Dual Environmental Regulation and Carbon Emission Reduction in Pig Breeding Industry: Synergistic Effect or Crowding-Out Effect? Evidence from China

Abstract

The prohibition zone policy restricts certain activities in specific areas to protect the environment, while the emission permit policy allows companies to trade emissions permits as a market-based approach to regulation. Can the two policies jointly promote the carbon emission reduction of the pig breeding industry (PBI)? Based on the panel data of 31 provinces and cities in China from 2010 to 2020, this paper adopts propensity score matching and multi-period difference-in-differences (PSM-DID) models to study the impact of the dual policy of the prohibition zone and emission permit on the carbon emissions of the PBI. Our theoretical and empirical findings suggest that the dual policy significantly reduces carbon emissions in the PBI. A mediating analysis reveals that industrial structure and government penalties regulate the impact of the dual policy. Further analysis shows that the prohibition zone policy and the emission permit policy play a synergistic role in the impact on the carbon emissions of the PBI. Regional heterogeneity is also explored, indicating that the carbon emission reduction effect is more significant in western China. Policy implementers should fully consider various policies, regional differences, and regulatory factors, formulating complementary policy combinations to jointly promote the green new quality productivity of the PBI.

1. Introduction

Given China’s significant role in global livestock production [1], environmental policies targeting carbon emissions are critical. In recent years, China has successively issued a large number of environmental regulations for animal husbandry, including command-and-control and market-oriented policies. For example, in January 2014, “the Regulations on the Prevention and Control of Pollution from Large-scale Livestock and Poultry Breeding” was implemented, and the prohibition zone policy was officially put forward, which is a command-and-control environmental regulation. The other two official documents issued in 2015, which are called “The Action Plan for the Prevention and Control of Water Pollution” and “the Notice on Cooperation in the Delineation of Prohibited Areas for Livestock and Poultry Breeding”, helped bring about the specific delimitation of prohibited breeding areas in various regions [2]. The emission permit policy is a market-oriented environmental regulation, which could be traced back to 2007. In 2011, the National Development and Reform Commission issued “the Notice of the National Development and Reform Office on the Pilot Work of Carbon Emission Trading”, which listed carbon emission rights as the subject matter of emission permits, and then the carbon emission trading pilot projects across the country were successively designated [3]. The pilot areas stipulated that the annual carbon emissions of all industries, besides key industries, should also be limited. If exceeding a certain equivalent, they ought to be offset by purchasing carbon emission rights or developing Chinese certified emission reduction (CCER) projects [4,5,6,7,8,9].

A large number of scholars have carried out extensive research on the effects of environmental regulations. The environmental regulations of developed countries have achieved good results in reducing environmental pollution [10,11,12,13]. However, for China, a developing country, the impact of environmental regulations on environmental pollution is controversial. Some scholars believe that environmental regulations can effectively reduce environmental pollution [14,15,16,17], while others argue that regulations may not mitigate pollution [18]. In addition, increasingly more scholars believe that the relationship between environmental regulations and environmental pollution is not a simple linear relationship [19,20,21,22]. Under the influence of technological innovation, industrial restructuring, and foreign investment, the emission reduction effect of environmental regulations may change structurally, showing a significant threshold feature. In other words, when the intensity of environmental regulations exceeds the critical value, environmental regulations can effectively curb pollution emissions.

In response to the social and environmental losses from livestock production, the Chinese government has implemented a series of livestock environmental regulations to control the pollution emissions resulting from livestock production [1,2,3,22,23,24,25,26,27,28,29,30,31,32]. Some scholars find that with the continuous promotion of environmental regulations, the awareness and willingness of livestock and poultry farmers to participate in waste utilization could be significantly improved [26]; the half-point source pollution of livestock and poultry breeding could be controlled [24,27]; and chemical oxygen demand (COD) [1,28], NH₃-N [1,22], and antibiotic resistance genes [29] could be reduced significantly. Others argue that the relationship between environmental regulation and agricultural non-point source pollution may be U-shaped [30]. As for the prohibition zone policy, few existing studies have shown that the policy could significantly reduce the emissions of pollutants from the pig breeding industry (PBI) and promote the high-quality development of the PBI through technological innovation [2,3,22]. The implementation of the policy of prohibited breeding areas has led to a rapid decline in the total amount of livestock and poultry breeding, the gradual withdrawal of retail investors, and the concentration of livestock and poultry production capacity to large farms. The direct reduction in the amount of aquaculture brought by the implementation of the policy could help alleviate the pressure of cultivated land pollution and water pollution caused by livestock and poultry breeding, but the risk of environmental pollution and pollution transfer is still high [31]. The latest research on the two restricted breeding zone policies indicates that these policies have a significantly negative impact on the ecological efficiency of pig farming [32]. Furthermore, the relaxation of these policies has a significantly negative effect on the green total factor productivity of pig farming [22]. Existing research on China’s emission permit policy demonstrates that the policy has significantly reduced regional carbon emissions [22,33,34,35,36,37,38,39,40,41]. Under the pilot policy of emission permits, animal husbandry, including the PBI, carries out carbon emission reductions and carbon offsets mainly through trading carbon emission rights or developing and trading the CCER projects [4,5,6,7,8,9,42,43,44].

According to previous relevant studies, the prohibition zone policy could reduce the carbon emissions of the PBI, and the emission permit policy could significantly achieve a regional carbon emission reduction effect, but can the superposition of the two policies also significantly reduce carbon emissions of the PBI? Is there a synergistic effect or crowding-out effect when the two policies are implemented together? What is the impact mechanism of the dual policy on the PBI? There is no relevant research on these issues, and if the dual policy is not as effective as a single policy, it will result in significant cost waste. This paper collects panel data of 31 provinces or cities in China, and employs the PSM-DID method to study the impact of the dual policy of the prohibition zone and emission permit on the carbon emissions of the PBI. The marginal contribution of this paper mainly includes the following three points. First, previous studies have only studied the impact of a single policy on the carbon emissions of the PBI in prohibited breeding areas [2,3,31]. This paper studies the common impact of the superimposed emission permit policy in the prohibition zone policy on the carbon emissions of the PBI and its impact mechanism. Secondly, regarding benchmark impact, impact mechanisms, and effect analysis, this paper first draws preliminary conclusions using the model deduction approach enhanced by [3]. It then proposes research hypotheses and ultimately confirms all these assumptions through empirical regressions. Thirdly, through selecting samples, this paper discusses in-depth whether the dual policy is more effective than a single one, finding that there is a synergistic effect between the two single policies.

