Impact of Environmental Regulation on the Green Total Factor Productivity of Dairy Farming: Evidence from China

Abstract

Environmental regulation is essential to promote green and sustainable development in dairy farming. Nevertheless, limited studies have focused on the impact of environmental regulation on the green total factor productivity (GTFP) of dairy farming. This study measures the GTFP of dairy farming in 27 provinces in China during 2009–2020 using the Slack Based Measure (SBM) model and the Malmquist–Luenberger (ML) productivity index. In addition, random effects and threshold regression models are used to measure the impact of environmental regulations on the GTFP of dairy farming. The results demonstrate the fluctuating growth of the GTFP of dairy farming and that technical efficiency is the primary driver of the GTFP growth. The annual growth rate of GTFP is the highest in large-scale dairy farming (3.27%), followed by medium-scale dairy farming (2.73%) and small-scale dairy farming (1.98%). Furthermore, environmental regulation positively affects the GTFP and has a threshold effect on the GTFP, with the urban–rural income gap as the threshold variable in medium-scale dairy farming and small-scale dairy farming. The impact on the GTFP can be significantly negative if the urban–rural income gap crosses the threshold value. Overall, this study provides some policy recommendations for attaining green and sustainable dairy farming development in China.

1. Introduction

Since the United Nations Development Programme proposed the concept of “Green Development” in 2002, the realization of green development and sustainable development has become a shared aspiration around the globe [1]. Green development is considered the primary way to attain economic development and environmental protection [2]. As the world’s largest developing country, severe conflict exists between economic growth and environmental protection in China [3,4]. Thus, the Chinese government has elevated “Green Development” as a national strategy at the Fifth Plenary Session of the 18th Party Central Committee and integrated it into its long-term development plan. The 19th National Congress report emphasized promoting green development and consolidating the prevention of agricultural surface source pollution [5].

China is a mainly dairy-consumption and dairy-farming country. According to data from China Dairy Industry Statistics, the consumption of milk per capita in China increased from 4.28 to 13 kg/year from 2000 to 2020, about three times as much as in 2000. Driven by the increment in dairy consumption demand, dairy cattle breeding maintained a high growth trend in China. The dairy cattle stock rapidly increased from 4.89 to 10.43 million, and the growth rate was as high as 113.29%. However, the development of dairy farming resulted in severe environmental problems in China [6]. The National Pollution Source Census Bulletin reported livestock farming as the primary source of agricultural pollution in China. As large ruminants are involved in livestock farming, dairy cattle add significantly higher manure and greenhouse gas emissions than other livestock and poultry animals, causing severe environmental pollution [7]. The growing environmental pollution is in stark contrast with China’s growing market for dairy farming. The “Opinions on Promoting the Revitalization of the Dairy Industry to Safeguard Dairy Quality and Safety” promulgated by the State Council in 2018 established the principle of green development of the dairy industry while recommending the synergistic development of dairy production and ecology. Enhancing the green total factor productivity (GTFP) of dairy farming has become the primary goal for the green development of China’s dairy industry. Achieving the goal depends on the necessary environmental regulations, which can reduce environmental pollution and improve the efficiency of resources’ comprehensive utilization. However, the impact of environmental regulations on the green total factor productivity of dairy farming is still unclear.

Researchers have increasingly focused on the incongruity between dairy farming and the ecological environment [8,9,10,11]. Reconciling the relationship between the dairy farming industry and environmental pollution has become an enormous challenge for the dairy industry [12]. Guided by the concept of green and high-quality development, the Chinese government implemented a series of environmental regulations to mitigate the conflict between dairy farming development and environmental pollution [13]. As one of the most effective means to address environmental pollution problems, environmental regulation has been extensively used in agriculture pollution control [14,15]. Environmental regulation denotes the policies and coercive instruments developed by the government to decrease pollutant emissions [16], which are specifically categorized into two types of instruments: formal environmental regulation and informal environmental regulation [17]. In addition, some studies claimed that implementing appropriate environmental regulations could promote green technology adoption and technological innovation, resulting in green production and sustainable development in recent years [18,19,20]. Lu et al. [21] found that environmental regulation could improve the adoption of green technology by farmers. Mbanyele et al. [22] applied the difference in the difference model to test the impact of environmental regulation on technological innovation and found that environmental regulation substantially promoted innovation productivity. The measurement of environmental regulation normally uses indicators such as investment in pollution control or the number of environmental policies to assess the intensity of environmental regulation [23,24]. Furthermore, some studies demonstrated a nonlinear relationship between environmental regulation and environmental pollution [25,26]. When the intensity of environmental regulation exceeds a certain threshold, the impact of environmental regulation on environmental pollution reduction is not satisfactory. Environmental regulations are not only an effective measure to address environmental pollution problems and promote green development but also an essential factor affecting the green total factor productivity. Previously, environmental regulations have been used widely in multiple fields with remarkable results [27,28,29].

Extensive research has been conducted on the total factor productivity and the impact of inputs on output [30,31,32,33,34]. The total factor productivity of dairy farming has been measured recently [35,36]. For example, McCormack et al. [37] found, relative to 2010, the total factor productivity of Irish dairy farms had increased by almost 18%. Balezentis et al. [38] found the total factor productivity of Lithuanian dairy farms grew at an average annual rate of 2%, with technical change and scale components being the main driver of the growth. Nevertheless, the traditional total factor productivity overlooks the environmental pollution due to production in the calculation process. Conversely, the green total factor productivity is measured based on environmental pollution, aligning more with green and high-quality development [39,40]. Green total factor productivity is defined as the total factor input–output efficiency, which takes pollution emissions into account [41]. Improving the green total factor productivity is the way to attain sustainable economic development [42,43]. Wang et al. [44] measured China’s agricultural GTFP between 2004 and 2016, reporting that the average annual growth rate of the GTFP of Chinese agriculture reached 3.1% and varied regionally. Ding et al. [45] demonstrated that China’s industrial GTFP exhibited an inverted “U”-shaped trend. Moreover, studies primarily focused on the issue of the total factor productivity of dairy farming and revealed that the total factor productivity of dairy farming closely correlated with factors such as farming inputs, farming technology, and farmers’ education [46,47]. The total factor productivity of dairy farming has been improved under technological progress [48]. However, the current literature on the GTFP of dairy farming is scarce and does not create a systematic theoretical framework.

