Research on the Threshold Effect of Green Technology Innovation on Fog–Haze Pollution in the Transfer of Air Pollution-Intensive Industries: A Perspective of Thermal Power

Abstract

Green technology innovation can effectively reduce the problem of pollution transfer in air pollution-intensive industries like thermal power and realize the green development of air pollution-intensive industries like thermal power. Based on green technology innovation, this paper analyzes the spatial–temporal characteristics of fog–haze in 31 provinces and municipalities. Taking the panel data of 31 provinces, municipalities, and autonomous regions from 2000 to 2017 as samples, this paper adopts the panel threshold regression method to examine the relationship between green technology innovation and fog–haze pollution in the transfer of air pollution-intensive industries like thermal power. The study found the following: China’s haze outbreak and the subsequent increasingly serious reasons for the implementation of weight detection haze policy seriously misled the haze prevention and control work, simple disorganized management aggravated the degree of haze pollution, and layer by layer, management methods caused the huge increase in secondary particulate matter; haze pollution aggregation occurs in the area of environmental self-purification capacity in the low air pollution-intensive industrial agglomeration to affect the atmospheric environment, a significant increase in the neighbouring industrial pollution agglomeration in resource-rich provinces; green technology innovation above the threshold has a significant inhibitory effect on the industrial transfer of haze pollution, and so on. There is a need for the scientific planning of pollution industry transfer to undertake the development of the place, the effective transfer of Beijing–Tianjin–Hebei haze pollution and other areas of air pollution-intensive industries, the development of targeted green technology innovation to strengthen policies, the scientific management of haze pollution, and the contribution of the scientific management of haze pollution in China.

1. Introduction

There is a question about the importance of combating haze pollution in China. China’s haze pollution is due to the high number of particles, and the weight of the particles has a very small effect. As the causes of haze pollution in China have not been clarified, the implementation of the policy of detecting haze pollution by weight method in 2011 and 2012 has seriously misled the haze prevention and control work (in order to reduce the weight surface of particles in the air, there has been a large amount of water vapour spraying and the full promotion of electrostatic dust removal technology). In order to complete the five-year goal of haze pollution control, a large number of the closed and limited production of scattered and polluted enterprises in 2017 and 2018 caused great pressure on China’s economic growth and social harmony, so the scientific management of haze pollution and the precise management of haze pollution is very important. In early 2020, when people stayed at home to avoid the novel coronavirus pneumonia, a large number of factories and schools were shut down, and motor vehicles were parked at home. However, many cities in northern China experienced severe smog-polluted weather. This was mainly because thermal power, petrochemical, steel, and other enterprises that heat in winter did not shut down because their shut down would lead to a remarkable impact [1], so the scientific transfer and elimination of thermal power and other air pollution-intensive industries has become an important means of combating haze pollution. As a result, the appropriate transfer of thermal power, steel, petrochemical, and other air pollution-intensive industries has become an important issue of concern. The relevant studies are as follows.

1.1. Research on Green Technology Innovation

With the aggravation of global climate change and environmental pollution problems, the view that green innovation can bring beneficial environmental effects has gained consensus in the academic community. Green innovation usually refers to innovation that can reduce the adverse impact on the ecological environment, which is characterized by double externality, unpredictable risk, high uncertainty, high complexity, and high preliminary resource investment [2,3]. Currently, green technology innovation has become the core driving force to promote sustainable development. Green technology refers to the realization of the goal of promoting green development and forming a green development model by scientific and technological innovation [4]. Specifically, green technology innovation has a significant inhibitory effect on pollutant emissions [5]. Especially in the industrial field, the breakthrough of desulphurisation, denitrification, and dust removal technology, and the synergistic innovation of haze management have become the focus of attention of the current academic community. In the field of green technology, scholars promote the breakthrough of technological boundaries by publishing papers, applying for patents, and participating in policy consultations. Green innovation talents, on the other hand, refer to the professional groups with practical ability in the field of green innovation, including engineers, technical workers, policymakers, and so on. Scholars and talents jointly promote the theoretical and practical development of green innovation.

1.2. Research on the Threshold Effect of Green Technology Innovation and the Transfer of Haze Pollution from Air Pollution-Intensive Industries

Domestic and foreign scholars have researched the threshold effect of green technology innovation on fog–haze pollution in the transfer of air pollution-intensive industries centres on three topics: the relationship between industrial transfer and environmental pollution; the relationship between technological innovation and industrial transfer; how technological innovation (as an intermediate variable) regulates the impact of industrial transfer on environmental pollution.

Firstly, there are two main types of views on the relationship between industrial transfer and environmental pollution: one believes that there exists a ‘pollution refuge hypothesis’ between industrial transfer and environmental pollution. Another view is that there is a ‘pollution halo hypothesis’ between industrial transfer and environmental pollution. Secondly, there are two views on the study of the relationship between technological innovation and industrial transfer: one believes that industrial transfer promotes technological innovation. Another view is that industrial transfer is not conducive to technological innovation. Finally, as to how technological innovation, as an intermediate variable, regulates the impact of industrial transfer on environmental pollution, different scholars have studied it from different angles. Some studies have found that when technological innovation is still in its infancy, there is an “inverted U-shaped” relationship between technological innovation and carbon emissions [6]. On the basis of confirming that technological innovation and environmental pollution are in an ‘inverted U-shaped’ relationship, some studies further found that there is a significant threshold effect of technological innovation on environmental pollution [7,8,9,10]. This threshold effect of technological innovation on the efficiency of environmental pollution management includes a single threshold and a double threshold [11].

