Study on the Impact of the Digital Economy on the Upgrading of Industrial Structures—Empirical Analysis Based on Cities in China

Abstract

This paper used the entropy method to measure the comprehensive level of China’s urban regional digital economy development and measured the transformation and upgrading of industrial structures from both quantitative and qualitative dimensions. Based on the panel data of 271 prefecture-level cities in China from 2011 to 2018, this paper investigated the impact and transmission mechanisms of digital economy development on the transformation and upgrading of industrial structures through a fixed effects model, mediating effect model, and threshold regression model. The results show the following: First, the development of the digital economy has a significant positive role in promoting the quantity and quality of the upgrading of industrial structures, and has a stronger explanatory power for the quality of the upgrading of industrial structures. Second, the mechanism analysis shows that the digital economy can accelerate the transformation and upgrading of industrial structures by stimulating the level of regional innovation. Third, the analysis of the threshold regression model shows that the impact of the digital economy on the upgrading of industrial structures has nonlinear characteristics. Fourth, the impact of the digital economy on the upgrading of industrial structures has regional heterogeneity. It has the greatest impact in the western region, followed by the central region and the eastern region.

1. Introduction

China’s economy has entered a new era. The report of the 19th National Congress of the Communist Party of China pointed out that China’s “economy has shifted from a stage of high-speed growth to a stage of high-quality development, and is in a critical period of transforming the development mode, optimizing the economic structure, and transforming the driving force of growth”. General Secretary Xi Jinping emphasized: “To promote high-quality economic development, we must focus on promoting the transformation and upgrading of the industrial structure.” This means that continuous industrial structure optimization is a necessary condition for economic growth, and it is also an inevitable choice for high-quality economic development. Thus, what is the driving force for the transformation and upgrading of China’s industrial structure? With a new round of scientific and technological revolutions sweeping the world, the digital economy has become a key driving force for China’s economic development, injecting new vitality into industrial transformation and upgrading.

China attaches great importance to the development of the digital economy and has elevated it to a national strategy. The Fifth Plenary Session of the 18th Central Committee of the Communist Party of China proposed to implement the strategy of strengthening the country through the internet and the national big data strategy. The ministries and commissions of the State Council have also issued a series of policies and measures to promote the development of the digital economy, and have built a relatively complete policy support system for the development of the digital economy. According to the relevant research data of the China Institute of Information and Communication in 2020, the scale of China’s digital economy had increased from 2.6 trillion in 2005 to 39.2 trillion, an increase of 3.3 trillion over 2019, accounting for 38.6% of the GDP; the growth rate of the digital economy was 3.2 times that of the nominal GDP in the same period. The digital economy has become the main driving factor in China’s economic growth and industrial structure adjustment [1]. Therefore, it is of practical significance to study and analyze the impact of the digital economy on the upgrading of the industrial structure. How to effectively release the dividends of the digital economy on the transformation and upgrading of the industrial structure has become a topic of concern for the academic community.

Although some studies have begun to explore the impact of the digital economy on the upgrading of the industrial structure, there are still many questions that need to be answered urgently. First, what is the internal mechanism for the digital economy to upgrade the industrial structure? Second, considering the differences in China’s geographic regions and city levels, is there any heterogeneity in the mechanism of the digital economy’s effect on the industrial structure? Third, considering the externalities of the network and the scale effect of the network, does the digital economy have nonlinear characteristics in upgrading the industrial structure? At present, these problems are relatively rare, and this is the focus of this paper. Compared with the previous literature, the main innovations of this paper are as follows: (1) using the data of Chinese prefecture-level cities, it provides a more detailed and reliable empirical analysis for research on the digital economy; (2) measuring the changing of industrial structures from the perspective of quality and quantity is an innovation in this field of research.

2. Literature Review

2.1. Summary of the Relevant Literature

In terms of in-depth exploration of the mechanism of the digital economy on the upgrading of the industrial structure, the relevant literature mainly focuses on the following three aspects:

First, the research related to the digital economy. At present, there is no unified consensus on the accounting of digital economy indicators. Most of the literature regarding the digital economy is qualitative [2,3]; relatively few quantitative studies have been conducted, and they all stay at the national and provincial levels. As for the analysis of the role of the digital economy, the existing literature has mainly studied the micro, meso and macro levels. The research at the micro level mainly focuses on the field of digital consumption. In the digital era, the needs of consumers are accurately identified and met, and the needs of the minority and personalized needs are gradually met, which stimulates a stronger long tail effect. The reasonable development of digitalization is conducive to generating positive externalities, generating the Matthew effect and expanding the consumption scale [4,5]. The research at the middle level mainly focuses on the impact of the digital economy on traditional industries. Digital transformation has led to intelligent and customized information processing, strengthened data analysis, and improved production efficiency, and has injected new vitality into traditional industries, and has reconstructed the Industrial Organization [6,7]. The digital transformation of traditional industries will promote cross-border integration of industries, form a digital ecology, and better strengthen the interaction between supply and demand [8]. The macro level research mainly focuses on the economic growth theory. As a new production factor, the combination of data and labor factors will improve production efficiency and quality [9]. In addition, to its sharing and repeatability, the data break through the limitations and exclusiveness of the original production factors and realizes the incremental marginal return [10].

Second is the study of industrial structure upgrading. This mainly includes the following three aspects: The first is the policy. Appropriate industrial policies are conducive to improving industrial innovation efficiency [11] and promoting industrial structure upgrading [12]. The optimization of the institutional environment can increase the proportion of investment in specialized investment industries and promote industrial transformation and upgrading [13]. The second is infrastructure. The role of high-speed rail is to improve the accessibility between regions, reduce the time cost, promote trade [14,15], and influence the industrial structure through the division of the labor effect, convergence effect, and learning effect [16]. The third is technological progress. Internet technology promotes the transformation and upgrading of industrial structure by promoting the improvement of total factor productivity, labor productivity, and technical efficiency [17,18].