To date, several studies have demonstrated that environmental regulation is an essential factor affecting the green total factor productivity, but the direction of the impact of environmental regulation on the green total factor productivity is not uniform [49,50]. One view is that implementing environmental regulations could promote green total factor productivity [51]; this is because apt environmental regulations can stimulate technological innovation and further expand the production capacity while decreasing the input costs of combating environmental pollution, thereby increasing the green total factor productivity. In addition, when producers are limited by the constraints of environmental regulations, they will adopt cleaner production technologies in order to reduce pollution emissions, which will also optimize the allocation of production factors and promote the green total factor productivity [52]. This view is supported by some recent studies. For example, Xu et al. [49] measured provincial GTFP in China with the Slack Based Measure-Malmquist–Luenberger model and found that environmental regulation influenced GTFP positively. Fan et al. [53] found that environmental regulation indirectly promoted GTFP by enhancing green technological innovation levels. Another view is that environmental regulations would decrease the green total factor productivity, as implementing strict environmental regulations would lead to extra costs for pollution control. Thus, the investment in production is reduced, and the cost of complying with environmental regulations is higher than the benefit from technological innovation, thereby arguing that environmental regulations inhibit the increase in the green total factor productivity. Several of the latest studies confirm this view. For example, Li et al. [54] found that strict environmental regulation would inhibit the growth of GTFP. Li et al. [28] found that the impact of environmental regulation on GTFP was negative when the economic development level reached a medium level. The existing literature has primarily focused on the impact of environmental regulations on environmental pollution and green total factor productivity (

Table 1. A brief summary of the recent literature.

Reference	Period	Study Object	Explanatory Variables	Outcome Variables	Methods	Main Conclusions
Liu, Y., et al. (2005) [25]	2004–2016	China	ER	EP	Spatial autoregressive model	An inverted U-shaped curve relationship exists between ER and EP
Chen, X., et al. (2021) [26]	2006–2017	China	ER	EP	PVAR model	ER has a nonlinear effect on EP
Xu, X.F., et al. (2021) [49]	2006–2017	China	ER	GTFP	SBM-ML model	ER influences GTFP positively
Junxia, H., et al. (2021) [50]	2009–2018	Yangtze River Economic Belt of China	ER	GTFP	SBM-ML model	ER has no significant impact on GTFP
Huang, Q.H., et al. (2021) [55]	2006–2018	110 Chinese cities	ER	GTFP	SBM-ML model	ER has a nonlinear regulatory effect on GTFP
Fan, M., et al. (2022) [53]	2004–2018	269 Chinese cities	ER	GTFP	Epsilon-based measure model	ER indirectly promotes GTFP by enhancing the green technological innovation level
Li, J., et al. (2021) [54]	2003–2019	41 Chinese cities	ER	GTFP	System generalized method of moments model and GML index	Too strict ER will inhibit the growth of GTFP
Li, B., et al. (2017) [56]	2003–2013	273 Chinese cities	ER	GTFP	Spatial Durbin model and ML index	Civil ER has positive effects in promoting GTFP
Gong, M.Q., et al. (2020) [57]	2005–2017	China’s 27 manufacturing industries	ER	GTFP	SBM-ML model	The impact of ER on the GTFP is first positive and then negative
Li, H., et al. (2020) [58]	2005–2015	China’s 35 industries	ER	GTFP	Threshold model and SBM-ML model	ER can promote GTFP by increasing market concentration
He, Q., et al. (2021) [59]	2005–2018	China’s 30 provinces	ER	GTFP	SBM-ML model	There is a nonlinear relationship between ER and GTFP
Zou, H., et al. (2022) [60]	2005–2017	China’s pollution-intensive industries	ER	GTFP	SBM-ML model	The effect of ER on GTFP is an inverted U-shaped curve

Note: The abbreviation “ER” means environmental regulation, “EP” means environmental pollution, “GTFP” means green total factor productivity, PVAR model refers to panel vector autoregressive model, ML index is Malmquist–Luenberger index, GML index is Global Malmquist–Luenberger index, SBM-ML model refers to Slack Based Measure-Malmquist–Luenberger model.

); it has not yet reached a consistent conclusion. The studies mentioned above provide an essential contribution to elucidating the correlation between environmental regulations and green total factor productivity.

What is the green total factor productivity level of dairy farming in China? Does environmental regulation affect the green total factor productivity of Chinese dairy farming? These questions are not explored in the current literature. Based on the data on Chinese dairy farming between 2009 and 2020, this study uses the Slack Based Measure-Malmquist–Luenberger (SBM-ML) model to measure the green total factor productivity of small-, medium-, and large-scale dairy farming. Furthermore, this study empirically examines the impact of environmental regulations on the green total factor productivity of dairy farming. Finally, using the urban–rural income gap as the threshold variable, this study investigates the nonlinear relationship between environmental regulations and the green total factor productivity of dairy farming.

Compared with previous studies, the possible innovations of this paper are as follows: Firstly, previous studies on dairy farming have focused on the total factor productivity and overlooked the impact of environmental pollution, and fewer studies have addressed the green total factor productivity of dairy farming. Meanwhile, the existing studies have not disintegrated the green total factor productivity of dairy farming and have not yet investigated the role of technological progress and technical efficiency in the green total factor productivity of dairy farming. To fill this research gap, we apply the SBM-ML model to measure the green total factor productivity of dairy farming. In addition, the effect of technological progress and technical efficiency on the green total factor productivity of dairy farming is further analyzed. Secondly, little research has been conducted on the nonlinear relationship between environmental regulations and the green total factor productivity of dairy farming, and the impact of environmental regulations on the green total factor productivity of dairy farming remains unclear. Different from the previous studies, the threshold effect model is applied to explore the nonlinear effect of environmental regulations on the green total factor productivity of dairy farming.

The subsequent section of the paper is organized as follows: Section 2 presents the model specification, data sources, and descriptive statistics of the variables used in the paper. Section 3 shows the results and further discussion. According to the analysis above, the conclusions and policy recommendations are drawn in Section 4.

