Heterogeneous Environmental Regulations and Green Total Factor Productivity: A Study on China’s Animal Husbandry Sector

Abstract

The rapid expansion of China’s livestock husbandry has boosted the supply of meat, eggs, and dairy products, while concurrently giving rise to environmental pollution issues. Research on the effects of various environmental regulations on the green total factor productivity (GTFP) of the livestock sector and their underlying mechanisms is still lacking, despite the Chinese government’s implementation of corresponding environmental regulatory policies to address this practical challenge. As a key instrument for fostering green economic transformation, examining the relationship between environmental regulation and the green total factor productivity (GTFP) of animal husbandry is crucial for the sector’s sustainable development. In order to estimate the GTFP of China’s livestock sector for the years 2010–2022, this study uses the super-slack-based measure (Super-SBM) methodology. It conducts an empirical analysis to examine the mechanisms through which different environmental regulations influence livestock GTFP, alongside an investigation of regional heterogeneity. The results show that different environmental regulations have different effects on animal husbandry GTFP, with notable regional differences. Specifically, incentive-based environmental regulations enhance livestock GTFP by facilitating technological innovation; however, the level of regional economic development negatively moderates the association between incentive-based environmental regulations and livestock GTFP. The findings confirm that incentive-based environmental regulations are successful in encouraging livestock GTFP through technical innovation. They further emphasize that regions should formulate context-specific environmental regulatory policies to balance environmental protection and industrial development, thereby supporting the green and sustainable growth of China’s livestock industry.

1. Introduction

The problem of livestock manure pollution has gotten worse as China’s total livestock production capacity rises and the size of animal husbandry grows, posing serious difficulties for the country in striking a balance between ecological preservation and Gross Domestic Product (GDP) growth [1]. China has released several policy guidelines to promote sustainable and environmentally friendly advancements in animal husbandry. There is a broad consensus that balancing animal production, resource protection, and environmental sustainability should be a top priority. This regulatory framework extends beyond pollution management, aiming to foster a more environmentally sustainable industry by promoting efficiency improvements, reducing carbon dioxide emissions, and encouraging technological innovation. At present, green growth in the livestock sector has emerged as a critical indicator within comprehensive economic sustainability assessments. Consequently, enhancing productivity through these strategies has transitioned from a discretionary option to an imperative priority. Furthermore, escalating international attention to ecological issues has driven the adoption of increasingly stringent environmental standards within the farming sector. Against this backdrop, environmental regulation itself becomes a powerful instrument for facilitating a broader green economic transition. Given these converging priorities, elucidating how environmental regulations shape the sector’s green growth trajectory constitutes a fundamental research question, the resolution of which is essential for steering the industry toward a genuinely sustainable future.

Green total factor productivity (GTFP) solves the problems of traditional total factor productivity, which only looks at economic output, by adding energy use and environmental damage to its framework. It is used to measure the synergies and trade-offs between economic production and environmental impacts, and it shows the idea of “green growth” that seeks a balance between economic growth and ecological sustainability [2]. In the agricultural sector, many factors work together to drive GTFP growth. Agricultural mechanization helps by improving management, making factor use more efficient, and helping industries diversify, and this effect is stronger in regions with advanced agricultural technology and strict environmental regulations [3]. The use of green agricultural technologies [4], the way data helps promote technological innovation and the movement of labor to non-agricultural jobs [5], and fixing resource misallocation and helping integrated agricultural service development [6,7] have all been shown to effectively raise agricultural GTFP. Also, common factors like technological innovation [8,9], improved energy efficiency [10], industrial structure optimization [11], human capital [12] and fiscal support [13] also have a big impact on agricultural GTFP growth. By working together to make resource allocation better, make production more efficient, and reduce environmental harm, these factors all help increase agricultural green total factor productivity.

Environmental regulation means government actions to stop activities that pollute the public environment. It is a major form of social regulation and a key policy tool for sustainable economic growth and environmental protection [10]. Existing research shows there are complex nonlinear relationships between environmental regulation and things like innovation activities [14], eco-efficiency [15], green total factor productivity (GTFP) [16], and the green transition in agriculture [17]. These relationships can be U-shaped or inverted U-shaped, and they vary a lot across different industries, regions, and types of regulation [18,19,20,21]. This nonlinear relationship comes from two opposing economic effects of environmental regulation: the “cost-compliance effect” and the “innovation-compensation effect” [22]. The cost-compliance effect says that environmental regulation can lower GTFP growth in the short term by raising costs and distorting resource allocation. Producers pay more for things like pollution control facilities [23], and that can hurt market confidence and production incentives. It can also threaten product supply and even make some livestock farmers leave the industry [24]. At the same time, as an outside policy tool, environmental regulation can take money away from technological innovation and business expansion. This makes it harder for industries like the livestock sector to go through a green transition. The innovation–compensation effect says that environmental regulation can push enterprises to use better green production technologies and more efficient management methods. This helps cut pollution and also raises productivity and resource use efficiency. The gains from this can pay for the long-term costs of following the rules, so GTFP can keep growing over time [19]. So, the overall effect of environmental regulation on GTFP depends on which of these two effects is stronger. If the innovation–compensation effect is bigger than the compliance costs, it helps GTFP grow. If not, it holds it back [25].

Different types of environmental regulation also have different effects on GTFP [26]. Researchers usually put them into three groups: command-and-control, market-based incentives, and voluntary public participation [27,28]. For the mining industry, the first two types of policy tools made its green transition go through a “first down, then up” path. In the construction industry, Bai et al. (2023) found that command-and-control and market-based rules had an inverted U-shaped link with GTFP, but rules based on public participation did not have a clear direct effect [29]. This shows that the effects of environmental regulation change with the type of rule and the industry. So, when making these rules, we need to think about what each industry is like and how the rules will work.

In conclusion, more empirical research is needed to confirm the precise effects of diverse environmental restrictions on green total factor production (GTFP). The majority of research has mostly concentrated on economic sectors like industry, manufacturing, and urban economies, despite the fact that the body of existing literature has produced significant insights into the types, strictness, and economic effects of environmental regulations as well as the factors that influence GTFP [30,31,32,33]. There is still a dearth of research that focuses on the animal husbandry industry. Since the animal husbandry industry contributes significantly to pollution, national environmental governance places a high priority on it. However, the impact of various environmental regulations on GTFP in this specific sector has not been sufficiently studied.

The aim of this study is to empirically investigate the impact of heterogeneous environmental regulations on the green total factor productivity (GTFP) of the livestock industry, as well as the underlying mechanisms. Specifically, it seeks to examine the potential mediating role of technological innovation in the relationship between environmental regulations and GTFP, and to assess the moderating effect of economic development levels. The study intends to offer both a theoretical framework and practical insights for optimizing the policy mix of environmental regulations and fostering the green transition of the livestock industry.

2. Material and Methods

2.1. Research Hypotheses

2.1.1. Examination of the Effects of Diverse Environmental Regulations on the Livestock Industry’s Green Total Factor Productivity

The dynamic equilibrium among the “cost-compliance effect” and the “innovation-compensation effect” is the fundamental mechanism by which environmental rules alter producer behavior through economic levers. The cost-compliance effect suggests that strict environmental regulations force firms to allocate capital to discharge fees and pollution prevention, raising production costs and potentially crowding out R&D investment [34]. Conversely, the innovation compensation effect argues that well-designed environmental rules can encourage firms to boost R&D and human capital investments, improving resource allocation and technological capabilities, thereby enhancing competitiveness and manufacturing efficiency [35].

