Research on the Impact Mechanism and Spatial Spillover Effect of Digital Economy on Rural Revitalization: An Empirical Study Based on China’s Provinces

Abstract

Digital Road is the new direction of rural revitalization, and the digital economy empowers rural revitalization as a new driving force for the sustainable development of agriculture and rural areas. Currently, the studies on the digital economy and rural revitalization are focused on a certain dimension and lack research on the influence mechanism of the digital economy driving rural revitalization and their spatial effects under the same framework. Based on the panel data of 31 provinces in China from 2013 to 2021, the paper explores the influence mechanism of rural revitalization driven by the digital economy and their spatial spillover effects by using the benchmark regression model, intermediary effect model and spatial Durbin model, and draws the following conclusions: (1) Digital economy can significantly promote rural revitalization. (2) Digital economy can significantly promote the revitalization of rural areas by transforming industrial structure and rural entrepreneurship. (3) The development of the digital economy will significantly promote the revitalization of villages in the region and the level of rural revitalization in neighbouring areas. After the endogenous treatment and robustness test, the above conclusions remain valid.

1. Introduction

The countryside is the ”stabilizer” and ”reservoir” for the sustainable development of China’s economy. Only by achieving comprehensive development of rural areas, narrowing the gap between urban and rural areas, and achieving integrated development between urban and rural areas can the overall economic level of the country be improved. Rural revitalization refers to vigorously developing rural industries, increasing farmers’ income, and making rural areas ecologically livable and demonstration areas for economic development. In rural revitalization, the role of the digital economy cannot be ignored. The digital economy has been very popular in recent years. It relies on data as the main element, uses high-tech to change the operation mode of traditional industries, and efficiently and conveniently promotes economic development. Therefore, we can fully utilize the digital economy to comprehensively promote rural revitalization, achieve modernization of rural industries, breed more emerging formats, bring development opportunities to rural areas, and promote employment for rural populations. Especially after the COVID-19 epidemic, it is essential to explore the mechanism of the digital economy driving rural revitalization. Therefore, the paper researches this issue, hoping to provide reference suggestions for rural revitalization.

The joint linkage between the digital economy and rural revitalization has become a hot issue, attracting continuous attention. In recent years, many scholars have conducted research on the issue of the digital economy: (1) In terms of conceptual connotation. Bukht and Heeks [1] define the digital economy as “the part of economic output that comes entirely or mainly from digital technology, and its business model is based on digital goods or services”. Cai Yue Zhou [2] believes that the digital economy is a new economic form of digital technology and related specific industrial sectors and the integration of digital elements (or information elements). Ouyang Rihui [3] divides the digital economy into five levels: new infrastructure layer, new production factor layer, new ecological environment layer, new entity economy layer and new economic form layer from the five mechanisms of “interconnection, elements, integration, transformation and innovation”, the concept and connotation of the digital economy are constructed. (2) In terms of digital economy measurement. Margherio et al. [4] discussed measuring the digital economy from the e-commerce perspective. Haltiwanger & Jarmin [5] and Moulton [6] discussed the calculation of the contribution of the digital economy to GDP from the aspects of data collection, survey statistics, capital stock estimation, price index adjustment and so on. Cai Yuezhou [2] calculated the scale of China’s digital economy based on the theory of value added. Given the difficulty of measuring the scale of the digital economy, some scholars have proposed to measure the level of the digital economy by constructing evaluation indicators. For example, Zhang Wang and Bai Yongxiu [7] selected the indicators of the digital economy from three aspects: digital infrastructure, agricultural digitization, and agricultural digital industrialization, and used the entropy method to measure the development level of the digital economy. Tian Ye et al. [8] select the digital economy index from two aspects of the digital basic level and digital application and uses the entropy method to measure the digital economic index. (3) Empirical studies on the digital economy mainly focus on digital economy and innovation [9], employment [10], productivity, and high-quality economic development. Explore the mechanism of the digital economy in various aspects [11,12,13,14].

The research on rural revitalization is also relatively rich: (1) Research on the evaluation index system of rural revitalization, most scholars start from five dimensions: industrial prosperity, ecological livability, rural civilization, effective governance, and affluent life, and build an evaluation index system of rural revitalization [15,16,17]. (2) Scholars qualitatively analyze the role of different subjects in rural revitalization from the perspectives of farmers [18], cultural and tourism enterprises [19], rural governance [20], social capital [21] and reciprocal symbiosis of individual and collective [22]. (3) In the research on the path of rural revitalization, scholars have theoretically expounded the multiple driving paths of rural revitalization, such as rural e-commerce path [23], rural financial development path [24], independent entrepreneurship path [25], rural tourism path [26] and science and technology support path [27].

Regarding the research on the impact of the digital economy on rural revitalization, it is found that the digital economy plays an essential role in promoting agricultural quality and efficiency, rural economic development, rural construction, and governance: (1) Digital economy can promote the development of rural economy. By innovating the development model of the rural economy, the digital economy weakens the barriers to urban-rural dual structure, stabilizes agricultural production, reduces poverty, and promotes the sustainable development of rural areas, thus promoting the high-quality development of the rural economy [28,29]. (2) Digital economy can promote rural construction and governance. Chugunov A V et al. [30] believe that information and communication technology is integrated into the construction of intelligent villages to improve the quality of life of farmers and advocate the use of mobile technology to make e-government services closer to rural residents; Shen Feiwei and Yuan Huan [31] expounded the practical logic of digital rural governance in big data era by using independent governance theory and put forward the optimization strategy of digital rural governance. (3) Digital economy can promote the quality and efficiency of agriculture. Through the technological synergy of scientific and technological innovation, the integrated development of the digital economy and rural industry can give full play to the multiplier and spillover effects in information technology innovation and obtain higher-quality output [32]. Kupriyanova M et al. [33] believe that rural areas suffer from serious digital discrimination and should strengthen the construction of rural digital facilities, reduce the digital gap between urban and rural areas, and alleviate the gap in the quality of life between urban and rural areas, to enhance the competitiveness and profitability of the agricultural sector.

