Study on the Evolution of Spatiotemporal Dynamics and Regional Differences in the Development of Digital Agriculture in China

Abstract

In the current study, an index system for digital agriculture growth was constructed. The index encompasses six key dimensions, namely production, operation, service, management, sustainability, and digital information infrastructure. Data from 30 Chinese provinces between 2011 and 2020 were collected and analyzed using the entropy method, Moran index, Dagum’s Gini coefficient, and the kernel density estimate. An in-depth analysis of the development level and spatial patterns, dynamic evolution and intra- and inter-regional differences in China (i.e., eastern, western, and central regions) was conducted. From the result, an overall growing trend of digital agriculture in China was observed, with a relatively more advanced status in the eastern region. A positive spatial dependence, showing a “high-high” and “low-low” (HH, LL) trend, was obtained. However, the regional spatial dependence has generally weakened since 2019. The intra-regional differences were large in western and eastern areas, while the greatest inter-regional differences were unveiled between western and eastern regions. The country’s overall differences mainly stemmed from inter-regional differences. The overall kernel density curves moved to the right over time, showing a trend of a gradual rise in digital agricultural growth, accompanied by a polarization pattern in the western region.

1. Introduction

Digital agriculture is the newest generation of farming technology, contributing to agricultural integration, high-quality agricultural development, and rural revitalization. First proposed in 1997, digital agriculture was defined as an information-based farming technique supported by geographic and knowledge technologies [1]. In recent years, the Chinese government has invested much effort in advancing local digital agriculture. A series of documents, regulations, and programs were released to consolidate the ecosystem of digital agriculture. Of note, the Ministry of Agriculture and Rural Affairs and the Central Network Information Office issued the Electronic Farming and Rural Growth Plan (2019–2025), emphasizing several key areas of development, i.e., improving digital literacy in rural areas, facilitating optimal agricultural resource allocation, and boosting operational effectiveness of the value chain [2]. The plan also highlighted the development objectives and strategic position of digital farming and countryside development [3]. Later, the 2021 White Paper on China’s Digital Economy Development unveiled a stunningly low level of agricultural digitization in China (i.e., 7.3%), which was attributed to the absence of a framework for digital agriculture and a shortage of digital talent. Given this, a subsequent document, 2020–2022 Central Document No. 1, outlined several key proposed measures, including an electronic village pilot scheme, electronic village building and growth projects, the establishment of IT-powered villages and countryside public services, and the promotion of sustainable growth in agriculture and the countryside. (Source: http://www.lswz.gov.cn/html/xinwen/2022-02/22/content_269430.shtml) (accessed on 22 February 2022). The top-down policy approach of Chinese digital farming was considered crucial in laying a solid developmental framework for local digital agriculture.

Therefore, it is of great significance to explore the development path of digital agriculture and narrow the development gap of regional digital agriculture by constructing a multidimensional comprehensive evaluation index system for digital agriculture development, exploring its spatial distribution characteristics, and comparing and analyzing the regional differences and dynamic evolution laws of digital agriculture development by region.

Currently, digital agriculture has served as a key driver in boosting the economic growth of the countryside in China. In the past, global research focuses were dedicated to farming machinery control, rational allocation of production factors, and utilization of production data [4]. The basics have laid a strong foundation in the recent worldwide adoption of the Internet of Things (IoT), big data, and sensors in digital farming [5]. The technology was expected to improve farming efficiency, marketability of agricultural goods, and farmers’ income. Specifically, digital agriculture, known as smart or precision agriculture [6,7], uses remote and near-sensing technologies in farm monitoring [4]. The technology also represents a total solution for digital governance, mobile connectivity, and data valorization [8]. The recent advancement also served to address preexisting challenges, like imperfect supply chain management, low operational efficiency, and information asymmetry [9]. Even better, management of sowing could be assisted by electronic equipment [10]. However, compared to most developed economies, China still lagged behind in this field. A local consensus on digital farming has yet to be established due to the differences in conceptual understanding among scholars and practitioners [11,12]. In recent years, local research efforts were concentrated on the concept of digital farming [13,14], value chains, and innovation [15,16], as well as environmental change and food security [17]. The discussion mainly focused on the connotation of digital farming at the theoretical level. For crop production, some scholars considered digital and communication tools were the key [18], meanwhile others [19] considered deep technological integration into agricultural processes was more important. Moreover, several key indexes were established to support the evaluation of digital agricultural output. Firstly, Meng et al. [20] constructed an evaluation index system having five aspects, i.e., digital agricultural infrastructure, human resources, agricultural digital production, agricultural informatization level, and agricultural production efficiency. Secondly, Hua et al. [21] constructed a digital economy index system from three factors, i.e., digital technology innovation, digital industry application, and digital transaction basis. Thirdly, Ren et al. [22] constructed a digital rural development indicator system from four aspects, i.e., facility digitization, economy digitization, administrative digitization, and life digitization. Next, Wang et al. [23] constructed an agricultural high-quality growth indicator system in terms of innovation, coordination, efficiency, openness, sharing, and sustainability. Most scholars build an evaluation index system mainly from the aspects of the digital economy, digital countryside, agricultural informatization [24], and high-quality growth. Based on previous findings, a novel index system is developed in the current study.

