The impact of industrial robots application on air pollution in China: Mechanisms of energy use efficiency and green technological innovation

Abstract

The battle against air pollution in China persists, and haze remains over cities. Whether industrial robots, as the core technology of intelligent manufacturing, can improve city air quality in the process of production has not been determined. Using the International Federation of Robotics data and Chinese city air pollution data (2013–2018), this study finds that industrial robots significantly reduce city air pollution levels (PM2.5, PM10, and SO₂), which remains robust after addressing endogeneity. The mechanism of action lies in the synergistic benefits of industrial robots in reducing city air pollution levels by effectively improving energy use efficiency and promoting green technological innovation. Heterogeneity analysis suggests that industrial robots, as the incarnation of green technology, can be an effective alternative tool to green policies, such as low-carbon piloting, resource planning, and environmental regulation. This study empirically confirms that industrial robots are environment-friendly technologies that can provide new policy ideas to promote air pollution prevention and control in the industrialization process.

Keywords

Industrial robot intelligence techniques air pollution energy use efficiency green technology innovation

Introduction

The haze that hangs over cities not only has a huge impact on production and life but also poses a serious threat to human health, exacerbates climate change, and damages the ecological environment. China attaches great importance to air pollution control, and according to the Three-Year Plan on Defending the Blue Sky released by the State Council in 2018, it was planned that by 2020 the total national SO₂ emissions would be reduced by more than 15% compared to 2015, the average PM2.5 concentration in cities exceeding the standard would be reduced by more than 18%, and the percentage of heavily and severely polluted days would be reduced by more than 25%. In fact, these three indicators decreased by 82.9%, 28.8%, and 62.5%, respectively, in 2020, exceeding the planned targets, and PM10 concentration also decreased by 39.8%. Despite these reductions, most air pollutant concentrations in Chinese cities remain above the levels set by the Air Quality Guidelines (AQG) (Figure 1), which can drive the formation of heavily polluted weather under unfavorable meteorological conditions. Thus, it is necessary to continue to focus on energy and industrial restructuring and the promotion of green and high-quality development in order to eliminate heavy pollution.

Figure 1.

Dynamic changes of concentrations of three air pollutants in China (2013–2021).

Industrial robots refer to multi-joint manipulators or high degree-of-freedom robots employed in industrial activities, such as serial manipulators widely used in the industrial assembly,¹ high-speed parallel robots widely used in industrial production lines such as food and pharmaceuticals,² and collaborative robots that maximize the collaborative advantages between humans and robots.³ As a key device for intelligent manufacturing, industrial robots are expected to aid the environmental improvement of industrial systems in cities, and the combined and coordinated development of industrial economy and environmental governance. According to the International Federation of Robotics (IFR), robots are predicted to contribute to achieving 13 sustainable development goals (out of the 17 set by the United Nations), including clean energy and climate action that are related to environmental governance.¹

IFR stated that the development of robot density in China has been the most dynamic worldwide, especially after 2013. In 2013, there was a historical inflection point of negative labor supply growth in China and subsequent explosive growth in robot applications, with a steep increase in robot installations, surpassing Japan as the world's largest consumer of industrial robots and accounting for more than one-fifth of global sales (Figure 2). As of 2021, China has been the world's top industrial robotics market for eight consecutive years.² Meanwhile, Chinese robot manufacturers reached a turning point in 2013. According to Cheng et al.,⁴ the number of registered companies producing or researching robots in China doubled annually from 2013 to 2015.

Figure 2.

Annual robot sales in China and the world.

Robotics development in China is also reflected at the policy level. As early as 2012, in the 12th Five-Year Plan, China included industrial robots as intelligent devices in its roadmap for the development of the intelligent manufacturing equipment industry. In 2013, China released its first policy document titled “Industrial Robotics” (Guidance on Promoting the Development of Industrial Robotics Industry), which elevated the development of the robotics industry to a national strategic level. Since then, industrial robotics development has been incorporated into the 13th and 14th Five-Year Plans, leading to China's digital development and intelligent upgrading. Whether in terms of practical applications or policy guidance, China's focus and determination on industrial robotics development and technological innovation are evident.

Notably, the role of industrial robots in overcoming air pollution, as a representative frontier technology in the booming city industrial system, has not been fully justified. Therefore, we examine the impact of industrial robots on cities' air pollution levels and explore the mechanisms and regional heterogeneity of their impact to explore city-level drivers of pollution reduction beyond pollution abatement policies. The contribution of this paper is in demonstrating the beneficial role of industrial robots in air pollution prevention and control at a Chinese city level, providing new ideas for the further promotion of the blue sky defense, and providing policymakers with suggestions that are in line with the trend of intelligence and the requirements of pollution reduction.

The innovations include a mechanism analysis from the perspective of automation and technology, which corroborates that industrial robots are the embodiment of green technology and can achieve synergistic benefits in reducing cities' air pollution levels by improving energy use efficiency and promoting green technology innovation. Another is a sample classification and heterogeneity analysis based on the relevance and policy value of cities' characteristics, taking into account the coupling and coordination of industrial robots with other environmental protection measures, which shows that industrial robots can be an effective alternative tool for green policies such as low-carbon piloting, resource planning, and environmental regulation.

Literature review

Since the introduction of the environmental Kuznets curve (EKC), there has been a proliferation of studies on how economic activities affect environmental pollution. Existing studies have analyzed the impact on environmental pollution and air quality in terms of economic growth,⁵ population density,^6,7 green policies,⁸ urbanization,^9,10 and industrial agglomeration.^11,12 Among them, studies on the correlation between industrial development and environmental pollution have attracted significant attention owing to the high pollution and high-emission nature of industrial activities. Panayotou found that by decomposing the EKC, poor countries with low industrialization emit less pollution and that there is a J-shaped curve relationship between SO₂ and industrial share when industrialization is in the range of 20%–40%, that is, SO₂ pollution emissions increase significantly with industrialization in this range.¹³ Li et al. found an inverted S-shaped curve relationship between industrialization and air pollution levels in China, which shows that increased industrialization in the low-level range exacerbates air pollution; however, when industrialization reaches a certain level, its continued increase can improve air quality, but continued industrialization at high levels leads to a serious rebound in air pollution.¹⁴ The improvement in air quality under moderate industrialization may be attributed to technological progress and environmental regulation; however, this level of technology or regulation is not sufficient to cope with the pollution generated by high industrialization. This indicates that there are environmental governance bottlenecks in highly industrialized countries or regions, and new solutions to mitigate pollution rebound and breakthrough governance bottlenecks should be sought urgently.

