Spatio-Temporal Characteristics and Influencing Mechanisms of China’s Digital Rural Development: A Panel Data Analysis Across 31 Provinces

Abstract

To advance human society towards a fully inclusive and accessible digital future, it is essential to foster the comprehensive and balanced development of digital villages, thereby addressing rural residents’ aspirations for a digitally enriched life. This study systematically investigates the spatiotemporal differentiation patterns and spatial spillover effects of China’s Digital Rural Development (DRD). Utilizing panel data from 31 provinces in China from 2013 to 2022, we construct a comprehensive evaluation framework covering digital infrastructure, economic digitization, governance digitization, and life digitization. The empirical analysis integrates entropy weighting, Dagum Gini coefficient decomposition, Moran’s I index, and spatial Durbin models. The findings indicate that China’s DRD has exhibited sustained overall improvement, progressing through three distinct phases: slow growth, rapid advancement, and fluctuating ascent. Significant regional disparities persist, with eastern regions consistently outperforming central, western, and northeastern areas. Inter-regional differences constitute the primary source of overall variation, and this gap has progressively widened over time. Spatially, DRD demonstrates a significant positive agglomeration effect alongside a negative spatial spillover effect (ρ = −1.3209), suggesting that advancements in neighboring regions may inhibit local development progress. Mechanism analysis identifies technological innovation, rural population size, and age structure as key local determinants, whereas industrial upgrading generates significant positive spillover effects on surrounding regions. Based on these results, at the same time, in order to ensure the sustainable development of DRD, we propose the following policy recommendations: implement regionally differentiated interventions, enhance the alignment of core local drivers, establish interregional coordination mechanisms, and develop dynamic monitoring and adjustment systems. These measures are expected to promote more balanced urban–rural and regional development, offering empirical evidence and policy insights for other developing countries pursuing similar pathways of rural digital transformation.

1. Introduction

The pervasive integration of digital technologies—including big data, cloud computing, and artificial intelligence—is fundamentally transforming the global economic and social fabric. Nations across the world are actively devising strategic plans to harness these innovations for competitive gains. Within this international landscape, China has emerged as a prominent actor, elevating Digital Rural Development (DRD) to the status of a national strategic priority and positioning it at the intersection of rural revitalization and the Digital China initiative. Formally initiated through the 2018 Central Document No. 1, which integrated the “Digital Village Strategy” into national policy, China has since rolled out a series of guiding frameworks, such as the Digital Village Development Strategy Outline and the Digital Agriculture and Rural Development Plan (2019–2025). The 2022 Central Document No. 1 further emphasized the need to anchor efforts in major national strategic priorities, stabilize the agricultural sector, advance work related to agriculture, rural areas, and farmers, and sustain comprehensive rural revitalization. During the recent National People’s Congress and Chinese People’s Political Consultative Conference sessions, members of the CPPCC also put forward practical proposals aimed at promoting rural revitalization. These suggestions focus on bridging the urban–rural digital gap, dismantling the urban–rural dual structure, fostering balanced regional development, and ultimately accelerating agricultural and rural modernization. Despite significant progress in recent years toward near-universal digital infrastructure coverage in rural China, a substantial digital divide persists between urban and rural areas—a challenge shared by many nations globally. This gap raises concerns that rural regions may risk becoming “digital slums” [1]. Moreover, China’s vast territory and uneven resource distribution have led to marked spatial disparities in DRD across regions, highlighting the need for deeper investigation into its underlying drivers and spatial interaction effects.

DRD represents a critical global issue focused on addressing the urban–rural digital divide and advancing sustainable rural growth. International research often explores this domain through lenses such as digital exclusion, policy intervention, technology adoption, and social capital, revealing the multidimensional complexity of the subject. For instance, Pavez et al. [2], in their study of rural Chile, observed that despite improvements in digital infrastructure, age and educational attainment remain key predictors of digital access. Di Stefano et al. [3], examining Italy’s agricultural digitization policies, identified uneven regional resource allocation and insufficient technical training as primary obstacles to digital transformation. Ferrari et al. [4] developed a driver-barrier-impact framework through expert interviews, highlighting the influence of socioeconomic characteristics and institutional environments on technology implementation. Warren [5] pioneered the “digital vicious cycle” theory, which underscores how digital exclusion and social exclusion can reinforce each other, particularly in remote rural contexts. Additionally, Visvizi and Lytras [6] proposed the “Smart Village” concept, advocating for the integration of information and communication technologies with rural governance and sustainable development. Kerras et al. [7] further validated the intrinsic link between the digital divide and the socioeconomic status of rural residents through structural equation modeling.

In recent years, DRD has become a prominent research focus within China’s academic community. Studies on its level measurement and spatiotemporal characteristics have progressively deepened. For instance, Zhao Chengwei et al. [8] constructed a three-dimensional indicator system covering infrastructure, industry, and governance, finding that eastern regions lead in DRD levels and that regional disparities continue to widen. Wang Liukun et al. [9] further applied entropy weighting and Moran’s I index to corroborate the spatial differentiation and agglomeration patterns of DRD in China, which is characterized by higher levels in the east and lower levels in the west. These conclusions align with those of Zhang and Zhou et al. [10], who utilized an intuitionistic fuzzy analytic hierarchy process (IFAHP) and a dynamic grey relational analysis (GRA) model to identify significant spatial correlation and network spillover effects in China’s DRD. Through geographic detector analysis, they also confirmed that investment in information infrastructure and fiscal expenditure on education serve as core driving factors. In a related vein, Zhao et al. [11] further elucidated that the coupling coordination between DRD and agricultural-rural modernization is influenced by multiple factors, including ecological livability and governance systems.

However, existing research still faces the following key limitations: First, most studies focus on describing regional disparities and positive spatial linkages, lacking systematic empirical examination of the “siphon effect” where neighboring development inhibits local growth. While Park’s [12] research on Australian communities provides micro-level evidence of the negative spatial externalities of digital technology, this perspective remains rare in domestic macro-level empirical studies. Second, analyses of influencing mechanisms have not sufficiently incorporated factors such as industrial structure upgrading, population size, and age structure into spatial econometric models to decompose their direct and indirect effects [13]. Furthermore, existing domestic indicator systems predominantly emphasize infrastructure and economic dimensions, with inadequate comprehensive consideration of governance digitization and digital livelihood dimensions [14].

In light of the above context, this study seeks to systematically uncover the spatial interaction mechanisms underlying the development levels of DRD in China. Specifically, it addresses the following two research questions: (1) What are the defining characteristics of the spatiotemporal differentiation in China’s DRD levels, and does a significant negative spatial spillover effect exist? (2) What core factors drive improvements in DRD levels, and how do the spatial mechanisms of these factors differ?

The structure of this paper is organized as follows: the Section 1 outlines the research design; the Section 2 measures the development levels of digital villages and analyzes their spatiotemporal characteristics; the Section 3 empirically examines the influencing mechanisms using a Spatial Durbin Model; and the Section 4 presents conclusions and policy recommendations. The potential contributions of this study are threefold: First, based on panel data from 31 provinces in China spanning 2013 to 2022, we construct a comprehensive evaluation system covering four dimensions—digital infrastructure, economic digitization, governance digitization, and lifestyle digitization. Second, by integrating the Dagum Gini coefficient, Moran’s index, and the Spatial Durbin Model, this research empirically verifies, at the macro provincial level, the existence of a significant negative spatial spillover effect in China’s DRD, thereby revealing a competitive “siphon effect” among regions. Third, through the analysis of both direct and indirect effects, the study elucidates the distinctive mechanism whereby key driving factors predominantly operate through spatial spillovers rather than through localized impacts.