As the current digital farming studies primarily focus on the connotations of digital farming, limited systematic studies on digital agricultural growth in China have been conducted. There is still an unmet need for empirical research on the structure of an indicator system covering the whole value chain in terms of production, management, service, and sustainable development. Digital agriculture, as an important initiative for achieving the strategy of rural revitalization, has emerged as an emerging catalyst for the modernization of farming and rural development. Herein, the current study exploratively analyzes the evolution of the spatiotemporal dynamics and regional variations in China’s advancement in digital agriculture. The key deliverables are as follows. First of all, a novel index system was scientifically constructed to analyze the growth level of digital farming in China. Six factors, namely, production intensification, operation networks, management refinement, service informatization, sustainable development level, and digital information infrastructure, were included. Secondly, the spatial distribution of Chinese digital agriculture was investigated via the Moran index. Thirdly, the data from 2011 to 2020 from 30 provinces, and Dagum’s Gini coefficient, were adopted to explore the spatial variation, heterogeneity among regions, and identify the main sources of differences. Further, the dynamic evolution characteristics of digital agriculture as a whole and in the three areas (i.e., central, eastern, and western regions) was analyzed through the distribution of kernel density curve. The objective was to provide theoretical and data support for Chinese digital agriculture development policy proposals.

2. Materials and Methods

2.1. Construction of an Indicator System for the Growth in Digital Agriculture and Its Rationale

Based on previous publications, the Outline of the Digital Rural Growth Strategy, the 2021 Assessment Report on the National Counties’ Agricultural and Rural Informatic Development Level, the Digital Farming and Rural Growth Plan (2019–2025), and the basic connotation of digital agriculture, a comprehensive evaluation index with 6 dimensions in digital agriculture development was selected, as shown in

Table 1. Evaluation indicator system of China’s digital farming growth level.

Dimension	Indicator	Indicator Metrics	Property
Production intensification	Effective irrigated area	Effective irrigated area (thousand hectares)	+
	Food productivity	Total food production/number of people employed in the primary sector (tonnes/person)	+
	Agricultural land productivity	Gross agricultural output/area sown to crops (yuan/ha)	+
	Mechanization level	The entire farm equipment power (10,000 kW)	+
	Footprint of facility agriculture	Total area of facility-based farming/total cultivated area (%)	+
	Agricultural labor productivity	Primary industries value added/number of workers in primary industries (yuan/person)	+
Operational Networking	Level of precision processing of agricultural products	Business income from agro-processing industry/gross agricultural output (%)	+
	Level of agricultural network payments	Digital Financial Inclusion Index (-)	+
	Digital transactions in agriculture	E-commerce sales + purchases (billion yuan)	+
	Informatization level of agricultural products	The volume of postal and telecommunication business (billion yuan)	+
Management refinement	Level of logistics and transport	Freight transport by road (10,000 tonnes)	+
	Level of traceability of agricultural product quality and safety	Number of professional farmers’ cooperatives per 10,000 people in the countryside (number)	+
	Level of application of information technology in agriculture	Share of administrative villages with postal service (%)	+
	Agro-meteorological observatories	Number of agrometeorological observation stations (number)	+
Service informatization	Digital Talent Ownership	Number of workers in scientific research and technical services/total number of workers in the urban non-private units (%)	+
		Number of workers in transport, storage, and postal services/total number of workers in the urban non-private sector (%)	+
		Number of workers in information transmission, computer services, and software villages/total number of workers in the urban non-private sector (%)	+
	Digital Agriculture Innovation Base	Number of comprehensive demonstration counties for e-commerce in villages (number)	+
	Scope of information technology services, such as the Internet of Things	Length of rural postal delivery routes (km)	+
Sustainable developmentlevel	Pesticide application	Pesticide application (10,000 tonnes)	−
	Fertilizer application	Fertilizer application (10,000 tonnes)	−
	Agricultural film usage	Agricultural film usage (10,000 tonnes)	−
Digital information infrastructure	Rural Internet penetration	Number of Internet users in the region/population of the region (%)	+
	Rural smartphone penetration	Average number of cellphones per 100 rural homes (units)	+
	Rural radio and television coverage	Average television ownership per 100 rural households (units)	+

, namely, production intensification, operational networks, management refinement, service informatization, sustainable development, and digital information infrastructure. Additionally, 23 secondary indicators were created under the 6 dimensions.

2.1.1. Selection of Production Intensification Indicators

Crop production intensification is the main goal of digital agriculture. Electronic farming adopts various applications, like 3S technology, IoT, and big data, to achieve intelligent planting, irrigation, and monitoring to intensify the irrigation area, agricultural output, and food productivity [25,26]. The aforementioned six indicators (i.e., the effective irrigated area, agricultural land output rate, food productivity, mechanization level, the area occupied by facility agriculture, and agricultural labor productivity) were selected to assess the degree of production intensification.

2.1.2. Selection of Indicators for Operational Networking

Business networking is a prerequisite to support modern digital farming. First of all, business networking technology has the potential to revolutionize the traditional mode of operation in terms of production, processing, and sales. Significant cost reduction is also implied. Secondly, better market supply and demand docking, mechanization, and automation of screening and cleaning in agricultural process are the key benefits of standardization, as well as an increase in the standard of farming and the precision processing of sideline products, and the encouragement of the advancement in digital farming. Furthermore, the selling channels are expanded to enhance the level of agricultural network payment transactions, thus improving the level of information technology of farming products. Thus [27,28], four indicators, namely the precision processing level of agricultural products, the agricultural network payment level, digital agricultural transactions, and the informatization level of farming products, were selected to assess the level of the operation network for digital agriculture.

2.1.3. Selection of Management Refinement Indicators

Management refinement is a basic element in digital farming processes, of which automation and precision are the major developmental directions. This enhances agricultural products’ output, as well as the turnover rate [29]. The adoption of agricultural technologies, such as IoT and cloud computing, was illustrated to support real-time monitoring and value chain management. Herein, four indicators, i.e., the level of logistics and transport, traceability for farming product safety and quality, the level of application of agricultural IT [30], and the number of agrometeorological observation stations, were selected to measure management refinement of digital agriculture.

