Global Value Chains, Digital Economy, and Upgrading of China’s Manufacturing Industry

Abstract

This study utilized panel data from 31 provinces in China from 2006 to 2020 to investigate the impact of the digital economy on the upgrading of the manufacturing industry’s global value chain. Two types of spatial weighting matrices were used to construct SAR, SEM, SAC, and SDM models. The results revealed that technological innovation plays a direct mediating role in the upgrading of the manufacturing industry, and the global value chain has a positive regulatory effect on the relationship between the digital economy and the manufacturing industry’s upgrading. Under the economic distance spatial weighting matrix, the spatial spillover effect of the digital economy on the manufacturing industry’s global value chain is not significant, whereas, under the geographic distance spatial weighting matrix, the digital economy has a positive and significant spatial spillover effect. The SDM model showed the best explanatory effect. This implies that geographic spatial dependence has a significant impact on the upgrading of the manufacturing industry’s industrial structure, and it is positively influenced by nearby provinces. Understanding the impact mechanism and spatial spillover effects of the digital economy on the manufacturing industry’s upgrading can help promote efficient, fair, and balanced regional development. It can also aid in constructing a new domestic and international “dual circulation” development pattern that evolves with the global manufacturing value chain, sharing the dividends of the digital economy’s impact on the global value chain’s development.

1. Introduction

The impact of the epidemic has caused temporary paralysis of the global supply chain, with a large number of enterprises ceasing production and work, resulting in bottlenecks in production and logistics, which has had a significant impact on the manufacturing industry [1]. In the reconstruction of the global value chain, competition among countries has become more intense [2]. In order to avoid production capacity shortages caused by similar epidemics, companies have begun to strengthen localized production to reduce their reliance on external supplies. This has also brought new challenges to the reconstruction of the manufacturing industry supply chain. The digital economy has had a significant impact on the reconfiguration of the global value chain of China’s manufacturing industry [3]. With the advent of new digital technologies, China’s manufacturing industry has been able to move up the value chain and increase its competitiveness in the global market. Digital technologies such as robotics, the Internet of Things (IoT), and artificial intelligence (AI) have enabled China to enhance its manufacturing capabilities, reduce costs, and improve product quality. In addition, digital platforms have facilitated the integration of supply chains, allowing Chinese manufacturers to better coordinate with their global partners.

The use of robotics in China’s manufacturing industry has increased dramatically in recent years [4]. As a result, the country has become the world’s largest market for industrial robots, with a 36.6% market share in 2020. The use of robots has allowed Chinese manufacturers to increase efficiency, reduce labor costs, and improve product quality. In addition, the development of AI technologies has enabled Chinese manufacturers to improve their production processes by analyzing vast amounts of data, optimizing production lines, and predicting maintenance needs.

The rise of digital platforms has also facilitated the integration of supply chains and enabled Chinese manufacturers to better coordinate with their global partners [5]. Platforms such as Alibaba, JD.com, and Pinduoduo have become important channels for Chinese manufacturers to sell their products to consumers around the world. These platforms have also enabled Chinese manufacturers to access global markets more easily, thereby expanding their customer base and increasing their competitiveness [6].

Moreover, the digital economy has facilitated the growth of new business models in China’s manufacturing industry [7]. For example, the “smart factory” model has emerged as a key strategy for Chinese manufacturers to increase efficiency and reduce costs. Smart factories leverage advanced digital technologies to optimize production processes, reduce downtime, and improve quality control. In addition, digital technologies have enabled Chinese manufacturers to provide more customized products and services to their customers, thereby increasing customer satisfaction and loyalty.

Overall, the impact of the digital economy on the reconfiguration of the global value chain of China’s manufacturing industry has been significant. Chinese manufacturers have been able to move up the value chain by leveraging digital technologies such as robotics, IoT, and AI. The integration of supply chains through digital platforms has also enabled Chinese manufacturers to expand their global reach and increase their competitiveness. Finally, the emergence of new business models, such as smart factories, has enabled Chinese manufacturers to improve their efficiency, reduce costs, and provide more customized products and services to their customers. As the digital economy continues to evolve, it is likely that China’s manufacturing industry will continue to play an increasingly important role in the global value chain.

The global market size of smart factories is gradually expanding. According to a report released by Market and Markets, the compound annual growth rate of the global smart factory market size from 2021 to 2025 will reach 11% (as shown in Figure 1). The key factors driving the market growth include the financial policy to maintain the normal operation of manufacturing facilities during the COVID-19 epidemic, the optimization of resources, and the reduction of production and operation costs, which have increased the demand for industrial robots in the market growth and the demand for digital technology in the manufacturing environment. Smart factories are the future development trend of the manufacturing industry and an optimized path to achieve industrial upgrading. According to a research report from the China Industrial Research Institute, the market size of China’s smart factories in 2021 was 138.4 billion yuan, and the compound growth rate in the coming years will exceed 10%. It is expected that this data will increase to 207.6 billion US dollars by 2025.

How to achieve rapid industrial development in the context of the global value chain is an important issue that needs to be addressed. Can the digital economy help upgrade the manufacturing industry under the global value chain? Moreover, how does it work? Few scholars have analyzed the relationship between the digital economy and manufacturing upgrading in the context of global value chains. This article considers the relationship between the three and conducts in-depth research on their mechanisms and spatial spillover effects, providing a meaningful experience for the upgrading path of manufacturing in the context of global value chains.

The problems addressed in this study are twofold. Firstly, China’s manufacturing industry is facing challenges due to the global epidemic impact, global value chain reconstruction, and new technology revolution [8]. These factors have led to declining traditional comparative advantage, excessive resource consumption, and low value-added of the manufacturing industry. Secondly, the study seeks to explore how the digital economy can be integrated into regional manufacturing industries in the process of global value chain reshaping and how this integration can lead to the high-end leap of the provincial manufacturing industry.

The study aims to address these problems by analyzing the mechanism and spatial effects of the digital economy on the upgrading of China’s manufacturing industry from the perspective of the global value chain. It seeks to explore how the digital resource elements and innovation resource elements in the global value chain can be used efficiently to promote the rationalization and advanced development of the manufacturing industry structure of each region. Through an endogenous breakthrough, the study aims to realize the industrial structure upgrade of the manufacturing industry with high innovation capacity and high added value. However, due to limitations in length, methods, and data, there is still room for improvement in the article. This article lacks a comparison of heterogeneity in the manufacturing industry worldwide and overlooks the differences in the industrial development foundations of different regions.

The remainder of the paper is organized as follows. Section 2 is a review of related literature. Section 3 shows the research hypothesis. Section 4 conducts research design. Section 5 displays Empirical Analysis. Finally, Section 6 concludes the paper and proposes policy recommendations.

2. Literature

2.1. The Impact of Global Value Chains on Manufacturing Upgrading

Studies on global value chains in manufacturing upgrading have mainly focused on the knowledge spillover effect [9], technological innovation effect [10], and foreign investment spillover effect [11]. The knowledge spillover effect can promote the upgrading of the manufacturing industry. The diffusion of knowledge among enterprises in the industrial chain is beneficial for promoting the upgrading of the manufacturing industry [12]. The technological innovation effect is an important path for the manufacturing industry to achieve industrial upgrading and move toward the high end of the global value chain [13]. The improvement of technological efficiency and total factor productivity can improve the relatively unfavorable position of China’s manufacturing industry in the global value chain. The manufacturing industry of developing countries can obtain technology spillover from developed countries by embedding themselves in the global value chain. Under the global value chain, the technological level of China’s manufacturing industry has been improved, but it still plays a “follower” role in some high-end manufacturing industries. The division of labor in the global value chain has a mechanism that promotes and suppresses the upgrading of the manufacturing industry in developing countries. The foreign investment spillover effect of the global value chain promotes the upgrading of the manufacturing industry [14]. Foreign direct investment promotes the upgrading of the value chain in the host country, and China’s participation in the global value chain promotes the improvement of its independent innovation capability. Embedding itself in the global value chain can promote the transformation and upgrading of China’s manufacturing industry.

2.2. The Impact of Digital Economy on Industrial Upgrading

Studies on the impact of the digital economy on industrial upgrading have mainly focused on theoretical aspects such as the high-quality development of industries, industrial chains, technological levels, and industrial formats. The digital economy provides new ideas for the high-quality development of industries, and implementing policies to promote digital transformation in industries has strategic significance and will promote industrial upgrading [15]. The industrial chain will be restructured and gradually achieve digital transformation [16]. In order to leverage the driving role of the digital economy in the upgrading of manufacturing, the integration and sharing of data in the industrial chain should be strengthened, and attention should be paid to the construction of intelligent manufacturing ecosystems [17]. The improvement of digital technology is the foundation of industrial development, and the popularization of digital infrastructure will improve the efficiency of resource allocation. The digital economy indirectly promotes industrial upgrading through technological innovation and human capital [18]. The digital economy is reshaping the industrial form and plays an important role in the formation of new industries and formats [19]. By solving the “pain points” of upgrading China’s manufacturing industry, the digital economy helps Chinese manufacturing move toward the high end of the global value chain [20].