2. Research Design

2.1. Development of the Indicator System

DRD refers to the comprehensive digital transformation of rural economies, governance, culture, and public services through modern information technologies, with the strategic aim of achieving agricultural and rural modernization alongside rural revitalization. Its core components include the digitalization of infrastructure, agricultural production, rural governance, public services, rural culture, and the ecological environment. This initiative aims to narrow the urban–rural digital divide, promote economic diversification in rural areas, and enhance the efficacy of rural governance. DRD constitutes a multidimensional and systematic process. To ensure the comprehensiveness and scientific rigor of the evaluation framework, this study identifies core dimensions across three levels: theoretical foundations, policy orientations, and practical applications. First, from a theoretical perspective, based on the Technology–Organization–Environment (TOE) framework, digital transformation must encompass technological infrastructure, organizational applications, and the external environment. Second, at the policy level, key national documents such as the Digital Rural Development Strategy Outline and the Digital Rural Development Action Plan explicitly identify infrastructure construction, the digital economy, governance capacity, and public services as focal areas for development. Finally, in terms of academic practice, this paper draws upon existing research [15,16,17,18] while aligning with the objectives outlined in relevant policy guidelines, including the Digital Village Development Strategy Outline and the Digital Rural Development Action Plan (2022–2025). The resulting evaluation system comprises 12 indicators across four dimensions: digital infrastructure, economic digitization, governance digitization, and life digitization. To ensure objectivity in weighting, the entropy weighting method was applied. Prior to weighting, Variance Inflation Factor (VIF) tests were conducted on all 12 indicators to assess multicollinearity. Results showed that all VIF values remained well below the critical threshold of 10, indicating no severe multicollinearity. Eleven indicators had VIF values below 5, while the “Digital Finance” indicator recorded a VIF of 5.29, reflecting its inherent correlation with the level of digital economic development. The average VIF across the system was 2.82, confirming low information overlap among indicators and their capacity to independently represent distinct aspects of DRD. These results satisfy the data requirements for comprehensive evaluation methods. Entropy weighting outcomes revealed that indicators such as “rural e-commerce service stations” and “e-commerce development level” contributed notably to the overall evaluation, underscoring the pivotal role of e-commerce in driving the digital transformation of rural economies. Detailed indicators and their weights are presented in

Table 1. Indicator System for Assessing the Level of DRD.

Dimension	Indicator Name	Indicator Explanation	Unit	Indicator Attributes	Weighting
Digital Infrastructure	Internet Penetration Rate	Rural Broadband Subscribers	104 households	Positive	0.1017
	Computer Penetration Rate	Number of Computers Per 100 Rural Households at Year-End	Units/100 households	Positive	0.0275
	Smartphone Penetration Rate	Mobile Telephone Ownership Per Hundred Rural Households at Year-End	Units/100 households	Positive	0.0161
Economic Digitalization	E-commerce Development Level	Agricultural Product E-Commerce Sales	Billion CNY	Positive	0.1377
	Digital Finance	Rural Inclusive Finance Index Total		Positive	0.0300
	Rural E-commerce Service Points	Indicator Number of Taobao Rurals	Number	Positive	0.2797
Digital Governance	Capital Investment in Rural Digital Governance	Local Government Expenditure on Urban and Rural Community Affairs	Billion CNY	Positive	0.0634
	Digital Government Human Resource Capacity	Agricultural Technicians in Public Economic Enterprise and Institutions	Name	Positive	0.0616
	Rural Digital Government Technology Application Level	Ecological and Agricultura Meteorological Experimental Stations	Number	Positive	0.0633
Digitalization of Living Standards	Digital Cultural Resources Reaching Rurals and Households	Proportion of Rural Cable Television Subscribers to Total Households	%	Positive	0.0605
	Level of Information Service Consumption	Per Capita Expenditure on Transport and Communications Among Rural Residents	Yuan	Positive	0.0415
	Information Technology Services	Total Telecommunications Business Revenue	Billion CNY	Positive	0.1170

.

2.2. Data Sources

This study utilizes panel data from 31 provinces in mainland China (excluding Hong Kong, Macao, and Taiwan) covering the period 2013–2022. Data are primarily drawn from the China Statistical Yearbook, the China Rural Statistical Yearbook, provincial statistical yearbooks, the National Bureau of Statistics, and Alibaba research reports. The Rural Digital Inclusive Finance Index is constructed by averaging the county-level indices from the Peking University Digital Inclusive Finance Index. Missing observations were imputed using linear interpolation to maintain a balanced panel dataset.

2.3. Research Methodology

2.3.1. Entropy Weight Method

The level of DRD is evaluated through a multi-indicator composite system. Traditional subjective weighting methods are often influenced by evaluator bias, which can compromise the objectivity of results. In contrast, the entropy weight method determines weights based on the intrinsic dispersion of the data, serving as an objective weighting technique that more accurately reflects the actual importance of each indicator, thereby improving the reliability and credibility of the evaluation outcomes. Accordingly, this study adopts the entropy weight method for indicator weighting. The specific computational steps are outlined below.

1.

Data standardization:

For positive indicators: $Y_{ij}={\displaystyle \frac{X_{ij}-\mathit{min}\left(X_j\right)}{\mathit{max}\left(X_j\right)-\mathit{min}\left(X_j\right)}}$.

For negative indicators: $Y_{ij}={\displaystyle \frac{\mathit{max}\left(X_j\right)-X_{ij}}{\mathit{max}\left(X_j\right)-\mathit{min}\left(X_j\right)}}$.

X_ij denotes the raw value of the i-th sample for the j-th indicator, and Y_ij represents the standardised value. min(X_j) and max(X_j) denote the minimum and maximum values for the j-th indicator, respectively.

2.

Calculate the proportion P_ij of sample i under the j-th indicator:

Specific gravity P_ij: $P_{ij}={\displaystyle \frac{Y_{ij}}{{\sum }_{i=1}^n{Y}_{ij}}}$, where n denotes the sample size.

3.

Calculate the information entropy Ej of the j-th indicator:

Information entropy E_j: $E_j=-k{\sum }_{i=1}^n{P}_{ij}\ln \left(P_{ij}\right)$, where $k={\displaystyle \frac{1}{\ln \left(n\right)}}$.

4.

Calculate the information entropy redundancy D_ij for the j-th indicator:

Information entropy redundancy D_ij: $D_{ij}=1-E_j$.

5.

Calculate the weight W_j for the j-th indicator:

Weight W_j: $W_j={\displaystyle \frac{D_j}{{\sum }_{j=1}^m{D}_j}}$, where m denotes the index number.