2.1.4. Selection of Service Informatization Indicators

The informatization of services is another important consideration for digital agricultural development. Of note, adequate digital talents and foundational infrastructure are the features established to sustain the growth in digital agriculture technology. Therefore, 3 secondary indicators, the population of digital talents, the number of digital agricultural innovation bases, and the scope of IoT and IT services, were included to gauge the degree of information technology in digital agriculture services.

2.1.5. Selection of Indicators of the Level of Sustainable Growth

Sustainable growth is the ultimate goal of digital agricultural development. Modern agricultural processes have brought about digital dividends with the help of various technologies such as IoT, big data, and cloud computing. The technology was expected to enable scientific decision making, thereby contributing to precise fertilizer application, spraying, and farming. Potentially, the process led to better environmental production, as well as positive growth in crop output. Aligned with previous findings [31,32], 3 secondary indicators, namely pesticide application, fertilizer application, and agricultural film usage, were adopted as a part of the indicator.

2.1.6. Selection of Digital Information Infrastructure Indicators

In previous research, continuous upgrading of digital information infrastructure was reflected by the rates of rural Internet penetration, rural smartphone penetration, and rural radio and television penetration. The penetration rate also mapped the degree of infrastructure support in a region [2], thus allowing the measurement of the digital information base of the corresponding study area. From a previous study [33], the 3 aforementioned indicators were utilized to evaluate the study area’s digital information infrastructure.

2.2. Study Area and Data Sources

In recent years, China’s digital agriculture has developed rapidly. As of 2019, the degree of county’s digital farming reached 36.0% and the rate of digitalized agricultural production reached 23.8%. For farming products, online retail sales accounted for 10.0% of the total trade volume. The traceability level of safety and quality attained 17.2%. Geographically, the coverage of rural information management service institutions reached 75.5%, while the coverage of e-commerce sites in administrative villages reached 74.0%. With a clear economic benefit, all levels of Chinese government placed much emphasis on digital agriculture to support local economic growth in rural areas.

In the study, data from 30 Chinese provinces (Macau, Hong Kong, Tibet, and Taiwan were excluded) between 2011 and 2020 were collected. The data were obtained from the Peking University Digital Inclusive Finance Index (2011–2020), China Rural Statistical Yearbook, China Statistical Yearbook, China Agricultural Product Processing Industry Development Report, the statistical yearbooks of various provinces and cities, the EPS database, the area of facility-based agriculture from the National Greenhouse Data System, the Agricultural Development Report, data from farmers’ professional cooperative societies, theme reports on the market development of various regions, the websites of various regional governments, etc. The data’s missing values were filled using the average annual growth rate or an interpolation method.

2.3. Entropy Method

The evaluation of digital farming’s progress was based on the entropy method, which was used to determine the relevant weights and evaluation. The entropy method was utilized to measure the comprehensive index of the growth in digital agriculture across years by region. The specific methods were as follows.

Indicator data were first standardized to remove dimensional interference.

Step 1—Positive indicators:

$$L_{ij}={\displaystyle \frac{b_{ij}-\min (b_{1j},b_{2j},\dots ,b_{nj})}{\max (b_{1j},b_{2j},\dots ,b_{nj})-\min (b_{1j},b_{2j},\dots ,b_{nj})}}$$

(1)

Step 2—Negative indicators:

$$L_{ij}={\displaystyle \frac{\max (b_{1j},b_{2j},\dots ,b_{nj})-b_{ij}}{\max (b_{1j},b_{2j},\dots ,b_{nj})-\min (b_{1j},b_{2j},\dots ,b_{nj})}}$$

(2)

where $b_{ij}$ is the original indicator value, $i$ represents the region, $j$ represents the indicator, and $L_{ij}$ is the standardized indicator value.

Step 3—Information entropy of the jth indicator:

$${\kappa }_j=-{\displaystyle \frac{1}{\ln (n)}}{\displaystyle \sum _{i=1}^{\mathrm{n}}(\frac{L_{ij}}{{\displaystyle {\sum }_{i=1}^n{L}_{ij}}}\ln \frac{L_{ij}}{{\displaystyle {\sum }_{i=1}^n{L}_{ij}}})}$$

(3)

Step 4—The $j$th indicator’s weight:

$$w_j=(1-{\kappa }_j)/{\displaystyle {\sum }_{j=1}^m(1-{\kappa }_j)}$$

(4)

Step 5—Calculation of the score level for each sample composite:

$$p_i={\displaystyle {\sum }_{j=1}^m{w}_j{L}_{ij}}$$

(5)

2.4. Spatial Correlation Test

$$\begin{gathered} \begin{array}{l}M\mathrm{oran}’\mathrm{s}\mathrm{I}={\displaystyle \frac{n{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n{w}_{ij}(y_i-\overline{y})(y_j-\overline{y})}}}{{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n{w}_{ij}}}{\displaystyle {\sum }_{i=1}^n{(y_i-\overline{y})}^2}}}\end{array} \\ ={\displaystyle \frac{{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n{w}_{ij}(y_i-\overline{y})(y_j-\overline{y})}}}{s^2{\displaystyle {\sum }_{i=1}^n{\displaystyle {\sum }_{j=1}^n{w}_{ij}}}}} \end{gathered}$$

(6)

Here, $s^2={\displaystyle \frac{1}{n}}{\displaystyle {\sum }_{i=1}^n{(y_i-\overline{y})}^2}$. $n$ is the total number of spatial individuals and $w_{ij}$ is the spatial matrix. Moran’s I index can be greater than zero, less than zero, or equal to zero, corresponding to the three spatial states of positive correlation, negative correlation, or independent distribution, respectively. From the three spatial matrices of the economy, economic geography, and inverse distance of the economy, the spatial correlation of the growth degree of Chinese digital farming from 2011 to 2020 was measured based on Moran’s I index.

