1. Introduction
With the continuous development of China’s economy, the scale of the environmental costs of the economic growth is increasing day by day. As the world’s second largest economy and the world’s largest real economy, how to control the environmental impact of development while ensuring economic development has become a problem that China must deal with at the present stage. In September 2020, Chinese President Xi Jinping, in his seventy-fifth United Nations general assembly speech, announced that China will strive for a carbon emissions peak by 2030 and become carbon neutral by 2060. This means that China will take an overall low-carbon development as an important goal of its national development strategy in the medium- and long-term development plan.
To achieve this goal, China has made corresponding planning arrangements for energy conservation and carbon reduction at different levels, including in industrial development and social management. The goal of sustainable development requires development that meets the needs of the present without jeopardizing the ability of future generations to meet their needs. It can be seen that the key to ensuring sustainable development under the carbon peaking and carbon neutrality goals lies in how to effectively control the regional carbon emissions, and that the control of carbon emissions cannot be separated from the background of regional development. At present, China is in the stage of industrial development and transformation, and the intelligent manufacturing industry has become the key to China’s industrial transformation, and while in this process, the promotion of production automation is particularly obvious. From the aspect of the labor force, the application of industrial robots has greatly changed China’s manufacturing production mode, and the industrial automation represented by them will become a medium and long-term trend of China’s industrial development. Analyzing the impact of this trend on carbon emissions will help in judging the development prospects of China’s carbon emissions peak and carbon neutrality goals.
Therefore, the research focus of this paper is on the role of automation industry development while in the process of a carbon emissions reduction in China. This study can help us to have a better understanding of the role of production automation in the process of carbon reduction, and to understand the path of carbon emissions reduction. Based on this analysis, we can provide a basis for the prediction of the prospect of the carbon neutrality goals, and it can also provide a new perspective on the construction of low-carbon development industry prospects for regions with similar goals and conditions. The paper’s structure arrangement is as follows:
Section 2 reviews the relevant research on carbon emissions and production automation, and explains the basic viewpoints of this research;
Section 3 analyzes the relationship between automation and a carbon reduction, and gives the analytical framework of this paper;
Section 4 establishes an empirical model, and analyzes the relationship between production automation and carbon emissions;
Section 5 discusses some conclusions of this paper and the relevant research results;
Section 6 states the main conclusions of this paper and gives some policy recommendations.
2. Literature Review
For the analysis of carbon emissions, the most representative research is carried out around the carbon Kuznets curve (CKC). The theory of the CKC originates from the hypothesis of the environmental Kuznets curve, and its basic point of view is that at the low income level, the regional environmental impact will increase with an increase in per capita output value, while at the high income level, it will decrease with an increase in per capita output value [
1]. This hypothesis has been confirmed in many studies [
2]. From the empirical analysis of different regions, the CKC hypothesis is generally applicable to explain the carbon emissions process in different regions [
3,
4], which shows that this relationship also widely exists in the relationship between various human activities and carbon emissions [
5]. What is certain is that the existence of the CKC as a phenomenon has been recognized by most studies. The current research results mainly include three types of explanations for the causes of inflection points in the process of a carbon emissions change: institutional adjustment in the process of development [
6], technological change that drives development [
7], and the impact of related events [
8]. In addition, there are explanations for the phased changes in the environmental impact on economic development from the perspective of civic awareness and other cultural aspects [
9]. However, some scholars question the accuracy of the carbon Kuznets curve from the perspective of model characteristics due to an ambiguity of the detail description [
10,
11]. In order to supplement and improve the theory of the CKC, Some scholars supplement the theory of the carbon Kuznets curve by exploring the intermediary channels between economic growth and carbon emissions [
12]. In general, to explain the relationship between socioeconomic development and carbon emissions, the carbon Kuznets curve is still applicable to explain the process of carbon emissions in most regions. At the same time, the extension model based on this theory has a good extensibility. However, as pointed out by Webber and Allen, the phenomenon of the environmental Kuznets curve generally exists in the process of changes of various environmental indicators, but the inflection points of changes caused by human activities on the environment are different in different cases [
13]. Therefore, according to the development characteristics of different regions, the influencing factors of the change process of carbon emissions are analyzed from the key development characteristics. Such analyses are important for projecting regional carbon emission prospects.