2.3. The Impact of Digital Economy on Industrial Upgrading under Global Value Chain

Scholars’ research on the impact of the digital economy on industrial upgrading in the global value chain is mainly conducted from the perspectives of global value chain division of labor, value chain upgrading, corporate social responsibility, and manufacturing service. The digital economy has significantly changed the spatial layout and governance model of the global value chain, making it exhibit new trends such as digitization, disintermediation, service-orientation, and customization [21]. The digital economy has elevated the status of the global value chain division of labor and has the most significant impact on low- to medium-technology manufacturing and information and communication services [22]. The deep integration of digital technology and manufacturing has become a new engine for the global value chain to climb to the high end, and investment in digitization has improved the global value chain division of labor of enterprises. China’s manufacturing industry is faced with development dilemmas such as being captured by low-end industries and being blocked by multiple barriers to upgrade, and digitization will reconstruct the labor division and value system of the global manufacturing industry [23].

Previous research has laid a solid theoretical foundation for this topic, but there are still some limitations and room for further expansion. Firstly, previous studies have focused on the development of ideas, strategies, technologies, and industrial forms that drive industrial upgrading through the digital economy but lack a more comprehensive measurement of the global value chain. Secondly, previous research has mostly analyzed the impact of the global value chain on manufacturing upgrading or the impact of the digital economy on manufacturing but lacks relevant studies that simultaneously consider the relationships between the global value chain, the digital economy, and manufacturing upgrading. Therefore, this study attempts to explore the mechanism analysis of the digital economy on the upgrading of the global value chain in China’s manufacturing industry and its spatial spillover effects.

3. The Research Hypothesis

3.1. Influence Mechanism of Technological Innovation on Digital Economy and Regional Manufacturing Upgrade

Data is the most important factor of production in the digital economy. It overcomes the inherent defects of traditional production factors in the manufacturing industry. With the help of technological innovation, the manufacturing industry uses digital resources such as massive data and information, as well as digital technologies such as big data, cloud computing, and artificial intelligence, to more reasonably allocate the production factors of the manufacturing industry [24]. Simultaneously, the development of the digital economy can reduce the transaction cost and marginal production cost of the manufacturing industry through technological innovation and carry out low-end transformation of the traditional manufacturing industry [25]. The “economies of scale” in production will change the original output structure and output efficiency, promote the formation of high-end manufacturing through the effects of technology spillovers and industrial linkages, realize demand-side reshaping, and drive regional manufacturing upgrades. Therefore, the development of the digital economy may have an economic effect on the upgrading of the manufacturing industry, with “technological innovation” as an intermediary variable [26]. Based on this, the first hypothesis is put forward:

Hypothesis 1 (H1).

Technological innovation plays an intermediary role between the digital economy and the upgrading of the manufacturing industry structure.

3.2. Influence Mechanism of Global Value Chain on Digital Economy and Manufacturing Upgrade

The global value chain has a profound impact on the rationalization and advanced development of China’s manufacturing industry [27]. The global value chain has played an important role in the digital economy and the upgrading effect of regional manufacturing [28]. By integrating digital resources and production factors with the most comparative advantages in the global value chain, the goal of maximizing profits is achieved. The digital economy is reshaping global value chains while changing how manufacturing operates and allocates resources [29]. China is an important participant in the division of labor in the global value chain. Relying on its comparative advantages, such as complete industrial classification and economies of scale, China has a significant advantage in the manufacturing industry in the global value chain. With strong manufacturing capabilities and a digital economic foundation, the global value chain uses the commercial application of disruptive manufacturing innovations to spawn new industries, new businesses, and new formats, promote China’s manufacturing industry to the mid-to-high end of the global value chain and realizes the upgrading of the industrial structure [30]. Based on this, a second hypothesis is proposed:

Hypothesis 2 (H2).

The global value chain has a positive regulating effect on the upgrading of the industrial structure of the digital economy and manufacturing industry.

3.3. Spatial Spillover Effects

Participating in the division of labor in global value chains is an effective channel for a country or region to obtain technological progress and economic growth [31]. In the context of the reshaping of the global value chain, China should use the massive digital resources and digital technologies of the digital economy to expand the manufacturing value chain and build a new pattern of domestic and foreign dual circulation that co-evolves with the global manufacturing value chain, thereby promoting regional efficiency, fairness, and balance in the development of the manufacturing industry, accelerating the upgrading of the manufacturing industry. In the process of “digitalization of the manufacturing industry”, the development of the digital economy breaks the constraints of time and space on information transmission and enhances the depth and breadth of the coordinated development of inter-regional manufacturing industries [32]. Existing studies have shown that global value chains have a spatial spillover effect on inter-regional economic development [33]. Does the digital economy have a spatial spillover effect on the upgrading of China’s regional manufacturing industries that participate in the division of labor in global value chains? Based on this, a third hypothesis is proposed:

Hypothesis 3 (H3).

The spatial spillover effect enables the digital economy to promote the upgrading of regional manufacturing industries.

4. Research Design

4.1. Data Sources

Data from 31 provinces in China (except Hong Kong, Macau, and Taiwan) from 2006 to 2020 were selected, including: “China Statistical Yearbook”, “China Industrial Statistical Yearbook”, “China Science and Technology Statistical Yearbook”, “China Environmental Statistical Yearbook”, EPS database, TiVA (Trade in Value Added) database. Among them, the relevant data of trade added value from 2006 to 2018 comes from the TiVA database, and the added value data from 2019 to 2020 is predicted by the GM (1, 1) model.

4.2. Variable Setting

(I).

Explained variable. The connotation of industrial structure rationalization emphasizes aspects such as the degree of industrial coordination and resource allocation efficiency. This paper selects the upgrading level of the manufacturing industry structure as the explained variable and uses the Theil index to measure the rationalization degree of the manufacturing industry structure [34]. This paper also uses the structural hierarchy coefficient to measure the advanced degree of the manufacturing industry structure.

(II).

Explanatory variables. Firstly, this paper uses the improved entropy method to weigh the Internet penetration rate, the total amount of telecom services, and the number of mobile phone users. This paper obtains a comprehensive score to represent the level of digital infrastructure construction. Then, this paper uses the improved entropy method to weigh the total GDP of the information transmission, software, and information technology service industry and the business income of the information service industry. This paper also obtains a comprehensive score to represent the level of digital development. Finally, this paper uses the improved entropy method to weigh the ICT R&D funds and the total number of talents with a bachelor’s degree or above and obtain a comprehensive score to represent the scientific research level of digital technology innovation.

(III).

Control variables. Referring to most current scholars, in this paper, GDP per capita is selected to represent the reference of the provincial economic development level. This paper selects the ratio of the expenditure in the general budget of the local government to GDP to represent the degree of government participation. This paper selects the actual use of foreign capital in the current year to represent the level of foreign investment. In this paper, the average exchange rate of the year is converted into RMB, and the logarithm is used to express its value. Considering the availability of data, this paper selects the ratio of total import and export to GDP to express the degree of dependence on foreign trade. This paper selects the ratio of the sum of the provincial financial scientific and educational expenditures to the expenditures in the general budget of the local finance to represent the level of education investment. This paper selects the ratio of energy consumption to GDP in each province to represent energy intensity.

(IV).

Mediating Variables. Drawing on the methods currently used by most scholars, this paper takes the number of patents and R&D expenditures of the manufacturing industry in each province as the two evaluation indicators, and then, by using the improved entropy method to weight the two indicators, this paper finally obtains a comprehensive score to represent the technological innovation level of the manufacturing industry.

(V).

Moderator variable. This paper selects the Global Value Chain Participation (GVC-Participation) Index, the Global Value Chain Position (GVC-Position) Index, and the Domestic Value-Added Comparative Advantage (RCA-DVA) Index to characterize the position level of China’s manufacturing industry in the global value chain and use them as moderator variables. The calculation formula is as follows:

①

Global value chain participation index

In exports, the sum of the proportion of indirect added value and the proportion of foreign added value included:

$$GVC\_participatio{n}_{ir}=\frac{I{V}_{ir}}{E_{ir}}+\frac{F{V}_{ir}}{E_{ir}}$$

(1)

IV_ir represents indirect value-added exports of industry i in country r specifically and the trade volume of intermediate goods exported by industry i in country r to other countries. FV_ir represents the value of foreign imported intermediate goods included in the final product exports of the country i industry. E_ir represents the value-added export value of industry i in country r. Among them, IV_ir/E_ir indicates the proportion of indirect added value in the total export of country r industry i, and FV_ir/E_ir indicates the proportion of foreign added value included in the total export of country r industry i.