2.5. Dagum’s Gini Coefficient and Its Decomposition

Dagum’s Gini coefficient provides a nuanced analysis of regional variations in the development of digital agriculture, while the figure reflects the relative differences only. Dagum’s Gini coefficient divides the total Gini coefficient (G) into the following components: $G_w$—contribution of within-group differences; $G_{nb}$—contribution of net between-group differences; and $G_t$—hypervariable density contribution. The result was calculated with the following formula: $G=G_w+G_{nb}+G_t$.

$$G={\displaystyle \frac{{\displaystyle {\sum }_{j=1}^k}{\displaystyle {\sum }_{h=1}^k}{\displaystyle {\sum }_{i=1}^{n_j}{\displaystyle {\sum }_{r=1}^{n_h}}\left|y_{ji}-y_{hr}\right|}}{2n^2\overline{y}}}$$

(7)

$$G_{jh}={\displaystyle \frac{{\displaystyle {\sum }_{i=1}^{n_j}{\displaystyle {\sum }_{r=1}^{n_h}}\left|y_{ji}-y_{hr}\right|}}{n_j{n}_h\left({\overline{y}}_j+{\overline{y}}_h\right)}}$$

(8)

where $j$ and $h$ represent two regions, $n_j$ and $n_h$ represent the number of provinces in the two regions, $y_{ji}$ and $y_{hr}$ represent the levels of digital agriculture in the $i$th province in region $j$ and in the $r$th province in region $h$, respectively; and ${\overline{y}}_j$ and ${\overline{y}}_h$ represent the mean of the level of the growth in digital farming of all provinces in each area. The result is $G_{jj}$ if region $j=h$.

$$G_{jj}={\displaystyle \frac{\frac{1}{2\overline{y_j}}{\displaystyle {\sum }_{i=1}^{n_j}{\displaystyle {\sum }_{r=1}^{n_j}}\left|y_{ji}-y_{jr}\right|}}{n_j^2}}$$

(9)

$$G_w={\displaystyle {\sum }_{j=1}^k{G}_{jj}{p}_j{s}_j}$$

(10)

$$G_{nb}={\displaystyle {\sum }_{j=2}^k}{\displaystyle {\sum }_{h=1}^{j-1}}{G}_{jh}(p_j{s}_h+p_h{s}_j)D_{jh}$$

(11)

$$G_t={\displaystyle {\sum }_{j=2}^k}{\displaystyle {\sum }_{h=1}^{j-1}}{G}_{jh}(p_j{s}_h+p_h{s}_j)(1-D_{jh})$$

(12)

$$D_{jh}={\displaystyle \frac{(d_{jh}-p_{jh})}{(d_{jh}+p_{jh})}}$$

(13)

$$d_{jh}={\displaystyle {\int }_0^{\infty }d{F}_j(y)}{\displaystyle {\int }_0^y(y-x)d{F}_h(x)}$$

(14)

$$p_{jh}={\displaystyle {\int }_0^{\infty }d{F}_h(y)}{\displaystyle {\int }_0^y(y-x)d{F}_j(x)}$$

(15)

where $p_j={\displaystyle \frac{n_j}{n}}$ is the ratio of the number of provinces $n_j$ in region $j$ to the sample $n$, and $s_h={\displaystyle \frac{n_h{\overline{y}}_h}{n\overline{y}}}$ is the ratio of the level of growth in digital farming in region $h$ to the level of growth in digital farming in all provinces in the sample. $D_{jh}$ is the relative influence between region $j$ and region $h$. The detailed calculation process and explanation of each equation can be found in the related literature [34].

2.6. Kernel Density Estimate

$$V(x)=(1/Nh){\displaystyle {\sum }_{i=1}^{N}K\left[\left(X_i-x\right)/h\right]}$$

(16)

$$K(x)={\displaystyle \frac{1}{\sqrt{2\pi }}}\exp (-{\displaystyle \frac{x^2}{2}})$$

(17)

where $V(x)$ is the density function of the composite index of the level of growth in digital farming, $N$ is the number of observed provinces in the region, $X_i$ is the independent distribution of observations, $x$ is the average of the observations of the level of growth in digital farming in a certain region in the period $i$, K(g) $K(·)$ is the kernel function, and $h$ refers to the bandwidth; the smaller the value of $h$, the higher the accuracy.

3. Results

3.1. The Development Level of Digital Agriculture in China

Table 2. Index of the development level of China’s digital agriculture.

