Industrial agglomeration and carbon neutrality in China: lessons and evidence

This study explored the impact of industrial agglomeration on carbon neutrality and its spatial spillover effects utilizing the expanded the output density theoretical model of Ciccone and Hall. The main findings are as follows: (1) In terms of long-term effects, industrial output has significantly reduced regional carbon emissions, while industrial labor and capital agglomeration have aggravated carbon emissions, and industrial technology agglomeration has no significant impact on carbon emissions. (2) From the perspective of lagging effects, industrial output agglomeration has significantly increased regional carbon emissions, while industrial technology agglomeration has effectively reduced carbon emissions. The lagging effects of industrial labor and capital agglomeration are not significant. (3) From the perspective of regional differences, there are significant regional differences in the impact of industrial output and capital agglomeration on regional carbon emissions, while this differential influence does not exist on industrial labor and technological agglomeration. Therefore, in carbon emissions governance, it is necessary to analyze the long-term impact of industrial agglomeration on carbon emissions but also to pay attention to the lag and regional differences of the impact.


Introduction
Carbon dioxide emissions have been the primary cause of climate change and have detrimental consequences on human life, irrespective of an economy being developed or underdeveloped (Rehman et al. 2021a, b, c, d, e). Countries around the world are making various efforts to reduce carbon emissions and achieve carbon neutrality (Rehfeldt et al. 2020;Subramanian et al. 2020). On September 22, 2020, during the 75th United Nations General Assembly, the Chinese government announced that it would adopt policies and measures that would enable the country to reach its carbon emissions peak no later than 2030 and achieve carbon neutrality by 2060 (Zhang and Li 2021). However, the reality of China's current energy and carbon emissions is that the energy consumption and carbon emissions in 2019 have increased by 69. 7% and47.2%, respectively, compared with 2006 (Hu 2021). Meanwhile, it is anticipated that China's economy will maintain a medium-to-high-speed growth (about 7%) for certain a long period of time . Therefore China's energy consumption and carbon emissions will also continue to grow in the future. In order to achieve carbon neutral by 2060, China need have near-zero emissions by 2050 and build an energy system with green and renewable energy (Sun et al. 2018;Alam and Murad 2020). The key factor in achieving this goal depends on the significant transformation of the country's economic development, industries, and energy system (Danish et al. 2020).
At present, China's economic development is gradually shifting from the extensive mode of pursuing growth speed to pursuing development quality and environmental efficiency (Shao et al. 2019). Industrial agglomeration and energy conservation and emission reduction have undoubtedly become the two major themes in the development of China's economic green transformation (Hong et al. 2020). On the one hand, the Chinese government is realizing resource sharing and optimized allocation through industrial agglomeration. On the other hand, as the world's largest country in primary energy consumption and carbon emissions, China's pressure on energy conservation and emission reduction is self-evident. In this context, what is the impact of industrial agglomeration on carbon emissions? Does industrial agglomeration hinder the goal of carbon neutrality? As the economic cooperation and ties between neighboring regions are close, and the spatial correlation characteristics of economic cooperation between regions are significant, the industrial agglomeration and carbon emissions in the region will have an impact on the surrounding regions. Is there a spatial spillover effect on the impact of industrial agglomeration on carbon emissions? What is more, does industrial agglomeration have a time lag in carbon emissions and thus show a lag effect? In order to answer the questions above, this study systematically analyzes the impact of industrial agglomeration on carbon emissions concerning with the following four aspects: industrial output agglomeration, industrial labor agglomeration, industrial capital agglomeration, and industrial technology agglomeration, utilizing the output density theoretical model of Ciccone and Hall and spatial measurement models. Furthermore, it explores the lagging influence and regional difference influence between them. The answers to these questions can help explain the relationship between industrial agglomeration and carbon neutrality and support in developing policies that would benefit economic development while strengthening carbon control.