Finally, the research regarding the impact of the digital economy on industrial structure adjustment is mainly reflected in the following aspects: First, some scholars believe that the digital economy can promote innovation ability [19,20,21] and improve total factor productivity [22], so as to promote the upgrading and evolution of industrial structures. Second, some scholars analyzed the impact of the digital economy on industrial structures from the perspective of inputs and outputs, and believe that digital technology leads to the transfer of the value chain, which leads to the changing of industrial structures [23,24]; of course, some scholars believe that the combination of digital technology and finance can lead to the upgrading of industrial structures [25]. Due to the great differences in regional economic development in China, the impact of the digital economy on industrial structures has certain regional differences [26].

To sum up, there are many valuable studies on the digital economy and the industrial structure, but there are still the following shortcomings: first, most of the relevant literature on the digital economy are qualitative studies, few are quantitative studies, and they stay at the national and provincial levels, with research lacking at the city level; second, there are few studies analyzing the transmission mechanism of the digital economy to industrial structure upgrading. In view of this, this paper tries to make up for the shortcomings of the existing literature. Based on the panel data of 271 prefecture-level cities in China from 2011 to 2018, this paper studies the impact of digital economy development on the upgrading of industrial structures by using econometric models, such as the threshold effect, and studies the transmission mechanisms of the two, with the level of regional innovation as an intermediary variable. Scientific evaluation of the impact of the digital economy on the upgrading of the industrial structure has an important guiding role and practical significance for the development planning of the digital economy, especially for the high-quality development of China’s economy.

2.2. Theoretical Analysis and Research Hypothesis

The upgrading of industrial structures mainly refers to the process of, or trend in industrial structures changing from a low-level form to a high-level form. It is also the changing of the economic growth mode and the economic development mode. As a new economic form, the digital economy impacts the existing economic operation mode, activity rules, and economic development mode [27]. The digital economy can give full play to the advantages of information intermediaries, reduce transaction costs, improve the efficiency of resource allocation, and promote the upgrading of industrial structures. In addition to having a direct impact on industrial structures by virtue of its own characteristics, the digital economy can also have an indirect impact on the upgrading of industrial structures by affecting the innovation level of the region. At the same time, considering the network externality effect and network scale effect, the impact of the digital economy on the upgrading of industrial structures may also have nonlinear characteristics.

2.2.1. Digital Economy and the Upgrading of Industrial Structures

The progress of digital technology can play a positive role in promoting the optimization of industrial structures from the macro, medium, and micro levels. First, from the perspective of the macro level, on the one hand, based on digital technology, the government can provide a better business environment and guide capital to more efficient departments [28]; on the other hand, the progress of digital technology can improve total factor productivity, promote economic growth, and optimize industrial structures [29,30]. From the perspective of meso-industry, the progress of digital technology has promoted the emergence of new business forms, reconstructed the industrial system through industrial digitalization and digital industrialization, cultivated new kinetic energy and new space for economic growth, and accelerated the changing of industrial structures [31]. From the microenterprise level, the application of digital technology has formed an economic environment with economies of scale, economies of scope, and long tail effects [3]. Digital transformation is conducive to the specialization of enterprises [32,33]; enterprises’ digital transformation can effectively develop customers and form a better match between supply and demand; thus, it can optimize resource allocation and promote the transformation and upgrading of the industrial structure [34].

Hypothesis 1 (H1).

The digital economy has a significant positive contribution to the upgrading of industrial structures.

2.2.2. Digital Economy, Regional Innovation Level, and the Upgrading of Industrial Structures

Combing the existing literature, we can see that the digital economy can have a positive impact on industrial structure upgrading and economic growth has basically reached a consensus, but the mechanism analysis of the digital economy on industrial structure upgrading is relatively small. Some scholars have pointed out that China’s digital transformation is essentially an acceleration of innovation [35]. The improvement of the regional innovation level can improve total factor productivity and promote the upgrading of industrial structures, so as to change the mode of economic growth and promote high-quality economic growth [36]. The development of the digital economy plays a positive role in promoting the level of regional innovation. On the one hand, the construction of digital infrastructure eliminates the limitations of geographical location, reduces information asymmetry, reduces the communication costs of enterprises, universities, and other innovative subjects, and thus improves innovation [37,38]. On the other hand, the wide application of digital technology makes the dissemination and application of knowledge more convenient, improves the possibility of knowledge acquisition, facilitates people searching for as well as absorbing external knowledge and accumulating experience, and promotes the improvement of innovation efficiency [39,40].

Hypothesis 2 (H2).

The digital economy can indirectly promote the upgrading of industrial structures by enhancing the level of regional innovation.

2.2.3. Nonlinear Transmission Mechanism of the Digital Economy

The impact of the digital economy on the transformation and upgrading of industrial structures is nonlinear, that is, in the different stages of economic development, the impact of the digital economy on the transformation and upgrading of industrial structures is different. On the one hand, the digital economy optimizes production, circulation, distribution, exchange, and other links to improve production efficiency; however, with the continuous improvement of the level of economic development, the contribution of the digital economy to economic growth will decrease marginally [41]. On the other hand, with the continuous improvement of the digital infrastructure and technical level, the industry tends to be “service oriented” [42]. At the same time, with the development of the Internet of things, artificial intelligence, and other technologies, digital technology also plays a great role in promoting the efficiency of manufacturing and agricultural production [43], and the dividend released by the digital economy on the transformation and upgrading of industrial structures will decrease.

Hypothesis 3 (H3).