Technological innovation is an important intermediary between industrial activities and environmental pollution,⁵ can improve energy utilization by mitigating resource mismatch on the input side, and can improve productivity by optimizing production methods and upgrading manufacturing equipment on the output side, thus promoting the optimization and upgrading of industrial structure and the transformation of economic development.^15–17 Many studies have provided new solutions from the perspective of integrating informatization and industrialization. As a combination of informatization, industrialization, and urbanization, smart cities have been proven by many studies to reduce environmental pollution, including SO₂, PM2.5, and PM10.^17–19 For example, through a study based on Chinese city panel data from 2004 to 2018, Cui and Cao found that smart city construction significantly reduced PM2.5 concentrations in local and surrounding non-pilot cities, and its effect on local pollution control was achieved by promoting city innovation, industrial optimization, and upgrading, driving technological progress, and increasing total factor productivity.¹⁷ Information and communication technology (ICT), a key integrating technology in smart city construction, can significantly improve air quality in eastern China through information infrastructure improvement and energy efficiency improvement effects but is weak in central, western, and northeastern regions.²⁰

Admittedly, new technological innovations such as ICT are considered by many studies to be important ways to improve air quality and control environmental pollution. However, there is little literature on a systematic analysis of the impact of automation technologies on air quality. Theoretically, there are at least two mechanisms by which industrial robots impact air pollution: energy use efficiency and green technology innovation. The positive effects of energy use efficiency and green technology innovation on air pollution management have been confirmed by copious literature.^21–29

In contrast, the literature on the impact of industrial robots on both is incomplete and relatively scarce. The existing literature analyzes the impact of industrial automation on energy efficiency, mainly at the national, industrial, or firm level, and lacks empirical evidence at the city level. For example, based on cross-sectional panel data from the Chinese industry, Liu et al. showed that industrial robots can improve energy efficiency by increasing output and reducing energy consumption and that this effect is more pronounced in energy-intensive industries such as paper and steel.³⁰ Based on a cross-country panel of manufacturing firms, Wang et al. showed that industrial robots can improve energy efficiency through technological effects and that this positive effect is amplified in the era of Industry 4.0, but this is only evident for non-renewable energy sources.³¹ Based on a panel of industrial firms in China, Huang et al. showed that industrial robots can improve energy efficiency by increasing production rather than reducing energy consumption while emphasizing that this effect is insignificant for clean energy.³²

Moreover, the literature on the effects of industrial robots on green technology innovation is scarcer. Liu et al. confirmed that industrial robots promote technological innovation using cross-sectional panel data from the Chinese manufacturing industry,³³ and Yang et al. confirmed that intelligent manufacturing promotes green technology innovation using provincial panel data from the Chinese manufacturing industry.³⁴ In addition, based on cross-country panel data of the manufacturing industry, to the best of our knowledge, only Lee et al. provided empirical evidence that industrial robot applications significantly promote green technology innovation.³⁵

Among the existing empirical studies, the study by Luan et al.³⁶ is the closest to this study. They studied the impact of industrial robot stockpiles on air quality and climate change based on cross-country panel data from 1993 to 2019 and found that PM2.5 concentrations and other pollutant emissions increased with increasing industrial robot stockpiles, and the indirect promoting effect of energy consumption on PM2.5 emission accounted for up to 76.9%. This conclusion is opposite to the environmental management effect of technological innovation. Is an industrial robot, as a representative technology of automation, the driver of air degradation or the embodiment of green technology? The jury is still out.

In this paper, we provide conclusive and direct empirical evidence using data on robots and air pollution in Chinese cities. Additionally, based on the mechanistic routes shown in Figure 3, we explore the mechanisms underlying the effect of the application of industrial robots on cities' air pollution levels from two perspectives: energy use efficiency and green technology innovation.

Figure 3.

Mechanistic routes of industrial robotic application to affect air pollution.

Data and methodology

Data

The data in this study came from IFR's industrial robot dataset, the industrial enterprise module of China's Second National Economic Census (2008), China Stock Market and Accounting Research Database, China Meteorological Data Service Centre, Chinese Research Data Services, and government work reports from various prefecture-level cities. It should be noted that the IFR dataset contains information on robot installations and stocks in productive and non-productive industries in 50 countries since 1993 and is able to cover 90% of the global industrial intelligent robot market installations. From this, we selected industrial robot inventory data for each industry in China from 2013 to 2018 and matched it with industrial enterprise data from the second national economic census to calculate the per capita robot inventory for each city.

Regarding the data selection period, we considered the following: first, as mentioned in the introduction, 2013 was considered the first year of widespread application and vigorous development of industrial robots in China. Second, in 2013, China implemented the Ambient Air Quality Standards to carry out air quality index (AQI) monitoring, which comprehensively covers the six air pollutant indicators of SO₂, NO_x, PM10, PM2.5, O₃, and CO, and has been gradually extended to cover the entire country in subsequent years. Therefore, we selected 2013 as the starting year for the sample data. Third, after 2018, China started to fully implement the Three-Year Plan on Defending the Blue Sky, and a large number of actions were deployed nationwide for air pollution control, which may interfere with the results of this study. Therefore, we do not consider data after 2018.

Model

To examine the impact of industrial robots on cities' air pollution, the following econometric model was constructed:

L n Y_{i t} = α_{0} + α_{1} R o b o t_{i t} + \sum {β X}_{i t} + μ_{i} + v_{t} + ε_{i t}

(1)

where the subscript i represents the city and t is the year. The dependent variable

L n Y

includes PM2.5, PM10, and SO₂ pollutant concentrations in the city. The core independent variable

R o b o t_{i t}

denotes the per capita robot inventory of city i in year t, and its estimated coefficient

α_{1}

is the focus of this study. X is a series of control variables,

μ

is the city fixed effect, v is the year fixed effect, and

ε

is the random perturbation term.