2.3.2. Dagum Gini Coefficient

This study utilizes the Dagum Gini coefficient to assess disparities in DRD across regions. It decomposes the overall disparity into three components: intra-regional inequality, inter-regional inequality, and super-inequality density. This enables precise identification of the contribution of intra-regional and inter-regional inequality to the overall disparity, thereby facilitating an understanding of the imbalance between different regions and its origins [19]. The specific methodology entails first calculating the intra-regional Gini coefficient G_w, the inter-regional Gini coefficient G_b, and the super-variability density G_t for each region. Subsequently, the overall Gini coefficient is computed: $G=G_w+G_b+G_t$.

$$G_w={\sum }_{r=1}^k{\displaystyle \frac{n_r}{n}}{\mu }_r{G}_r$$

(1)

$$G_b={\sum }_{r=1}^{k-1}{\sum }_{s=r=1}^k{\displaystyle \frac{n_r{n}_s}{n^2}}\left|{\mu }_r-{\mu }_s\right|\times D_{rs}$$

(2)

$$G_t={\sum }_{r=1}^{k-1}{\sum }_{s=r+1}^k{\displaystyle \frac{n_r{n}_s}{n^2}}|{\mu }_r-{\mu }_s|\times (1-D_{RS})$$

(3)

$$G_r={\displaystyle \frac{1}{2n_r^2{\mu }_r}}{\sum }_{i=1}^{n_r}{\sum }_{j=1}^{n_r}|x_{ri}-x_{rj}|$$

(4)

$$D_{rs}={\displaystyle \frac{{\mu }_r+{\mu }_s}{2\mu }}\times {\left({\displaystyle \frac{{\mu }_r}{{\mu }_s}}+{\displaystyle \frac{{\mu }_s}{{\mu }_r}}\right)}^{-1}\times \left(1+{\displaystyle \frac{2\mu \times {∆}_{rs}}{{\mu }_r+{\mu }_s}}\right)$$

(5)

where ${∆}_{rs}={\displaystyle \frac{1}{n_r{n}_s}}{\sum }_{i=1}^{n_s}\left({\displaystyle \frac{x_{ri}-x_{sj}}{\mu }}\right)\times I\left(x_{ri}\ge x_{sj}\right)$ denotes the indicator function, which takes the value 1 when the condition holds, and 0 otherwise.

The study area is categorized into k = 4 regions. For the r-th and s-th regions, the sample sizes are denoted as n_r and n_s, with regional mean DRD scores of μ_r and μ_s, respectively. Let n represent the total sample size and μ the overall mean score. Furthermore, x_ri and x_rj denote the comprehensive DRD scores of the i-th and j-th samples within region r.

2.3.3. Moran’s I Index

Moran’s I statistic serves as a spatial autocorrelation measure to assess whether observed spatial data exhibit clustering, dispersion, or random distribution patterns. In this study, both the global Moran’s I and local Moran’s I (LISA) are calculated to examine the presence of spatial clustering effects and to identify local spatial dependencies in the level of DRD across regions.

• Global Moran’s I: $I={\displaystyle \frac{n{\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{ij}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}{{\sum }_{i=1}^n{\sum }_{j=1}^n{w}_{ij}{\sum }_{i=1}^n{\left(x_i-\overline{x}\right)}^2}}$.
• Local Moran’s I: $I_i={\displaystyle \frac{\left(x_i-\overline{x}\right){\sum }_{j=1}^n{w}_{ij}\left(x_j-\overline{x}\right)}{S^2}}$, where $S^2={\displaystyle \frac{{\sum }_{i=1}^n{\left(x_j-\overline{x}\right)}^2}{n}}$.

Here, n denotes the number of spatial units (provinces), x_i and x_j represent the DRD composite scores of units I and j, $\overline{x}$ is the mean DRD score, and w_ij is the spatial weight matrix element defining the connectivity between units i and j.

2.3.4. Spatial Durbin Model

The spatial Durbin Model (SDM) is an advanced spatial econometric specification that extends both the spatial autoregressive (SAR) and spatial error (SEM) models. Unlike these conventional approaches, the SDM incorporates spatial lags of not only the dependent variable but also the independent variables, allowing for a more complete representation of spatial interaction and spillover mechanisms. Its general formulation can be expressed as:

$$y_{it}=\rho {\sum }_{j=1}^n{w}_{ij}{y}_{it}+\beta x_{it}+\theta {\sum }_{j=1}^n{w}_{ij}{x}_{jt}+{\mu }_i+{\vartheta }_t+{ϵ}_{it}$$

(6)

where y_it denotes the composite score for DRD in province i during year t, ρ represents the spatial autoregressive coefficient, w_ij constitutes the spatial weighting matrix, x_it comprises a series of explanatory variables potentially influencing DRD levels, β corresponds to the coefficient vector, μ_i signifies the province-specific fixed effect, ϑ_t indicates the time-specific fixed effect, and ϵ_it constitutes the random error term.

3. Measurement and Analysis of DRD Levels

3.1. Temporal Evolution of DRD Levels

Table 2. Comprehensive Scores for DRD Levels by Province, 2013–2022.