The digital economy has nonlinear characteristics in promoting the upgrading of industrial structures.

3. Materials and Methods

3.1. Econometric Model

In order to analyze the impact of the digital economy on the upgrading of industrial structures, an empirical analysis model that includes the digital economy and the upgrading of industrial structures is constructed in this paper:

$$lnai{s}_{it}={\beta }_0+{\beta }_{1}lndg{c}_{it}+{\beta }_{2}ln{X}_{it}+u_i+{\epsilon }_{it}$$

(1)

where $lnai{s}_{it}$ represents the level of the upgrading of industrial structures of city i in period t; $lndg{c}_{it}$ represents the level of the digital economy of city i in period t; $ln{X}_{it}$ represents a series of control variables that may affect the upgrading of industrial structures; $u_i$ represents the unobservable individual fixed effects of city $i$; ${\epsilon }_{it}$ represents the random disturbance term; ${\beta }_0$ is the intercept of the model term; and ${\beta }_1$ reflects the impact of the level of digital economy development on the upgrading of industrial structures.

To test whether the regional innovation level ($inn$) plays a mediating role in the process of promoting the upgrading of industrial structures in the digital economy, a mediating effect regression model was set using a stepwise regression method with reference to the study of [44]:

$$lnin{n}_{it}=a_0+a_{1}lndg{c}_{it}+a_{2}ln{X}_{it}+u_i+{\epsilon }_{it}$$

(2)

$$lnai{s}_{it}=y_0+y_{1}lndg{c}_{it}+y_{2}lnin{n}_{it}+y_{3}ln{X}_{it}+u_i+{\epsilon }_{it}$$

(3)

Based on the model assumption, it implies that the influence of the digital economy and the regional innovation level on regional industrial structure upgrading is multidimensional, and its influence may be characterized differently with the level of digital economy development and the innovation technology level being in different intervals, i.e., there may be a nonlinear relationship between the variables. To test whether there is a nonlinear relationship between the variables, the panel threshold model proposed by [45] is used here for testing, and the specific regression model is as follows:

$$lnai{s}_{it}={\delta }_0+{\delta }_{1}lndg{c}_{it}\times I\left(Ad{j}_{it}\le \theta \right)+{\delta }_{2}lndg{c}_{it}\times I\left(Ad{j}_{it}\ge \theta \right)+{\delta }_3{X}_{it}+u_i+{\epsilon }_{it}$$

(4)

where $Ad{j}_{it}$ is the threshold variable; $\theta$ is the unknown threshold; and $I$ is the indicator function, which takes the value of 1 when the condition is satisfied and 0 otherwise. Equation (4) can be expanded from a single-threshold to a multi-threshold model according to the corresponding measurement test.

3.2. Variable Measurement and Description

3.2.1. Explanatory Variable: The Upgrading Level of Industrial Structures

This paper intends to use the panel data of 271 prefecture-level cities in China from 2011 to 2018 for empirical testing, and the explanatory variable is the upgrading level of industrial structures. The main indicators of the upgrading level of industrial structures are the industrial structure rationalization index, the industrial structure advanced index, and the industrial structure upgrading index. In this paper, referring to the literature of [29,45], the industrial structure advanced index was used to measure the level of the upgrading of industrial structures. Many previous studies have measured the industrial structure advanced index with a simple share, ignoring the essence of the industrial structure evolution and easily resulting in a quantitative “false height”. In fact, in addition to the increase in quantity, the upgrading of industrial structures should include the improvement of quality. The quality of the upgrading of industrial structures includes the evolution of the proportional relationship and the improvement of labor productivity. Therefore, this paper measures the impact of the digital economy on different attributes of the upgrading of industrial structures from two views of quantitative and qualitative aspects. Among them, the quantity index of the upgrading of industrial structures is expressed by the industrial structure hierarchy coefficient with the following formula:

$$ais{1}_{it}={{\displaystyle \sum }}_{m=1}^3{y}_{imt}\times m,m=1,2,3$$

(5)

where $y_{imt}$ denotes the proportion of industry $m$ in region i to the regional GDP in period $t$ and $ais{1}_{i,t}$ denotes the quantity of the upgrading of industrial structures in region $i$ in period $t$. This index reflects the dynamic development process of the three major industries in China, in which the evolution from the dominance of primary industry to the dominance of secondary and tertiary industries occurs gradually, which is the connotation of the amount of highly advanced industrial structures.

The qualitative concept of the upgrading of industrial structures is defined as the weighted value of the product of the proportional relationship between industries and the labor productivity of each industry:

$$ais{2}_{it}={{\displaystyle \sum }}_{m=1}^3{y}_{imt}\times l{p}_{imt},m=1,2,3$$

(6)

$$l{p}_{imt}=\raisebox{1ex}{$Y_{imt}$}\!\left/ \!\raisebox{-1ex}{$L_{imt}$}\right.$$

(7)

where $ais{2}_{it}$ denotes the quality of the upgrading of industrial structures in region $i$ in period $t$; $l{p}_{imt}$ denotes the labor productivity of industry m in region i in period t; $Y_{imt}$ denotes the value added of industry m in region $i$ in period $t$; and $L_{imt}$ denotes the employees in industry m in region $i$ in period $t$. Here, $y_{imt}$ does not have a dimension, while $l{p}_{imt}$ has a dimension. In this paper, we adopt the method of homogenization to eliminate the dimension, so that the quality of advanced industrial structures does not have a dimension.

3.2.2. Explanatory Variable: Development Level of the Digital Economy

There is less literature on the measurement of the level of digital economy development, and it is more focused on the provincial level. In this paper, with reference to the method of [20] and combining the availability of relevant data at the city level, the level of digital economy development was measured in terms of both internet development and digital inclusive finance. The level of digital finance adopts the digital financial inclusion index, jointly compiled by the Digital Finance Research Center of Peking University and the Ant Group [46]. The specific measurement indicators are shown in

Table 1. Evaluation index system of the digital economy development level.