Literature review
In the existing research, most researchers have paid attention to the influence of factors such as economic growth and energy consumption on carbon emissions Rehman et al. 2021a, b, c, d, e). With the deepening of research, researchers began to further explore the impact of industrial agglomeration on carbon emissions from the perspective of economic factors. Whether industrial agglomerations reduce the overall carbon emissions remains to be highly controversial, some scholars have found that energy conservation and emissions reduction are not the goals of industrial agglomeration (Zhang and Wang 2014;Ya and Meng 2019;Zhang et al. 2019;Dong et al. 2020a, b;Ding et al. 2020;Lu et al. 2021;Li et al. 2021). The endogenous power of industrial agglomeration is to realize the sharing of infrastructure, improve the matching efficiency of production factors, and promote the spatial spillover of professional knowledge, thereby improving the production efficiency of enterprises (Bie et al. 2017;Ding et al. 2019). Industrial agglomerations fail to promote carbon emissions reduction and may even distort existing allocations of production factors, which further exacerbate carbon emissions (Zhang and Wang 2014;Zhang et al. 2019). Lu et al. (2021) applied spatial econometrics and extended STIRPAT model to explore the impact of the industrial agglomeration on haze pollution using panel data from critical cities of Bohai Sea economic region (BSER) in China between 2006 and 2018 and found that while there is a significant positive relationship between the agglomeration of primary and secondary industries and haze pollution in the Bohai Sea economic region, the agglomeration of the tertiary industry has no significant impact on haze pollution (Lu et al. 2021). In addition, Dong et al. (2020a, b) and Li et al. (2021) also confirmed that industrial agglomeration increases regional environmental pollution (Dong et al. 2020a, b;Li et al. 2021).
On the other hand, New Economic Geography believes that industrial agglomerations, as compact economic spaces, can improve enterprises' production and resource utilization efficiencies with externalities as the link (Li and Zhang 2020). It is believed that industrial agglomeration may aggravate environmental pollution within a certain period, but once it reaches the threshold, industrial agglomeration benefits the environment (Yan et al. 2011;Yang 2015). Chen et al. (2017) found that industrial agglomeration reduces industrial carbon dioxide, which can help realize the emission reduction targets in China's prefecture-level cities (Chen et al. 2017). Zhang et al. (2018) concluded that industrial agglomeration in Henan Province is conducive to breaking the lock-in of high-carbon industries and reducing energy consumption (Zhang et al. 2018). Zheng and Lin (2018) found that for China's paper industry, industrial agglomeration can improve energy efficiency and reduce environmental pollution after it crosses the threshold (Zheng and Lin 2018). Liu et al. (2019), Guo et al. (2020), andChen et al. (2020) also identified an inverted U-shaped relationship between industrial agglomeration and industrial pollution at the city-level, and when a certain threshold is exceeded, industrial agglomeration is conducive to reducing industrial carbon emissions Guo et al. 2020;Chen et al. 2020).
Some scholars believe that the relationship between industrial agglomeration and carbon emissions is uncertain. Under different scenario factors, the impact of urban industrial agglomerations on carbon emissions efficiency varies (Yang 2017). Industrial agglomeration may increase carbon emissions in some regions while exert the restraining effect on carbon emissions of neighboring regions (Ya and Meng 2019). The relationship between industrial agglomeration and different types of emissions can also be heterogeneous . Wang and Wang (2019) suggest that the relationship of industrial agglomeration with sulfur dioxide and dust pollutants is heterogeneous in urban China. Shen and Peng (2020) used the spatial panel model to analyze the impact of industrial agglomeration externalities on environmental efficiency. Their study found that different systems and degrees of industrial agglomeration can have varying emissions reduction effects (Shen and Peng 2020). The emission reduction effect of industrial agglomeration is a double threshold effect Ahmad et al. 2021).
Previous studies have explored the impact of industrial agglomeration on carbon emissions at different levels. However, after conducting an extensive review of the existing literature (see Table 1), it is found that there are still some shortcomings with the current research. First, few studies have directly studied industrial agglomeration and carbon emissions. Studies have mainly focused on analyzing the impact of industrial agglomeration on environmental pollution at the macro-level perspective, but different types of pollutants have different characteristics. For example, carbon emissions have strong transboundary and spatial spillover effects, which may not be significant for other pollutants, such as industrial smoke and dust. When studying the impact of industrial agglomeration on pollution levels, the type of pollutants and their characteristics must be fully considered. Second, most studies use only a single indicator when analyzing the impact of industrial agglomeration on environmental pollution, often overlooking the differential impact of different agglomeration structures. Different industrial agglomeration characteristics have differentiated effects on the environment. Industrial agglomerations with labor as the main feature would impact pollution levels differently compared with capital and technology type agglomerations. Third, the impact of industrial agglomeration on carbon emissions exhibits a particular lag, which has often been neglected in previous research. This impact is often not obvious during the formation of industrial agglomerations and becomes more pronounced over time. These cumulative effects and lagging characteristics have to be considered when exploring the overall impact of industrial agglomeration on the environment.
To address these current research limitations, by using provincial panel data of 30 provinces in China from 2002 to 2018 and spatial econometric models, this study explores the spatial spillover effects, lag effects, and regional differences effects using different type of industrial agglomeration on carbon emissions. The contributions of this study are as follows. First, based on internal structure, industrial agglomerations were analyzed using four aspects: industrial output agglomeration, industrial labor agglomeration, industrial capital agglomeration, and industrial technology agglomeration. Each agglomeration aspect was analyzed separately to provide a more comprehensive overview of the impact of industrial agglomeration on carbon levels. Second, this study fully considered the transboundary impact and spatial spillover characteristics of carbon emissions and used spatial econometric models in analyzing the effects. Third, this study explored the lag effect and cumulative effects of industrial agglomeration on regional carbon emissions and examined the regional differences. The results of this study can be used to provide an explanation for the inconsistent research conclusions in the existing research and provide a reference for future research.