The rest of this paper is arranged as follows. Section 2 carries out theoretical analysis and puts forward research hypotheses. Section 3 introduces the source of data and provides the definition of variables and the model used. Section 4 shows the benchmark regression and, in order to verify it, the robustness tests are arranged subsequently, and then the moderating effect is analyzed. In order to explore the synergy or squeezing effect of the mechanism and whether it has heterogeneity, this paper makes a further analysis in Section 5. The final part is the conclusions and policy recommendations.

2. Theoretical Analysis and Research Hypothesis

This paper expands the environmental regulation model of [3], mainly including the following: (1) Assume that there is a functional relationship between the area of pig farms and the carbon emissions of pigs, because there is a positive correlation between the area of pig farms and the scale of breeding. (2) Under the dual policy, the total cost of pig breeding is not the accumulation of all costs under the two policies, but part of the accumulation. Based on this, this paper theoretically analyzes the impact of the dual policy on the carbon emissions of the PBI.

2.1. Various Costs of Pig Breeding Under the Dual Policy

Under the dual policy of the prohibition zone and emission permit, a pig farm mainly has the following costs.

(1) Cost of Relocation. This refers to the cost for a pig farm to move out of the prohibited breeding area and choose another location to establish the farm. Assuming that the unit cost of plant construction is $b$ and the unit government subsidy is $s$, generally, $b>s$. Under the control of other factors, assuming that the area of the pig farm is a function $\delta (e)$ of the carbon emissions $e$ generated from the farm, the larger the breeding area, the greater its carbon emissions, and with the expansion of the area, the greater the added value of the breeding amount would be; therefore, it could be assumed that $\delta \prime (e)>0$ and $\delta ″(e)>0$. Thus, the relocation cost $(b-s)\times \delta (e)$ could be obtained.

(2) Cost of emission reduction. This refers to the cost paid by a pig farm to reduce emissions. Assume that $g$ represents the reciprocal of the low carbon technology level of the pig farm. The larger it is, the lower the level of low-carbon technology is, and the greater the reduction cost of the farm. At the same time, assume that the reduction cost of the farm after controlling the technology level and other factors is a function $\rho (e)$ of the emissions $e$. For the farm, a larger $e$ means that the reduction efforts are smaller and the reduction costs are smaller; thus, this paper assumes that $\rho \prime (e)<0$ and $\rho ″(e)>0$. Therefore, the emission reduction cost of the farm is $g\times \rho (e)$.

(3) Cost of emission permit. This refers to the cost borne by a pig farm in excess of the pollutant discharge permit, which is equal to $p\times (r-d)$. $p$ represents the price of the pollutant discharge permit, $r$ denotes the pollutant discharge amount reported by the farm, $d$ means the quantity of the pollutant discharge permit obtained by the pig farm, and $r-d$ stands for the carbon emission equivalent that the farm needs to purchase the pollutant discharge permit power.

(4) Cost of Penalty. This refers to the cost of fines paid by the environmental protection department due to the false reporting of carbon emissions by a pig farm. The penalty cost is equal to $\pi \times \upsilon (e-r)$, where $\pi$ is the probability that the administrative department detects the misreporting of carbon emissions from the farm, and $\upsilon (e-r)$ is the penalty function. Because there is a heavier penalty for greater misreporting, it is assumed to be a strictly concave function, that is, $\upsilon \prime (e-r)>0$ and $\upsilon ″(e-r)>0$.

2.2. Optimal Carbon Emissions of Pig Breeding Under Dual Policy

Assuming that the total cost of a pig farm is $C$, then

$$C=(b-s)\times \delta (e)+g\times \rho (e)+p\times (r-d)+\pi \times \upsilon (e-r),e-r>0.$$

(1)

The Lagrange function is constructed as follows:

$$L=(b-s)\times \delta (e)+g\times \rho (e)+p\times (r-d)+\pi \times \upsilon (e-r)-\lambda \times (e-r).$$

(2)

In order to obtain the optimal actual carbon emissions and the optimal reported carbon emissions, we have the following expressions:

$${\displaystyle \frac{\partial L}{\partial e}}=(b-s)\times \delta \prime (e)+g\times \rho \prime (e)+\pi \times \upsilon \prime (e-r)-\lambda =0,$$

(3)

$${\displaystyle \frac{\partial L}{\partial r}}=p-\pi \times \upsilon \prime (e-r)+\lambda =0.$$

(4)

Adding (3) to (4), the following is obtained:

$$(b-s)\times \delta \prime (e)+g\times \rho \prime (e)+p=0.$$

(5)

It can be seen from (5) that the optimal actual carbon emissions $e*$ of the pig farm is

$$e*=\left\{e\right|(b-s)\times \delta \prime (e)+g\times \rho \prime (e)+p=0\}.$$

(6)

2.3. Hypothesis Formulation

Let the optimal actual carbon emissions $e*$ of a pig farm take the derivative of the unit cost of plant construction $b$ and the emission permit price $p$, obtaining

$${\displaystyle \frac{\partial e*}{\partial b}}={\displaystyle \frac{-1}{(b-s)\times \delta ″(e)+g\times \rho ″(e)}}\times \delta \prime (e)<0,$$

(7)

$${\displaystyle \frac{\partial e*}{\partial p}}={\displaystyle \frac{-1}{(b-s)\times \delta ″(e)+g\times \rho ″(e)}}<0.$$

(8)

It can be seen from (7) and (8) that when the prohibition zone policy leads to the relocation of a pig farm, the higher the unit construction cost ($b$), the lower the actual carbon emissions ($e*$). Similarly, the higher the emission permit price ($p$), the lower the actual carbon emissions of the pig farm. Based on this, the following research assumption could be proposed:

Hypothesis 1 (H1).