Variables

City air pollution indicators (PM2.5, PM10, SO₂), the dependent variables. In existing studies on city pollution, the following two types of indicators were mainly used as proxy variables for city pollution levels. The first is the pollutant emission indicators of industrial enterprises,^16,18,37 because outdoor pollution mainly comes from industrial production processes; however, the enterprise pollution emission statistics are manipulable, and SO₂ and fumes in industrial pollutant emissions are not sufficient to represent the air pollution level of the entire city.¹⁹ The second type is the city six pollutants concentration index (AQI),³⁸ as such satellite monitoring data can more objectively reflect the actual level of city air pollution. Moreover, it is the same caliber as the AQG from the World Health Organization (WHO), which is more helpful in controlling the pollution governance process, where SO₂ and PM2.5 concentrations are more commonly used as metrics.^7,12,14 In addition, PM10, one of the causes of hazy weather, has a much higher annual average level in Chinese cities than the safety value set by WHO. Therefore, we selected PM2.5, PM10, and SO₂ concentrations as representative indicators and constructed a logarithmic treatment to comprehensively measure the air pollution levels in cities and ensure the reliability of the empirical results.

Per capita robot inventory at the city level (Robot), the independent variable. As the robot data provided by the IFR are based on the industry level, to calculate the robot coverage situation in each city, we drew on the idea of Bartik³⁹ and the common practice in the existing literature^40,41 to match the IFR dataset with industrial enterprise data from the Second National Economic Census (2008); thus, the per capita robot inventory was estimated for each city. The specific estimation method is as follows:

R o b o t_{i t} = \sum_{C = 1}^{C} \frac{e m p l o y_{c, i, t = 2008}}{e m p l o y_{i, t = 2008}} \times \frac{R o b o t_{c t}}{e m p l o y_{c, t = 2008}}

(2)

where C denotes the ensemble of industries,

R o b o t_{i t}

is the number of robot inventories in city i in year t,

R o b o t_{c t}

is the number of robot inventories in industry c in year t,

e m p l o y_{c, t = 2008}

is the total employment in industry c in 2008,

e m p l o y_{i, t = 2008}

is the total employment in city i in 2008, and

e m p l o y_{c, i, t = 2008}

is the total employment in industry i in city c in 2008. Additionally, we selected the following city-level control variables, considering that these factors may have an impact on city air quality.

Level of output per capita (PGDP)

According to the EKC hypothesis, the output per capita and environmental pollution are closely related.

Industrial structure (IND)

It is obvious that the share of the secondary industry reflects the air pollution level to some extent,⁵ but the boom of the tertiary industry also brings difficult air pollution problems, such as the transportation industry.³⁷ Therefore, we used the sum of the shares of the secondary and tertiary sectors to construct the city industrial structure variable.

Government scale (GOV)

Local governments are an important repository for implementing environmental protection policies. Bao and Liu argued that local government investment in air pollution control has increased, resulting in improving air quality.⁴² We used the ratio of government expenditure to GDP to measure the size of local governments.

Internet development status (Internet), taken as a logarithm. The application of Internet technology can effectively achieve environmental regulation, and the integrated development of the Internet and pollution control is an effective way for cities to achieve green transformation in the information era.⁴³ We used the logarithm of the number of international Internet users to measure the status of Internet development in cities.

The sunshine duration (Sunshine) and precipitation (Rainfall), taken as logarithms. Referring to Li et al.⁴⁴ and Borck and Schrauth,⁶ we included two control variables, sunshine duration, and precipitation, considering the effects of geographical conditions and meteorological variables. Statistical information on the main variables is presented in Table 1.

Table 1.

Statistical information on the variables.

Variables	Unit	N	Mean	SD	p25	p50	p75
PM2.5	μg/m³	1541	54.725	27.357	36.333	48.55	64.667
PM10	μg/m³	1541	94.332	40.958	64.583	85.5	113.625
SO₂	μg/m³	1541	28.216	22.786	13.333	20.825	34.125
Robot	per 1000 people	1541	2.246	2.442	0.708	1.454	2.953
PGDP	10⁴ CNY	1541	5.592	3.582	3.144	4.621	6.74
IND	%	1541	88.514	7.353	84.19	89.435	94.14
GOV	%	1541	20.078	10.117	13.393	17.416	23.534
Ln(Internet)	households	1541	13.501	0.862	12.948	13.43	14.062
Ln(Sunshine)	hours	1541	7.571	0.208	7.417	7.572	7.74
Ln(Rainfall)	millimeters	1541	9.16	0.485	8.764	9.181	9.568

Results

Baseline regression results

The baseline regression results in Table 2 show the effects of industrial robot applications on the concentrations of the three air pollutants. Overall, we found that the estimated coefficients of robot inventory per capita for all three air pollutant concentrations are negative at the 1% significance level after controlling for the city- and year-fixed effects. This suggests that industrial robot applications can significantly reduce the overall city air pollution levels while controlling for other variables. Specifically, an increase of one robot inventory per 1000 people reduced PM2.5, PM10, and SO₂ concentrations in city air by 2.8%, 1.97%, and 2.68%, respectively. This suggests that fine PM (PM2.5) and exhaust gases (SO₂) are more affected by industrial robotics than respirable particles (PM10), which may be related to the pollutant sources—PM2.5 and SO₂ mainly originate from industrial production, daily power generation, and vehicle exhaust emissions; therefore, the emission intensity is more impacted by industrial robotics, while PM above 2.5 µm in PM10 mainly comes from direct emissions from pollution sources, such as wind-induced dust, chimneys, and vehicles. This type of PM outside the industrial system weakens the pollution reduction effect of the industrial robots. In contrast to existing studies, this study provides empirical evidence that industrial robots affect city-level air pollution levels in China.

Table 2.

Baseline regression results.