Model building
Drawing lessons from Dong et al. (2015) incorporating environmental factors into the industrial production model and using carbon emissions as a by-product of industrial output, this study incorporated the environment as an output factor into the output density model of Ciccone and Hall (1996) (Ciccone and Hall 1996;Dong et al. 2015). This study used the theoretical mechanism of action between industrial agglomeration and carbon emissions. The model can be expressed as follows: where, C i , N i , K i , andE i represent the total carbon emissions, industrial employment scale, industrial capital scale, and industrial energy consumption of area i, respectively. C i ∕A i , N i ∕A i , K i ∕A i , and E i ∕A i represent per unit area of carbon emissions, industrial labor density, industrial capital density, and industrial energy consumption density of area i, respectively. i is the production efficiency of area i; A i is the total area of area i; α is the return to scale of industrial labor, capital, and energy per unit area. When 0 < α < 1, the return to scale is decreasing; when α = 1, the return to scale is unchanged; when α > 1, the return to scale is increasing. β is the contribution rate of industrial labor output per unit area in region i, γ is the contribution rate of industrial capital-output per unit area in region i, 0 < β ≤ 1.0 < γ ≤ 1. λ is the parameter of carbon emissions concentration. When λ > 1, carbon emissions have externalities to the regional economy. After converting Eq. (1), it can be expressed as follows: Equation 2 is then converted to obtain as follows: (1) Equation 3 can then be converted to as follows: Taking the logarithm of both sides of Eq. (5), it can be expressed as follows: Equation (6) can be expressed as follows: In Eq. (7), ln(C i ∕A i ) indicates unit carbon emissions, ln i indicates industrial production efficiency, lnk i and lne i represent the efficiency of industrial resource use, and lnN i ∕A i represents industrial labor concentration.
The resulting equation shows an interactive relationship between carbon emissions and industrial efficiency and industrial agglomeration. This is consistent with the findings of Ya and Meng (2019), Dong et al. (2020a, b), and Lu et al. (2021) that found industrial agglomeration as a major factor leading to environmental problems. These studies also suggest that carbon emissions have spatial spillover characteristics and that the spatial measurement model can be used to construct the empirical model. Based on the spatial lag model (SLM) and the spatial error model (SEM) of Anselin (1995) (Anselin 1995), this study constructs the SLM and SEM models of the impact of industrial agglomeration on carbon emissions as follows: where lnC it represents carbon emissions, lnIA it represents industrial agglomeration, and lnX it represents the collection of control variables. To investigate the lagged effect of industrial agglomeration on carbon emissions, this study added the lagged 1st and lagged 2nd of industrial agglomerations to Eqs. (8) and (9) and formed the SLM and SEM models to analyze the lagged effects:

Carbon emissions
Based on the "Guidelines for National Greenhouse Gas Inventories" issued by the IPCC in 2006, carbon dioxide emissions are measured based on emissions from different energy sources, which include coal, hard coke, crude oil, gasoline, kerosene, diesel, fuel oil, and natural gas (Wu 2015;Yang et al. 2020). Consumption is calculated from various energy sources and multiplied each with the respective emissions coefficient, then added the values to obtain the total carbon emissions. The carbon emission coefficients were obtained from the China Energy Statistical Yearbook (2008), and the various energy sources were standardized before calculations. The calculation method is as follows: where C it is the total amount of carbon dioxide emissions of province i in year t; E irt denotes the consumption of energy r in year t of province i, and η r is the carbon emissions coefficient of the rth type of energy.

Industrial agglomeration
Production factors have considerable influence on the development of industrial agglomeration. Labor, capital, and technology are the primary production factors affecting economic agglomeration (Dong et al. 2015;Shen and Peng 2020). In this study, the degree of industrial agglomeration was first measured as a whole by industrial output aggregation (OA), measured as the industrial output per unit area (Dong et al. 2015). Then, industrial agglomeration was evaluated from three aspects: industrial labor agglomeration (LA), industrial capital agglomeration (CA), and industrial technology agglomeration (TA). Industrial labor agglomeration is measured by the number of industrial employees per unit area, industrial capital agglomeration is measured by the scale of industrial capital stock per unit area, and industrial technology agglomeration is measured by research and development (R&D) investment per unit area (Huang et al. 2019;Yao et al. 2020).