China’s dual policy of the prohibition zone and emission permit can significantly curb the carbon emissions from the PBI.

In addition, the stricter the government’s punishment on pig enterprises that do not comply with the policy, the smaller the $g$—because, in order to avoid purchasing excess carbon emission rights, they need to focus on developing low-carbon technologies—and thus the greater the impact of $p$ on $e*$, and the greater the impact of $b$ on $e*$. In other words, the stricter the penalties for defaults, the more proactive local pig farmers will be in developing low-carbon technologies to ensure compliance. This will negatively moderate the suppressive effect of the dual policies on carbon emissions from pig farming, further reducing carbon emissions in the industry. On the other hand, the high industrial structure of pigs means that there may be a large proportion of large-sized pig farms in the province. The area with a large scale of pig breeding may be the main producing area of pigs, and there is a large demand for pigs. Therefore, the farm may tend to adopt low-cost breeding technologies, resulting in more carbon emissions. Then, the higher the pig industrial structure is, the lower the low-carbon technology level of the farms in the region may be, the larger $g$ is, and the greater $\rho ″(e)$ is, so the smaller the impact of $p$ on $e*$ is, and the smaller the impact of $b$ on $e*$ is. In other words, the higher the structural development of the pig industry, the more likely it is that the area will be a major production zone for pigs. Compared to carbon reduction, the priority of ensuring stable production and supply becomes more significant in that region. As a result, the government’s carbon reduction requirements for local pig farmers may be somewhat compromised, ultimately leading to increased carbon emissions due to the higher industrial structure of pig farming. This will positively moderate the impact of the dual policies on carbon emissions from pig farming, thereby weakening the suppressive effect of these policies on carbon emissions. Based on the analysis, we put forward the second hypothesis to be tested.

Hypothesis 2 (H2).

The regional pig industrial structure and the extent of government punishment for non-performing enterprises play a regulatory role in the impact of the dual policy on the carbon emissions of the PBI.

If a province is only a single pilot of a prohibition zone policy, the total cost of a pig farm in the province is

$$CPro=(b-s)\times \delta (e)+g\times \rho (e).$$

(9)

It can be calculated from the above formula that the optimal actual carbon emissions ($ePro*$) of the farm is

$$ePro*=\left\{e\right|(b-s)\times \delta \prime (e)+g\times \rho \prime (e)=0\}.$$

(10)

If a province is only a pilot of an emission permit policy, the total cost of the farm is

$$CPer=g\times \rho (e)+p\times (r-d)+\pi \times \upsilon (e-r),e-r>0.$$

(11)

In this case, the optimal actual carbon emissions ($ePer*$) of the farm is

$$ePer*=\left\{e\right|p+g\times \rho \prime (e)=0\}.$$

(12)

Comparing (6), (10), and (12), we have

$$e*<ePro*\mathrm{and}e*<ePer*.$$

(13)

According to (13), we find that the dual-policy approach achieves greater carbon emission reductions than either individual policy alone. This demonstrates that the carbon emission policy and livestock prohibition zones exhibit a synergistic effect rather than a crowding-out effect. Based on this, the third research hypothesis is proposed.

Hypothesis 3 (H3).

The carbon emission reduction effect of the dual policy on the PBI is better than that of any single policy. In other words, the prohibition zone policy and the emission permit policy play a synergistic effect in the carbon emission reduction of the PBI.

3. Materials and Methods

3.1. Sample Selection and Data Sources

Referring to [45,46], the data from Hong Kong, Macao, and Taiwan were excluded from the sample due to unavailability. Consequently, this paper selected the carbon emissions of the PBI in 31 provinces and cities in China as the research sample, which ranged from 2010 to 2020. The common DID method uses the average method to select the control group, which may lead to sample selection bias, while the PSM-DID method can reduce sample selection bias by selecting the control group with more similar characteristics to the treatment group, and it is also more conducive to eliminating the impact of other joint intervention policies through differences. This study therefore constructed a treatment group comprising provinces and cities participating in both the livestock prohibition zones and emission permit policy pilots. Using PSM to select comparable controls from other regions, we employed a DID approach to examine the dual policy’s impact on PBI carbon emissions. In the original data, the sample included 341 observations, while after PSM, the sample contained a total of 288 observations. The data used in this paper were derived from the China Rural Statistical Yearbook (https://navi.cnki.net/knavi/yearbooks/YMCTJ/detail?uniplatform=NZKPT accessed on 31 December 2023) and the Collection for the Cost and Profit of China Agricultural Products (https://www.zgtjnj.org/navibooklist-n3024112112-1.html accessed on 31 December 2023).

3.2. Description of Variables

3.2.1. Explained Variable

Referring to [20], this paper used the emission factor method to calculate the carbon emissions of the PBI, as shown in the following formula.

$$\begin{array}{ll}E& =E_{C{H}_4}+E_{N_{2}O}\\ & =e_{C{H}_4}\times Days\times {\kappa }_1+e_{C{H}_4}\times Days\times {\kappa }_2+e_{N_{2}O}\times Days\times {\kappa }_3,\end{array}$$

(14)

where $E$ represents the carbon emissions of the PBI, and $E_{C{H}_4}$ and $E_{N_{2}O}$ denote the CO₂ equivalent converted from $C{H}_4$ and $N_{2}O$, respectively. $e_{C{H}_4}$ and $e_{N_{2}O}$ signify the global warming potential values of $C{H}_4$ and $N_{2}O$, respectively, taking 21 and 310. $Days$ refers to the annual average number of pigs raised in a province. ${\kappa }_1=1,{\kappa }_2=3.5,$ stand for the emission coefficient of $C{H}_4$ produced by the gastrointestinal and fecal fermentation of pigs, respectively, and ${\kappa }_3=0.53,$ denotes the emission coefficient of $N_{2}O$ produced by the fecal fermentation of pigs.

