumental variable is exogenous. It can be seen that the selection of the two instrumental variables above is more reasonable. After accounting for endogeneity, the digital village level still has a positive and significant impact on agricultural resilience. This significance at the 5% statistical level indicates that the conclusion of this study remains valid even after controlling endogenous problems.

4.5. Robustness Test

To validate the robustness of our research findings, three different methods are employed for robustness checks (

Table 7. Robustness test results.

Variable	Replacement Model	Tailing Processing	Replacement Variable
			Lagged One-Period FE	System GMM Model
Digital Village Level	0.0994 **	0.5563 ***	0.7182 ***	0.6844 ***
	(2.182)	(15.188)	(4.635)	(3.651)
Control variable	YES	YES	YES	YES
Province/Year Fixed Effects	YES	YES	YES	YES
Observations	121	121	110	121
F-value	123.72	—	40.81	—
p-value	0.0000	—	0.0000	—

Note: Robust standard errors are in parentheses; *** p < 0.01, ** p < 0.05.

). Firstly, model replacement is employed, using OLS models to conduct robustness tests for the research hypotheses. Secondly, all variables in the study are trimmed by 1% to address potential biases in regression test results caused by extreme values in the original model variables. Thirdly, variable replacement involves recalculating agricultural resilience scores and digital village scores using the entropy method. Subsequently, fixed-effects tests are conducted separately using the lagged one-period digital village level, and regression analysis is performed on agricultural resilience and digital village levels using the system GMM. The results indicate that there is no significant change in the direction of the regression for the digital village level, only slight variations in the estimated coefficient sizes and significance levels. All three methods further affirm the robustness of the empirical findings in this study.

4.6. Heterogeneity Test

Considering the evolving trends of agricultural resilience and digital village development levels in provinces and cities during the sample period, along with the differences in socio-economic development, this study anticipates regional heterogeneity in the relationship between agricultural resilience and digital village development levels. According to

Table 8. Heterogeneity test results.

Variable	Northern Region	Southern Region
Digital Village Level	0.6546 ** (3.335)	0.1609 * (2.134)
Control variable	YES	YES
Province/Year Fixed Effects	YES	YES
Observations	66	55
R2	0.867	0.958
Adj. R2	0.823	0.949

Note: Robust standard errors are in parentheses; ** p < 0.05, * p < 0.1.

, the impact of digital village development levels on agricultural resilience has passed significance tests at the 5% and 10% levels in the northern and southern regions, respectively, showing a positive effect. The impact of digital village development levels on agricultural resilience is more intense in the northern region than in the southern region. Compared to the southern region, which has higher levels of digital village development, the northern region with lower levels of digital village development experiences a more significant powerful promoting effect on agricultural resilience through the improvement of digital village development.

5. Discussion

5.1. Digital Village Level and Agricultural Resilience Score

Studies indicate that the advancement of digital village construction and the enhancement of agricultural resilience are crucial for the region. On one hand, digital technology has become a crucial approach globally for advancing agricultural and rural modernization [51]. As the world’s second-largest country in terms of its farming population [52], China is rapidly promoting agricultural modernization. The digital village construction by the Chinese government can play a massive role in this endeavor. On the other hand, the introduction and application of resilience issues bring a new systemic perspective to the sustainable development of agricultural systems, posing sharp questions about the long-term survival and development capabilities of modern agricultural systems [53]. The proportion of young and middle-aged populations in ecologically vulnerable ethnic areas is nearly 90%. From 2010 to 2020, the average years of education per capita in this region increased from 8.17 to 9.17 years. The scale of internet users has grown by approximately 40% since 2016. With a reasonable population structure and increased education levels, this region has become a “blue ocean” for China’s digital economic growth and information technology development. Therefore, this study formulated a more comprehensive evaluation system for digital village development and agricultural resilience. Drawing on relevant research on digital village development and agriculture, the assessment of digital village development is determined by examining aspects of inclusive finance, trade services, technological applications, and infrastructure. The evaluation system for agricultural resilience is based on the resource endowment and ecological development policies in ecologically fragile ethnic areas. It characterizes the ecological environment, development capabilities, and rural changes through six aspects across three dimensions: resistance, adaptability, and transformability.