Primary Index	Secondary Index	Indicator Description
Digital economy development level	Internet penetration rate (+)	Number of internet users per 100 people
	Number of internet-related employees (+)	Computer service and software employees as a percentage of urban unit employees
	Internet-related output (+)	Total telecommunications services per capita
	Number of mobile internet users (+)	Number of cell phone users per 100 people
	Digital access to inclusive development (+)	China digital financial inclusion index

Note: “+” in parentheses represents positive indicators.

.

This paper adopts the entropy method of the objective assignment method to measure the level of digital economy development, which can avoid the inaccuracy of the subjective assignment method measurement. The level of digital economy development is measured using the entropy method, and the specific measurement method is as follows:

Standardization of the indicators:

$$P_{ijt}=\frac{Z_{ijt}-Min(Z_{ijt})}{Max\left(Z_{ijt}\right)-Min\left(Z_{ijt}\right)}$$

(8)

$Z_{ijt}$ is the observed value of indicator j of the evaluated city i in year t. Equation (9) is the indicator weight, and Equation (10) is the entropy formula:

$$Q_{ijt}=\frac{P_{ijt}}{{{\displaystyle \sum }}_{i=1}^{242}{{\displaystyle \sum }}_{t=2011}^{2018}{P}_{ijt}}$$

(9)

$$E_j=-k{{\displaystyle \sum }}_{i=1}^{271}{{\displaystyle \sum }}_{t=2011}^{2018}{Q}_{ijt}\mathit{ln}\left(Q_{ijt}\right)$$

(10)

To ensure that $ln\left(Q_{ijt}\right)$ in Equation (10) is meaningful, it is replaced by 0.00001 when taking the value of 0. Q is the indicator weight and E is the entropy value. Where $k=\raisebox{1ex}{$1$}\!\left/ \!\raisebox{-1ex}{$ln\left(hm\right)$}\right.$, $h$ represents the number of cities and $m$ represents the time span. The indicator coefficient of variation calculated using $G_j=1-E_j$, and then the indicator weight, $W_j$ is obtained:

$$W_J=\frac{G_j}{{{\displaystyle \sum }}_{j=1}^5{G}_j}$$

(11)

The composite index value, $I_{it}$, is calculated by summing up the weighting method:

$$I_{it}=100\ast {{\displaystyle \sum }}_{j=1}^5{W}_j{P}_{ijt}$$

(12)

In Formula (2), the core explanatory variable, $dg{c}_{it}$, is substituted by $I_{it}$.

3.2.3. Mediating Variable: Innovation Index ($inn$)

The mediating variable in this paper is the regional innovation index, which is jointly compiled by the National Development Institute of Peking University and the Longxin Data Research Institute. The index integrates the micro characteristics of enterprise setup, enterprise investment, and enterprise innovation output to evaluate the entrepreneurship level of a region comprehensively. Specifically, it includes seven sub-dimensional indicators, such as the number of new enterprises, the number of patents granted, and the number of trademarks registered. The results calculated by using big data analysis methods can reflect the innovation level of a region well.

3.2.4. Control Variables

This paper draws on the studies of [47] as well as [48] to set control variables that may have an impact on the upgrading of industrial structures: (1) economic development ($pgdp$), expressed as GDP per capita; (2) foreign direct investment ($fdi$), expressed as actual foreign investment; (3) labor force level ($emp$), expressed as the number of people employed in urban units at the end of the year; (4) social consumption ($soc$), expressed as the ratio of social retail consumption to regional GDP; (5) infrastructure ($road$), expressed as the ratio of actual urban road area to built-up area at the end of the year; (6) human capital ($human$), expressed as the ratio of the number of general undergraduate students to the total population at the end of the year; and (7) government intervention ($gov$), expressed as the ratio of local government fiscal expenditure to regional GDP. To reduce heteroskedasticity, the above control variables were taken as natural logarithms.

3.3. Data Sources and Description

The empirical sample of this paper was prefecture-level cities in China; due to missing data (Due to data availability, some cities in Tibet and Xinjiang are not included), the total sample finally selected was 2168, and the study period was 2011–2018. In order to eliminate the effect of sample outliers, a 1% tailing process was applied to all of the continuous variables in the model. The sources of variables involved in the empirical analysis were: in the digital economy index, except for the data of digital inclusive finance from the Digital Finance Research Center of Peking University and the Ant Group, the other four indicators are from the China City Statistical Yearbook; The regional innovation index comes from the National Development Research Institute of Peking University and Longxin Data Research Institute; The industrial structure indicators and other control variables are from the China City Statistical Yearbook and the statistical yearbooks of each prefecture-level city. Some missing data were filled in by consulting the statistical yearbooks of each province and by extrapolation based on the interpolation method and relevant data. The descriptive statistics of the main variables in this paper are shown in

Table 2. Descriptive statistics of the variables.

Variable	Obs	Mean	SD	Min	Max
lnais1	2168	−1.047	0.854	−2.88	1.238
lnais2	2168	−3.919	1.328	−7.544	−0.964
lndgc	2168	−2.306	0.442	−3.447	−1.153
lninn	2168	−2.055	1.51	−5.298	1.603
lnpgdp	2168	1.458	0.547	0.252	2.818
lnfdi	2168	−4.694	1.355	−9.605	−2.658
lnemp	2168	−3.269	0.755	−4.702	−1.191
lnsoc	2168	−1.004	0.3	−2.013	−0.37
lnroad	2168	−6.496	1.863	−10.45	−1.628
lnhuman	2168	−4.279	0.942	−6.946	−2.131
lngov	2168	−1.746	0.394	−2.573	−0.747

.