ID	Province	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020	Average Value
1	Beijing	0.438	0.455	0.471	0.485	0.492	0.505	0.513	0.525	0.550	0.558	0.499
2	Tianjin	0.301	0.328	0.347	0.361	0.381	0.386	0.389	0.412	0.429	0.466	0.380
3	Hebei	0.343	0.358	0.369	0.388	0.400	0.407	0.429	0.453	0.472	0.498	0.412
4	Shanxi	0.299	0.317	0.327	0.344	0.357	0.367	0.380	0.391	0.401	0.426	0.361
5	Inner Mongolia	0.316	0.334	0.351	0.377	0.394	0.396	0.435	0.458	0.470	0.490	0.402
6	Liaoning	0.314	0.331	0.363	0.376	0.399	0.399	0.408	0.411	0.424	0.439	0.386
7	Jilin	0.297	0.320	0.330	0.334	0.356	0.365	0.379	0.385	0.408	0.433	0.361
8	Heilongjiang	0.318	0.339	0.369	0.385	0.409	0.424	0.440	0.456	0.489	0.513	0.414
9	Shanghai	0.390	0.388	0.409	0.419	0.418	0.437	0.460	0.465	0.485	0.513	0.438
10	Jiangsu	0.347	0.374	0.387	0.400	0.424	0.441	0.461	0.473	0.509	0.532	0.435
11	Zhejiang	0.350	0.368	0.376	0.379	0.399	0.418	0.438	0.460	0.496	0.521	0.421
12	Anhui	0.268	0.295	0.323	0.341	0.351	0.372	0.396	0.415	0.431	0.459	0.365
13	Fujian	0.304	0.327	0.349	0.362	0.383	0.396	0.407	0.416	0.444	0.462	0.385
14	Jiangxi	0.260	0.275	0.299	0.313	0.327	0.343	0.365	0.387	0.407	0.426	0.340
15	Shandong	0.298	0.323	0.331	0.361	0.368	0.380	0.408	0.430	0.450	0.473	0.382
16	Henan	0.291	0.310	0.321	0.331	0.350	0.367	0.391	0.419	0.445	0.471	0.370
17	Hubei	0.272	0.298	0.324	0.345	0.364	0.377	0.400	0.428	0.449	0.466	0.373
18	Hunan	0.261	0.298	0.315	0.320	0.343	0.361	0.382	0.405	0.423	0.457	0.356
19	Guangdong	0.343	0.361	0.379	0.391	0.414	0.431	0.454	0.490	0.518	0.537	0.432
20	Guangxi	0.281	0.300	0.317	0.324	0.333	0.347	0.368	0.384	0.401	0.415	0.347
21	Hainan	0.267	0.289	0.297	0.312	0.336	0.357	0.370	0.390	0.407	0.416	0.344
22	Chongqing	0.229	0.273	0.319	0.343	0.360	0.370	0.390	0.410	0.397	0.412	0.350
23	Sichuan	0.292	0.313	0.340	0.352	0.375	0.396	0.419	0.454	0.466	0.487	0.390
24	Guizhou	0.213	0.227	0.253	0.287	0.307	0.327	0.352	0.387	0.405	0.423	0.318
25	Yunnan	0.250	0.265	0.290	0.297	0.314	0.330	0.354	0.374	0.400	0.417	0.329
26	Shanxi	0.317	0.327	0.343	0.363	0.378	0.393	0.411	0.422	0.430	0.457	0.384
27	Gansu	0.245	0.267	0.282	0.293	0.316	0.335	0.357	0.395	0.405	0.424	0.332
28	Qinghai	0.241	0.253	0.279	0.316	0.351	0.362	0.377	0.380	0.394	0.413	0.337
29	Ningxia	0.253	0.295	0.313	0.325	0.343	0.336	0.357	0.365	0.373	0.390	0.335
30	Xinjiang	0.272	0.290	0.323	0.318	0.332	0.343	0.356	0.376	0.397	0.415	0.342
Average value		0.296	0.317	0.337	0.351	0.369	0.382	0.402	0.421	0.439	0.460	0.377
Eastern region		0.336	0.355	0.371	0.385	0.401	0.414	0.431	0.448	0.471	0.492	0.410
Central region		0.283	0.307	0.326	0.339	0.357	0.372	0.392	0.411	0.432	0.456	0.368
Western region		0.264	0.286	0.310	0.327	0.346	0.358	0.380	0.400	0.413	0.431	0.351

shows the comprehensive index of the level of growth in China’s digital agriculture. The pattern of changes in the degree of growth in digital farming in each study area during the period of 2011–2020 shows a clear trend of “east > central > west”. The overall levels of the east, central, west, and the country as a whole shows a gradual upward trend.

3.2. A Chronological Characterization of the Level of Development of Digital Agriculture in China

Table 2 depicts the level of growth in digital farming and the trends of its changes in 30 provinces between 2011 and 2020. Over time, the degree of digital farming growth in each province generally showed an upward trend. Beijing, Shanghai, Jiangsu, Guangdong, and Zhejiang from the eastern region top the list. This result may be due to a higher degree of economic progress, favorable resource endowment, and well-developed transport network in the eastern region, providing strong support for the growth in digital farming in the area. On the other hand, the provinces located at the back of the pack, such as Ningxia, Gansu, Guizhou, Yunnan, Qinghai, and Xinjiang, lag and are concentrated in the western region. Possible reasons include the low level of agricultural informatization infrastructure, the uneven growth in the digital economy, and the low degree of centralization of the growth in digital farming in the area. These factors slowed the advancement of digital agriculture as a whole.

Figure 1 shows an overall upward trend in digital farming growth in China, as well as the western, eastern, and central areas, from 2011 to 2020. In Table 2, the national mean level of digital agricultural growth is 0.377, and the mean levels of digital agricultural growth in the eastern, central, and western areas are 0.410, 0.368, and 0.351, respectively. A comparison reveals that the degree of digital agricultural growth in the eastern area is far higher than the national average, while the central region is comparable to the national average level. The western area has the lowest degree of digital farming growth, indicating significant room for improvement.

3.3. Spatial Characterization of the Level of Development of Digital Agriculture in China

3.3.1. Global Spatial Autocorrelation Test

In

Table 3. Moran’s I index of the development of digital agriculture in China (2011–2020).