The evaluation system addresses the shortcomings of previous research, which examined regions too broadly, resulting in a lack of specificity in constructing indicators for local resource endowment and targeted policy characterization. It also aims to avoid focusing solely on community farmer issues, which could lead to an oversight of the external policy environment. The evaluation system systematically examines indicators for ecologically fragile ethnic areas, with a specific focus on industrial stability, ecological sustainability, and agricultural growth dynamics. Past research has commonly employed the entropy method for evaluating digital village level and agricultural resilience [54,55,56]. However, this approach faces challenges in reducing the dimensionality of the evaluation indicator system, which may overlook the importance of certain indicators and lead to discrepancies between indicator weights and expectations. This study further employs the projection pursuit model to assess and measure both, projecting high-dimensional data into a low-dimensional subspace based on genetic algorithms. This approach achieves multidimensional global optimization, providing a more objective estimation and optimized calculation of the digital village level and agricultural resilience. Examining the average provincial changes, the digital village level gradually increased from 2010 to 2020, while agricultural resilience exhibited a slight downward trend. A possible reason is that the most stringent environmental protection law in Chinese history was introduced in 2015, which made the digital village and agricultural system subject to strict environmental constraints, resulting in certain fluctuations and declines. Regionally, both digital village level and agricultural resilience exhibit a characteristic of “Northern region < Southern region.” The gap in agricultural resilience between different regions is gradually narrowing, while the gap in digital village level is expanding further. This variation in gaps directly reflects the differences in economic foundation conditions across regions. According to the calculation data from the China Academy of Information and Communications Technology, 5G base stations require three times the number and investment compared to 4G base stations for the same coverage. This high cost limits the construction of digital infrastructure to some extent. Technological progress is at the core of economic growth, and the limited research and development capabilities of basic software and key equipment constrain improvements in digital village level. The data indicate that Sichuan, a prominent province in western China, has the highest level of digital villages and agricultural resilience. Tibet, as a relatively underdeveloped region, ranks lowest of all provinces and cities. In addition, the global agricultural product supply chain has been affected by factors such as climate change, local wars, and epidemic shocks in recent years, showing a trend of fluctuating agricultural resilience and a slight overall decline, indicating that the measurement indicators are reasonable and scientific.

5.2. The Impact of Digital Village Level on Agricultural Resilience

The findings of this study provide valuable empirical experience and policy insights for digital village development and the enhancement of agricultural resilience in ecologically fragile ethnic regions in China. Strategies such as increasing the utilization of renewable agricultural resources and industrialization, expanding channels for farmers’ income growth, and emphasizing the implementation of region-specific policies contribute to enhancing agricultural resilience in the region.

In fact, the development of agricultural resilience reflects the interaction within the internal agricultural system and the flow of elements. It involves externally mobilizing resources to cope with shocks, adapting to disturbances in the system, balancing internal deficiencies, and achieving the capacity for sustainable development. For ecologically fragile ethnic minority areas, in the face of protecting ecological and economic development, more consideration is given to enhancing the resilience of agricultural economic development after poverty alleviation and preventing large-scale return to poverty caused by exogenous environmental uncertainty and internal vulnerability of the system; more urgently, it is necessary to enhance the resilience of the agricultural economic system within the region, which is in line with the policy trend of digital village development. By alleviating the asymmetry of traditional agricultural information, promoting the flow of elements in the agricultural system, and releasing the environmental, resource, and energy constraints in ecologically fragile ethnic areas, the resilience of the agricultural system can be enhanced to lay a solid foundation for the future high-quality and modern development of agriculture.

This study demonstrates that digital village level has a positive impact on agricultural resilience in ecologically fragile ethnic regions at a 1% statistical significance level. Previous studies have indicated that strengthening digital infrastructure inspections can boost regional GDP and productivity [57,58]. However, during the construction process of digital village development, the high investment costs may become a significant obstacle hindering the progress of digital village development. The development of digital villages still faces constraints due to strict digital industrialization policies and the long-term maintenance costs of digital infrastructure. This has led to a gradual improvement in the digital village level in regions with limited human capital and economic endowments, further constraining the development space for agricultural resilience in these areas. Hence, the government must augment dedicated funding for the development of digital villages, establish long-term construction plans, and reduce the policy costs for local governments and rural areas. Only through these measures can the “digital divide” resulting from regional economic disparities be mitigated, fostering the enhancement of agricultural resilience. In addition, when considering the factors that influence agricultural resilience, it is evident that agricultural resilience has a positive impact on the utilization level of rural renewable resources and the level of industrialization. The agricultural sector can effectively mitigate land pollution, pesticide abuse, and atmospheric pollution caused by straw burning through the recycling of waste. Simultaneously, the increased level of industrialization brings about significant spillover effects, enabling the agricultural sector to adopt new agricultural technologies and enhance the value of agricultural products.

