Effect and Mechanism of Environmental Decentralization on Pollution Emission from Pig Farming—Evidence from China

Abstract

Pollutants produced by pig breeding are among the important pollutions. It is necessary to explore the factors influencing the emission intensity of pollutants from pig breeding and find ways to decrease them. Using the provincial panel data of China from 2001 to 2017, this paper analyzed the mechanism impact of environmental decentralization on pollution emission from pig farming. The results showed that environmental decentralization could significantly reduce the emission intensity of pollutants from pig farming. Decentralization of environmental monitoring had a certain effect on reducing such emission intensity, while decentralization of environmental administration has not played this role. In addition, the scale of pig farming can effectively strengthen the effect of environmental decentralization on pollution reduction in pig farming. From the perspective of regional heterogeneity, environmental decentralization can reduce such emission intensity in restricted development areas, potential growth areas, and moderate development areas. Also, decentralization of environmental monitoring can also, obviously, reduce pollutant emissions in potential growth areas. The research results have reference value for determining the reasonable level of environmental decentralization between regions and improving environmental regulation policies.

1. Introduction

The environmental problems caused by the pollution of livestock and poultry (LAP) breeding are becoming more and more serious and have become a hot topic in the world [1]. With the development of large-scale livestock breeding in China, livestock production in China is no exception. According to the Bulletin of the Second National Survey of Pollution Sources published by the Chinese government in 2020, the chemical oxygen demand, ammonia nitrogen, total nitrogen, and total phosphorus emissions from China’s LAP industry accounted for 93.76%, 51.30%, 42.14%, and 56.46% of agricultural source emissions, respectively, in 2017. Livestock and poultry farming is an important source of greenhouse gases. The total emissions of ${\mathrm{CO}}_2$,${ \mathrm{N}}_2\mathrm{O},$ and ${ \mathrm{CH}}_4$ per pig during its life cycle are 6.75 kg [2]. Figure 1 shows the growth and change in pollutant emission intensity from pig farming in China’s provinces (municipalities) during 2007–2017. The data from this study show that 22 of the 30 provinces increased their pollutant emission intensity, of which 16 provinces experienced a greater increase in pollutant emission intensity. This showed the seriousness of the pollution emission problem of LAP farming, which is a source of environmental pollution that produces waste that has a negative impact on environmental resources (water, soil, air, etc.) and causes the greenhouse effect [3,4]. According to Chmielowiec-Korzeniowska et al. [5], LAP breeding pollutants affect the health of breeders and increase the health threats for other residents. Taking reasonable measures to deal with the pollution emission of LAP breeding is an imperative approach to achieving sustainable development of agriculture [6].

The Chinese government actively solves the environmental problems caused by pollution emissions from LAP farming. Relevant departments of the Chinese government have strengthened supervision and issued a series of policies and regulations to control pollution from LAP farming. These include the Regulations on Pollution Prevention and Control of Large-Scale LAP Farming, Emission Standards of Pollutants from LAP Farming, and Notice on Further Regulating the Zoning and Management of Prohibited LAP Farming to Promote the Development of Pig Production, etc. This regulates the utilization and discharge processes of LAP farming wastes to transform China’s LAP farming. The China Animal Husbandry and Veterinary Yearbook shows that the scale of pig farming in China has significantly improved from 2007 to 2020. The production of pig manure in China has exceeded 600 million tons and the actual utilization rate is low, causing serious pollution. Moreover, China is the largest consumer of pork. Pig breeding plays an important role in China’s LAP breeding [7], hence, the significance of studying pollution emissions from pig farming. Jiang postulated that China has a special system of environmental decentralization (ED) [8], thus the distribution of power between central and local governments in ecological and environmental governance affects pollution control in pig farming. What is the impact of ED on pollution emissions from pig farming? What is the mechanism of action? The purpose of this paper is to analyze the panel data of 30 provincial administrative units (excluding Tibet, Hong Kong, Macao, and Taiwan) in China from 2001 to 2017, and to reveal the mechanism of ED on the emission intensity of pollutants from pig breeding, so as to provide reference for the formulation of environmental policies.

This article has several contributions, as follow. This paper explores the impact mechanism of ED on pollutant emissions from hog farming, and, further, clarifies different types of ED, which make a useful policy assessment of ED in the context of livestock pollution. On the other hand, by analyzing the moderating effect of the pig farming scale, this study provides a new perspective on how ED can better contribute to pollution control.

The structure of this study is as follows: the second section is the literature review and the theoretical analysis. The third is the research methods. The next section is the results and discussion, and the last is the conclusions and implications.

2. Literature Review and Theoretical Hypothesis

As for the impact of ED on pollution control, there are, mainly, two kinds of opinions in the academic circle at present. One is that ED has disadvantages in pollution control, and the other is that ED will have favorable effects.