Because they use different mechanisms, different environmental initiatives have varying degrees of efficacy. This study follows the conventional taxonomy by dividing environmental regulations into market-based incentive rules and command-and-control regulations. In order to create legally binding environmental standards, laws, and regulations, command-and-control regulations mostly rely on government administrative authority. They enforce compliance by directly monitoring and managing businesses that harm the environment. Typical policy tools in this category include technical specifications and environmental standards. Government-established laws and regulations must be followed by economic entities, including cattle operations. Small-scale operations with limited financial resources are disproportionately burdened by the necessary environmental investments, which raise production costs for cattle firms. As a result, some financially strapped businesses may be forced to leave the market due to these growing environmental expenses [36]. Long-term, nevertheless, this procedure may make it easier to choose livestock businesses with a competitive edge. Green production methods are more likely to be used by the remaining companies, which will raise GTFP. On the other hand, market-based incentive rules follow the “polluter pays” principle and are based on market mechanisms [37]. In order to lower overall pollution levels and direct society pollution toward an optimal and manageable state, they use market signals to affect business emission behavior. Environmental taxes, tradable emission permits, and subsidies are examples of common policy tools [38]. Market-based incentive rules provide businesses more freedom to make decisions. By affecting the market environment where economic activity occurs, the government seeks to internalize environmental externalities rather than directly influencing production decisions. This framework enables businesses to make independent production and operational decisions, enabling them to adopt cost-effective compliance measures, including emission limits, based on their particular circumstances. We proposed the following theory in view of the several mechanisms already mentioned:

H1. 

Heterogeneous environmental regulations have a nonlinear impact on GTFP in the livestock sector.

H2. 

Command-and-control and market-based incentive regulations exert differential effects on the GTFP of the livestock industry.

2.1.2. Mediating Effect of Technological Innovation

There are “strong” and “weak” versions of the Porter Hypothesis, which holds that carefully drafted environmental laws can promote innovation and boost competitiveness [18]. According to the “strong” version, livestock businesses may increase their technological innovation efforts in response to well-crafted environmental restrictions. This perspective claims that by changing cost–benefit analyses, environmental restrictions stimulate enterprise innovation. For example, the implementation of pollution fines drives up the cost of polluting inputs, forcing businesses to look for and implement factor-substituting solutions. In order to minimize their overall compliance costs, businesses are encouraged to implement technologies whose marginal abatement cost is less than the pollution fee rate [39]. Thus, this procedure fosters business innovation and improves resource efficiency, both of which support the expansion of GTFP [40,41,42]. These improvements can improve overall firm competitiveness in addition to offsetting the higher expenses imposed by environmental legislation. On the other hand, the “weak” version admits that eco-friendly inventions may be sparked by environmental restrictions but argues that the transmission mechanism by which these innovations effect GTFP is less direct and the overall result is still unclear. The final result depends on a difficult trade-off between the advantages of innovation and the costs of regulations, and it is further mitigated by firm-specific capabilities. Environmental restrictions are acknowledged for their public usefulness in enhancing general societal welfare from a cost–benefit standpoint. However, due to limited resources, the related private expenses incurred by businesses to adhere to environmental regulations may discourage R&D investments, thereby impeding technological progress [43]. By merging the concepts of the two variations in the Porter Hypothesis, we propose that technological innovation serves as an important mediator. This raises the following theory:

H3. 

The relationship between diverse environmental legislation and GTFP in animal husbandry is mediated by technological innovation.

2.1.3. Moderating Effect of Economic Development Level

One of the most important indicators of a region’s economic strength is its level of economic development. The regional gross domestic output per capita is used in this study to measure this statistic. However, the degree of economic development may have a negative moderating influence on the process by which environmental rules promote advances in the green TFP of animal husbandry. Firstly, economically developed regions typically adopt stringent environmental regulations at an earlier stage. The possibility for further improvement of green TFP in animal husbandry is limited once principal pollution sources are under control since the marginal cost of additional emissions reduction rises dramatically [44]. When environmental regulation strength exceeds a specific level, the decreasing phase of the inverted U-shaped connection may be initiated [45]. Second, in economically developing regions, the pressure on investment in technological innovation could increase significantly due to stricter environmental regulations, strong economic capacity and clear growth targets [46]. Local governments may emphasize resource allocation to industries that provide quick GDP gains in order to satisfy these strict economic growth targets, putting short-term growth ahead of environmental quality. This strategy takes funds away from long-term investments needed to achieve strategic goals like green growth and innovation-driven development [47]. As a result, this may lead to institutional obstacles that stifle creativity, which would ultimately have a detrimental overall effect on the green TFP of animal husbandry. We put out the following theory in light of this analysis:

H4. 

The link between diverse environmental rules and GTFP in the livestock business is negatively moderated by economic development.

2.2. Empirical Testing Model Construction

2.2.1. Benchmark Regression Model

In the livestock sector, the green total factor productivity (GTFP) is a continuous, non-negative variable that falls under the category of restricted dependent variables. Based on Zhao’s methods, this study used a fixed effect model [27], which was configured as follows:

$${\mathrm{G}\mathrm{T}\mathrm{F}\mathrm{P}}_{\mathrm{i},\mathrm{t}}=\alpha +\beta {\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{j},\mathrm{i},\mathrm{t}}+\sum \mu {\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}_{\mathrm{i},\mathrm{t}}+{\mathrm{u}}_{\mathrm{i}}+{\mathrm{v}}_{\mathrm{t}}+{\epsilon }_{\mathrm{i},\mathrm{t}}$$

(1)

The green total factor productivity for the livestock sector for province i in year t is represented in the model by ${\mathrm{G}\mathrm{T}\mathrm{F}\mathrm{P}}_{\mathrm{i},\mathrm{t}}$. $\beta {\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{j},\mathrm{i},\mathrm{t}}$ is the j-th type of environmental regulation variable, and β is its corresponding coefficient. ${\mathrm{u}}_{\mathrm{i}}$ and ${\mathrm{v}}_{\mathrm{t}}$ capture province-specific fixed effects and year-fixed effects, respectively. ${\epsilon }_{\mathrm{i},\mathrm{t}}$ is the idiosyncratic error term. The analysis encompasses 30 provincial-level regions in China.

$${\mathrm{G}\mathrm{T}\mathrm{F}\mathrm{P}}_{\mathrm{i},\mathrm{t}}=\alpha +{\beta }_1{\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{j},\mathrm{i},\mathrm{t}}+{\beta }_2{\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{j},\mathrm{i},\mathrm{t}}^2+\sum \mu {\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}_{\mathrm{i},\mathrm{t}}+{\mathrm{u}}_{\mathrm{i}}+{\mathrm{v}}_{\mathrm{t}}+{\epsilon }_{\mathrm{i},\mathrm{t}}$$

(2)

2.2.2. Mediating Effect Model

According to theoretical research, environmental laws may have an indirect impact on the GTFP of the livestock industry through technical innovation. To empirically test this mediating mechanism, we follow the established procedure by Wen et al. (2014) and introduce technological innovation (INN) as the mediating variable [48]. Building upon Equation (1), we construct the following mediation effect model:

$${\mathrm{G}\mathrm{T}\mathrm{F}\mathrm{P}}_{\mathrm{i},\mathrm{t}}={\alpha }_0+{\beta }_0{\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{j},\mathrm{i},\mathrm{t}}+\sum {\mu }_0{\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}_{\mathrm{i},\mathrm{t}}{+\mathrm{u}}_{\mathrm{i}}+{\mathrm{v}}_{\mathrm{t}}+{\epsilon }_{\mathrm{i},\mathrm{t}}$$