4. Results

4.1. Analysis of the Baseline Regression Results

The Hausman test was used to choose between a fixed effects model or the random effects model, and the result was that a fixed effects model was chosen, and the specific regression results are shown in

Table 3. Effects of the digital economy on the upgrading of industrial structures.

Variable	(1)	(2)	(3)	(4)
	Lnais1	Lnais1	Lnais2	Lnais2
lndgc	0.486 ***	0.147 ***	0.926 ***	0.448 ***
	(29.610)	(8.830)	(15.590)	(6.900)
lnpgdp		0.747 ***		1.448 ***
		(21.040)		(11.690)
lnfdi		(0.003)		(0.012)
		(−1.217)		(−0.913)
lnemp		0.058 **		−0.823 ***
		(2.331)		(−7.046)
lnsoc		(0.029)		−0.219 **
		(−1.066)		(−2.319)
lnroad		0.025 ***		0.055
		(2.790)		(1.080)
lnhuman		−0.071 ***		(0.074)
		(−3.322)		(−1.311)
lngov		−0.089 **		(0.054)
		(−2.543)		(−0.437)
Constant	0.074 *	−1.947 ***	−1.783 ***	−8.022 ***
	(1.950)	(−14.864)	(−13.008)	(−12.108)
Obs	2168	2168	2168	2168
N	271	271	271	271
City fixed effect	YES	YES	YES	YES
r2_a	0.597	0.857	0.306	0.445
F	876.5	989.7	243	58.29

Note: (1) *, **, and *** denote significance at 10%, 5%, and 1% significance levels, respectively; (2) t-values are in parentheses; and (3) all of the regressions use clustering robust standard errors with prefecture-level cities as clustering variables.

. Table 3 gives the regression results with or without the inclusion of control variables, respectively, and it can be concluded that the regression coefficient of the digital economy is still significant before and after the inclusion of control variables, which is no reason to reject Hypothesis 1.

Columns (2) and (4) of Table 3 present the benchmark results of digital economy development on the advancement of industrial structures. The coefficients of the effects of lndgc on the quantity and quality of the advancement of industrial structures are 0.147 and 0.448, respectively, which pass the significance test. This indicates that digital economy development is beneficial for promoting the quantity and quality of the advancement of industrial structures, but it promotes the quality of the advancement of industrial structures more. It indicates that, from the proportional share of industrial structures, the share of service-oriented tertiary industry gradually increases, and the development of the digital economy brings industrial innovation and technological progress, which promotes the advancement of industrial structures. This is mainly due to the fact that, with the development of the digital economy, it has accelerated the updating of big data, cloud computing and other technologies, integrated into various economic fields, reorganized factor resources, extended the industrial chain, and promoted industrial digitization and digital industrialization. In addition, as a new production factor, data can improve industrial labor productivity, promote enterprise competition, and then promote industrial structure adjustment.

The regression results of the control variables show that the effects of the economic development level, the labor force level, and social consumption on industrial structure advancement pass the significance test. The economic development level has a significant contribution to the advancement of industrial structures at both quantitative and qualitative levels, but the effect of foreign investment on the advancement of industrial structures is not significant, while human capital and government regulation are not conducive to the advancement of industrial structures.

4.2. Robustness Test

It is usually believed that there is an endogenous relationship between the level of digital economy development and the upgrading of industrial structures. The digital economy affects the upgrading of industrial structures, and, in turn, the upgrading of industrial structures promotes the development of the digital economy through structural effects. There may be a reverse causality relationship between digital economy development and the upgrading of industrial structures. Considering the possible endogeneity problem in the model, this paper conducts robustness tests from two aspects.

4.2.1. Instrumental Variable Method

Referring to the literature of [29,49,50], the number of landline telephones per 100 people in 1984 in each city is multiplied by the number of regional internet users in the previous year to constitute panel data as an instrumental variable for the core explanatory variables, which is consistent with the study of this paper. There are reasons for this: (1) The number and layout of landline telephones affects the internet usage of local residents, thus affecting the development level of the local internet and satisfying the correlation requirement. (2) Relative to the rapid development of the internet, the number of fixed phones historically has hardly had an impact on the upgrading of the industrial structures in a region, satisfying the requirement of exclusivity.

Using 2SLS for the empirical analysis, the regression results are presented in

Table 4. Regression results for the instrumental variables.

Variables	(1) Lndgc	(2) Lnais1	(3) Lnais2
lndgc		0.473 ***	0.745 ***
		−9.24	−3.48
IV	0.187 ***
	−11.42
CV	YES	YES	YES
City fixed effect	YES	YES	YES
Obs	1680	1680	1680
Kleibergen-Paap rk LM		72.894 ***	59.714 ***
Kleibergen-Paap rk Wald F		83.937 ***	41.584 ***

Note: (1) *** denote significance at 1% significance levels; (2) t-values are in parentheses; and (3) all of the regressions use clustering robust standard errors with prefecture-level cities as clustering variables.

. Column (1) indicates the results of the first-stage regression, where the regression coefficient of the instrumental variable was 0.187, which is significantly positive at the 1% level. Meanwhile, the F-statistics obtained from the C–D Wald test, the K–P rank LM test, and the K–P rank Wald test are much greater than 10, excluding the possibility of the existence of weak instrumental variables. Columns (2) and (3) indicate the results of the second-stage regression, and the results of the regression of the instrumental variables on the quantity of the advancement of industrial structures over the quality are 0.473 and 0.745, respectively; furthermore, both are significant at the 1% level, indicating that the digital economy can indeed promote the upgrading of industrial structures, and the qualitative driving effect on the advancement of industrial structures is greater.