The opposing viewpoint of ED holds that the management mode of ED is not conducive to the central government, owing to the overall planning. Under the decentralized governance model, local governments have a “free rider” behavior in environmental governance based on their own interests [9]. ED may lead to rent-seeking corruption of local governments, which would influence the efficiency of local governments in dealing with environmental emergencies and lead to regional pollution [10,11]. ED will affect the use efficiency of green factors and bring a negative influence on regional green innovation efficiency [12,13]. The unified supply of environmental public goods by the central government can effectively avoid the entrustment cost caused by ED. This purpose is to bring full play to the advantages of economies of scale, increase the efficiency of ecological environment management, and fully meet the social demand for a high-quality natural environment [14]. In order to pursue short-term political achievements, local governments may neglect environmental governance and protection, or even over-use and sacrifice environmental resources, reduce environmental standards, and create recompenses in some industries, thus forming a “race to the bottom” effect in local environmental governance, which further worsens the environment [15,16]. Moreover, local governments lack the motivation to clean up the environment as the implementation efficiency of environmental policies is not high, and ED will, eventually, increase carbon emissions [17].

Proponents of ED believe that ED can clarify the powers and responsibilities of various sections of government and optimize environmental governance [18]. Hence, local governments have advantages in mastering and acquiring information related to the local environment and can accurately grasp the situation of environmental governance. On this basis, local governments can formulate their own environmental governance policies according to their own economic and geographical characteristics, which can have the effect of adapting to local conditions, so as to realize the overall effective allocation of resources [19]. Due to the continuous improvement of China’s environmental assessment system, environmental management has been valued by the local government. In this context, the ED model will enable local governments to form healthy competition in environmental governance, and environmental pollution will also be weakened [20]. Furthermore, ED can promote enterprises to increase environmental protection investment, which is conducive to enterprises’ green production innovation [21,22]. Also, ED can promote the continuous optimization of industrial structure and control environmental pollution while promoting and intensifying development [23]. The establishment of ED agencies can collect more effective information, coordinate environmental protection measures among different provinces, and significantly reduce the environmental pollution behaviors of enterprises at provincial boundaries [24].

Literature on the influencing factors of pollution emission from pig farming indicates that the current academic circle mainly discusses the influencing factors of pollution emission from the aspects of individual pig breeders, farming scale, relevant policies, and modes. From the perspective of individual breeders, their higher education level and easier access to information will strengthen their willingness to deal with pig breeding pollution, thus this would be conducive to the reduction of pig breeding pollution emissions [25]. Other studies show that the higher the risk to breeders, the more likely they are to adopt environmentally friendly LAP pollution emission treatment with a large investment. Similarly, the deeper the awareness of the environmental impact on human health, the more inclined farmers are to take actions to deal with the pollution emissions from LAP production [26]. The social and economic characteristics of smallholder farmers who manage individual farms have an impact on the pollution discharge of pig breeding. For example, individual farmers often adopt the mode of combination breeding, so as to enhance soil fertility and recycle the manure generated by pigs, hence, forming a circular ecological breeding mode [27].

The pig breeding scale has different impacts on the emission of pig breeding pollution. It was found that the medium-scale pig breeding subjects had the lowest utilization rate of aquaculture emissions and the highest pollution risk. The main body of small-scale and retail pig breeding has higher resource utilization, which causes the most direct pollution [27]. The environmental efficiency of large, medium, and small pig farming has improved over time [28]. The scale of pig farming affects pollution emission through intensive factors, which shows that, while the scale of pig farming increases, the progress of technology and the sharing of infrastructure have positive externalities which can effectively reduce the cost of pollution treatment of surrounding pig farming [29].

From the perspective of related policies and models, different policies and models have an impact on pollution emissions from pig farming. Biogas subsidies, livestock manure treatment technical support, and other policies have an emission reduction effect on pig breeding, while livestock manure emission technical standard policy has a poor emission reduction effect [30]. According to Zhu, government environmental regulation positively affects the recycling of pig manure [31]. Thus, cooperative organizations have no apparent influence on the carbon emission reduction of large-scale pig farming, and the “leading enterprise + farmer” model also has this role under certain conditions [32]. The strengthening of the government’s responsibility in the process of making relevant laws can promote the recycling of pig breeding pollution. Likewise, Chen and Li asserted that the strengthening of the government’s responsibility in the process of supervision will have a negative impact on the recycling of pig breeding pollution [33].

The previous research has laid the theoretical foundation for our research. Extant literature on ED mostly focused on its impact on industrial environmental pollution, but few studies centered on its impact on the agricultural field pollution emission from pig farming. Moreso, literature on the influencing factors of pollution in pig farming mostly focused on the scale of pig farming and seldom considered the impact of ED. In view of this, this paper discusses ED, pig farming scale, and pig farming pollution emissions in a unified context. Furthermore, it explores the mechanism of ED on pig farming pollution emissions, thus providing a reference to adjust the environmental management model.