Although the digital economy and rural revitalization are relatively rich, there are still the following shortcomings:

(1)

Most existing theoretical analyses focus on the concept definition, connotation interpretation, policy design and implementation path of the digital economy and rural revitalization. Still, there is a lack of empirical research on them.

(2)

The few existing empirical studies only focus on the local influence of a particular dimension but do not consider the relationship between the two from a global perspective, let alone an in-depth influence mechanism.

(3)

Existing studies lack spatial effect studies on the digital economy and rural revitalization.

Given the above shortcomings, the paper attempts to construct the evaluation index system of the digital economy and rural revitalization, respectively and empirically discusses the relationship between them. The possible marginal innovations are as follows:

(1)

For the first time, the paper brings the digital economy and rural revitalization into the same empirical analysis framework and explores whether the digital economy drives rural revitalization.

(2)

The paper explores the impact mechanism of digital economy-driven rural revitalization using a mediating effect model by selecting industrial structure transformation and rural entrepreneurship as mediating variables.

(3)

The spatial spillover effects of the digital economy and rural revitalization are explored using the spatial Durbin model.

The remaining parts of the paper are organized as follows: Section 2 performs a theoretical analysis and raises some hypotheses; Section 3 presents the Research method and data resource; Section 4 presents the results and analysis, beginning with the time and region double fixed effect model is used to analyze the relationship between the digital economy and rural revitalization, and by using the intermediary effect model, the paper probes into the mechanism of digital economy promoting rural revitalization; Section 5 present the spatial spillover effect. The spatial spillover effects of the digital economy and rural revitalization are explored using the spatial Durbin model. Section 6 summarizes this study and raises suggestions against the conclusions.

2. Theoretical Analyses and Research Assumptions

The five levels of “industrial prosperity, ecological livability, rural civilization, effective governance, and prosperous life” are commonly used and recognized indicators to measure rural revitalization. Therefore, the paper explains the direct impact of the digital economy on rural revitalization from the perspectives of improving the development level of the digital economy, promoting industrial prosperity and effective governance, and helping to achieve ecological livability, prosperous life, and developing rural civilization. The digital economy can promote the prosperity of rural industries and effective governance. Digital inclusive finance is an essential component of the digital economy, making it easier for farmers to access agricultural funds. And the low-cost and widely covered financial credit system injects vitality into rural development [34].

Moreover, combining the digital economy and traditional industries can increase opportunities to optimize the internal structure of various rural industries. The Internet is an essential carrier of information dissemination in the digital economy, which can reduce information asymmetry between different regions and subjects [35]. On the one hand, it can reduce the cost of searching for information, and on the other hand, it can provide power for information transmission and accelerate the cycle. By using digital technology to govern rural areas, the openness of rural affairs information can be improved, the threshold for villagers to access information resources can be lowered, and residents can consciously participate in rural public affairs governance. At the same time, the digitalization of rural governance facilitates villagers’ supervision of village affairs, finance, and other aspects, enhances the transparency of village affairs, and makes more residents participate in rural governance [36]. The digital economy can play a great role in realizing rural ecological livability, rural civilization, and farmers’ affluence. Compared with traditional industry, the pollution caused by the digital economy to the environment is very small. The digital economy will gradually infiltrate the traditional industry and the strong support of digital inclusive finance to the green and low-carbon industries, which greatly reduces the survival and development space of enterprises with high pollution emissions, and then improves the rural environment. At the same time, using a digital economy can improve the efficiency of resource utilization and transformation efficiency, thus promoting the transformation of green and low-carbon agriculture [37]. The development of the digital economy has brought many kinds of network sales channels, such as rural e-commerce, live broadcast with goods, and so on, which provides a way for farmers to increase their income by selling characteristic agricultural products. And the financial credit system under digital inclusive finance can help the countryside to form a trustworthy and responsible spirit. As mentioned earlier, the development of e-commerce helps farmers get rich. In addition, the digital economy can improve the transaction rate of the market, overcome the position obstacles of being in the mountains and isolated from the outside world, and improve the effective connection between the broad market and farmers. To sum up, “industrial prosperity, ecological livability, rural civilization, effective governance and affluent life” should be incorporated into the comprehensive system of rural revitalization. Based on this, the following hypothesis is put forward.

Hypothesis 1 (H1).

The digital economy can significantly promote the revitalization of rural areas.

The digital economy can promote rural entrepreneurship: First, the digital economy can broaden access to information for entrepreneurs by providing an information exchange platform so that rural personnel can easily and timely obtain entrepreneurial information and grasp accurate market trends. At the same time, applying digital technology can provide more advanced financial services, thus reducing entrepreneurial costs and stimulating entrepreneurial passion and activity. Second, digital technology can promote consumption’s diversified and personalized needs, give birth to many entrepreneurial opportunities, and provide a good foundation and market environment for entrepreneurial activities [38]. Third, the digital economy, through digital technology, builds a convenient learning platform through the online platform to search for entrepreneurial learning and training content needed by entrepreneurs, improve the professional literacy of entrepreneurs, and significantly reduce the time cost and material cost of entrepreneurship [39]. Rural entrepreneurship can promote rural revitalization because returning entrepreneurship is essential for sustainable industrial revitalization. Returning entrepreneurs fully combine the advantages of local resources to guide capital, technology, and other factors to gather in rural areas, continue to enhance the intensity of entrepreneurship in rural areas, expand employment channels for rural residents, and promote regional economic development [39], thus enabling rural revitalization. Based on this, the following hypothesis is put forward.

Hypothesis 2 (H2).