(3)

$${\mathrm{I}\mathrm{N}\mathrm{N}}_{\mathrm{i},\mathrm{t}}={\alpha }_1+{\beta }_1{\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{j},\mathrm{i},\mathrm{t}}+\sum {\mu }_1{\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}_{\mathrm{i},\mathrm{t}}+{\mathrm{u}}_{\mathrm{i}}+{\mathrm{v}}_{\mathrm{t}}+{\epsilon }_{\mathrm{i},\mathrm{t}}$$

(4)

$${\mathrm{G}\mathrm{T}\mathrm{F}\mathrm{P}}_{\mathrm{i},\mathrm{t}}={\alpha }_2+{\beta }_2{\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{j},\mathrm{i},\mathrm{t}}+\mathsf{\gamma }{\mathrm{I}\mathrm{N}\mathrm{N}}_{\mathrm{i},\mathrm{t}}+\sum {\mu }_2{\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}_{\mathrm{i},\mathrm{t}}+{\mathrm{u}}_{\mathrm{i}}+{\mathrm{v}}_{\mathrm{t}}+{\epsilon }_{\mathrm{i},\mathrm{t}}$$

(5)

In Equation (3), coefficient β₀ captures the total effect of environmental regulations ${\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{j},\mathrm{i},\mathrm{t}}$ on ${\mathrm{G}\mathrm{T}\mathrm{F}\mathrm{P}}_{\mathrm{i},\mathrm{t}}$. In Equation (4), coefficient ${\beta }_1$ measures the effect of ${\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{j},\mathrm{i},\mathrm{t}}$ on the mediating variable, ${\mathrm{I}\mathrm{N}\mathrm{N}}_{\mathrm{i},\mathrm{t}}$. In Equation (5), coefficient $\mathsf{\gamma }$ represents the effect of the mediating variable ${\mathrm{I}\mathrm{N}\mathrm{N}}_{\mathrm{i},\mathrm{t}}$ on ${\mathrm{G}\mathrm{T}\mathrm{F}\mathrm{P}}_{\mathrm{i},\mathrm{t}}$, after controlling for the independent variable ${\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{j},\mathrm{i},\mathrm{t}}$. Coefficient ${\beta }_2$ signifies the direct effect of ${\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{j},\mathrm{i},\mathrm{t}}$ on ${\mathrm{G}\mathrm{T}\mathrm{F}\mathrm{P}}_{\mathrm{i},\mathrm{t}}$, after controlling for the mediating variable ${\mathrm{I}\mathrm{N}\mathrm{N}}_{\mathrm{i},\mathrm{t}}$. The term ${\epsilon }_{\mathrm{i},\mathrm{t}}$ denotes the idiosyncratic error term across all equations. Therefore, through technological innovation, the impact of the β₁ × γ product on the environmental protection rules surrounding GTFP is measured.

2.2.3. Moderating Effect Model

To examine the suggested moderating effect of economic development level (PGDP) (H4), we include an interaction term between environmental limitations and the moderator in our empirical design. The following is the specification of the moderating effect model:

$${\mathrm{G}\mathrm{T}\mathrm{F}\mathrm{P}}_{\mathrm{i},\mathrm{t}}={\alpha }_3+{\beta }_3{\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{j},\mathrm{i},\mathrm{t}}+{{\beta }_4\mathrm{P}\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i},\mathrm{t}}\ast {\beta }_5{\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{i},\mathrm{t}}+\sum {\mu }_3{\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}_{\mathrm{i},\mathrm{t}}+{\mathrm{u}}_{\mathrm{i}}+{\mathrm{v}}_{\mathrm{t}}+{\epsilon }_{\mathrm{i},\mathrm{t}}$$

(6)

$${\mathrm{G}\mathrm{T}\mathrm{F}\mathrm{P}}_{\mathrm{i},\mathrm{t}}={\alpha }_4+{\beta }_6{\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{j},\mathrm{i},\mathrm{t}}+{{\beta }_7\mathrm{P}\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i},\mathrm{t}}\ast {\beta }_8{\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{i},\mathrm{t}}+{{\beta }_9\mathrm{P}\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i},\mathrm{t}}\ast {\beta }_6{\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{i},\mathrm{t}}^2+\sum {\mu }_4{\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}_{\mathrm{i},\mathrm{t}}+{\mathrm{u}}_{\mathrm{i}}+{\mathrm{v}}_{\mathrm{t}}+{\epsilon }_{\mathrm{i},\mathrm{t}}$$

(7)

In Equation (6), the term ${\mathrm{P}\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i},\mathrm{t}}\ast {\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{i},\mathrm{t}}$ represents the interaction term between environmental regulation and the moderating variable (economic development level). In Equation (7), ${\mathrm{P}\mathrm{G}\mathrm{D}\mathrm{P}}_{\mathrm{i},\mathrm{t}}\ast {\mathrm{R}\mathrm{e}\mathrm{g}\mathrm{u}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{i},\mathrm{t}}^2$ represents the interaction term between the moderating variable and the squared term of environmental regulation. The definitions of all other variables remain consistent with the previous equations.

2.3. Variable Selection and Description

2.3.1. Dependent Variable

Green Total Factor Productivity (GTFP) in animal husbandry. By selecting relevant indicators in accordance with the research objectives and building upon well-established methodologies documented in the literature, specifically those developed by Lin et al. (2023), Geng et al. (2013), Xie et al. (2025), Hu and Wang (2010), and Xu et al. (2019), we systematically constructed a comprehensive evaluation index system [49,50,51,52,53]. As shown in Table 1.

Table 1. Indicator System for Green Total Factor Productivity (GTFP) in Animal Husbandry.

Category	Variable	Evaluating Indicator
Input	Labor input	The Number of employees in the animal husbandry sector
	Intermediate Consumption	Intermediate Consumption in Animal Husbandry
	Capital Input	Capital Stock in Animal Husbandry
output	Desirable output	Total Output of the Animal Husbandry Sector
	Undesirable output	Pollutant Emissions from Animal Husbandry
		Greenhouse Gas Emissions from Animal Husbandry

This study adopts the following input variables for the livestock sector:

(1) Labor input: The number of employees in the livestock industry is estimated by multiplying the proportion of livestock output value in the total output value of the primary industry by the total number of employees in the primary industry.

(2) Intermediate consumption: Intermediate consumption includes various inputs required for production, such as feed and veterinary drugs. Due to the unavailability of detailed intermediate consumption data in statistical yearbooks after 2018, intermediate consumption is effectively derived by subtracting the value added of the livestock industry from its gross output value.