Region	Province	2013	2014	2015	2016	2017	2018	2019	2020	2021	2022	Average	Rank
Eastern Region	Beijing	0.1567	0.1658	0.2012	0.2081	0.2351	0.2615	0.2917	0.3004	0.3124	0.3214	0.2454	5
	Tianjin	0.0865	0.0922	0.1037	0.1228	0.1389	0.1318	0.1466	0.1533	0.1690	0.1698	0.1315	22
	Hebei	0.1228	0.1425	0.1543	0.1757	0.1965	0.2359	0.2950	0.3244	0.3151	0.3378	0.2300	6
	Shanghai	0.1341	0.1702	0.1824	0.1861	0.1913	0.2284	0.2354	0.2395	0.2525	0.2745	0.2094	7
	Jiangsu	0.2195	0.2456	0.2909	0.2984	0.3376	0.4072	0.4818	0.5096	0.4684	0.4899	0.3749	3
	Zhejiang	0.1733	0.1893	0.2473	0.2848	0.3401	0.4212	0.5257	0.5392	0.5746	0.6092	0.3905	2
	Fujian	0.1073	0.1232	0.1371	0.1536	0.1824	0.2100	0.2692	0.2586	0.2584	0.2856	0.1985	10
	Shandong	0.1904	0.2004	0.2403	0.2704	0.3157	0.3802	0.4017	0.4442	0.4327	0.4645	0.3341	4
	Guangdong	0.1802	0.2002	0.2357	0.2610	0.3500	0.4292	0.5610	0.5819	0.5490	0.5868	0.3935	1
	Hainan	0.0414	0.0508	0.0627	0.0630	0.0740	0.0939	0.1134	0.1258	0.1375	0.1592	0.0922	28
	Average	0.1412	0.1580	0.1856	0.2024	0.2362	0.2799	0.3322	0.3477	0.3470	0.3699	0.2600
Central Region	Shanxi	0.0815	0.0869	0.0974	0.1100	0.1196	0.1433	0.1600	0.1732	0.1964	0.2311	0.1400	20
	Anhui	0.0801	0.0976	0.1157	0.1210	0.1467	0.1835	0.2168	0.2268	0.2104	0.2175	0.1616	15
	Jiangxi	0.0916	0.1056	0.1219	0.1411	0.1613	0.1879	0.2095	0.2107	0.1969	0.2096	0.1636	12
	Henan	0.1076	0.1314	0.1435	0.1612	0.1924	0.2222	0.2863	0.2919	0.2625	0.2684	0.2067	8
	Hubei	0.1099	0.1145	0.1340	0.1426	0.1587	0.1909	0.2383	0.2544	0.2491	0.2603	0.1853	11
	Hunan	0.0908	0.1012	0.1173	0.1374	0.1513	0.1813	0.2131	0.2223	0.1963	0.2083	0.1619	14
	Average	0.0936	0.1062	0.1216	0.1356	0.1550	0.1848	0.2207	0.2299	0.2186	0.2326	0.1698
Northeast Region	Liaoning	0.1208	0.1348	0.1440	0.1479	0.1573	0.1750	0.1903	0.1940	0.1831	0.1879	0.1635	13
	Jilin	0.0959	0.1093	0.1190	0.1206	0.1268	0.1415	0.1483	0.1616	0.1475	0.1464	0.1317	21
	Heilongjiang	0.0976	0.0980	0.1101	0.1150	0.1260	0.1359	0.1555	0.1477	0.1382	0.1399	0.1264	24
	Average	0.1048	0.1140	0.1243	0.1278	0.1367	0.1508	0.1647	0.1678	0.1563	0.1581	0.1405
Western Region	Guangxi	0.0797	0.0868	0.1046	0.1064	0.1221	0.1547	0.1922	0.2012	0.1878	0.1964	0.1432	19
	Inner Mongolia	0.1015	0.1113	0.1171	0.1215	0.1243	0.1295	0.1503	0.1571	0.1484	0.1519	0.1313	23
	Chongqing	0.0914	0.1059	0.1137	0.1208	0.1390	0.1565	0.1745	0.1887	0.1809	0.2082	0.1480	17
	Sichuan	0.1136	0.1258	0.1484	0.1579	0.1879	0.2184	0.2679	0.2908	0.2537	0.2845	0.2049	9
	Guizhou	0.0560	0.0653	0.0753	0.0860	0.1082	0.1338	0.1657	0.1742	0.1575	0.1745	0.1197	25
	Yunnan	0.0873	0.0997	0.1093	0.1077	0.1308	0.1609	0.1894	0.2042	0.1747	0.1791	0.1443	18
	Tibet	0.0152	0.0220	0.0307	0.0328	0.0359	0.0507	0.0626	0.0634	0.0682	0.0685	0.0450	31
	Shaanxi	0.0937	0.1021	0.1128	0.1239	0.1383	0.1643	0.1873	0.1983	0.1819	0.2018	0.1505	16
	Gansu	0.0595	0.0741	0.0805	0.0878	0.0995	0.1186	0.1354	0.1411	0.1334	0.1398	0.1070	27
	Qinghai	0.0409	0.0515	0.0592	0.0612	0.0676	0.0694	0.0810	0.0850	0.0857	0.0858	0.0687	30
	Ningxia	0.0475	0.0524	0.0582	0.0633	0.0718	0.0803	0.0873	0.0916	0.0937	0.0923	0.0738	29
	Xinjiang	0.0685	0.0855	0.0951	0.1003	0.1068	0.1043	0.1293	0.1326	0.1269	0.1365	0.1086	26
	Average	0.0712	0.0819	0.0921	0.0975	0.1110	0.1284	0.1519	0.1607	0.1494	0.1599	0.1204
National	Average	0.1014	0.1143	0.1311	0.1416	0.1624	0.1904	0.2246	0.2351	0.2272	0.2415	0.1770

presents the comprehensive scores of DRD across China’s provinces from 2013 to 2022. The national average DRD score increased from 0.1014 in 2013 to 0.2415 in 2022, representing an average annual growth rate of 10.12%, which reflects the sustained progress driven by national policy initiatives in this domain. When examined by phase, China’s DRD exhibited a three-stage trajectory: slow growth (2013–2016)-rapid advancement (2016–2019)-fluctuating ascent (2019–2022), closely mirroring major policy developments. During the initial exploratory phase (2013–2016), growth was moderate with an average annual rate of 11.77%, as policy focus centered primarily on infrastructure deployment. Following 2016, and particularly after the official launch of the “Digital Village Strategy” in 2018, enhanced investment and regional pilot programs accelerated development, resulting in a rapid advancement phase with an average annual growth rate of 16.62%. After 2019, despite pandemic-related disruptions to offline promotion, rising demand for online services helped sustain momentum, though growth decelerated to an annual average of 2.45%. Jiangsu, Zhejiang, Shandong, and Guangdong consistently ranked highest in DRD, achieving 2022 composite scores of 0.3749, 0.3985, 0.3341, and 0.3935, respectively. In contrast, Qinghai, Ningxia, and Tibet recorded the lowest scores, at 0.0687, 0.0738, and 0.0450, respectively. Throughout the study period, coastal provinces with stronger economic foundations consistently demonstrated higher levels of DRD compared to inland regions, highlighting a correlation between DRD performance and regional economic development.

3.2. Spatial Characteristics of DRD Levels

As shown in Figure 1, the 31 provinces are categorized into four major regions: the Eastern, Central, Western, and Northeastern regions. Throughout the study period, DRD levels in all regions have shown significant improvement, reflecting strong nationwide momentum in advancing digital rural initiatives. However, notable regional disparities persist. Specifically, the average annual growth rates for the Eastern, Central, Western, and Northeastern regions were 11.29%, 10.64%, 9.40%, and 4.67%, respectively. The Eastern region’s leading growth rate can be attributed to multiple synergistic factors, including robust local fiscal support, digital industry clusters, mature market-oriented mechanisms, and a strong human capital base. The Central region has demonstrated steady progress, benefiting from industrial transfers from the East and the expansion of agricultural e-commerce. Despite its relatively weak foundation, the Western region has achieved relatively rapid growth, driven in part by targeted national infrastructure investments. Nevertheless, geographical constraints have led to high network construction costs in remote rural areas, while population outflows have resulted in insufficient demand for digital services. In contrast, the Northeastern region has experienced sluggish growth, which is closely associated with the delayed transformation of traditional industries, slow progress in digital industrialization, and underdeveloped marketization and professionalization in rural digital governance and living services. These factors collectively hinder the region’s ability to generate sustained development momentum. The persistent inter-provincial disparities in DRD are fundamentally rooted in path dependency effects, shaped by a combination of factors such as resource endowments, historical development foundations, and policy implementation efficiency. Given these structural differences, relying solely on uniform policy measures is unlikely to rapidly reverse existing imbalances. Instead, targeted and differentiated support mechanisms are essential to promote more equitable and coordinated development across regions.

3.3. Regional Variation Analysis

Table 3. Dagum Gini Coefficient Measurement Results.

Dagum Gini Coefficient Measurement Results
Year	Gini Coefficient				Contribution Rate
	Total	Intra-Regional Gini Coefficient Gw	Inter-Regional Gini Coefficient Gb	Trans-Regional Gini Coefficient Gt	Intra-Regional Contribution Rate Gw	Inter-Regional Contribution Rate Gb	Trans-Regional Density Contribution Rate Gt
2013	0.242	0.055	0.158	0.029	22.787%	65.218%	11.995%
2014	0.232	0.052	0.151	0.029	22.378%	65. 148%	12.474%
2015	0.24	0.053	0.161	0.026	22.115%	66.919%	10.966%
2016	0.244	0.052	0.168	0.024	21.378%	68.762%	9.860%
2017	0.254	0.055	0.175	0.024	21.587%	68.817%	9.596%
2018	0.267	0.058	0.182	0.027	21.816%	68.060%	10.125%
2019	0.278	0.063	0.184	0.031	22.555%	66.345%	11.100%
2020	0.278	0.064	0.183	0.032	22.883%	65.619%	11.499%
2021	0.278	0.058	0.199	0.02	21.063%	71.711%	7.227%
2022	0.282	0.06	0.2	0.022	21.175%	70.993%	7.832%

reports the measurement results of the Dagum Gini coefficient. The overall Gini coefficient rose from 0.232 in 2014 to 0.282 in 2022, indicating that regional disparities in DRD are widening. Regarding the sources of disparity, inter-regional differences consistently contributed over 65% to the overall gap, aligning with China’s fundamental reality of uneven regional development. This widening disparity primarily stems from the “siphon effect” of digital factor concentration in eastern regions. Resources such as digital talent, capital, and high-quality projects are drawn to eastern areas offering superior development environments and higher returns, perpetuating regional development gaps. Meanwhile, the contribution rate of intra-regional disparities has remained stable at around 22%, while that of hyper-density disparities is less than 12%. This indicates relatively good development coordination within the same region, suggesting that inter-provincial gaps are primarily driven by the development divide between regions rather than intra-regional polarization. This finding reveals that the key to narrowing the development gap in DRD lies in addressing the imbalance in resource allocation between regions.