In the model of ED, local governments have more power to govern the environment and formulate environmental policies. Local governments are better able to manage the environment according to regional preferences, have a stronger initiative in pollution control, and can more effectively supervise the discharge of pollution from pig farming within their jurisdiction [34]. Local governments can also leverage their information advantages to make pollution control more efficient. In the context of more effective environmental supervision and management, pig producers will adjust their production and investment decisions due to higher compliance costs, and put more resources into the upgrading of environmental protection equipment and development of environmental protection technology [35]. On the one hand, these measures will promote green technology innovation [36], and, on the other hand, improve the utilization of pig waste as a resource, so as to reduce pollution emissions from pig farming. ED is conducive to local governments’ careful supervision of environmental pollution and can encourage pig producers to choose environment-friendly breeding strategies with carbon emission reduction [29]. The decentralization of the central environmental authority can streamline the supervision and management process to a certain extent. This will reduce the cost of environmental regulation and enforcement, as well as ease resource constraints and break bottlenecks in technology research and development [37]. A decentralized environmental governance model can reduce the negative effects of multi-department orders and overcorrection, promote the development of the dual embeddedness governance model mediated by the pig industry association, and promote the development of the pig breeding model in the direction of green wisdom [38]. Therefore, this paper proposes:

Hypothesis 1.

ED can reduce pollutant emissions from pig farming.

Based on ED, large-scale pig producers are not only better able to promote green technology research and development but they also have financial advantages to accelerate the upgrading of environmental equipment and implement environmentally friendly investment strategies. They can give full play to the competitive advantages of the system and promote pollutant reduction in pig farming. This paper proposes:

Hypothesis 2.

The scale of pig farming can strengthen the effect of ED on pollution reduction in pig farming.

According to the current management of environmental affairs in China, the decentralization of environmental management in China can be divided into environmental administrative decentralization (EAD), environmental monitoring decentralization (EMD), and environmental supervision decentralization (ESD) [14]. EAD mainly includes the formulation of environmental policies and the arrangement of related personnel. Within their limited tenure, local government officials often choose projects with short investment cycles and quick returns to quickly enhance their performance and increase their chances of promotion, thus neglecting long-term investment in environmental protection projects. Under the role of EAD, local officials have more regulatory power, which may lower the standards of environmental protection in pursuit of economic development, resulting in more pollution [17]. EMD mainly includes the reform and perfection of the EMD system and the monitoring of environmental pollution behavior [8]. The environmental monitoring department is the most direct environmental pollution control department [17]. EMD helps local governments to quickly prevent pollution behavior and helps local governments to adopt their own monitoring methods according to regional differences. In addition, local governments have timely and accurate access to local information, and more environmental monitoring powers are conducive to their effective control of environmental pollution. ESD mainly involves the monitoring and evaluation of environmental governance. In the context of the increasingly perfect China’s environmental assessment system, local governments have more autonomy in environmental monitoring, which has a strong motivation to modify or hide environmental monitoring data [39]. Environmental monitoring has high requirements on government funds and technology. For local governments with limited financial resources and technology, ESD is not conducive to reducing pollution. Therefore, this paper proposes:

Hypothesis 3.

Among EAD, EMD, and ESD, only EMD can significantly reduce pollutant emission intensity of pig breeding animals.

3. Research Method

As shown in Figure 2, the horizontal axis represents the changes in ED values in different provinces from 2007 to 2017, and the vertical axis represents the changes in POLL values in different provinces from 2007 to 2017. From Figure 2, it can be preliminarily seen that the larger the ED, the smaller the POLL. Therefore, it is necessary to conduct multiple linear regression to further study the relationship between ED and POLL.

3.1. Sample and Model

Through the above theoretical analysis, it can be deduced that ED may have a negative impact on the pollutant emission intensity of pig farming. In order to empirically test the above hypothesis, the STATA software is used for regression analysis, and the model is set as follows:

$${\mathrm{POLL}}_{\mathrm{it}}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{\mathrm{ED}}_{\mathrm{it}-1}+{\mathsf{\gamma }\mathrm{CV}}_{\mathrm{it}}+{\mathrm{RE}}_{\mathrm{i}}+{\mathrm{YE}}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(1)

Model (1), $\mathrm{i}$ represents a region,$\mathrm{t}$ represents time, ${\mathsf{\beta }}_0$ is the intercept term, and ${\mathsf{\beta }}_1$ represents the correlation coefficient. ${ \mathrm{POLL}}_{\mathrm{it}}$ represents the emission intensity of pollutants from pig farming. ${\mathrm{ED}}_{\mathrm{it}-1}$ represents the degree of environmental decentralization in year $\mathrm{t}-1$ of region $\mathrm{i}$; ${\mathrm{CV}}_{\mathrm{it}}$ represents a group of control variables, mainly including environmental regulation, the abundance of pig feed supply, education level of farmers, carrying capacity of pig breeding land, transportation convenience, scientific and technological progress, disease risk, per capita income, and urbanization rate. ${\mathrm{RE}}_{\mathrm{i}}$ and ${\mathrm{YE}}_{\mathrm{t}}$ represent the fixed effect of region and year, severally. ${\mathsf{\epsilon }}_{\mathrm{it}}$ represents the error term.

