The Impact of Industrial Robots on Green Total Factor Energy Efficiency: Empirical Evidence from Chinese Cities

Abstract

Improving energy utilization efficiency is a crucial means to achieve energy conservation, emission reduction, and green development. At present, to establish a high-quality development framework and satisfy the growing need for a better life among all its people, China must steadfastly pursue the path of green development. Although China’s substantial economic scale and achievements in ecological civilization construction provide favorable conditions for green transformation, there remains a significant gap compared to developed countries in the application of green and clean technologies. Confronted with technological bottlenecks, leveraging emerging technologies such as industrial robots from the new round of scientific and technological revolutions to improve the green total factor energy efficiency (GTFEE) is of critical importance to China’s green development. This study explores the potential impact of industrial robots on enhancing China’s GTFEE. It begins by reviewing the current research landscape in this field, highlighting its shortcomings, and theorizing potential impact pathways of industrial robots. Subsequently, the paper analyzes data from 2010 to 2019 on the usage of industrial robots and GTFEE across 276 cities at the prefectural level or above in China. Through empirical regression models that incorporate control variables and interaction terms, the study investigates the specific impacts of industrial robots on energy efficiency and their mechanisms of action. The results indicate that industrial robots significantly enhance the GTFEE of Chinese cities, especially in the Northeastern region. Industrial robots notably improve the GTFEE in resource-based cities, old industrial bases, and low-carbon pilot cities. Additionally, robots indirectly boost GTFEE by increasing labor productivity. Enhanced levels of green innovation and environmental regulations also positively moderate the effectiveness of industrial robots in improving energy efficiency. The findings of this research can assist local government agencies in coordinating and implementing policies that are conducive to green development, making better use of industrial robots to serve the people, and are of significant importance for promoting the transformation of China’s economy and society towards high-quality development.

1. Introduction

With the rapid advancement of artificial intelligence technology, traditional extensive economic models are facing unsustainable challenges, compelling China to pivot towards a path of high-quality development. This transformation involves not only technological innovation and industrial upgrading but also the creation of a sustainable future economic system. In the pursuit of Chinese modernization, the stability of economic growth should be balanced with rational energy use and reduced pollution emissions to foster improvements in the ecological environment. Hence, setting scientifically sound environmental emission standards is crucial for enhancing the country’s sustainable development capabilities and urban energy efficiency. According to research by Gokmenoglu et al. [1], the recent continuous growth in energy consumption in China mainly stems from the expansion of industrial production activities. Therefore, accelerating industrial upgrading and optimizing energy structures are central to deepening and implementing the concept of green development.

In this context, the advancement of smart manufacturing and Industry 4.0 strategies has made industrial robots a key component of automated and intelligent production. Data from the International Federation of Robotics (IFR) shows that the global stock of industrial robots has grown from 756,500 units in 2001 to 3.907 million units in 2022, with an average annual growth rate of 7.48%, highlighting their increasingly prominent role in modern manufacturing and the overall economy. China has actively integrated into this technological revolution, promoting the widespread use of industrial robots nationwide to help enterprises reduce production costs and enhance efficiency. In fact, China has become the largest robot market in Asia, with an installation of nearly 300,000 units in 2022, up 5% year-on-year, and for two consecutive years, the annual installation of industrial robots in China has surpassed the total of all other countries.

At the technological forefront, industrial robots, used as mechanical devices designed to automate tasks, are increasingly integral to modern industrial systems. Their widespread adoption has notably bolstered China’s capabilities in environmental protection and resource conservation, profoundly influencing both the nation’s industrial advancement and the daily lives of its citizens. This has injected fresh vigor into China’s initiatives for green development. However, this capital-intensive technological progress has escalated the demand for highly skilled labor, and, concurrently, the extensive deployment of industrial robots may introduce environmental challenges such as pollution and climate change, posing new hurdles for enhancing energy efficiency. A thorough review of the pertinent literature reveals significant research gaps: primarily, there is a notable dearth of in-depth analysis on the specific impacts of industrial robots on green development. Research into whether industrial robots confer an environmental benefit remains sparse, characterized by underdeveloped theoretical frameworks and embryonic stages of empirical research. Additionally, the interaction between industrial robots and China’s green development is poorly explored, with research perspectives remaining quite narrow. Moreover, empirical studies on green total factor energy efficiency in China are largely confined to the provincial level, whereas city-level research mainly concentrates on assessing the impact of various environmental and energy conservation policies, with detailed investigations into urban green total factor energy efficiency being particularly rare. Consequently, determining whether the application of industrial robots can rejuvenate China’s green development and potentially elevate China’s green strategies to a new phase warrants further profound investigation.

Thus, in the context of rapid advancements in Industry 4.0 and artificial intelligence technologies, examining the role of industrial robots in enhancing energy efficiency and scientifically formulating policies to reduce energy intensity is not only an essential requirement for achieving global economic and environmental sustainability goals but also a critical necessity to address the conflict between substantial industrial energy consumption and green sustainable development. This study focuses on the impact of industrial robots on green total factor energy efficiency, thereby pioneering a new field in robotics research that demonstrates how such technology can advance sustainable development goals. By analyzing the penetration of industrial robots and their specific effects on green energy efficiency from the perspective of prefecture-level cities, this paper not only reveals the intrinsic links between robotics technology and environmental efficiency but also examines how this technology elicits differentiated policy responses across various regions and types of cities. Furthermore, this research provides new evidence on how factors such as labor productivity, green innovation, and environmental regulation act as mediating and moderating variables in this relationship, thus offering theoretical and practical guidance for the formulation of more precise and effective environmental policies. The findings not only enhance our understanding of the role of robotics technology in boosting green productivity but also lay the groundwork for further exploration of its applications in other sustainable domains.

The remaining structure of this paper is outlined as follows: Section 2 reviews relevant literature concerning the national policy background of industrial robots and green total factor energy efficiency, clearly defines the research route and innovative aspects of the study. Section 3 explores the potential impact mechanisms of industrial robots on green total factor energy efficiency, proposing foundational research hypotheses. Section 4 constructs a baseline regression model, clarifying variable selection and measurement methods, and introduces data sources and processing. This Section previews the relationships between variables through descriptive statistics and scatter plots, laying the groundwork for subsequent empirical analysis. Section 5 examines the effects of industrial robots on green total factor energy efficiency through baseline regression analysis, endogeneity analysis, robustness tests, and heterogeneity analysis, validating the theoretical discussions and research hypotheses presented in Section 3. Section 6 analyzes how industrial robots influence green total factor energy efficiency through mediation effects and moderation effects involving labor productivity, green innovation, and environmental regulatory standards. Section 7 presents the main conclusions and policy recommendations of this paper, while also highlighting potential issues and shortcomings.

2. Literature Review

The measurement of energy efficiency can be categorized into single-factor energy efficiency, total factor energy efficiency (TFEE), and green total factor energy efficiency (GTFEE). Single-factor energy efficiency considers only one input factor, typically measured by energy productivity or energy intensity, as noted in sources [2,3]. However, as this approach neglects the input of other production factors such as labor and capital, it fails to comprehensively assess their impact on output, potentially leading to less accurate efficiency evaluations. Hu and Wang [4] incorporated multiple input factors including capital, labor, and energy into their model, defining the resulting measure as TFEE, which assesses the ratio of actual energy input to optimal energy input. A ratio close to one indicates that the actual energy input is nearly optimal.

Building on the foundation of TFEE, Yang [5] has further integrated resource and environmental constraints into the assessment framework for economic development quality, thus deriving the metric of GTFEE. In calculating GTFEE, most scholars predominantly employ the Data Envelopment Analysis (DEA) method, coupled with the Malmquist–Luenberger (ML) index, which decomposes the efficiency into components of technological progress and efficiency improvements. Park and Kim [6] used energy data from South Korea from 1993 to 2000 to measure energy efficiency through a hybrid model combining DEA and Artificial Neural Networks (ANNs). Huang and Wang [7] utilized a three-stage SBM model to measure TFEE across various regions in China, revealing generally low energy efficiency with a ”U-shaped” developmental trend and significant regional imbalances. Tao et al. [8], using historical data from China’s industrial sectors between 2004 and 2015, employed the SBM model and ML productivity index to measure green TFEE from an industry perspective.