To sum up, the existing scholarship has made considerable achievements in the relationship between technological innovation, industrial transfer, and environmental pollution, yet lacks systematic research on the relationships between them. In the research on technological innovation and industrial transfer, scholars mostly apply traditional measurement models, failing to consider the time–space effect of industrial transfer. In the research on technological innovation and environmental pollution, scholars separately explore the effects of environmental pollution and technological innovation. In the process of industrial transfer, there are both positive and negative spillover effects, which need to be analyzed in a systematic way. Compared with the existing literature, the innovation in this paper lies in that it uses a threshold effect model to discuss the relationship between the transfer of air pollution-intensive industries like thermal power and fog–haze pollution. Then, it scrutinizes the spatial–temporal characteristics of fog–haze, determines high–high fog–haze regions in China, and explains the relevant reasons. Finally, this paper provides reasonable suggestions on policies on how to avoid the transfer of pollution, improve green technology innovation, and accelerate the green transfer of air pollution-intensive industries like thermal power.

2. Model Construction and Variable Selection

2.1. The Construction of the Threshold Model

Relevant studies show that Hansen’s non-dynamic panel threshold regression model can be used to draw the conclusion that inter-regional industrial transfer has a positive technology spillover effect on the underdeveloped western region [12]. In addition, Hansen’s threshold panel regression model was used to analyze the relationship between high-polluting industrial relocation and environmental pollution by introducing relative environmental regulation intensity as the threshold variable [13]. The results of these studies show that there is a nonlinear relationship between high-pollution industrial transfer and environmental pollution. With relative environmental regulation intensity rising from low threshold to high threshold, the environmental pollution that arises from the transfer of high-pollution industries intensifies [14].

Hansen’s threshold regression model was proposed by Hansen in 1999, and its essence is to divide the data into different intervals by setting one or more threshold values through threshold variables and using different regression equations to describe the data characteristics in each interval. This model captures the nonlinear relationships between variables and accurately characterizes the changes in variables at different intervals. Through the model, the threshold value between variables can be identified, which provides a basis for scientific decision-making. Based on the existing literature, this paper uses Hansen’s threshold regression model to construct the threshold regression model of this study and conducts a statistical analysis of the corresponding data to determine the threshold value. Hansen’s self-help method generates a sequence of dependent variables by simulation, calculates the LM statistics of the simulation, and repeats it many times to obtain an asymptotic distribution and then obtain the p value. This paper adopts Hansen’s Bootstrapping to acquire the critical value of the approximate distribution and then the p value based on the likelihood ratio test. If the p value reaches a low level (e.g., less than 0.1) and significantly rejects the original hypothesis, the threshold effect is justified.

This paper probes into the relationship between green technology innovation, the transfer of air pollution-intensive industries like thermal power, and fog–haze pollution. Taking the transfer of air pollution-intensive industries like thermal power as the threshold value, this paper delves into the impact of green technology innovation on fog–haze pollution. In order to eliminate the influence of heteroscedasticity, the threshold formula is calculated using the logarithm of the data. The formula is expressed as follows:

$$\begin{gathered} {lny1}_{it}={\beta }_0+{\beta }_1{k}_{it}\left(k_{it}<\mathsf{\gamma }\right)+{\beta }_2{k}_{it}\left(k_{it}\ge \mathsf{\gamma }\right)+{\beta }_3{lnx}_{it}+ \\ {\beta }_4{lnm1}_{it}+{\beta }_5{lnm2}_{it}+{\beta }_6{lnm3}_{it}+{\beta }_7{lnm4}_{it}+{\epsilon }_{it} \end{gathered}$$

(1)

In particular, ${\beta }_0$ stands for the constant term, lny1 stands for the natural logarithm of Geographic–Mean PM2.5, $k_{it}$ stands for the transfer of air pollution-intensive industries like thermal power, ${\epsilon }_{it}$ stands for the disturbance term, and other control variables cover population density, per capita GDP, per capita car ownership, and the industrial structure, respectively.

2.2. The Selection of Variables

The corresponding measurement indicators are selected according to the research design and needs, as shown in

Table 1. Indicator selection table.

Variable	Indicator	Indicator Measurement
Explained variable	Geographical average PM2.5 (lny1)	PM2.5 divided by the area
	Population average PM2.5 (lny2)	PM2.5 divided by the population of the area
Threshold variable	Green technology innovation (lnx)	Number of green technology patents
Core explanatory variable	Transfer of air pollution-intensive industries such as thermal power (k)	Pollution-intensive industries transfer output value
Control variable	Industrial structure (lnm1)	The value added of the secondary industry accounted for the proportion of local GDP
	GDP per capita (lnm2)	GDP divided by regional population
	Car ownership per capita (lnm3)	Cars divided by the population of the area
	Population density (lnm4)	Population divided by area

:

2.2.1. The Explained Variable: PM2.5

The fog–haze results from the interaction between specific climatic conditions and human activities are as follows [15]. Among the main components of fog–haze, nitrogen oxide, sulphur dioxide, and PM2.5, PM2.5 plays a major role in aggravating fog–haze pollution [16]. PM2.5 has a significant impact on air quality and visibility. Therefore, this paper sets PM2.5 as the explained variable.