2. Materials and Methods

2.1. SBM-ML Model

Previous studies mainly used stochastic frontier analysis (SFA) and data envelopment analysis (DEA) to measure the green total factor productivity [42,43]. However, the SFA method might have issues in parameter-setting, and the DEA method is highly prone to errors owing to different radial and angular choices. To solve this problem, a nonradial and nonangular Slack Based Measure (SBM) model, which introduced slack variables into the function for efficiency measurement, is developed by Tone et al. [61]. For measuring the GTFP of dairy farming, pollutants need to be taken into account as an undesirable output; SBM model is therefore used in this study. We constructed the SBM model in the manner described by Zhou et al. [62]. Assuming that dairy farming requires M inputs while producing S₁ consensual outputs and S₂ nonconsensual outputs, the SBM model can be presented as follows:

$$p*=\min \frac{1-\frac{1}{M}{\displaystyle {\sum }_{i=1}^M\frac{S_m^x}{x_{m_0}}}}{1+\frac{1}{S_1+S_2}\left({\displaystyle {\sum }_{r=1}^{S_1}\frac{S_r^g}{y_{r_0}^g}+{\displaystyle {\sum }_{k=1}^{S_2}\frac{S_k^b}{b_{k_0}^b}}}\right)}$$

(1)

$$s.t.\{\begin{array}{l}{\displaystyle \sum _{j=1}^J{\lambda }_j^t{Y}_{rj}^t-S_r^g=y_{rj}^t,r=1,\mathrm{\dots },S_1;}\\ {\displaystyle \sum _{j=1}^J{\lambda }_j^t{b}_{kj}^t+S_k^b=b_{bj}^t,K=1,\dots ,S_2;}\\ {\displaystyle \sum _{j=1}^J{\lambda }_j^t{X}_{mj}^t+S_m^x=x_{mj}^t,m=1,\dots ,M;}\\ {\displaystyle \sum _{j=1}^J{\lambda }_j^t=1,{\lambda }_j^t\ge 0,}{S}_k^b\ge 0,S_r^g\ge 0,S_m^x\ge 0,j=1,\dots ,J;\end{array}$$

(2)

where p* denotes the efficiency evaluation indicator; S_m^x means input excess; S_r^g denotes the desired output deficiency; S_k^b signifies undesirable output redundancy; λ_j^t means the weight vector; and p* represents ∈ (0, 1). The SBM model applies to the green total factor productivity measurement in a fixed time horizon and cannot reflect the dynamic evolution of the GTFP. To better reflect the dynamic evolution characteristics of the GTFP of dairy farming, the SBM-ML model, which combined the nonradial and nonangular SBM model with the directional distance function containing undesirable outputs, is applied for measurement [63,64]. We drew on the idea of Chung et al. [65], combining the directional distance function with the Malmquist index to construct the Malmquist–Luenberger index (ML). The equation can be presented as follows:

$$\begin{array}{l}M{L}_t^{t+1}={\left\{\frac{1+D_0^t(x_i^{i,t},y_i^{i,t},b_i^{i,t};y_i^{i,t},-b_i^{i,t})}{1+D_0^t(x_i^{i,t+1},y_i^{i,t+1},b_i^{i,t+1};y_i^{i,t+1},-b_i^{i,t+1})}\times \frac{1+D_0^{t+1}(x_i^{i,t},y_i^{i,t},b_i^{i,t};y_i^{i,t},-b_i^{i,t})}{1+D_0^{t+1}(x_i^{i,t+1},y_i^{i,t+1},b_i^{i,t+1};y_i^{i,t+1},-b_i^{i,t+1})}\right\}}^{\frac{1}{2}}\\ =MLT{C}_t^{t+1}\times MLT{E}_t^{t+1}\end{array}$$

(3)

where the ML index denotes the growth rate of the GTFP of dairy farming, which is categorized into the technical progress index (MLTC) and technical efficiency index (MLTE). If ML > 1, it indicates an improvement in the GFTP of dairy farming. As the ML index denotes the GTFP growth rate, we assumed that the GTFP of dairy farming in 2008 is one, then the GTFP of dairy farming in 2009 is one multiplied by the ML index in 2009, and so on.

2.2. Empirical Model

The green total factor productivity is largely affected by environmental regulation. To assess the impact of environmental regulation on the GTFP of dairy farming, we used the GTFP measured in the SBM-ML model as the explained variable and environmental regulation as the core explanatory variable. Drawing on Zhang et al. [66], we applied the level of economic development, farming technology, level of education, market demand, urban–rural income gap, and production structure as control variables. The model is constructed as follows:

$$\begin{array}{l}\ln GTFP=\alpha +{\beta }_1\ln E{R}_{it}+{\beta }_2\ln EC{O}_{it}+{\beta }_3\ln F{T}_{it}+{\beta }_4\ln ED{U}_{it}+{\beta }_5\ln M{D}_{it}+{\beta }_6\ln GA{P}_{it}\\ +{\beta }_7\ln P{S}_{it}+{\mu }_i+{\nu }_t+{\epsilon }_{it}\end{array}$$

(4)

where GTFP denotes the explained variable for the GTFP of dairy farming and ER denotes the core explanatory variable for environmental regulation. The control variables ECO, FT, EDU, MD, GAP, and PS signify the level of economic development, farming technology, level of education, market demand, urban–rural income gap, and production structure, respectively; i denotes the province; t denotes the year; α and β represent parameters to be estimated; μ and ν represent individual fixed effects and time fixed effects, respectively; and ε represents the random disturbance term.