Related closely to this paper is another area of research focusing on the penetration of industrial robots and their impact on the labor market. The International Federation of Robotics defines robots as multipurpose manipulative devices capable of automatic control, reprogramming, and performing multiple tasks. Current research on industrial robots primarily concentrates on how they influence the labor supply, enhance total factor productivity, drive economic growth, improve employment structure, and narrow gender wage gaps [9,10,11,12,13]. Acemoglu and Restrepo [9] found that the application of industrial robots reduces the labor force, specifically indicating that for every additional robot per thousand workers, unemployment could rise by 0.2% and wages might decrease by 0.42%. The application of industrial robots not only relieves traditional labor constraints on corporate production factors but also improves the neoclassical economic growth model through the application of intelligent technology, promoting balanced economic development. However, Ballestar et al. [14] found that the introduction of industrial robots did not negatively impact employment; between 2008 and 2015, small and medium-sized manufacturing enterprises in Spain actually created more job opportunities after adopting industrial robots.

As a result of technological innovation, robots not only demonstrate technological progress but also embody the characteristics of capital, showing clear technological bias with varying degrees of substitutability for different types of workers. The impact of industrial robot technology on the labor income ratio essentially reflects the elasticity of substitution between capital and total labor [15,16]. Specifically, for workers with vocational or undergraduate degrees, the substitutive effect of industrial robots is stronger; for those with graduate degrees or higher, the impact is nearly negligible; and for those with high school or lower educational levels, the application of robots significantly propels them forward. Wang and Dong [17] analyzed data from Chinese industrial enterprises to study the impact of industrial robots on China’s labor market, revealing significant effects on labor demand across different skill levels, characterized by “employment polarization”. This phenomenon is marked by increased demand for high-skilled and low-skilled workers, while the demand for medium-skilled workers decreases.

Moreover, the impact of industrial robots on employment largely depends on the quality of human capital. Song and Zuo [18] explored the relationship between changes in labor supply and the introduction of industrial robots, concluding that when the labor supply is insufficient, societal progress and the pursuit of higher living standards will drive industrial enterprises to continually foster technological innovation. Yang and Hou [13] incorporated robot pricing behavior and scale effects into the task-based model, theoretically dissecting how industrial robots can not only directly influence economic growth but also affect it through total factor productivity, further substantiating their positive impact on economic growth through empirical analysis. Xu et al. [19] examined the effects of industrial robot application on the gender wage gap, finding that the deployment of industrial robots significantly raised wage levels in China and notably narrowed the gender wage gap.

In this context, some scholars have also investigated the energy consumption of industrial robots. Brossog et al. [20] discovered that industrial robots can enhance the productivity of manufacturing enterprises, save production and operational costs, and ultimately reduce energy consumption, fostering green sustainable development. Li et al. [21] proposed that industrial robots help reduce greenhouse gas emissions by lowering energy consumption. Huang and Jiang [22] constructed a theoretical model including industrial robots and energy inputs, finding that industrial robots could facilitate urban industrial carbon emission reductions. Zhang and Yan [23] indicated that while the large-scale adoption of industrial robots can increase enterprise productivity, it might also elevate energy consumption, thereby exacerbating the environmental burden. Wang et al. [24] further explained from the perspectives of population mobility and energy rebound effects that the application of industrial robots has a certain inhibitory effect on regional carbon emissions. Therefore, the impact of industrial robots on energy efficiency remains inconclusive and warrants further analysis and verification.

The marginal contributions of this paper are as follows: First, from the perspective of the research angle, the existing literature primarily focuses on the impact of industrial robots on the labor market, with insufficient exploration of their role in environmental and energy aspects. This paper aims to investigate the influence of industrial robots on green total factor energy efficiency, providing a new perspective and direction for this emerging research field. Second, regarding the research subjects, this paper employs a more granular analytical approach by assessing green total factor energy efficiency and the penetration of industrial robots at the prefecture-city level. It further explores the causal relationships between these variables to discern variations in policy responses across diverse regions and urban classifications. Third, for the empirical analysis, this paper emphasizes the intrinsic relationship between robot application and green total factor energy efficiency and verifies these relationships through robustness tests and analysis of endogeneity issues. Furthermore, addressing the gaps in channel analysis and mechanism exploration found in existing research, this paper selects labor productivity as a mediating variable and uses the level of green innovation and environmental regulatory standards as moderating variables, thus more precisely analyzing the effects of mediating and moderating mechanisms.

3. Theoretical Analysis and Research Hypotheses

3.1. Industrial Robots and Green Total Factor Energy Efficiency

Green sustainable development has become a widely recognized critical issue, making the enhancement of urban green total factor energy efficiency particularly important. According to the methods for measuring green total factor energy efficiency, the primary ways to improve efficiency include reducing undesired outputs (such as decreasing environmental pollution) and increasing desired outputs (such as fostering economic growth). In this process, the extensive use of industrial robots is a key method to achieve these objectives. First, the use of industrial robots helps reduce environmental pollution; as the stock of industrial robots increases, urban industrial emissions of waste gases and solid waste are reduced [25]. Second, the application of industrial robots not only significantly enhances a company’s environmental performance but also achieves emission reductions by improving energy productivity and optimizing internal management efficiency, further promoting the greening of industries [26]. Lastly, industrial robots enhance productivity and resource utilization efficiency, reduce time costs, and can increase output without additional inputs, thus effectively enhancing input–output efficiency [27]. However, the production, transportation, and maintenance of industrial robots themselves may require substantial energy and raw materials, particularly during the early stages of their lifecycle. Moreover, the potential for a rebound effect should not be overlooked, where enterprises might tend to increase output through the use of robots, consequently raising overall energy consumption, especially in the absence of sufficient environmental protection measures. This increase in production-driven energy consumption could, paradoxically, undermine the potential benefits of robotic technology in improving energy efficiency [21,23].

Therefore, while industrial robots possess significant potential to drive urban green total factor energy efficiency, their comprehensive impact requires careful consideration within a broader lifecycle assessment and policy guidance. Only through thorough evaluation and prudent planning can we maximize their positive role in sustainable development while limiting their potential adverse effects. Accordingly, Hypothesis 1 is proposed in this paper.

Hypothesis 1. (H1)

The application of industrial robots can enhance urban green total factor energy efficiency.

3.2. Industrial Robots, Labor Productivity, and Green Total Factor Energy Efficiency

Industrial robots not only can directly enhance green total factor energy efficiency but also indirectly promote it by improving labor productivity. With the widespread deployment of industrial robots, the labor structure is also transformed, thereby optimizing the production efficiency and resource allocation. Li and Xu [28], through matching enterprise data with customs data, found that an increase in the use of robots significantly improved the labor productivity in Chinese manufacturing enterprises. Furthermore, Ma and Lai [29] studied the relationship between minimum wage standards and foreign direct investment, revealing the positive effects of increased labor productivity on residents’ work benefits and corporate product benefits. The growth in labor productivity not only raises the income levels of residents but also promotes an upgrade in consumption structure. As labor productivity continues to grow, the competitive advantages of China’s three major industries are also gradually improving, which is conducive to promoting the transformation of industries towards new directions in green development. While industrial robots have enhanced production efficiency in specific areas, the absence of timely advancements in accompanying industrial structures and resource allocation optimization may lead to uneven resource distribution and waste of productivity. Industries that excessively rely on robots tend to overlook the training and development of human resources, ultimately resulting in inefficient resource utilization, thereby reducing the overall energy efficiency of the industry. On this basis, Hypothesis 2 is proposed:

Hypothesis 2. (H2)

The application of industrial robots promotes urban green total factor energy efficiency by enhancing labor productivity.