• ②

  Global Value Chain Status Index

Indicators to measure the international division of labor of a country or region in the global value chain of an industry:

$$GVC\_positio{n}_{ir}=\ln (1+\frac{I{V}_{ir}}{E_{ir}})-\ln (1+\frac{F{V}_{ir}}{E_{ir}})$$

(2)

• ③

  Domestic value-added comparative advantage index

Indicators of explicit comparative advantage to measure domestic and international division of production:

$$\begin{array}{ll}RCA\_DV{A}_i^r& =\frac{(DD{V}_i^r+ID{V}_i^r+RI{M}_i^r)/{\displaystyle {\sum }_i^n(DD{V}_i^r+ID{V}_i^r+RI{M}_i^r)}}{{\displaystyle {\sum }_r^G(DD{V}_i^r+ID{V}_i^r+RI{M}_i^r)/{\displaystyle {\sum }_r^G{\displaystyle {\sum }_i^n(DD{V}_i^r+ID{V}_i^r+RI{M}_i^r)}}}}\\ & =\frac{DV{A}_i^r/{\displaystyle {\sum }_i^{n}DV{A}_i^r}}{{\displaystyle {\sum }_r^{G}DV{A}_i^r/{\displaystyle {\sum }_r^G{\displaystyle {\sum }_r^{n}DV{A}_i^r}}}}\end{array}$$

(3)

The above formula represents the comparative value of the domestic added value of a certain industry in a country or region to the total domestic added value of the country’s exports relative to the total domestic added value of the industry in global exports. Specific variable descriptions are shown in

Table 1. Variable Description.

Variable Type	Abbreviation	Variable Explanation
Explained variable	tl	Rationalization Level of Manufacturing
	tn	Advanced level of manufacturing
Explanatory variables	infra	Digital infrastructure construction level
	indus	digital development level
	sci	digital technology research level
control variable	pgdp	economic development level
	gov	Degree of government involvement
	inv	foreign investment level
	tra	degree of dependence on foreign trade
	edu	educational investment level
	ene	energy regulation (represented by energy intensity)
mediating variable	inno	technical innovation level
moderator variable	gvcpar	global value chain participation index
	gvcpos	global value chain status index
	rca	Domestic value-added comparative advantage index

.

4.3. Model Settings

(I)

Mechanism analysis

In this paper, the Theil index is used to measure the rationalization degree of China’s manufacturing industry structure, and the structural hierarchy coefficient is used to measure the advanced degree of industrial structure.

①

Basic regression model

$$\begin{gathered} t{l}_{i,t}=\alpha +{\beta }_{1}di{g}_{i,t}+{\beta }_{2}pgd{p}_{i,t}+{\beta }_{3}go{v}_{i,t}+{\beta }_{4}in{v}_{i,t}+{\beta }_{5}tr{a}_{i,t}+{\beta }_{6}ed{u}_{i,t}+{\beta }_{7}en{e}_{i,t}+{\epsilon }_{i,t} \\ t{n}_{i,t}=\alpha +{\beta }_{1}di{g}_{i,t}+{\beta }_{2}pgd{p}_{i,t}+{\beta }_{3}go{v}_{i,t}+{\beta }_{4}in{v}_{i,t}+{\beta }_{5}tr{a}_{i,t}+{\beta }_{6}ed{u}_{i,t}+{\beta }_{7}en{e}_{i,t}+{\epsilon }_{i,t} \end{gathered}$$

(4)

In the above model, α is a constant, β is the influence coefficient of each variable, the subscript i represents the Chinese province (municipality, autonomous region), and t represents the year. The explained variable tl_i,t is the rationalization degree of the industrial structure of the manufacturing industry, and tn_i,t is the advanced degree of the industrial structure of the manufacturing industry. The explanatory variable di_gi,t is the development level of the digital economy. The control variable pgdp_i,t is the level of economic development, gov_i,t is the government participation level, inv_i,t is the foreign investment level, tra_i,t is the foreign trade dependence, edu_i,t is the education input level, ene_i,t is the energy intensity, and ε_i,t is the random error that reflects the influence of other factors affecting the upgrading of the manufacturing industry′s industrial structure. Equation (4) is used to test the impact of the digital economy on the upgrading of the manufacturing industry’s structure in each province. If it is significant, then the development of the digital economy will impact the upgrading of the manufacturing industry structure.

②

Moderated mediating effect based on Bootstrap

Selecting the mediation test method with a moderating effect based on Bootstrap, this paper discusses the mediating effect of the digital economy on the upgrading of China’s manufacturing industry through technological innovation. Bootstrap’s confidence interval for calculating the product of coefficients has higher power in moderated mediation tests. The basic model is as follows [35]:

$$mi{u}_{i,t}={\alpha }_1+{\beta }_{1}di{g}_{i,t}+{\epsilon }_{i,t}$$

(5)

$$inn{o}_{i,t}={\alpha }_2+{\gamma }_{1}di{g}_{i,t}+{\gamma }_{2}co{n}_{i,t}+{\epsilon }_{i,t}$$

(6)

$$mi{u}_{i,t}={\alpha }^{’}{}_1+{\beta }^{’}{}_{1}di{g}_{i,t}+{\beta }^{’}{}_{2}inn{o}_{i,t}+{\beta }^{’}{}_{3}gv{c}_{i,t}+{\beta }^{’}{}_{4}di{g}_{i,t}\ast gv{c}_{i,t}+{\beta }^{’}{}_{5}co{n}_{i,t}+{\epsilon }_{i,t}$$

(7)

Among them, dig_it represents the digital economy, miu_it represents the upgrading of the manufacturing industry, inno_it represents the technological innovation of the mediator variable, con_it represents the control variable, gvc_it represents the moderator variable, β₁, γ₁, γ₂, β′₁, β′₂, and β′₃ are estimated parameter, α₁, α₂, and α′₁ are all constant, and ε_it is a random error term.

③

Moderating effect

$$\begin{array}{l}mi{u}_{i,t}=\alpha +{\beta }_{1}di{g}_{i,t}+{\beta }_{2}gv{c}_{i,t}+{\beta }_{3}di{g}_{i,t}\ast gv{c}_{i,t}+{\beta }_{4}pgd{p}_{i,t}+{\beta }_{5}go{v}_{i,t}+{\beta }_{6}in{v}_{i,t}+{\beta }_{7}tr{a}_{i,t}\\ +{\beta }_{8}ed{u}_{i,t}+{\beta }_{9}en{e}_{i,t}+{\epsilon }_{i,t}\end{array}$$

(8)

Equation (8) is used to test whether the global value chain plays a moderating role between the development of the digital economy and the upgrading of the manufacturing industry, and the significance of the regression coefficient of the interaction term is mainly examined.

(II)

Spatial spillover effects

This paper will use the economic distance matrix W₁ and the geographic distance matrix W₂ for spatial econometric analysis. The economic distance matrix describes the economic gap between the two regions, that is, the “distance” in the economic space. The economic distance matrix represents the economic gap between provinces, where g_i is the average value of real GDP in region i from 2006 to 2020. The formula is as follows:

$$w_{ij}=\{\begin{array}{ll}\frac{1}{\left|g_i-g_j\right|}& i\ne j\\ 0& i=j\end{array}$$

(9)

The elements on the principal diagonal of the geographic distance space weight matrix are 0, and the elements in the remaining positions are 1/d, where d is the geographic distance between the two provincial capitals. Compared with the geographic adjacency matrix, the geographic distance matrix can effectively measure the relationship between more distant spatial units. Let the latitude and longitude of the two points P and Q be (Lonp, Latp) and (Lonq, Latq), respectively, where R is the radius of the earth, and R = 6378.13 km. Then the geographic distance between two points can be expressed as:

$$d=2R\times arcsin\left(\sqrt{si{n}^2\left[\left(Latq-Latp\right)\times \frac{\pi }{360}\right]+cos\left(Latq\times \frac{\pi }{180}\right)\times cos\left(Latp\times \frac{\pi }{180}\right)\times si{n}^2\left[\left(Lonq-Lonp\right)\times \frac{\pi }{360}\right]}\right)$$

(10)

Generally speaking, spatial econometric models can be divided into spatial autoregressive model (SAR), spatial autocorrelation (SAC), spatial error model (SEM), and spatial Durbin model (SDM).

$$\{\begin{array}{l}mi{u}_{it}=\tau mi{u}_{i,t-1}+\rho w{\prime }_{i}mi{u}_t+dig{\prime }_{it}\beta +d{\prime }_{i}DI{T}_t\delta +u_i+{\gamma }_t+{\epsilon }_{it}\\ {\epsilon }_{it}=\lambda m{\prime }_i+v_{it}\end{array}$$

(11)

Among them, miu_it is the dependent variable of region i at time t, and miu_i,t−1 is the first-order lag of the explained miu_it. d′_iDIT_tδ represents the spatial lag of the explanatory variable, w′_imiu_t reflects the influence of other regional dependent variables on the local dependent variable, dig_itt represents the independent variable of region i at time t, and d′_i is the i-th row of the corresponding spatial weight matrix D. Variable u_i is the individual effect, γ_t is the time effect, and m′_i is the i-th row of the spatial weight matrix M of the disturbance term. ρ and λ are the spatial autocorrelation coefficient and the spatial error coefficient, respectively. If λ = 0, it is an SDM; if λ = 0 and δ = 0, it is a SAR; if τ = 0 and δ = 0, it is a SAC; if τ = ρ = 0 and δ = 0, it is an SEM.