	(1)	(2)	(3)
	Ln(PM2.5)	Ln(PM10)	Ln(SO₂)
Robot	−0.0280***	−0.0197***	−0.0268***
	(0.0082)	(0.0058)	(0.0082)
PGDP	−0.0359**	−0.0247**	−0.0034
	(0.0169)	(0.0121)	(0.0170)
PGDP_2	0.0009**	0.0005**	0.0000
	(0.0004)	(0.0003)	(0.0004)
IND	0.0107	−0.0017	−0.0170*
	(0.0087)	(0.0063)	(0.0088)
GOV	−0.0076**	−0.0030	−0.0084**
	(0.0035)	(0.0025)	(0.0035)
Ln(Internet)	−0.0938***	−0.0691***	−0.0785**
	(0.0352)	(0.0251)	(0.0353)
Ln(Sunshine)	0.7992***	0.5170***	−0.2266
	(0.2092)	(0.1494)	(0.2099)
Ln(Rainfall)	0.2835***	0.2060***	−0.0324
	(0.0878)	(0.0627)	(0.0881)
FE CITY	Yes	Yes	Yes
FE YEAR	Yes	Yes	Yes
Cons	−4.1830*	−0.1408	7.5063***
	(2.2402)	(1.6004)	(2.2480)
N	1541	1541	1541
R²	0.3775	0.5311	0.6882

Note: Standard errors in parentheses, *p < 0.1, **p < 0.05, ***p < 0.01.

Noteworthy, the estimated results of the control variables are consistent with theoretical expectations or existing research findings. The significant reductions in PM concentrations were 3.59% (PM2.5) and 2.47% (PM10) for every 10,000 Yuan increase in per capita output, indicating that China is at the right side of the “inflection point” of the EKC model, that is, the stage where per capita output is negatively correlated with environmental pollution. This indicates that China's green transformation has begun to bear fruit in recent years. For every 1% positive change in industrial structure, SO₂ concentration was significantly reduced by 1.7% in the cities. This is consistent with Stern's conclusion that industrial structure is the main factor influencing SO₂ emissions.⁴⁵ Moreover, a possible explanation for the increase in the share of secondary and tertiary industries leading to a decrease in SO₂ pollutant emissions is that industrial restructuring brings about environmental spillover effects that alter the impact of economic development on air pollution.⁴⁶ Each 1% increase in government size significantly reduced PM2.5 and SO₂ by 0.76% and 0. 84%, respectively, which is consistent with Halkos and Paizanos, who concluded that government spending is significantly and negatively associated with SO₂ emissions per capita.⁴⁷

For every 1% increase in the number of international Internet users, the PM2.5, PM10, and SO₂ concentrations were significantly reduced by 0.0938%, 0.0691%, and 0.0785%, respectively. This indicates that Internet development helps reduce cities' air pollutant concentrations, and Internet application is an effective measure for optimizing cities' air quality. This is in line with Yang et al.'s finding that with the widespread use of the Internet; its impact on air pollution breaks through the tipping point and exhibits an environment-friendly nature.⁴³

Robustness tests

Instrumental variable estimation

We calculated the robot penetration in China by the major import source countries of robots (America, Japan, Germany, Sweden, and Korea) as an instrumental variable, drawing on Acemoglu and Restrepo.⁴⁰ Table 3 shows the results of the instrumental variables estimation. The estimated coefficient from the first stage shows that the coefficient of the instrumental variable is positive (0.1862) and significant at the 1% level, which is consistent with theoretical expectations. The weak instrumental variable test reveals that the F-statistic is 27, which exceeds the critical value of 10, and excludes the possibility that the selected instrumental variable is a weak instrumental variable, meaning that the instrumental variable selected in this study has excellent explanatory power for the original independent variable. As shown by the second-stage estimation, the estimated coefficients of robot inventory per capita are all significantly negative, further confirming that industrial robot applications can reduce cities' air pollution.

Table 3.

Instrumental variable estimation results.

	(1)	(2)	(3)
	Ln(PM2.5)	Ln(PM10)	Ln(SO₂)
Robot	−0.1878***	−0.1218***	−0.0929***
	(0.0461)	(0.0333)	(0.0330)
Control Variables	Yes	Yes	Yes
FE CITY&YEAR	Yes	Yes	Yes
	Stage 1
IV: Robots of the import source countries	0.1862***
	(0.0352)
Cragg-Donald Wald F	72.605
Kleibergen-Paap rk Wald F	27.954
N	1541	1541	1541
Centered R ²	−0.2559	−0.2037	−0.0323

The control variables are the same as in the baseline regression.

Standard errors in parentheses, *p < 0.1, **p < 0.05, ***p < 0.01.

Plausibly exogenous test of the instrumental variable

Considering that the instrumental variable used is more likely to be only plausibly exogenous rather than strict, the test was based on the Union of Confidence Intervals (UCI) proposed by Conley et al.⁴⁸ The results in Table 4 show that the estimated coefficients of “Robot” are relatively stable and remain robustly negative at high significance levels, even when the range of δ values is changed under the assumption that strict exogeneity of the instrumental variable is violated. Thus, the approximate exogeneity of “Robots of the import source countries” has little effect on the estimation results and is a reasonable instrumental variable.

Table 4.

The robust confidence intervals of “Robot” based on Conley et al. (2012).

The range of $δ$	Ln(PM2.5)		Ln(PM10)		Ln(SO₂)
The range of $δ$	Lower	Upper	Lower	Upper	Lower	Upper
$δ \in (- 0.1, 0)$	[−0.2581	−0.0687]	[−0.1708	−0.0230]	[−0.1562	0.0225]
$δ \in (0, 0.1)$	[−0.3180	−0.1176]	[−0.2305	−0.0728]	[−0.2133	−0.0296]

Control of shocks from parallel policies

Considering the impact of pollutant control policies in recent years, the parallel policies of the Action Plan for Prevention and Control of Air Pollution (APPCAP),³ the pollutant emissions trading pilot (PET),⁴ and the comprehensive demonstration work on fiscal policies for energy conservation and emission reduction (FP) were controlled here. The results in the first three columns of Table 5 demonstrate that the pollution reduction effect of industrial robots still has significant importance and is even further amplified after controlling for three major parallel policy shocks. This indicates that industrial robot applications are well coupled with pollutant emission reduction policies and consistent with the policy trend of air pollution prevention and control governance. Moreover, APPCAP and PET have a significant and strong reduction effect on PM concentrations, confirming the effectiveness of China's major pollution reduction policies to improve air quality.

Table 5.

Robustness tests.