As shown in Figure 2, regional disparities in DRD exhibit a pronounced pattern of divergence. The East–West disparity is most pronounced, with the Gini coefficient reaching 0.413 in 2022 and continuing to widen. This stems primarily from the growing gap between the East’s high-quality development and the West’s foundational application stage. The East–Northeast disparity has surged rapidly, rising to 0.407 in 2022, mainly due to the intensified “siphon effect” of digital factors exacerbating the imbalance between Northeast China and the East. The trends in East–Central and Central–West disparities remained relatively stable, as the central region acted as a bridge to mitigate extreme gaps. Notably, the Northeast–West disparity showed a wave-like decline, while the Northeast–Central disparity surged significantly from 0.083 to 0.191. This highlights the stark contrast between the accelerated digital development in the central region and the lagging progress in Northeast China, indicating a significant reshaping of the regional disparity landscape.

Within each region, disparities exhibit significant heterogeneity. The eastern region shows the greatest internal variation, with a Gini coefficient reaching 0.232 in 2022, primarily driven by disparities in innovation capacity and resource integration efficiency among provinces. The western region saw a slight narrowing of internal differences, benefiting from targeted national infrastructure investments. The central region experienced fluctuating disparities before declining, decreasing by 0.056 in 2022, indicating improved development coordination. The Northeast region exhibited the smallest and most stable disparities, with a coefficient of 0.067 in 2022, reflecting strong development homogeneity but insufficient breakthrough innovation. Specific details are shown in Figure 3.

3.4. Spatial Correlation Analysis of DRD Levels

Table 4. Moran’s I Measurement Results.

Year	Moran’s I Index	Expected Value E(I)	Standard Deviation SD (I)	z-Value	p-Value
2013	0.063	−0.033	0.034	2.807	0.005
2014	0.072	−0.033	0.034	3.074	0.002
2015	0.072	−0.033	0.034	3.094	0.002
2016	0.077	−0.033	0.034	3.225	0.001
2017	0.058	−0.033	0.034	2.669	0.008
2018	0.059	−0.033	0.034	2.708	0.007
2019	0.047	−0.033	0.034	2.391	0.017
2020	0.041	−0.033	0.034	2.217	0.027
2021	0.055	−0.033	0.034	2.623	0.009
2022	0.055	−0.033	0.034	2.630	0.009

presents the results of the Moran’s I index measurements. From 2013 to 2022, the global Moran’s I index for China’s DRD consistently remained positive and statistically significant at the 5% level, confirming significant positive spatial autocorrelation in DRD levels across provinces. This indicates that DRD progress is not isolated at the provincial level but is significantly influenced by the development of neighboring regions, exhibiting a distinct agglomeration pattern characterized by “high–high” and “low–low” clustering. Temporally, the degree of spatial clustering followed a dynamic trajectory of initial strengthening, subsequent weakening, and eventual stabilization. From 2013 to 2016, Moran’s I increased steadily from 0.063 to 0.077, reflecting enhanced spatial concentration during the early phase of policy piloting and resource allocation. Between 2017 and 2020, the index gradually declined from 0.058 to 0.041, suggesting a weakening of agglomeration as interregional diffusion effects began to take hold, enabling some lagging regions to catch up through technology spillovers and policy support. From 2021 to 2022, the index stabilized around 0.055, indicating that the spatial pattern of DRD had entered a phase of relative equilibrium. This evolution reflects a transition in DRD from localized breakthroughs to a more coordinated, nationwide advancement. It not only illustrates the spatial spillover and regional catch-up mechanisms within DRD but also aligns with the gradual implementation of China’s regional coordinated development strategy. Nevertheless, the persistent spatial correlation highlights the path dependence inherent in the “center-periphery” structure. While high-level clusters generate positive spillovers that can uplift neighboring areas, low-level clusters risk being trapped in a cycle of underdevelopment due to the overall backwardness of their surrounding environments. Therefore, DRD remains significantly constrained by pre-existing regional development patterns. Moving forward, policy efforts should emphasize targeted cross-regional collaboration and institutional innovation to facilitate a spatial transition from fragmented agglomeration toward more integrated and balanced regional advancement.

Based on the Moran scatter plot, the spatial clustering pattern of China’s DRD levels is visualized in Figure 4. To objectively capture the evolution of DRD clustering over the study period, four representative years—2013, 2016, 2019, and 2021—were selected in accordance with major developmental phases, given the inherent time lag in spatial pattern transformation. As summarized in

Table 5. Local Moran’s I Index Measurement Results.

Year	H-H Pattern	L-H Type	L-L Type	H-L Type
2013	Beijing, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Henan	Tianjin, Shanxi, Jilin, Anhui, Jiangxi, Hunan, Guangxi, Hainan	Heilongjiang, Chongqing, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang	Inner Mongolia, Hubei, Guangdong, Sichuan
2014	Beijing, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Henan	Tianjin, Shanxi, Jilin, Anhui, Jiangxi, Guangxi, Hainan	Inner Mongolia, Heilongjiang, Hunan, Chongqing, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang	Hubei, Guangdong, Sichuan
2015	Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Henan	Tianjin, Shanxi, Anhui, Jiangxi, Guangxi, Hainan	Inner Mongolia, Jilin, Heilongjiang, Hunan, Chongqing, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang	Beijing, Liaoning, Hubei, Guangdong, Sichuan
2016	Beijing, Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Henan	Tianjin, Shanxi, Anhui, Jiangxi, Hunan, Guangxi, Hainan	Inner Mongolia, Jilin, Heilongjiang, Chongqing, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang	Liaoning, Hubei, Guangdong, Sichuan
2017	Beijing, Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Henan	Tianjin, Shanxi, Anhui, Jiangxi, Hunan, Guangxi, Hainan	Inner Mongolia, Liaoning, Jilin, Heilongjiang, Hubei, Chongqing, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang	Guangdong, Sichuan
2018	Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Henan	Tianjin, Anhui, Jiangxi, Hunan, Guangxi, Hainan	Shanxi, Inner Mongolia, Liaoning, Jilin, Heilongjiang, Chongqing, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang	Beijing, Hubei, Guangdong, Sichuan
2019	Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Henan	Tianjin, Shanxi, Anhui, Jiangxi, Hunan, Guangxi, Hainan	Inner Mongolia, Liaoning, Jilin, Heilongjiang, Chongqing, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang	Beijing, Hubei, Guangdong, Sichuan
2020	Beijing, Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Henan	Tianjin, Shanxi, Anhui, Jiangxi, Hunan, Guangxi, Hainan	Inner Mongolia, Liaoning, Jilin, Heilongjiang, Chongqing, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang	Hubei, Guangdong, Sichuan
2021	Beijing, Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Henan	Tianjin, Anhui, Jiangxi, Hunan, Guangxi, Hainan	Shanxi, Inner Mongolia, Liaoning, Jilin, Heilongjiang, Chongqing, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang	Hubei, Guangdong, Sichuan
2022	Beijing, Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Henan	Tianjin, Anhui, Jiangxi, Hunan, Guangxi, Hainan	Shanxi, Inner Mongolia, Liaoning, Jilin, Heilongjiang, Chongqing, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang	Hubei, Guangdong, Sichuan