(3) Capital input: Capital input is measured using the fixed capital stock of the livestock industry, estimated via the perpetual inventory method. The base year is set as 2010, and all relevant data are deflated using the agricultural means of production price index. The depreciation rate of capital stock (δ) is set at 5.42%. The calculation formula is:

$${\mathrm{K}}_{\mathrm{t}}={\mathrm{K}}_{\mathrm{t}-1}(1-\delta )+{\mathrm{I}}_{\mathrm{t}}$$

(8)

where ${\mathrm{K}}_{\mathrm{t}}$ and ${\mathrm{K}}_{\mathrm{t}-1}$ represent the capital stock in the current and previous periods, respectively; ${\mathrm{I}}_{\mathrm{t}}$ is the investment in year t (100 million yuan); and $\delta$ is the asset depreciation rate. The initial capital stock (${\mathrm{K}}_0$) is represented by the fixed asset investment in agriculture, forestry, animal husbandry, and fishery. Livestock fixed asset investment (${\mathrm{I}}_{\mathrm{t}}$) is approximated by multiplying the fixed asset investment in agriculture, forestry, animal husbandry, and fishery by the ratio of livestock output value to the total output value of agriculture, forestry, animal husbandry, and fishery.

The output variables in this study are divided into desirable and undesirable outputs:

(1) Desirable output: The gross output value of the livestock industry is used as the desirable output indicator.

(2) Undesirable output: Greenhouse gas emissions and pollutant emissions from the livestock industry are adopted as undesirable output indicators. The emission coefficients for CH₄ and N₂O are presented in the following Table S1. Table S2 lists the representative emission coefficient values for major livestock and poultry categories.

GTFP is computed using the MATLAB 2022b version of the super-efficiency SBM model with undesired outputs. The following describes each indicator’s particular measurement techniques:

$$\mathrm{m}\mathrm{i}\mathrm{n}\rho ={\displaystyle \frac{\frac{1}{\mathrm{m}}{\sum }_{\mathrm{i}=1}^{\mathrm{m}}\frac{\mathrm{x}}{{\mathrm{x}}_{\mathrm{i}\mathrm{k}}}}{\frac{1}{{\mathrm{l}}_1+{\mathrm{l}}_2}\left({\sum }_{\mathrm{r}=1}^{{\mathrm{l}}_1}\frac{{\mathrm{y}}^{\mathrm{w}}}{{\mathrm{y}}_{\mathrm{r}\mathrm{k}}^{\mathrm{w}}}+{\sum }_{\mathrm{q}=1}^{{\mathrm{l}}_2}\frac{{\mathrm{y}}^{\mathrm{b}}}{{\mathrm{y}}_{\mathrm{q}\mathrm{k}}^{\mathrm{b}}}\right)}}$$

(9)

$$\mathrm{s}.\mathrm{t}.{\mathrm{x}}_{\mathrm{i}\mathrm{k}}\ge {\sum }_{\mathrm{j}=1,\ne \mathrm{k}}^{\mathrm{n}}\lambda {\mathrm{x}}_{\mathrm{i}\mathrm{j}};{\mathrm{y}}_{\mathrm{r}\mathrm{k}}^{\mathrm{w}}\le {\sum }_{\mathrm{j}=1,\ne \mathrm{k}}^{\mathrm{n}}\lambda {\mathrm{y}}_{\mathrm{r}\mathrm{j}}^{\mathrm{w}};{\mathrm{y}}_{\mathrm{q}\mathrm{k}}^{\mathrm{b}}\ge {\sum }_{\mathrm{j}=1,\ne \mathrm{k}}^{\mathrm{n}}{\lambda }_{\mathrm{j}}{\mathrm{y}}_{\mathrm{q}\mathrm{j}}^{\mathrm{b}};{\lambda }_{\mathrm{j}}\ge 0;\mathrm{i}=1,2,\dots ,\mathrm{m};\mathrm{j}=1,2,\dots ,\mathrm{n};\mathrm{r}=1,2,\dots ,{\mathrm{l}}_1;\mathrm{q}=1,2,\dots ,{\mathrm{l}}_2$$

The number of decision-making units (DMUs) in this model is denoted by n. For the production of l₁ suitable outputs and l₂ undesirable outputs, each DMU uses m inputs. The slack variables of inputs, desired outputs, and undesirable outputs are represented by the symbols x, ${\mathrm{y}}^{\mathrm{w}}$, and ${\mathrm{y}}^{\mathrm{b}}$, respectively. Let ${\mathrm{x}}_{\mathrm{i}\mathrm{j}}$ be the i-th input of the j-th DMU; ${\mathrm{y}}_{\mathrm{r}\mathrm{k}}^{\mathrm{w}}$ the r-th desirable output of the j-th DMU; and ${\mathrm{y}}_{\mathrm{q}\mathrm{j}}^{\mathrm{b}}$ the q-th undesirable output of the j-th DMU. The terms ${\mathrm{x}}_{\mathrm{i}\mathrm{k}}$, ${\mathrm{y}}_{\mathrm{r}\mathrm{k}}^{\mathrm{w}}$, and ${\mathrm{y}}_{\mathrm{q}\mathrm{k}}^{\mathrm{b}}$ denote the i-th input, r-th desirable output, and q-th undesirable output of the k-th DMU after projection onto the efficiency frontier, respectively. The vector λ represents the intensity vector.

2.3.2. Core Explanatory Variables

According to the existing literature [54], this study categorizes environmental regulations into two types: those based on management and control (CER) and those driven by market incentives (IER). IERs are measured using a composite index consisting of three components. These components include the ratio of tax revenue from pollution control measures to industrial emissions reduction value added, the share of GDP in environmental investment, and the ratio of environmental investment in industry to industrial output value [55]. The number of local environmental laws and the frequency of phrases associated with regulatory intensity in regional policy documents, both processed using logarithms, are used to create a composite index that quantifies CER [18,55]. The environmental regulation indicator system is shown in Table 2.

Table 2. Environmental Regulation Indicator System.

Primary Indicator	Secondary Indicator
CER	Number of Local Environmental Regulations
	Stringency of Environmental Regulation
IER	Share of Investment in Environmental Pollution Treatment in GDP (%)
	Share of Pollution Discharge Fee Revenue in Industrial Value-Added (%)
	Share of Completed Industrial Pollution Treatment Investment in Value-Added (%)

When environmental regulations do not explicitly target the agricultural sector, methodologically identifying the causal effects of industrial policies on agricultural outcomes poses a substantial challenge. Li et al. (2024) [56] address this issue by leveraging China’s Two Control Zones policy, a flagship initiative aimed at regulating industrial sulfur dioxide emissions, and employing a difference-in-differences approach. Their analysis, which accounts for both intensive and extensive margins, reveals that the policy led to an approximate 9% increase in agricultural value added and an 11.6% rise in rural per capita net income. These effects are primarily attributable to adjustments in land allocation and shifts in fertilizer use, rather than to increases in the yield of individual crops. This study thus provides both a credible identification strategy and robust empirical evidence supporting the causal pathway from industrial regulation to agricultural outcomes.

Against the backdrop of China’s fiscal decentralization and administrative accountability system, total pollutant emission reduction targets are assigned at the regional level as integrated mandates. Under this system, both industrial sources and agricultural sources, particularly livestock and poultry operations, fall within the scope of regulatory assessment. Local governments that develop extensive experience in industrial pollution control and make adept use of policy instruments such as pollution levy systems often extend analogous approaches, including manure treatment subsidies and ecological compensation mechanisms, to guide the green transformation of the livestock sector. This dynamic gives rise to notable policy spillovers across sectors. Consequently, the stringency of industrial regulation can serve as a valid proxy for the intensity of environmental oversight applied to livestock operations and for the level of institutional capacity in agricultural environmental governance. As such, it indirectly reflects the substantive enforcement outcomes and the effectiveness of regulatory efforts aimed at fostering environmental compliance and transition within the agricultural sector.