Furthermore, some studies demonstrated that environmental regulation and green total factor productivity might be nonlinear, with threshold effects on other variables [67]. Thus, we used a threshold effect model for estimation. The primary view is that environmental regulation could stimulate the development of technological innovation [68] and attain the GTFP [69]. Nevertheless, the effects of technological innovation showed significant differences under the impact of different income disparities [70], resulting in a possible inconsistent effect of environmental regulation on the GTFP. Chinese dairy farms are mostly located in rural areas, far from urban areas. When the income gap between urban and rural areas is large, the original dairy farming technical labor and capital investment in rural areas would rapidly shift to urban areas, further decreasing the technological innovation capacity of dairy farming, and dairy farmers will disregard the environmental pollution problems stemming from the farming process [71]. Even if environmental regulations are implemented, it will be challenging to promote innovation and development of green farming technologies, hindering the improvement of the GTFP of dairy farming. Considering this, we took the urban–rural income gap as the threshold-dependent variable and drew on Hansen [72] to construct the following threshold effect model:

$$\begin{array}{l}\ln GTFP=\alpha +{\alpha }_1\ln E{R}_{it}\times I(\ln GA{P}_{it}\le {\gamma }_1)+{\alpha }_2\ln E{R}_{it}\times I({\gamma }_1<\ln GA{P}_{it}\le {\gamma }_2)+\\ {\alpha }_3\ln E{R}_{it}\times I(\ln GA{P}_{it}>{\gamma }_2)+{\displaystyle \sum \beta }{X}_{it}+{\mu }_i+{\nu }_t+{\epsilon }_{it}\end{array}$$

(5)

where GAP denotes the threshold variable, representing the urban–rural income gap, measured by the ratio of urban per-capita disposable income to rural per-capita net income; γ₁ and γ₂ denote the two threshold values; I denotes the indicative function, taking the value of 0 or 1; X denotes the control variable; and other variables have the same meaning as Equation (4).

2.3. Data and Variables

2.3.1. Data Sources

Based on the data accessibility, we selected the panel data of 27 provinces in China from 2009 to 2020 as the sample. All data were obtained from the National Compilation of Information on Costs and Benefits of Agricultural Products, the Handbook of Source Production and Discharge Coefficients of Animal and Poultry Farming, the China Statistical Yearbook, the China Rural Statistical Yearbook, and the China Environmental Yearbook. To ensure data continuity, we used the linear interpolation method and the mean value method to fill in the missing values. Thus, small-, medium-, and large- dairy farming data were transformed into balanced panel data.

2.3.2. Variables

(1)

Input and Output Variables

The SBM-ML model requires the setting of input variables, desired output variables, and undesirable output variables for dairy farming, which are set out below:

Input variables: The total input of dairy farming, which can be categorized into labor input and capital input, while the former is the labor cost of dairy farming, the latter includes direct costs, such as feed, water, and fuel power costs in dairy farming, as well as indirect costs such as fixed asset inputs and insurance costs. The costs are deflated by the price index to eliminate the effect of price factors. Desired output variable: Production of milk. Undesirable output variable: The sum of pollutants from dairy farming.

(2)

Explanatory and Control Variables

To estimate the impact of environmental regulation on the green total factor productivity of dairy farming, the explanatory variables, explanatory variables, and control variables should be set as follows:

Explained variable: The green total factor productivity of dairy farming. Based on the SBM-ML model, the green total factor productivity of dairy farming in China was derived. Explanatory variable: Environmental regulation, expressed as the share of investment in environmental pollution control in the region’s GDP. Control variables: The level of economic development, farming technology, level of education, market demand, urban–rural income gap, and production structure were selected as control variables.

Table 2. Descriptive statistics of variables.

Variable	Variable Description	Mean	Minimum	Maximum
Environmental Regulation (ER)	The amount of investment in environmental pollution control as a percentage of the Gross Domestic Product (unit: percentage)	1.5103	0.3000	4.5700
Level of economic development (ECO)	The Gross Domestic Product of each province (unit: one billion yuan)	2354.6422	108.1270	11,076.0940
Farming Technology (FT)	The value of output generated per cow (unit: yuan)	25,218.4126	10,480.4100	61,880.9300
Level of education (EDU)	Average education years	7.8366	6.1188	9.8380
Market Demand (MD)	Population density (unit: numbers of people per square kilometer)	469.6302	4.7066	3949.5556
Urban–rural income gap (GAP)	The ratio of urban per capita disposable income to rural per capita net income	2.6990	1.8451	4.2809
Production Structure (PS)	The proportion of the value of dairy products to Gross Domestic Product	0.0038	0.0001	0.0227

presents the descriptive statistics of the variables.

3. Results and Discussion

3.1. Green Total Factor Productivity of Dairy Farming in China

Table 3. The green total factor productivity index for dairy farming in China during 2009–2020.

Year	Small Scale			Medium Scale			Large Scale
	GTFP	MLTC	MLTE	GTFP	MLTC	MLTE	GTFP	MLTC	MLTE
2008–2009	0.9341	0.8530	1.0967	0.9413	0.9730	0.9728	0.9690	0.9316	1.0410
2009–2010	0.9518	0.9675	0.9836	0.9357	0.7733	1.2142	0.9691	0.9162	1.0601
2010–2011	0.9712	0.9401	1.0328	0.9799	1.0463	0.9350	1.0086	0.9956	1.0206
2011–2012	1.0637	1.0654	0.9985	1.0730	1.0208	1.0515	1.0544	1.0603	0.9950
2012–2013	1.0434	0.9960	1.0478	1.0933	1.1298	0.9670	1.0466	1.1213	0.9338
2013–2014	0.9778	1.0035	0.9731	0.9687	0.9703	1.0000	0.9960	1.0060	0.9896
2014–2015	1.0162	1.0124	1.0038	1.0377	1.1244	0.9238	1.0091	0.9503	1.0626
2015–2016	1.0198	1.0172	1.0028	1.0427	1.0111	1.0312	1.0154	0.9789	1.0373
2016–2017	0.9995	0.9751	1.0254	0.9737	0.9088	1.0734	0.9969	0.9788	1.0189
2017–2018	1.0319	1.084	0.9525	1.0204	1.0274	0.9925	1.0564	1.0615	0.9966
2018–2019	1.0712	1.0268	1.0431	1.1276	1.1049	1.0209	1.0886	1.1179	0.9747
2019–2020	1.1567	1.1314	1.0230	1.1342	1.1068	1.0222	1.1823	1.2423	0.9514
Average value	1.0198	1.0060	1.0153	1.0273	1.0164	1.0170	1.0327	1.0300	1.0068

shows the green total factor productivity index of dairy farming in China. The average annual growth rates of the GTFP of small-, medium-, and large-scale dairy farming between 2009 and 2020 were 1.98%, 2.73%, and 3.27%, respectively. The GTFP of dairy farming demonstrated a positive correlation with the degree of the farming scale. The technical progress index (MLTC) and technical efficiency index (MLTE) of small-, medium-, and large-scale dairy farming were both >1, suggesting that the GTFP of small-, medium-, and large-scale dairy farming was driven by both technological progress and technical efficiency improvement.