, DRD exhibits sustained and structurally distinct spatial clustering. Its “core–periphery” configuration aligns with the trajectory of the global Moran’s I index. Eastern coastal provinces consistently demonstrate “high–high” clustering, forming a core radiation zone for DRD. These regions not only lead in their own development but also drive neighboring provinces through industrial collaboration, technology transfer, and knowledge sharing, as exemplified by the coordinated DRD networks in the Yangtze River Delta and Pearl River Delta regions. In contrast, western and northeastern regions are predominantly characterized by “low–low” clustering. These areas commonly face structural constraints such as underdeveloped digital infrastructure, limited digital industry support, and lower levels of human capital. Moreover, due to the similar developmental levels across neighboring provinces and a lack of high-value digital resource inflows, they remain trapped in a low-level equilibrium. Notably, the number and composition of “low–high” and “high–low” provinces have fluctuated over time, reflecting ongoing adjustments in local spatial structures. “Low–high” regions, situated on the periphery of high-value clusters, hold locational advantages that enable them to absorb spatial spillovers and achieve upward development mobility. In contrast, “high–low” clusters, despite their advanced local development, are constrained in their radiating influence due to the underdevelopment of surrounding areas. From a dynamic perspective, the number of provinces classified as “high–high” or “low–low” increased from 19 in 2013 to 22 in 2022, indicating a strengthening of spatial connectivity and a consolidation of regional differentiation. Without strengthened cross-regional coordination and targeted policy interventions, spatial disparities in DRD may further widen, leading to a Matthew effect where advanced regions grow stronger while lagging regions fall further behind. Overall, the current spatial structure reflects both the radiating influence of high-value regions and the structural constraints faced by low-value regions. This underscores how DRD remains deeply embedded within—and constrained by—China’s existing regional development patterns.

4. Mechanisms Influencing DRD Levels

4.1. Variable Selection

This study employs the regional innovation ecosystem theory as its analytical framework [20]. This theory emphasizes that regional development levels constitute complex systems with self-organizing and dynamic evolutionary characteristics [21,22]. Within specific geographic spaces, these systems emerge through the continuous exchange and interaction of material resources, information, and knowledge among diverse innovation actors, the innovation environment, and external demands. DRD is essentially a regional innovation practice that drives the digital transformation of rural economies, governance, and lifestyles within rural areas, with digital technology serving as the key enabling factor. The formation and evolution of its development level and spatial patterns are not merely outcomes of infrastructure investment or single policies, but rather the result of the coordinated evolution of supply, demand, and environment within the region. Based on this theory and drawing from related rural digitalization research, a three-dimensional analytical framework—comprising innovation supply, demand structure, and environmental support—is constructed to guide variable selection. The variable selection is shown in

Table 6. Variable Selection.

Dependent Variable	Comprehensive Score for DRD
Explanatory Variables	Urban–rural income disparity, ratio of per capita disposable income of urban residents to that of rural residents
	Level of Urbanisation, Ratio of Urban Population to Rural Population
	Level of scientific and technological innovation, ratio of local government expenditure on science and technology to general budget expenditure
	Level of education, ratio of local education expenditure to local general budget expenditure
	Industrial structure upgrading, characterised by the proportion of secondary and tertiary industry value-added in GDP
	Rural population size, measured by the proportion of rural residents relative to the total regional population
	Population age structure, measured using the rural old-age dependency ratio

.

On the innovation supply side, this study selects technological innovation level and industrial structure upgrading as key variables. The former reflects the intensity of a region’s investment in technology research, development, and application, and serves as a core driver for digital technology integration and local innovation [23]. The latter is measured by the proportion of secondary and tertiary industries in the economy, capturing the shift toward higher value-added sectors, which provides an essential industrial foundation and market demand for DRD [24].

From the perspective of demand and structure, we focus on rural population size and age structure. Population size determines the potential market scale and human capital reserves, forming the demand base for digital rural initiatives [24]. Population age structure captures the demographic composition of rural society, with potential mechanisms of impact including specific service demands driven by the “silver economy” and the intra-household “intergenerational transfer” effect [7].

Regarding environmental support, three variables are included: education level, urbanization rate, and urban–rural income disparity. Education level reflects human capital accumulation and foundational digital literacy [23]; urbanization rate describes the population shift between urban and rural areas, which may have dual effects—resource siphoning and spatial spillovers [23]; and urban–rural income disparity indicates the extent of balanced economic development, with underlying mechanisms that may be complex and multifaceted [25].

4.2. Model Selection

To determine the appropriate panel specification, we first conducted a Hausman test, which strongly favored the fixed-effects model over the random-effects model (χ² = 167.74, p < 0.01). We then performed fixed-effects F-tests to decide whether to include individual (province), time (year), or both types of fixed effects. The tests for individual fixed effects (F = 84.08, p < 0.01) and time fixed effects (F = 363.72, p < 0.01) were both statistically significant at the 1% level. Therefore, a two-way fixed-effects model—controlling for both province and year—is adopted in the subsequent spatial regression analysis.

As reported in

Table 7. LM Test.

	Name	Statistic	p-Value
Spatial Error Model	LM Test RLM test	91.286 86.890	0.000 0.000
Spatial Lag Model	LM test RLM test	9.545 5.149	0.002 0.023

, the Lagrange Multiplier (LM) tests for the spatial error model and spatial lag model yield statistics of 91.286 and 86.890, respectively, and their robust counterparts (RLM) are 9.545 and 5.149. All are statistically significant at the 1% or 5% level, confirming the presence of both spatial error dependence and spatial lag dependence in DRD. In light of the coexistence of these spatial effects, the spatial Durbin Model (SDM) is adopted for regression analysis, as it accommodates spatial lags of both the dependent and independent variables and nests the spatial error and spatial lag models as special cases.

To further validate the choice of the spatial Durbin Model (SDM), Wald and Likelihood Ratio (LR) tests were performed. The results, summarized in

Table 8. Wald Test and LR Test.

	Wald Test	LR Test
Can it be reduced to SAR?	43.03 ***	129.57 ***
Can it be reduced to SEM?	46.65 ***	10.14 ***

Note: *, **, and *** denote significance at the 10%, 5%, and 1% statistical levels, respectively.

, reject the null hypotheses that the SDM can be simplified to a spatial lag model (SAR) or a spatial error model (SEM) at conventional significance levels. This confirms that the SDM provides a more general specification and that restricting it to either a pure spatial lag or pure spatial error form would result in a misspecified model.

4.3. Analysis of Influencing Mechanisms

The regression results (

Table 9. Spatial Durbin Model Regression Results.