3.3. The Mechanisms of Action between Industrial Robots and Green Total Factor Energy Efficiency

Green innovation primarily involves product and process innovations in energy conservation, emission reduction, environmental protection, and waste recycling, aimed at effectively addressing environmental issues. According to the theory of endogenous growth, technological advancement is seen as a core factor affecting energy efficiency, with technological innovation serving as the catalyst. Green technological innovations may impede the improvement of green total factor energy efficiency in industrial robots. Amjadi et al. [30] has revealed that such innovations could hinder increases in energy efficiency due to the rebound effect. Specifically, when green innovations lead to reduced energy market prices, firms might opt to substitute cheaper energy for other more expensive production factors, thereby increasing total energy consumption. Additionally, as energy efficiency improvements reduce the cost per unit of energy, this could further escalate consumer demand for energy, subsequently increasing overall energy consumption and diminishing the level of green total factor energy efficiency. The application of industrial robots is a significant marker of technological progress, accelerating the acquisition of advanced technologies by enterprises and expanding the channels for technological spillovers. Nie et al. [31] revealed the significant role of industrial robots in optimizing labor skill structures and task allocation, demonstrating that the use of industrial robots effectively enhances firms’ green innovation capabilities. Liu and Shen [32] employed a non-parametric index decomposition method to measure the biased technological progress index of prefecture-level cities, validating the promotive effect of environmentally biased technological innovations on urban emission reduction and energy efficiency enhancement. Therefore, industrial robots not only raise the starting point of urban technological innovation but also lay a solid technical foundation for deeper green innovative activities. As the level of green innovation continues to rise, the human capital in the field of green technological innovation in cities also increases, further promoting the development of new energy-saving robots and providing necessary support for the technological advancement in the field of industrial robots [33]. Moreover, the government’s growing emphasis on the economic and environmental benefits brought by green innovation technologies and its increased support for industries that enhance green innovation levels will further improve the city’s green total factor energy efficiency on the basis of industrial restructuring. Hence, we propose Hypothesis 3:

Hypothesis 3. (H3)

The level of green innovation has a positive moderating effect on the impact of industrial robots on improving green total factor energy efficiency.

To enhance the green total factor energy efficiency of a nation or region and strengthen its green core competitiveness, it is crucial to formulate appropriate environmental policies [34]. Xie et al. [35], using provincial panel data, revealed the significant role of environmental regulation in China’s green development. Environmental regulation refers to the measures of direct or indirect control and intervention by the government in socioeconomic activities for environmental protection and resource conservation. This mainly includes administrative regulations, market mechanisms, and raising public awareness. However, some have posited opposing views, Li et al. [36] through empirical analysis of provincial panel data in China, found that the relationship between environmental regulation and energy efficiency is positively U-shaped. This means that a positive effect on total factor energy efficiency is only observed when the intensity of environmental regulation exceeds a specific turning point.

On one hand, environmental regulations set emission standards for enterprises through governmental mandates, which, although costly, have proven effective in compelling enterprises to meet energy conservation and emission reduction standards. Moreover, command-type environmental regulations significantly promote enterprises’ energy conservation, emission reduction, and green productivity enhancement [37]. On the other hand, environmental regulations leverage market mechanisms to incentivize enterprises to voluntarily reduce pollution through policies like effluent fees and emissions trading, encouraging enterprises to adopt green clean technologies, thus reducing the costs of environmental pollution control and enhancing environmental efficiency. Additionally, as the concept of green development gains widespread acceptance in society, the environmental awareness of the public and enterprises continues to strengthen, helping to create a proactive environmental atmosphere, pushing forward technological progress and optimizing resource allocation [38].

Overall, not only does the implementation of environmental regulations play a complementary and strengthening role in enhancing the green total factor energy efficiency of industrial robots, but the application of industrial robots also provides certain technical support for the enforcement of environmental regulations. This bidirectional interaction greatly promotes the overall process of green development in China, highlighting the importance of combining environmental regulation with advanced technology. Hence, we propose Hypothesis 4:

Hypothesis 4. (H4)

The level of environmental regulation also has a positive moderating effect on the impact of industrial robots on improving green total factor energy efficiency.

3.4. Theoretical Framework

Figure 1 presents a graphical representation of the theoretical framework. In the following sections, we will utilize this framework to empirically examine the relationship between industrial robots and green total factor energy efficiency.

4. Research Design

4.1. Empirical Model Construction

To verify the impact of industrial robots on green total factor energy efficiency, based on the theoretical model discussed previously, we establish the following baseline regression model, as shown in Equation (1):

$$GTFE{E}_{it}={\alpha }_0+{\alpha }_{1}AP{R}_{it}+\sum _{i=2}^n{\alpha }_i{X}_{it}+{\varphi }_r+{\phi }_t+{\epsilon }_{it}$$

(1)

where $GTFE{E}_{it}$ represents the GTFEE of region i during period t, the core explanatory variable is the penetration rate of industrial robots in prefecture-level cities $AP{R}_{it}$, and the control variables $X_{it}$ include the level of real estate development, economic development level, industrial structure, population density, urbanization development level, and government constraint level. ${\varphi }_r$ represents city fixed effects, ${\phi }_t$ represents time fixed effects, and ${\epsilon }_{it}$ is the random disturbance term. The regression coefficient ${\alpha }_1$ measures the extent of the impact of industrial robot penetration on the city’s GTFEE.

4.2. Variable Selection and Data Sources

4.2.1. City Green Total Factor Energy Efficiency

Traditional total factor energy efficiency assessments primarily focus on input factors such as capital and labor, and their output evaluations typically consider only the total economic output. In contrast, the GTFEE assessment model is more comprehensive, considering the interactions between capital, labor, land, and other production factors, as well as incorporating energy consumption and unintended outputs (such as environmental pollutants) into the evaluation framework. Therefore, this study uses the global reference SBM-GML index method to measure GTFEE, providing a more comprehensive and accurate assessment of energy efficiency. The variables used are as shown in

Table 1. Input–output variable table.

Variables Category	Variables Name	Description
Input Variables	Labor	Total population at year-end of each prefecture-level city
	Capital	Capital stock of the prefecture-level city
	Energy	Standard coal consumption of the prefecture-level city
	Land	Built-up area within the jurisdiction of the city
Desired Output Variable	Regional GDP	Constant price regional gross domestic product
Undesired Output Variables	Industrial solid waste emissions	Total annual emissions of industrial solid waste
	Sulfur dioxide emissions	Total annual emissions of sulfur dioxide
	Industrial wastewater emissions	Total annual emissions of industrial wastewater

:

Firstly, drawing upon the research methodology of Zhang et al. [39], this study employs the perpetual inventory method to calculate the capital stock of prefecture-level cities, using 2006 as the base year for deflation. Simultaneously, following the approach of Wu et al. [40], we utilize the global stable nighttime light values to estimate the energy consumption data for prefecture-level cities. The brightness of nighttime lights serves as a proxy variable; its intensity reflects the scale of nocturnal economic activities, thereby indirectly measuring the extent of energy consumption. Since urban districts are generally regarded as the core areas of cities, playing central roles in administrative and economic activities and representing urban planning and management, this paper uses the land area of urban districts to represent the total land area of the cities. Additionally, to eliminate the effects of inflation and more accurately reflect real economic growth, all regional Gross Domestic Product (GDP) data are converted to constant prices with 2006 as the base year.

After collecting relevant input and output variable data, this study calculates green total factor energy efficiency using Matlab software R2021a. The preliminary results represent annual growth rates, primarily reflecting interannual fluctuations. To accurately assess the green total factor energy efficiency of each prefecture-level city, according to the method of Qiu et al. [41], using 2006 as the base year, the time series data of green total factor energy efficiency are cumulatively adjusted to ensure the authenticity and continuity of the data.