Since the feeding period of live pigs is 200 days, less than one year, it is necessary to adjust the average annual feeding number of pigs. As in [20], the annual average number of pigs was adjusted as follows:

$$Days={\displaystyle \frac{200}{365}}\times Num,$$

(15)

where $Num$ denotes the volume of annual sales of pigs.

3.2.2. Explanatory Variables

This paper took the product of the implementation time of dual environmental regulations including prohibition zone and emission permit policies (Time) and whether the province is in the pilot areas (Treated) as the core explanatory variable (DID = Time × Treated). Both Time and Treated are 0/1 variables.

3.2.3. Control Variables

With reference to [2,3,20,27] and [47,48], this paper selected pig industrial structure (Stru), pig price (LnPrice), education level of farmers (Edu), level of scale (Scale), urbanization level (Urban), and industrial economic development level (Indus) as control variables.

See

Table A1. The definitions, calculation methods, and references of variables.

Type	Variable	Definition	Calculation Method	Reference
Explained variable	LnCE	Carbon emissions	Logarithm of annual carbon emissions, which is calculated by formula (14). Unit: tCO2 e.	IPCC [20]
Explanatory variable	DID	Product terms of time and treated	Implementation time of dual environmental regulations multiplied by whether the province is a pilot, DID = Time × Treated. Unit: 0/1.	China government network
Control Variable	Stru	Pig industrial structure	Total output value of pigs divided by total output value of animal husbandry. Unit: %.	[47]
	LnPrice	Pig price	Logarithm of average selling price of live pigs. Unit: Yuan.	[48]
	Edu	Education level of farmers	(Number of unschooled population × 1 + population with primary education background × 6 + population with junior high school education × 12 + population with college degree or above × 16) divided by number of rural population aged over 6. Unit: Year.	[3,47]
	Scale	Level of scale	Number of pig farmers with more than 50 sold divided by total number of pig farmers. Unit: %.	[2,27]
	Urban	Urbanization level	Provincial urbanization rate. Unit: %.	[2,3,23]
	Indus	Industrial economic development level	Total output value of animal husbandry divided by rural population. Unit: 10⁸ yuan/10⁴ population.	[47]
Moderating Variable	Stru	Pig industrial structure	Total output value of pigs divided by total output value of animal husbandry. Unit: %.	[47]
	Punish	Punishment intensity	Set as 6 in Beijing, Shanghai, and Shenzhen, and as 5, 4, 3, 2, 1, and 0 in Chongqing, Hubei, Guangdong, Fujian, Tianjin, and the rest, respectively. Unit: None.	[49]

for detailed definitions of the aforementioned variables.

3.3. Empirical Models

3.3.1. Basic Regression Models

To check H1 and H3, this paper adopted the following models to explore the influence of dual environmental regulations on the carbon emissions of the PBI at first, and then compared the effects of a single policy and dual policy on the carbon emission reduction of the PBI.

$$LnC{E}_{it}={\alpha }_0+{\alpha }_{1}DI{D}_{it}+{\alpha }_2{X}_{it}+Z_t+{\lambda }_i+{\epsilon }_{it},$$

(16)

$$LnC{E}_{it}={\alpha }_0\prime +{\alpha }_1\prime DIDPr{o}_{it}+{\alpha }_2\prime X_{it}+Z_t+{\lambda }_i+{\epsilon }_{it},$$

(17)

$$LnC{E}_{it}={\alpha }_0″+{\alpha }_1″DIDPe{r}_{it}+{\alpha }_2″X_{it}+Z_t+{\lambda }_i+{\epsilon }_{it},$$

(18)

where $DI{D}_{it}$ denotes the cross term of the implementation time of the dual policy ($Tim{e}_t$) and whether the province is a dual pilot area ($Treate{d}_i$); $DIDPr{o}_{it}$ signifies the cross term of the implementation time of the prohibition zone policy ($TimePr{o}_t$) and whether the province is a single pilot area for the policy ($TreatedPr{o}_i$); $DIDPe{r}_{it}$ stands for the cross term of the implementation time of the emission permit policy ($TimePe{r}_t$) and whether the province is a pilot area for the policy ($TreatedPe{r}_i$); and $X_{it}$ represents a set of control variables including Stru, LnPrice, Edu, Scale, and Urban. $Z_t$ and ${\lambda }_i$ denote time and individual fixed effects, respectively. ${\epsilon }_{it}$ is the residual term.

3.3.2. Moderating Effect Models

In order to test H2, checking whether the variables Stru and Punish have regulatory effects, this paper added some cross terms into model (16) as follows:

$$LnC{E}_{it}={\beta }_0+{\beta }_{1}DI{D}_{it}+{\beta }_{2}DI{D}_{it}\times Str{u}_{it}+{\beta }_3{X}_{it}+Z_t+{\lambda }_i+{\epsilon }_{it},$$

(19)

$$LnC{E}_{it}={\gamma }_0+{\gamma }_{1}DI{D}_{it}+{\gamma }_{2}Punis{h}_{it}+{\gamma }_{3}DI{D}_{it}\times Punis{h}_{it}+{\gamma }_4{X}_{it}+Z_t+{\lambda }_i+{\epsilon }_{it},$$

(20)

where $Punis{h}_{it}$ is an ordinal measure that reflects the varying intensity of non-compliance penalties across China’s regional carbon market pilots. The values were calibrated according to each jurisdiction’s regulatory framework, as described in [49]: (i) Beijing, Shanghai, and Shenzhen (score = 6) have enacted the most stringent policies, which include (a) monetary penalties ranging from three to five times the prevailing carbon market price, (b) mandatory deductions of future emission allowances, and (c) public disclosure of non-compliant entities; (ii) Chongqing (score = 5), Hubei (score = 4), and Guangdong (score = 3) enforce intermediate-level sanctions that involve fines of one to three times the market price, along with mandatory allowance redemption; (iii) Fujian (score = 2) and Tianjin (score = 1) impose minimal deterrents through nominal fixed fines; and (iv) non-pilot regions (score = 0) completely lack formal penalty structures.