In order to explore whether the pig farming scale affects pollutant emission intensity, the cross term of ED and pig farming scale is added to the model (1). ED is divided into EAD, EMD, and ESD [14]. Model (2) is set as follows:

$${\mathrm{POLL}}_{\mathrm{it}}={\mathsf{\theta }}_0+{\mathsf{\theta }}_1{\mathrm{ED}}_{\mathrm{it}-1}+{\mathsf{\theta }}_2{\mathrm{scale}}_{\mathrm{it}-1}+{\mathsf{\theta }}_3\mathrm{ED}\times \mathrm{scale}+{\mathsf{\sigma }\mathrm{CV}}_{\mathrm{it}}+{\mathrm{RE}}_{\mathrm{i}}+{\mathrm{YE}}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(2)

Model (2), $\mathrm{ED}\times \mathrm{scale}$ represents the cross term between the pig breeding scale and ED, EAD, EMD, and ESD. ${\mathsf{\theta }}_0$ is the intercept term, and ${\mathsf{\theta }}_{\mathrm{i}}\left(\mathrm{i}=1,2,3\right)$ represents the correlation coefficient.

3.2. Variable Selection

The explained variable was the emission intensity of pollutants from pig farming, which can be expressed by the sum of chemical oxygen demand, ammonia nitrogen, and total phosphorus [40].

3.2.1. Explanatory Variables

ED refers to the effective division of powers and responsibilities between the central government and local governments in the supervision and management of environmental protection, pollution prevention, and control. It is a management mechanism based on a decentralization system. In this paper, the number distribution and dynamic change characteristics of the personnel of environmental agencies in each region are used to measure the degree of ED among governments at all levels [14]. The measurement formula is set as follows:

$${\mathrm{ED}}_{\mathrm{it}}=\frac{{\mathrm{ESP}}_{\mathrm{it}}/{\mathrm{PA}}_{\mathrm{it}}}{{\mathrm{HSP}}_{\mathrm{t}}/{\mathrm{PA}}_{\mathrm{t}}}\left(1-\frac{{\mathrm{GDP}}_{\mathrm{it}}}{{\mathrm{GDP}}_{\mathrm{t}}}\right)$$

(3)

$${\mathrm{EAD}}_{\mathrm{it}}=\frac{{\mathrm{ESA}}_{\mathrm{it}}/{\mathrm{PA}}_{\mathrm{it}}}{{\mathrm{HSA}}_{\mathrm{t}}/{\mathrm{PA}}_{\mathrm{t}}}\left(1-\frac{{\mathrm{GDP}}_{\mathrm{it}}}{{\mathrm{GDP}}_{\mathrm{t}}}\right)$$

(4)

$${\mathrm{EMD}}_{\mathrm{it}}=\frac{{\mathrm{ESM}}_{\mathrm{it}}/{\mathrm{PA}}_{\mathrm{it}}}{{\mathrm{HSM}}_{\mathrm{t}}/{\mathrm{PA}}_{\mathrm{t}}}\left(1-\frac{{\mathrm{GDP}}_{\mathrm{it}}}{{\mathrm{GDP}}_{\mathrm{t}}}\right)$$

(5)

$${\mathrm{ESD}}_{\mathrm{it}}=\frac{{\mathrm{ESS}}_{\mathrm{it}}/{\mathrm{PA}}_{\mathrm{it}}}{{\mathrm{HSS}}_{\mathrm{t}}/{\mathrm{PA}}_{\mathrm{t}}}\left(1-\frac{{\mathrm{GDP}}_{\mathrm{it}}}{{\mathrm{GDP}}_{\mathrm{t}}}\right)$$

(6)

${\mathrm{ESP}}_{\mathrm{it}}$, ${\mathrm{ESA}}_{\mathrm{it}}$, ${\mathrm{ESM}}_{\mathrm{it}},$ and ${\mathrm{ESS}}_{\mathrm{it}}$ are the total number of environmental protection system personnel, environmental protection administrative personnel, environmental protection supervisory personnel, and environmental protection monitoring personnel in the year $\mathrm{t}$ of the $\mathrm{i}$ province, respectively. ${\mathrm{HSP}}_{\mathrm{t}}$, ${\mathrm{HSP}}_{\mathrm{t}}$, ${\mathrm{HSM}}_{\mathrm{t}},$ and ${\mathrm{HSS}}_{\mathrm{t}}$, respectively, represent the total number of personnel in the national environmental protection system in year $\mathrm{t}$, environmental protection administrative personnel, environmental protection supervisory personnel, and environmental protections monitoring personnel in year $\mathrm{t}$. ${\mathrm{PA}}_{\mathrm{it}}$ represents the population size of region $\mathrm{i}$ in year $\mathrm{t}$, and ${\mathrm{GDP}}_{\mathrm{it}}$ means the GDP of region $\mathrm{i}$ in year $\mathrm{t}$, respectively. Meanwhile, in order to avoid endogenous problems, $1-{\mathrm{GDP}}_{\mathrm{it}}/{\mathrm{GDP}}_{\mathrm{t}}$ is used to deflate the above measurement indicators.