Control variables
Industrial energy consumption directly affects industrial solid waste and the total carbon emissions (Huang et al. 2019;Afridi et al. 2019;Yao et al. 2020;Murshed et al. 2021). At the same time, the level of carbon emissions in a region is also affected by the local economic development level (GDP), foreign direct investment (FDI) level, and industrial structure (IS) (Rehman et al. 2021a, b, c, d, (12) The higher the level of regional economic development and the greater the total economic volume, the more energy is consumed and the greater the carbon emissions (Yang et al. 2020;Regmi and Rehman 2021). The level of foreign investment reflects the degree of economic openness in a region. Economic openness and technology can significantly affect industrial agglomeration and impact the environment (Yang et al. 2018). The proportion of the secondary industry in the industrial structure can also have a considerable impact on energy consumption and pollution emissions. As a high-energy consumer, the secondary industry sector is the primary source of carbon emissions (Dong et al. 2020a, b). The level of economic development is measured by per capita GDP, FDI is measured by actual foreign investment, and industrial structure is measured by the proportion of the secondary industry sector (Dong et al. 2020a, b;Li et al. 2021). Data from 30 provinces in China from 2002 to 2018 were used as the research sample. The data were obtained from the "China Industry Yearbook," "China Statistical Yearbook," "China Environment Statistical Yearbook," "China Population and Employment Statistical Yearbook," and provincial statistical yearbooks. The missing data was calculated using the mean value method. In data processing, the logarithm of individual variables was used to eliminate the difference in variable measurement units. The definition and description of each variable are shown in Table 2.

Results and discussion
Are there spatial spillover characteristics of carbon emissions?
Before analyzing the spatial econometric model, it is necessary determined whether the explanatory variables have spatial spillover characteristics (see Fig. 1). To gauge spatial autocorrelation characteristics, exploratory data analysis methods were used in calculating the global spatial autocorrelation index (Global Moran's I) and Moran scatter plot of carbon emissions. Figure 2 reports the global Moran's I index of carbon emissions from 2002 to 2018. The average Moran's I for carbon emissions is 0.2450, and both indexes are significant at the 5% level. As shown in Fig. 3, most of the Moran's I for carbon emissions falls in the first and three quadrants of the scatterplot. For 2002For , 2007For , 2012, more than 60% of the areas were in the first and three quadrants with significant and positive spatial spillover characteristics. This suggests that carbon emissions in these regions exhibit some spillover effects on neighboring regions. This finding is consistent with the research conclusions of Wu (2015) and Yang et al. (2018). Therefore, when analyzing the impact of industrial agglomeration on carbon emissions, spatial factors must be considered.