Year	Economy			Economic Geography			Inverse Distance of the Economy
	I	Z	p	I	Z	p	I	Z	p
2011	0.464	4.831	0.000	0.286	4.445	0.000	0.314	4.766	0.000
2012	0.433	4.566	0.000	0.288	4.502	0.000	0.318	4.844	0.000
2013	0.434	4.594	0.000	0.240	3.854	0.000	0.269	4.201	0.000
2014	0.446	4.711	0.000	0.247	3.951	0.000	0.282	4.381	0.000
2015	0.401	4.230	0.000	0.215	3.465	0.000	0.244	3.813	0.000
2016	0.381	4.041	0.000	0.163	2.744	0.003	0.193	3.118	0.001
2017	0.358	3.753	0.000	0.133	2.302	0.011	0.171	2.771	0.003
2018	0.226	2.475	0.007	0.095	1.756	0.040	0.132	2.221	0.013
2019	0.204	2.251	0.012	0.115	2.033	0.021	0.148	2.437	0.007
2020	0.199	2.182	0.015	0.165	2.682	0.004	0.190	2.958	0.002

Note: p is the probability and z is the statistical value.

, the result unveils a positive Moran’s I index with statistically significant (p < 0.05), suggesting an obvious spatial dependence on the degree of growth in digital farming in China’s provinces from 2011 to 2020. The areas with a higher level of digital farming growth are demonstrated to drive neighboring provinces’ digital farming development. Secondly, the gradual reduction in Moran’s I index indicates a decrease in the spatial dependence of digital farming growth. Meanwhile, aligned with Table 2, Beijing, Shanghai, and Jiangsu tended to outperform in terms of the degree of digital farming and economic strength. The result suggests a positive spatial correlation between the level of digital farming and the extent of economic development.

3.3.2. Local Spatial Autocorrelation Test

The current study further analyzed the Moran scatterplot of the economic matrix for digital agricultural development.

In Figure 2, scatter plots of the local Moran index on Chinese digital farming growth are presented, showing a clustering trend from 2011, 2014, 2017, and 2020 data. Most provinces demonstrate “high–high”, and “low–low” (HH, LL), and are located in the first and third quadrants. The possible reasons for this are as follows. In the eastern region, the area experienced a greater degree of digital agriculture development fueled by better economic strength. Diversified social capital and agricultural resources were attracted to the region, eventually promoting the advancement of digital farming in the area. In the western region, the financially disadvantaged location had less economic strength to support digital farming. Eventually, the growth remained at a low level. Of note, the spatial dependence of digital farming development appeared to be weakened in the recent 2 years, despite the disturbance of the COVID-19 pandemic. However, the results still suggest that the advancement of digital farming in China had a global spatial positive correlation as a whole, which also tended to be consistent with the empirical results.

3.4. Regional Differences in the Growth in Digital Farming in China and Their Decomposition

This study calculated Dagum’s Gini coefficient and further divided it into intra-group and inter-group differences, and the contribution rates used to derive the overall Gini coefficient of digital farming progress in China between 2011 and 2020. The detailed results are displayed in

Table 4. Gini coefficient of digital agricultural development in China and the results of its decomposition.

Year	Overall Difference G	Differences within Groups			Difference between Groups			Contribution (%)
		Eastern Part	Central Part	Western Part	East–Central	East–West	Central–West	Gw	Gnb	Gt
2011	0.087	0.073	0.039	0.069	0.092	0.126	0.065	25.418	64.606	9.976
2012	0.076	0.062	0.033	0.062	0.079	0.111	0.057	25.191	66.247	8.562
2013	0.067	0.061	0.028	0.053	0.074	0.096	0.047	26.097	62.095	11.808
2014	0.063	0.055	0.031	0.048	0.072	0.088	0.045	25.814	60.627	13.558
2015	0.058	0.049	0.031	0.045	0.067	0.080	0.043	25.641	60.126	14.233
2016	0.055	0.049	0.028	0.040	0.062	0.078	0.042	25.456	61.878	12.666
2017	0.053	0.049	0.028	0.041	0.057	0.071	0.041	26.960	55.844	17.196
2018	0.050	0.048	0.030	0.041	0.054	0.066	0.040	28.299	51.946	19.755
2019	0.054	0.052	0.034	0.036	0.057	0.074	0.042	26.742	57.729	15.528
2020	0.053	0.049	0.032	0.036	0.054	0.074	0.045	26.171	57.858	15.970
Average level	0.062	0.055	0.031	0.047	0.067	0.086	0.047	26.179	59.896	13.925

Source: Calculated and collated from Dagum’s Gini coefficient and its decomposition.

.

3.4.1. Overall Regional Differences and Trends in the Development of Digital Agriculture in China

In Table 4, The overall regional differences in China’s digital agricultural growth shows a gradual reduction over the study period, with an average annual decline of about 5.4%. Specifically, the overall variance is manifested as a gradual decrease from 0.087 in 2011 to 0.050 in 2018. Due to the impacts of uncontrollable external factors, such as the COVID-19 epidemic and trade friction, the overall difference in 2019–2020 shows a small rebound compared to 2018. The overall regional variations in the growth in digital farming in China show a shrinking trend, while the spatial imbalance decreases.