2.3.3. Mediating Variable

The mediating variable in this study is technological innovation (INN). This study adopts the methodology of Wu et al. (2019) and uses the natural logarithm of the number of granted invention-patent applications to represent technological innovation, whereas the existing literature typically measures technological innovation using regional R&D expenditure or patent counts [57,58].

2.3.4. Moderating Variable

The degree of regional economic development, as measured by per capita GDP (PGDP), is the moderating variable. The China Statistical Yearbook provides the yearly GDP figures for every province. To remove the impact of price fluctuations, all nominal GDP numbers are deflated to constant 2010 prices using provincial GDP deflators.

2.3.5. Control Variables

Drawing on previous research, a set of control factors is included to the model to reduce potential omitted variable bias [59,60]. The selected controls and their measurements are as follows:

(1) Trade Openness (TO): The indicator is measured as a percentage of the region’s gross domestic product.

(2) Human Capital (HC): Measured by the average years of schooling of the rural population.

(3) Fiscal Support (FS): This indicator reflects the extent of government intervention and support in the agricultural sector, measured by the proportion of total government budget expenditures to gross domestic product.

(4) Industrialization Level (IL): Measured as the share of industrial value-added in regional GDP.

2.4. Data Sources

The primary data used in this study are derived from official statistical sources including the China Statistical Yearbook, China Animal Husbandry and Veterinary Yearbook, China Taxation Yearbook, China City Statistical Yearbook, China Environmental Statistical Yearbook, China Rural Statistical Yearbook, and China Agricultural Statistics, and China Population and Employment Statistics Yearbook, supplemented by data from the National Bureau of Statistics of China official website and various provincial statistical yearbooks. To address missing data for certain indicators in specific provinces or years, which account for less than five percent of the total research sample, a linear interpolation method was applied to complete the dataset, thereby ensuring data integrity and the validity of empirical analysis. The descriptive statistics for every variable utilized in this investigation are shown in Table 3.

Table 3. Descriptive Statistics.

VARIABLES	Sample Size	Average	Std.	Min	Max
GTFP	390	0.691	0.293	0.000	1.333
IER	390	0.182	0.131	0.010	0.854
CER	390	0.684	0.094	0.398	0.925
TO	390	0.276	0.291	0.008	1.464
HC	390	7.819	0.627	5.88	10.11
IL	390	0.325	0.083	0.101	0.556
FS	390	0.251	0.105	0.106	0.758

3. Results

3.1. Multicollinearity Test

To assess the model’s reliability, we first check for potential multicollinearity among the variables using the Variance Inflation Factor (VIF). A VIF value below 10 indicates the absence of significant multicollinearity. All independent variables show low VIF values, with a maximum of 2.330, which is well below the typical criterion of 10, as shown in Table 4. The robustness of the ensuing conclusions is thus supported by the likelihood that multicollinearity will not skew the parameter estimations in our empirical model.

Table 4. Multicollinearity Test Results.

VARIABLES	VIF	1/VIF
FS	2.330	0.430
HC	1.880	0.531
IL	1.510	0.663
TO	1.470	0.679
IER	1.290	0.774
CER	1.170	0.857
Mean VIF	1.610

3.2. Benchmark Regression

We begin by estimating a fixed-effects model to examine the relationship between environmental regulations and GTFP in the livestock industry. The regression results are presented in Table 5. As shown in column (1), which reports the results for market-based incentive environmental regulations (IER), the coefficient of the linear term of IER is positive and statistically significant at the 1% level, whereas the coefficient of the quadratic term is negative and significant at the 5% level. This confirms an inverse U-shaped association between market-based incentive environmental regulations and green total factor productivity. Based on the estimated coefficients, the turning point of this relationship is calculated to be at a market-based incentive environmental regulations value of 0.732. An examination of the sample distribution reveals that this turning point lies between the 75th and 90th percentiles of the market-based incentive environmental regulations variable, indicating that for the majority of sample observations, approximately 75% to 90%, market-based incentive environmental regulations remain on the left side of the curve, where their effect on green total factor productivity is positive. This implies that within the prevailing regulatory intensity range, market-based incentive environmental regulations predominantly encourage green total factor productivity growth through the innovation compensation effect. By offering livestock businesses financial incentives to implement clean technologies such as biogas projects and manure recycling, appropriate market-based incentive environmental regulations improve resource efficiency and environmental performance, which supports the expansion of green total factor productivity. However, for the subsample of regions with the highest market-based incentive environmental regulations intensities beyond the turning point, the rising costs of compliance, including those related to investing in cleaner alternatives and phasing out polluting behaviors, may begin to displace other profitable investments. At this stage, the cost-compliance effect starts to dominate, inhibiting green total factor productivity growth.

Table 5. Baseline Regression Results.

VARIABLES	(1)	(2)
	GTFP	GTFP
CER	−1.071 ***
	(−2.88)
CER2	0.871 **
	(2.29)
IER		1.043 ***
		(2.67)
IER2		−0.713 **
		(−2.07)
TO	−0.293 **	−0.129
	(−2.45)	(−0.84)
HC	0.041	0.017
	(0.72)	(0.31)
IL	−0.028	0.180
	(−0.12)	(0.67)
FS	−1.392 ***	−1.552 ***
	(−3.78)	(−3.85)
Constant	1.109 **	0.767
	(2.25)	(1.55)
Observations	390	390
R-squared	0.733	0.737
Year FE	YES	YES
City FE	YES	YES

Note: Values in parentheses represent t-statistics; ** and *** denote significance levels of 5% and 1%, respectively.

The findings for command-and-control environmental regulations (CER) are shown in column (2). The quadratic term has a significantly positive coefficient, whereas the linear term has a significantly negative coefficient. This trend suggests that CER and GTFP have a U-shaped connection. The estimated turning point is at a command-and-control environmental regulations value of 0.614. A comparison with the descriptive statistics shows that this turning point is located near the median of the command-and-control environmental regulations distribution. Consequently, for regions with command-and-control environmental regulations intensity below this threshold, the relationship is negative; for those above it, the relationship becomes positive. In the initial stage, which corresponds to regions with lower command-and-control environmental regulations intensity, direct compliance expenses including equipment upgrades, management adjustments, and potential fines increase production costs for companies. In the short term, this cost burden suppresses green total factor productivity by crowding out alternative profitable investments and decreasing production efficiency. Small and medium-sized businesses with limited resources may be particularly affected by this suppression effect. However, for regions that have crossed the turning point, the long-term effect of command-and-control environmental regulations begins to manifest. These stricter regulations force businesses to innovate technologically and undergo structural changes to meet compliance standards. The adoption of cutting-edge technologies and streamlined production procedures ultimately enhances green total factor productivity. Additionally, by increasing industry concentration and phasing out obsolete production capacity, these regulatory pressures improve the sector’s overall efficiency. Thus, the U-shaped relationship captures an initial suppression followed by a long-term enhancement effect, with the turning point serving as a critical threshold for policy design. When combined, these results offer strong empirical evidence in support of Hypotheses H1 and H2.