4.2.2. Replacing the Digital Economy Index

Referring to the literature of [51], the lag period of the development level of the digital economy is used as the core explanatory variable and regressed on the basis that there is no reciprocal causality between the current period of the upgrading of industrial structures and the lag period of the development level of the digital economy. As can be seen from Table 4, the impact of the development level of the digital economy lagging behind the first stage on the upgrading of the industrial structure has not changed significantly compared with the previous regression results. Therefore, the core findings of this paper are still verified when the endogeneity issue is considered.

4.2.3. Changing the Measurement Method of the Digital Economy

In order to avoid the different calculation methods of digital economy development indicators affecting the empirical results, this paper also changed the calculation methods of digital economy indicators to test the robustness. Specifically, this paper used the entropy method to evaluate the development level of the digital economy. Refer to [29] to measure digital economic indicators using the principal component analysis. The specific regression results are shown in

Table 5. Variable substitution regression results.

Variables	(1)	(2)	(3)	(4)
	Lnais1	Lnais2	Lnais1	Lnais2
L.lndgc	0.138 ***	0.259 ***
	(7.92)	(3.82)
lndgc			0.044 ***	0.077 **
			(5.85)	(1.99)
CV	YES	YES	YES	YES
Obs	1897	1897	2168	2168
N	271	271	271	271
City fixed effect	YES	YES	YES	YES
r2_a	0.8	0.36	0.823	0.346
F	643.4	43.73	1295	177.9

Note: (1) ** and *** denote significance at 5% and 1% significance levels, respectively; (2) t-values are in parentheses; and (3) all of the regressions use clustering robust standard errors with prefecture-level cities as clustering variables.

. The results calculated on this basis are unchanged from the previous regression results, and the results are robust.

4.3. Mechanism Analysis

Further analysis of the mediating effect of the regional innovation index is presented in

Table 6. Analysis of the mediating effects of the regional innovation indices.

Variable	(1)	(2)	(3)	(4)
	Lninn	Lnais1	Lninn	Lnais2
lndgc	0.466 ***	0.120 ***	0.466 ***	0.402 ***
	(11.601)	(11.034)	(11.601)	(6.912)
lninn		0.059 ***		0.097 ***
		(9.777)		(3.023)
CV	YES	YES	YES	YES
Constant	(0.382)	−1.925 ***	(0.382)	−7.985 ***
	(−1.113)	(−21.475)	(−1.113)	(−16.644)
Obs	2168	2168	2168	2168
R-squared	0.661	0.864	0.661	0.45

Note: (1) *** denote significance at 1% significance levels; (2) t-values are in parentheses; and (3) all of the regressions use clustering robust standard errors with prefecture-level cities as clustering variables.

. From columns (2) and (4) of Table 6, the coefficients of the effect of the digital economy on the quantitative and qualitative aspects of the advancement of industrial structures were 0.147 and 0.448, respectively, both of which pass the 1% significance level test. In Table 6, Models (1) and (3) further verify whether the digital economy contributes to the improvement of the regional innovation index in two dimensions, showing positive results and significance at the 1% level. The mediating variable of the innovation index was then put into the equation for the regression analysis, and observing the coefficient values and significance levels of the digital economy in Models (2) and (4), it can be seen that the coefficient of the impact of the digital economy on the upgrading of industrial structures has decreased compared to Model (1); this indicates that the level of innovation is an important channel through which the digital economy promotes the upgrading of industrial structures. Consequently, there is no reason to reject Hypothesis 2. From Table 6, it can be seen that for every 1 unit increase in the digital economy, the innovation level can increase by 0.466 units, and for every 1 unit increase in innovation performance, the amount of advancement of industrial structures increases by 0.059 units, which means that for every 1 unit increase in the digital economy, the amount of advanced industrial structures can be effectively promoted by 0.027 units through the innovation level; similarly, it can be concluded that for every 1 unit increase in the digital economy, the advancement of industrial structures can be effectively promoted through the innovation level. Furthermore, for every 1 unit increase in the digital economy, the quality of the advancement of industrial structures increases by 0.045 units through the level of innovation. This is mainly due to the fact that, in the digital economy era, the open source strategy has become a popular trend. With its shared development mode, it reduces the development cost of enterprises, enhances the resource complementarity among enterprises, improves the innovation efficiency and, finally, optimizes the transformation and upgrading of the industrial structure through horizontal and vertical technology diffusion.

In order to verify the robustness of the mechanism analysis, the Sobel test and the bootstrap self-help method were also conducted in this paper, and the results are shown in

Table 7. Sobel test and bootstrap method mediation test results.

IV		Lnais1	Lnais2
Sobel test	Sobel value	0.129 *** (Z = 11.01)	0.456 *** (Z = 10.72)
Bootstrap method	95% confidence interval	[0.007, 0.029]	[0.251, 0.649]
	Intermediary effect value	0.016 *** (2.963)	0.060 *** (2.83)

Note: *** denote significance at 1% significance levels.

. The results of both the Sobel test and the bootstrap self-help method are significant at the 1% level, so the level of innovation and entrepreneurship plays a mediating role in the impact of the digital economy on the upgrading of industrial structures.

4.4. Nonlinear Characteristic Analysis

4.4.1. Test of the Threshold Effect

In order to test whether there are nonlinear characteristics between the digital economy and the upgrading of industrial structures, this paper adopts the threshold effect for the empirical analysis. Considering that the impact of the digital economy on industrial structures will be influenced by the economic development level of the region, this paper adopts the lagged period of economic development level as the threshold variable. Before conducting the regression, referring to [45] study, the bootstrap self-help method was used to determine the threshold number of the model, and the tests of single threshold, double threshold, and triple threshold were conducted in turn, with 300 iterations of sampling; the regression results are shown in

Table 8. Threshold regression test results.