3.4.2. Intra-Regional Variations in China’s Digital Farming Growth Rate and Their Changing Trends

From 2011 to 2020, the trend of intra-regional variations is revealed to be “east > west > central” in Table 4. Several tendencies are unveiled. Firstly, the eastern region showed an inverted “N”-shaped development trend—first decreasing, then rising slowly by a small margin, and then slowly decreasing again—and the overall trend of intra-regional differences showed a gradual decrease. Key cities or provinces like Beijing, Shanghai, and Zhejiang had a better economic strength, resource endowment and digital infrastructure, compared to individual provinces in the region. Meanwhile, certain regional provinces demonstrated an absolute advantage. Thus, the intra-regional difference fluctuated, with an average annual decrease of about 4.3%. Second, the central region generally showed an inverted “N”-type development trend. The result demonstrated a slightly upward and downward fluctuations between 2011 and 2019. The figure decreased to 0.032 by 2020. The annual average reduction rate was about 2.2%. The central region was actively involved in the construction of digital farming and countryside regions in response to the national call. Further, the western area showed a trend of a slow decline, a brief rise, and a subsequent decline. The data declined from 0.069 in 2011 to 0.040 in 2016, briefly rose to 0.041 in 2018, and then dropped to 0.036 in 2019. The overall trend of intra-regional differences in the western area indicated an average annual decline of 7.0%. To conclude, western and eastern regions are the major contributors to intra-group variations in China.

3.4.3. Inter-Regional Variations in China’s Digital Farming Growth Level and Their Changing Trends

The variations in China’s digital farming growth, as well as the trends among the three major inter-regional areas (i.e., east–central, east–west, and central–west), are illustrated in Table 4. The trend of the difference among the three major areas is similar, showing a first declining and then rising trend. An overall shrinking trend was found over time. From the result, firstly, the east–west region fell from 0.126 in 2011 to a low of 0.066 in 2018, and then briefly rose to 0.074 in 2020. The average annual rate of decline was about 5.7%. Secondly, the east-central region declined from 0.092 in 2011 to a low of 0.054 in 2018, rose briefly to 0.057 in 2019, and then declined to 0.054 in 2020. The average annual rate of decline was about 5.7%. Thirdly, the central–western region declined from 0.065 in 2011 to a low of 0.040 in 2018, and then rose to 0.045 in 2020. The average annual decline was about 4.0%. Next, the mean values of the differences between central–east, east–west, and central–west were 0.067, 0.086, and 0.047, respectively. In particular, west–east showed the largest inter-regional variations. The primary rationale may be attributed to the strong economic strength, favorable resource endowment, and superior location conditions in the eastern areas, yet the disadvantageous natural conditions, underdeveloped infrastructure, and a lack of digital agriculture professionals in the western region, leading to the result.

3.4.4. Sources of Regional Differences in China’s Digital Agricultural Growth Level and Their Contributions

Table 4 reports inter-regional, intra-regional, and hypervariable densities of variations in the advancement and current trends of digital farming in China. The contribution rate between regions showed a gradually shrinking trend from 64.606% in 2011 to 51.946% in 2018, and then a rebound to 57.858% in 2020. The contribution rate within the region showed an overall upward trend. It rose from 25.418% in 2011 to 28.299% in 2018, then slightly declined from 26.742% to 26.171% between 2019 and 2020. Overall, the hypervariable density initially increased, then decreased. It rose gradually from 9.976% in 2011 to 19.755% in 2018, and then it declined to 15.970% in 2020.

The inter-regional differences accounted for the largest share of contributions and became the main source of the variations in Chinese digital farming growth. Secondly, the impacts of uncontrollable external factors were seen in this analysis, including the COVID-19 pandemic and trade frictions. To reduce the regional variations in China, it would be necessary to redesign the local adaptation strategy and complement it with a top-down national policy from the central government. Several key areas, including rural information infrastructure upgrade, digital technology support, and the cultivation of farmers with digital acumen, are required to be embedded into the policy design.

3.5. Analysis of the Dynamic Evolution of the Level of Development of Digital Farming in China

The kernel density estimation approach was used to dynamically characterize the dynamic evolution of China’s digital farming development.

3.5.1. Distribution Dynamics of National Digital Farming Development

The result is reported in Figure 3. First of all, the central peak of the kernel density curve of the national digital agricultural growth level shifted to the right over time. The result suggested that the level of digital farming growth in the 30 provinces continuously improved. Secondly, the kernel density curve demonstrated a broadening characteristic when trailed to the right. The result indicated the increasing difference in China’s digital agricultural development level. More advanced provinces could far exceed some provinces with a slower pace of growth. For a number of peaks, a transition from double to triple peaks was seen from 2017 to 2020, which indicated a “digital divide” and emergence of differentiation. Matthew’s effect was seen in the growth in digital farming in some provinces.

3.5.2. Distribution Dynamics of Digital Farming Development in the Eastern Region

The result is reported in Figure 4. The central tendency of the kernel density curve of the digital agricultural development level in the eastern area moved to the right over time, suggesting a steady growth in digital farming growth. Second, in terms of waveform, the wave’s height exhibited an overall rising tendency between 2011 and 2014. Meanwhile, the main peak’s width became narrower, indicating a centralized trend of the growth. Moreover, from 2017 to 2020, the height of the wave decreased with an increased span, showing a gradual increase in the growth difference. For the number of peaks, from single peak to double peak, there was not much alternation over time. It shows that the polarization of the development of digital agriculture in this region is not serious.

3.5.3. Distribution Dynamics of Digital Farming Development in the Central Region

The result is reported in Figure 5. From the data, firstly, the central tendency of the kernel density curve shifted slightly to the right from 2011 to 2020, indicating a positive growth in digital farming over time. The trend of development was aligned with national and eastern regions. Second, a small decline in peak level was found from 2017 to 2020, while the width of distribution showed a broadening trend over time. The result indicated a certain difference and differentiation of the growth in the central area, although the trend was not significant.