Generally, the GTFP of medium- and small-scale dairy farming showed an “M” type trend. Before 2008, a large number of farmers were involved in the dairy farming industry. Dairy farmers mainly included free-range households, accounting for >70%. The scale and environmental standardization level of dairy farming were low. After the melamine dairy contamination incident in 2008, China’s dairy farming industry was severely affected, with dairy farmers’ enthusiasm for production severely dampening and the number of free-range dairy farmers plummeting. Consequently, the GTFP index for medium- and small-scale dairy farming only reached 0.9413 and 0.9341 in 2009, respectively, which decreased by 5.87% and 6.59% compared with the previous year. The Chinese government started focusing on augmenting quality control of the dairy industry while introducing policies and regulations such as the “Regulations on Supervision and Management of Dairy Quality and Safety” and the “Outline of the Plan for Rectification and Revitalization of the Dairy Industry”. After a long adjustment period, the GTFP of medium- and small-scale dairy farming started exhibiting a positive growth trend in 2011. Along with the withdrawal of a large number of free-range dairy farmers, a severe shortage of raw milk supply occurred in several Chinese provinces in 2013, forcing dairy farmers to increase their production capacity rapidly and disrupting the process of green production in dairy farming. Therefore, there was a decline in the GTFP of medium- and small-scale dairy farming between 2013 and 2014. After optimal adjustment of the farming structure, the GTFP of medium- and small-scale dairy farming increased from 2014 to 2016. With the import of dairy products in China surging in 2016, the dairy farming industry was hit by the dual impact of import squeeze and weak domestic demand, leading to dairy farming losses of >50%. The GTFP of medium- and small-scale dairy farming declined marginally in 2016 and maintained a stable growth trend after 2016.

The large-scale dairy farming GTFP index was similar to small- and medium-scale dairy farming, although it was less affected by unexpected events and generally showed a stable growth trend because large-scale dairy farms are more mature in farming technology and epidemic prevention conditions and are relatively more resilient to dairy farming risks. In 2018, the Chinese government proposed a dairy revitalization policy to support the development of dairy farming, under which the growth rate of the GTFP of small-, medium-, and large-scale dairy farming reached 15.67%, 13.42%, and 18.23% from 2019 to 2020, respectively, with the highest growth level in the past decade.

Table 4. Decomposition of the green total factor productivity of dairy farming of 27 provinces in China.

Province	Small Scale			Province	Medium Scale			Province	Large Scale
	GTFP	MLTC	MLTE		GTFP	MLTC	MLTE		GTFP	MLTC	MLTE
Hebei	0.9762	0.9762	1.0000	Sichuan	0.9787	1.0025	0.9748	Yunnan	0.9852	1.0286	0.9590
Jilin	0.9979	1.0134	0.9871	Yunnan	0.9859	1.0156	0.9723	Qinghai	1.0118	1.0242	0.9943
Shandong	1.0123	1.0028	1.0097	Inner Mongolia	1.0063	1.0165	0.9924	Liaoning	1.0129	1.0287	0.9886
Inner Mongolia	1.0168	0.9981	1.0208	Fujian	1.0117	1.0141	1.0043	Sichuan	1.0145	1.0289	0.9865
Yunnan	1.0228	1.0090	1.0097	Jilin	1.0132	1.0172	1.0057	Gansu	1.0188	1.0269	0.9939
Fujian	1.0236	1.0172	1.0110	Zhejiang	1.0228	1.0190	1.0084	Anhui	1.0203	1.0213	1.0041
Hunan	1.0237	1.0033	1.0208	Tianjin	1.0238	1.0041	1.0358	Zhejiang	1.0257	1.0327	0.9964
Heilongjiang	1.0268	0.9971	1.0348	Shanxi	1.0252	1.0162	1.0172	Beijing	1.0264	1.0111	1.0237
Shanxi	1.0271	1.0155	1.0133	Jiangsu	1.0278	0.9948	1.0457	Fujian	1.0293	1.0163	1.0191
Guangxi	1.0288	1.0048	1.0246	Liaoning	1.0278	1.0192	1.0123	Guangdong	1.0306	1.0385	0.9966
Ningxia	1.0314	1.0081	1.0246	Guangxi	1.0280	1.0168	1.0243	Tianjin	1.0338	1.0219	1.0166
Liaoning	1.0322	1.0188	1.0157	Hunan	1.0283	1.0164	1.0152	Xinjiang	1.0367	1.0357	1.0009
Henan	1.0374	1.0143	1.0267	Guizhou	1.0288	1.0193	1.0183	Heilongjiang	1.0392	1.0308	1.0110
				Beijing	1.0301	1.0050	1.0118	Shanghai	1.0406	1.0294	1.0130
				Ningxia	1.0332	1.0178	1.0208	Henan	1.0436	1.0333	1.0148
				Chongqing	1.0332	1.0157	1.0157	Guizhou	1.0466	1.0297	1.0216
				Gansu	1.0333	1.0139	1.0334	Shanxi	1.0485	1.0389	1.0103
				Shaanxi	1.0378	1.0169	1.0211	Jiangsu	1.0579	1.0338	1.0320
				Henan	1.0384	1.0165	1.0253	Shandong	1.0651	1.0597	1.0127
				Heilongjiang	1.0390	1.0161	1.0261	Inner Mongolia	1.0663	1.0308	1.0404
				Shanghai	1.0480	1.0161	1.0451
				Anhui	1.0519	1.0175	1.0531
				Xinjiang	1.0757	1.0708	1.0128

shows the GTFP of dairy farming in 27 provinces in China during 2009–2020. Among the small-scale dairy farming, Henan province had the fastest average annual growth rate of the GTFP of dairy farming, reaching 3.74%, with technical efficiency improvement contributing 2.67% and technological progress contributing 1.43%. The province with the lowest growth rate of the GTFP of dairy farming was Hebei province. The MLTC index in Hebei province was <1, suggesting technological regression, thereby inhibiting the improvement of the GTFP of small-scale dairy farming. Compared with small-scale dairy farming, the GTFP of most medium- and large-scale dairy farms was higher; this is because medium- and large-scale dairy farms are equipped with more advanced green farming techniques and pollution-prevention technologies, and are therefore more likely to attain technological progress and increase the GTFP.