Variable Name	Coefficient	Variable Name	Coefficient
Urban–Rural Income Gap	−0.3029 (0.2282)	Wx Urban–rural income disparity	0.5953 (1.6222)
Level of Urbanisation	0.1436 (0.1945)	Wx Urbanisation Level	−0.5422 (1.1628)
Level of Technological Innovation	2.7909 *** (0.3626)	Wx Level of Technological Innovation	7.1008 *** (2.5995)
Industrial Structure Upgrading	0.1896 (0.2877)	Wx Industrial Structure Upgrading	5.7718 *** (2.0258)
Educational Attainment	0.1983 (0. 1989)	Wx Educational Attainment	−2.8498 * (1.5453)
Rural Population Size	0.2762 *** (0.0757)	Wx Rural Population Size	4. 1654 *** (0.5879)
Population Age Structure	0.0047 *** (0.0016)	Wx Population Age Structure	0.0344 *** (0.0099)
p	−1.3209 *** (0.2704)	R2	0.0185

Note: *, ** and *** denote significance at the 10%, 5% and 1% statistical levels, respectively; figures in brackets represent standard errors; individual time effects have been controlled for.

) indicate that the spatial autoregressive coefficient ρ is −1.3209, passing the 1% significance level test. This reveals a significant negative spatial spillover effect in DRD, consistent with the findings of Li Yanling et al. [23]. Regarding local effects, technological innovation level (2.7909) exhibits the most significant positive driving force, indicating that technological advancement serves as the core engine propelling DRD. Its high coefficient reflects the crucial role of knowledge spillovers and innovation clustering in regional development. Rural population size (0.2762) and population age structure (0.0047) also demonstrate significant promotional effects. An expanding rural population provides a broader market and human capital foundation, while shifts in age structure generate demand for age-friendly digital services, compelling service upgrades. Their relatively lower coefficients suggest these factors primarily exert influence through scale accumulation and incremental adjustments, falling short of the structural transformation driven by technological innovation. From the perspective of spatial spillover effects, technological innovation (7.1008), industrial structure upgrading (5.7718), and rural population size (4.1654) all exhibit significant positive spatial spillovers. Their coefficients generally exceed local effects, reflecting these factors’ strong spillover attributes. They can significantly drive the development of surrounding areas through regional collaboration, factor mobility, and experience sharing. Specifically, enhanced technological innovation in a region not only advances local DRD but also positively radiates to adjacent areas. Economic restructuring and resource allocation efficiency gains from industrial upgrading may stimulate DRD initiatives in surrounding regions through interregional economic linkages. Areas with larger rural populations may drive DRD-related industries in neighboring regions via labor mobility and market demand spillovers. Notably, the spillover effects of technological innovation and industrial upgrading are the strongest, indicating that technology and industry play a leading role in cross-regional linkages. However, the spatial lag term for education levels (−2.8498) exhibits a significant negative spillover, possibly due to the competitive or siphoning effects of high-quality educational resources across regions, which may inhibit the accumulation and enhancement of educational resources in neighboring areas. Overall, DRD benefits from endogenous drivers like local innovation and demographic structure while being profoundly influenced by interregional factor flows and spatial interactions. Policy attention must prioritize cross-regional coordination and structural integration.

According to the spatial Durbin model decomposition effects (

Table 10. Spatial Durbin Model Effect Decomposition.

Variable Name	Direct Effect	Indirect Effect	Total Effect
Urban–Rural Income	−0.3350	0.3559	0.0209
Gap	(0.2583)	(0.7501)	(0.6614)
Level of urbanization	0.1682	−0.3130	−0.1448
	(0.2260)	(0.6740)	(0.5512)
Level of Technological	2.5647 ***	1.6057	4.1704 ***
Innovation	(0.4565)	(1.2614)	(1.1330)
Industrial Structure	−0.0213	2.6277 ***	2.6064 ***
Upgrading	(0.2718)	(0.9391)	(0.9735)
Level of education	0.3388 *	−1.4208 *	−1.0820
	(0. 1957)	(0.7645)	(0.7277)
Rural Population Size	0.1009	1.8512 ***	1.9521 ***
	(0.0750)	(0.3765)	(0.3592)
Population Age	0.0035 **	0.0135 **	0.0170 ***
Structure	(0.0018)	(0.0054)	(0.0049)

Note: *, ** and *** denote significance at the 10%, 5% and 1% statistical levels, respectively; standard errors are shown in parentheses; individual time effects have been controlled for.

), the level of DRD is differentially influenced by both direct and indirect effects of various factors. Regarding direct effects, technological innovation levels (2.5647) and educational attainment (0.3388) significantly promote local development. The high coefficient for technological innovation further confirms its pivotal role as the core engine in DRD. while the direct effects of rural population size (0.1009) and population age structure (0.0035) are relatively weaker but still significant, reflecting the supportive influence of demographic foundations, primarily through long-term structural adaptation and scale accumulation. Regarding indirect effects: Industrial structure upgrading (2.6277) exhibits the strongest indirect effect, with its coefficient significantly exceeding its direct effect. This indicates that industrial synergy and regional linkage play a dominant role in driving cross-regional development. Rural population size (1.8512) also exhibits a significant indirect effect, reflecting its spillover impact on surrounding areas through labor mobility and demand spillovers. Conversely, educational attainment shows a pronounced negative effect (−1.4208), consistent with its negative coefficient in spatial regression, further confirming the competitive nature and brain drain phenomenon of educational resources across regions. Differences in coefficient magnitudes reflect varying influence attributes: knowledge-intensive factors like technological innovation exhibit strong direct effects, yet their indirect effects remain limited due to institutional and geographical constraints on technology diffusion. Conversely, factors such as industrial structure and population possess greater mobility, facilitating regional synergies more readily. Overall, DRD is shaped by both local innovation drivers and regional factor linkages. Policy efforts should integrate local capacity building with cross-regional collaborative governance, particularly focusing on establishing mechanisms for regional coordination in educational resource allocation and industrial linkage.

This study reveals that China’s DRD exhibits significant negative spatial spillover effects, exposing a “siphoning” competitive dynamic between regions amid scarce digital resources. This phenomenon parallels Park’s theory of the “digital vicious cycle,” where digital exclusion and social exclusion mutually reinforce each other. However, the Chinese context more prominently features structured regional competition driven by both policy initiatives and market mechanisms. Unlike the positive technology diffusion patterns commonly observed in studies of Chile and Italy [1,2], China exhibits an absorption effect where developed regions draw in surrounding factors, reflecting the heightened risk of regional imbalances that developing countries may face during rapid digitalization. Among core drivers, the roles of technological innovation and population size/structure validate the “supply–demand dual-engine” logic, aligning with the “application layer barriers” phase proposed by Ferrari et al. [3]. Industrial upgrading’s impact through spatial spillovers underscores regional industrial synergy, resonating with Visvizi & Lytras’ [5] systemic integration perspective on “smart villages.” Furthermore, the dual effects of local promotion and neighborhood suppression observed in educational attainment levels further reveal the complexity of human capital mobility and the necessity of policy coordination within the context of regional development imbalances.

4.4. Robustness Test

To verify the sensitivity of the research conclusions to measurement methods, this paper recalculates the comprehensive score for DRD using the Critic objective weighting method and re-estimates the spatial Durbin model with this score as the dependent variable. The robustness test results are shown in

Table 11. Robustness Test.