	(1)	(2)	(3)	(4)	(5)	(6)
	Ln(PM2.5)	Ln(PM10)	Ln(SO₂)	Ln(PM2.5)	Ln(PM10)	Ln(SO₂)
Robot	−0.0299***	−0.0199***	−0.0282***
Robot	(0.0081)	(0.0058)	(0.0083)
Robot_lag				−0.0328***	−0.0242***	−0.0331***
				(0.0101)	(0.0072)	(0.0101)
APPCAP	−0.1894***	−0.0727**	−0.0664
	(0.0465)	(0.0336)	(0.0474)
PET	−0.3110***	−0.1534***	0.0206
	(0.0605)	(0.0437)	(0.0618)
FP	−0.0841	0.0865	0.0150
	(0.0741)	(0.0535)	(0.0756)
Control	Yes	Yes	Yes	Yes	Yes	Yes
Variables	Yes	Yes	Yes	Yes	Yes	Yes
FE CITY	Yes	Yes	Yes	Yes	Yes	Yes
FE YEAR	Yes	Yes	Yes	Yes	Yes	Yes
N	1541	1541	1541	1541	1541	1541
R²	0.3987	0.5380	0.6888	0.3770	0.5311	0.6882

The control variables are the same as in the baseline regression.

Standard errors in parentheses, *p < 0.1, **p < 0.05, ***p < 0.01.

Adoption of the lagged one-period term of the independent variable

To avoid the potential risk of endogeneity caused by the correlation between the current period independent variable and the current period residual term, the lagged one-period term of the per capita robot inventory was used to replace its current period term. The estimation results in columns (4)–(6) of Table 5 show that there is still a significant negative correlation between the per capita robot inventory and the concentrations of the three air pollutants, with this negative effect somewhat larger than that of the baseline model, further confirming the core findings of this paper.

Mechanism exploration

Industrial robots, as the representative technology for automation, inherently possess the characteristics of efficiency and technology. Therefore, we selected two proxy variables, energy use efficiency, and green technology innovation to identify mechanisms for industrial robot application to reduce cities' air pollutant concentrations. The estimated results of the mechanism analysis are reported in Table 6.

Table 6.

Mechanism exploration.

	(1)	(2)
	Energy use efficiency	Green technology innovation
Robot	0.0676***	0.0189*
	(0.0105)	(0.0112)
Control Variables	Yes	Yes
FE CITY&YEAR	Yes	Yes
N	1514	1515
R²	0.5841	0.5929

The control variables are the same as in the baseline regression.

Standard errors in parentheses, *p < 0.1, **p < 0.05, ***p < 0.01.

Energy use efficiency

Patterson defined the ratio of useful output to energy input in production as energy efficiency.⁵² With reference to this, we used the logarithmic form of the ratio of gross product to electricity consumption to measure the energy use efficiency of cities, that is, Ln(GDP/Electricity), and used this as the dependent variable to test the effect of per capita robot inventory on energy use efficiency, as shown in column (1) of Table 6. An increase in one robot inventory per 1000 people results in a significant increase of 6.76% in energy use efficiency, thus confirming the consistent finding from the existing literature at the city level.^30–32 This suggests that the increase in city output from industrial robot applications far outweighs their energy consumption and can improve energy use efficiency, leading to a synergistic effect of reducing air pollution.^21,22

Green technology innovation

Although the academic community has not yet reached a consensus on the definition of green technology, it cannot be denied that green technology has the property of promoting green development and building an ecological civilization, and green technology innovation has environmental benefits.²⁸ With advances in information technology, the utilization, and reliability of patent statistics as proxy indicators of innovation activity have increased.⁵³ In view of this, we used the logarithmic form of the number of green invention patents filed by cities in the current period to measure the level of green technology innovation in cities and used this as the dependent variable to test the impact of the robot inventory per capita on green technology innovation, as shown in column (2) of Table 6. An increase of one robot inventory per 1000 people will significantly increase the level of green technology innovation by 1.89%, which is consistent with Lee et al.'s finding.³⁵ This shows that industrial robot is not only a representative technology for automation but also the embodiment of green technology, which can effectively improve the level of green technology innovation and thus systematically reduce air pollutant emissions.²⁹

Heterogeneity analysis

Theoretically, the impact of robots' intervention in cities' industrial systems on local air pollution levels will vary across cities due to differences in industrial development, natural endowments, and government management. Therefore, a heterogeneity analysis was conducted here for cities with different characteristics to confirm which characteristics are beneficial for the air pollution reduction effect of industrial robots. The estimated results are shown in Table 7.

Table 7.

Heterogeneity analysis.

	(1)	(2)	(3)	(4)	(5)	(6)
	Ln(PM2.5)	Ln(PM10)	Ln(SO₂)	Ln(PM2.5)	Ln(PM10)	Ln(SO₂)
	Low Carbon Pilot Cities			Non-Low Carbon Pilot Cities
Robot	−0.0071	0.0009	−0.0135	−0.0408***	−0.0260**	−0.0300**
	(0.0082)	(0.0066)	(0.0100)	(0.0150)	(0.0103)	(0.0141)
N	669	669	669	872	872	872
R²	0.6563	0.7164	0.7971	0.2475	0.4257	0.6012
	Resource-based Cities			Non-resource-based cities
Robot	0.0400	−0.0078	−0.0084	−0.0179**	−0.0097^#	−0.0216**
	(0.0310)	(0.0258)	(0.0346)	(0.0088)	(0.0061)	(0.0088)
N	559	559	559	982	982	982
R²	0.1517	0.1803	0.4845	0.5232	0.6686	0.7683
	Strong environmental regulation			Weak environmental regulation
Robot	−0.0093	−0.0086	−0.0273**	−0.0515***	−0.0278***	−0.0360**
	(0.0098)	(0.0083)	(0.0116)	(0.0160)	(0.0102)	(0.0141)
N	758	758	758	767	767	767
R²	0.4707	0.5010	0.7029	0.3205	0.5536	0.6729

The control variables are the same as in the baseline regression and have been controlled as well.

Standard errors in parentheses, #p < 0.12, *p < 0.1, **p < 0.05, ***p < 0.01.