The digital economy can promote rural entrepreneurship and then drive rural revitalization.

The fact that the digital economy can promote industrial transformation and upgrading is mainly reflected in the following aspects: First, the digital economy can promote the adjustment of industrial structure in rural areas and establish a new modern rural industrial model. For example, it can build a “Internet+” industrial model, deeply explore the advantages of local resources, landscape, and culture, develop leisure tourism with local characteristics, promote upgrading industrial structure, and realize the integrated development of primary, secondary, and tertiary industries [40]. Second, the wide application of digital technology can accelerate the optimization and upgrading of industrial structure, eliminate backward industries, and give birth to new industries, business types and business models [39]. Third, the digital economy provides convenient and low-cost learning opportunities and platforms for rural personnel. On the one hand, it can cultivate new farmers so that pillar industries can continue to develop, and high-quality crops can be developed at a high quality and high level. High-quality crops can be sold at a better price to avoid hurting farmers.

On the other hand, through the Internet, farmers can broaden their horizons and no longer rely solely on agriculture for a living. They can develop the service industry, broaden their business scope and channels, increase their income, and even rely on e-commerce platforms, live broadcast with goods, and sales of characteristic agricultural products. All these will change the previous single industrial structure in rural areas. So that its development model is richer, and the industrial structure can be transformed and upgraded. The upgrading of the industrial structure can provide more employment opportunities for the rural areas, and the income level will be raised accordingly; farmers can also take advantage of this to achieve employment on their doorstep so that all kinds of expenses and expenses arising from going out to work can be reduced, life will become rich, and the quality of life will be improved, thus promoting the revitalization of the countryside. Based on this, the following hypothesis is put forward.

Hypothesis 3 (H3).

The digital economy can promote the transformation of industrial structures and revitalize rural areas.

3. Research Method and Data Resource

3.1. Benchmark Regression Model

To investigate the impact of the development level of the digital economy on rural revitalization, the models are established as follows:

$$R_{it}={\alpha }_0+{\alpha }_1{D}_{it}+{\alpha }_c{C}_{it}+{\mu }_i+{\nu }_t+{\epsilon }_{it}$$

(1)

where $R_{it}$ represents the level of the rural revitalization of the ith province in tth year; $D_{it}$ represents the digital economy level of the ith province in tth year; $C_{it}$ represents the set of other control variables affecting; ${\mu }_i$ and ${\nu }_t$ represent region and time fixed effect respectively; ${\epsilon }_{it}$ represents random disturbance term.

3.2. Mediating Effect Model

Mediating effect means that if there is a variable M, if the variable X can affect the variable M, and the variable M will affect the variable Y, that is, the variable X affects the variable Y through the influence of the variable M, then M is the intermediary variable. The model is as follows:

$$M_{it}={\beta }_0+{\beta }_1{D}_{it}+{\beta }_C{C}_{it}+{\mu }_i+{\nu }_t+{\epsilon }_{it}$$

(2)

$$R_{it}={\gamma }_0+{\gamma }_1{D}_{it}+{\gamma }_2{M}_{it}+{\gamma }_C{C}_{it}+{\mu }_i+{\nu }_t+{\epsilon }_{it}$$

(3)

where $M_{it}$ indicates the mediating variable level of the ith province in tth year, including rural entrepreneurship and industrial structure transformation, and the meaning of other variables is the same as above.

3.3. Description of Data and Variables

3.3.1. Explained Variable

The five-in-one structure of “Industrial prosperity, ecologically livable, civilized rural style, effective governance and affluent life” has been recognized to reflect the level of rural revitalization very well. At the same time, it points out how to realize the prosperity and development of rural agriculture and increase farmers’ income. Rural revitalization is a dynamic process of sustainable development. We should examine the external factors that affect rural revitalization and explore the internal driving force driving its development. Therefore, we cannot simply measure the level of rural revitalization through a single index. According to the actual situation of rural areas, we should analyze rural revitalization in an all-round and multi-level way. Referring to He Leihua et al. [41], the paper selects industrial prosperity, ecological livability, rural civilization, effective governance, and affluence as the first-level index. And in each dimension, two indicators are selected as secondary indicators to construct the evaluation index system of rural revitalization. Please refer to

Table 1. List of variable indicators.

Variable Type	Variable Meaning	Primary Index	Secondary Index	Unit
Explained variable (Rural)	Rural revitalization	Industrial prosperity	Per capita output value of agriculture, forestry, animal husbandry and fishery	Ten thousand yuan per person
			Total power of agricultural machinery per capita	Kilowatts per person
		Ecologically livable	Per capita park green space area	Square meters per person
			Number of public toilets per 10,000 people	Per ten thousand people
		Rural civilization	Per capita cultural and entertainment consumption expenditure of farmers	Yuan/person
			The proportion of rural cable radio and television users	%
		Effective Governance	Number of villagers’ committees per unit area	Per ten thousand square kilometres
		Affluent life	Income ratio of rural residents to urban residents	%
			Engel coefficient of rural residents	%
Explanatory variables (Digital)	Digital Economy	Traditional Digital Infrastructure	Number of Internet broadband access ports	Ten thousand
			Number of Internet Broadband access users	Ten thousand households
		New Digital Infrastructure	Mobile phone penetration rate	One per hundred people
			Mobile phone base station	Ten thousand
Mediating variable	—	Rural entrepreneurship (Entrepre)	The ratio of rural self-employed to the rural population	%
		Transformation of industrial structure (Industrial)	The sum of the proportion of people employed in the second and third industries	%
Control variable	—	Population structure (Elderly)	Elderly dependency ratio	%
		Enterprise structure (Indu-struct)	Number of industrial enterprises above scale	-
		Industrial structure (Second)	The proportion of secondary production	%
		Degree of openness (Open)	The proportion of total imports and exports of goods	%

for details.