5. Empirical Analysis

5.1. Mechanism Analysis

Benchmark regression analysis of the digital economy on China’s manufacturing upgrade

In

Table 2. Regression results of benchmark model.

	(1)	(2)	(3)	(4)	(5)	(6)
	tl	tl	tl	tn	tn	tn
infra	8.9871 ***			2.7108 *
	(3.18)			(1.92)
indus		5.6650 **			2.9572 **
		(2.34)			(2.12)
sci			0.4142			0.0493
			(0.81)			(0.26)
pgdp	0.0794	0.0779	0.0339	−0.2452 ***	−0.9828 ***	−0.2452 ***
	(1.50)	(1.24)	(0.48)	(−6.66)	(−14.52)	(−8.39)
gov	0.1607 **	0.1547	0.1538	0.9177 ***	0.2744 ***	0.9177 ***
	(2.31)	(1.28)	(1.33)	(8.11)	(2.81)	(8.11)
inv	0.1429 *	0.1480 *	0.1485 *	−0.0035	−0.0577	−0.0035
	(1.90)	(1.74)	(1.73)	(−0.08)	(−1.15)	(−0.07)
tra	0.2483 **	0.1407	0.1345	0.4106 ***	0.6854 ***	0.4106 ***
	(2.13)	(0.91)	(1.20)	(6.02)	(6.81)	(6.09)
edu	1.0058 ***	0.9163 **	0.9155 **	−0.4787 **	−0.4787 *	−0.4787 **
	(2.64)	(2.30)	(2.32)	(−2.16)	(−1.78)	(−2.18)
ene	−0.1706 **	−0.0594	−0.0569	−0.2628 ***	−0.2628 **	−0.3073 ***
	(−1.97)	(−0.63)	(−0.53)	(−3.69)	(−3.73)	(−4.30)
_cons	−0.2848 **	−0.3869 ***	−0.3874 ***	−0.0107	−0.0107	−0.0107
	(−2.06)	(−2.98)	(−3.01)	(−0.04)	(−0.13)	(−0.04)
N	434	434	434	434	434	434
R2	0.6358	0.6162	0.6253	0.5247	0.5168	0.5166

Standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.001.

, (1), (2) and (3) are listed as the test results of the relationship between the digital economy and the rationalization level of the industrial structure of China’s manufacturing industry, and (4), (5), (6) are listed as the test results of the relationship between the digital economy and the advanced level of the industrial structure of the manufacturing industry.

From columns (1) to (2), we conclude that the levels of digital infrastructure and digital development are significantly positively correlated with the rationalization level of the manufacturing industry structure in China’s provinces. The regression coefficients were 8.9871 and 5.6650, which were significant at the 1% and 5% levels, respectively. However, the scientific research level of digital technology fails the significance test. It can be seen from column (3) of Table 2 that although the scientific research level of digital technology has a positive impact on the rationalization of manufacturing in each province, it has not passed the significance test.

It can be seen from column (4) that the level of digital infrastructure construction is positively correlated with the advanced level of China’s manufacturing industry structure, and the regression coefficient is 2.7108, which is significant at the 10% level. It can be seen from column (5) that the level of digital development is positively correlated with the advanced level of China’s manufacturing industry structure, and the regression coefficient is 2.9572, which is significant at the 5% level. It can be seen from column (6) that although the scientific research level of digital technology has a positive impact on the advanced level of China’s manufacturing industry structure, it has not passed the significance test.

5.2. Analysis of the Mediating Effect of Technological Innovation on Digital Economy and Manufacturing Upgrade

Based on Bootstrap, the moderated intermediary test method is used to explore the impact mechanism of technological innovation on the digital economy and the upgrading of the industrial structure of the manufacturing industry.

①

Analysis of the mediating effect of technological innovation on the rationalization level of the digital economy and the industrial structure of the manufacturing industry

Table 3. Results of the mediating effect of technological innovation on the rationalization of the digital economy and the industrial structure of the manufacturing industry.

		Coefficient Value	Bootstrap Std. Err.	95% Confidence Interval
				Lower Limit	Upper Limit
infra	direct effect	−0.3113	−0.0112	−1.173	0.4325
	indirect effect	1.4031	0.0161	0.4336	2.5262
indus	direct effect	−0.1119	−0.0112	−1.0949	0.71
	indirect effect	1.2036	0.0318	0.2312	2.3278
sci	direct effect	−0.4057	−0.0846	−1.6491	0.3071
	indirect effect	1.4974	0.0965	0.3802	3.0427

shows the test results of the mediation effect of technological innovation on the rationalization level of the digital economy and the industrial structure of the manufacturing industry. The 95% confidence intervals of the direct effects of digital infrastructure construction level, digital development level, and the scientific research level of digital technology are [−1.1730, 0.4325], [−1.0949, 0.7100], and [−1.6491, 0.3071], respectively. The confidence interval includes 0, and there is no direct mediation effect. The 95% confidence interval of the indirect effect are [0.4336, 2.5262], [0.2312, 2.3278], and [0.3802, 3.0427]. The confidence interval does not contain 0, and there is an indirect mediation effect.

Therefore, technological innovation plays an intermediary role between the level of digital infrastructure construction, the scientific research level of digital technology, and the rationalization of the manufacturing industry structure.

②

Analysis of the mediating effect of technological innovation on the digital economy and the advanced level of manufacturing

Table 4. Results of the mediating effect of technological innovation on the digital economy and the advanced industrial structure of manufacturing.

		Coefficient Value	Bootstrap Std. Err.	95% Confidence Interval
				Lower Limit	Upper Limit
infra	direct effect	−0.0885	−0.0112	−0.3994	0.2502
	indirect effect	0.0458	0.0161	0.4765	1.3735
indus	direct effect	−0.2439	−0.0112	−0.5391	0.0556
	indirect effect	0.1096	0.0318	0.2851	1.528
sci	direct effect	−0.0878	−0.0846	−0.5095	0.1773
	indirect effect	0.0465	0.0965	0.4109	1.4901

shows the test results of the mediating effect of technological innovation on the digital economy and the advanced level of manufacturing industrial structure. The 95% confidence intervals of the direct effects of digital infrastructure construction level, digital development level, and the scientific research level of digital technology are [−0.3994, 0.2502], [−0.5391, 0.0556], and [−0.5095, 0.1773], respectively. The confidence interval includes 0, and there is no direct mediating effect. The 95% confidence intervals of the indirect effect are [0.4765, 1.3735], [0.2851, 1.5280], and [0.4109, 1.4901]. The confidence interval contains 0, and there is an indirect mediation effect. Therefore, technological innovation plays an intermediary role between the level of digital infrastructure construction, the level of digital development, the scientific research level of digital technology, and the advanced industrial structure of the manufacturing industry.

As shown in Figure 2, the manufacturing industry uses digital resources and digital technology to reasonably allocate the production factors of the regional manufacturing industry and reduce the transaction cost and marginal production cost of the manufacturing industry through technological innovation. In addition, the manufacturing industry carries out the low-end transformation of the traditional manufacturing industry to change the original output structure and output efficiency. It promotes the formation of a high-end manufacturing industry, reshapes the demand side, and promotes the upgrading of the regional manufacturing industry through technology spillover effects and industrial linkage effects.

5.3. Analysis of the Adjustment Effect of the Global Value Chain on the Upgrading of the Digital Economy and Manufacturing Industry

In

Table 5. Results of the Moderating Effect of Global Value Chains.