Does industrial agglomeration increase carbon emissions?
OLS regression is performed using the LM test and the robust LM test for SLM and SEM models for model selection. The p-values of the LM test and robust LM test for the SLM model are significant, while not significant for the SEM model. It is also found that the fixed-effects model yielded better results than the random-effects model. Based on these preliminary results, the SLM model with fixed effects was used in the remainder of this study. To avoid multicollinearity problems, this study conducted separate regression analyses on the four indicators of industrial agglomeration. Table 3 summarizes the analysis results of the impact of industrial agglomeration on carbon emissions.
The results show that industrial output agglomeration effectively suppressed regional carbon emissions and exhibited a significant lag effect. Table 3 reports the impact of industrial output agglomeration on carbon emissions. The coefficient for industrial output agglomeration is negative and statistically significant at the 5% level, indicating that industrial output agglomeration reduces carbon emissions. The agglomeration of industrial output generally reflects the level and degree of industrial agglomeration in a region. The negative coefficient suggests that the industrial output agglomeration has realized the optimal allocation of regional resources, improved resource utilization efficiency, and Moran's I combined regional and resource agglomeration advantages and was able to improve the environment while also promoting economic development.
The coefficient for industrial output agglomeration with one lag period is positive and significant at the 5% level. The coefficient for industrial output agglomeration with two lag periods is also positive but not significant. This suggests that industrial output agglomeration has a lag effect on regional carbon emissions and that this lag effect is significant for the one lag period. The lag effect is manifested by the aggravation of regional carbon emissions and increased environmental pollution. But over time, the lag effect of industrial output agglomeration on carbon emissions becomes not significant.
For industrial labor agglomeration, its lagging impact is negative and not significant. As shown in Table 3, the coefficient for industrial labor agglomeration is positive and significant at the 1% level, indicating that industrial labor agglomeration has increased regional carbon emissions. The results suggest industrial labor agglomerations exhibit negative externalities on carbon emissions levels, which is consistent with the research conclusions of Zhang and Wang (2014) Ding et al. (2020), and Lu et al. (2021). The concentration of industrial labor is not conducive to achieving regional carbon control targets. China is still dominated by labor-intensive industries, where the proportion of low-end industrial workers remains relatively large and concentrated (Dong et al. 2020a, b). In comparison, capital and technology-intensive industries are less concentrated where the proportion of high-end industrial workers is lower (Ding et al. 2020). The coefficients for industrial labor agglomeration with one and two lag periods are both negative and not significant. This suggests that the lag effect of industrial labor agglomeration is not significant and that its effect on carbon levels will not considerably change over time.
For industrial capital agglomeration, its impact on carbon emissions was found to be not significant. In Table 3, the coefficients for industrial capital agglomeration and two lagging periods are positive and non-significant. The impact of industrial capital agglomeration on regional carbon emissions is not significant mainly because the degree of industrial capital agglomeration is far lower than labor in China (Dong et al. 2015). The agglomeration of industrial capital is inefficient and has limited direct effect on the regional environment (Zhao et al. 2020). Industrial capital is often reflected in the introduction of technology and equipment, affecting regional carbon levels and ultimately resulting in labor, technology, and output concentrations of the regional industrial agglomeration. Industrial technology agglomeration was found to intensify regional carbon emissions and cause environmental degradation. However, its lag effect is positive, which suggests that, over time, regional carbon emissions are effectively suppressed. As shown in Table 3, the coefficient for industrial technology agglomeration is 0.1453, significant at the 10% level, which indicates that it increases prevailing regional carbon emissions. Industrial technology agglomeration normally does not immediately produce innovation effects and is mainly manifested as the cost effects, resulting in environmental degradation (Yang et al. 2021). The estimated coefficients for the one and two lagging periods are negative and statistically significant. This indicates that industrial technology agglomeration has a positive lag effect on carbon emissions. Comparing the significance level of the coefficients, it is found that the one with the two lagging periods has higher significance level than the one with only one lagging period. This suggests that over time, industrial technology agglomeration shifts from increasing carbon emissions towards reducing carbon levels (Hong et al. 2020). Industrial technology agglomeration gradually exerts an innovative compensation effect, which can be conducive to improving the regional environment and help control carbon emissions (Pei et al. 2020). This finding supports the conclusions of Chen et al. (2020) and Pei et al. (2020) to a certain extent. Technology agglomeration may cause a transitory increase in environmental pollution but would later result in a positive environmental effect and reduce carbon emissions in the long-term.
The spatial correlation coefficient ρ for carbon emissions is significant at the 10% level in all four models. This suggests that carbon emissions have significant spatial spillover effects, where the region's carbon emissions affect the carbon levels of its surrounding areas. The ρ-values are 0.2560, 0.1010, 0.0910, and − 0.0070, which indicates that spatial correlation of carbon emissions is dominated by positive autocorrelation. Regional carbon emissions show high agglomeration in high-carbon emission regions and low agglomeration in low-carbon emission regions. The spatial correlation coefficient ρ is significant, highlighting the importance of considering the spatial dimensions in these types of research. The results also suggest that the level of regional carbon emissions is affected not only by its own industrial agglomeration but also by the level of agglomeration in the surrounding areas.
In terms of control variables, energy consumption, economic development level, and industrial structure were found to significantly intensify regional carbon emissions. The coefficients for these three parameters are all positive, although not significant. The coefficient for FDI (see Table 3) is positive in all models, although two are not significant. These findings suggest that the introduction of FDI plays a positive role in controlling regional carbon emissions.
In general, industrial output agglomeration plays a significant role in restraining the prevailing regional carbon emissions and is conducive to regional environmental governance. In line with the goals and policies of the Chinese government, the development of more industrial clusters and industrial parks can be used to promote economic development and help improve the environment. However, industrial labor and technology agglomerations exert negative external effects, leading to increased regional carbon emissions. Industrial output and technology agglomerations have significant lag effects, while the direct and lag effects of industrial capital agglomeration on carbon emissions are not significant.

The impact of industrial agglomeration on carbon emissions from the perspective of regional differences
To further explore the regional differences in the impact of environmental regulations on carbon emissions, the research areas are divided into three  Table 4 reports the impact of industrial output agglomeration on carbon emissions. The coefficient for industrial output agglomeration in the eastern and central regions is negative and significant at the 1% level, indicating that industrial output agglomeration reduces carbon emissions in these regions. This result confirms the robustness of the national panel data regression results in Table 3. For the western region, the industrial output agglomeration coefficient is significant and positive, suggesting that industrial output agglomeration increases regional carbon emissions in the western provinces of China. The main reason for this difference is the regional distribution of industries in China. Due to the country's regional development policy and partly due to its war preparation strategies in the early days of the founding of the People's Republic of China, industries were roughly evenly dispersed throughout the country ). But after the economic reform and opening up, the distribution of industries began to show gradual agglomeration in the eastern and central regions, particularly in the southeast coastal areas (Yi and Zhou 2020). In addition, the lag effect of industrial output agglomeration significantly increased carbon emissions in the eastern and western regions, while the effect is not significant for the central region.  Industrial labor agglomeration affects regional differences in carbon emissions Table 5 reports the results of the impact of industrial labor agglomeration on carbon emissions for different regions. The coefficients for industrial labor agglomeration in the eastern, central, and western regions are 1.133, 1.1186, and 0.9173, respectively, significant at the 1% level. The results suggest that for every 1% increase in industrial labor agglomeration, carbon emissions increase by 1.133%, 1.1186%, and 0.9173%, consistent with the national results. In all three regions, industrial labor agglomeration increased environmental pollution. One possible reason is that government-initiated industrial parks and employment-oriented industrial agglomerations are largely dominated by low-end labor (Dong et al. 2015). This kind of labor agglomeration is unable to attract high-level talents and cannot produce the corresponding "technical effect," which is not conducive to the innovation of clean, energy-saving, and emission-reducing technologies (Yao et al. 2020). For example, in the eastern coastal areas, the labor sector in many industrial clusters is composed mainly of unskilled migrant workers. These clusters are unable to achieve knowledge innovation and talent agglomeration effects. The influx of migrant workers to the eastern and central industrial agglomeration regions may also lead to increased energy and raw material consumption, more waste discharges (e.g., household garbage), and higher environmental pollution. The industrial labor agglomeration with two lag periods has a significant control effect for the central region, while not significant for the other regions.