Dimensionality	Threshold Type	F-Value	p-Value	BS Times	Confidence Interval
					1%	5%	10%
lnais1	Single threshold	43.39	0.0033	300	24.9431	19.7202	16.6053
	Double threshold	25.85	0.0333	300	35.6355	21.3317	17.3129
	Triple threshold	29.79	0.4967	300	68.6341	62.0704	52.4589
lnais2	Single threshold	22.96	0.0633	300	28.8362	23.3716	20.0965
	Double threshold	20.14	0.0867	300	29.6009	21.1307	19.0015
	Triple threshold	17.08	0.5633	300	31.721	35.437	43.4539

.

Observing the p-value shows that there is a double threshold for the effect of the digital economy on the quantity of the advancement of industrial structures, and the double threshold passes the 1% test, while the triple threshold effect is not significant; there is also a double threshold for the effect of the digital economy on the quantity of advanced industrial structures. The double threshold passed the 10% test, and the triple threshold effect was also insignificant. It shows that the digital economy is not a simple linear relationship on the transformation and upgrading of industrial structures, and that there is a significant double threshold feature between the two. Based on this, the double threshold is set for testing, and the specific regression results are shown in

Table 9. Threshold model regression results.

Varname	Lnais1	Varname	Lnais2
Threshold variable	$lnpgd{p}_{t-1}$	Threshold variable	$lnpgd{p}_{t-1}$
$lnpgd{p}_{t-1}$$(lnpgd{p}_{t-1}\le 1.9001$)	0.190 ***	$lnpgd{p}_{t-1}$ $(lnpgd{p}_{t-1}\le 1.2986$)	0.254 ***
	(13.376)		(3.543)
$lnpgd{p}_{t-1}$$(1.9001<lnpgd{p}_{t-1}\le 2.2562$)	0.163 ***	$lnpgd{p}_{t-1}$$(1.2986<lnpgd{p}_{t-1}\le 2.2562$)	0.199 ***
	(10.656)		(2.667)
$lnpgd{p}_{t-1}$$(lnpgd{p}_{t-1}\ge 2.2562$)	0.116 ***	$lnpgd{p}_{t-1}$$(lnpgd{p}_{t-1}\ge 2.2562$)	0.143 *
	(6.637)		(1.813)
CV	YES	CV	YES
R-squared	0.779	R-squared	0.259
Obs	1897	Obs	1897

Note: (1) * and *** denote significance at 10% and 1% significance levels, respectively.

.

4.4.2. Threshold Regression Results

From Table 9, the effects of the digital economy on the quantity and quality of the advancement of industrial structures are different under different levels of economic development. Under the double threshold, the development level of the digital economy was positive at a 1% or 10% significance level in all intervals, but the effect of the digital economy on the quantity and quality of the advancement of industrial structures showed a nonlinear change from strong to weak. The effect of the digital economy on the advancement of industrial structures was weakened when the level of economic development exceeds a certain degree. In the quantitative model of the advancement of industrial structures, when the level of economic development was less than 1.9001, the regression coefficient was 0.19; when it was in the interval of [1.9001, 2.2562], the coefficient was 0.163; and when the level of economic development was greater than 2.2562, the coefficient was 0.116. The positive promotion result was still significantly positive, and the promotion effect decreased. From the qualitative model of the advancement of industrial structures, the regression coefficient was 0.254 when the level of economic development was lower than 1.2986; when in the interval of [1.2986, 2.2562], the positive contribution decreased from 0.254 to 0.199; and when the level of economic development was greater than 2.2562, the promotion effect decreased from 0.163 to 0.143. The main reasons for the weakening of the effect of the digital economy are the following: (1) When the level of economic development is at a low level, the digital economy can optimize resource allocation and improve labor productivity to a certain extent, but as the level of economic development increases and industrial development is relatively perfect, the dividend released by the digital economy decreases. (2) When the level of economic development is low, the ratio between primary, secondary, and tertiary industries is not reasonable, and the digital economy can promote the optimization of industrial structures; when the economic development level increases, the industries tend to be “service-oriented” and relatively perfect, and the impact of the digital economy on industrial structures is relatively weaker.

4.5. Heterogeneity Analysis

4.5.1. The Test of Regional Heterogeneity

Since the level of digital economy development in each city is different, the level of the upgrading of industrial structures in different regions will also be different. According to the traditional division method, this paper divided the 271 cities into three subsamples, which are the eastern, central, and western regions, and the specific regression results are shown in

Table 10. Analysis of regional heterogeneity.

	(1)	(2)	(3)	(4)	(5)	(6)
	East		Central		West
	Lnais1	Lnais2	Lnais1	Lnais2	Lnais1	Lnais2
lndgc	0.126 ***	0.136 ***	0.151 ***	0.404 ***	0.476 ***	0.503 ***
	(5.69)	(4.30)	(6.17)	(4.03)	(4.26)	(4.58)
CV	YES	YES	YES	YES	YES	YES
Obs	768.00	800.00	600.00	768.00	800.00	600.00
N	96.00	100.00	75.00	96.00	100.00	75.00
City fixed effect	YES	YES	YES	YES	YES	YES
r2_a	0.76	0.91	0.92	0.53	0.42	0.43
F	206.40	632.80	675.50	51.37	33.82	38.72

Note: *** denote significance at 1% significance levels.