4. Results

4.1. Descriptive Statistics

Table 1. Descriptive statistics of variables.

	Variable	Sample Size	Mean Value	Standard Deviation	Minimum	Maximum
	Stru	341	0.402	0.175	0.023	0.778
	LnPrice	341	6.690	0.260	6.275	7.493
	Edu	341	0.037	0.022	0.007	0.163
Before PSM	Scale	341	0.210	0.277	0.003	0.991
	Urban	341	0.574	0.134	0.226	0.896
	Indus	341	8.387	0.496	7.011	9.827
	Stru	288	0.453	0.132	0.087	0.778
	LnPrice	288	6.697	0.266	6.312	7.493
After PSM	Edu	288	0.037	0.023	0.007	0.163
	Scale	288	0.224	0.281	0.004	0.991
	Urban	288	0.589	0.131	0.315	0.896
	Indus	288	8.359	0.496	7.011	9.827

describes the main variables. Stru ranges from 0.023 to 0.778, reflecting the obvious differences in the industrial structure of pigs in different regions. The minimum and maximum of Urban are 0.226 and 0.896, respectively, which indicates that the urbanization level of different provinces in China is quite different. After PSM, the minimum values of Stru and Urban increase significantly to 0.087 and 0.315, respectively, and the gap between the maximum and minimum values has been narrowed to a certain extent.

4.2. PSM and Balance Test

By means of the kernel matching method, which belongs to PSM methods, this paper selects the control group from non-pilot provinces and cities to reduce sample selection bias, and then employs the logit model to estimate the possible probability that the sample is a pilot province. Considering the main influencing factors of carbon emissions of the PBI, this paper selects Stru, Edu, and Indus as matching variables, and selects the one which is closest to its probability value as the control group based on the propensity score. Based on the research of [50], the absolute value of the standard deviation after matching needs to be less than 20% to achieve the matching effect. From

Table 2. Balance test results of PSM.

Variable	Sample	Mean Value		Standard Deviation (%)	Reduction Rate of Standard Deviation (%)	T Value
		Treatment Group	Control Group
Stru	No Matching	0.476	0.322	96.8	99.1	8.99
	Matching	0.476	0.477	−0.9		−0.1
Edu	No Matching	0.039	0.034	24.6	57.8	2.25
	Matching	0.039	0.037	10.4		0.96
Indus	No Matching	8.329	8.449	−24.4	46.6	−2.26
	Matching	8.329	8.393	−13.0		−1.38

, we can see that all the standard deviations of matching variables are less than 20%. Among them, the standard error of Stu decreases by 99.1%, and the standard error of Edu decreases by 57.8%. They suggest that, after tendency value matching, the variable characteristics of the treatment and control groups are relatively similar; therefore, they meet the balance hypothesis of the DID model.

In addition, as shown in Figure 1, in the treatment and control groups, the vast majority of samples are within the common value range, while the tendency scores of samples not within the common range are extreme.

4.3. Parallel Trend Test

Before the implementation of the dual policy, the change trend of the carbon emissions of the PBI in pilot and non-pilot areas should be parallel, which is called the parallel trend hypothesis, and is the key premise of the DID model. With reference to [51], this paper adopts the event study to carry out a parallel trend test, which can be expressed as

$$LnC{E}_{it}=\delta +{\displaystyle \sum _{j=2}^3{\delta }_{current-j}}DI{D}_{i(t-j)}+{\displaystyle \sum _{k=0}^5{\delta }_{current+k}}DI{D}_{i(t+k)}+\delta \prime X_{it}+Z_t+{\lambda }_i+{\epsilon }_{it},$$

(21)

where $DI{D}_{i(t-j)}$ is equal to 1 if the province or city becomes a pilot area after $j$ years, and is 0 otherwise. $DI{D}_{i(t+k)}$ is equal to 1 if the province or city becomes the pilot area before $k$ years, and is 0 otherwise. The coefficient ${\delta }_{current-j}$ reflects the difference in carbon emissions of the PBI in pilot and non-pilot areas in the $j$ years before the implementation of the dual policy. If the regression results show that each ${\delta }_{current-j}$ is not significantly different from 0, then it could be determined that the treatment and control groups conform to the parallel trend assumption, and the regression results of model (16) have no obvious selectivity error.

This paper takes the first year before the implementation of the dual policy as the base period, and the results of the parallel trend test are presented in

Table 3. Parallel trend test.

Variable	LnCE
DID (current − 3)	−0.066 (0.115)
DID (current − 2)	0.014 (0.115)
DID (current)	−0.111 (0.118)
DID (current + 1)	−0.192 (0.116)
DID (current + 2)	−0.330 *** (0.117)
DID (current + 3)	−0.560 *** (0.119)
DID (current + 4)	−1.282 *** (0.138)
DID (current + 5)	−2.176 *** (0.173)

Note: *** indicates that they have passed the significance test at the level of 1%. The figures in brackets are standard errors. DID (current + m) denotes $DI{D}_{i(current+m)}$, m = −3, −2, 0, 1, 2, 3, 4, 5.

and Figure 2. As seen from the table, the coefficients of ${\delta }_{current-3}$ and ${\delta }_{current-2}$ are −0.066 and 0.014, respectively, which are not significant at least at the 10% level. This indicates that there are no significant trend differences in the carbon emissions of the PBI in the pilot and non-pilot regions before the implementation of the dual policy in China. Further, the coefficient of ${\delta }_{current+k}$ becomes increasingly larger, and significant at the level of 1% from ${\delta }_{current+2}$ to ${\delta }_{current+5}$, showing that the effect of the dual policy is gradually strengthened. On the other hand, from Figure 2, we can see that the 95% interval of the corresponding coefficients ${\delta }_{current-j}$ from the third year to the second year before the implementation of the dual policy contain 0, all of which fail the significance test, signifying that both the variation trends of the treatment and control groups meet the parallel trend test. Further, it can be seen from the figure that each coefficient ${\delta }_{current+k}$ becomes increasingly more significant and its absolute value becomes increasingly larger year by year, which demonstrates the persistence and effectiveness of the impact of the dual policy on the carbon emissions of the PBI. In conclusion, the application of the PSM-DID method in this paper conforms to the parallel trend hypothesis.

