ETFP is also called green total factor productivity, eco-efficiency or green development efficiency, it measured the efficiency of production, which refers to generating more output with less input (capital, labor, energy) while emitting less pollution. The literature on ETFP mainly includes two aspects, namely the calculation of ETFP and the analysis of ETFP influencing factors. The literature for measuring ETFP mainly uses parametric and non-parametric methods, and calculates ETFP from the provincial level or the enterprise level. For example, Chen et al. (2016) used the analytic hierarchy process (AHP) and constructed an evaluation index system for China’s industrial green development. Feng et al. (2017) used a hybrid model and window analysis to measure the efficiency of China’s regional industrial green transformation. Li et al. (2016) used data envelopment analysis (DEA) to evaluate the sustainable development of resource-based cities. Cheng and Li (2018) used non-directional distance function (NDDF) and the meta-Frontier method to measure the green growth of China’s manufacturing industry. Hu et al. (2019) used ultra-relaxation Super-Slack Based Measure (SBM) function and the Global Malmquist-Luenberger (GML) index to calculate the environment total factor productivity of China’s manufacturing.

Other document mainly studies the constituent elements and influencing factors of ETFP. The literature related to this research mainly focuses on the impact of input-output factors, technological progress, and government regulations on ETFP. The literature’s research conclusions on the influence of input-output factors are relatively consistent, and they believed that factors such as resources and environment have a significant impact on ETFP. For example, excessive use of energy and excessive emissions of SO₂, CO₂, and COD are the main sources of environmental inefficiency (He et al., 2016), and the regional economic development level and ETFP show a strong positive correlation (Zhang and Ye, 2015). He et al. (2016) also found through specific calculations that provinces with higher environmental efficiency are concentrated in the eastern economically developed regions, while the overall ETFP growth level in the central and western regions is relatively low. Xia and Xu (2020) found through further research that the decrease in the output value of the secondary industry with high energy consumption and high pollution and the increase in the output value of the tertiary industry can effectively promote the growth of ETFP.

In terms of the impact of technological progress, Liu and Xin (2019) used provinces in the “One Belt, One Road” strategy as the research object and found that technological progress is the main reason for the increase in ETFP. Yan et al. (2020) further studied the technological innovation level of renewable energy at the provincial level and found that only after the urban economy has developed to a certain stage, technological progress can significantly promote the growth of ETFP. Feng et al. (2018) conducted a detailed analysis of China’s metallurgical industry and found that technological progress is the most important factor in promoting the growth of ETFP. Wang et al. (2020) further studied 34 industrial sectors in China and found that technological innovation is the main driving factor for the growth of ETFP. Part of the literature also analyzed from the perspective of enterprise R and D investment. For example, Chen and Golley (2014) conducted a study on 38 industrial sectors in China and found that R and D investment intensity has a significant role in promoting ETFP, while Xia and Xu (2020) found through further research that the higher the proportion of R and D investment in GDP, the more it can promote the growth of ETFP.

The literature on the research conclusions on the effect of government intervention is quite divergent. He et al. (2016) believe that the government’s environmental management capacity and ETFP show a negative correlation; Zhang and Jiang (2019) found through empirical research that the ETFP of key environmental protection cities in China is lower than that of non-key protection cities, while the Porter Effect of government environmental regulations still exists. Compared with “command-and-control” environmental regulatory policies, market-based environmental regulatory policies can exert a greater Porter Effect and have a greater role in promoting ETFP (Xie et al., 2017). Wang et al. (2019) found through further research that government environmental regulation can significantly promote the growth of ETFP under appropriate intensity, but excessive environmental regulation will offset the innovation effect, thereby inhibiting ETFP. Liu et al. (2018) also found an inverted U-shaped relationship between the intensity of government environmental regulations and environmental efficiency. Jin et al. (2020) and Song et al. (2018) found an inverted U-shaped relationship in the relationship between the intensity of intergovernmental competition, fiscal decentralization, and ETFP.