2.2.2. The Threshold Variable: Green Technology Innovation

Green technology innovation includes clean production technology innovation and end-treatment technology innovation, and it is an important means to control air pollution and improve environmental quality. Since green technology reduces energy consumption, green technology innovation is measured by the R & D investment required for unit energy consumption [16]. In terms of the R & D investment, green technology innovation is measured by the number of scientific and technological personnel [17]. This paper chooses the number of green patent licenses to measure the degree of green technology innovation.

2.2.3. The Core Explanatory Variable: The Transfer of Air Pollution-Intensive Industries like Thermal Power

Industrial transfer means that governments at various levels regulate the changes in resource supply and product demand via policies to fuel regional coordinated development, upgrade industrial structure, and boost economic development, which triggers the spatial transfer of industrial investment and production base, and ultimately reshapes the industrial economy and the number of enterprises in the region [18]. In the transfer-in regions, when the income level of air pollution-intensive industries remains low, the inflow of industries will exacerbate environmental pollution [19], and industrial transfer will have a significant spillover effect on pollution [20]. This paper takes the transfer of air pollution-intensive industries like thermal power as the core explanatory variable to research its impact on environmental pollution.

2.2.4. The Control Variables: Industrial Structure, per Capita GDP, per Capita Car Ownership, and Population Density

On the basis of the relevant literature, this paper takes industrial structure, per capita GDP, per capita car ownership, and population density as the control variables, which all affect fog–haze pollution. Presently, China stays at a critical stage of industrial restructuring, where contradictions arising from industrial structures are prominent and overcapacity and shortage coexist. China’s extensive industrial economic development mode has created environmental pollution [21], and the environmental pollution in areas with a large proportion of secondary industry is more severe [22]. The impact of the industrial structure on the temporal and spatial evolution of environmental pollution varies greatly, and different cities in the Beijing–Tianjin–Hebei region have great potential in industrial structure adjustment [23].

Simultaneously, this paper uses per capita GDP to measure the scale of economic development. When economic development reaches a high level, the emissions of industrial pollutants also have a negative impact on environmental pollution. As the number of motor vehicles soars, the exhaust or particulate matter emitted from the motor vehicles and the fuel consumption caused by vehicle congestion probably lead to air pollution [24]. Additionally, economic growth increases population density, which intensifies the emissions of sulphur dioxide and carbon dioxide and has a negative impact on environmental pollution [25].

2.3. Data Sources

Considering the impact of the large-scale shut down of polluting enterprises in China by administrative means in 2017 and 2018 on this study, this paper selected panel data of 31 provinces, municipalities, and autonomous regions from 2000 to 2017. The data on the haze pollution come from the annual average global PM2.5 concentrations published by the Socioeconomic Data and Application Center of Columbia University and based on satellite monitoring (https://www.earthdata.nasa.gov/centers/sedac-daac, accessed on 13 January 2025); the number of green patents authorized comes from the China National Intellectual Property Administration and is manually sorted out by the author; other data are derived from the China Environmental Statistics Yearbook, China Urban Statistics Yearbook, China Urban Construction Statistics Yearbook, China Regional Economic Statistics Yearbook, and provincial statistics yearbooks.

3. The Analysis of Empirical Results

3.1. Descriptive Statistics

Table 2. Descriptive statistics of variables.

Variable	Mean Value	Standard Deviation	Minimum Value	Maximum Value
Geographic–Mean PM2.5 (lny1)	3.579207	0.457681	1.960095	4.434382
Population–Mean PM2.5 (lny2)	3.753287	0.42793	2.091864	4.486387
Green technology innovation (lnx)	5.883514	1.719067	0.693147	9.835904
The transfer of air pollution-intensive industries Thermal power (k)	−1.12 × 10−7	0.003153	−0.01602	0.028452
Industrial structure (lnm1)	−0.78809	0.202389	−1.66	−0.4865
Per Capita GDP (lnm2)	10.02109	0.844047	7.886667	11.76752
Per Capita car ownership (lnm3)	−2.43614	0.931238	−4.67893	−0.66344
Population density (lnm4)	5.429892	1.320673	1.968187	8.249705

shows the results of the descriptive statistics of the variables:

As shown in Table 2, the standard deviation of the geographically averaged PM2.5 variable is 0.457681, the minimum value is 1.960095, and the maximum value is 4.434382, and the pollution level of each region has a large spatial difference in the sample period. The mean value of the green technology innovation variable is 5.883514, the minimum value is 0.693147, and the maximum value is 9.835904, indicating that the level of technological innovation in China is at a medium level. Combined with the relevant research of our research group, it is found that there is a huge gap in the level of green technology innovation in various regions of China, and the spatial differentiation is obvious [26].

3.2. The Analysis of the Temporal–Spatial Characteristic of Fog–Haze Pollution

3.2.1. The Time Sequential Characteristics of PM2.5

Figure 1 shows the average concentration value and median of PM2.5 in various provinces from 2001 to 2018. The dashed lines in the figure represent the turning points. Evidently, PM2.5 underwent three stages.