3.2.2. Adjusting Variable

Pig breeding scale: At present, there is no unified measurement for the scale of pig breeding. This paper uses the measurement method of Zhang Yuanyuan for Reference [41], and the formula is set as follows:

$${\mathrm{scale}}_{\mathrm{it}}=\frac{{\mathrm{SFG}}_{\mathrm{it}}}{{\mathrm{ALL}}_{\mathrm{it}}}\times 100$$

(7)

Among them, ${\mathrm{SFG}}_{\mathrm{it}}$ represents the number of pig farm households with more than 50 pigs in the $\mathrm{t}$ year of region $\mathrm{i}$, and ${\mathrm{ALL}}_{\mathrm{it}}$ represents the total number of pig farm households in the $\mathrm{t}$ year of region $\mathrm{i}$.

In terms of control variables: (1) Environmental regulation. The environmental regulation index can measure the degree of environmental regulation because the intensity of government environmental regulation will affect the intensity of pollutant emissions from pig farming. (2) Abundance of feed supply for pigs. With the high food consumption coefficient of pigs, feed supply capacity and convenience will affect pig breeding, and, thus, affect pollutant emission intensity of pig breeding. Therefore, the sum of regionally concentrated feed and compound feed was used as a proxy index for the abundance of feed supply for pigs. (3) Education level of farmers. Studies have pointed out that the education level with high school culture as the dividing line will significantly affect the scale of pig breeding [42], thus affecting the emission intensity of pollutants from pig farming. Therefore, the proportion of the labor force with high school education of rural residents’ families was used to reflect farmers’ education level. (4) Carrying capacity of the land for pig breeding. The proportion of cultivated land supporting pig farms reflects the capacity of manure treatment [27]. The larger the proportion of cultivated land, the higher the degree of resource utilization of pig manure, and the pollution discharge intensity of pig breeding will be weakened. Therefore, the proportion of cultivated land area in each province to the total cultivated land area of the whole country can express the carrying capacity of pig breeding land in each province. (5) Transportation convenience. The more developed the regional transportation is, the more conducive it is to the transportation of environmental protection equipment, thus affecting the emission of pollutants from pig breeding. Therefore, the research method of Li Xuesong is used for reference, and the ratio of the total mileage of regional roads, railways, and inland waterways to land area is used to measure transportation convenience [43]. (6) Scientific and technological progress. The higher the level of science and technology and the more developed the environmental protection technology, the stronger the ability of pig breeding enterprises to reduce pollution emissions from pig breeding. Therefore, the ratio of patent grant volume to the GDP of each province is used to measure the level of scientific and technological progress. (7) Disease risk. The disease is one of the important causes of pig death, which affects the number of live pigs, and, thus, affects the emission of pollutants in pig breeding. The sum of the number of pig deaths and the volume of culling caused by the common 8-min pig blight included in the Veterinary Bulletin was used as a proxy index to measure the risk of pig blight [8]. (8) Per capita income. The higher the per capita income level, the higher the human cost of reducing pollution emissions of pig breeding enterprises will be, thus affecting the degree of pollution emissions of pig breeding. Therefore, provincial per capita GDP can mean the level of provincial per capita income. (9) Urbanization rate. The higher the urbanization rate, the more perfect the infrastructure construction of pig breeding pollutant treatment. Therefore, the proportion of the urban population in each province is used to measure the urbanization rate of each province.

3.3. Data and Time Span

Considering that the latest data used to measure the number of environmental administrative personnel, environmental supervisors, and environmental monitoring personnel in each province of environmental decentralization can only be obtained up to 2017, considering the availability and scientific nature of the data, panel data of 30 provincial administrative units in China (excluding Tibet, Hong Kong, Macao, and Taiwan) from 2001 to 2017 were selected as the research samples. The data used in this article to measure the intensity of pollutant emissions from pig farming and the data related to environmental protection system personnel are from the China Environmental Yearbook; the measurement data for environmental regulations, land carrying capacity for pig farming, and transportation convenience are from the China Statistical Yearbook; the measurement data of rural residents’ income level come from the China Rural Statistical Yearbook; the measurement data of technological progress level come from the China Agricultural Machinery Industry Yearbook; the measurement data of the abundance of pig feed supply come from the China Feed Industry Yearbook; the measurement data of epidemic risk come from the Veterinary Bulletin; the measurement data of the scale of pig farming come from the China Animal Husbandry and Veterinary Yearbook; the measurement data for the cultural level of farmers come from the China Rural Household Survey Yearbook; and the measurement data of urbanization rate and per capita income level are from the China Economic and Social Big Data Research Platform and ERA-Interim database. Some missing data are supplemented using linear interpolation and cubic spline function methods.