4.4. Basic Regression Results

Table 4. Basic regression results.

Variable	LnCE
	(1)	(2)
DID	−0.485 ** (0.181)	−0.252 *** (0.073)
Control Variables	NO	YES
Time FE	YES	YES
Individual FE	YES	YES
Observations	288	288
R-squared	0.50	0.67

Note: ***, and **, respectively, indicate that they have passed the significance test at the levels of 1%, and 5%. The figures in brackets are standard errors (consistent in all tables).

presents the impact of the dual policy on the carbon emissions of the PBI based on the basic model (16). Column (1) controls a set of time and individual fixed effects, and column (2) adds control variables mentioned above. The results in Table 4 show that regardless of whether control variables are included or not, the core coefficient is persistently significant and negative at the level of 1%, which suggests that the implementation of the dual policy in pilot areas is conducive to lowering the carbon emissions of the PBI. Quantitatively, the estimated coefficients of DID in columns (1) and (2) indicate that the carbon emissions of the PBI decrease by 0.485 and 0.252, respectively, after the implementation of the dual policy. Therefore, both statistically and economically, these results corroborate H1.

4.5. Robustness Checks

4.5.1. Replace PSM Method

The above analysis confirms that the impact of the dual policy on the carbon emissions of the PBI is significantly negative, where the PSM method adopts nuclear matching with higher matching accuracy. Based on the original data, this part further adopts the nearest neighbor matching and difference-in-differences method for regression.

Table 5. Regression results by the nearest neighbor matching and difference-in-differences method.

Variable	LnCE
	(1)	(2)
DID	−0.473 *** (0.181)	−0.257 *** (0.072)
Control Variables	NO	YES
Time FE	YES	YES
Individual FE	YES	YES
Observations	341	341
R-squared	0.41	0.54

Note: *** indicates that they have passed the significance test at the level of 1%.

reports the new PSM-DID regression results of the matched samples. From the table, we can see that the cross terms of the policy without and with control variables are negative at the significance level of 1%, in line with the previous conclusions, indicating that the empirical method is robust.

4.5.2. Time Placebo Test

To further verify the robustness of the benchmark regression results, by changing the time point of policy implementation, this paragraph carries out the counterfactual test. Referring to [52], this paper narrows the sample range from 2010 to 2016 and sets the policy implementation time as two years before the actual implementation time, and then carries out PSM-DID regression again.

Table 6. Regression results by counterfactual test.

Variable	LnCE
	(1)	(2)
DID	−0.121 (0.077)	−0.002 (0.930)
Control Variables	NO	YES
Time FE	YES	YES
Individual FE	YES	YES
Observations	160	160
R-squared	0.73	0.76

reports the results, which demonstrates that after changing the year of policy impact, the coefficients of policy cross terms are not significant. These results verify that the reduction in carbon emissions of the PBI comes from the pilot policy, rather than from other random factors.

4.5.3. Treatment Group Placebo Test

A potential criticism of this paper’s conclusion is that the carbon emission reduction effect of dual policies in the pig industry may stem from certain unobservable factors. The above regressions only alter the timing of the policy occurrence once, and the results may be coincidental, which is insufficient to confirm the complete robustness of the dual policy effect. To address this, we draw on the research methodology of [53] to construct an additional placebo test for the treatment group. This test aims to identify whether the carbon emission reduction effect of the dual policies in the pig industry is attributable to other factors. Specifically, based on the actual number of experimental group provinces each year, we randomly select provinces equal to that number to generate a ‘pseudo-experimental group’ and then repeat the regression 500 times. Figure 3 illustrates the distribution of estimated coefficients from the 500 regroupings, along with the corresponding p-values, where the horizontal axis represents the size of the core explanatory variable coefficients, and the vertical axis represents the p-value magnitude, with dots indicating the distribution of p-values. The results show that most of the estimated p-values are greater than 0.1, which are not statistically significant, indicating that the estimation results of this paper are not obtained by chance.

4.5.4. Replace Explained Variable

To further test the robustness of the basic regression, the global warming potential value ($C{H}_4:21,$ $N_{2}O:310$) is replaced by the greenhouse benefit index ($C{H}_4:25,$ $N_{2}O:298$) as in [20], and the carbon emissions of the PBI are recalculated. The regression results are shown in

Table 7. Regression results by replacing explained variable.

Variable	LnCE		LnCE(1) (3)	LnCE(2) (4)
	(1)	(2)
DID	−0.477 *** (0.180)	−0.256 *** (0.071)	−0.342 *** (0.088)	−0.401 *** (0.147)
Control Variables	NO	YES	YES	YES
Time FE	YES	YES	YES	YES
Individual FE	YES	YES	YES	YES
Observations	288	288	258	232
R-squared	0.53	0.70	0.70	0.71

Note: LnCE(i) denotes the annual carbon emissions lag by i years. *** indicates that they have passed the significance test at the level of 1%.

, including the situation that the explained variables are not lagged, lagged by one year, and lagged by two years, all of which indicate that the policy effect estimated above is robust.

4.5.5. Exclude the Impact of COVID-19

On 31 December 2019, the COVID-19 outbreak began in Wuhan, China. The pandemic led to a decrease in carbon emissions [54,55]. To assess the robustness of our findings, this paper excludes the data from 2020 and re-runs the regression analysis, with the results presented in

Table 8. Regression results for excluding the impact of COVID-19.

Variable	LnCE
	(1)	(2)
DID	−0.368 *** (0.105)	−0.204 *** (0.057)
Control Variables	NO	YES
Time FE	YES	YES
Individual FE	YES	YES
Observations	260	260
R-squared	0.58	0.70

Note: *** indicates that they have passed the significance test at the level of 1%.