4.2.2. Urban Industrial Robot Penetration Rate

In this paper, we use APR to represent urban industrial robot penetration rate. Referencing the measurement methods of Acemoglu and Restrepo [42] and Wei et al. [43], this paper uses industrial robot data provided by the International Federation of Robotics. Initially, the IFR’s industry classifications are matched with the “National Economic Industry Classification” to obtain industry-level industrial robot penetration indicators, denoted as $P{R}_{j,t}$:

$$P{R}_{j,t}={\displaystyle \frac{Ro{b}_{j,t}}{Em{p}_{j,t=2010}}}$$

(2)

where $Ro{b}_{j,t}$ represents the stock of industrial robots in industry j for year t, and $Em{p}_{j,t=2010}$ denotes the employment figures in industry j for the year 2010. Using 2010 as the base year, and following the logic of the Bartik instrumental variable approach, the industrial robot penetration rate for each city is calculated based on the employment shares by industry in each city during the base year and the industry-specific robot penetration rates. The formula for calculating the industrial robot penetration rate $AP{R}_{it}$ for city i in year t is as follows:

$$AP{R}_{it}=\sum {\displaystyle \frac{Em{p}_{i,j,t=2010}}{Em{p}_{j,t=2010}}}\times P{R}_{j,t}$$

(3)

where $Em{p}_{i,j,t=2010}/Em{p}_{j,t=2010}$ represents the share of employment in industry j within region i in 2010 relative to the national employment in industry j for the same year. Unlike existing studies that focus on the provincial-level robot penetration, this paper disaggregates the industry-level industrial robot penetration to the city level, allowing for a more detailed examination of the impact of industrial robots on green total factor energy efficiency from a micro-level perspective. Additionally, this study employs the stock-based industrial robot penetration indicator to conduct relevant robustness checks.

4.2.3. Control Variables

Based on the theoretical model and the studies by Yu et al. [44] and Huang and Jiang [22], this paper selects six control variables $X_{it}$: (1) real estate development level (house), represented by the logarithm of real estate development investment; (2) economic development level (rgdp), measured by the ratio of regional GDP to the urban population; (3) industrial structure (industry), expressed as the proportion of the secondary industry’s added value to regional GDP; (4) population density (PD), calculated as the ratio of the urban resident population to the land area, and logarithmically transformed; (5) urbanization level (urban), measured by the proportion of the urban population to the total regional population; (6) government constraint (MC), represented by the ratio of local fiscal expenditure to regional GDP.The introduction of the explanatory variables, dependent variables, and control variables is presented in

Table 2. Introduction of the variables.

Variables	Definition	Description
GTFEE	Green total factor energy efficiency	Measure using the global reference SBM-GML index method.
APR	Industrial robot penetration rate	Calculated based on the employment shares by industry in each city during the base year and the industry-specific robot penetration rates.
house	Real estate development level	Represented as the logarithm of real estate development investment.
rgdp	Economic development level	The ratio of regional GDP to the urban population.
industry	Industrial structure	Defined as the proportion of the secondary industry’s added value to regional GDP.
PD	Population density	The ratio of the urban resident population to the land area.
urban	Urbanization level	The proportion of the urban population to the total regional population.
MC	Government constraint	The ratio of local fiscal expenditure to regional GDP.

.

4.2.4. Data Sources and Processing

Given that the large-scale application of industrial robots in China began in 2010, this study selects that year as the starting point for analysis to comprehensively capture the development trends during the initial application of industrial robots. To ensure the completeness and reliability of the data, cities with missing data for three consecutive years or more were excluded from the sample. The empirical analysis was conducted using panel data from 276 prefecture-level and above cities from 2010 to 2019. The data sources include the EPS China Data Database, CNRDS China Data Database, China City Statistical Yearbook, China Energy Statistical Yearbook, China Environmental Statistical Yearbook, and the statistical yearbooks and bulletins of individual cities.

4.2.5. Descriptive Statistics of Variables

Table 3. Descriptive statistics of variables.

Variable Type	Variables	N	Mean	Std. Dev.	Min	Max
Dependent Variable	GTFEE	2760	0.971	0.123	0.488	2.514
Independent Variable	APR	2760	0.307	0.619	0.002	8.468
Control Variables	house	2760	14.19	1.232	9.721	17.61
	rgdp	2760	2.740	4.228	0.078	44.18
	industry	2760	0.475	0.106	0.117	0.897
	PD	2760	5.766	0.908	1.612	7.876
	urban	2760	0.554	0.150	0.204	1.000
	MC	2760	0.704	0.641	0.053	5.852

presents the descriptive statistics for the main variables, including GTFEE and APR. The mean value of GTFEE is 0.971, with a standard deviation of 0.123. The noticeable difference between the minimum and maximum values indicates significant variation in green total factor energy efficiency across different cities and years. The mean value of APR is 0.307, with a substantial gap between the minimum and maximum values, reflecting regional and temporal differences in the usage of industrial robots.

4.3. Correlation between Industrial Robots and Green Total Factor Energy Efficiency

Based on the measurement results of industrial robot penetration and green total factor energy efficiency across various prefecture-level cities in China, the corresponding scatter plot and fitted regression line are depicted in Figure 2. As observed in Figure 2, there is a significant positive correlation between industrial robot penetration and green total factor energy efficiency. This suggests that the widespread adoption of industrial robots in cities is associated with improvements in green total factor energy efficiency. To further test the stability and reliability of this relationship, this paper will conduct empirical testing and in-depth analysis using more rigorous statistical methods based on Equation (1).

5. Empirical Results and Analysis

5.1. Empirical Analysis Methods

In this section, we employed a series of complex empirical analysis steps to ensure the accuracy and reliability of our research findings. Initially, a baseline regression analysis was conducted to establish a preliminary model for examining the impact of key explanatory variables on dependent variables. Subsequently, given the potential endogeneity issues, this study introduced instrumental variables into the model to more accurately estimate the causal relationships between variables. Furthermore, robustness tests were performed by adjusting the dependent and explanatory variables and narrowing the observation years to meet specific analytical needs. Finally, an in-depth heterogeneity analysis was conducted to investigate how regional and urban typologies affect the model outcomes, with particular focus on the effects in the Eastern, Central, Western, and Northeastern regions, as well as in resource-based cities, old industrial bases, and low-carbon pilot cities. The combination of these steps not only strengthened the rigor of the analysis but also enhanced the interpretability of the results and their policy relevance. The flowchart for the empirical analysis is shown in Figure 3, with the bolded sections indicating the start and end of the empirical analysis.

5.2. Baseline Regression Results

Table 4. Baseline regression of industrial robot penetration on green total factor energy efficiency.

Variables	GTFEE
	(1)	(2)	(3)	(4)
APR	0.1157 ***	0.0964 ***	0.1020 ***	0.0694 ***
	(0.0031)	(0.0031)	(0.0038)	(0.0034)
Constant	0.9360 ***	0.9419 ***	1.3497 ***	1.6469 ***
	(0.0021)	(0.0015)	(0.0298)	(0.1535)
Control Variables	No	No	Yes	Yes
Year FE	No	Yes	No	Yes
City FE	No	Yes	No	Yes
N	2760	2760	2760	2760
$R^2$	0.3366	0.7965	0.4222	0.8293

Note: Standard errors clustered at the region level in parentheses. FE indicates Fixed Effects model. *** indicates statistical significance at the 1% level, respectively.

reports the baseline regression results for the impact of industrial robot penetration on urban green total factor energy efficiency. In column (1), ordinary least squares regression is used solely with industrial robot penetration and green total factor energy efficiency, where the estimated coefficient for APR is significantly positive at the 1% statistical level. This indicates that the application of industrial robots significantly enhances regional green total factor energy efficiency. Column (2) builds upon column (1) by adding control variables; the results show that the estimated coefficient for APR remains significantly positive at the 1% level, further corroborating the correlation shown in Figure 1 and indicating a significant promotional effect of industrial robots on green total factor energy efficiency. In column (3), based on column (1), city and year fixed effects are introduced, and the positive impact of industrial robot penetration on green total factor energy efficiency remains significant at the 1% confidence level. In column (4), based on column (2), city and year fixed effects are added. Compared to column (2), there is no change in the estimated coefficient or significance level for APR, suggesting that the significant effect of industrial robots on green total factor energy efficiency persists. Intuitively, an increase of one unit in urban industrial robot penetration corresponds, on average, to an increase of about 0.0694 units in urban green total factor energy efficiency. In summary, the APR coefficients in columns (1) to (4) of Table 1 are all significantly positive. These results effectively support Hypothesis 1, demonstrating that increased industrial robot penetration significantly boosts urban green total factor energy efficiency.