3.4. Description of the Study Variables

The meanings and descriptive statistics of all the variables are displayed in

Table 1. Meaning of variables and descriptive statistics.

Variable Name	Code	Specific Meaning	Mean Value	Standard Deviation
Emission intensity of pollutants from pig breeding	POLL	Total chemical oxygen demand, ammonia nitrogen and total phosphorus on pig breeding (10,000 tons)	1.4922	1.2278
Environmental decentralization	ED	Environmental decentralization index	0.9778	0.3595
Environmental administrative decentralization	EAD	Environmental administrative decentralization index	1.3363	1.2773
Environmental monitoring decentralization	EMD	Environmental monitoring decentralization index	1.3773	1.4512
Environmental supervision decentralization	ESD	Environmental supervision decentralization index	1.4009	1.4098
Environmental regulation	ER	Environmental regulation index	8.4964	0.9689
Pig breeding scale	Scale	Scale index	3.3333	3.2022
Feed supply for pigs	lnSLF	Aggregate of concentrated feed and compound feed (tons)	14.8792	1.1274
Education level of farmers	YWD	Proportion of the labor force with high school education in rural families (%)	10.6026	3.5248
Carrying capacity of land	ZYC	Proportion of cultivated land area of provinces in total area of cultivated land of China (%)	3.0155	2.1805
Transportation convenience	JBD	Ratio of total mileage of roads, railways and inland waterways to land area (%)	0.8132	0.5789
Scientific and technological progress	JSP	Ratio of patent grants to GDP (%)	1.0481	0.9380
Disease risk	YBF	Total number of pig deaths and culls due to eight common diseases (Head)	1236.9900	3546.4590
Per capita income level	lnJGDP	Per capita GDP (Yuan)	3.8066	0.5368
Urbanization rate	CZL	The percentage of urban population (%)	48.6878	15.3648

Source: Author’s own conception, using STATA 16 software.

. This descriptive statistical result was calculated using STATA 16 software. The standard deviations of variables in the sample are quite small, indicating that the sample data are stable. The average emission intensity of pollutants from pig farming reached 14,922 tons, indicating serious pollution from pig farming in China. The indexes of ED, EAD, EMD, and ESD 0.9778, 1.3363, 1.3773, and 1.4009, respectively, whose effects on the pollutant emissions from pig farming remain to be studied.

4. Results and Discussions

4.1. Results of Baseline Regression

This study examined the effect of ED on pollution emissions from pig farming. The regression results of ED, EAD, EMD, and ESD on pollution emissions are shown in

Table 2. Results of baseline regression.

Variable	(1)	(2)	(3)	(4)
ED	−0.2250 *** (0.0554)
EAD		−0.0036 (0.0075)
EMD			−0.0118 * (0.0062)
ESD				−0.0072 (0.0071)
Control	Yes	Yes	Yes	Yes
Time	Yes	Yes	Yes	Yes
Region	Yes	Yes	Yes	Yes
N	510	510	510	510
${\mathrm{R}}^2$	0.2460	0.2200	0.2260	0.2210

Note: Standard errors in parentheses. * p < 0.1 and *** p < 0.01. Source: Author’s own conception, using STATA 16 software.

. The coefficient between ED and pollutant emissions is negative and significant at a 1% level. The coefficient of ED indicates that for every 0.1 increase in ED, POLL is expected to decrease by 0.02250. The R² corresponding to ED is 0.246, indicating a good fit for the model. Further, only the EMD coefficient is negative and significant at a level of 10%, however, EAD and ESD have no significant effect. This shows that ED can clarify the responsibilities and rights of local governments, improve the allocation efficiency of local environmental personnel and funds, adapt to local conditions, and formulate environmental policies suitable for their own conditions according to local economic and geographical characteristics. Also, ED enables accurate supervision and management of regional pollution sources of pig farming and, ultimately, reduces the emission of pollutants from pig farming. From the perspective of different decentralization types, EAD and ESD have little impact on the pollutant emission intensity of pig breeding. EMD enables local governments to directly take measures against pollution behaviors, timely formulate environmental monitoring policies that meet their own needs, and effectively reduce pollutant emission intensity of pig breeding.

4.2. The Moderating Effect of the Pig Breeding Scale

In order to further explore whether the pig farming scale will affect the effect of ED on the pollutant emissions of pig farming, the cross terms of ED and pig farming scale are added, and the regression results are shown in

Table 3. Regulating effect of pig breeding scale.