Low-carbon pilot

We divided the sample into low-carbon pilot cities and non-low-carbon pilot cities based on whether the cities were included in the low-carbon city pilot list. The subsample estimation results show that industrial robots exhibit significant air pollution reduction effects only in the non-low-carbon pilot cities, which are even more pronounced than in the full sample cities, and it appears as if the low-carbon pilot cities in the full sample offset the pollution reduction effects of industrial robot applications.

Resource endowment type

We divided the sample into resource-based cities and non-resource-based cities based on whether the cities were included in the National Sustainable Development Plan for Resource-based Cities (2013–2020). The subsample estimation results show that industrial robots exhibit a significant air pollution reduction effect only in non-resource-based cities, which appear to be using technology to compensate for their lack of resources.

Environmental regulation intensity

Generally, the multidimensional nature of environmental regulation makes it difficult to quantify, and academics have not yet reached a consensus on measuring the level of environmental regulation. However, as environmental regulation has currently been implemented in China mainly through the issuance of relevant policy documents, local government work reports tend to reflect the willingness, direction, and strength of local environmental governance to a certain extent. Therefore, we referred to Chen et al. to measure the intensity of city environmental regulation in terms of the proportion of relevant environmental vocabulary to the total vocabulary of local government work reports and used a machine learning approach to judge the sample into cities with strong environmental regulations and cities with weak environmental regulations.⁵⁴ The subsample estimation results show that industrial robots exhibit a significant and comprehensive air pollution reduction in cities with weak environmental regulations, whereas they only significantly reduce SO₂ concentrations in cities with strong environmental regulations.

Discussion

Addressing climate change and combating air pollution are coupled and coordinated

Pollution abatement and carbon reduction are often considered to have co-benefits due to the large overlap between the sources of air pollutants and carbon dioxide.^26,55 Our findings validate this view, suggesting that low-carbon pilot policies can achieve synergistic management of climate change and air pollution. Since low-carbon pilot cities improved air quality while exploring carbon reduction pathways,⁵⁶ the reduction effect of air pollutant concentrations by industrial robots is not significant. In other words, in non-low-carbon pilot cities, industrial robotics can exert pollution reduction effects comparable to those of low-carbon pilot policies.

Resource-based cities rely on inherent resources, while non-resource-based cities rely on technological advances

Resource-based cities are cities with mining and processing of local natural resources as the leading industry, and as important strategic guarantee bases for energy resources, their sustainable development is crucial for promoting new industrialization and urbanization processes and ecological civilization construction. In 2013, the State Council issued the National Sustainable Development Plan for Resource-based Cities (2013–2020), which planned resource-based cities, including 126 prefecture-level administrative regions (prefecture-level cities, regions, autonomous regions, leagues, etc.).

The heterogeneity analysis shows that the intensity of response to introducing industrial robotics in resource-based cities differs from that in non-resource-based cities under the influence of resource endowment and governmental planned development. Specifically, resource-based cities lack the incentive to pursue technological development because of the relatively low energy use costs associated with abundant natural resources. In contrast, non-resource-based cities rely more on technology due to energy cost pressures, and robots are more deeply integrated into the industrial systems of such cities, thus significantly reducing pollution.

Industrial robot applications can replace environmental regulations for air pollution control

Environmental regulation in China is mainly conducted through the issuance of policy documents to regulate the polluting behavior of enterprises. However, “imperative” environmental regulation may lead to the curtailment of productive investments and innovation expenditures by enterprises,⁵⁷ or it may force firms to upgrade their production technology in order to seek development.⁵⁸

The findings from the heterogeneity analysis in this paper similarly suggest that the intensity of environmental regulation affects the pollution reduction effect of industrial robotics applications in cities' industrial systems. Specifically, in cities with strong environmental regulation, firms' pollution emission behavior is well controlled, and thus the environmental gain of introducing industrial robotics is not significant. In contrast, in cities with weak environmental regulation, industrial robotics can be used as an alternative tool to environmental policies to effectively reduce air pollutant concentrations and help restore unpolluted skies.

The above discussion reveals an interesting phenomenon: in cities where green policies such as low-carbon pilots, resource planning, and environmental regulation are not or partially covered, the pollution reduction effect of industrial robot applications is undoubtedly evident and even magnified in most cases. Therefore, it is reasonable to assume that cities with imperfect green policies can use technology applications such as industrial robots to achieve air pollution control under the pressure of pollution reduction requirements.

Conclusions and policy implications

China's average air pollution levels are still higher than international safety values; haze remains above cities, and air pollution prevention and control should be continuously promoted. In the process of integration into the city industrial system, industrial robots may be able to bring synergistic benefits in controlling pollution. However, there is a lack of research on the effects of industrial robots on city air pollution levels. This study examines the relationship between the two and argues for the mechanism of action of industrial robots. First, we constructed a per capita robot inventory indicator at the city level in China using the IFR dataset and the National Economic Census China (2008) industrial enterprise panel data, and then analyzed its effects and mechanisms on city air pollutant concentrations. The study finds that, first, industrial robots are a booster army in the Blue Sky Defense War, and industrial robot applications significantly reduce city air pollutant concentrations. The conclusions still hold after controlling for various factors that may affect city pollutant concentrations and a series of robustness tests. Second, industrial robots are the embodiment of green technology, and a mechanism analysis shows that industrial robots can achieve synergistic benefits in reducing cities' air pollution by improving energy use efficiency and promoting green technology innovation. Third, city heterogeneity exists in the pollution reduction effect of industrial robots, and the heterogeneity analysis shows that industrial robots reduce air pollution more prominently in non-low-carbon pilot cities, non-resource-based cities, and cities with weak environmental regulations, suggesting that industrial robots can be used as an alternative to green policies to reduce cities' air pollution.

Based on the Chinese city level, this study examines the importance of industrial robots for controlling air pollution, thereby enriching research related to industrial robots and air pollution. The findings of this study echo the national sustainable development goals of promoting new industrialization and urbanization and establishing an environment-friendly society at the policy level. This study presents several policy implications as follows.

First, policy support is critical. The implementation of extremely effective pollution reduction policies should first be ensured, and then the broader coverage of these policies should be actively and consistently promoted. Comprehensive policy support, coupled with the introduction of industrial robotics, is the two-pronged approach that can help a city's industrial system adapt more quickly to national trends in air pollution prevention and control.