. The findings indicate that, after excluding the impact of the pandemic, the dual policies continue to exert a significant suppressive effect on carbon emissions in the pig industry, remaining statistically significant at the 1% confidence level.

4.5.6. Endogeneity Test

Many studies have found a significant negative correlation between precipitation and air pollution [56,57]—where air pollution is low, environmental regulations are naturally weaker. The rainfall in an area is an exogenous natural phenomenon that is expected to have no substantial impact on carbon emissions in the pig breeding industry. Therefore, we believe that the annual precipitation (rainfall) in the sample area is a reasonable instrumental variable. This paper uses the Heckman two-step method to perform regression on the original problem. First, we set the AveCE variable such that if the carbon emissions of a province or city in the previous year are greater than the average emissions of all sample provinces and cities, it is assigned a value of 1; otherwise, it is assigned a value of 0. Next, in the first-stage Probit model, we include the annual precipitation of that location (IV, measured in meters, https://data.cma.cn/data/detail/dataCode/A.0053.0002.S007.html accessed on 31 December 2020) as an additional variable that influences whether the government establishes pilot programs for dual environmental regulation. Finally, we incorporate the calculated Inverse Mills Ratio (IMR) into the second stage for regression analysis. The results from the second stage in

Table 9. Regression results for endogeneity test.

Variable	AveCE	LnCE
	(1)	(2)
IV	−4.792 * (−1.755)
DID		−0.299 *** (−3.478)
IMR		0.001 (0.055)
Control Variables	YES	YES
Time FE	YES	YES
Individual FE	YES	YES
Observations	234	234
R-squared	0.65	0.68

Note: * and *** respectively indicate that they have passed the significance test at the levels of 10% and 1%.

show that the IMR does not exhibit significance (p > 0.1), and the coefficients and significance levels of the dual policies’ impact on carbon emissions in the pig breeding industry are similar in magnitude to those of the baseline regression. Together, this confirms that the problem of sample selection bias endogeneity in our research is not serious.

4.6. Moderating Effect Analysis

This part empirically explores the moderating effect for the policy. H2 is tested with the aid of models (19) and (20), and the estimation results are presented in

Table 10. Regression results of regulatory effect.

Variable	LnCE
	(1)	(2)
DID × Stru	3.856 *** (1.879)
DID × Punish		−0.043 * (0.024)
Control Variables	YES	YES
Time FE	YES	YES
Individual FE	YES	YES
Observations	288	288
R-squared	0.78	0.67

Note: * and *** respectively indicate that they have passed the significance test at the levels of 10% and 1%.

. The coefficient of the cross term DID × Stru is positive, indicating that the pig’s industrial structure curbs the influence of the dual policy on carbon emissions of the PBI. The higher the industrial structure of pigs and the larger the scale of pig breeding, the more difficult it is to carry out low-carbon transformation in a short time due to the high cost. Moreover, areas with a large scale of pig breeding may be the main producing areas of pigs, and the demand for pigs is large; therefore, the farms may tend to adopt low-cost breeding technologies, resulting in more carbon emissions, thus playing a negative role in the impact of environmental regulations on the carbon emissions of the PBI. By contrast, as seen from column (2), the stricter the implementation of environmental regulations in the pilot areas, the more significant the policy effect would be. These regression results support H2.

5. Further Analysis

5.1. Comparison of the Effects of Single and Dual Policies

In order to further study whether the dual policy is better than the single policy, with reference to [58], this paper first supplements the carbon emission reduction effect tests for the single policy pilot of the prohibition zone and emission permit. The specific operation is as follows. First, the samples of the emission permit policy pilots are removed, while the samples of the prohibition zone policy pilots and the samples that are neither the prohibition policy pilots nor the emission permit policy pilots are retained. After regression with the remaining samples, the coefficient of the cross item DIDPro reflects the net impact of the prohibition zone policy on the carbon emissions of the PBI in the policy pilot provinces. In the same way, after excluding the samples from the prohibited breeding areas, the coefficient of the multi-period double difference variable DIDPer in the new regression reflects the net effect of the emission permit policy on the carbon emissions of the PBI in the policy pilot provinces.

Table 11. Net effects of the single policy including the prohibition zone policy and carbon emission policy.

Variable	LnCE (1)	LnCE(1) (2)	LnCE(2) (3)	LnCE (4)	LnCE(1) (5)	LnCE(2) (6)
DIDPro	−0.146 *** (0.053)	−0.117 *** (0.038)	−0.113 *** (0.038)
DIDPer				−0.099 * (0.048)	−0.078 (0.045)	−0.065 (0.039)
Control Variables	YES	YES	YES	NO	NO	NO
Time FE	YES	YES	YES	YES	YES	YES
Individual FE	YES	YES	YES	YES	YES	YES
Observations	223	200	179	112	98	88
R-squared	0.69	0.76	0.76	0.77	0.79	0.75

Note: LnCE(i) denotes the annual carbon emissions lag by i years. * and *** respectively indicate that they have passed the significance test at the levels of 10% and 1%.

reports the regression results. Column (1) reveals that the coefficient of DIDPro is significantly negative at the level of 1%, signifying that the prohibition zone policy can significantly inhibit the carbon emissions of the PBI. Columns (2) and (3) report the regression results of explained variables with a lag of 1 year and 2 years, respectively, which is in line with column (1), indicating that the influence of the prohibition zone policy on the carbon emissions of the PBI is sustainable, in accordance with [2]. Columns (4)–(6) provide the regression results for the impact of the emission permit policy on carbon emissions of the PBI, indicating that the emission permit policy can also inhibit the carbon emissions of the PBI, but the inhibition degree is weak and gradually decreases.

Then, this paper further tests and proves whether the dual pilot is more effective than each single pilot. The specific operation is as follows: eliminate the samples that are neither the prohibition zone pilots nor the emission permit pilots, and retain the samples that are already the prohibition zone pilots or the emission permit pilots. After regression with the remaining samples, the coefficient of the multi-period dual differential variable DID captures the net effect for the impact of the single pilot provinces becoming the dual pilot provinces on the carbon emissions of the PBI. As shown in

Table 12. Net effect of single pilot provinces becoming dual pilot provinces.