Since 2009, China’s Ministry of Agriculture has identified 13 regions, such as Heilongjiang province, Henan province, and Shanghai City, as advantageous areas for dairy farming. Table 3 shows that the GTFP indexes in these regions are relatively high overall. Of note, the GTFP indexes are also high in some provinces in nonadvantageous areas, mainly because the GTFP of dairy farming is not only affected by national dairy policies and regional distribution but also by the level of economic development and environmental regulation policies in each region, which might also contribute to the dairy GTFP.

3.2. Environmental Regulation on Green Total Factor Productivity of Dairy Farming in China

3.2.1. Unit Root Test

The panel data might have a pseudoregression problem; thus, we conducted unit root tests on each variable before constructing the model. The panel unit root tests can be divided into the tests for the same unit root and different unit roots. We used the LLC and IPS tests for unit root tests based on the data characteristics.

Table 5. Unit root test results.

Variable	Small Scale		Medium Scale		Large Scale
	LLC Test	IPS Testing	LLC Test	IPS Testing	LLC Test	IPS Testing
lnGTFP	−4.6411 ***	−2.3057 **	−7.2319 ***	−3.8423 ***	1.6493 *	−0.7086 *
lnER	−4.9516 ***	−2.1903 **	−7.6816 ***	−2.7856 ***	−7.4204 ***	−2.4366 ***
lnECO	3.0431 ***	5.1655 ***	2.4129 **	5.7309 ***	1.9303 *	4.8050 *
lnFT	0.5485 *	1.2684 *	−0.5081 **	0.5789 *	−2.1857 **	−0.9583 *
lnEDU	−1.3535 *	−0.4240 **	−3.6483 ***	−1.6394 *	−4.0583 ***	−1.8966 **
lnMD	7.1275 *	6.2738 ***	4.0843 *	3.3320 **	−0.4741 *	0.7127 *
lnGAP	−2.0363 **	0.7120 **	−6.6500 ***	−0.6591 **	−6.1326 ***	−2.5925 ***
lnPS	−0.2778 **	0.6072 *	−2.3829 **	−0.3882 ***	−0.6119 **	1.5103 *

Note: *, **, and *** mean significance at the levels of 10%, 5%, and 1%, respectively.

shows the test results. All the variables passed the LLC and IPS tests with at least a 10% significance level, suggesting the stability of the data selected in this study.

3.2.2. Panel Regression Results and Analysis

The panel data model was used to estimate the impact of environmental regulation on the green total factor productivity of dairy farming. The fitting models of the panel data were mainly fixed- and random-effects models and the Hausman test was used to determine which model was more appropriate.

Table 6. Panel regression results.

Variable	Small Scale		Medium Scale		Large Scale
	FE	RE	FE	RE	FE	RE
lnER	0.0417	0.0350 *	0.0929 ***	0.0664 **	0.0876 ***	0.0842 ***
	(0.0382)	(0.0365)	(0.0368)	(0.0332)	(0.0325)	(0.0292)
lnECO	0.1845 **	0.1785 ***	0.2283 ***	0.1382 **	0.0312	0.0016
	(0.0851)	(0.0638)	(0.0937)	(0.0570)	(0.0842)	(0.0520)
lnFT	0.0697 *	0.0462 *	0.0512 *	0.1053 ***	0.1964 ***	0.1970 ***
	(0.0711)	(0.0644)	(0.0675)	(0.0622)	(0.0564)	(0.0512)
lnEDU	0.4772 *	0.5615	1.3783 ***	1.2440 ***	0.4986 *	0.2529 *
	(0.4782)	(0.4484)	(0.4084)	(0.3471)	(0.3699)	(0.3184)
lnMD	0.1467	0.0931	0.4959 **	0.0569 *	0.1329 *	0.0072 *
	(0.3543)	(0.0781)	(0.2580)	(0.0371)	(0.2307)	(0.0337)
lnGAP	−0.6834 ***	−0.5736 ***	−0.2772 **	−0.2994 **	−0.4931 ***	−0.5226 ***
	(0.1399)	(0.1224)	(0.1429)	(0.1247)	(0.1416)	(0.1246)
lnPS	0.0645 *	0.0201	0.0276	0.0088	−0.0278	0.0116 *
	(0.0389)	(0.0295)	(0.0384)	(0.0251)	(0.0376)	(0.0262)
Cons.	3.1826	1.0863	−2.9100 *	−0.3890	2.2336	2.8357 ***
	(2.3500)	(1.3189)	(1.6936)	(1.0903)	(1.5950)	(0.9899)
Hausman test	chi2 = 5.28		chi2 = 7.69		chi2 = 4.30
	p = 0.7273		p = 0.4645		p = 0.8289

Note: *, **, and *** mean significance at the levels of 10%, 5%, and 1%, respectively.

shows the empirical results. The Hausman test showed that the model cannot reject the null hypothesis, suggesting that the random-effect model was better than the fixed-effect model. Thus, we used the random-effect model as the benchmark regression model.

Regarding the core explanatory variables, the regression coefficients of environmental regulations in small-, medium-, and large-scale dairy farming were 0.0350, 0.0664, and 0.0842, respectively, which was significant at the 10% level at least, suggesting that environmental regulations exerted a positive impact on the green total factor productivity of dairy farming. The GTFP of large-scale dairy farming was mostly affected by environmental regulations, while the impact on medium- and small-scale dairy farming was relatively small; this could be explained by the fact that large-scale dairy farmers have a larger dairy farming scale than small- and medium-scale dairy farmers. Thus, the response of dairy farmers to environmental regulations was also more pronounced. In addition, large-scale dairy farmers have more advanced farming techniques and farming experiences. Under environmental regulation, large-scale dairy farmers prefer adjusting the farming structure while adopting green and efficient farming techniques to enhance the dairy farming production efficiency. Furthermore, the pollution emissions from dairy farming would be reduced, while the efficiency of dairy farming would be improved, resulting in a higher GTFP of dairy farming.