	(1)	(2)	(3)	(4)	(5)	(6)
	tl	tl	tl	tn	tn	tn
infra	6.1884 ***	6.2173 ***	4.3793 ***	6.4221 ***	9.5319 ***	6.4221 ***
	(2.82)	(2.98)	(2.94)	(3.91)	(4.21)	(3.63)
indus	7.1944 **	7.1055 ***	6.0888 ***	8.1703 ***	10.3403 ***	8.1703 ***
	(2.56)	(2.79)	(2.74)	(5.19)	(4.62)	(4.75)
sci	0.4327	0.3692	0.8724	0.5614 *	0.8348 *	0.5614 **
	(0.56)	(0.57)	(0.67)	(1.66)	(1.72)	(2.03)
gvcpar	58.8495 ***	52.1162 ***	35.3343 ***	49.0029 ***	42.8276 ***	34.9190 ***
	(22.10)	(21.41)	(19.98)	(10.80)	(10.62)	(9.99)
gvcpos	21.1240 ***	17.8219 ***	14.5937 ***	39.9308 ***	35.8772 ***	34.9896 ***
	(7.43)	(7.16)	(6.71)	(3.63)	(3.55)	(3.36)
rca	22.6854 ***	20.1805 ***	12.9229 ***	18.2000 ***	15.9606 ***	12.6934 ***
	(10.23)	(9.93)	(9.33)	(4.99)	(4.93)	(4.67)
gvcpar × infra	644.5272 ***			883.0751 ***
	(491.09)			(240.00)
gvcpos × infra	−464.8493 ***			−322.4585 ***
	(−146.34)			(−71.52)
rca × infra	−216.4589 ***			−358.4389 ***
	(−214.36)			(−104.76)
gvcpar × indus		320.6789 ***			534.5590 ***
		(420.81)			(208.73)
gvcpos × indus		−329.8900 ***			−168.1908 ***
		(−122.06)			(−60.54)
rca × indus		−89.5490 ***			−218.2698 ***
		(−183.65)			(−91.09)
gvcpar × sci			201.8589 ***			246.8885 ***
			(−298.10)			(149.14)
gvcpos × sci			−219.8763 ***			−122.1024 ***
			(−98.64)			(−49.35)
rca × sci			135.6437 ***			−94.9595 ***
			(132.58)			(−66.33)
pgdp	0.0009	0.0091	0.0237	−0.4838 ***	−0.2452 ***	−0.4838 ***
	(0.01)	(0.13)	(0.34)	(−14.32)	(−6.73)	(−10.13)
gov	0.1809 **	0.0832	0.0811	0.1667 ***	0.1549 **	0.1767 ***
	(2.17)	(1.21)	(1.17)	(2.58)	(2.13)	(2.62)
inv	0.1897 **	0.1917 **	0.2278 ***	0.0008	0.0035	0.0008
	(2.26)	(2.23)	(2.78)	(0.02)	(0.08)	(0.02)
tra	0.1649	0.1657	0.2124 *	0.5165 ***	0.4106 ***	0.5165 ***
	(1.42)	(1.41)	(1.81)	(8.89)	(6.09)	(4.95)
edu	1.1016 ***	1.1024 ***	0.9852 ***	−0.6898 ***	−0.4787 **	−0.6898 ***
	(2.96)	(2.95)	(2.67)	(−3.80)	(−2.18)	(−3.93)
ene	−0.1936 **	−0.1612 *	−0.1214	−0.1621 *	−0.1686 **	−0.2021 **
	(−1.99)	(−1.77)	(−1.23)	(−1.86)	(−1.96)	(−2.31)
_cons	−0.1103	−0.1371	−0.1220	0.3908 ***	0.4121 ***	0.4121 **
	(−0.78)	(−0.97)	(−0.85)	(5.38)	(5.70)	(2.37)
N	434	434	434	434	434	434
R2	0.5413	0.5409	0.5414	0.6464	0.6443	0.5439

Standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.001.

, columns (1)~(3) and (4)~(6) are the test results of the adjustment effect of the global value chain on the digital economy and the rationalization and advancement of the industrial structure of the manufacturing industry.

According to the results in columns (1)~(3), in the main effect, the regression coefficients of digital infrastructure construction and digital development level are 6.1884 and 7.1055, respectively, both of which are significant at the 1% level. Only the regression coefficient of digital technology research level 0.8724 failed the significance test. By examining the moderating effect, it is found that the interactive regression coefficients of the global value chain participation index and digital infrastructure construction, digital industry development, and digital technology research level are 644.5272, 320.6789, and 201.8589, all of which are significant at the 1% level. The above shows that the global value chain participation index strengthens the positive relationship between the construction of digital infrastructure and the level of digital development and the rationalization of the industrial structure of the manufacturing industry. The interactive regression coefficients of the global value chain status index and digital infrastructure construction, digital development level, and the scientific research level of digital technology are −464.8493, −329.8900, and −219.8763, respectively, all of which are significant at the 1% level. The above shows that the global value chain status index weakens the positive relationship between digital infrastructure construction and digital development level on the rationalization level of the manufacturing industry structure. The interactive regression coefficients of the added value advantage index and digital infrastructure construction, digital development level, and the scientific research level of digital technology are −216.4589, −89.5490, and 135.6437, respectively, all of which are significant at the 1% level. The above shows that the value-added advantage index weakens the positive relationship between digital infrastructure construction and digital industry development on the rationalization level of the manufacturing industry structure. The global value chain only has no regulating effect on the scientific research level of digital technology and the rationalization of the industrial structure of the manufacturing industry.

According to the results of (4)~(6), in main effects, the regression coefficients of digital infrastructure construction, digital development level, and the scientific research level of digital technology are 6.4221, 10.3403, and 0.5614, respectively. They are significant at the level of 1%, 1%, and 5%, respectively. Through adjustment effect, we found that the regression coefficient of the global value chain participation index and digital infrastructure construction, digital development level, and the scientific research level of digital technology are 883.0751, 534.5590, and 246.8885, respectively, all of which are significantly at a level of 1%. The above shows that the Global Value Chain Participation Index strengthens the positive relationship between digital infrastructure construction, digital development level, and the scientific research level of digital technology in the advanced manufacturing industry structure. The interactive regression coefficients of the GVC status index and digital infrastructure construction, digital development level, and the scientific research level of digital technology are −322.4585, −168.1908, and −122.1024, respectively, all of which are significant at the 1% level. The above shows that the global value chain status index weakens the positive relationship between digital infrastructure construction, digital development level, and the scientific research level of digital technology in the advanced manufacturing industry structure. The interactive regression coefficients of the added value advantage index and digital infrastructure construction, digital development level, and the scientific research level of digital technology are −358.4389, −218.2698, and −94.9595, respectively, all of which are significant at the 1% level. The above shows that the added value advantage index weakens the positive relationship between digital infrastructure construction, digital development level, and the scientific research level of digital technology in the advanced manufacturing industry structure.

As shown in Figure 3, the digital economy has reshaped the global value chain while changing the operating and resource allocation models of the manufacturing industry. The global value chain achieves the goal of maximizing profits by integrating advantageous digital resources and production factors. The global value chain promotes the development of the digital economy, and the commercial application of disruptive manufacturing innovation will spawn new industries, enterprises, and forms, promoting the rise of the manufacturing industry towards the high end of the global value chain, thereby achieving structural upgrading of the manufacturing industry.

5.4. Robustness Test

In order to improve the accuracy and reliability of the model test, the data from 2010 to 2020 are used to repeat the empirical part above to test the robustness of the relationship between the digital economy, technological innovation, global value chain, and manufacturing industry upgrading. As shown in

Table 6. Robustness test results.

	−1	−2	−3	−4	−5	−6
	tl	tl	tl	tn	tn	tn
infra	2.8855 **			9.9634 **
	−2.3988			−3.3694
indus		1.2173 **			5.2208 **
		−2.5311			−1.8251
sci			1.15			0.0019
			−0.633			−0.2621
pgdp	0.0165	0.0189	0.0314	−0.3419 ***	−0.4420 ***	−0.2435 ***
	−0.0799	−0.0875	−0.0881	(−7.5826)	(−8.0565)	(−6.6441)
gov	0.1665 **	0.1287 *	0.0701	0.9177 ***	0.2531 ***	0.9177 ***
	−2.2649	−1.7848	−0.085	−8.11	−2.642	−8.11
inv	0.1429 *	0.1480 *	0.1485 *	−0.0965	−0.1970 **	−0.0207
	−1.9	−1.74	−1.73	(−0.0893)	(−2.0864)	(−0.0428)
tra	0.3843 **	0.1137	0.1504	0.2238	0.3280 **	0.5275 ***
	−2.1218	−1.1205	−1.1165	−0.1635	−2.1629	−2.6654
edu	0.4233 **	0.5719 ***	0.5891 ***	−0.5064	−0.5893	−1.0539 ***
	−2.52	−2.5918	−2.5978	(−0.5574)	(−0.5570)	(−3.2615)
ene	−0.2815 **	−0.3731 **	−0.3705 **	−0.1407	−0.3089 **	−0.2505 **
	(−2.1221)	(−2.1643)	(−2.1645)	(−0.1354)	(−2.1315)	(−2.0889)
_cons	−0.0119	−0.1508	−0.1375	−0.3039	−0.2309	−0.3887 ***
	(−0.1918)	(−0.2006)	(−0.2005)	(−0.2448)	(−0.2232)	(−0.1087)
N	310	310	310	310	310	310
R2	0.59	0.5802	0.5461	0.4831	0.452	0.588

Standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.001.