Industrial capital agglomeration affects regional differences in carbon emissions
While industrial capital agglomeration was not significant in the eastern and central regions, it significantly increased regional carbon emissions in the western region (see Table 6). This is probably because of China's current industrial agglomerations having many labor-intensive, low-capital, and low-technology industries. In recent years, the Chinese government has initiated measures focused on developing the western region. The western development strategy is mainly characterized by infusing more government funds and economic agglomerations from crossadministrative economic belts (Chen et al. 2018). While this type of government-led industrial agglomeration has capital agglomeration characteristics, the western region's economic development has been comparatively low (Zhang et al. 2018). To promote industrial development and support industrial agglomeration, many infrastructure projects, such as roads and railways, would have to be undertaken, which may cause greater environmental damage (Ya and Meng 2019). The results also show that the lagged impact of industrial capital agglomeration on carbon emissions is not significant in the three regions. The effect of capital agglomeration on regional carbon emissions is mainly manifested in the immediate term but is not significant in the long term.

Industrial technology agglomeration affects regional differences in carbon emissions
The estimated coefficients of industrial technology agglomeration for the eastern, central, and western regions are 0.5030, 0.3902, and 0.2693 (see Table 7). The effect is  significant for the eastern and western regions but not significant in the central region. The findings suggest industrial technology agglomeration has not exerted its technical effects and has not achieved technological innovation and technological diffusion. One possible reason for this is that even with industrial agglomeration, the level of independent R&D and technological innovation remains low (Yan et al. 2011). Advanced, high-end, low-carbon technologies are mainly obtained through imports, and the current industrial clusters cannot achieve technological innovation and technological diffusion (Han 2020). Technology agglomeration is mainly manifested as a cost effect, and it is difficult to compensate for the innovation effect. In terms of lag effects, the second-stage industrial technological innovation significantly reduced carbon levels only in the eastern region and was not significant for the other two regions.

Discussion
1) In terms of long-term effects, industrial output has significantly reduced regional carbon emissions, while industrial labor and capital agglomeration have aggravated carbon emissions, and industrial technology agglomeration has no significant impact on carbon emissions. First, the agglomeration of industrial output has significantly reduced regional carbon emissions, which is conducive to achieving the goal of carbon neutrality. Table 3 and Fig. 5, the industrial output agglomeration has significant and negative correlations with carbon emissions, which means it is able to lower carbon emissions. Industrial output agglomeration is the concentrated expression of result-oriented industrial agglomeration (Dong et al. 2015). The original intention of China's industrial agglomeration, which is guided mainly by market forces and site determination, is based largely on the principle of efficiency. A  certain degree of industrial agglomeration realizes the connection between enterprises and local comparative advantages, which can effectively stimulate agglomeration effects, promote economic development, and reduce environmental pollution (Yang 2015;Han and Li 2019). In China, many industrial agglomerations are based on market-driven industrial configuration, which selects suitable industries and promotes industrial agglomeration using market rules and efficiency principles to form agglomeration advantages (Lin et al. 2019;Li and Zhang 2020). Second, industrial labor and technology agglomerations significantly increased regional carbon emissions. In China, in order to achieve economic development and maximize fiscal incentives, many local governments actively encourage foreign companies to set up industries in clusters within their jurisdictions (Ya and Meng 2019). These agglomerations initiated by local governments promote rapid economic growth in the short term. However, most of these industrial clusters are unable to produce agglomeration effects because they did not fully adhere to prevailing market forces (Guo et al. 2020). Instead, these industrial agglomerations are accompanied by an accelerated rise in urban labor, resulting in the expansion of public infrastructure and intensified urban utilization, which consequently leads to environmental damage and higher pollution levels (Dong et al. 2020a, b). Moreover, government-initiated industrial agglomerations are not always consistent with the local and regional comparative advantages Lu et al. 2021). This, in turn, often leads to low technological innovation and poor technological agglomeration and instead yields negative environmental externalities. In the short term, industrial labor agglomeration and technology agglomeration are not conducive to zero carbon emissions and carbon neutrality (Zhang and Wang 2014).