5.3. Endogeneity Concerns

Considering the potential for omitted variable bias and the possibility that improvements in regional total factor energy efficiency might increase demand for industrial robots, thereby causing reverse causality among the variables, this study employs the instrumental variable (IV) method to analyze endogeneity. Two approaches are used to select instrumental variables: first, following the research by Lin and Li [45] on industrial robots and higher education enrollment rates, the lagged value of the explanatory variable is chosen as an instrumental variable for endogeneity analysis. This selection is based on the correlation between the historical values of the core explanatory variable and its current values, meeting the relevance requirement for an instrumental variable; furthermore, since the lagged variable is unaffected by other variables in the model, it satisfies the condition of exogeneity. Additionally, this paper also uses the USA industry-level industrial robot stock data and employment numbers for each industry to calculate the penetration rate of industrial robots in Chinese cities, as illustrated in Equation (4):

$$AP{R}_{i,t}^{IV}=\sum {\displaystyle \frac{Em{p}_{i,j,t=2010}}{Em{p}_{j,t=2010}}}\times {\displaystyle \frac{Ro{b}_{j,t}^{US}}{Em{p}_{j,t=2010}^{US}}}$$

(4)

where $Ro{b}_{j,t}^{US}$ represents the stock of industrial robots in industry j in year t in the USA and $Em{p}_{j,t=2010}^{US}$ denotes the employment numbers in industry j in the USA for the base year 2010. Given the widespread use of robots in the USA’s industrial sectors and the technological characteristics shared among similar industries, this significantly influences the application of robots in related sectors in China, fulfilling the relevance requirement. The industrial robot penetration rate calculated using the USA robot data does not independently increase China’s green total factor energy efficiency, meeting the exogeneity requirement. Hence, $AP{R}_{i,t}^{IV}$ is chosen as the instrumental variable for industrial robot penetration at the city level in China, effectively addressing the endogeneity issue present in the model.

The study employs a two-stage least squares (2SLS) method for the endogeneity analysis, with specific results presented in

Table 5. Regression results using instrumental variables.

Variables	$\mathit{L}.\mathit{APR}$		${\mathit{APR}}^{\mathit{IV}}$
	APR	GTFEE	APR	GTFEE
	(1)	(2)	(3)	(4)
APR		0.0690 ***		0.0628 ***
		(0.0034)		(0.0037)
Instrumental Variable	1.2092 ***		1.1899 ***
	(0.0028)		(0.0124)
Constant	−0.0758	1.4209 ***	−1.6503 ***	1.5744 ***
	(0.1048)	(0.1840)	(0.4103)	(0.1719)
Control Variables	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
City FE	Yes	Yes	Yes	Yes
N	2484	2484	2760	2760
R2	0.9976	0.8486	0.9512	0.8291
First-stage F statistic	$1.9\times {10}^4$ ***		9232.74 ***

Note: Standard errors are clustered at the region level in parentheses. *** indicate statistical significance at the 1% levels, respectively.

. To test the validity of the chosen instrumental variables, a weak instrument variable test is subsequently conducted, and the Cragg–Donald Wald F statistic is reported in Table 5 to ensure the strength and appropriateness of the instrumental variables.

As shown in Table 5, in columns (1) and (2), the lagged value of APR is used as the instrumental variable, and in columns (3) and (4), $AP{R}_{i,t}^{IV}$ is used as the instrumental variable. Initially, the first-stage regression results in the table indicate that the selected instrumental variable shows a positive relationship at the 1% significance level, fully satisfying the relevance condition for instrumental variables. More importantly, the F-statistic is significantly above the conventional threshold of 10, strongly indicating that the chosen instrumental variable is sufficiently robust, eliminating the possibility of a weak instrumental variable problem. After introducing the aforementioned instrumental variables in the model, the second-stage regression analysis indicates that the estimated coefficient for industrial robot penetration has a positive impact at the 1% significance level. This finding not only highlights the significant positive influence of industrial robots on enhancing urban green total factor energy efficiency but also reiterates the support for Hypothesis 1. These results provide solid statistical backing for the environmental benefits of industrial robots, affirming their key role in driving urban sustainable development.

5.4. Robustness Checks

To ensure the reliability of the baseline regression results, this paper conducts multiple robustness checks by replacing core explanatory variables, adjusting the dependent variables, and reducing the years considered.

5.4.1. Replacing Core Explanatory Variables

In this study, we employ two methods to replace the explanatory variables for robustness checks. (a) Referencing the study by Xu et al. [19] on industrial robots and the gender wage gap, the penetration rate of industrial robots is constructed using the annual installation data of industrial robots:

$$AP{R}_{i,t}^I=\sum {\displaystyle \frac{Em{p}_{i,j,t=2010}}{Em{p}_{j,t=2010}}}\times {\displaystyle \frac{Ro{b}_{j,t}^I}{Em{p}_{j,t=2010}^I}}$$

(5)

where $Ro{b}_{jt}^I$ represents the annual installation volume of industrial robots in industry j in year t in China. (b) By accessing the advanced search on the “Patent Pool” official website, the number of “robot” patents from 2010 to 2019 across 276 prefecture-level cities represents the regional level of robot usage. According to the empirical results in Table 6, column (1) shows that when replacing the stock measure with annual installation data of industrial robots in China, the impact coefficient on green total factor energy efficiency is significantly positive at the 1% level. Upon replacing the explanatory variable, this robust finding supports Hypothesis 1, demonstrating a significant positive impact of industrial robot usage on enhancing urban green total factor energy efficiency. Column (2) uses the number of industrial robot patents to measure the development level of industrial robots, and the results similarly show a significant positive impact on green total factor energy efficiency. This further confirms Hypothesis 1, demonstrating the positive role of industrial robot technological advancement in enhancing green total factor energy efficiency.

5.4.2. Adjusting the Dependent Variable

In the process of adjusting the dependent variable, this study inspired by the method of Sun and Zhou [46], divides green total factor energy efficiency into two components: technical efficiency and technical progress efficiency. By focusing on technical progress efficiency as the dependent variable replacing the measurement of prefecture-level green total factor energy efficiency, the results in column (4) indicate that even after replacing the dependent variable, the impact of industrial robots on green total factor energy efficiency remains significantly positive at the 1% level. This observation reiterates the pivotal role of industrial robots in substantially augmenting green total factor energy efficiency, thereby substantiating Hypothesis 1.

5.4.3. Reducing the Years

When analyzing economic panel data, the initial data often show a certain level of instability due to policy changes and technologies in the early stages of development. As time progresses, the related technologies and policies mature, and the data tend to stabilize. By selecting data from years that are more stable and representative, the analysis can more accurately capture the prevalence of robot technology and its effects. To more truly reflect long-term trends and the actual effectiveness of policies, this paper follows the research method of Duan and Zhuang [47], excluding data from the earliest three years of the study period for robustness checks. From the regression results in column (3), the coefficient for APR remains significantly positive at the 1% level. This indicates that even after excluding data from the initial years of the study, the application of industrial robots continues to have a significant promotional effect on green total factor energy efficiency. This result once again confirms Hypothesis 1 and provides strong evidence in support of the conclusions drawn earlier in the paper.

Table 6. Robustness test.