Variable	(5)	(6)	(7)	(8)
ED*Scale	−0.0581 *** (0.0141)
EAD*Scale		−0.0142 *** (0.0033)
EMD*Scale			−0.0056 (0.0039)
ESD*Scale				−0.0034 (0.0039)
Control	Yes	Yes	Yes	Yes
Time	Yes	Yes	Yes	Yes
Region	Yes	Yes	Yes	Yes
N	510	510	510	510
${\mathrm{R}}^2$	0.3250	0.2910	0.2710	0.2650

Note: Standard errors in parentheses. *** p < 0.01. Source: Author’s own conception, using STATA 16 software.

. The results indicate that the coefficient between the cross term of the pig farming scale and ED and the pollutant emissions is −0.0581 and significant at a 1% level. The coefficient of the cross term between pig farming scale and environmental decentralization indicates that when ED is constant, for every 1 increase in scale, it is expected that POLL will decrease by 0.0581. The corresponding R² is 0.3250, indicating a good fit for the model. Therefore, the pig farming scale strengthens the negative impact of ED on pollutant emissions. The reason may be that large-scale breeding entities have advantages in capital and scale and are more able to deal with pig breeding pollutants in accordance with local environmental policies. The technological spillover effect generated by their technological innovation will reduce the emission reduction cost of surrounding pig farming pollutants and, ultimately, reduce the emissions of pig farming pollutants. The cross terms of three kinds of decentralization and pig breeding scale had a negative effect on the pollutant emission intensity of pig breeding, but only EAD was significant, and the rest were not significant. The reasons are that EAD helps local governments adapt to local conditions, formulate suitable environmental policies, and optimize the efficiency of environmental supervision. Hence, promoting the innovative breeding behaviors of large-scale pig breeding subjects while reducing the pollutant exclusion intensity of pig breeding.

4.3. Heterogeneity Analysis

In November 2017, the Ministry of Agriculture issued the National Plan for Pig Production and Development (2016–2020), which divided the country’s pig breeding into key development zones, restricted development zones, potential growth zones, and moderate development zones, according to regional environmental carrying capacity and resource endowment for large-scale pig breeding. This was carried out in order to explore the impact of ED on pollutant emissions in regions with different pig farming policies. Grouping regression was conducted, and the empirical regression results are shown in

Table 4. Heterogeneity analysis of pig breeding area planning.

Variable	Key Zones	Restricted Zones	Potential Zones	Moderate Zones	Key Zones	Restricted Zones	Potential Zones	Moderate Zones
ED	−0.1280	−0.2330 ***	−0.3450 **	−0.1960 ***
	(0.1440)	(0.0881)	(0.1500)	(0.0656)
EAD					0.0159	−0.0231	0.1150 **	0.0002
					(0.0175)	(0.0189)	(0.0530)	(0.0033)
Control	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Region	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	119	187	102	102	119	187	102	102
R2	0.4600	0.3240	0.6470	0.5590	0.4600	0.3010	0.6450	0.5130
Variable	Key Zones	Restricted Zones	Potential Zones	Moderate Zones	Key Zones	Restricted Zones	Potential Zones	Moderate Zones
EMD	0.0046	−0.0158	−0.0752 ***	−0.0026
	(0.0181)	(0.0115)	(0.0244)	(0.0027)
ESD					0.0096	−0.0133	−0.0264	−0.0034
					(0.0164)	(0.0137)	(0.0229)	(0.0033)
Control	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Time	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Region	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	119	187	102	102	119	187	102	102
R2	0.4560	0.3030	0.6630	0.5180	0.4580	0.2990	0.6310	0.5190

Note: Standard errors in parentheses. ** p < 0.05 and *** p < 0.01. Source: Author’s own conception, using STATA 16 software.

. In key pig development areas, ED had little effect on the pollutant emission intensity of pig farming, but in restricted development areas, potential growth areas, and moderate development areas, ED causes an obvious reduction in pollutant emissions. In potential growth areas, decentralization of environmental monitoring can also decrease the pollutant emissions from pig farming.

4.4. Endogenous Analysis

According to theoretical and empirical tests, ED can effectively inhibit carbon emissions from pig farming. Meanwhile, higher carbon emissions from regional pig farming will, in turn, improve the regional ED system, so explanatory variables and explained variables may be mutually causal. In order to solve the endogenous problems brought by ED, this paper refers to the treatment method of Liu [44]. The mean value of ED levels of neighboring provinces in the same year is used as the instrumental variable of ED in this region. The reason is that the carbon emission of pig breeding in this region is not affected by the ED of other provinces, which meets the exogeneity requirements. Moreover, due to the mutual imitation behavior characteristics between the governments of neighboring provinces, ED in this region is also affected by neighboring regions, and the decentralization settings, and pollution treatment methods are very similar. As shown in

Table 5. Results of endogeneity test.

Variable	First Stage	Second Stage
ED		−0.3892 *** (0.0167)
IVED	0.2573 *** (0.0229)
Control	Yes	Yes
Time	Yes	Yes
N	510	510
Wald		95.3148
F	72.1903 ***
R2	0.3842	0.2819

Note: Standard errors in parentheses. *** p < 0.01. Source: Author’s own conception, using STATA 16 software.