For different regions, researchers have conducted a large number of analyses on their carbon emissions according to the important factors of regional development, and these analyses all show that the key development trends of the region have an important impact on the regional carbon emissions. For example, in the cases of developed countries or regions, by analyzing the impact of various socio-economic factors on the carbon emissions in Central and Eastern Europe, Atici believes that the per capita energy consumption in central and Eastern Europe is the main cause of local carbon emissions [
14]. Vieira et al. analyzed the carbon emissions of European manufacturing and energy industries, and they pointed out that the main difficulty for EU countries to achieve the goal of net zero emissions lies in the emissions control of large emitters [
15]. Miura et al. analyzed the carbon emissions contributions of the economic sectors in different regions of Japan and pointed out that there are regional differences in the carbon emissions contributions of different economic sectors, depicting the impact of carbon emissions in the modernization of Japanese industries [
16]. These studies all show that the dominant factors of carbon emissions are different within the cases of the same developed countries. In contrast, the economic structure of developing countries is more diverse, and the factors for carbon emissions are more complex. For example, India’s international tourism industry has grown rapidly in recent years, and Jayasinghe and Selvanathan’s study points out that international tourist spending is also an important factor in India’s carbon emissions [
17]. As a manufacturing power, China’s carbon emissions have attracted extensive attention from scholars around the world. Jalil and Mahmud analyzed China’s carbon emissions and pointed out that the per capita income and energy consumption are the long-term determinants of China’s carbon emissions [
18]. Even within China, there are still differences in the carbon emissions of different regions. The study by Lu et al. points out that most of China’s eight economic zones have already crossed the turning point of the CKC and have entered the ranks of low-carbon development; however, the northwest region still needs to be improved, and the reason for this regional difference is China’s differentiated regional development strategy [
19]. Therefore, the analysis of the impact of China’s key development trends on carbon emissions is of great significance for the analysis of China’s carbon emission prospects.
At present, the world is in the stage of a rapid development of emerging industrial technology, and at the production level, production automation is undoubtedly one of the most eye-catching trends [
20]. Compared with the traditional production mode, the most significant advantage of production automation lies in its improvement of production efficiency [
21,
22], while at the same time, this process will bring about a reduction in the total share of the labor-added value [
23]. In the early stages of production automation, its application has mainly been in the automobile manufacturing industry, machinery manufacturing industry and other industries that are easier to promote already using Fordism [
24]. Obviously, there is a technical basis for promoting production automation in a wider range of fields, such as the planting industry [
25], breeding industry [
26], and food production industry [
27]. The increase in global risks represented by the global COVID-19 epidemic in recent years has become an opportunity for the further promotion of production automation technology [
28,
29]. These changes in production technology will undoubtedly have an impact on a carbon emissions reduction. Shin et al.‘s study on the relationship between production automation and economic growth also shows that production automation contributes to the sustainable development of society [
30]. Moreover, Van and Morlet’s study also points out that the traditional environmental costs are gradually changing with the development of production automation [
31]. As the world’s largest manufacturing economy, production automation is an important trend in China’s manufacturing transformation; therefore, analyzing the impact of the development of production automation on carbon emissions is of great significance for judging the prospect of China’s dual carbon strategy.
In order to analyze the details of the environmental external effects generated by a development trend, an important perspective is to start from the representative industry of this transformation. Focusing on the impact of key polluting industries [
32] and the change of carbon emissions in industries with high energy consumption [
33], or the change of carbon emissions in the process of service industry development [
34,
35], can reflect the impact of the carbon emissions caused by an industrial structural transformation. For example, the research on the carbon emissions of the construction industry can reflect the details of the changes of carbon emissions in the process of urbanization [
36,
37]. Meanwhile, an analysis of the carbon emissions changes in the transportation industry can be used to describe the changes in carbon emissions during the development of the trade circulation network [
38,
39]. In general, to analyze the environmental externality of a development trend, it is necessary to observe the environmental impact of the industry that can represent the trend.