The first stage lasted from 2001 to 2007, when MP2.5 showed an upward trend. The reason was that in the 21st century, China’s fast economic development mainly relied on the support of the secondary industry, especially heavy industry. In this stage, the scale of the air pollution industry continuously expanded. The development model of the extensive economy with high investment and high pollution generated high energy consumption, emitted a substantial number of air pollutants, and increased the concentration of PM2.5. Meanwhile, after 2000, China experienced a rapid yet short-term process of urbanization as well as intense human activities, which induced the pollution of the atmospheric environment. The emissions of vehicle exhaust in various regions shot up and produced PM2.5. The second stage lasted from 2007 to 2012, when PM2.5 showed a downward trend. Governments laid down a series of policies and measures on environmental protection, which helped to lower the level of PM2.5. Coupled with the financial crisis in 2008, the demand for industrial products decreased, and so did the emissions of the pollution from industrial enterprises. The third stage was from 2012 to 2017, and this stage of PM2.5 showed up and down fluctuations; the general trend of fluctuations was up in 2012–2014, and its reasons are as follows: First, the development and implementation of the weight method to detect haze pollution was developed and implemented in 2011 and 2012 to seriously mislead China’s haze prevention and control work. In order to reduce the weight of particulate matter in the air in China’s haze-polluted areas, water vapour was constantly sprayed into the air (the more serious the haze the more water vapour was sprayed in the area); the water sprayed was tap water and recycled water, resulting in a large number of ultrafine particles that were contained in this water being sprayed into the air; in order to reduce the weight of particulate matter and the use of electrostatic precipitation methods in polluting enterprises, electrostatic precipitation was achieved mainly through particulate matter with a negative charge adsorbing particulate matter, with the larger particulate matter that is more negatively charged being more likely to be adsorbed, and vice versa, which also caused dozens to hundreds (the number of) of ultrafine particulate matter to be discharged up to the standard [26]. Secondly, the simple and rough management aggravates the degree of haze pollution. For example, in order to prevent polluting enterprises from using the bypass flue and intentionally not using the desulphurisation device to emit exhaust gas, on 17 June 2010, the environmental protection department issued ‘a notice on the implementation of lead sealing on thermal power enterprises in desulphurisation facilities for bypass flue baffle’, and decided to implement lead sealing of all ‘thermal power enterprises desulphurisation facilities using bypass flue baffle’. All levels of the environmental protection department and the power group companies should actively encourage thermal power enterprises to gradually remove the built desulphurisation facilities’ bypass flue, all new coal-fired units should not be equipped with desulphurisation bypass flue. Thermal power generation accounts for 70% of China’s total power generation, and the coal it uses accounts for nearly 50% of China’s total coal use. After the cancellation of the bypass flue system, the heat exchanger element of the flue gas heat exchanger (GGH) would be easily scaled, corroded, or blocked when the FGD unit is in operation. With China’s introduction of foreign wet FGD technology, there was not a clear provision for the flue gas moisture content, and therefore, there was a requirement of enterprises to cancel the GGH, and the environmental protection department therefore also cancelled the introduction of FGD technology when the temperature control requirements for the tail gas emissions. Following the cancellation of the GGH, the thermal power plants did not have to dewater the flue gas emitted by them (thermal power companies are using the recycled water, and the recycling of particulate matter in the water is saturated), resulting in the water vapour being self-contained. Later, a large amount of ultrafine particles are discharged into the air at no cost; these large amounts of water vapour and sky-high ultrafine particles in the air, meeting the static weather, caused the concentrated outbreak of haze in China in 2012. The layered environmental management resulted in a number of places having too high standards of offsetting emissions, and in order to meet the standard emissions of ammonia spraying, the relevant enterprises led to the ammonia escape phenomenon which is serious, resulting in secondary ultrafine particulate matter. The amount of secondary ultrafine particulate matter has increased dramatically, with secondary particulate matter accounting for more than 50% of the heavy haze days in some places. On 1 January 2013, after the implementation of strict online monitoring and punitive measures using the weight method in China, thermal power and other high haze polluters began to rapidly launch desulphurisation, decommissioning, and dedusting facilities, which resulted in the emissions of more ultrafine particulate matter and water vapour into the air. After 2014, China’s implementation of environmental protection tariffs on thermal power enterprises incentivised thermal power enterprises to build or renovate desulphurization and dedusting facilities, and these newly built and renovated desulphurization and dedusting facilities were gradually put in place to counteract the impact of China’s haze prevention and control measures [26]. There was a decline in 2015–2017 due to the fact that after the severe haze event, the State Council introduced the action plan for the Prevention and Control of Air Pollution and other air pollution prevention and control measures, the industrial structure was upgraded, the use of high energy-consuming resources began to decline, and the use of green and clean energy was widely promoted in production and life. In particular, a large number of scattered polluting enterprises shut down and restricted production. For example, in order to complete the five-year goal of combating haze pollution, Hebei shut down 68,000 and remediated 38,000 scattered polluting enterprises in 2017, Shandong shut down and transferred more than 100,000 scattered polluting enterprises in the two years from 2017 to 2018, Henan comprehensively remediated and banned 83,000 scattered polluting enterprises in 2017, Tianjin shut down 15,000 enterprises in 2017, and so on. China’s 2017 heating season production restrictions on hundreds of thousands of enterprises in Beijing, Tianjin, Hebei, and other regions of the autumn and winter comprehensive treatment of air pollution offensive action plan for Hebei, Shanxi, Shandong, and Henan, with the four provinces having a ‘2 + 26’ city heating season steel production capacity limit of 50%. For example, the annual output value of Hebei Iron and Steel in 2018 was 116.62 billion yuan. However, due to the production restrictions of Hebei Iron and Steel, the losses during the heating season may exceed 150 billion yuan. An electrolytic aluminum plant production limit of 30% or more, ceramics, glass wool, gypsum board, and other industries in the heating season needing to implement production suspension, and other cities referring to the independent decision to give the minimum production limit. PM2.5 concentration began to enter a slow decline; however, due to the excessive economic and social impact of haze pollution management in 2017 and 2018 partially cancelling the one-size-fits-all haze prevention and control and management measures, haze pollution rose again after 2018. PM2.5 from 2001 to 2018 passed through three stages. It showed an upward trend from 2001 to 2007, a downward trend from 2007 to 2012, a fluctuation from 2012 to 2017, an increase from 2012 to 2014, and then a decline and an increase. The increase in 2012 was due to the implementation of the haze detection policy by weight method, resulting in a large number of water vapour and ultrafine particles emitted at no cost, while the decrease from 2015 to 2017 was mainly due to the implementation of prevention and control measures such as closing and limiting production enterprises.