In the literature related to the influence of industrial agglomeration on ETFP, Chang and Oxley (2009) found that industrial agglomeration can promote ETFP through innovation spillover. Cheng and Jin (2020) found that the industrial complementarity in industrial agglomeration has the strongest promotion effect on ETFP. Lu et al. (2021) believes that the common agglomeration of producer services and manufacturing has significantly promoted the improvement of regional GTFP and has a positive spatial spillover effect on surrounding areas. Xie et al. (2019) thought that the professional agglomeration of producer services not only increases the GTFP of cities and surrounding cities, but also weakens the promotion of GTFP in cities and surrounding cities by the professional agglomeration of producer services. Guo et al. (2020) studied the industrial agglomeration and green development efficiency in Northeast China, found that there is a u-shaped relationship between them, and analyzed the impact of economic development, environmental regulation, and other factors on ETFP.

This paper uses the industry clusters as the carrier to explore the facilitation of China’s industrial agglomeration on the ETFP in prefecture-level cities. Three paths have been constructed: technological innovation, government intervention and industrial structure. It may be through economies of scale, knowledge spillover effects, technological innovation effects, industrial upgrading effects, industrial transformation effects, and resource sharing effects on city economic performance, social performance, and environmental performance. Thereby accelerating environmental innovation, improving energy efficiency, and promoting the improvement of the production efficiency of city environmental total factors. The specific mechanism diagram (Figure 1) and the mechanism analysis are as follows.

First, industrial agglomeration has a positive impact on innovation through the Marshall externalities agglomerated within the industry (Arrow, 1971; Romer, 1986). Practitioners through face-to-face technical exchanges and information interactions in agglomeration areas are conducive to the spillover of environmental protection knowledge and green technology among enterprises. Second, industrial agglomeration is conducive to the interaction, sharing, and diffusion of energy-saving information and clean technology among enterprises (Wang et al., 2021; Xie et al., 2019). At the same time, cross-industry agglomeration activities are conducive to promoting the deep integration of information and communication technologies such as the Internet and big data with industrial industries, thereby promoting technological innovation. For example, Baptista and Swann (1998) found that the British manufacturing industry has a higher level of agglomeration and a higher level of innovation. In addition, industrial agglomeration strengthens the degree of competition among enterprises. Industrial agglomeration encourages enterprises not only to carry out technological and process innovation but also to learn from each other in the fields of knowledge and technology, thereby giving birth to new ideas and knowledge, accelerating the progress of low-carbon and environmentally friendly technologies, and increasing the green productivity of the industry. In summary, through the spillover effects of clean technology and knowledge, industrial agglomeration is conducive to improving the level of regional technological progress (especially the promotion of environmentally friendly innovative technologies), which can increase the ETFP (Xie et al., 2020). Therefore, this paper proposes the following hypothesis:

Hypothesis 1: Industrial agglomeration could grow the ETFP through the improvement of technological level.

Second, industrial agglomeration can influence ETFP through government intervention. The increase in production efficiency brought about by industrial agglomeration has significantly improved the income and living standards of residents, as well as the ecological and environmental problems that have begun to become prominent. Therefore, the government will raise environmental standards for industrial production, driving industrial agglomeration to gradually transform into green ecological agglomeration. Government intervention is mainly through the formulation of relevant laws and regulations, environmental subsidies, and other means to achieve city governance to facilitate pollution control (Ghazouani et al., 2021; Song et al., 2019). On the one hand, environmental regulations are considered to be the most effective way to solve environmental pollution. The government establishes an active intervention and supervision mechanism by formulating a series of corresponding environmental protection policies and regulations. For example, the government uses fines and shuts down high-polluting enterprises to increase the cost of environmental pollution by enterprises (Guo and Wang, 2018), thereby improving environmental pollution (Shen and Wang, 2018). On the other hand, the government encourages enterprises to introduce advanced production technologies through direct financial compensation for the R and D and application of green technology innovations in enterprises (Cheng et al., 2017), thereby enabling enterprises in agglomeration areas to achieve a “win-win” (Porter and Linde, 1995). In summary, this paper proposes the following hypothesis:

Hypothesis 2: Industrial agglomeration can facilitate the growth of ETFP by strengthening government intervention.