3.3.2. Core Explanatory Variable

Referring to Huang Qunhui et al. [42] and Meng Weifu et al. [39], this paper determines the measurement indicators of the digital economy from two aspects, the traditional digital infrastructure and the new digital infrastructure. Regarding traditional digital economic facilities, the number of Internet broadband access ports and Internet broadband access users are selected. Mobile phone penetration and the number of mobile phone base stations are selected as measurement indicators in the new digital infrastructure.

3.3.3. Control Variables

Regarding He Leihua et al. [41] and Tian Ye et al. [8], the paper selects four indicators: openness, industrial structure, population structure and enterprise structure as control variables, which are measured by the proportion of total import and export of goods to GDP, the proportion of secondary industry, the dependency ratio of the elderly and the number of industrial enterprises above scale.

3.3.4. Mediating Variable

Regarding the index selection methods of Zhao Tao et al. [14], Yuan Fang and Shi Qinghua [43], this paper selects rural entrepreneurship and the transformation of industrial structure as intermediary variables to explore the mechanism of the impact of the level of the digital economy on the level of rural revitalization, which is measured by the ratio of rural individual employment to the number of rural people at the end of the year and the proportion of employment in secondary and tertiary industries. The variables and indicators involved in the above are detailed in Table 1.

3.3.5. Data Sources

Given the availability and continuity of the data, this paper uses the panel data of 31 provinces except Hong Kong, Macao, and Taiwan from 2013 to 2021. The data mainly come from the China Statistical Yearbook, the China Rural Statistical Yearbook, the Ministry of Industry and Information Technology, the official website of China’s agricultural and rural areas, the National Bureau of Statistics of China, and the Local Bureau of Statistics.

3.3.6. Entropy Weight Method

Because there are great differences in the dimensions of each index, the entropy method is used to measure the comprehensive index of rural revitalization and the digital economy. The entropy weight method calculates the weight through information entropy and then weighted average to get the final comprehensive index. Referring to the research of Jiang Xuebin et al. [44], the calculation method of entropy method is as follows:

(1)

Index selection: assuming that there are t years, i years, and j indicators, $a_{ijt}$ is the j index value of province i in the t year.

(2)

Standardization of indicators: because different indicators have different dimensions, they need to be standardized.

$$\mathrm{Positive} \mathrm{index}: a_{ijt}^{\prime }=\frac{a_{ijt}-\min a_{ijt}}{\max a_{ijt}-\min a_{ijt}}$$

$$\mathrm{Negative} \mathrm{index}: a_{ijt}^{\prime }=\frac{\min a_{ijt}-a_{ijt}}{\max a_{ijt}-\min a_{ijt}}$$

(3)

To avoid the logarithm being meaningless when calculating the entropy $e_j$, the data is translated: $x_{ijt}=a_{ijt}^{\prime }+0.0001$

(4)

Calculate normalized p value: $p_{ijt}=\frac{x_{ijt}}{{\displaystyle \sum _t{\displaystyle \sum _j{x}_{ijt}}}}$

(5)

Calculate the entropy value of the jth indicator: $e_j=-k{\displaystyle \sum _t{\displaystyle \sum _i{p}_{ijt}\ln p_{ijt}}}$, where $k=\frac{1}{\ln (m\cdot n)}$, m represents the number of provinces, n represents the number of years.

(6)

Calculate the information utility value of the jth indicator: $g_j=1-e_j$.

(7)

Calculate the weight of each indicator: $w_j=\frac{g_j}{{\displaystyle \sum _j{g}_j}}$

(8)

The method of weighted summation of weight and index is used to calculate the comprehensive evaluation index: $s_j={\displaystyle \sum _j{w}_j\cdot x_{ijt}}$

In this paper, the rural revitalization index and digital economy index calculated by entropy weight are shown in Figure 1 and Figure 2.

Table 2. Descriptive statistical analysis of the used variables.

VarName	Obs	Mean	SD	Min	Median	Max
Rural	279	0.292	0.097	0.115	0.277	0.599
Digital	279	0.256	0.176	0.025	0.213	0.934
Elderly	279	15.520	4.198	7.010	14.850	26.700
Industrial	279	12,409	13,814	70	6426	66,307
Second	279	0.411	0.082	0.158	0.420	0.573
Open	279	0.036	0.038	0.001	0.021	0.206
Entrepre	279	0.074	0.051	0.006	0.064	0.311
Indu-struct	279	0.697	0.139	0.407	0.677	1.182

reports the descriptive statistics of the original variables. Table 2 shows the minimum value of the explained variable rural revitalization level is 0.115, and the maximum value is 0.599. There are significant differences in rural development among different regions. The minimum value of the core explanatory variable digital economic level is 0.025, and the maximum value is 0.934, indicating that the digital economic level is also quite different from place to place. From the perspective of control variables, there are also significant differences in population structure, enterprise structure, industrial structure, and openness in different regions.

3.3.7. Spatial Durbin Model

The spatial Durbin model is as follows:

$$\ln rura{l}_{it}={\alpha }_0+{\beta }_1\ln digita{l}_{it}+{\rho }_1{w}_{ij}\ln digita{l}_{it}+{\beta }_i{\displaystyle \sum C_{it}}+{\rho }_i{{\displaystyle \sum w_{ij}C}}_{it}+{\mu }_i+{\phi }_t+{\epsilon }_{it}$$

where ${\alpha }_0$ is a constant term, $\epsilon$ is a disturbance term, i represents a province, t represents a year, C is a control variable, w is a weight matrix, $\rho$ is a spatial autoregressive coefficient, ${\mu }_i$ and ${\phi }_t$ represent an individual disturbance term and a time disturbance term, respectively.

4. Results and Analysis

4.1. Benchmark Regression Analysis

This paper uses the time and region double fixed effect model to analyze the relationship between the digital economy and rural revitalization. The test results are shown in

Table 3. Two-Way Fixed Effects Benchmark Regression.