3.2.2. Global Spatial Autocorrelation Test of PM2.5

Moran’s index method (Moran’s I) is a statistical method used to measure spatial autocorrelation, which determines the spatial distribution pattern of data by analyzing the similarity between the attribute value of a spatial unit and the attribute value of its neighbouring unit. Under the assumption of a spatial random distribution, the expected value of the Moran index is as follows: E(I) = −1/(n − 1). The Z value is used to measure the degree to which the Moran index deviates from its expected value. The p-value represents the probability of observing the current Moran index value or a more extreme value under the hypothesis of a spatial random distribution, and the smaller the p-value, the stronger the evidence for rejecting the spatial random distribution hypothesis.

Table 3. 2001–2017 Moran index of PM2.5.

Year	Moran’s I	Expected Value	Standard Deviation	Z Value	p Value
2001	0.343	−0.033	0.110	3.407	0.001
2002	0.358	−0.033	0.110	3.544	0.000
2003	0.410	−0.033	0.110	4.039	0.000
2004	0.334	−0.033	0.110	3.349	0.001
2005	0.404	−0.033	0.110	3.986	0.000
2006	0.393	−0.033	0.109	3.902	0.000
2007	0.446	−0.033	0.109	4.380	0.000
2008	0.391	−0.033	0.110	3.873	0.000
2009	0.374	−0.033	0.109	3.727	0.000
2010	0.373	−0.033	0.110	3.709	0.000
2011	0.413	−0.033	0.110	4.071	0.000
2012	0.377	−0.033	0.110	3.745	0.000
2013	0.407	−0.033	0.109	4.027	0.000
2014	0.357	−0.033	0.110	3.561	0.000
2015	0.428	−0.033	0.110	4.207	0.000
2016	0.423	−0.033	0.109	4.168	0.000
2017	0.396	−0.033	0.110	3.905	0.000

shows the Moran’s I of PM2.5.

The formula of global Moran’s I is expressed as follows.

$$\mathrm{M}\mathrm{o}\mathrm{r}\mathrm{a}\mathrm{n}’\mathrm{s}\mathrm{I}={\displaystyle \frac{\sum _{\mathrm{i}=1}^{\mathrm{n}}\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{W}}_{\mathrm{i}\mathrm{j}}({\mathrm{g}\mathrm{a}\mathrm{p}\mathrm{m}}_{\mathrm{i}}-\overline{\mathrm{g}\mathrm{a}\mathrm{p}\mathrm{m}})({\mathrm{g}\mathrm{a}\mathrm{p}\mathrm{m}}_{\mathrm{j}}-\overline{\mathrm{g}\mathrm{a}\mathrm{p}\mathrm{m}})}{{\mathrm{S}}^2\sum _{\mathrm{i}=1}^{\mathrm{n}}\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{W}}_{\mathrm{i}\mathrm{j}}}}$$

(2)

In particular, in ${\mathrm{S}}^2={\displaystyle \frac{1}{\mathrm{n}}}\sum _{\mathrm{i}=1}^{\mathrm{n}}({\mathrm{g}\mathrm{a}\mathrm{p}\mathrm{m}}_{\mathrm{i}}-\overline{\mathrm{g}\mathrm{a}\mathrm{p}\mathrm{m}})\mathrm{a}\mathrm{n}\mathrm{d}\overline{\mathrm{g}\mathrm{a}\mathrm{p}\mathrm{m}}={\displaystyle \frac{1}{\mathrm{n}}}\sum _{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{g}\mathrm{a}\mathrm{p}\mathrm{m}}_{\mathrm{i}},{\mathrm{g}\mathrm{a}\mathrm{p}\mathrm{m}}_{\mathrm{i}}$ stands for the PM2.5 value of the i-th province, and ${\mathrm{W}}_{\mathrm{i}\mathrm{j}}$ stands for the adjacent space weight matrix. As the statistical results in Table 2 suggest, the Moran’s I of PM2.5 proves positive, and the p value reaches less than 0.01, indicating that PM2.5 in various provinces has a high degree of spatial positive correlation. The value of Moran’s I fluctuates. From 2001 to 2007, Moran’s I was in the stage of a rise in volatility. This was mainly because in the early 20th century, China adopted an extensive development model that centred on high energy consumption and high-pollution industries. From 2008 to 2017, Moran’s I rose first and then declined. This was mainly because the economic crisis in 2008 bred China’s CNY four trillion investment into infrastructure to stimulate economic development, hedging against the impact of air pollution-intensive industries like thermal power. For the 2008 Beijing Olympics, five provinces, autonomous regions, and municipalities, including Hebei, Tianjin, Shanxi, Shandong, and Inner Mongolia, implemented measures such as joint prevention and control and the vigorous development of clean energy, which led to a significant decrease in the spatial correlation of RSP, which lasted until 2011. After the 2008 economic crisis in China, to stimulate the economic development of the CNY 4 trillion investment into other industries, to play the role of diffusion, China’s economy returned to the fast lane of disorganized development, and coupled with the causes of haze pollution being unclear, the use of the weighting method to detect the haze policy implementation was not effective and disorganized management and other reasons made the haze pollution serious [26].