Third, industrial agglomeration can promote ETFP by optimizing the industrial structure. Economic structure and economic activities can have an important impact on environmental pollution and carbon emissions (Ikram et al., 2021; Sharma et al., 2021), and further affect environmental sustainability (Rafique et al., 2021). industrial agglomeration causes enterprises to obtain innovation monopoly profits, and various departments will compete to increase R and D investment to improve production efficiency and production technology level, thereby promoting the adjustment of industrial structure from the supply side. At the same time, industrial diversification and agglomeration optimize the structure of input factors through horizontal and vertical cooperation between industrial chains, thereby promoting the optimization and upgrading of industrial structure. Optimizing the industrial structure can reduce energy consumption (Yu et al., 2018) and reduce pollutant emissions (Cole et al., 2005), thus promoting the improvement of ETFP. Based on this, this paper proposes the following hypothesis:

Hypothesis 3: Industrial agglomeration is conducive to the optimization and upgrading of the industrial structure, thereby promoting ETFP.

In order to study the causal effect between industrial agglomeration and ETFP, we set up the following model based on the method of Lin et al. (2011), Li et al. (2021), Hu et al. (2015) and Liu and Zhang (2021): E T F P i t = α 0 + α 1 A g g i t + α 2 X i t + β 1 γ i + β 2 σ t + ε i t (1) where i represents city, t represents year, ETFP represents the environmental total factor productivity growth, Agg represents the degree of industrial agglomeration, X represents control variables, γ _i represents city fixed effect, σ _t represents year fixed effect, andε _it represents the random error term.

Agglomeration can be understood as the relative concentration of a certain element in a region. Common indicators used in existing studies to measure the degree of agglomeration of economic variables include Herfindahl index, Gini index, location entropy index, and Ellision-Glaeser index (Kamal and Sundaram, 2019). Among them, Herfindahl index, Gini index, and Ellision-Glaeser index can only measure the degree of agglomeration of elements at a global level but cannot effectively observe the degree of agglomeration of elements within each region, so they are not suitable for this study. The location entropy index can measure the spatial distribution of elements in a certain area, and is suitable as an effective index to measure the spatial distribution and agglomeration level of the research object, and it is comparable between different geographical units within the area (O’Donoghue and Gleave, 2004), so as to meet the needs of this research. Therefore, we draw on the idea of location entropy index and follow the method of Liu et al. (2021) and Mo et al. (2020) to construct the ratio of city secondary industry employees to the national secondary industry employees. This industry agglomeration index measures the degree of industrial agglomeration in each city, and its specific calculation formula is as follows: A g g i t = e m p _ sec i t / e m p i t ∑ i = 1 N e m p _ sec i t / ∑ i = 1 N e m p i t (7) where e m p _ sec i t represents the number of employees in the secondary industry in city i in year t, and e m p i t represents the total number of employees in the primary, secondary and tertiary industries in city i in year t.

(1) Technological progress (Tec). Regional industrial gatherings have a significant impact on the level of technological progress, and industrial gatherings produce knowledge and technology spillover effects, which are conducive to the integration of innovative elements, saving innovation costs, and improving innovation performance to facilitate regional technological innovation (Amiti, 2005). We use the number of patents related to green innovation in each prefecture-level city to indicate the level of technological progress in the region.

(1) Foreign direct investment (Fdi). Existing research has two main views on the impact of foreign direct investment on environmental pollution: the “pollution refuge” hypothesis holds that foreign direct investment increases environmental pollution in the host country (Cole et al., 2005). The “pollution halo” hypothesis (Balsalobre-Lorente et al., 2019) pointed out that the adoption of standardized management and production technologies by foreign-funded enterprises can help promote the construction of environmental management systems by domestic enterprises through spillover effects, and this would produce a pollution halo effect and improves environmental quality. Therefore, the sign of the coefficient of foreign direct investment is uncertain. Without loss of generality, we use the proportion of foreign direct investment in the city’s GDP to measure it.