Considering the research results of these literatures, we propose three basic understandings as follows: firstly, although there are differences in the understanding of the causes and development process, most studies agree that the CKC exists as a phenomenon; secondly, on the basis of recognizing the CKC as a universal law, most studies agree that the main development trend of a region greatly affects the local carbon emissions process; thirdly, as an important form of the modern manufacturing revolution, production automation is an important trend that will affect the development of China’s future productivity. To sum up, the contribution of this paper is to add variables about production automation to the traditional CKC model, to analyze the impact of production automation on carbon emissions changes, and to provide a new viewpoint for the expectation of China’s carbon emission reduction prospects.
3. Methodology and Data
To analyze the role of production automation in the process of carbon emissions reduction, it is necessary to first clarify the relationship between the two. By reviewing the existing studies, we can confirm that there is a functional relationship between the per capita income and the carbon emissions in line with the CKC hypothesis [
40], and the per capita income has a strong impact on the carbon emissions [
41]. Previous studies have usually associated it with changes in the consumption levels and consumption preferences, and have believed that changes in the per capita income achieve a carbon emissions reduction through changes in consumer behavior [
42]. However, when the research perspective is narrowed to the level of individual behavior, it may be difficult to explain the mechanism of the CKC [
43]. Therefore, from the perspective of income changes for future industrial development trends, this paper attempts to put forward an analytical framework for the relationship between the per capita income and the total carbon emissions mediated by production automation, so as to discuss how an increase in the per capita income affects the total carbon emissions. The analytical framework is shown in
Figure 1.
In this analysis framework, the effect of the per capita income on the carbon emissions is divided into two paths: (1) an increase in the per capita income leads to an increase in the total carbon emissions. The logic of this path is relatively simple, that is, an increase in the per capita income will lead to an increase in personal consumption, which in turn will affect the production scale of consumer goods and, thus, lead to the growth of carbon emissions; (2) the increase in per capita income will push up the labor cost in the production of products, and the increasing consumer demand will put forward new demands for improving production efficiency, thus, promoting the development of production automation. This will promote the application of more carbon reduction technologies in the production process, thus. helping to achieve a carbon emissions reduction. There are many related studies on the former path [
44]; therefore, that path is not the focus of this paper. The second path is the key of this paper because it is able to explain the role played by production automation in the carbon reduction process, and can explain the cause of formation of the CKC from the perspective of production mode reform.
Therefore, the problem to be demonstrated in this paper consists of two parts: firstly, whether an increase in income levels can promote production automation; secondly, with production automation as the intermediary, how does the change of per capita income affect the total amount of carbon emissions? Since production automation contains more specific market behaviors, this paper will also observe the effects of different market developments on carbon emissions from different market behaviors.
3.1. How Does the Demand for Labor Substitution Due to Rising Incomes Affect the Development of Production Automation
In the analytical framework proposed in this paper, the first relationship to be tested is whether an increase in the per capita income will lead to the development of production automation. In this analysis framework, the impact of the per capita income on the production automation industry has two paths: (1) an increase in the per capita income will become the labor cost in the production process, which will stimulate the demand of manufacturing enterprises for labor substitution in the production process, and then promote the development of the production automation industry [
45]; (2) an increase in the per capita income can stimulate the consumption willingness of residents, resulting in an increase in market consumption demand, and then an expansion in the production scale of consumer goods [
23]. These theories are based on studies in regions with more developed productivity. When we look at the longer term development, this unilateral promotion may have its limits. Consequently, we attempt to take a longer view of the impact of increasing income on the development of production automation.
Production automation is essentially a reform of the production mode of the manufacturing industry. In this trend, the influence of wage income (the price of labor) on the process of production automation is different at different stages. The basis of this judgment is as follows: firstly, the scale of production is proportional to the wage income. The larger the scale of production, the higher the wage income [
46]; secondly, according to the theory of a marginal effect of scale, there is an intersection between the marginal revenue and wage income. On the basis of these two conclusions, we can see that with an expansion of the scale of production (with the rise of the price of labor), the process of the change can be depicted as in
Figure 2. In the diagram we can see two important points where the first is the apex of the marginal benefits of scale (
X1), and the second is the intersection of the labor costs and marginal benefits (
X2). These points, especially
X2, can be helpful to understand the relationship between the labor price and production automation. The impact of labor price on production automation is manifested in two directions, which can be regarded as the characteristics of two development stages. In the early stage of large-scale development, the price of labor is relatively low, and the boundary effect generated by an expansion of the production scale makes it profitable for enterprises to expand production by increasing the labor;, therefore, there will not be a strong demand for the realization of production automation at this stage (in the left range of
X2). With an increase in the labor price, the scale effect of expanding production will be gradually exceeded by the increase of the labor cost. In contrast, production automation becomes one of the options for cost control; therefore, there will be an increasing demand for production automation (in the right range of
X2).