3.2.3. The Agglomeration Regions of PM2.5

Figure 2 shows the agglomeration of PM2.5.

As Moran’s I confirms, PM2.5 has a positive spatial correlation. This paper chooses the inflection point of the time sequential characteristic of PM2.5 to explore the agglomeration of PM2.5. The low self-purification capacity of the atmospheric environment coupled with the rough management of thermal power and other air pollution-intensive industry aggregation results in thermal power and other air pollution-intensive industry aggregation being too high compared to the self-purification capacity of the atmospheric environment, leading to serious haze pollution, specifically from four provinces and cities in the region in 2001, to seven provinces and cities in 2007, and then to six provinces and cities in 2012.

From the LISA cluster diagram of PM2.5 in 2001, it can be seen that there are four main provinces in the high–high neighbourhood, namely Beijing, Hebei, Shandong, and Henan, and the rapid growth of the rough economy in the Beijing–Tianjin–Hebei region and Henan Province is accompanied by huge energy consumption; Beijing is not conducive to the diffusion of pollutants due to its geographic location, Hebei Province is rich in mineral resources and has convenient transport, and the industrial structure is dominated by the secondary industry, which causes serious industrial pollution. The four cities of Beijing, Hebei, Shandong, and Henan are geographically close to each other, and air pollutants will flow into each other as the air flows; there are only two provinces that are close to each other, namely Jilin and Yunnan, which have beautiful environments, a strong self-purifying ability of the atmosphere, are located in the southwest of China, and did not yet have a perfect transport network in 2001, with fewer polluting industries and relatively low concentrations of PM2.5. The concentration of PM2.5 is also relatively low, the environmental carrying capacity of the nearby area of Yunnan Province is high, and there are fewer polluting industries, thus forming a low–low neighbouring province; Jilin Province has gradually achieved industrial upgrading in the national industrial strategy policy, and the advantages of heavy industry which it previously possessed have gradually disappeared, and although there is a certain degree of industrial pollution, the pollution is not agglomerated, thus making it a low–low neighbouring province.

According to the LISA agglomeration map of PM2.5 in 2007, there are mainly seven high–high neighbouring provinces, i.e., Jiangsu, Shandong, Beijing, Hubei, Henan, Anhui, and Hebei. This is mainly due to the low self-purification capacity of the atmospheric environment in this region, and the concentration of air pollution-intensive industries, such as thermal power, is too high to exceed the self-purification capacity of the corresponding region. For example, there are too many high-polluting industries such as thermal power and iron and steel in Hebei Province, and the air self-purification capacity is low, which makes the PM2.5 concentration in the southern part of Hebei Province the highest.

Meanwhile, the transfer of polluting industries from Beijing to Hebei aggravates the PM2.5 concentration in Hebei. Henan possesses rich resources, which provide raw materials and fuel for the development of heavy industry. Henan lies in the Central Plains of China, with convenient transportation. This facilitates the transportation of products and intensifies the production of PM2.5. Rich in mineral resources, Anhui attracts high energy consumption industries and increases the emission of pollutants. As required by industrial development, Jiangsu and Hubei develop heavy industry, which intensifies the PM2.5 as well as the spatial spillover effect of pollution. There are two low–low neighbouring provinces, i.e., Jilin and Yunnan. Guided by national policies, Jilin energetically adjusts its industrial structure. As polluting industries decrease, the agglomeration of industrial pollution weakens. Yunnan lies in southwestern China. The inconvenient transportation means a disadvantage of industrial development. In addition, Yunnan boasts a beautiful environment, high forest coverage, and a strong atmospheric–environmental self-purification capacity. Coupled with low-level pollution in neighbouring provinces, the concentration of PM2.5 remains relatively low in Yunnan. The LISA agglomeration map of PM2.5 in 2012 has only one less province in the high–high neighbouring provinces, and the others are the same as the LISA agglomeration map of PM2.5 in 2007. From the LISA concentration map of PM2.5 in 2017, it can be seen that the provinces in the high–high neighbourhood are Hebei, Beijing, Shandong, Jiangsu, Henan, and Anhui. Driven by the high concentration of traditional polluting industries in Hebei and Inner Mongolia, the level of industrial polluting industries in the surrounding resource-rich provinces such as Shandong and Henan have also increased significantly, and the corresponding concentration of PM2.5 is also relatively high. The only provinces adjacent to the low and low are Sichuan and Yunnan, which have a weak industrial base and a low level of industrial agglomeration, and the surrounding provinces and regions also have a low level of air pollution-intensive industrial agglomeration.