, the conclusion on the relationship between the digital economy and the rationalization and development of the manufacturing industry structure has not changed. Technological innovation still plays a significant mediating role between the digital economy and manufacturing upgrading, and the global value chain has a significant moderating effect on the relationship between the digital economy and manufacturing upgrading. A significant part of the control variable has changed, but it does not affect the overall conclusion.

5.5. Spatial Spillover Effect

(I)

The constructed model is as follows:

①

The established spatial autoregressive model

$$\left(SAR\right) mi{u}_{it}=\rho w{\prime }_{i}mi{u}_t+{\beta }_{1}x{\prime }_{it}+{\beta }_{2}co{n}_{it}+{\beta }_{3}inn{o}_{it}+{\beta }_{4}gv{c}_{it}+u_i+{\gamma }_t+{\epsilon }_{it}$$

(12)

Among them, miu_it is the dependent variable of region i at time t, and miu_i,t−1 is the first-order lag of the explained miu_it. x′_it represents the independent variable of region i at time t, con_it is the control variable, inno_it is the mediator variable, and gvc_it is the moderator variable. u_i is the individual effect, and r_t is the time effect.

②

The established spatial error model (SEM) is as follows:

$$\{\begin{array}{l}mi{u}_{it}={\beta }_{1}x{\prime }_{it}+{\beta }_{2}co{n}_{it}+{\beta }_{3}inn{o}_{it}+{\beta }_{4}gv{c}_{it}+u_i+{\gamma }_t+{\epsilon }_{it}\\ {\epsilon }_{it}=\lambda m{\prime }_i{\epsilon }_t+v_{it}\end{array}$$

(13)

In the spatial error model (SEM), the spatial effect is caused by the random disturbance term in other regions, ε_it satisfies the assumption of homoscedasticity, m′_i is the ith row of the spatial weight matrix M of the disturbance term, and λm′_iε reflects the influence of random disturbance terms in other regions on the dependent variables in this region.

③

The SAC is as follows:

$$\{\begin{array}{l}mi{u}_{it}=\rho w{\prime }_{i}mi{u}_t+{\beta }_{1}x{\prime }_{it}+{\beta }_{2}co{n}_{it}+{\beta }_{3}inn{o}_{it}+{\beta }_{4}gv{c}_{it}+u_i+{\gamma }_t+{\epsilon }_{it}\\ {\epsilon }_{it}=\lambda m{\prime }_i{\epsilon }_t+v_{it}\end{array}$$

(14)

Among them, ρw′_imiu_t reflects the influence of other regional dependent variables on the local dependent variables.

④

The SDM is as follows:

$$mi{u}_{it}=\rho w{\prime }_{i}mi{u}_t+{\beta }_{1}x{\prime }_{it}+{\beta }_{2}co{n}_{it}+{\beta }_{3}inn{o}_{it}+{\beta }_{4}gv{c}_{it}+d{\prime }_i{X}_t\delta +u_i+{\gamma }_t+{\epsilon }_{it}$$

(15)

The ρw′_imiu_it represents the spatial effect of the dependent variable, d′_i is the ith row of the response spatial weight matrix D, and d′_iX_tδ represents the spatial lag of the explanatory variable, reflecting the spatial effect of the independent variable.

In order to improve the accuracy of the regression results, the spatial panel SAR, SEM, SAC, and SDM models are established and estimated based on the economic distance matrix W₁ and the geographic distance matrix W₂, and the best model is selected to analyze the global value chain, spatial spillover effects of the digital economy and manufacturing industry upgrading.

(II)

Spatial correlation test

According to the previous mechanism analysis of the global value chain, digital economy, and manufacturing upgrade, will the digital economy have an impact on the upgrade of the global value chain of manufacturing in China’s provinces due to the interaction between regions? The study will use a spatial econometric model to verify the above relationship.

From

Table 7. Spatial measurement results of industrial structure rationalization.

	SAR		SEM		SAC		SDM
	Economic Distance	Geographic Distance	Economic Distance	Geographic Distance	Economic Distance	Geographic Distance	Economic Distance	Geographic Distance
infra	8.244 ***	7.736 ***	3.529 **	7.959 **	2.635 **	8.042 **	7.946 ***	6.535 ***
	[2.8844]	[2.8904]	[2.5317]	[2.4263]	[2.3700]	[2.4554]	[2.5858]	[7.0075]
indus	8.548 ***	7.915 ***	4.634 **	6.826 ***	4.291 **	6.839 ***	7.896 ***	9.592 ***
	[5.4949]	[6.3019]	[2.3897]	[4.7264]	[2.2271]	[4.7164]	[5.7669]	[5.6662]
sci	0.342	0.286	1.154	0.259	1.047	0.276	0.511	1.697 *
	[0.7306]	[0.7222]	[0.5601]	[0.4793]	[0.5253]	[0.4716]	[0.7153]	[1.6710]
pgdp	0.0165	−0.0156	−0.163	0.0107	−0.0994	0.0278	0.00814	0.236
	[0.0725]	[0.0719]	[0.1330]	[0.0893]	[0.1297]	[0.0995]	[0.0721]	[0.2579]
gov	0.016	0.198 *	0.0315	0.1650 *	0.0708	0.1823 *	0.1515 *	0.2420 *
	[0.1392]	[1.9399]	[0.0864]	[1.7348]	[0.0717]	[1.7579]	[1.6516]	[1.7220]
inv	0.138	0.154	0.306	0.128	0.266	0.12	0.146	0.212
	[0.1068]	[0.1216]	[0.0742]	[0.0870]	[0.0722]	[0.0874]	[0.1141]	[0.1679]
tra	−0.138	0.132	−0.329	0.24	0.307	0.272	0.143	0.275
	[0.1825]	[0.2031]	[0.1259]	[0.1413]	[0.1186]	[0.1443]	[0.1700]	[0.2450]
edu	0.2041	0.304 *	0.483 *	0.505 **	0.389 *	0.576 **	0.648 **	0.327 *
	[1.3819]	[1.6991]	[1.8451]	[1.9857]	[1.7256]	[2.3193]	[2.4732]	[1.7173]
ene	0.0525	0.118	−0.0793	0.121	−0.0754	0.117	−0.0536	−0.102
	[0.1419]	[0.1871]	[0.1003]	[0.0931]	[0.0929]	[0.0940]	[0.1469]	[0.0881]
inno	0.46 *	0.515 **	2.677	0.7	2.535	0.681	0.439 *	1.533 *
	[1.9581]	[2.0021]	[0.4979]	[0.5656]	[0.4766]	[0.5629]	[1.9441]	[1.6839]
gvcpar	66.47 ***	60.16 ***	86.80 ***	78.15 ***	76.16 ***	87.39 ***	65.82 ***	24.03 ***
	[21.5652]	[20.8710]	[23.0811]	[15.4833]	[20.7737]	[17.9695]	[21.2139]	[10.7414]
gvcpos	−0.741 ***	3.422 ***	13.68 ***	10.33 ***	−10.34 ***	13.22 ***	−0.674 ***	17.91 ***
	[6.8273]	[8.1202]	[7.1724]	[6.5582]	[6.6326]	[7.6153]	[6.9671]	[13.3844]
rca	28.27 **	26.14 **	38.38 ***	34.42 ***	32.91 ***	38.67 ***	28.04 ***	38.26 ***
	[9.5752]	[9.5548]	[10.7569]	[7.2702]	[9.6716]	[8.4428]	[9.4243]	[14.9415]
W × infra							10.96 ***	839.4 ***
							[9.3866]	[563.1620]
W × indus							4.762 ***	1066.1 ***
							[10.5826]	[650.8169]
W × sci							−0.907 **	−159.4 ***
							[2.0052]	[113.1578]
rho	0.0222	4.699 ***			0.41	6.413 ***	0.017	6.858 ***
	[0.0653]	[7.5274]			[0.1237]	[7.4258]	[0.0631]	[8.0768]
lambda			0.17	11.39 ***	0.493	15.38 ***
			[0.1133]	[4.6500]	[0.1047]	[5.4344]
delta							−1.564 *	−1.12
							[1.8994]	[0.5681]
R2	0.5203	0.5556	0.5082	0.5221	0.4115	0.5463	0.6163	0.6907
Log-L	275.7411	262.9753	276.0833	265.0894	137.8259	264.69	289.3986	339.422

Standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.001.