Variable	(1)	(2)
	Rural	Rural
Digital	0.399 *** (28.72)	0.283 *** (8.35)
Elderly		−0.119 ** (−2.04)
Industrial		0.173 *** (4.02)
Second		−0.216 ** (−2.02)
Open		0.036 * (1.3)
_cons	−0.649 *** (−28.43)	−3.004 *** (−5.92)
Year Effect	Yes	Yes
Individual Individual	Yes	Yes
N	279	279
R2	0.381	0.369
Hausman test	p = 0.022 (chi2 = 5.28)	p = 0.000 (chi2 = 1596.62)

Notes: Please mention that the t-values are in parentheses; The triple, double, and single asterisks (***, **, and *) signify that the test passes the 1%, 5%, and 10% significance levels, respectively.

. To avoid the data differences caused by different data dimensions and alleviate the model’s heteroscedasticity and multiple collinearity problems, this paper takes logarithms for all the variables involved.

First of all, to separately examine the impact of the digital economy on rural revitalization, this paper establishes a model (1). Its regression coefficient is 0.399, and it has passed the test with a confidence level of 1%, indicating that the digital economy can promote rural revitalization. Hypothesis 1 was verified. Then control variables such as elderly dependency ratio and number of industrial enterprises above scale are included in the regression, and the model (2) is obtained. It is found that the coefficients of the number of industrial enterprises above scale and the proportion of imports and exports are positive, indicating that the industrial structure and degree of openness can all drive rural revitalization, and farmers’ income and industrial structure are related. The industrial structure can adjust and balance farmers’ income; the number of industrial enterprises above scale can bring employment opportunities to rural people and help them increase their income and become rich; trade opening can drive the export of agricultural products and rural surplus labor, thus increasing their income. The difference in the size of their regression coefficients shows the difference in the degree of driving. Also, from the negative regression coefficient of population structure in model (2), it can be seen that the aging of rural society and the proportion of secondary industry cannot promote rural revitalization and may even affect the development of rural revitalization. It is self-evident that population aging inhibits rural revitalization. Population aging has caused a large loss of labor force, farmers have no economic income, and enterprises cannot operate efficiently, so the countryside cannot become rich. The proportion of secondary industry mainly reflects the development of secondary industry. The higher the proportion of secondary industry is, the lower the proportion of local primary industry may be, which in turn affects the development of rural areas.

4.2. Mediation Mechanism Test

The previous article theoretically analyzes the transmission mechanism of the digital economy promoting rural entrepreneurship and industrial structure transformation to promote rural revitalization. The paper uses the Sobel test and Bootstrap test to prove the existence of the Mediation effect, which shows that it is reasonable to choose rural entrepreneurship and industrial structure transformation as intermediary variables, and the test results are shown in

Table 4. Mediation Test Results.

Variable	(3)	(4)	(5)	(6)
	Entrepre	Rural	Indu-Struct	Rural
Digital	0.324 ***	0.155 ***	0.075 ***	0.118 ***
	(2.61)	(3.25)	(2.79)	(2.65)
Entrepre		0.036 *
		(1.72)
Indu-struct				0.642 ***
				(6.52)
Elderly	−0.002	−0.320 ***	0.023	0.305 ***
	(−0.01)	(−4.29)	(0.54)	(4.37)
Industrial	0.085	0.008	−0.025	0.011
	(1.07)	(0.26)	(−1.42)	(0.39)
Second	−0.446 *	−0.137	−0.042	−0.126
	(−1.73)	(−1.39)	(−0.75)	(−1.38)
Open	0.071	0.068 ***	0.117 **	−0.005
	(1.26)	(3.19)	(9.63)	(−0.21)
_cons	−3.232 ***	−1.602 ***	0.298	−1.910 ***
	(−2.68)	(−3.46)	(1.14)	(−4.46)
Ind	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes
R2	0.259	0.468	0.474	0.535
Adj. R2	0.246	0.456	0.464	0.525
N	279	279	279	279
Sobel Test	Z = 1.74 *		Z = 2.537 **
	(0.082)		(0.011)
Bootstrap Test	[−0.010–0.089]		[−0.010–0.154]

Notes: The triple, double, and single asterisks (***, **, and *) signify that the test passes the 1%, 5%, and 10% significance levels, respectively; the items in parentheses are t-statistical values.

.

In Table 4, models (3) and (4) are the result of the intermediary variable-rural innovation, which shows that the direct effect of the digital economy in promoting rural revitalization is 0.155, and the indirect impact of the digital economy in promoting rural entrepreneurship is 0.012 (0.324 × 0.036). The proportion of intermediary effect is 7.03% (0.324 × 0.036/(0.155 + 0.324 × 0.036)). Hypothesis 2 is verified. Models (5) and (6) are the result of the transformation of industrial structure, which shows that the direct effect of the digital economy in promoting rural revitalization is 0.118. The intermediary effect of the digital economy in promoting rural revitalization by promoting the transformation and upgrading of industrial structure is 0.048 (0.075 × 0.642), accounting for 28.99% (0.075 × 0.642/(0.118 + 0.075 × 0.642)), Hypothesis 3 is proved.

4.3. Robustness Test

4.3.1. Endogenous Test

A critical issue of empirical research is to alleviate endogenesis. As far as this article is concerned, the development of the digital economy promotes rural revitalization, and rural revitalization will, in turn, promote the development of the digital economy. That is, there may be a potential reverse causal relationship between the development level of the digital economy and the development level of rural revitalization. To ensure the reliability of the empirical results, this paper refers to the instrumental variable method of Huang Qunhui et al. [42]. It selects the number of post offices per 10,000 people in each region in 1984 as the instrumental variable of the development level of the digital economy. Its logic is: on the one hand, the development of information and communication technology is based on traditional communication technology, and the distribution of post offices as a telecommunications infrastructure will affect the future development of information and communication technology and the degree of application of information and communication technology by enterprises in terms of technical level and habit formation, so the tool variable meets the correlation requirements.