3.3. Threshold Panel Regression

According to the research on the spatial correlation effect analysis of green technology innovation and haze pollution by the spatial autoregressive model, there is a spatial correlation between green technology innovation and haze pollution [26]. To analyze the threshold of Geographic–Mean PM2.5, this paper firstly determines the number of thresholds, and tests the results with the self-sampling test method. According to the test results of

Table 4. Threshold effect tests.

y1	The Critical Value
	F	p-Value	10%	5%	1%
Single Threshold	12.57	0.0133	7.7101	9.148	14.4517
Double Threshold	12.57	0.0067	7.3247	9.229	11.8852
	6.58	0.2400	9.3609	11.675	14.4909

, the single threshold effect passed the 1 percent significance test, and the double threshold effect did not pass the 10 percent significance test, as shown in Table 4 and

Table 5. Single threshold and double threshold regression results.

	(1)	(2)
	lny1	lny1
lnx	−0.061 ***	−0.060 ***
	(−2.721)	(−2.682)
lnm1	−0.130 *	−0.125 *
	(−1.945)	(−1.863)
lnm2	0.211 ***	0.201 ***
	(4.763)	(4.509)
lnm3	−0.037	−0.031
	(−0.782)	(−0.646)
lnm4	−0.031	−0.043
	(−0.210)	(−0.290)
0._cat#c.k	9.291 **	7.720 **
	(2.437)	(2.007)
1._cat#c.k	−8.297 ***	28.553 **
	(−3.036)	(2.528)
2._cat#c.k		−8.170 ***
		(−2.999)
_cons	1.822 **	2.002 ***
	(2.487)	(2.706)
N	510	510
R2	0.163	0.168
adj. R2	0.099	0.102

t statistics in parentheses; * p < 0.1, ** p < 0.05, and *** p < 0.01.

below. Therefore, the hypothesis proves valid when the variable k has a single threshold effect.

When green technology innovation is below the threshold, its inhibitory effect on haze pollution is small. As shown in Table 5, when green technology innovation remains lower than the threshold value of −0.0017, the impact coefficient of green technology innovation on the Geographic–Mean PM2.5 reaches 9.291. When green technology innovation reaches a low level, it significantly increases the Geographic–Mean PM2.5 of air pollution-intensive industries like thermal power. The reason mainly lies in that this happens mostly in developed regions, where the trend of the transfer of air pollution-intensive industries like thermal power to less developed regions is not evident. When developed regions achieve economic progress, air pollution problems that arise from air pollution-intensive industries like thermal power become prominent. Simultaneously, less developed regions suffer from environmental pollution and neglect green technology innovation. Therefore, green technology innovation has a relatively low inhibitory effect on PM2.5. The lack of initial green capital investment and poor decisions on investment have worsened environmental pollution.

When green technology innovation is above the threshold, its inhibitory effect on haze pollution is enhanced. As shown in Table 5, when green technology innovation remains higher than the threshold value of −0.0017, the impact coefficient of green technology innovation on Geographic–Mean PM2.5 reaches −8.297. When green technology innovation reaches a high level, it significantly inhibits the Geographic–Mean PM2.5 of air pollution-intensive industries like thermal power. The reason mainly lies in that as the impact of air pollution-intensive industries like thermal power on the atmospheric environment intensifies, developed regions improve the intensity of environmental regulation, and the cost of air pollution-intensive industries like thermal power keeps rising. The demand for industrial upgrading in developed regions augments the trend of the transfer of air pollution-intensive industries like thermal power to less developed regions. In less developed regions, the green technology of air pollution-intensive industries like thermal power matures. With stricter environmental regulations and a stronger environmental self-purification capacity, green technology innovation can effectively curb PM2.5 and ease atmospheric–environmental pollution. When green technology innovation is lower than the threshold value, its inhibitory effect on haze pollution is small, and when green technology innovation is higher than the threshold value, its inhibitory effect on haze pollution is significantly enhanced.

4. Research Conclusions and Suggestions on Policies

4.1. Research Conclusions

Based on the analysis of the causes of haze and the analysis of the characteristics of spatial and temporal distribution, this article used the panel data from 2000 to 2017 to verify the threshold effect of green technology innovation on the transfer of haze pollution in air pollution-intensive industries such as thermal power, etc., and drew the following conclusions:

First, the great haze outbreak is due to unclear causes and disorganized management. China’s big haze outbreak and the serious reasons for weight detection haze policy implementation seriously misled the Chinese haze prevention and control work, such as water vapour containing a large number of ultrafine particles being sprayed into the air, polluting enterprises in order to reduce the weight of the particles, and the use of the electrostatic precipitation method caused by dozens to hundreds of (number) ultrafine particles being discharged above the standard. Secondly, simple and rough management has aggravated the degree of haze pollution. In order to prevent polluting enterprises from using the bypass flue and deliberately not using the desulphurisation device emissions exhaust, environmental protection departments mandated lead sealing by thermal power enterprises’ desulphurisation facilities of bypass flue baffle and the cancellation of the flue gas heat exchanger, resulting in the cost-free emissions of these water vapours and a large amount of ultrafine particulate matter. Third, the cascading management results in a number of places having emission standards that are too high, and the ammonia escape is serious, resulting in a sharp increase in the amount of secondary ultrafine particulate matter, etc.