To sum up the above analysis, we can make a basic inference: at the stage of a low labor price, with an increase in the labor price, the demand for production automation will decrease; at the stage of a high labor price, the demand for production automation will increase with an increase in the labor price. Therefore, in terms of function form, there should be a U-shaped curve between the labor price and the development of production automation. In order to verify this relationship, this study uses a static panel model for preliminary test:
where,
PA represents the development of production automation, and the explanatory variables are
GDP,
EC and
AveWage, which, respectively, represent the regional gross product, the total regional energy consumption and the average salary of employees.
X is the control variable, including the population urbanization rate, total bank loans, manufacturing output value and the population’s education level. By comparing the results of the two equations, we can test the relationship between the number of market players in the industrial robot industry and the average wage level. If both
β3 and
β4 of Equation (2) are significant, and the goodness of fit of Equation (2) is better than that of Equation (1), then the relationship between the labor price and production automation can be considered as a U-shaped curve. The turning point of the curve can be regarded as the position of
X2 in
Figure 2. When the per capita income moves to the right range of the turning point, the increase in per capita income will promote the progress of production automation. In other words, the need for automation due to an increase in the per capita income does not occur until the turning point is crossed.
Based on this model, we can further observe the details of the impact of economic development on production automation. Because of the highly detailed division of labor in the production automation industry, enterprises engaged in related work may be affected differently by changes in the per capita income according to their own work in the industrial chain. From the perspective of the process of industrial demand, the market activities related to production automation can be divided into the research on production automation, the manufacturing of equipment, and all kinds of related services, such as promotion, marketing, after-sales, maintenance, etc. These demands should occur in a sequential order: before the actual application, there will be a demand for research and development of related technology, and only after that, will there be a demand for manufacturing and other services. In the early stages of development, when the technology is not yet perfect and the market demand is still relatively small, the manufacturing enterprises of relevant equipment both need and have the ability to provide the services, such as the marketing, maintenance, and installation, but when the wage further increases, the market demand expands, and it will then produce the demand of a specialized service. This inference can be expressed in the form shown in
Figure 3.
Continuing the prediction, it can be believed that with an increase in average wages, the demand for production automation will continually increase, and the process of production automation will persistently accelerate. This prediction can be verified by the inflection point of Equation (4). The demand for production automation can be divided into research and development, manufacturing and service, which are respectively, substituted into Equation (4) for testing, when the difference in their turning points can be analyzed. Thus, the demand generated in the different stages can then be verified.
3.2. The Mediating Role of Production Automation in Carbon Reduction
In the analytical framework of this paper, another relationship that needs to be verified is the mediating role of production automation in a carbon emissions reduction. Of the reasons for believing that production automation contributes to carbon emissions reductions, the direct one is that the essence of production automation is to further standardize the production process, and to cut more randomness from the subjective judgment of working human labor. Production automation has the characteristic of being modular, so that technical updates can be integrated into the production module, and are easier to be applied to the production process. Another context is that China’s energy structure has changed significantly over the past twenty years. As a production mode that relies on electricity as the main energy, automated production means an increase in the electricity energy consumption. From 2000 to 2018, China’s energy structure changed significantly. Among them, the share of coal energy decreased from 68.5% to 59.0%, and that of petroleum energy decreased from 22.0% to 18.9%, which are major sources of carbon emissions. In contrast, in terms of clean energy, the share of natural gas increased from 2.2% to 8.4%, and that of other clean electricity increased from 7.3% to 14.5%. In the context of this changing energy structure, we can be sure that production automation means that more consumer goods are produced by using cleaner energy.