Second, enhancement of the scale of application and the degree of intervention of industrial robots in the city industrial system, strengthening the introduction of fully automated production lines and the operation of closed production workshops to achieve full collection and treatment of pollutants after discharge, and ultimately, reducing the emission of air pollutants and secondary pollution from the source of industrial activities.

Third, encouraging enterprises to target the introduction of industrial robots with different degrees of automation, such as human–robot collaboration robots that are easy to operate and configure to help more enterprises unlock automation applications and make full use of the energy efficiency improvement effect of automated production, while driving the synergistic development of independent R&D and green technology innovation of industrial robots in China.

Finally, optimize the development path of industrial transformation and upgrading by focusing on intelligent technology application and green process transformation, especially in resource-based cities with huge space for technology introduction, cities lacking policy pilots, and cities with weak environmental regulations, and make full use of the high marginal pollution reduction effect of industrial robots in such cities to maximize the pollution reduction value of industrial robot applications.

The primary limitation of this study is that it mainly analyzed the impact of industrial robots on three major haze-causing substances (PM2.5, PM10, and SO₂) and did not investigate the impact on other air pollutants (CO, O₃, and NO_x). Therefore, a more comprehensive examination of the impact of industrial robots on air pollution can be considered in future studies.

Footnotes

Availability of data and material

City-level data on robot inventory for this paper is not publicly available but can be collected from the publicly available industrial robot dataset of the International Federation of Robotics. Besides, raw data for the other variables at the city level are obtained from the China Stock Market and Accounting Research Database (CSMAR), the China Meteorological Data Service Centre, Chinese Research Data Services (CNRDS), and government work reports for each prefecture-level city. All of the above data sources have been explicitly identified in the main text.

Authors’ contribution

Lingzheng Yu: Conceptualization, Writing – review & editing. Chenyu Zeng: Writing – original draft. Xiahai Wei: Methodology, Formal analysis.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The paper is supported by the National Social Science Fund of China (No: 22BJL075).

ORCID iD

Chenyu Zeng

Notes

Author biographies

Lingzheng Yu, PhD, is an Associate Professor at the School of Economics and Finance, Huaqiao University, Quanzhou, Fujian, 362021, China.

Chenyu Zeng is a postgraduate student at the School of Economics and Finance, Huaqiao University, Quanzhou, Fujian, 362021, China.

Xiahai Wei, PhD, is a Professor at the Institute of Economic Development and Reform, Huaqiao University, Xiamen, Fujian 361021, China

References

Shang

, et al. Resonant suppression method based on PI control for serial manipulator servo drive system. Sci Prog 2020; 103: 0036850420950130.

Mei

Zhang

Zang

, et al. Trajectory optimization of the 6-degrees-of-freedom high-speed parallel robot based on B-spline curve. Sci Prog 2020; 103: 0036850419880115.

Yang

Yan

. Comparison of deep learning-based methods in multimodal anomaly detection: a case study in human–robot collaboration. Sci Prog 2021; 104: 00368504211021192.

Cheng

Jia

, et al. The rise of robots in China. J Econ Perspect 2019; 33: 71–88.

Guo M, Li H, Lin W. The impact of economic growth, FDI, and innovation on environmental efficiency of the logistics industry in provinces along the belt and road in China: An empirical study based on the panel Tobit model. Sci Prog 2021; 104(2): 00368504211018054.

Borck

Schrauth

. Population density and urban air quality. Reg Sci Urban Econ 2021; 86: 103596.

Chen

Wang

Huang

, et al. The influence of increased population density in China on air pollution. Sci Total Environ 2020; 735: 139456.

Zhou

Zhong

Chen

, et al. Can the joint regional air pollution control policy achieve a win-win outcome for the environment and economy? Evidence from China. Econ Anal Policy 2022; 74: 13–33.

Liu

Cui

Zhang

. Exploring the causal relationship between urbanization and air pollution: evidence from China. Sustain Cities Soc 2022; 80: 103783.

10.

Zhao

Wang

. How does new-type urbanization affect air pollution? Empirical evidence based on spatial spillover effect and spatial Durbin model. Environ Int 2022; 165: 107304.

11.

. Can industrial collaborative agglomeration reduce haze pollution? City-level empirical evidence from China. Int J Environ Res Public Health 2021; 18: 1566.

12.

Chen

Zhu

Cheng

. Industrial agglomeration and haze pollution: evidence from China. Sci Total Environ 2022; 845: 157392.

13.

Panayotou

. Demystifying the environmental Kuznets curve: turning a black box into a policy tool. Environ Dev Econ 1997; 2: 465–484.

14.

, et al. Mining of the association rules between industrialization level and air quality to inform high-quality development in China. J Environ Manage 2019; 246: 564–574.

15.

Pham

Huynh

TLD

Nasir

. Environmental consequences of population, affluence and technological progress for European countries: a malthusian view. J Environ Manage 2020; 260: 110143.

16.

Chen

Wang

. The impact of technological innovation on air pollution: firm-level evidence from China. Technol Forecast Soc Change 2022; 177: 121521.

17.

Cui H-yu, Cao Y-qun. Do smart cities have lower particulate matter 2.5 (PM2.5)? Evidence from China. Sustain Cities Soc 2022; 86: 104082.

18.

Chu

Cheng

. A smart city is a less polluted city. Technol Forecast Soc Change 2021; 172: 121037.

19.

Gao

Yuan

. Is the sky of smart city bluer? Evidence from satellite monitoring data. J Environ Manage 2022; 317: 115483.

20.

Qiao

Fei

. Information infrastructure and air pollution: empirical analysis based on data from Chinese cities. Econ Anal Policy 2022; 73: 563–573.

21.

Zhang

Worrell

Crijns-Graus

, et al. Co-benefits of energy efficiency improvement and air pollution abatement in the Chinese iron and steel industry. Energy 2014; 78: 333–345.

22.

Zhang

Worrell

Crijns-Graus

. Cutting air pollution by improving energy efficiency of China’s cement industry. Energy Procedia 2015; 83: 10–20.

23.