The coefficients of FS are significantly negative, indicating that under the size preference orientation of fiscal support for agriculture, relatively large-scale enterprises with lower technical efficiency are more likely to obtain subsidies by leveraging their stronger capacity for policy negotiation, whereas smaller-scale operators with high innovation potential are crowded out of the market. This leads to a reverse flow of fiscal resources from high-efficiency to low-efficiency sectors [61]. Consequently, the intensity of fiscal support is negatively correlated with green productivity, creating a lock-in effect for low-efficiency, high-pollution farms and thereby delaying the process of environmental selection and industrial upgrading. Furthermore, against the backdrop of an underdeveloped financial market, excessive reliance on fiscal support induces a path dependence on subsidies among agricultural operators, weakening their willingness and ability to finance through market-oriented instruments such as green credit and green bonds. This substitution of market mechanisms by fiscal resources further reduces the overall marginal efficiency of capital allocation.

3.3. Robustness Test

3.3.1. Spatial Lag Model Robustness Check

To account for potential spatial dependence, this study employs the Spatial Lag Model (SAR) as a robustness check. Table 6 presents the estimation results of the SAR model. The estimated direct effects of environmental regulations in the SAR model are highly consistent with those from the two-way fixed effects model in terms of sign, magnitude, and statistical significance, indicating that the baseline findings are robust to alternative specifications that explicitly model spatial dependence. Moreover, the indirect effects are not statistically significant, further supporting the validity of our research design, which focuses on local direct effects. Overall, whether or not spatial lag effects are taken into account, the impact of local environmental regulations on the green total factor productivity of the livestock industry remains stable, confirming that the baseline conclusions are not contingent on a specific spatial model specification.

Table 6. Spatial Lag Test Results.

VARIABLES	(1)		(2)		(3)
	Direct		Indirect		Total
CER		−1.015 ***		−0.135		−1.150 **
		(0.386)		(0.100)		(0.458)
CER2		0.799 **		0.108		0.907 **
		(0.378)		(0.088)		(0.445)
IER	1.033 ***		0.104		1.137 ***
	(0.311)		(0.074)		(0.335)
IER2	−0.773 **		−0.077		−0.850 **
	(0.347)		(0.064)		(0.381)
TO	−0.129	−0.290 **	−0.015	−0.038	−0.145	−0.328 **
	(0.125)	(0.116)	(0.019)	(0.026)	(0.141)	(0.134)
HC	0.016	0.035	0.001	0.004	0.018	0.040
	(0.052)	(0.052)	(0.007)	(0.008)	(0.057)	(0.058)
IL	0.157	0.008	0.018	0.004	0.175	0.013
	(0.309)	(0.306)	(0.044)	(0.049)	(0.346)	(0.350)
FS	−1.480 ***	−1.338 ***	−0.153	−0.169 *	−1.633 ***	−1.507 ***
	(0.304)	(0.297)	(0.106)	(0.097)	(0.332)	(0.331)

Note: Standard errors in parentheses; * p < 0.1, ** p < 0.05, *** p < 0.01.

3.3.2. Proxy Variable Robustness Check

We use a proxy variable technique to evaluate the robustness of our results. In particular, instrumental proxies are the one-period delayed values of the linear and quadratic terms of CER and IER. These proxy variables are used to re-estimate the regression models. Table 7 presents the robustness check’s findings.

Table 7. Robustness Test Results.

VARIABLES	(1) The Lagged Term of CER Employed as a Proxy Variable.		(2) The Lagged Term of IER Employed as a Proxy Variable.
	CER _{t−1}	CER _{t−2}	IER_{t−1}	IER_{t−2}
L.CER	−1.108 ***
	(−2.74)
L.CER2	1.005 **
	(2.46)
L2.CER		−1.263 ***
		(−2.79)
L2.CER2		1.256 **
		(2.51)
L.IER			1.046 ***
			(3.05)
L.IER2			−0.767 **
			(−2.47)
L2.IER				0.684 **
				(2.34)
L2.IER2				−0.422
				(−1.60)
TO	−0.333 ***	−0.408 ***	−0.184	−0.314 **
	(−2.71)	(−3.33)	(−1.27)	(−2.48)
HC	0.036	0.030	0.024	0.015
	(0.64)	(0.52)	(0.42)	(0.27)
IL	0.001	−0.026	0.049	−0.142
	(0.00)	(−0.10)	(0.19)	(−0.53)
FS	−1.120 ***	−1.016 ***	−1.341 ***	−1.240 ***
	(−3.12)	(−2.73)	(−3.38)	(−2.94)
Constant	1.061 **	1.113 **	0.716	0.887 *
	(2.24)	(2.40)	(1.45)	(1.83)
Observations	360	330	360	330
R-squared	0.753	0.762	0.757	0.762
Year FE	YES	YES	YES	YES
City FE	YES	YES	YES	YES

Note: Values in parentheses represent t-statistics; *, **, and *** denote significance levels of 10%, 5%, and 1%, respectively.

The findings demonstrate the qualitative consistency of our main conclusions. For the important variables, the coefficients’ signs and statistical significance levels are still in accord with the initial estimations. This validates the strength of our previous findings.

3.4. Heterogeneity Analysis

Significant geographical differences in China’s institutional settings, resource endowments, and economic development result in wide variations in the strictness and application of environmental laws. Because of this variety, various regions may have varying effects from environmental constraints on cattle GTFP. We use Li et al. (2020)’s regional classification to examine these possible heterogeneous effects, and we estimate our model independently for China’s eastern, central, and western regions [62]. Table 8 shows the regression results.

Table 8. Heterogeneity Analysis Results.

VARIABLES	(1) CER			(2) IER
	Eastern	Central	Western	Eastern	Central	Western
CER	−1.294 **	−0.613	−0.924
	(−2.26)	(−0.93)	(−0.99)
CER2	1.249 **	0.517	0.699
	(2.14)	(0.80)	(0.78)
IER				0.511	−0.165	1.511 **
				(0.63)	(−0.35)	(2.23)
IER2				0.183	0.452	−1.459 **
				(0.09)	(0.66)	(−2.08)
TO	−0.359 **	0.603	0.915	−0.254	0.800 **	1.064 *
	(−2.44)	(1.34)	(1.55)	(−1.53)	(2.01)	(1.87)
HC	0.049	−0.235 ***	0.054	−0.007	−0.213 ***	0.069
	(0.71)	(−3.18)	(0.41)	(−0.11)	(−3.00)	(0.52)
IL	−0.083	0.175	−0.061	−0.359	0.212	−0.050
	(−0.13)	(0.66)	(−0.11)	(−0.61)	(0.75)	(−0.08)
FS	−1.091	−0.094	−2.310 ***	−1.260	−0.159	−2.394 ***
	(−1.41)	(−0.25)	(−3.35)	(−1.62)	(−0.42)	(−3.52)
Constant	1.163 *	2.613 ***	1.118	1.291 **	2.252 ***	0.515
	(1.79)	(3.91)	(0.99)	(1.99)	(3.94)	(0.46)
Observations	143	104	143	143	104	143
R-squared	0.783	0.910	0.584	0.779	0.910	0.596
Year FE	YES	YES	YES	YES	YES	YES
City FE	YES	YES	YES	YES	YES	YES

Note: Values in parentheses represent t-statistics; *, **, and *** denote significance levels of 10%, 5%, and 1%, respectively.