Variable Name	Coefficient	Variable Name	Coefficient
Urban–Rural Income Gap	−2.0296 ** (0.8918)	Wx Urban–rural income disparity	9.9550 (6.3167)
Level of Urbanisation	1.9125 ** (0.7572)	Wx Urbanisation Level	5.9847 (4.6771)
Level of Technological Innovation	2.7909 *** (0.3626)	Wx Level of Technological Innovation	10.0171 (9.7639)
Industrial Structure Upgrading	11.7251 *** (1.4146)	Wx Industrial Structure Upgrading	24.6355 *** (7.8647)
Educational Attainment	0.8679 (1.1237)	Wx Educational Attainment	−13.4848 ** (6.0566)
Rural Population Size	1.2205 *** (0.2942)	Wx Rural Population Size	15.7694 *** (2.3599)
Population Age Structure	0.0184 *** (0.0060)	Wx Population Age Structure	0.0800 ** (0.0391)
p	−0.9413 *** (0.2560)	R2	0.0482

Note: *, ** and *** denote significance at the 10%, 5% and 1% statistical levels, respectively; standard errors are shown in parentheses; individual time effects have been controlled for.

. The test reveals that the key findings remain highly robust. The spatial autoregressive coefficient ρ shifted from −1.3209 to −0.9413, remaining significantly negative at the 1% level. This indicates that neighborhood suppression is not attributable to specific weighting methods, confirming the genuine existence of negative spatial spillover effects. The direction and significance of core local drivers remained fully consistent. The direct effects of technological innovation level, rural population size, and population age structure all retained significant positive impacts at the 1% level, confirming these three factors as intrinsic stable drivers of DRD. The spatial spillover mechanism of key variables is robust. The spatial lag coefficients for industrial structure upgrading and rural population size are both significantly positive at the 1% level, reaffirming the mechanism through which these factors positively influence surrounding areas via spatial spillover. Although coefficient estimates vary due to differences in weighting principles, the direction of influence, statistical significance, and spatial effect characteristics of core variables remain unchanged, ensuring strong reliability of the research conclusions.

5. Conclusions and Policy Implications

This study utilizes panel data from China’s 31 provinces spanning 2013–2022 to construct an evaluation framework encompassing four dimensions: digital infrastructure, economic digitization, governance digitization, and digital living. By integrating entropy analysis, the Dagum Gini coefficient, the Moran index, and spatial Durbin models, it systematically examines the spatiotemporal evolution characteristics and influencing mechanisms of China’s DRD. Key findings include:

First, the development level has generally evolved through three stages: “slow growth—rapid advancement—fluctuating rise.” The 2018 “Digital Rural Development Strategy” marked a critical policy turning point. Regional disparities are pronounced. Eastern regions leverage digital industry clusters and human capital advantages to achieve dual empowerment through policy dividends and market effects. In contrast, central, western, and northeastern regions face geographical constraints and rigid industrial structures, leading to low-level lock-in. This divergence stems not merely from developmental stage differences but from imbalanced resource allocation and path dependence.

Second, spatially, there exists a pronounced positive agglomeration effect and negative spatial spillover effect (ρ = −1.3209). At its core, this reflects regional competition driven by the scarcity of digital factors. The “siphon effect” exerted by core eastern provinces like Guangdong and Zhejiang on surrounding areas’ digital talent and capital has both spurred localized efficiency gains and amplified systemic imbalance risks. Robustness tests confirm this characteristic, indicating it is not a measurement bias but an inevitable contradiction in the deepening process of DRD. The solidification of “high–high” and “low–low” agglomeration patterns further warns that without effective intervention, DRD may fall into a Matthew Effect where the strong grow stronger and the weak grow weaker.

Third, regarding the influencing mechanisms: technological innovation levels, rural population size, and population age structure serve as core local drivers, confirming the importance of technological empowerment and demand support; industrial structure upgrading exerts its influence through positive spatial spillover, highlighting the value of regional coordination; Education exhibits dual effects—positive locally but negative across sectors—revealing the zero-sum nature of human capital competition. This complex relationship, previously understudied, offers critical policy implications.

Based on these findings, the following policy recommendations are proposed:

First, implement differentiated regional interventions. Targeted measures should address the root causes of regional disparities, leveraging negative spatial spillovers and “high–high”/”low–low” clustering patterns. For core provinces like Guangdong and eastern Zhejiang, advance policies such as tax breaks for technology spillovers and cross-regional joint fund initiatives. Provide targeted infrastructure upgrades for “low–high” provinces like Anhui and Jiangxi. For western locked-in regions like Tibet and Guizhou, subsidize network construction and maintenance while supplementing with e-commerce support and digital literacy training. Grant subsidies for smart agriculture projects in Northeast China’s three provinces, while strengthening cross-regional coordination in central provinces to balance equity and efficiency through benefit compensation.

Second, strengthen core drivers through precise alignment. Align policies with mechanisms based on empirical evidence. For technological innovation as a core driver, establish special funds to support R&D and commercialization. Leverage rural population size and age structure advantages to cultivate digital consumer markets and advance age-friendly renovations. Balance the dual effects of education through two-way talent incentives, eliminating blanket interventions to enhance policy fit.

Third, promote the establishment of spatial coordination mechanisms. Based on spatial interconnectivity, mitigate negative spillovers while amplifying positive effects. Form interprovincial collaborative alliances to coordinate factor mobility and industrial cooperation; integrate digital infrastructure development with unified standards and data interfaces; build a national data-sharing platform to enhance positive spillovers from industrial upgrading; eliminate local protectionism through unified market rules, reduce cross-regional factor mobility costs, and optimize spatial interaction patterns.

Fourth, strengthen policy trade-offs and dynamic risk management. Establish a closed-loop control system to prevent unintended consequences. Implement a national coordination mechanism with interest compensation to balance regional benefits; introduce project evaluation and subsidy exit mechanisms to avoid resource waste and market distortions; build a dynamic monitoring platform integrating data on agglomeration types and spillover effects across all 31 provinces. Conduct biennial policy evaluations and adjustments to anticipate new imbalance risks, ensuring policies precisely align with development needs.

China’s DRD practices offer a reference model for the world, particularly developing nations. In public services, the “Internet + Healthcare” model leverages remote consultation and diagnosis to overcome geographical barriers, effectively alleviating the supply–demand imbalance of rural medical resources. In education, an “Internet + Education” system bridges the urban–rural development gap by sharing high-quality digital resources and utilizing remote teaching methods. In governance, internet platforms showcase regional development achievements comprehensively, significantly enhancing government transparency and public participation.

We acknowledge this study’s limitations: the indicator system does not fully encompass soft dimensions like digital security and digital literacy; provincial-level data struggles to capture county-level variations and micro-level transmission mechanisms; and spatial econometric analysis may face issues of bidirectional causality and spatial interaction-induced endogeneity. Although we employed a dual fixed-effects model for provinces and time to control for regional heterogeneity and temporal trends, limitations in research design and data prevented the use of stricter methods like instrumental variables to fully eliminate endogeneity. Future research could expand indicator dimensions, incorporate county-level microdata, and integrate qualitative analysis. Furthermore, leveraging exogenous policy shocks to construct quasi-natural experiments or employing instrumental variable methods could strengthen causal inference. This approach would provide more refined empirical evidence and policy recommendations for Digital Rural Development.