Muhammad

Pan

Agha

, et al. Industrial structure, energy intensity and environmental efficiency across developed and developing economies: the intermediary role of primary, secondary and tertiary industry. Energy 2022; 247: 123576.

24.

Taskin

Dogan

Madaleno

. Analyzing the relationship between energy efficiency and environmental and financial variables: a way towards sustainable development. Energy 2022; 252: 124045.

25.

Feng

Wang

Liang

. How does environmental information disclosure affect economic development and haze pollution in Chinese cities? The mediating role of green technology innovation. Sci Total Environ 2021; 775: 145811.

26.

Gao

Liu

Hua

. Environmental protection tax law on the synergy of pollution reduction and carbon reduction in China: evidence from a panel data of 107 cities. Sustain Prod Consum 2022; 33: 425–437.

27.

Liao

. Spatial division of labor, specialization of green technology innovation process and urban coordinated green development: evidence from China. Sustain Cities Soc 2022; 80: 103778.

28.

Fan

. Renewable energy technology innovation, industrial structure upgrading and green development from the perspective of China’s provinces. Technol Forecast Soc Change 2022; 180: 121727.

29.

Zhou

Bai

Tian

. Study on the impact of electric power and thermal power industry of Beijing–Tianjin–Hebei region on industrial sulfur dioxide emissions—from the perspective of green technology innovation. Energy Rep 2022; 8: 837–849.

30.

Liu

Yang

Fujii

, et al. Artificial intelligence and energy intensity in China’s industrial sector: effect and transmission channel. Econ Anal Policy 2021; 70: 276–293.

31.

Wang

Lee

. Assessing the impact of industrial robots on manufacturing energy intensity in 38 countries. Energy Econ 2022; 105: 105748.

32.

Huang

Lin

. Robot adoption and energy performance: evidence from Chinese industrial firms. Energy Econ 2022; 107: 105837.

33.

Liu

Chang

Forrest

JYL

, et al. Influence of artificial intelligence on technological innovation: evidence from the panel data of China’s manufacturing sectors. Technol Forecast Soc Change 2020; 158: 120142.

34.

Yang

Liu

. The effect of manufacturing intelligence on green innovation performance in China. Technol Forecast Soc Change 2022; 178: 121569.

35.

Lee

Qin

. Does industrial robot application promote green technology innovation in the manufacturing industry? Technol Forecast Soc Change 2022; 183: 121893.

36.

Luan

Yang

Chen

, et al. Industrial robots and air environment: a moderated mediation model of population density and energy consumption. Sustain Prod Consum 2022; 30: 870–888.

37.

Huang

Zhang

, et al. Impact of transportation infrastructure on industrial pollution in Chinese cities: a spatial econometric analysis. Energy Econ 2020; 92: 104973.

38.

Song

Liu

, et al. The effect of environmental regulation on air quality in China: a natural experiment during the COVID-19 pandemic. Atmos Pollut Res 2021; 12: 21–30.

39.

Bartik

. Who Benefits from State and Local Economic Development Policies? Michigan, USA: W.E. Upjohn Institute, 1991. Accessed August 2, 2022. https://www.jstor.org/stable/j.ctvh4zh1q

40.

Acemoglu

Restrepo

. Robots and jobs: evidence from US labor markets. J Polit Econ 2020; 128: 2188–2244.

41.

Goldsmith-Pinkham

Sorkin

Swift

. Bartik instruments: what, when, why, and how. Am Econ Rev 2020; 110: 2586–2624.

42.

Bao

Liu

. How does government attention matter in air pollution control? Evidence from government annual reports. Resour Conserv Recycl 2022; 185: 106435.

43.

Yang

Ren

, et al. Does the development of the internet contribute to air pollution control in China? Mechanism discussion and empirical test. Struct Chang Econ Dyn 2021; 56: 207–224.

44.

Zheng

, et al. The new smart city programme: evaluating the effect of the internet of energy on air quality in China. Sci Total Environ 2020; 714: 136380.

45.

Stern

. Explaining changes in global sulfur emissions: an econometric decomposition approach. Ecol Econ 2002; 42: 201–220.

46.

Zheng

Peng

Xiao

, et al. Industrial structure transformation and provincial heterogeneity characteristics evolution of air pollution: evidence of a threshold effect from China. Atmos Pollut Res 2020; 11: 598–609.

47.

Halkos

Paizanos

EΑ

. The effect of government expenditure on the environment: an empirical investigation. Ecol Econ 2013; 91: 48–56.

48.

Conley

Hansen

Rossi

. Plausibly exogenous. Rev Econ Stat 2012; 94: 260–272.

49.

Dai

Wei

, et al. Air pollution prevention and control action plan substantially reduced PM2.5 concentration in China. Energy Econ 2022; 113: 106206.

50.

Tao

Zhu

. Identifying FDI spillovers. J Int Econ 2017; 107: 75–90.

51.

Zeng

. Re-examine the air pollution control effect of China’s pollutant emissions trading system: evidence from a multi-period causal inference approach. J Clean Prod 2022; 371: 133467.

52.

Patterson

. What is energy efficiency? Energy Policy 1996; 24: 377–390.

53.

Pavitt

. Patent statistics as indicators of innovative activities: possibilities and problems. Scientometrics 1985; 7: 77–99.

54.

Chen

Kahn

Liu

, et al. The consequences of spatially differentiated water pollution regulation in China. J Environ Econ Manage 2018; 88: 468–485.

55.

Dong

Xia

Fang

, et al. Effect of the carbon emissions trading policy on the co-benefits of carbon emissions reduction and air pollution control. Energy Policy 2022; 165: 112998.

56.

Liu

Song

, et al. An integrated approach to evaluating the coupling coordination degree between low-carbon development and air quality in Chinese cities. Adv Clim Chang Res 2021; 12: 710–722.

57.

Qiao

Zhao

Guo

, et al. Factor price distortions, environmental regulation and innovation efficiency: an empirical study on China’s power enterprises. Energy Policy 2022; 164: 112887.

58.

Cai

Zhu

Zhang

, et al. Can direct environmental regulation promote green technology innovation in heavily polluting industries? Evidence from Chinese listed companies. Sci Total Environ 2020; 746: 140810.