, from the regression results of the economic distance spatial weight matrix, the spatial term coefficients of the SAR, SEM, and SAC models are 0.0222, 0.17, 0.410/0.493, respectively, which are not significant. The spatial coefficient of the SDM model is 0.017, which is not significant, and the spatial lag coefficient is −1.564, which is negative at the 10% level. From the regression results of the geographic distance spatial weight matrix, the spatial term coefficients of the SAR, SEM, SAC, and SDM models are 4.699, 11.39, 6.413/15.38, and 6.858, which are all significantly positive at the 1% level.

In this paper, the Wald test, LR test, Robust-LM test, and LM test are used to test the fitting effect of the model so as to judge whether the SDM model can be simplified into SAR, SEM, and SAC models. The results show that LM passed the 1% significance test, indicating that compared with the SAR, SEM, and SAC models, the SDM has a better explanatory effect, so the SDM model was selected for further analysis.

In the economic distance matrix, the spatial correlation coefficient of the SDM model is 0.017, which failed the significance level test. In the geographic distance matrix, the spatial correlation coefficient of SDM is 6.858, which passes the 1% confidence level and is significantly positive, indicating that the rationalization of China’s manufacturing industry has a significant geographic and spatial dependence. Additionally, the rationalization of the manufacturing structure of each province is also affected by neighboring provinces. The general regression coefficient and spatial regression coefficient of the digital infrastructure construction level and digital development level passed the 1% significance test under the spatial weight matrix of geographic distance, which means that the positive spillover effect of the digital infrastructure construction level and digital development level on the rationalization of manufacturing is established in the geographic distance matrix. The scientific research level of digital technology only passes the 10% significance level test, and the general regression coefficient is positive, indicating that the digital technology level has promoted the rational development of the manufacturing industry. However, because its spatial regression coefficient is negative, it shows that the scientific research level of digital technology has a negative spillover effect on the rationalization of manufacturing in neighboring provinces.

From

Table 8. Spatial measurement results of advanced industrial structure.

	SAR		SEM		SAC		SDM
	Economic Distance	Geographic Distance	Economic Distance	Geographic Distance	Economic Distance	Geographic Distance	Economic Distance	Geographic Distance
infra	0.658	1.102	0.0675	−0.032	0.393	1.404	0.206 *	0.106 **
	[0.9660]	[1.1779]	[0.6499]	[0.6237]	[0.7432]	[1.1118]	[1.7151]	[2.1577]
indus	4.468 **	4.957 **	0.504	0.268	2.693	5.755 **	0.303 *	0.862 *
	[2.1441]	[2.4479]	[0.6107]	[0.6183]	[1.5138]	[2.2356]	[1.6774]	[1.7749]
sci	−0.141	0.13	−0.183	0.197 *	0.103	0.161	0.153	0.274 *
	[0.1461]	[0.1619]	[0.1420]	[1.8460]	[0.1638]	[0.2400]	[0.1597]	[1.7718]
pgdp	−0.122	−0.218	0.0992	0.0862	−0.063	−0.206	0.0066	−0.0334
	[0.0460]	[0.0849]	[0.0330]	[0.0329]	[0.0202]	[0.0380]	[0.0536]	[0.1056]
gov	0.144	0.367 *	0.0482	0.076	0.0566	0.404 *	0.0665	0.2993 *
	[0.0975]	[1.7409]	[0.0174]	[0.0467]	[0.0416]	[1.8983]	[0.0254]	[1.7499]
inv	0.0331	0.0662	0.0498	0.079	0.00798	0.0688	0.0549	0.0614
	[0.0310]	[0.0300]	[0.0184]	[0.0200]	[0.0235]	[0.0383]	[0.0222]	[0.0209]
tra	0.0183	0.0432	−0.169	−0.202	0.0375	0.0499	0.0903	0.084
	[0.0719]	[0.1162]	[0.0315]	[0.0338]	[0.0354]	[0.0587]	[0.0747]	[0.1387]
edu	0.0182	0.137	0.122	0.138	0.0697	0.164	0.0359	0.00632
	[0.1316]	[0.1726]	[0.0873]	[0.0860]	[0.0742]	[0.1258]	[0.0991]	[0.1754]
ene	−0.12	−0.151	−0.15	−0.209	−0.091	−0.15	−0.116	−0.187
	[0.1145]	[0.1714]	[0.0255]	[0.0269]	[0.0238]	[0.0416]	[0.0945]	[0.1590]
inno	0.282 *	0.0173	−0.0617	0.0303	0.0816	0.1732	0.0035	0.2905 *
	[1.8068]	[0.2412]	[0.1212]	[0.1214]	[0.1721]	[1.2549]	[0.1837]	[1.8307]
gvcpar	0.954 **	0.44 ***	8.504 ***	9.978 ***	−0.42 **	0.0529 ***	7.185 **	8.914 ***
	[2.2109]	[2.8202]	[6.2911]	[5.4236]	[2.2290]	[4.0172]	[2.3992]	[5.1835]
gvcpos	6.333 *	10.05 ***	0.514 *	0.761	4.643 ***	9.360 *	2.53 ***	3.266 ***
	[1.7385]	[2.5975]	[1.9511]	[1.7360]	[1.1325]	[1.8643]	[2.0594]	[3.1674]
rca	1.477	2.349	−4.53 ***	−5.081 **	1.43	2.598 *	3.608 ***	4.336 **
	[1.2673]	[1.3551]	[2.9201]	[2.5204]	[1.0283]	[1.8945]	[1.0490]	[1.9780]
W × infra							4.568 ***	371.8 ***
							[4.1072]	[160.2788]
W × indus							4.82 ***	60.50 ***
							[4.4692]	[52.9102]
W × sci							0.0834	1.143 ***
							[0.5628]	[41.9729]
rho	0.805	22.79 **			0.908	23.58	0.665	17.75 ***
	[0.0576]	[2.2823]			[0.0162]	[1.0068]	[0.0785]	[3.0937]
lambda			−0.346	−33.14 ***	−0.981	−11.71 ***
			[0.0936]	[6.1869]	[0.0738]	[5.9168]
delta							13.37 ***	16.13
							[3.6137]	[0.7003]
R2	0.7545	0.6604	0.6579	0.6366	0.6914	0.678	0.7953	0.728
Log-L	636.6656	569.1976	736.3047	726.0684	697.4565	571.1742	658.4559	572.4805

Standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.001.

, from the regression results of the economic distance spatial weight matrix, the spatial term coefficients of the SAR, SEM, and SAC models are 0.805, −0.346, and −0.908/−0.981, respectively, which are not significant. The spatial coefficient of the SDM model is 0.665, which is not significant, and the spatial lag coefficient is 13.37, which is significant at the 1% level. From the regression results of the spatial weight matrix of geographic distances, the spatial term coefficients of the SAR and SEM models are 22.79 and −33.14, respectively, both of which are significant at the 1% level. The spatial autocorrelation coefficient of the SAC model is 23.58, which is not significant, and the spatial error coefficient is −11.71, which is significant at the 1% level. The spatial coefficient of the SDM model is 17.75, which is significant at the 1% level. In this paper, the Wald test, LR test, Robust-LM test, and LM test were used to test the fitting effect of the model. LM passed the 1% significance test, indicating that compared with the SAR, SEM, and SAC models, SDM has a better explanatory effect, so the SDM model is selected for further analysis.

In the economic distance matrix, the spatial coefficient of the SDM model is 0.665, which fails the significance level test. In the geographic distance matrix, the spatial coefficient of the SDM model is 17.75, which passed the 1% significance test, and the spatial spillover effect was significantly positive, indicating that the industrial structure of the manufacturing industry in each province is highly dependent on geographic space. The general regression coefficient of the digital infrastructure construction level passed the 5% significance test under the geographic distance matrix, and its spatial regression coefficient passed the 1% significance test, indicating that the positive spillover effect of digital infrastructure construction level on the advanced manufacturing industry is established in the geographic distance matrix. The general regression coefficient of the digital development level passed the 10% significance level test, while the spatial regression coefficient passed the 1% significance level test, and the regression coefficient was negative. It shows that the level of digital development has promoted the advanced development of manufacturing. However, since the spatial regression coefficient is negative, it shows that the level of digital development has a negative spillover effect on the advanced manufacturing of the neighboring provinces. The general regression coefficient of the scientific research level of digital technology has passed the 10% significance test, and the regression coefficient is positive, indicating that the scientific research level of digital technology has promoted the advanced development of the manufacturing industry. However, since its spatial regression coefficient is negative, it shows that the level of digital technology research has a negative spillover effect on the advanced manufacturing of the neighboring provinces.

From the geographic distance matrix in

Table 9. Direct effect, indirect effect, and total effect of SDM of manufacturing rationalization and advancement.