Secondly, PM2.5 is mainly concentrated in areas with low environmental self-purification capacity. The environmental self-purification capacity of the lower regions is mainly in Beijing, Tianjin, Hebei, and other surrounding areas, and such areas are the main gathering places of PM2.5.

Thirdly, the diffusion effect of regions with a high agglomeration of air pollution-intensive industries like thermal power becomes prominent. This basically results from the fact that the agglomeration of pollution-intensive industries affects industrial development and significantly increases the agglomeration of industrial pollution in the neighbouring provinces with rich resource reserves, and the degree of fog–haze pollution intensifies.

Fourthly, when the green technology innovation is lower than the threshold value, it has no inhibitory effect on the haze pollution of industrial transfer, and only when green technology innovation proves higher than the threshold value can it remarkably inhibit fog–haze pollution in industrial transfer.

4.2. Suggestions on Policies

The great haze outbreak was caused by the unclear cause of haze, disorganized management, and layer upon layer of various management styles. The most effective measures to prevent and control haze pollution are to innovate technological innovations in desulfurization, destocking, and dust removal, and strictly control the quantity and quality of water vapor emissions and scientifically transfer haze pollution industries.

First, the scientific planning of the transfer of polluting industries to undertake the development of the site.

In order to achieve the goal of scientifically planning industrial transfer to sustain development, governments should establish and improve the overall coordination and regulation mechanisms of industrial transfer, analyze the differentiation advantages of resources in various regions, and formulate the overall goals of industrial transfer by researching the relationship between the transfer-in and transfer-out regions and respecting local conditions. After determining the overall goals, governments should set the goals of industrial transfer in various regions in line with the environmental self-purification capacities of the transfer-in regions. Simultaneously, in combination with the goal of high-quality industrial development, governments can consider the overall features of ecological and environmental systems in a coordinated way to achieve scientific, effective, and green industrial transfer. Governments can build an exchange platform for industrial development in the Beijing–Tianjin–Hebei region, make full use of the advantages in the Internet era, release the information on industrial transfer in a timely manner, provide information channels for local governments in the transfer-in regions, facilitate the transfer-in region’s learning from the experience of the development of green industry in developed regions, and improve the efficiency and scientificalness of industrial transfer. Additionally, in the process of industrial transfer, the transfer-in regions should seize the opportunity to optimize the industrial chain, strengthen the overall management of key industrial platforms and transfer projects for industrial transfer, and ensure the practical effect of industrial transfer and sustainable development.

Secondly, effectively transferring air pollution-intensive industries like thermal power to the Beijing–Tianjin–Hebei region.

When air pollution-intensive industries like thermal power are transferred, governments should clarify the correlation of the industrial chain in the transfer-in regions and avoid the transfer of air pollution-intensive industries like thermal power to poor regions. The transfer-in regions can formulate the access principle for industrial transfer and selectively accept air pollution-intensive industries like thermal power that meet national industrial policies and pollutant emission standards. Meanwhile, the transfer-out regions should reduce the possibility of pollution transfer in the process of industrial transfer, and the transfer-in regions should give priority to introducing industries with environmental protection and technological advantages to ensure that both parties reduce the possibility of pollution transfer. In addition, in the process of the transfer of air pollution-intensive industries like thermal power, governments need to determine the bearing capacity of resources and the environment in the transfer-in regions, formulate differentiated policies on industrial transfer based on such capacity, and reasonably transfer and accept relevant air pollution-intensive industries like thermal power based on environmental self-purification capacity so as to reduce the emissions of pollutants, guarantee green industrial transfer, and strengthen the protection of the ecological environment.

Thirdly, formulating targeted green technology innovation and strengthening policies to scientifically manage haze pollution.

Desulphurisation and dust removal technological innovation is the most effective way to combat haze pollution, and we need to increase the research and development of desulphurisation and dust removal technological innovation. Firstly, we should improve the wet desulphurisation technology and research and develop the dry desulphurisation technology so as to achieve a reasonable substitution of wet desulphurisation. This can be achieved by adding baghouse dust collectors, GGH equipment, and flue towers in one, increasing the flue gas emission temperature, enhancing the chimney emission capacity, realizing high altitude emission, and enhancing the flue gas diffusion capacity; secondly, we should promote the technological innovation of desulphurisation and denitrification processes and scientifically control ammonia emissions. We should encourage the optimization of the desulphurisation system and the research and development of ammonia escape monitoring technology through the introduction of advanced control algorithms, process optimization, automated intelligent ammonia spraying, etc., the research and development of stable and reliable ammonia escape monitoring technology in complex flue gas environments, in particular, automatic monitoring technology, the reasonable setting of the monitoring point position, the improvement of the response speed between the ammonia escape monitoring system and the ammonia spraying control system, the innovation of more effective oxidizing agents other than ammonia, and the denitrification process. Optimization reduces the possibility of ammonia escape and other innovative denitrification processes. Third, increasing the intensity of the research and development of the core technology of ultrafine particulate matter dust removal, such as electrostatic dust removal and agglomeration technology combined to strengthen the combination of bag dust removal technology and electrostatic dust removal, innovative bag type dust removal technology to increase the core technology of ultrafine particulate matter dust removal, and the R & D strength to effectively improve the dust removal efficiency.