On the other hand, with the rapid development of information and communication technology, the impact of the number of post offices on rural revitalization gradually disappears. Post offices’ main function is to provide communication services for society, which does not directly affect the process of rural revitalization. Therefore, the tool variable meets the exclusive requirements. Because the sample of this paper is panel data, and the number of post offices per ten thousand people belongs to cross-section data, we draw lessons from the processing method of Nunn and Qian [45] to construct panel tool variables by introducing variables that vary with time. Specifically, the interactive items are constructed between the amount of Internet investment in the previous year and the number of post offices per 10,000 people in each province in 1984, expressed as iv, which is used as a digital economic index tool variable.

Table 5. Endogenesis test results.

Variable	(1)		(2)
	Digital		Rural
iv	0.321 ***
	(8.93)
Digital			0.307 ***
			(2.92)
Elderly	0.079		0.263 ***
	(0.84)		(2.61)
Industrial	0.531 ***		−0.073
	(21.92)		(−1.37)
Second	−1.04 ***		0.001
	(−10.69)		(0.01)
Open	−0.072 ***		0.081 ***
	(−2.96)		(3.63)
_cons	−12.22 ***		−0.564
	(−26.55)		(−0.63)
Ind	Yes		Yes
Year	Yes		Yes
Kleibergen-Paap rk LM		79.677 ***
Observations		279

Notes: The triple asterisks (***) signify that the test passes the 1% significance levels; the items in parentheses are t-statistical values.

shows the regression results of the instrumental variable method. After considering the possible endogenous problems between the development level of the digital economy and rural revitalization, columns (1) and (2) show that the development of the digital economy can still significantly promote rural revitalization. In addition, for the test of “insufficient tool variable recognition” of the original hypothesis, Kleibergen-Paaprk’s LM statistic p value is 0.000, which significantly rejects the original hypothesis; in the test of tool variable weak recognition, the WaldF statistic of Kleibergen-Paaprk is greater than the critical value of Stock-Yogo weak recognition test. It shows that it is reasonable to select the cross-item number of post offices per 10,000 people in history and the scale of national Internet investment in the same year as the tool variable of the development level of the digital economy.

4.3.2. Robustness Test

The robustness test examines the robustness of the interpretation ability of evaluation methods and indicators, that is, whether the evaluation methods and indicators still maintain a consistent and stable interpretation of the evaluation results when some parameters are changed. Although the benchmark regression analysis shows that the digital economy can promote rural revitalization, there may be a variety of possible reasons that may lead to the inconsistency of our results, so we need to conduct a robustness test. The paper uses the method of eliminating the four municipalities directly under the Central Government to carry on the robustness test because the economic development speed of the municipalities directly under the Central Government is relatively fast, so if we add it to the full sample size for benchmark regression, it is possible to exaggerate the driving role of the digital economy, so it is necessary to eliminate the four municipalities for robustness test, the test results are shown in

Table 6. Robustness Test Results.

Variable	Rural
Digital	0.275 ***
	(7.13)
Elderly	−0.119 *
	(−1.7)
Industrial	0.173 ***
	(3.8)
Second	−0.303 ***
	(−2.62)
Open	0.034
	(1.11)
_cons	−3.12 ***
	(−5.58)
Year	Yes
N	243
R2	0.435

Notes: The triple and single asterisks (*** and *) signify that the test passes the 1% and 10% significance levels, respectively.

.

As can be seen from the above test results, the robustness is good. The conclusion that the digital economy promotes rural revitalization after excluding municipalities directly under the Central Government is still valid, and it is established at a significant level of 1%.

5. Spatial Spillover Effect

5.1. Global Moran’s Index

The Moran’s Index is an overall test of the spatial autocorrelation between the digital economy and rural revitalization, and its calculation method is as follows:

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

(4)

where s² is the sample variance, x_i and x_j are the observations in regions i and j, respectively, n is the number of spatial cells, and w_ij is the spatial weight matrix. In this paper, we choose the spatial weight matrix as the economic, geographic distance nested matrix and calculate the global Moran’s index of rural revitalization index and digital economy index under this matrix for the spatial correlation test. The calculation results are shown in

Table 7. Digital Economy and Rural Revitalization Moran Index.

Year	Economic Geography Nested Matrix
	Digital Economy Index	Rural Revitalization Index
2013	0.104 ***	0.210 ***
	(2.892)	(5.026)
2014	0.085 **	0.216 ***
	(2.482)	(5.147)
2015	0.066 *	0.218 ***
	(2.061)	(5.194)
2016	0.041 **	0.227 ***
	(1.665)	(5.373)
2017	0.034 *	0.235 ***
	(1.527)	(5.524)
2018	0.028 *	0.240 ***
	(1.465)	(5.622)
2019	0.024 *	0.243 ***
	(1.472)	(5.648)
2020	0.159 *	0.231 ***
	(1.658)	(5.394)
2021	0.162 *	0.222 ***
	(1.668)	(5.208)

Notes: The triple, double, and single asterisks (***, **, and *) signify that the test passes the 1%, 5%, and 10% significance levels, respectively; the items in parentheses are z-statistical values.

. The calculation method of the economic geography nesting matrix is as follows:

(1)

Economic distance matrix (U): matrix element $u_{ij}=\{\begin{array}{cc}\frac{1}{\left|PGD{P}_i-PGD{P}_j\right|}& i\ne j\\ 0& i=j\end{array}$, where $PGD{P}_i$ represents the per capita GDP of the province ith from 2013 to 2021.