Regarding command-and-control environmental regulations, the estimation results reveal pronounced regional heterogeneity. In the eastern region, the coefficient for command-and-control environmental regulation is significantly negative at the 5 percent level, while its quadratic term is significantly positive at the 5 percent level. This suggests a possible U-shaped relationship between command-and-control environmental regulation and green total factor productivity. The initial negative effect may reflect a compliance cost effect, whereas over time, regulated entities may experience an innovation offset effect as they adopt cleaner production technologies to meet regulatory standards. The eastern region, characterized by advanced technological infrastructure and well-developed financial markets, likely possesses the absorptive capacity to transform regulatory pressure into productive innovation. In the central region, neither the linear nor the quadratic term for command-and-control environmental regulation is statistically significant. As a traditional livestock production area with low technology adoption rates and a high proportion of small-scale household farming, the independent effect of command-and-control environmental regulation may be masked by concurrent implementation of other policy instruments. In the western region, both coefficients fail to achieve statistical significance, suggesting that the short-term impact of administrative regulation on green total factor productivity remains unclear. This may be partly attributable to the relatively low baseline level of regulatory enforcement in the region, where regulatory pressure may fall below the threshold necessary to induce behavioral changes among producers.

Regarding market-based incentive environmental regulation, the regression results also exhibit significant regional heterogeneity. In the western region, the coefficient for market-based incentive environmental regulation is significantly positive at the 5 percent level, while its quadratic term is significantly negative at the 5 percent level, indicating an inverted U-shaped relationship. A moderate level of incentive-based regulation may initially enhance productivity by offering flexible compliance options, yet beyond the optimal intensity, the region’s limited technological and economic absorptive capacity may constrain further positive effects. In the eastern region, neither the linear nor the quadratic term for market-based incentive environmental regulation is significant. This may suggest that such regulation has not yet reached a critical intensity sufficient to exert a significant impact, or alternatively, that green total factor productivity in this region is primarily driven by other factors, such as advanced technological innovation capability and the externalities associated with industrial agglomeration. Similarly, in the central region, both coefficients are insignificant. One possible explanation is that the region is undergoing a transition from small-scale decentralized farming to large-scale intensive operations. During this transitional period, the institutional environment remains dynamic and market mechanisms are imperfect, which may delay the full effectiveness of market-based incentive environmental regulation.

3.5. Analysis of Mechanism Effects

3.5.1. Testing Mediating Effects

To further explore the underlying mechanisms, this study employs the stepwise regression approach developed by Wen et al. (2014) to examine mediating effects, with a particular focus on how market-based incentive environmental regulations influence green total factor productivity [48]. The mediating variable is chosen to be technological innovation (INN). Table 9 displays the findings. Table 9 illustrates that command-and-control environmental rules (CER) do not promote the mediating effect of technological innovation. The results in Table 8 indicate that CER does not exert a significant effect on technological innovation. The coefficient for CER in the first-stage regression from CER to INN was −0.107, with a t-statistic of −0.65, indicating a lack of statistical significance. Therefore, the mediating pathway through technological innovation is not supported for CER. A plausible explanation for this finding lies in the cost-compliance effect inherent to rigid command-and-control regulations. Unlike market-based instruments that provide flexible incentives, CER typically mandates uniform emission standards or specific technologies. In response, agricultural producers, particularly small-scale livestock farms with limited capital and technical capacity, are incentivized to adopt end-of-pipe compliance strategies such as installing basic waste treatment facilities rather than investing in process-innovative R&D like precision feeding systems or manure valorization technologies. This pattern reflects a capital misallocation effect: constrained by finite resources, firms prioritize immediate compliance to avoid penalties over long-term, uncertain returns from green innovation. In the agricultural sector, where profit margins are narrow and risk aversion is high, this tendency is especially pronounced. Thus, rather than facilitating technological upgrading, CER may inadvertently crowd out the resources needed for genuine innovation, rendering the mediating pathway inoperative.

Table 9. Mediating Effect Regression Results.

VARIABLES	(1) CER			(2) IER
	GTFP	INN	GTFP	GTFP	INN	GTFP
CER	−0.221 **	−0.107	−0.211 **
	(−2.12)	(−0.65)	(−2.04)
IER				0.492 ***	0.750 ***	0.435 ***
				(3.27)	(3.18)	(2.86)
INN			0.091 ***			0.077 **
			(2.69)			(2.25)
TO	−0.293 **	0.584 ***	−0.346 ***	−0.181	0.727 ***	−0.237 **
	(−2.52)	(3.18)	(−2.97)	(−1.54)	(3.94)	(−1.99)
HC	0.032	0.008	0.031	0.014	−0.020	0.016
	(0.62)	(0.09)	(0.61)	(0.28)	(−0.25)	(0.31)
IL	−0.128	0.590	−0.182	0.179	1.169 **	0.088
	(−0.47)	(1.36)	(−0.66)	(0.60)	(2.51)	(0.30)
FS	−1.444 ***	−1.180 **	−1.336 ***	−1.590 ***	−1.407 ***	−1.481 ***
	(−4.28)	(−2.22)	(−3.97)	(−4.71)	(−2.66)	(−4.37)
Constant	1.034 **	9.490 ***	0.169	0.880 *	9.344 ***	0.160
	(2.25)	(13.12)	(0.30)	(1.94)	(13.14)	(0.29)
Observations	390	390	390	390	390	390
R-squared	0.729	0.972	0.735	0.734	0.973	0.738
Year FE	YES	YES	YES	YES	YES	YES
City FE	YES	YES	YES	YES	YES	YES

Note: Values in parentheses represent t-statistics; *, **, and *** denote significance levels of 10%, 5%, and 1%, respectively.

For IER, a different pattern emerges. In the first-stage regression, the coefficient for IER on INN is 0.750 and statistically significant at the 1% level, indicating a strong positive association between market-based incentives and technological innovation. In the second-stage regression, after controlling for INN, the coefficient for IER on GTFP remains positive and significant (0.435, p < 0.01), while the coefficient for INN on GTFP is 0.077 and significant at the 5% level. This pattern suggests that technological innovation partially mediates the relationship between IER and GTFP. The underlying logic aligns with the innovation compensation effect posited by the Porter Hypothesis. By internalizing environmental costs through price signals such as emissions fees or tradable permits, IER creates continuous incentives for producers to seek cost-effective abatement solutions. In response, agricultural operators, particularly those with sufficient scale and technical capacity, are motivated to adopt or develop green technologies, including intelligent feeding equipment, low-pollution feed formulations, and efficient manure treatment systems. These innovations enhance resource efficiency and reduce environmental impact per unit of output, ultimately contributing to GTFP growth.

This study adopts the Bootstrap approach for extra testing in order to confirm the robustness of the mediating influence of technical innovation. With a 95% confidence level, the sampling number is set at 1000. The testing principle states that if the indirect impact’s 95% confidence interval does not include zero, a mediating effect has been proven. As reported in Table 10, the bias-corrected 95% confidence intervals are [0.048, 0.102] for the direct effect and [−0.039, −0.012] for the indirect effect. Additionally, the zero value is not included in the intervals for either the direct effect or the indirect effect. The trustworthiness of our findings about the impact of technical innovation is further confirmed by this result, which offers strong statistical evidence for the existence of the mediating effect.

Table 10. Bootstrap Mediating Effect Test Results.

Mediating Variables	Pathway	Effect Coefficient	Standard Error	Z-Statistic	p-Value	95% Confidence Intervals
INN	indirect effect	−0.026	0.007	−3.69	0.000	[−0.039, −0.012]
	direct effect	0.075	0.014	5.43	0.000	[0.048, 0.102]

3.5.2. Moderation Effect Test

To examine whether the relationship between environmental regulations and GTFP is conditional on the level of economic development, we introduce interaction terms between economic development level (PGDP) and both the linear and quadratic terms of the regulation variables. The results of the moderation effect test are presented in Table 11.