Mechanism studies have found that the improvement of the digital village level can increase farmers’ income, narrow the urban–rural income gap, and further enhance the agricultural resilience of ecologically fragile ethnic areas. The construction of digital village areas can expand rural entrepreneurial activities [59] and incentivize farmers to adopt proactive resource allocation for maximized operation [60]. Coupled with information technology that brings information connectivity and barrier removal, it can broaden farmers’ income channels, strengthening the agricultural resilience of the region. Furthermore, the continuous improvement of digital village infrastructure can reduce the outflow of rural talents [61], attracting more young human capital to return and participate in regional agricultural production, thereby energizing rural economic development. Studies on regional heterogeneity reveal that while overall levels of digital village development and agricultural resilience are higher in the southern region than the northern region, the impact coefficient of digital village development on agricultural resilience is 0.494 units greater in the northern region, exhibiting higher statistical significance. This implies that the differences in digital village development are primarily attributed to regional disparities, confirming and aligning with the results of Leong et al. [62] and Zhao et al. [63].

However, this study has some limitations. Firstly, the research focus is concentrated on ecologically fragile ethnic areas in China, and there has not been relevant research on digital village development and agricultural resilience in ecologically fragile and ethnically concentrated areas globally. Secondly, there are tangible differences among rural social groups, and the province-level evaluation indicators and calculation results may not necessarily apply to more microscopic rural areas. Finally, there is a lack of attention to individual farmers and families in ecologically fragile ethnic regions, and future research can focus on these aspects. Despite these limitations, this study’s theoretical foundation is robust, and the research methods are scientifically sound, not compromising the effectiveness of the research results.

6. Conclusions

In this study, the projection pursuit model is used to calculate the digital village level and agricultural resilience of ecologically fragile ethnic areas. After obtaining scores for the digital village level and agricultural resilience, this study employs the instrumental variable method and mediation effect model to analyze the impact of digital village level on agricultural resilience and its mechanisms. Finally, regional heterogeneity analysis is conducted on the sampled provinces. Based on this, the following conclusions are drawn:

• In ecologically fragile ethnic areas, both the digital villages level and agricultural resilience scores display an initial rise followed by a decline, with considerable overall volatility and unstable growth;
• The digital village level has a significant positive effect on agricultural resilience, and the conclusion is still robust after controlling for endogeneity;
• The improvement of the digital village level promotes the enhancement of agricultural resilience by narrowing the urban–rural income gap. The elevation of the digital village level optimizes operational methods and expands income channels, contributing to an increase in farmers’ income levels and consequently reducing the urban–rural income disparity, fostering healthy development of agricultural resilience;
• There is evident regional heterogeneity in the impact of digital village level on agricultural resilience in ecologically fragile ethnic areas. The influence of digital village areas on agricultural resilience is more pronounced in the northern region compared to the southern region, indicating that the improvement of the digital village level has a more significant and powerful effect on the development of agricultural resilience in the northern region.

Based on the research conclusions, the following policy recommendations are proposed:

• Prioritize the dynamic trends of digital village level and agricultural resilience, leveraging the leading role of provinces with elevated digital village levels and robust agricultural resilience. In ecologically fragile ethnic areas where there are fluctuations in both digital levels and agricultural resilience, policies should dynamically track and analyze evolving patterns through policy instruments. Additionally, mobilize the leading role of provinces with high digital village levels and strong agricultural resilience, explore the intrinsic potential of agricultural development, harness their first-mover advantages in digital villages and agricultural resilience, facilitate information sharing and resource exchange among provinces, and radiate to drive the common development of surrounding provinces;
• Strengthen investment in digital village construction and accelerate the rural digitization process. Utilize data as a crucial element in the digital economy and focus on infrastructure construction such as 5G base stations and internet accessibility for digital development in rural areas. Improve the digital network in rural areas, deeply integrate digital technology and information devices in the agricultural sector, enhance agricultural production efficiency, reduce production costs, and mitigate losses from risk impacts, thereby promoting farmers’ income growth, narrowing the urban–rural income gap, and improving agricultural risk resilience and recovery capability;
• Comprehend the disparities in digital village sophistication and agricultural resilience among regions, striving for precise implementation. There are significant disparities in digital village levels and agricultural resilience among different regions. Government decision making should consider differences in economic foundations, resource endowments, and ecological environments between regions. Develop reasonable and scientific digital village policies based on regional development requirements and paths. Introduce subsidies and financial credit policies tailored to local conditions for the implementation of digital technology. Strengthen the role of digital village areas in stimulating agricultural resilience in the northern regions while leveraging the economic and talent advantages in the southern regions. Narrow regional gaps to establish a rational pattern for digital economic development.