(2)

Geographical distance Matrix (V): matrix element $v_{ij}=\frac{1}{d_{ij}}$, where dij indicates the distance between the provincial capitals of the two provinces.

(3)

This article takes the economic geographic distance nested matrix (W): $W=\phi U+(1-\phi )V$, $\phi$ is the weight, considering the equal importance of the economic distance matrix and geographical distance matrix $\phi =0.5$.

As seen from Table 7, Moran’s digital economy index is all Significantly greater than 0, which can still indicate the existence of some spatial autocorrelation.

5.2. Moran Scatter Plot

To reflect the degree of correlation more directly between adjacent spatial units, this paper uses a Moran scatter plot to analyze the local spatial correlation. In the Moran scatter plot, the first, second, third and fourth quadrants represent high-high, low-high, low-low and high-low gathering areas, respectively. This paper selects the Moran scatter plot of the four-year (2013, 2015, 2018 and 2021) rural revitalization index and digital economy index as an example, and the results are shown in Figure 3 and Figure 4.

The Moran scatter plot of the rural revitalization index reports the following two points. First, the rural revitalization indexes of 31 Chinese provinces are almost equally distributed in the first and third quadrants, showing the clustering phenomenon of high being surrounded by high and low being surrounded by low; second, through the Moran diagram, we find that the provinces clustered in the third lower quadrant are mostly in the central and western regions, while the provinces in the more developed eastern region are distributed in the first quadrant.

The Moran scatter plot of the digital economy index also reports the following two points. Firstly, most of the digital economy indices of 31 Chinese provinces are distributed in the first and third quadrants, while some individual provinces are scattered in the second and fourth quadrants, indicating that there is also a high-high and low-low agglomeration of the digital economy. Secondly, most of the low-low areas falling in the third quadrant are in the central and western regions, indicating their low level of digital economy. Accordingly, most of the provinces falling in the high-high aggregation areas belong to the eastern regions, indicating that the level of the digital economy in these places is more developed.

5.3. Spatial Spillover Effect and Model Determination

Spillover effects are those where behavior within a region or organization has an impact not only on itself but also on its neighboring regions or organizations.

The spatial Durbin model (SDM) can simultaneously examine the effects of explanatory variables on the explained variables, including the effects from itself (direct effect) and the effects from neighboring regions (indirect effect), and the indirect effect is the spillover effect referred to in the paper. The value adding up the direct effect and the indirect effect is the total effect, which reflects the level of impact of the digital economy in a certain region on rural revitalization in all regions.

Some scholars point out that the digital economy promotes local rural revitalization and radiates to the surrounding areas. Therefore, based on the exploration of many scholars, this paper again verifies that there is a spatial spillover effect between the digital economy and rural revitalization. Following the test idea of Elhorst [46], the LM test (the statistic of LM-error is 51.976 and the statistic of LM-lag is113.572, both significant at 1% statistical level), LR test (individual effect: chi2 = 3.013; time effect: chi2 = 602.69 both significant at 1% statistical level), and Hausman test (chi2 = 13.26, significant at 5% statistical level) are used to determine the time spatial Durbin model with individual double fixed effects, the test results are shown in

Table 8. Model Determination.

	SAR	SEM	Choice
LM test	113.572 *** (0.000)	51.976 *** (0.000)	The model has both a spatial error effect and a spatial autoregressive effect, so the spatial Durbin model is chosen.
Hausman test	Chi2 = 13.26 (0.0216)		Hausman test results showed that the original hypothesis was rejected at a 5% significance level, so the fixed effect model was chosen.
LR test	individual effect: chi2 = 3.013 (0.003) time effect: chi2 = 602.69 (0.000)		LR test results show that the Durbin model has individual and time effects, so the Two-Way fixed effect is chosen.
Wald test	Chi2 = 46.04 (0.000)	Chi2 = 51.70 (0.000)	Wald test results show that the Durbin model refuses to degenerate into the spatial autoregressive and spatial error models.

Note: p-value in parentheses. The triple asterisks (***) signify that the test passes the 1%, significance level.

and the regression results are shown in

Table 9. Regression Results of Spatial Durbin Model.

Variable	Economic Geography Nested Matrix
	Direct Effect	Indirect Effect	Total Effect
Digital	0.091 *	1.069 ***	1.16 ***
	(1.95)	(3.43)	(3.61)
Elderly	−0.256 ***	−0.673	−0.930 **
	(−4.04)	(−1.4)	(−1.84)
Industrial	−0.022	0.336	0.314
	(−0.6)	(1.22)	(1.12)
Second	−0.135 *	−0.505	−0.64
	(−0.274)	(−0.90)	(−1.1)
Open	−0.0713 ***	0.080	0.009
	(−3.1)	(0.52)	(0.05)
Ind	Yes	Yes	Yes
Year	Yes	Yes	Yes
N	279	279	279

Notes: The triple, double, and single asterisks (***, ** and *) signify that the test passes the 1%, 5%, and 10% significance levels, respectively; the items in parentheses are t-statistical values.

.

Table 9 reports that under the Economic Geography Nested Matrix, the coefficients of the digital economy are all positive under the three effects and are significant at the significant levels of 1% and 10% respectively. The direct effect refers to the impact of the digital economy on the rural revitalization of the region, the indirect effect refers to the impact of the digital economy on the rural revitalization of the surrounding areas, that is, the spatial spillover effect, and the total effect refers to the sum of the two. Among them, the coefficient of indirect effect is 1.069. The proportion of indirect effect reaches 92.2% (1.069/1.16), indicating that the contribution rate of indirect effect is relatively large. There is a spatial spillover effect in the digital economy. The contribution rate of indirect effect is greater than that of direct effect. That is, the development of a digital economy in this region can not only promote the level of rural revitalization in this region but also enhance the level of rural revitalization in neighboring areas and will take the lead in driving the revitalization of rural areas in neighboring areas.