By this analysis, we made a basic inference that production automation plays a partial mediating role in the process of carbon emissions reduction, that is, production automation is a means to assist the realization of a carbon emissions reduction. Before building the model, we first needed to select the variables to be modeled. First of all, the core issue to be explained in this study was the role of production automation in the CKC phenomenon; therefore, it contained the variable, PA, that represented the production automation. It also needed to include the AveWage, a core explanatory variable representing labor prices [
47]. In addition, the impact of the production scale on carbon emissions has been confirmed in a large number of empirical studies [
48], while the energy consumption is a more direct factor affecting carbon emissions [
49]; therefore, both were included as explanatory variables. The above four variables were the explanatory variables of this study, while in order to eliminate the influence from other factors (which are some common influencing factors in CKC-related research), four control variables were selected [
2,
50,
51,
52]. To verify this inference, refer to the research on the mediating role [
53]. Here, we designed Equations (3) and (4) to verify the partial mediating role:
where,
CE is the regional carbon emissions, and the explanatory variables are
GDP,
EC,
AveWage, and
PA, which, respectively, represent the regional gross product, energy consumption, average wage, and production automation.
X is the control variable, including the population urbanization rate, total bank loans, manufacturing output value and the population’s education level. By comparing the difference between the correlation coefficient in Equations (3) and (4), we can verify the role of production automation in the process of carbon emissions reduction. According to our analysis, if
β3 and
β4 are significant in Equation (3), this indicates that the hypothesized phenomenon of the CKC exists, there will be a turning point in China’s carbon emissions, when the average wage will be higher than the turning point, and that the carbon emissions will reach a peak and start to decline gradually as the average wage increases. In Equation (4), if
β3 and
β4 are still significant and
β5 is also significant, this indicates that production automation is one of the pathways in the CKC hypothesis processes to reduce the carbon emissions, and that it has a partial mediating role.
3.3. Data
According to the validation model built by the above analysis, we selected 9 variables for analysis. The research time range was from 2000 to 2019, and the research sample was 289 cities with districts in China. The description of each variable is shown in
Table 1.
- (1)
Regional carbon emissions. This paper chose the Open-Data Inventory for Anthropogenic Carbon dioxide (ODIAC) Fossil Fuel Emission Dataset (ODIAC2020b) as an indicator of the regional carbon emissions, and this dataset draws a 1 × 1 km distribution image of the global fossil fuel carbon dioxide emissions through emissions modeling methods [
54]. The impact at this resolution can effectively reflect the carbon dioxide emissions at the spatial scale of prefecture-level cities. Using the carbon emissions raster data of the ODIAC superimposed on the administrative boundaries of prefecture-level cities, the carbon emissions of 289 sample cities from 2000 to 2019 could be extracted.
- (2)
Production automation. For the observation of production automation, this paper chose the perspective of enterprises engaged in production automation-related market activities, and took the number of relevant enterprises in the region as the index. The number of enterprises engaged in an industry can reflect the market demand for that industry; therefore, the number of relevant enterprises engaged in providing production automation services in a region can well reflect the development of production automation in that region. Through the online data platform of Qcc.com (
https://www.qcc.com/), the national industrial and commercial registration data was retrieved according to the standard that the business scope included the business content related to production automation. The industrial and commercial registration data was used as the basis for judging the existence of enterprises to determine the number of production automation enterprises in the different years and different regions. After filtering, there were 77,484 enterprises in 289 sample areas. Then, the number of enterprises engaged in production automation-related work in each region from 2000 to 2019 was obtained. According to the different types of enterprises engaged in production automation service work, the research enterprises, production enterprises and service enterprises were classified, denoted by PA
R, PA
M and PA
S, respectively..
- (3)
The explanatory variables. The core explanatory variables of this article was the urban worker’s average wage, and the reason the urban worker’s average wage was selected rather than the manufacturing worker’s average wage was that due to the lack of obvious industry barriers in the modern labor force, the flow of the labor force between industries is more and more frequent. Therefore, the average wage of urban workers, which can reflect the overall level of labor income, could better reflect the impact of wage changes on production automation. Other explanatory variables also included the regional gross product and energy consumption, reflecting the impact of the regional economic size and energy use, respectively. These data were acquired from the EPS data platform (
https://www.epsnet.com.cn).
- (4)
Control variables. The control variables selected in this paper included the population’s urbanization rate, financial industry activity, manufacturing output value and the population’s education level. The urbanization rate of the population was expressed by the population with an urban household registration divided by the total population of the region. The financial industry activity was represented by the total social loans. The education level of the population was represented by the proportion of the population with a college education or above in the total population in the sample population survey of the concept area.