Regarding the control variables, the regression coefficients of economic development levels in small- and medium-scale dairy farming were 0.1785 and 0.1382, respectively. The improvement of economic development level stimulated dairy farming technology development and innovation, contributing to the efficient and green development of dairy farming. The regression coefficients of farming technology in all three scales were positive. The improvement of farming technology helped farmers optimize the allocation of breeding resources, save breeding input costs, and decrease pollution emissions. In addition, the coefficient of the level of education and market demand was significantly positive in medium- and large-scale dairy farming, suggesting that the enhancement of education and increased market demand could improve the GTFP. Nevertheless, the level of education and market demand coefficients were not significant in small-scale dairy farming; perhaps, compared with medium- and large-scale dairy farmers, small-scale dairy farming is mainly constrained by factors such as capital and technology shortage rather than the educational experience of breeders. Meanwhile, small-scale dairy farmers were slower to respond to changes in market demand and adjust their production decisions, making it challenging to significantly improve the farming effect in the short term. The regression coefficient of the urban–rural income gap was negative. The widening income gap might discourage farmers from producing and lead to the loss of dairy farming labor in rural areas, disrupting the market order and stability of the dairy farming industry. The regression coefficient of production structure in large-scale dairy farming was 0.0116, significant at the 10% level, suggesting that the GTFP was higher in the region where dairy farming was the primary industry; the likely reason is that if the share of dairy farming in the GDP is high, the dairy farming industry will be valued. Hence, the policy supporting the dairy farming industry will likely be strengthened in that region. Furthermore, the large-scale dairy farmers will be subject to more significant policy benefits than small- and medium-scale dairy farmers.

3.3. Threshold Effect of Environmental Regulation on the Green Total Factor Productivity of Dairy Farming in China

3.3.1. Threshold Effect Test

To test the nonlinear impact of environmental regulation on the green total factor productivity of dairy farming, we conducted a self-sampling (Bootstrap) test, which used the urban–rural income gap as the threshold variable to test whether a threshold effect of environmental regulation existed on the GTFP.

Table 7. Threshold effect test.

Dairy Farming	Number of Thresholds	F-Value	p-Value		Threshold Values
				10%	5%	1%
Small scale	Single	23.41	0.0327	15.9514	20.2844	28.2546
	Double	14.02	0.0680	12.2577	15.4749	21.3220
	Triple	5.17	0.6800	13.7564	19.5347	34.5539
Medium scale	Single	17.97	0.0933	17.4424	21.0679	26.6982
	Double	8.07	0.3767	14.1031	18.8439	29.8340
Large scale	Single	9.35	0.3300	15.3645	19.1039	31.7692

shows the test results of the threshold effect.

With the urban–rural income gap as the threshold variable, a double threshold occurs in small-scale dairy farming. The F-values of the single and double threshold passed the significance test at the 5% level. In medium-scale dairy farming, we observed a single threshold effect of environmental regulation on the GTFP of dairy farming based on the urban–rural income gap. However, the threshold effect in large-scale dairy farming was not significant. Hence, environmental regulation exerted a nonlinear effect on the GTFP of small- and medium-scale dairy farming owing to the change in the urban–rural income gap.

3.3.2. Threshold Estimates and Confidence Intervals

Based on the threshold effect, the threshold values and confidence intervals (CI) were measured (

Table 8. Threshold estimates and confidence intervals.

Dairy Farming	Number of Thresholds	Threshold	Lower Confidence Limit	Upper Confidence Limit
Small scale	Single	0.7999	0.7878	0.8025
	Double	0.9040	0.8830	0.9044
Medium scale	Single	1.3484	1.2698	1.3563

). The double threshold values of 0.7999 and 0.9040 in small-scale dairy farming were located within the CI of (0.7878, 0.8025) and (0.8830, 0.9044), respectively. In medium-scale dairy farming, the single threshold value of 1.3484 was located within the CI of (1.2698, 1.3563).

3.3.3. Threshold Model Regression Results and Analysis

The regressions were estimated separately for small- and medium-scale dairy farming using the urban–rural income gap as the threshold variable.

Table 9. Threshold model regression results.

Variable	Small Scale		Variable	Medium Scale
	Coefficient	p-Value		Coefficient	p-Value
lnER (lnGAP ≤ 0.7999)	0.0253 *	0.0620	lnER (lnGAP ≤ 1.3484)	0.0705 *	0.0940
lnER (0.7999 < lnGAP ≤ 0.9040)	−0.2147 ***	0.0000	lnER (1.3484 < lnGAP)	−0.4751 ***	0.0010
lnER (0.9040 < lnGAP)	0.0759	0.7800
lnECO	0.2338 ***	0.0050	lnECO	0.2297 **	0.0140
lnFT	0.0739 ***	0.0090	lnFT	−0.0203	0.7630
lnEDU	0.6132	0.1860	lnEDU	1.5383 ***	0.0000
lnMD	0.1234	0.7220	lnMD	0.4332 *	0.0910
lnPS	0.0398	0.3070	lnPS	−0.0249	0.5120
Cons.	1.7870	0.4370	Cons.	−3.1067 *	0.0640
R2	0.8137		R2	0.9263

Note: *, **, and *** mean significance at the levels of 10%, 5%, and 1%, respectively.

shows the estimated results. The impact of environmental regulations on the GTFP of dairy farming varied with the changes in the urban–rural income gap. In small-scale dairy farming, the impact of environmental regulation on the GTFP of dairy farming was positive when the urban–rural income gap was below the first threshold (0.7999). When the urban–rural income gap was between the first threshold value (0.7999) and the second threshold value (0.9040), the impact of environmental regulation on the GTFP changed from positive to negative. When the urban–rural income gap was greater than the second threshold (0.9040), the impact of environmental regulation on the GTFP of dairy farming became positive again, but the result was not significant. In medium-scale dairy farming, the impact of environmental regulation on the GTFP of dairy farming changed from positive to negative when the urban–rural income gap exceeded 1.3484, suggesting that the impact of environmental regulation on the GTFP of dairy farming was not a simple linear relationship and that a threshold effect existed based on the urban–rural income gap. Environmental regulation positively affected the GTFP of dairy farming only when the urban–rural income gap was within a reasonable degree.

3.4. Robustness Tests

To test the robustness of the regression results, we reran the random effect model by substituting the core explanatory variable. The core explanatory variable “environmental regulation” was substituted by the number of environmental protection regulations in each region. In addition, the impact of environmental regulations on the GTFP was retested.