. In Models (1)–(6), the quantitative and qualitative regression coefficients of the level of digital economy development on the advancement of industrial structures were significantly positive for the eastern, central, and western regions, but they varied with different regions. Specifically, the quantitative and qualitative effects of the digital economy on the advancement of industrial structures showed a trend of western > central > eastern. The reason for this is that the eastern region has a higher level of development and a more reasonable industrial structure, and its industrial level is more dominated by tertiary and high-tech industries, so the dividends of the impact of the digital economy on the upgrading of industrial structures may be released in advance, while the central and western regions are relatively backward in development, and their industries are mostly low-level industries, so the impact of digital economy development on the upgrading of industrial structures is still at the stage of increasing marginal utility. It also means that, if the digital economy is developed rationally, it will not only help to promote the advanced industrial structures, but may also effectively reduce the wealth gap between developed and less-developed regions.

4.5.2. Urban-Level Heterogeneity Analysis

In this paper, referring to the relevant literature of [52], cities are split into quartiles by their economic development levels and divided into low, medium–low, medium–high, and high levels, and the specific regression results are shown in

Table 11. Heterogeneity analysis of city classes.

Lnais1	(1)	(2)	(3)	(4)
	Low	Medium–Low	High	Medium–High
lndgc	0.199 ***	0.119 ***	0.052 ***	0.169 ***
	(9.77)	(4.53)	(3.42)	(2.80)
CV	YES	YES	YES	YES
Obs	542	542	542	542
N	121	170	149	110
City fixed effect	YES	YES	YES	YES
r2_a	0.90	0.83	0.87	0.55
F	195.20	297.00	508.70	81.94
lnais2	(5)	(6)	(7)	(8)
	Low	Medium–Low	High	Medium–High
lndgc	0.522 ***	0.15	0.454 ***	0.266 *
	(5.50)	(0.96)	(3.59)	(1.78)
CV	YES	YES	YES	YES
Obs	542	542	542	542
N	121	170	149	110
City fixed effect	YES	YES	YES	YES
r2_a	0.46	0.20	0.31	0.45
F	22.80	7.43	32.59	41.36

Note: (1) * and *** denote significance at 10% and 1% significance levels, respectively.

. It can be seen that the digital economy is still significantly positive on the upgrading of industrial structures. The reason for this is that cities with a high ranking have good resource allocation and perfect industrial structures, and their development is more focused on being high-quality, green, and service-oriented, while cities with a low ranking are more focused in the central and western regions, which have rich resources, low factor costs, and high market potential. With the advantages of abundant resources, low factor costs, and high market potential, the digital economy has a stronger driving effect on lower-ranked cities.

5. Discussion

The development of the digital economy has become an irreversible trend, and there are many related studies on the digital economy. The digital economy is a typical manifestation of technological progress, which has an important impact on the economic structure [43,53]. The development of digital technology can reduce the mismatch level of production factors, reduce transaction costs, and improve the efficiency of factor allocation; additionally, it can create new elements to replace traditional elements [54]. While the development of digital technology has an impact on the factor market, it will also have an impact on the upgrading of industrial structures, in addition to the upgrading and rationalization of industrial structures [55]. The development of digital technology has an impact on industrial structures through four channels, including changes in relative prices, total demand structures, input–output structures, and trade structures [53,56]. The digital economy developed very quickly in China. Some studies have evaluated the impact of the digital economy on industrial structures from the perspective of total factor productivity [21], and some have evaluated the impact of the digital economy on the industrial structure from the perspective of an input–output model [24]. Considering the regional differences and nonlinear characteristics of the impact of the digital economy on the upgrading of industrial structures, this paper used the data of prefecture-level cities in China to carry out empirical research, mainly to verify the positive role of the digital economy on promoting the upgrading of the industrial structure and the diminishing marginal effect. Compared with previous research [1,47,48], this paper has some innovations in research methods and perspectives. In terms of research methods, this paper uses the threshold effect model to illustrate the nonlinear characteristics of the impact of the digital economy on industrial structures. In terms of research perspectives, this paper first distinguished the quantity and quality of the upgrading of industrial structures, and then explored the intermediary mechanism of the impact of the digital economy on the upgrading of industrial structures; This paper also further analyzed the regional differences of the impact of the digital economy on the upgrading of industrial structures. This study may be helpful for local governments to formulate regional economic development policies. Local governments should increase investment in digital technology infrastructure construction, strengthen talent training in digital technology, and encourage enterprises’ technological innovation through financial subsidies and other measures.

6. Conclusions

Based on the panel data of 271 cities in China from 2011 to 2018, this paper analyzed the mechanism of the role and degree of the influence of the digital economy on the upgrading of industrial structures, and draws the following conclusions:

The digital economy has a significant positive contribution to both the quantity and the quality of the advancement of industrial structures, and the conclusion is reliable.

(1)

The digital economy can accelerate the upgrading of industrial structures by stimulating the level of regional innovation.

(2)

The digital economy has nonlinear characteristics in the transformation and upgrading of industrial structures.

(3)

From a regional perspective, the regional heterogeneity between the east and west is obvious, with the greatest impact being on the western region, followed by the central region, and the smallest impact being on the eastern region.

Limitations of the Work and Future Research Directions

The limitation of the paper is that it does not take into account the impact of the spatial spillover effect of digital technology on the economic development of adjacent areas. Further research directions: (1) The impact of digital technology spatial spillovers on the industrial structures of neighboring regions. Digital technology can be diffused through personnel flow, technical cooperation, and investment activities. At the same time, it will also change the industrial structures of neighboring areas. This problem is worthy of in-depth research. (2) The impact of digital economy development gaps on regional income gaps. This paper verifies that the difference in digital economy development in different regions leads to different industrial structures. Will this impact lead to the expansion of regional income gaps? This issue is also worthy of in-depth research. (3) Will the development of the digital economy have an impact on the environment, and will the green total factor productivity be improved through the spatial spillover effect? This issue is also worth exploring in detail.