Table 11. Moderating Effect Regression Results.

VARIABLES	(1) CER			(2) IER
	GTFP	GTFP	GTFP	GTFP	GTFP	GTFP
CER	−0.221 **	−0.293	−0.198
	(−1.98)	(−1.39)	(−0.90)
PGDP		0.147	0.230 **		0.191 **	0.162 *
		(1.62)	(2.15)		(2.26)	(1.72)
PGDP*CER		0.096	0.468
		(0.58)	(1.54)
PGDP*CER2			−0.447
			(−1.46)
TO	−0.293 **	−0.148	−0.213	−0.181	−0.060	−0.048
	(−2.38)	(−1.03)	(−1.41)	(−1.33)	(−0.47)	(−0.38)
HC	0.032	0.020	0.030	0.014	0.004	0.006
	(0.55)	(0.38)	(0.56)	(0.25)	(0.07)	(0.12)
IL	−0.128	0.142	0.155	0.179	0.482	0.448
	(−0.52)	(0.46)	(0.51)	(0.66)	(1.51)	(1.38)
FS	−1.444 ***	−1.185 ***	−1.226 ***	−1.590 ***	−1.305 ***	−1.313 ***
	(−3.83)	(−3.28)	(−3.39)	(−3.89)	(−3.66)	(−3.68)
IER				0.492 ***	1.069 ***	1.016 ***
				(2.92)	(3.58)	(3.29)
PGDP*IER					−0.679 **	−0.796 **
					(−2.31)	(−2.34)
PGDP*IER2						0.130
						(0.69)
Constant	1.034 **	0.754	0.704	0.880 *	0.451	0.465
	(2.11)	(1.57)	(1.46)	(1.81)	(0.95)	(0.98)
Observations	390	390	390	390	390	390
R-squared	0.729	0.732	0.734	0.734	0.741	0.741
Year FE	YES	YES	YES	YES	YES	YES
City FE	YES	YES	YES	YES	YES	YES

Note: Values in parentheses represent t-statistics; *, **, and *** denote significance levels of 10%, 5%, and 1%, respectively.

For CER, the coefficient on the interaction term between PGDP and CER² (PGDP×CER²) is −0.447 with a t-statistic of −1.46, which is not statistically significant at conventional levels. Similarly, the coefficient on the linear interaction term (PGDP×CER) is 0.468 (t = 1.54), also not significant. These results suggest that, within the observed sample, the level of economic development does not exert a statistically detectable moderating effect on the U-shaped relationship between CER and GTFP. In other words, the inflection point of the CER-GTFP curve does not vary systematically with regional economic development in a manner that can be reliably estimated from the current data.

For IER, a more nuanced pattern emerges. The coefficient on the linear interaction term (PGDP×IER) is negative and statistically significant at the 5% level (−0.796, t = −2.34), indicating that economic development negatively moderates the marginal effect of IER on GTFP. However, the coefficient on the quadratic interaction term (PGDP×IER²) is 0.130 with a t-statistic of 0.69, which is not statistically significant. This suggests that while PGDP significantly attenuates the positive marginal effect of IER, it does not significantly alter the curvature (i.e., the inverted U-shape) of the IER-GTFP relationship.

Taken together, these findings imply that the inverted U-shaped relationship between IER and GTFP that identified in the baseline analysis is relatively stable across different levels of economic development. However, the position on that curve at which a given region operates depends on its economic development level. Specifically, for regions at higher levels of economic development, the marginal benefit of IER tends to be lower, and the threshold beyond which IER becomes counterproductive may be reached at a lower level of IER intensity. A possible explanation draws on the concept of regulatory saturation. In regions with higher economic development, environmental regulations, including both command-and-control and market-based instruments that may already be relatively stringent. In such contexts, introducing additional market-based incentives may yield diminishing marginal returns, as the most cost-effective abatement opportunities have already been exploited [47]. Moreover, in highly developed regions, firms may face competing pressures: the need to maintain rapid economic growth can divert resources toward short-term, low-risk projects rather than long-term green innovation, potentially crowding out the innovation capacity that underpins GTFP growth. The negative moderating effect of PGDP on the IER-GTFP relationship is therefore consistent with the presence of such competing dynamics. Overall, the results provide partial support for Hypothesis H4, with the caveat that the moderating effect is primarily linear in nature and does not fundamentally alter the nonlinear structure of the IER-GTFP relationship.

4. Conclusions

This study examines the impact mechanisms of heterogeneous environmental legislation and uses the super-efficiency SBM model to evaluate the green total factor productivity (GTFP) of the livestock industry based on panel data from 30 Chinese provinces (2010–2022). The results show that environmental regulations have nonlinear effects on GTFP. Market-based incentive regulations exhibit an inverted U-shaped relationship with GTFP, whereas command-and-control regulations show a U-shaped relationship, reflecting short-term costs followed by long-term efficiency gains. Environmental regulation exhibits significant regional heterogeneity: command-and-control regulations are more pronounced in the eastern region, while incentive-based regulations are more significant in the western region. Technological innovation serves as a key mediating mechanism through which incentive regulations enhance GTFP. Moreover, the level of economic development negatively moderates the relationship between incentive regulations and GTFP, as higher regulatory intensity in economically advanced regions imposes additional economic burdens that constrain further productivity growth.

This study offers a comparative exploration into how different types of environmental regulations affect the green total factor productivity in livestock farming. Still, it is important to acknowledge its limitations, which in turn open up some new directions for future work. First, the relevance of environmental regulation indicators can be improved. In this study, incentive-based environmental regulation is defined using a composite of pollutant discharge fee revenue as a share of industrial value added and completed investment in industrial pollution control as a share of value added. This composite indicator may deviate from actual conditions. As data availability improves, future research should refine the environmental regulation indicator system to achieve more accurate measurements of regulatory stringency. Second, the GTFP data we used here is at the provincial level. While this approach yields findings with broad applicability, future research would surely benefit from working with more granular data. Moving down to the city or county level, for instance, could give us a sharper picture of the regional variations in GTFP. Finally, this study examines the direct impact of environmental regulation on the green total factor productivity of the livestock industry. However, both livestock pollution and environmental regulation exhibit spatial autocorrelation. Future research could incorporate the spatial spillover effects of environmental regulation into the analytical framework to provide a more comprehensive understanding of how environmental regulation influences the green development of the livestock industry.

Based on the empirical findings, this study proposes the following policy recommendations to enhance green total factor productivity in China’s livestock industry:

(1) Establish a dynamic monitoring mechanism for market-based regulations, shifting from increasing stringency to optimizing policy structure as intensity approaches the inflection point, while leveraging the coercive function of command-and-control regulations to set industrial access and emission baselines in highly polluting or underdeveloped regions.

(2) Strengthen fundamental and applied research on green technologies, establish platforms for technology transfer and pilot programs, and accelerate the commercialization and diffusion of technological innovations through government-subsidized services such as equipment rental and technical training.

(3) Formulate spatially differentiated environmental policies, enhance regional coordination, reduce enforcement costs through digital monitoring, and link targeted subsidies with green certification mechanisms to establish a policy framework that balances incentives with binding constraints.

(4) Establish a regulatory framework aligned with local economic development levels, create a risk compensation fund to mitigate the impact of regulatory cost shocks, conduct periodic policy evaluations, incorporate regional adjustment coefficients, and dynamically adjust the type and intensity of regulatory instruments.