First, we use the baseline regression in Table 2 to study causal relationship between industrial agglomeration and ETFP. Column (1) reports the results of POOL-OLS. We find that the impact of industrial agglomeration on ETFP is not significant without considering the fixed effects of cities and time. Columns (2) to (5) further control multiple effects. Specifically, column (2) controls city fixed effects and year fixed effects based on column (1). Columns (3) and (4) control macroeconomic variables and meteorological factors on the basis of the column (2), respectively. Column (5) controls macroeconomic variables, meteorological factors, city fixed effects, and year fixed effects.

We can find from column (5) that, after considering all control variables and fixed effects, industrial agglomeration can significantly facilitate ETFP. Specifically, on average, for each additional industrial agglomeration (10.63%), the ETFP increases by 11.44% (3.32%). This corresponds to an elasticity between industrial agglomeration and ETFP of 31.23%. Also, we can convert the magnitude using standard deviations. The estimates indicate that a one standard deviation increase in industrial agglomeration increases ETFP by 0.023 standard deviations.

We investigated the three mechanisms of industrial agglomeration affecting ETFP, and Table 3 reports the estimated results. Column (1) shows the impact of industrial agglomeration on technological innovation. We find that industrial agglomeration can significantly facilitate regional technological innovation levels related to green development. Specifically, due to the knowledge spillover effect that exists in the process of industrial agglomeration, industrial agglomeration can increase the level of technological innovation by 47.96% at the 5% significance level. Industrial agglomeration strengthens the upstream and downstream linkages of industries, which is conducive to enhancing input-output linkages, expanding knowledge spillover effects, and improving the availability of human capital, thereby reducing the cost of technological knowledge dissemination, generating technological innovation effects, improving the level of green technological innovation, and promoting ETFP. This research conclusions is consistent with Guo et al. (2021), who found that green innovation and CO₂ are cointegrated, and green innovation is one of the key contributors in explaining CO₂ emissions.

Column (2) reports the impact of industrial agglomeration on government intervention. We can find that industrial agglomeration can significantly facilitate the degree of local government intervention, thereby achieving an increase in ETFP. This is inseparable from the Chinese government’s continued attention to environmental and ecological issues in recent years. In terms of environmental investment, the government has continuously strengthened its investment in pollution control and increased its investment in green technology research and development for enterprises in concentrated areas. In terms of system construction, the government has continuously improved environmental regulations in agglomeration areas and improved relevant laws and regulations, fully implemented the pollution discharge permit system in agglomeration areas, and continued to facilitate market-oriented transactions such as pollution rights, energy rights, water rights, and carbon emissions rights, thereby promoting the city environment improvement of total factor production efficiency. This conclusion is also in line with Farooq et al. (2019) and Shahzad (2020). The burning of fossil fuels causes the emission of greenhouse gases and environmental pollutants, which in turn leads to environmental degradation and affects health. The strengthening of government environmental regulations such as environmental taxes (Farooq et al., 2019) and afforestation (Shahzad, 2020) will reduce environmental pollution and improve environmental quality.

Column (3) shows the impact of industrial agglomeration on the industrial structure. We can find that the improvement of industrial agglomeration can significantly facilitate the upgrading of industrial structure. More companies have transitioned from the secondary industry to the tertiary industry and further improved the regional ETFP. This may be caused by the following reasons. On the one hand, the increase in industrial agglomeration has strengthened the degree of market competition among enterprises, forcing enterprises to continuously improve technology and update equipment to gain competitive advantages, thereby achieving regional industrial structure upgrades. On the other hand, industrial diversification and agglomeration contribute to the formation of the entire industrial chain, improve the production efficiency of the manufacturing industry, and realize the upgrading of China’s manufacturing industry to the high-end of the global value chain. At the same time, it facilitates the vigorous development of the tertiary industry such as productive services. Finally, the ETFP is improved. Similar to the conclusion of Doğan et al. (2021) that economic complexity can alleviate environmental degradation, in the process of promoting the optimization and upgrading of industrial structure, industrial agglomeration has also promoted the growth of labor force, the improvement of production technology and the use of clean energy to a certain extent, thereby promoting the growth of ETFP.