3.5.4. Distribution Dynamics of Digital Farming Development in the Western Region

The result is reported in Figure 6. First and foremost, the central tendency of the kernel density curve gradually shifted to the right from 2011 to 2020, indicating growth in digital farming in the western area. Secondly, the distribution pattern of the kernel density curve trailed to the right, suggesting an advancement of digital farming with an unbalanced state. The span of the distribution curve increased. The data demonstrated a wide discrepancy in the digital farming level. Of note, the western region comprised both highly developed areas (e.g., Inner Mongolia, Sichuan) and areas with limited digital agriculture activities (e.g., Ningxia, Gansu), giving rise to large differences in the developmental status for digital farming. Moreover, the peak level increased significantly from 2011 to 2020, indicating progressive growth in digital farming in a concentrated manner. Furthermore, a change in pattern from a single to a double peak was observed during the period, which indicates a trend of growth polarization in the area.

4. Discussion

This paper constructed a comprehensive evaluation index system for the development level of digital agriculture, estimated its development level via the entropy method, compared and analyzed the regional differences in the development level of digital agriculture by combining the Dagum Gini coefficient and nuclear density estimation, and revealed the spatiotemporal dynamic evolution characteristics. This research shows that in terms of development level, the development level of digital agriculture in China continues to improve, with Jiangsu, Guangdong, and other provinces in the forefront, and Ningxia, Qinghai, and other provinces at the bottom, which is consistent with the research results of Meng et al. [20]. In previous studies, digital benefits, environment, talents, and innovation were found to be the key attributes to measure the growth level of digital agriculture [35,36,37]. In the current study, a myriad of considerations was further considered to build a novel and comprehensive index, in order to objectively assess the development of digital agriculture in China. From an industrial chain view, the current study aimed to examine the extent of integration for digital agricultural technology, as well as production intensification, operation networking, refinement management, and service informatization. Further, digital information infrastructure and sustainable development were also considered.

Regarding spatial distribution characteristics, previous studies mainly adopted a standard deviation ellipse or spatial adjacency matrix [20,38], which supported the three spatial weight matrices (i.e., economy, economic geography, and economic inverse distance) used in the current study to investigate the spatial distribution features. The study results also suggested a spatial correlation for digital agricultural growth. Most of the “high-high” (HH) agglomeration provinces were found in the eastern region, which aligned with the possible causes of spatial agglomeration, namely, better agricultural conditions, the availability of digital talents, and more established information infrastructure. More advanced regions were revealed to assume a driving role in supporting neighboring provinces.

Regarding regional differences, the current study adopted the Dagum Gini coefficient to analyze regional variations and their sources. The variations between the eastern and the western regions were found to be the largest, and the inter-regional variations were the main sources of country differences in the development of digital agriculture [39]. In addition, a trend of growth polarization was found in the western region. More importantly, the analysis provided an overview of China’s intra- and inter-regional variations in digital agriculture. From a policy view, further investigation would be warranted to minimize the growth gap in digital farming between regions in China.

Therefore, this paper puts forward the following opinions for further improving the development level of digital agriculture and narrowing regional differences.

Firstly, a strong rural information infrastructure would be a key driver for digital agricultural growth. Feasible measures include an increase in coverage of broadband networks, fiber-optic networks, 5G base stations, artificial intelligence, and additional rural digital infrastructure, etc. The measures are expected to lay a strong technological foundation and minimize the “digital divide” in the countryside, supporting the growth in digital agriculture.

Secondly, nurturing technology–agricultural talents is a key factor of success. Even better, agriculture-related enterprises, research institutes, and universities are encouraged to collaborate with grass-roots organizations to provide on-the-job training, build a robust digital agricultural ecosystem, and create job opportunities in the industry. Meanwhile, extra support for, and retention of, digital and farming professionals are highly encouraged. Support for continuous professional development is also suggested, aiming to boost their on-job abilities and qualities, as well as their endogenous impetus for digital agriculture.

Thirdly, localization of digital farming development is a priority. Due to regional variations, it is necessary to practically explore the region’s resource endowment, the economic and technological level, and the industrial development. With an in-depth understanding of regional characteristics, relevant policies, such as financial support and tax exemptions, could be administered accordingly and address the unmet need in a precise way.

5. Conclusions

Based on the data from 30 Chinese provinces (Macau, Hong Kong, Tibet, and Taiwan excluded) between 2011 and 2020, a system of an index for digital farming growth was developed. An entropy approach was utilized to elucidate a comprehensive index of the growth level. Moran’s index, Dagum’s Gini coefficient, and kernel density estimation were used to analyze the spatiotemporal dynamic evolution characteristics and regional differences in digital agriculture growth in China. Key findings included the following. First, in terms of spatial and temporal characteristics, a positive trend of digital farming growth was observed in China. The eastern region topped the list (0.410), followed by central and western regions, for which the average comprehensive scores of digital agricultural development were 0.368 and 0.351. Digital agricultural development in the eastern region was significantly higher than the national average (0.377). Secondly, regarding spatial characteristics, a positive spatial dependence was characterized in China, yet the spatial dependence appeared to weaken. From the local Moran index scatter plot, most provinces were identified as the “high–high” and “low–low” (HH, LL) type, and concentrated in the first and third quadrants. Thirdly, regarding regional differences in China, the geographical difference in digital farming growth was found to be narrowing. The larger differences were within the western and eastern groupings, where the average level was 0.047 and 0.055, and the largest inter-regional variations were between western and eastern regions, with the average level of 0.086. The overall differences mainly stemmed from the inter-regional differences. Next, regarding distributional dynamics, the kernel density curve shifted to the right over time, showing gradual digital agricultural growth for the country and all three regions. In particular, the eastern and western regions demonstrated trends of concentration and polarization, respectively.