As shown in
Finally, the effect of capital agglomeration on carbon emissions is not significant. In the agglomeration zone, capital agglomeration is conducive to the sharing of production materials and equipment, which can effectively reduce energy consumption (Yang 2017). On the other hand, capital agglomeration can help attract highly skilled talents and encourage competition. However, in China, although there is a certain level of resource sharing in industrial agglomeration zones, the use of capital as a core element of enterprise development may not necessarily promote entrepreneurial cooperation . This means that capital agglomeration effect cannot be achieved in industrial clusters and that energy conservation cannot be effectively reduced. This also shows that industrial capital agglomeration cannot be the main path to lower enterprise energy consumption and achieve carbon neutrality. 2) From the perspective of lagging effects, industrial output agglomeration has significantly increased regional carbon emissions, while industrial technology agglomeration has effectively reduced carbon emissions. The lagging effects of industrial labor and capital agglomeration are not significant. Firstly, the lag effect of industrial output agglomeration increases regional carbon emissions. The results indicate that the industrial output agglomeration with one lag period significantly aggravated regional carbon emissions, while output with two lag period had a positive but not significant impact. This suggests that while industrial agglomeration promotes optimal resource allocation, it can also generate employment demand, leading to urban population growth Chen et al. 2017). The accelerated rise in urban population will lead to higher demands for public facilities and housing and more urban land expansion, consequently increasing regional carbon emissions (Zheng and Lin 2018).
What is more, the lag effect of industrial technology agglomeration helps reduce carbon emissions. First, industrial technology agglomeration is conducive to the imitation and innovation of knowledge and technology (Zheng and Lin 2018). Especially in the agglomeration areas, new technologies are often quickly learned and imitated by other companies, which are further improved through innovation (Ding et al. 2020). The proliferation of technological innovation spread in the agglomeration area and ultimately increases enterprises' overall technological innovation level Ullah et al. 2021). Over time, technological innovation promotes research and development of low-carbon technology, improves production efficiency, and supports energy conservation. Second, the agglomeration of industrial technology is conducive to the diffusion of innovation (Shen and Peng 2020). Compared to the dispersed spatial pattern, agglomeration is more conducive to the dissemination and diffusion of new knowledge and technologies (Finnerty et al. 2018). Industrial technology agglomerations would eventually innovate towards low-carbon technology, become more energy-efficient, and reduce their carbon emissions. Technological innovation is key to low-carbon technologies and the driving force for enterprises to achieve energy conservation and emissions reduction.
Then, industrial labor and capital agglomerations do not exhibit statistically significant lag effects. Which is mainly because China's industrial agglomerations are still dominated by labor-intensive industries with relatively large proportions of low-end industrial prac-titioners (Dong et al. 2015). This leads to low levels of labor sharing in industrial agglomerations and achieves minimal talent sharing advantages. In China, industrial agglomerations are dominated by low-end labor enterprises, making it difficult to realize industrial labor sharing and capital sharing and improbable to exert agglomeration effect.
3) From the perspective of regional differences, there are significant regional differences in the impact of industrial output and capital agglomeration on regional carbon emissions, while this differential influence does not exist on industrial labor and technological agglomeration (see Fig. 6).
First, the impact of industrial output agglomeration on carbon emissions is different in the three regions. The industrial output agglomeration in the eastern and central regions has a significant carbon mitigation effect, while in the western region, agglomeration exacerbates regional carbon emissions. In recent years, with the rapid industrial advancement and structural reforms in eastern and central China, the economy and the environment have had a relatively good coordinated development (Yao et al. 2020). This development of industrial agglomerations yielded a win-win situation for both economic growth and environmental protection . However, since the western region is still in the initial stage of industrialization, its development model still capitalizes on industrial agglomeration primarily to achieve economic growth, even at the expense of the environment.
Then, the impact of industrial capital agglomeration on carbon emissions is also different in the three regions. Industrial capital agglomeration was to have no significant impact on regional carbon emissions in the eastern and central regions, while in the western regions, it significantly increased emissions. In recent years, the Chinese government has implemented the Western Development Program to vigorously promote the economic development of the western region . The plan uses government funding to construct industrial belts and industrial parks as the main method. Several industrial parks oriented by government policies have been constructed in the western region. The construction of many industrial parks would require extensive use of steel and cement and destroy vegetated areas, resulting in increased environmental pollution (Ding et al. 2020). Due to differences in industrial agglomeration characteristics, economic development levels, and regional advantages, the impact of industrial agglomeration on carbon emissions can significantly vary . These variations can lead to considerable differences in the speed and method of achieving carbon neutrality goals in various regions.
However, there is no significant regional difference in the impact of industrial labor and technological agglomeration on carbon emissions in the three regions. Industrial labor and technology agglomerations were shown to aggravate regional carbon emissions in all three regions. China's industrial agglomerations are still heavily laborintensive, and many are considered low-end labor agglomerations (Dong et al. 2020a, b). The agglomeration of lowend labor forces makes it difficult to realize labor sharing and knowledge sharing in industrial parks and thus cannot exert talent accumulation effect ). On the other hand, the accumulation of low-end labor will increase the consumption of resources and energy in the industrial park and increase environmental pollution (Yang et al. 2020(Yang et al. , 2021. The accumulation of industrial technology in the three regions has increased carbon emissions, which reflects the cost effect of industrial technology agglomeration. The main reason is that a long period is needed for industrial technology agglomeration to play the compensation innovation effect and exhibit substantial carbon control effects.