, from the regression results of the first stage, the F value is greater than 10 and passed the 1% significance level test, indicating that there is no weak instrumental variable problem. According to the results of the second stage regression, the negative impact of ED on the pollutant emission intensity of pig farming is more obvious.

4.5. Robustness Test

To enhance the reliability of our findings, we conducted a series of robustness tests. Firstly, we replaced the explained variables. We used the chemical oxygen demand of pig breeding to replace the explained variable to reduce indicator bias. Secondly, we deleted some samples. Considering the policy advantages of municipalities directly under the central government, we deleted these samples. Then, we added the time trend of the control variable. In order to avoid the endogeneity problem caused by the change in the macro system, the time trend of the control variable was added to alleviate this potential endogeneity problem. In addition, the method of the high-dimensional fixation effect was used to alleviate potential endogeneity problems. As shown in

Table 6. Robustness test results.

Variable	Replace the Explained Variable	Exclude the Municipalities Directly under the Central Government	Add Control Variable Time Trend	High-Dimensional Fixation Effect
ED	−0.4892 *** (0.0237)	−0.3356 *** (0.0372)	−0.5381 *** (0.0829)	−0.3305 *** (0.0168)
EAD	−0.0829 (0.1639)	−0.0831 (0.1274)	−0.5893 (0.9872)	−0.6285 (0.9173)
EMD	−0.4261 ** (0.2048)	−0.5379 *** (0.0378)	−0.6382 * (0.3514)	−0.5918 ** (0.2704)
ESD	−0.3722 (0.9816)	−0.0368 (0.9935)	−0.8816 (0.7934)	−0.4893 (0.7429)
Control	Yes	Yes	Yes	Yes
Time	Yes	Yes	Yes	Yes
Region	Yes	No	Yes	Yes
N	510	442	510	510
R2	0.2109	0.2235	0.2091	0.2638

Note: Standard errors in parentheses. * p < 0.1, ** p < 0.05 and *** p < 0.01. Source: Author’s own conception, using STATA 16 software. The National Pig Production Development Plan (2016–2020) clearly states that the key development zones include Hebei, Shandong, Henan, Chongqing, Guangxi, Sichuan, and Hainan; restricted development zones include Beijing, Tianjin, Shanghai, Jiangsu, Zhejiang, Fujian, Anhui, Jiangxi, Hubei, Hunan, Guangdong; potential growth zones include Liaoning, Jilin, Heilongjiang, Inner Mongolia, Yunnan, and Guizhou; and moderate development zones include Shanxi, Shaanxi, Gansu, Xinjiang, Tibet, Qinghai, and Ningxia.

, despite the difference in coefficient size and significance level, the research conclusion is not affected, which, again, demonstrates the reliability of the results of this paper.

5. Conclusions and Implications

In this paper, ED, pig farming scale, and pig farming pollutant emission intensity are included in the unified analysis framework. This study analyzed the influence of ED on pollution emissions from pig farming from both theoretical and practical perspectives. The results showed that ED could significantly reduce the pollutant emissions from pig farming, and EMD has a certain effect on reducing pollutants from pig farming, while EAD and ESD have little effect. In addition, the scale of pig farming can effectively strengthen the effect of ED on pollution reduction in pig farming. From the perspective of regional heterogeneity, ED can obviously reduce the pollutant emissions from pig farming in restricted development areas, potential growth areas, and moderate development areas. Moreso, EMD can also achieve pollutant reduction from pig farming in potential growth areas, however, EAD and ESD do not play such an effect.

Based on the above research conclusions, this paper proposes the following policy suggestions: first, decentralize the authority of environmental affairs management in appropriate regions. The environmental management mode of decentralization should be adopted in the restricted development area, potential growth area, and moderate development area of pig breeding. So, to clarify the rights and responsibilities of each local government in environmental governance, constantly improve the environmental assessment system, and maximize the advantages of local government in environmental governance; second, we should reasonably handle the decentralization of different types of environments and consider regional heterogeneity. The power of environmental supervision shall be appropriately delegated to lower levels in areas with potential growth of pig breeding, and local governments shall be empowered to reform and improve the environmental supervision system. At the same time, EAD and EMD should be prudently managed in each region to prevent local governments from excessively pursuing short-term economic development while neglecting environmental protection, to make up for the shortage of funds and technologies of local governments in environmental supervision; third, we will encourage the large-scale development of pig breeding, such as promoting the construction of infrastructure related to large-scale pig breeding, providing more funds for updating environmental protection equipment of pig breeding enterprises, and supporting the large-scale development of pig breeding.

The limitation of this study is that the research data in this paper are from 2001 to 2017, and we did not make a comparative analysis of other periods. In addition, it may not be perfect to measure ED only by the personnel structure of environmental protection departments, so future research could optimize these issues. Undeniably, our study can be a good reference for studying ED on pollution in other industries.