Table 10. Robustness tests.

Variable	Small Scale	Medium Scale	Large Scale
lnER	0.0525 *	0.1086 ***	0.0304 *
	(0.0422)	(0.0300)	(0.0266)
lnECO	0.1818 ***	0.1831 ***	0.0365
	(0.0639)	(0.0582)	(0.0536)
lnFT	0.0205 *	−0.0131	0.1920 ***
	(0.0697)	(0.0657)	(0.0532)
lnEDU	0.5222	1.0463 ***	0.0020 *
	(0.4491)	(0.3483)	(0.3289)
lnMD	0.0964	0.0952 **	0.0362 *
	(0.0793)	(0.0389)	(0.0351)
lnGAP	−0.5113 ***	−0.2017 *	−0.5461 ***
	(0.1323)	(0.1248)	(0.1269)
lnPS	0.0333	0.0235	−0.0226
	(0.0304)	(0.0256)	(0.0277)
Cons.	0.8318	−0.7904	2.2312 **
	(1.3460)	(1.0705)	(0.9855)
R2	0.8328	0.6895	0.6055

Note: *, **, and *** mean significance at the levels of 10%, 5%, and 1%, respectively.

shows the regression results. The results show that environmental regulations exerted a significant positive impact on the GTFP, proving the robustness of the previous empirical results.

To confirm the effectiveness of empirical results, this paper compares the results of this study with previous studies, as shown in

Table 11. Comparison of previous results and our results.

Reference	Previous Results	Our Results
Huang, X., et al. (2022) [73]	The mean value of agricultural green total factor productivity growth in China is 0.0296.	The average annual growth rates of the GTFP of small-, medium-, and large-scale dairy farming is 1.98%, 2.73%, and 3.27%, respectively.
Li, J., et al. (2021) [74]	The value of green total factor productivity on Chinese laying hens in Zhejiang province and Jiangsu province is 1.00958 and 1.00397, respectively.	The value of green total factor productivity on medium-scale dairy farming in Zhejiang province and Jiangsu province is 1.0228 and 1.0278, respectively.
Liang, Z., et al. (2020) [75]	The technical progress index and technical efficiency index of the GTFP in the logistics industry is 1.0167 and 0.9911, respectively.	The technical progress index and technical efficiency index of the GTFP in large-scale dairy farming is 1.0300 and 1.0068, respectively.
Li, X.F., et al. (2021) [76]	The impact coefficient of environmental regulations on GTFP is 0.391.	The regression coefficients of environmental regulations on GTFP in small-, medium-, and large-scale dairy farming were 0.0350, 0.0664, and 0.0842, respectively.
Li, X.F., et al. (2021) [28]	When per capita GDP is low (GDP ≤ 12,873), the coefficient of environmental regulation on GTFP is 0.238. When per capita GDP is at a medium level (12,873 < GDP ≤ 55,447), the coefficient of environmental regulation on GTFP is −0.025.	In small-scale dairy farming, the coefficient of environmental regulation on the GTFP is 0.0253 when the urban–rural income gap is below the first threshold (0.7999). When the urban–rural income gap is between the first threshold value (0.7999) and the second threshold value (0.9040), the coefficient of environmental regulation on the GTFP is −0.2147.

. Previous studies have mainly focused on the green total factor productivity of the Chinese laying hens industry, agricultural industry, and so on. We found that some previous results are similar to this paper’s results.

4. Conclusions

Based on panel data from 27 provinces in China from 2009 to 2020, this study applies the SBM-ML model to measure the green total factor productivity of dairy farming and further analyzes the impact of environmental regulations on the green total factor productivity of dairy farming. The main conclusions are as follows: (1) the GTFP of dairy farming in China exhibited a fluctuating upward trend from 2009 to 2020, and technical efficiency is the main driving factor of GTFP growth; (2) environmental regulations positively affect the GFTP of dairy farming in China. In addition, a threshold effect exists between environmental regulations and GTFP based on the urban–rural income gap in small- and medium-scale dairy farming. These findings provide the necessary evidence for improving GTFP in dairy farming. In the future, the government may adjust the intensity of environmental regulations to increase dairy farming efficiency and reduce environmental pollution at the same time.

This study proposes the following policy recommendations: First, the Chinese government should fully promote technological innovation in dairy farming and support farming technology research. The results of the SBM-ML model suggest that the technological progress index of the green total factor productivity of dairy farming is low. Thus, green technology in dairy farming warrants urgent improvement. The government should provide subsidies for green technology innovation in dairy farming and encourage the diffusion of green farming technologies, eventually leading dairy farmers to realize the manure resource utilization and promote the green and efficient development of dairy farming. Second, environmental regulators should implement differentiated environmental regulation policies based on the actual situation of dairy farms. The empirical results demonstrate that the impact of environmental regulation on dairy farms varies by scale. The regulators should implement differential pollution emission standards for dairy farms based on the scales and pollution emissions of dairy farming while ensuring that the intensity of environmental regulations is reasonable. Third, the government must protect dairy farmers’ income stability and augment the farming technology training for dairy farmers. Environmental regulation will exert a negative impact on the green total factor productivity as the urban–rural income gap widens. Thus, the government must stabilize milk prices and provide subsidies for dairy farming, while also encouraging dairy enterprises to execute the industrial layout in rural areas to drive the economic development of dairy farming regions. Besides, the empirical results demonstrate that factors such as breeding technology and level of education exert a positive impact on the green total factor productivity of dairy farming. Accordingly, the Chinese government should provide guidance services on green technologies for dairy farming and conduct technology training on ecological dairy farming. Furthermore, farmers must enhance environmental awareness to attain green and sustainable dairy farming.

This paper provides direct evidence that the environmental regulation makes a great contribution to improving the GTFP of dairy farming in China. However, some limitations exist in this paper. We measure the GTFP of dairy farming in China from 2009 to 2020. As part of the data in 2021 has not been published, it is difficult to obtain the latest data on dairy farming in China. We will continue to collect the data for measurement and analysis in the future. In addition, environmental regulations are composed of command-and-control environmental regulations and market-incentive environmental regulations, whose impact on GTFP in dairy farming is not explored in this paper. Future research should evaluate the impact of command-and-control environmental regulations or market-incentive environmental regulations on GTFP.