Conclusions and policy implications
This study expanded the output density theoretical model of Ciccone and Hall and constructed a theoretical model of the relationship between industrial agglomeration and carbon emissions. Using data from 30 provinces in China from 2002 to 2018, the theoretical model was verified using spatial measurement methods. The main findings are as follows: 1) In terms of long-term effects, industrial output has significantly reduced regional carbon emissions, while industrial labor and capital agglomeration have aggravated carbon emissions, and industrial technology agglomeration has no significant impact on carbon emissions. This suggests that industrial labor and technological agglomerations have negative environmental externalities, while industrial output agglomeration has mitigation effects on regional carbon emissions. 2) From the perspective of lagging effects, industrial output agglomeration has significantly increased regional carbon emissions, while industrial technology agglomeration has effectively reduced carbon emissions. Industrial output agglomeration promotes optimal resource allocation; it can also generate employment demand, leading to urban population growth and environmental pollution. While industrial technology agglomeration is conducive to the imitation and innovation of knowledge and technology, the proliferation of technological innovation spreads in the agglomeration area and ultimately increases enterprises' overall technological innovation level. But the lagging effects of industrial labor and capital agglomeration are not significant. 3) There are significant regional differences in the impact of industrial output and capital agglomerations on regional carbon emissions. Industrial output agglomeration in the eastern and central regions has a significant carbon control effect, while it exacerbates regional carbon emissions in the western region. Industrial capital agglomeration in the eastern and central regions has no significant impact on carbon emissions, while it significantly increased regional carbon emissions in the western region. However, industrial labor and technological agglomerations have no significant regional differences.
Based on the above research results, this study proposes the following policy recommendations. First, the government should pay attention to the relationship between industrial agglomeration, economic development, and carbon neutrality in developing new industrial zones. Increased industrial agglomeration can facilitate more effective energy and resource consumption and encourage new technologies, which is an important way to achieve carbon neutrality. However, industrial labor and technological agglomerations aggravate regional carbon emissions, resulting in negative environmental effects. The government should consider industrial structural reforms and management policy modifications towards more coordinated economic development and carbon neutrality, particularly concerning industrial agglomerations. Then, in carbon emission governance, coordinated strategies and joint governance mechanisms should be established. This study shows that carbon emissions have a significant spatial spillover effect, which means that a particular region's carbon levels affect its neighboring areas. In China, there are clear delineations in management and governance between administrative regions, which may cause substantial gaps in environmental pollution control and management. Since carbon emissions have strong spatial characteristics, intra-regional coordination strategies and collaboration must be established to jointly plan the layout of industrial clusters and implement environmental governance. What is more, for carbon emission control, the differences between regions should be considered, and different regions have different ways to achieve carbon neutrality goals. In China, the eastern region is relatively developed and has significant industrial agglomeration characteristics. The central and western regions are comparatively less economically developed, and the agglomeration effect of their industrial sector is not as pronounced. Different regions have varying industrial structures and distinct differences in industrial agglomeration methods. Such regional differences result in industrial agglomerations having heterogeneous effects on carbon emissions. These regional and local differences have to be further explored and considered, particularly in developing policies and strategies on industrial agglomerations. Finally, companies should fully measure carbon emissions and use industrial energy conservation and emission reduction as the main method to achieve 0 carbon emissions and carbon neutrality. The transformation of the industry to low-carbon and the realization of industrial upgrading are the foundation of the decarbonization of the regional economy.

Limitations and improvements
The limitations of this paper mainly lie in the following aspects. First, when analyzing the impact of industrial agglomeration on carbon neutrality, this paper mainly considers the control variables such as energy consumption, economic development, and industrial structure. While other social factors such as population will also have an impact on carbon neutrality, further exploration is needed (Rehman et al. 2021a, b, c, d, e). Second, in the process of measuring industrial agglomeration, due to the availability of data, this paper has selected relatively macro data to measure