	Manufacturing Rationalization			Manufacturing Advancement
Variable	Direct Effect (1)	Indirect Effect (1)	Total Effect (1)	Direct Effect (2)	Indirect Effect (2)	Total Effect (2)
infra	7.78 ***	62.97 ***	70.75 ***	0.906	29.32 ***	30.22 ***
	[7.2051]	[37.2927]	[39.0368]	[1.2201]	[11.7655]	[12.3890]
indus	11.03 ***	78.55 ***	89.58 ***	2.668 ***	−1.55 **	1.118 *
	[6.1604]	[43.6041]	[46.6130]	[2.6892]	[2.3164]	[1.6582]
sci	1.895 **	−1.328 ***	1.567 *	0.229	0.171 ***	0.4 ***
	[2.0550]	[2.8062]	[1.7099]	[0.2729]	[3.6769]	[3.9197]
pgdp	−0.231	−0.151	−0.382	−0.0575	−0.817	−0.875
	[0.2755]	[0.2528]	[0.4801]	[0.1131]	[0.3584]	[0.4294]
gov	0.3311 *	0.0172	0.4483 *	0.0367	2.361 **	2.3977 **
	[1.6737]	[0.3294]	[1.8661]	[0.0663]	[2.4831]	[2.5329]
inv	0.207	0.112	0.32	0.0696	0.248	0.317
	[0.1777]	[0.1356]	[0.2823]	[0.0257]	[0.4371]	[0.4532]
tra	0.285	0.214	0.499	−0.0593	0.993	0.934
	[0.2443]	[0.2627]	[0.4681]	[0.1283]	[0.4482]	[0.4429]
edu	0.656 *	0.0641	0.7201 **	0.0205	0.976	0.997
	[1.8209]	[0.4574]	[1.9671]	[0.1550]	[1.3811]	[1.3694]
ene	−0.101	−0.0685	−0.169	−0.226	−1.075	−1.3
	[0.0855]	[0.0936]	[0.1608]	[0.1460]	[1.1899]	[1.2235]
inno	1.623 *	0.92	2.44 **	1.74 ***	0.147	1.888 ***
	[1.9354]	[1.4619]	[2.4869]	[3.0945]	[0.1771]	[3.2100]
gvcpar	84.09 ***	58.49 ***	142.6 ***	8.377 ***	−5.08 ***	3.297 ***
	[30.0689]	[48.8187]	[66.8641]	[5.5654]	[3.5279]	[2.6487]
gvcpos	18.41 ***	−12.85 ***	5.56 ***	4.601 ***	47.28 ***	51.88 ***
	[12.9485]	[14.3047]	[4.4696]	[2.9736]	[13.6727]	[12.5782]
rca	38.23 ***	−16.84 ***	21.39 ***	17.46 ***	−3.813 *	13.65 ***
	[14.7914]	[11.2163]	[12.5659]	[8.2128]	[1.9434]	[7.8100]

Standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.001.

, we conclude that the spatial autocorrelation coefficient of the SDM model is significantly positive at the 1% confidence level, which indicates that there is a significant geographic spatial dependency in the rationalization and advancement of China’s manufacturing structure. Moreover, the upgrading of the manufacturing structure in each province is also affected by neighboring provinces to a certain extent. From the direct effect (1), indirect effect (1), and total effect (1), it can be found that the coefficients of the digital infrastructure construction level and digital development level all pass the 1% significant level and have a positive effect. This shows that the level of digital infrastructure construction and digital development can promote the rationalization of manufacturing in the region and provinces with similar geographical distances. From the direct effect (1), we found that the coefficient of the scientific research level of digital technology is 1.895, which has a positive effect through the 5% significance level. However, from the indirect effect (1), the coefficient is −1.328, and it has a negative effect through the 5% significance level. This shows that the level of scientific research on digital technology has promoted the rationalization level of manufacturing in the region, but it has a certain inhibitory effect on the rationalization level of manufacturing in its neighboring provinces. The technological innovation of the intermediary variable promotes the level of manufacturing rationalization in this region, but the promotion effect of its neighboring provinces is not significant. From the direct effect (1), indirect effect (1), and total effect (1), it can be seen that the coefficients of the global value chain participation index all passed the 1% significant level and had a positive effect. That shows that the global value chain participation index has a promoting role in the relationship between the digital economy and the rationalization level of manufacturing in the region and also has a certain role in promoting its neighboring provinces. From the direct effect (1), we conclude that the coefficients of the global value chain status index and the domestic value-added comparative advantage index both through the 1% significance level and have a positive effect. However, from the indirect effect (1), we conclude that its coefficient through the 1% significance level has a negative effect. That shows that the global value chain status index and the domestic value-added comparative advantage index have a promoting effect on the relationship between the digital economy and the rationalization level of the manufacturing industry in the region but have a certain inhibitory effect on its neighboring provinces.

From the direct effect (2), the coefficient of digital infrastructure has a positive effect but is not significant. Indirect effects (2) and total effects (2) passed the 1% level of significance. That shows that digital infrastructure has a certain role in promoting the advanced manufacturing of provinces with similar geographical distances. From the direct effect (2), indirect effect (2), and total effect (2), we conclude that the coefficient of scientific research level of digital technology has passed the 1% significant level and has a positive effect. That shows that the level of scientific research level of digital technology has a certain role in promoting the advanced manufacturing of the region and provinces with similar geographical distances. From the direct effect (2), we conclude that the coefficient of digital development level is 1.118, and only passed the 10% significance level and has a positive effect. However, from the indirect effect (2), we conclude that its coefficient is −1.55, and it passed the 5% significance level and has a negative effect. That shows that the scientific research level of digital technology has promoted the advanced level of manufacturing in the region, but it has a certain inhibitory effect on the advanced level of manufacturing in its neighboring provinces. Technological innovation as the intermediary variable promotes the relationship between the digital economy and the advanced level of manufacturing in the region and geographically close regions. From the direct effect (2), indirect effect (2), and total effect (2), we conclude that the global value chain status index is used as an adjustment variable, and its coefficients all pass the 1% significant level and have a positive effect. That shows that the global value chain status index has a role in promoting the relationship between the digital economy and advanced level of manufacturing in the region and also has a certain role in promoting its neighboring provinces. From the direct effect (2), we conclude that the coefficients of the global value chain participation index and the domestic value-added comparative advantage index have both passed the 1% significance level and have a positive effect. However, from the indirect effect (2), we conclude that its coefficient has a negative effect through the 1% significance level. Tha0074 shows that the global value chain participation index and the domestic value-added comparative advantage index have a promoting effect on the relationship between the digital economy and the advanced level of manufacturing in the region but have a certain inhibitory effect on its neighboring provinces.

6. Conclusions

This paper selects the panel data of 31 provinces in China from 2006 to 2020 and analyzes the mechanism of the digital economy on the upgrading of China’s manufacturing global value chain. Then, two spatial weight matrices, an economic distance matrix and a geographic distance matrix, are used to construct SAR, SEM, SAC, and SDM models to explore the spatial spillover effect of the digital economy on the upgrading of the global value chain of manufacturing in China’s provinces.

Technological innovation plays an intermediary role between the level of digital infrastructure construction, digital development, scientific research of digital technology, and the rationalization and advancement of the manufacturing industry structure. The development of the digital economy enables digital resources to promote the rational use and allocation of various resources. Technological innovation has driven the application of the digital economy to all fields of the manufacturing industry, improving the rationalization and efficiency of resource allocation. The rational allocation of digital resources can effectively promote the digitization process of manufacturing in various regions. Technological innovation has promoted the digital economy to upgrade the manufacturing industry in the region, but the spillover effect of the rationalization space on its neighboring provinces is not significant. The degree of government participation as a control variable has a positive spatial spillover effect on the advanced manufacturing industry structure, and the spatial spillover effect of other control variables is not significant. The global value chain has a positive regulating effect on the relationship between the development of the digital economy and the upgrading of the manufacturing industry structure. Under the economic distance spatial weight matrix, there is no significant spatial spillover effect of the digital economy on the upgrading of the global value chain in China’s provinces’ manufacturing industry. However, under the geographic distance spatial weight matrix, there is a spatial spillover effect of the digital economy on the upgrading of the global value chain in China’s provinces’ manufacturing industry, and compared to SAR, SEM, and SAC models, the SDM model has a better explanatory effect and is significantly positive. This indicates that there is a significant geographic spatial dependence on the industrial structure upgrading of China’s provinces’ manufacturing industry, and the industrial structure upgrading of each province’s manufacturing industry is also positively influenced by the nearby provinces to a certain extent. Therefore, the upgrading of the manufacturing industry in one province will drive the manufacturing industry in the surrounding underdeveloped areas towards rationalization and sophistication.