Variables	GTFEE	GTFEE	GTFEE	Technological Change
	(1)	(2)	(3)	(4)
$AP{R}_{it}^I$	0.3133 ***
	(0.0163)
IRP		1.0963 ***
		(0.0733)
APR			0.0643 ***	0.0691 ***
			(0.0039)	(0.0026)
Constant	1.5632 ***	1.6558 ***	0.8488 ***	0.8700 ***
	(0.1538)	(0.1607)	(0.2238)	(0.1161)
Control Variables	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
City FE	YES	YES	YES	YES
N	2760	2760	1932	2760
$R^2$	0.8272	0.8178	0.8934	0.8117

Note: Standard errors are clustered at the region level in parentheses. *** indicate statistical significance at the 1% level.

Table 6. Robustness test.

Variables	GTFEE	GTFEE	GTFEE	Technological Change
	(1)	(2)	(3)	(4)
$AP{R}_{it}^I$	0.3133 ***
	(0.0163)
IRP		1.0963 ***
		(0.0733)
APR			0.0643 ***	0.0691 ***
			(0.0039)	(0.0026)
Constant	1.5632 ***	1.6558 ***	0.8488 ***	0.8700 ***
	(0.1538)	(0.1607)	(0.2238)	(0.1161)
Control Variables	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
City FE	YES	YES	YES	YES
N	2760	2760	1932	2760
$R^2$	0.8272	0.8178	0.8934	0.8117

Note: Standard errors are clustered at the region level in parentheses. *** indicate statistical significance at the 1% level.

5.5. Heterogeneity Analysis

5.5.1. Regional Heterogeneity Analysis

It is well known that China’s vast territory is rich and diverse in resources, with the eastern coastal areas compared to the central regions having more significant development advantages over the remote western areas. The relative remoteness of the western regions and their distance from major ports pose substantial challenges to trade development. Furthermore, the flow of high-end talent, technological innovation activities, financial resources, and foreign direct investments tend to concentrate in coastal cities, exacerbating regional imbalances.

In light of these regional development disparities, this paper conducts an in-depth analysis of the specific impact of industrial robot applications on green total factor energy efficiency across different regions. Previous research often divides China’s geographic areas into east–west or north–south to study green total factor energy efficiency. According to the regional division guide by the National Bureau of Statistics, this paper divides cities into four major regions: East, Central, West, and Northeast. The results are presented in

Table 7. Heterogeneity analysis of different regions.

Variables	GTFEE
	(1)	(2)	(3)	(4)
	Eastern Region	Central Region	Western Region	Northeast Region
APR	0.0725 ***	0.0596 ***	0.0615 ***	0.1132 ***
	(0.0056)	(0.0121)	(0.0084)	(0.0154)
Constant	0.1077	1.7215 ***	1.3304 ***	2.1192 ***
	(0.4947)	(0.2506)	(0.1812)	(0.6266)
Control Variables	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
City FE	YES	YES	YES	YES
N	860	790	780	330
$R^2$	0.8488	0.8323	0.8147	0.8963

Note: Standard errors areclustered at the region level in parentheses. *** indicate statistical significance at the 1% levels, respectively.

:

Table 7 shows that the coefficients of APR from column (1) to column (4) are significantly positive, indicating that the widespread application of industrial robots can significantly enhance green total factor energy efficiency across regions. A further observation reveals that the APR coefficient in the Northeast region is as high as 0.1132, which is significantly higher than the other three regions. This indicates that in the Northeast, the impact of industrial robots on energy efficiency is particularly notable, primarily because this region has a very solid industrial base. Since the Northeast has a concentration of heavy and manufacturing industries, which typically consume a lot of energy, there is a higher potential for improvement in green total factor energy efficiency. With the introduction and application of industrial robots, production processes have been optimized, significantly enhancing green total factor energy efficiency.

5.5.2. Urban Heterogeneity Analysis

When studying the impact of industrial robots on urban green total factor energy efficiency, considering the heterogeneity of cities is crucial. This research selects three typical city types—resource-based cities, industrial base cities, and low-carbon pilot cities—to explore the specific impact differences of industrial robot applications under different urban contexts.

For energy, as a key strategic resource, its efficiency improvement relies not only on the dynamic changes in the urban economic structure but also deeply on the endowment of resources. For example, the economies of resource-based cities are often driven by local mineral resources and forestry development, and this high dependence on natural resources significantly affects the city’s development characteristics and future trajectory. According to the “National Resource-Based Cities Sustainable Development Plan (2013–2020)”, this paper classifies the research samples into resource-based and non-resource-based cities to further explore the heterogeneity of urban resource endowments on the application of industrial robots and their impact on green total factor energy efficiency. Consequently, a dummy variable for city resource endowment is constructed, assigning a value of 1 to resource-based cities and 0 to non-resource-based cities, along with constructing an interaction term between the resource endowment dummy variable and APR.

Old industrial base cities are primarily defined as industrial centers built by the state, relying on heavy industry backbone enterprises during the planned economy period in China. The rise of these cities was concentrated during the “First Five-Year Plan”, the “Second Five-Year Plan”, and the “Third Front” construction periods, aimed at enhancing the national industrial base and self-sufficiency in production to ensure the stability of industrial production. Typical old industrial base cities, dominated by heavy industries such as steel, machinery manufacturing, and chemicals, have formed a relatively singular industrial structure. Benefiting from national support, these cities have shown significant advantages, especially in infrastructure, transportation, energy supply, and industrial facilities. Following the classification and definition by the National Development and Reform Commission’s “National Old Industrial Base Adjustment and Transformation Plan (2013–2022)”, similar to resource-based cities, this paper continues to construct a dummy variable to distinguish whether a city is an old industrial base city, assigning a value of 1 to old industrial base cities and 0 to non-old industrial base cities, and constructs an interaction term between this dummy variable and APR.

The low-carbon city pilot policy is one of China’s strategic measures to respond to global climate change challenges and promote sustainable urban development. This policy aims to optimize the urban energy consumption structure and foster innovation and the development of green technologies, thereby achieving a transformation and upgrade in industrial and energy structures. This, in turn, effectively enhances the green total factor energy efficiency of cities, laying a solid foundation for their sustainable development. In 2010, five provinces, including Guangdong and Liaoning, along with eight cities, including Tianjin and Chongqing, were designated as the first batch of low-carbon city pilot areas. Subsequently, the National Development and Reform Commission issued the second and third batches of low-carbon city pilot areas in 2012 and 2017, respectively. In reference to the above, this paper continues to construct a dummy variable to identify low-carbon pilot cities, assigning a value of 1 to low-carbon pilot cities and 0 otherwise, and constructs an interaction term with APR. The urban heterogeneity regression model is illustrated in Equations (6)–(8):

$$GTFE{E}_{it}={\alpha }_0+{\alpha }_{1}AP{R}_{it}+{\alpha }_{2}AP{R}_{it}\times Resourc{e}_{it}+\sum _{i=3}^n{\alpha }_i{X}_{it}+{\varphi }_i+{\phi }_t+{\epsilon }_{it}$$

(6)

$$GTFE{E}_{it}={\alpha }_0^{\prime }+{\alpha }_1^{\prime }AP{R}_{it}+{\alpha }_2^{\prime }AP{R}_{it}\times Bas{e}_{it}+\sum _{i=3}^n{\alpha }_i^{\prime }{X}_{it}+{\varphi }_i+{\phi }_t+{\epsilon }_{it}$$

(7)

$$GTFE{E}_{it}={\alpha }_0^{″}+{\alpha }_1^{″}AP{R}_{it}+{\alpha }_2^{″}AP{R}_{it}\times L{C}_{it}+\sum _{i=3}^n{\alpha }_i^{″}{X}_{it}+{\varphi }_i+{\phi }_t+{\epsilon }_{it}$$

(8)

where $Resourc{e}_{it}$ represents the dummy variable for resource-based cities, $Bas{e}_{it}$ denotes the dummy variable for old industrial base cities, and $L{C}_{it}$ signifies the dummy variable for low-carbon pilot cities. Additionally, ${\alpha }_2$, ${\alpha }_2^{\prime }$, and ${\alpha }_2^{″}$ quantify the impact of these dummy variables interacting with APR on green total factor energy efficiency. The results of the heterogeneity analysis are shown in

Table 8. Heterogeneity analysis of different city types.