Robustness Test

The previous article is the regression of the spatial Durbin model under the geo-economic nested matrix to examine whether the result is reliable. Hence, the paper refers to the idea of Meng Weifu et al. [39], carries on the robustness test by changing the spatial weight matrix, and selects the geographical distance spatial weight matrix to carry on the regression again. The test result is shown in

Table 10. Robustness Test of Spatial Durbin Model.

Variable	Economic Geography Nested Matrix
	Direct Effect	Indirect Effect	Total Effect
Digital	0.213 ***	1.788 **	2.001 **
	(4.04)	(2.32)	(2.49)
Elderly	−0.27 ***	−1.663	−1.932 *
	(−3.75)	(−1.62)	(−1.82)
Industrial	−0.034	−0.303	−0.337
	(−0.82)	(−0.52)	(−0.56)
Second	−0.154 *	−0.843	−0.997
	(−1.71)	(−0.64)	(−0.73)
Open	−0.083 ***	−0.502	−0.585
	(−3.22)	(−1.26)	(−1.41)
Ind	Yes	Yes	Yes
Year	Yes	Yes	Yes
N	279	279	279

Notes: The triple, double, and single asterisks (***, ** and *) signify that the test passes the 1%, 5% and 10% significance levels, respectively; the items in parentheses are t-statistical values.

.

As can be seen from Table 10, the regression results of the digital economy under the three effects of direct effect, indirect effect and total effect are still significant, showing that the previous results are reliable and robust and that the digital economy can improve rural revitalization. And it will promote the rural revitalization of the surrounding areas.

6. Discussion and Conclusions

Based on the panel data of 31 provinces for nine years, the paper discusses the mechanism and spatial spillover effect of rural revitalization driven by the digital economy and draws the following conclusions:

First, the results of benchmark regression report the conclusion that the digital economy can promote the development of rural revitalization. Whether or not the control variable is added, the coefficient between the digital economy and rural revitalization is always positive and significant at a 1% level. The digital economy has injected new vitality into revitalizing the countryside. After deleting the four municipalities directly under the Central Government, the main regression was carried out again, and the results still passed the significance test at the 1% level. That is, the robustness was good, which confirmed the previous conclusion again. In addition, the establishment of this conclusion indirectly supports the research conclusions of many scholars, such as Zhang Xiaoyu [47], Yang Shilin [48], Hong Rufei and Wu Jianhua [49].

Second, using the Sobel test and Bootstrap test to prove the existence of the intermediary effect, the digital economy can improve the level of intermediary variables, and the accumulation of intermediary variables can better promote rural revitalization. Among them, rural entrepreneurship and industrial structure transformation intermediary variables show different intermediary roles in this process. Comparatively speaking, the intermediary role of industrial structure transformation is greater.

Third, there are spatial spillover effects in the promotion of rural revitalization by the digital economy, and the Moran index of the digital economy and rural revitalization is significantly positive, indicating the existence of spatial autocorrelation, showing the phenomenon of aggregation. The three effects of the decomposition of the spatial Durbin model have all passed the significance test. The proportion of indirect effect is higher than that of direct effect, indicating that the greatest role of the development of the digital economy in a certain region is not to promote the rural revitalization of the region but to improve the rural revitalization level of the surrounding areas greatly. The indirect effect shows that the local digital economy can not only improve the level of rural revitalization but also promote the development of rural revitalization in neighboring areas because of spatial spillover.

According to the above conclusions, to achieve the grand goal of rural revitalization, the digital economy is an efficient and green means and way. In the future, we should actively develop a digital economy and release dividends. To this end, the following suggestions are put forward:

First, it is necessary to strengthen the construction of digital infrastructure in all provinces. To give full play to the role of the digital economy in rural revitalization, it is necessary to have the basis of digital facilities—especially digital infrastructure in rural areas. Although the infrastructure construction in China’s rural areas has achieved great results compared with before, the problem of regional imbalance is still prominent. In addition, there is still a gap in digital infrastructure between rural areas and cities in China. Therefore, the state should increase investment in rural digital facilities, especially in remote areas, to build high-quality, wide coverage of the Internet, 5G technology, information service platform, e-commerce logistics and other systems.

Second, it is necessary to introduce technical talents to enhance the governance ability of the main body in rural areas. The carrier of the digital economy is digital technology, and the main user of digital technology is people in the final analysis. Introducing technical talents can broaden the scope of digital technology, which can prosper the digital economy and promote rural revitalization. Pepsi thrives when people can make the best of their talents. Excellent talents are the fresh blood of the development of enterprises, especially in rural areas with a talent shortage. Therefore, the state should improve the talent security system in remote rural areas, earnestly consider their needs from the perspective of talents, and attract talents to return and take root in rural areas. On the one hand, the presence of talents has directly brought high-quality new governance subjects to the countryside. On the other hand, under their leadership and influence, traditional grass-roots cadres’ old concepts and working methods have been changed so that with their joint efforts, they can use new technologies and take new roads to revitalize the countryside.

Third, rebalance the gap between development and over-development. Everything has its law of development. We should not blindly and aimlessly develop the digital economy just because the digital economy can promote the revitalization of rural areas. We should pay special attention to the negative effects that may be brought about by the development process and excessive development. Once it is found that it should be stopped in time, and the direction of work should be adjusted, the work plan should be reformulated according to the actual situation of various places. The digital economy should be developed pertinently, efficiently, and green. And then promote the revitalization of the countryside.

Fourth, give full play to the role of the spatial spillover effect, neighboring areas can form a relationship of assistance. Just as distant relatives are not as good as close neighbors, different provinces should form a friendly neighborhood relationship of harmonious coexistence and mutual promotion and develop in one place to promote the development of the surrounding areas. Instead of widening the distance and making a gap, we should develop in a balanced manner.