To make our results more credible, we have conducted a large number of robustness tests through a variety of methods. The results are shown in Table 4. First, the undesired output in the ETFP calculation process includes the 3-year moving average data of the PM_2.5 of China’s prefecture-level cities calculated by Columbia University using satellites. We replaced the relevant PM_2.5 concentration data with the annual PM_2.5 concentration data calculated by the atmospheric composition analysis group of Dalhousie University in Canada. We find from column (1) that, after replacing the calculation content of ETFP, industrial agglomeration can still facilitate ETFP at the 5% significance level.

As China’s prefecture-level cities or provincial capital cities are the facades of this region, the effects of any national policies must be first shown in provincial capital cities. This also caused the provincial government to concentrate all political resources in the provincial capital to ensure that the provincial capital can fulfil the relevant national policy requirements. If this kind of government behavior is not eliminated, it will cause an overestimation of the effect of industrial agglomeration on ETFP. Therefore, we exclude four municipalities and all provincial capital cities, and the regression results are reported in column (2). We can find that after excluding government behaviors in provincial capital cities, industrial agglomeration can still facilitate ETFP at the 5% significance level.

Second, China has implemented transformational development in recent years, and more environmental policies have further strengthened the regulation of environmental pollutant emissions. The Chinese government has implemented the strictest environmental regulations and policies for the “2 + 26” cities with the most serious environmental pollution. Strict environmental regulation policies can significantly improve regional air quality. If these 28 regions are not excluded, we cannot distinguish the role of environmental regulation policies and overestimate the impact of industrial agglomeration on ETFP. Column (3) reports the estimated results after excluding 28 areas with serious environmental pollution. We can find that after excluding the impact of environmental regulatory policies, industrial agglomeration can still facilitate ETFP at the 1% significance level.

Finally, since the sample time of this paper started in 2003, it includes the migration of polluting enterprises from the east to the west in the Western Development Policy. The development of the secondary industry in the eastern region has been saturated, while the western region has a large amount of resources needed for the development of manufacturing enterprises. With the implementation of the western development policy, enterprises with serious pollution in the eastern region have moved to the western region. This caused an overestimation of the effects of industrial agglomeration policies in the eastern region. Column (4) excludes the heavy industrial areas in the east to excludes the influence of the western development policy. We can find that industrial agglomeration can still increase ETFP at the 5% significance level.

We also conducted a heterogeneous analysis of three aspects of the impact of industrial agglomeration on ETFP in Table 5. First, we rank the industrial agglomeration of all cities each year. We estimate the areas where the industrial agglomeration is higher than the average and lower than the average, respectively, to see the impact of different level of agglomerations on ETFP. We find from columns (1) and (2) of Table 5 that regions with high industrial agglomeration can significantly facilitate ETFP at the 1% significance level, while regions with low agglomeration have no significant effect on promoting ETFP.

Further, we separately examined the impact of industrial agglomeration in the eastern, central, and western regions on ETFP. From columns (3) to (5), it can be found that industrial agglomeration in the eastern region can have a significant impact on ETFP at the 5% significance level, while industrial agglomeration in the central and western regions has no significant impact on ETFP.

Finally, we examine the heterogeneity between resource-based cities ² and non-resource-based cities. The reason for the agglomeration of industries in resource-based cities may be to get closer to the geographical location of local mineral resources. The industrial agglomeration of non-resource-based cities can not only reduce the cost between the upstream and downstream of the industrial chain but also produce the Porter effect and achieve technology spillover, thereby improving production technology and achieving higher-quality development. Therefore, the impact of industrial agglomeration on ETFP will vary with city types. Columns 6) and 7) report the estimated results. We find that in non-resource-based cities, industrial agglomeration can indeed facilitate the increase of ETFP at the 5% significance level, while this facilitation effect is not significant in resource-based cities.