Variables	GTFEE
	(1)	(2)	(3)
APR	0.0698 ***	0.0479 ***	0.0489 ***
	(0.0034)	(0.0041)	(0.0068)
APR × Resource	0.0573 ***
	(0.0132)
APR × Base		0.0448 ***
		(0.0048)
APR × LC			0.0231 ***
			(0.0066)
Constant	1.6011 ***	1.4239 ***	1.6684 ***
	(0.1533)	(0.1527)	(0.1532)
Control Variables	YES	YES	YES
Year FE	YES	YES	YES
City FE	YES	YES	YES
N	2760	2760	2760
$R^2$	0.8306	0.8351	0.8301

Note: Standard errors are clustered at the region level in parentheses. *** indicate statistical significance at the 1% levels, respectively.

.

According to the results shown in column (1) of Table 8, compared to non-resource-based cities, industrial robots significantly enhance green total factor energy efficiency in resource-based cities. Given that resource-based cities depend on high-energy-consuming industries, they have a high baseline of energy consumption. In such a high-energy consumption context, the introduction of technologies or management improvements like industrial robots can significantly increase energy utilization efficiency and reduce absolute energy consumption, often exceeding the effects seen in non-resource-based cities. Additionally, facing severe environmental pollution and ecological pressures, resource-based cities are more proactive in adopting technologies that reduce environmental impact. Thus, resource-based cities demonstrate considerable potential in driving economic structural transformation and sustainable development.

From column (2) of Table 8, it can be observed that in old industrial base cities, the interaction coefficient between the application of industrial robots and green total factor energy efficiency is significantly positive. This indicates that these cities, with the increasing application of industrial robots, exhibit more pronounced improvements in green total factor energy efficiency. This difference primarily stems from the additional positive impacts generated by industrial robot technology in old industrial base cities. Due to these cities’ solid industrial foundations and effective utilization of robot technology, they play a crucial role in enhancing production efficiency and optimizing energy usage. In old industrial base cities, typically dominated by heavy industries like steel and machinery manufacturing, although these industries are concentrated energy consumers, the adoption of efficient automation and robot technology not only enhances energy utilization efficiency but also reduces energy wastage and optimizes production processes.

From the results in column (3) of Table 8, in low-carbon pilot cities, as the penetration of industrial robots increases, the enhancement of green total factor energy efficiency becomes more evident. Low-carbon cities typically support the development of low-carbon industries through industrial upgrading and structural adjustment, optimizing traditional high-carbon emission industries and rearranging industrial structures. With the introduction of automation technologies like industrial robots, particularly in manufacturing and production processes, these technologies significantly help these cities reduce energy consumption and pollutant emissions, thus enhancing the region’s green total factor energy efficiency.

6. Channel and Mechanism Analysis

6.1. Impact Channel Examination

Following the empirical results described above, the application of industrial robots significantly enhances urban green total factor energy efficiency. To further understand the interactions between variables, this paper introduces mediating variables to explore potential mediating effects. Mediating effect analysis helps reveal how the explanatory variable influences the dependent variable through one or more intermediary variables and provides a more comprehensive perspective for understanding the intrinsic mechanisms of the causal chain. Drawing on the research by Song and Zuo [18] on the relationship between industrial robot usage and labor productivity, as well as the measurement methods of labor productivity by Fan [48], this paper calculates the ratio of non-agricultural output to non-agricultural employment in prefecture-level cities to measure labor productivity. Referring to Wen et al. [49], a mediation effect model is constructed, as shown in Equations (9)–(11):

$$GTFE{E}_{it}={\alpha }_0+{\alpha }_{1}AP{R}_{it}+\sum _{i=2}^n{\alpha }_i{X}_{it}+{\varphi }_i+{\phi }_t+{\epsilon }_{it}$$

(9)

$$L{P}_{it}={\beta }_0+{\beta }_{1}AP{R}_{it}+\sum _{i=2}^n{\beta }_i{X}_{it}+{\varphi }_i+{\phi }_t+{\epsilon }_{it}$$

(10)

$$GTFE{E}_{it}={\gamma }_0+{\gamma }_{1}AP{R}_{it}+{\gamma }_{2}L{P}_{it}+\sum _{i=3}^n{\gamma }_i{X}_{it}+{\varphi }_i+{\phi }_t+{\epsilon }_{it}$$

(11)

where $L{P}_{it}$ represents the labor productivity in region i in year t; ${\beta }_1$ measures the impact of industrial robot penetration on labor productivity; and ${\gamma }_1$ and ${\gamma }_2$, respectively, represent the impact of industrial robots and labor productivity on urban green total factor energy efficiency. The results of the mediation effect analysis are shown in

Table 9. Analysis of influence channels: a perspective on labor productivity.

Variables	GTFEE	LP	GTFEE
	(1)	(2)	(3)
APR	0.0694 ***	1.4874 ***	0.0610 ***
	(0.0034)	(0.2667)	(0.0033)
LP			0.0044 ***
			(0.0002)
Constant	1.6469 ***	7.2482	1.5892 ***
	(0.1535)	(11.7933)	(0.1444)
Control Variables	YES	YES	YES
Year FE	YES	YES	YES
City FE	YES	YES	YES
N	2760	2760	2760
$R^2$	0.8293	0.8195	0.8450

Note: Standard errors are clustered at the region level in parentheses. *** indicate statistical significance at the 1% levels, respectively.

.

The baseline regression results in column (1) of Table 9 have already tested and confirmed that the application of industrial robots significantly enhances urban green total factor energy efficiency, thereby satisfying the conditions for testing mediation effects. Next, from the perspective of labor productivity, we explore the channels through which industrial robots contribute to the growth of green total factor energy efficiency. The results in column (2) of the table show that the application of industrial robots significantly increases labor productivity. The results in column (3) indicate that labor productivity has a positive impact on urban green total factor energy efficiency. Since ${\beta }_1$, ${\gamma }_1$, and ${\gamma }_2$ are all significantly positive, it can be concluded that industrial robots enhance urban green total factor energy efficiency by improving labor productivity. The observed reduction in the coefficient of industrial robots on green total factor energy efficiency, from 0.0694 in column (1) to 0.0610 in column (3), upon the inclusion of labor productivity, underscores the mediating role of labor productivity. This relationship not only persists but also highlights the substantial positive link between labor productivity and green total factor energy efficiency. Consequently, these results provide strong empirical support for Hypothesis 2, which posits that industrial robots boost urban green total factor energy efficiency primarily through improving labor productivity.