Variable	LnCE
	(1)	(2)
DID	−0.447 *** (0.185)	−0.334 *** (0.093)
Control Variables	NO	YES
Time FE	YES	YES
Individual FE	YES	YES
Observations	219	219
R-squared	0.51	0.66

Note: *** indicates that they have passed the significance test at the level of 1%.

, compared with the single policy, the dual policy has a more significant and effective inhibition on the carbon emissions of the PBI, which verifies hypothesis H3.

5.2. Regional Heterogeneity Test

The above analysis verifies the impact of the dual policy on the carbon emissions of the PBI. However, the effect of policy implementation is often heterogeneous at the regional level, due to the unbalanced regional development in China. According to [38], we divide 31 provinces and cities in China into eastern and western regions, and the PSM-DID regression results by region are shown in

Table 13. Regression results based on different regions.

Variable	Eastern Region LnCE (1)	Western Region LnCE (2)
DID	−0.161 (0.166)	−0.183 *** (0.030)
Control Variables	YES	YES
Time FE	YES	YES
Individual FE	YES	YES
Observations	115	68
R-squared	0.75	0.56

Note: *** indicates that they have passed the significance test at the level of 1%.

. From the table, we notice that in the western region, the impact of the dual policy on the carbon emissions of the PBI is consistent with the full sample, while in the eastern region, the influence of the policy is still negative, but not significant. A possible reason is that the western region is the traditional pig production area in China, with a large amount of pig breeding and a relatively backward economy, so the proportion of low-carbon pig breeding is relatively small. Strict environmental regulations can promote the PBI in the western region to introduce relatively advanced low-carbon technologies in the eastern region, thus effectively reducing carbon emissions. In contrast, in the eastern region, on one hand, the production of pigs is relatively low, so the carbon emissions are lower, and it is not easy to further reduce them. On the other hand, the eastern region has adopted more advanced low-carbon technology and low-carbon feed; therefore, it is not easy for environmental regulations to play a significant role.

From 2022 to 2024, the project team conducted extensive research on pig farming enterprises and large-scale family farms in western regions such as Chongqing and Sichuan. The findings indicate that government departments in these areas are increasingly implementing environmental regulations. The adoption of stringent environmental regulations can indeed compel pig farming enterprises, especially leading companies, to actively develop low-carbon and circular farming practices. Policy coordination can achieve better carbon reduction effects, as pig companies and farmers need to adopt different low-carbon behaviors in response to each specific environmental regulation, and each of these low-carbon behaviors contributes to carbon reduction. Currently, the ecological technology and low-carbon farming ratios in pig breeding in western China are still very low, indicating significant potential for improvement. However, the effectiveness of government regulations is often disrupted by policies aimed at maintaining production stability and supply, particularly in major pig-producing provinces. For instance, during periods of low production capacity, such as when the COVID-19 pandemic significantly impacted production in 2020, there may be a relaxation of environmental regulations and a slowdown in the progress of low-carbon farming in favor of ensuring supply stability. Therefore, achieving the dual goals of carbon reduction and maintaining production stability and supply may be the most important objective of policy coordination.

5.3. Comparative Analysis with the EU ETS

With the implementation of the EU ETS, its impact has aroused widespread concern in previous studies. The initial research mainly focused on the environmental effects of the ETS; for example, some studies focused on the role of the ETS in carbon emission reduction [59,60,61,62], basically reaching a consensus that the EU ETS has significantly contributed to emission reductions. In addition, scholars continued to study the spillover effects of the EU ETS, such as its impact on PM2.5 emission reduction from the perspective of policy tools, and common policy tools include resource tax [63] and carbon tax [64]. It is generally agreed that there is a fairly good synergistic effect. Compared with the EU, China’s carbon emission trading market started relatively late and is less mature. There has been relatively little research on the synergistic emission reduction effects between carbon trading policies and other policies in China. The findings of this study on the synergistic emission reduction effects of China’s carbon permit and livestock prohibition zone policies are largely consistent with relevant EU research.

6. Conclusions and Implications

Based on the panel data of 31 provinces and cities in China from 2010 to 2020, this paper adopts PSM-DID models to study the impact of the dual policy of the prohibition zone and emission permit on the carbon emissions of the PBI. The combined implementation of the prohibition zone and emission permit policies reduced carbon emissions by at least 72.6% compared to single-policy regions. The mechanism analysis shows that the punishment degree of the government on non-performance pig enterprises and the pig industrial structure play a regulatory role in the impact of the dual policy on the carbon emissions of the PBI. Further analysis shows that the prohibition zone policy and the emission permit policy play a synergistic role in the impact on the carbon emissions of the PBI. In addition, the results of regional heterogeneity regression show that the carbon emission reduction effect is more significant in western China.

The conclusions of this paper have great implications for the policy makers of carbon emission reduction. First, the prohibition zone policy and the emission permit policy exhibit a significant synergy in restraining the carbon emissions of the PBI. The two policies should be appropriately combined to jointly promote the carbon emission reduction of the PBI, but in the meantime, we should pay attention to the intensity of implementation, promote the high-quality development of the PBI by promoting low-carbon technological innovation, and finally realize the “Potter effect” in China, as has been verified under the single prohibition zone policy [2,3,39] or emission permit policy [65,66,67,68]. Secondly, the government’s punishment on pig enterprises that fail to comply with the contract plays a positive role in regulating the impact of the dual policy on the carbon emissions of the PBI. Therefore, after a policy is issued, we must find ways to strictly implement it to achieve good policy effects. By contrast, the pig industrial structure plays a reverse role in the impact of the dual policy on the carbon emissions of the PBI. Therefore, the industrial structure of major pig producing areas in China should be adjusted reasonably to reduce the pressure of carbon emission reduction in the areas. Thirdly, due to the differences in economic development between eastern and western China, in order to ensure the green and low-carbon transformation of the PBI in China, regional differentiated environmental regulations should be implemented.