6.2. Mechanism Examination

To validate Hypotheses 3 and 4, this study introduces levels of green innovation and environmental regulation for a moderation effect analysis. Compared to utility and design patents, green invention patents exhibit greater innovation and technical depth. Based on city-level accessible data, following the research approach by Wang et al. [50] and Li [51], this paper uses the number of green invention patent applications per 10,000 people to measure the level of green innovation in cities. A higher value indicates a higher level of green innovation in the city, and vice versa. Moreover, due to the diversity of government interventions and the variability of environmental regulation tools, measuring environmental regulation presents certain complexities. Following the measurement methods by Yu and Peng [52] and Cai and Wang [53], this paper uses the inverse of the ratio of total urban wastewater, waste, and solid dust emissions to city GDP (in tons per ten thousand yuan) to measure the intensity of environmental regulation. A higher value of this indicator suggests stronger environmental regulation intensity. Referencing Jiang [54], a moderation effect model is constructed, as shown in Equations (12) and (13):

$$GTFE{E}_{it}={\delta }_0+{\delta }_{1}AP{R}_{it}+{\delta }_{2}AP{R}_{it}\times G{I}_{it}+\sum _{i=3}^n{\delta }_i{X}_{it}+{\varphi }_i+{\phi }_t+{\epsilon }_{it}$$

(12)

$$GTFE{E}_{it}={\delta }_0^{\prime }+{\delta }_1^{\prime }AP{R}_{it}+{\delta }_2^{\prime }AP{R}_{it}\times E{R}_{it}+\sum _{i=3}^n{\delta }_i^{\prime }{X}_{it}+{\varphi }_i+{\phi }_t+{\epsilon }_{it}$$

(13)

where $G{I}_{it}$ represents the green innovation level in region i in year t and $E{R}_{it}$ indicates the environmental regulation intensity in region i in year t. ${\delta }_2$ and ${\delta }_2^{\prime }$, respectively, quantify the impact of the moderation variables with APR interaction on green total factor energy efficiency. The results of the moderation effect analysis are presented in

Table 10. Impact mechanisms: green tech innovation and environmental regulation intensity.

Variables	GI	GTFEE	ER	GTFEE
	(1)	(2)	(3)	(4)
APR	0.0966 ***	0.0698 ***	0.4045 ***	0.0597 ***
	(0.0021)	(0.0051)	(0.0681)	(0.0037)
APR × GI		0.0181 ***
		(0.0061)
GI		−0.1455 ***
		(0.0427)
APR × ER				0.0018 ***
				(0.0004)
ER				0.0040 ***
				(0.0012)
Constant	−0.2397 **	1.6104 ***	−14.2093 ***	1.8609 ***
	(0.0936)	(0.1535)	(3.0312)	(0.1537)
Control Variables	YES	YES	YES	YES
Year FE	YES	YES	YES	YES
City FE	YES	YES	YES	YES
N	2760	2760	2760	2760
$R^2$	0.8780	0.8302	0.4339	0.8343

Note: Standard errors are clustered at the region level in parentheses. *** and ** indicate statistical significance at the 1% and 5% levels, respectively.

:

The results, as analyzed from columns (1) to (2) in Table 10, indicate that the APR coefficient ${\delta }_1$ and the interaction term coefficient ${\delta }_2$ are significantly positive at the 1% confidence level. This finding strongly supports the notion that the level of green innovation serves as a key mechanism through which industrial robots enhance the urban green total factor energy efficiency. As cities advance in green innovation, the developed robotic technologies become more efficient and energy-saving. This not only optimizes production processes and management efficiency but also promotes the advancement of energy-saving and emission reduction systems, thereby significantly improving the effective utilization of energy. These results solidly back Hypothesis 3, suggesting that enhancements in the level of green innovation directly foster an increase in urban green total factor energy efficiency.

The regression analysis results from columns (3) to (4) in Table 10 indicate that ${\delta }_1^{\prime }$ and ${\delta }_2^{\prime }$ are positive at the 1% significance level, suggesting that environmental regulation plays a positive moderating role in enhancing the green total factor energy efficiency of industrial robots in cities. In regions with stricter environmental regulations, the positive impact of industrial robots on green total factor energy efficiency is more pronounced. Industrial robots contribute to saving resources and reducing pollution emissions by promoting technological innovation and upgrading production technologies, thereby improving energy efficiency. The government encourages enterprises to adopt more advanced technological solutions by setting strict emission and energy efficiency standards to achieve energy saving and emission reduction goals. Furthermore, the government also encourages public participation in environmental activities, jointly reducing environmental pollution and promoting sustainable urban development. These findings support Hypothesis 4, affirming that the level of environmental regulation positively moderates the impact of industrial robots on enhancing green total factor energy efficiency.

7. Conclusions, Policy Implications, and Limitations

7.1. Conclusions

This paper employs empirical analysis methods, utilizing data from prefecture-level and above cities from 2010 to 2019, to thoroughly examine the impact of industrial robot applications on urban green total factor energy efficiency. The research findings underscore the crucial role that industrial robots play in enhancing green total factor energy efficiency. Robots directly promote the efficient use of energy and environmental protection by optimizing production processes and reducing energy consumption. Additionally, industrial robots also indirectly affect urban green total factor energy efficiency by enhancing regional labor productivity. With the gradual increase in the application of industrial robots, both urban labor productivity and green total factor energy efficiency have seen significant improvements. Moreover, the continuous enhancement of urban green innovation levels and environmental regulation also further strengthens the positive impact of industrial robots on green total factor energy efficiency, promoting the sustainable use of energy.

Furthermore, through regional heterogeneity analysis, this paper also finds that the impact of industrial robots on enhancing green total factor energy efficiency is most significant in the Northeast region. This is primarily due to the Northeast region’s lead in the application and technical support of industrial robots compared to other regions. Specifically, in resource-based cities, industrial base cities, and low-carbon pilot cities, the impact of industrial robots on green total factor energy efficiency is more pronounced. This finding highlights the differentiated needs and potential of different types of cities in advancing the application of industrial robots and enhancing energy efficiency.

7.2. Policy Implications

Based on the findings, the application of industrial robots can drive improvements in green total factor energy efficiency, indicating that promoting the deployment of industrial robots nationwide to achieve green development is feasible. To further tap into the potential of industrial robots in fostering green development, this study offers three following recommendations.

7.2.1. Enhance Investment in Technology R&D and Talent Development

First, investments should be bolstered in domestic robotics research and development, and through homegrown innovation, decrease reliance on foreign operating systems and chips, which will significantly reduce the production and operational costs of domestic industrial robots. Second, businesses should continually refine their employment structures, hire a substantial number of professional technical talents in the field of robotics, and boost investment in internal talent training to expand employees’ expertise and skills in robot operation, maintenance, and energy management, thereby ensuring effective utilization of robot technology. Finally, sustainable urban development can be realized through a dual strategy of equipment upgrading and talent development, expediting the application and widespread adoption of artificial intelligence and emerging industry technologies.

7.2.2. Adopt Differentiated Environmental Policies

It is recommended that different environmental policies be adopted according to regional characteristics. For central and western regions, it is advised to accelerate the phasing out of high-energy-consuming and technologically outdated industries and continue to promote the development of the urban industrial internet. Additionally, the government should provide comprehensive support for enterprises in funding, technology, and talent training to adopt robotic manufacturing, enabling non-resource-based cities, non-industrial base cities, and non-low-carbon pilot cities to promote intelligent transformation in industry and other sectors through knowledge and technology, thus enhancing the potential of industrial robots in energy conservation and emission reduction.

7.2.3. Boost Urban Green Innovation and Environmental Regulation

Cities are encouraged to adopt green technologies to replace traditional technologies and attract more talented individuals to the green innovation sector. This is crucial for enhancing a city’s green innovation capabilities. Furthermore, perfecting the environmental regulatory policy framework is a key component in enhancing urban green total factor energy efficiency. Government agencies need to continuously leverage advanced technological means to strengthen the enforcement of environmental regulations; they should also guide the public and the media to monitor and evaluate the implementation and effects of environmental policies, thereby forming an effective external supervision and constraint mechanism. For major environmental protection projects, the government should disclose information and regularly hold public hearings to fully listen to and integrate public opinions and suggestions, thereby improving regulatory transparency and public participation, and continuously promoting sustainable urban development.

7.3. Limitations

We have explored the enhancement of GTFEE through the use of industrial robots, but there are still areas that need further investigation. First, we examined the impact of industrial robots on the GTFEE of cities from a macro perspective but did not delve into how these robots could reduce resource wastage and improve energy efficiency within regional trade activities. Second, we have not yet analyzed how micro-enterprises can improve their production processes and energy management by adopting industrial robots, thereby enhancing their green productivity. Consequently, future research could broaden the current study framework to more comprehensively investigate how industrial robots can enhance resource efficiency in trade activities and how technological innovations in micro-enterprises can promote environmental sustainability.