1. Introduction
Along with the increasing attention paid to environmental issues, the agricultural environment has gradually become one of the most attractive research topics in the agricultural field in China. Agricultural environment efficiency, which brings environmental factors into traditional agricultural technical efficiency, is an effective index to balance agricultural development and agricultural environment. Especially since China has set forth the aims of carbon peak in 2030 and carbon neutrality in 2060, agricultural environmental efficiency based on carbon emissions and its determinants have caused increasing concern. Existing fruitful studies on the determinants of agricultural environmental efficiency are developed from two aspects of the external agricultural environment and the internal productive factors. In terms of the external environment, agricultural development level [
1,
2], urbanization, natural disasters, production risk [
3], innovation [
4], and energy price [
5] are taken into account. In terms of internal factors, agricultural structure [
2], agricultural labor force, reservoir infrastructure construction, agricultural mechanization service organizations [
5], as well as mechanization [
5,
6,
7] are frequently mentioned.
At the same time, due to the ubiquity of small-scale and fragmented farming, agricultural productive services adapted to China’s national conditions have spread and popularized rapidly. Agricultural productive services are an essential driving force for improving agricultural productivity and promoting agricultural modernization [
8], for its benefits of realizing economies of scale and reducing agricultural production costs without changing the social security functions of cropland [
9,
10]. Nevertheless, agricultural productive services mainly rely on large machinery that consumes diesel and other fuels, emitting large amounts of carbon dioxide. It is worth noting that the fuel consumption of energy activity has been one of the most fundamental origins of agricultural carbon emissions [
11]. However, few studies concentrate on whether the promotion of agricultural productive services exerts a negative impact on the agricultural environment and causes the variation of agricultural environmental efficiency. To date, it is still unclear whether agricultural productive services will lead to the improvement of the agricultural environment. Therefore, more imperative exploration of whether agricultural productive services improve agricultural environmental efficiency from a holistic evaluation perspective can not only conduce to understanding the formation and evolution of agricultural environmental efficiency, but also help to promote the relevant policies for the promotion and application of agricultural productive services.
Agricultural environmental efficiency is determined by agricultural inputs, outputs, and carbon emissions [
7], which are influenced by agricultural productive services. In re-ality, agricultural productive services, also known as agricultural production outsourcing, refer to outsourcing some or all agricultural production stages to service providers or other farmers [
12]. Due to its incremental, overlapping, complementary advantages [
13], agri-cultural productive services can help optimize agricultural factor inputs while keeping agricultural output increased or unchanged. This not only reduces agricultural production costs [
14], but also reduces carbon emissions by optimizing agricultural chemical inputs. In addition, more generally, the cost of agricultural productive services to a service provider is lower than the cost of labor to complete the same task because of specialization. Accordingly, through the adoption of productive services, farmers can not only save production costs, but also save labor time to engage in non-agricultural work to obtain more income [
15]. Moreover, agricultural production services embed green production techniques and green production materials in farmers’ production processes, change their customs and experience with fertilizer application, and thus achieve a reduction in agricultural chemical materials. In addition, agricultural productive services have a mix of benefits, such as increasing the speed of operations, enhancing the timeliness of crucial production stages, improving the ability to cope with weather-related risks, and reducing losses in the harvest process, which exert positive effects on agricultural output [
13]. Machila et al. revealed that agricultural outsourcing had a significant impact on farmers’ crop income and net crop income in Zimbabwe [
16]. Agricultural outsourcing contributes significantly and substantially to household crop income and the net income of farmers who participated in the program of outsourced extension service [
17]. From the perspective of economics, farmers share land operation rights with service providers by purchasing socialized agricultural services, which benefit from specialization arising from the division of labor [
14] and will inevitably affect agricultural production efficiency and agricultural environmental efficiency. Agricultural productivity increased by 25.61% for China’s farmer households who chose agricultural productive services, and the productivity would increase by 10.86% if non-outsourcing farmer households chose to outsource [
18].
In summary, the literature on the impact of agricultural productive services on agricultural inputs and outputs are growing. However, the research on agricultural environmental impact is still scarce, which needs further theoretical analysis and empirical verification. In particular, carbon emissions are the inevitable product of modern agricultural production for null-jointness and weak disposability imposed on agricultural production. It means that the impact of agricultural productive services on agricultural environmental efficiency requires not only further theoretical analysis but also more micro-empirical results for verification. To fill this observed deficiency, this paper empirically examines the effects of productive agricultural services on agricultural environmental efficiency. Furthermore, comprehensive theoretical analyses and empirical tests are needed to explain the environmental effects of agricultural productive services, i.e., the mechanisms by which agricultural productive services contribute to agricultural environmental efficiency. To make up for this theoretical gap, this paper presents a comprehensive overview of the possible influencing pathways, and empirically validates some of them.
In these contexts, based on the panel data of 30 provinces in mainland China from 2004 to 2019, this paper investigates the role of agricultural productive services in improving agricultural environmental efficiency. The findings indicate that agricultural productive services significantly improve agricultural environmental efficiency, which holds steady after endogeneity treatment and a series of robust analyses. Meanwhile, the impact of agricultural productive services on agricultural environmental efficiency shows a marginal decreasing trend. Further research conclusions suggest that the spread of agricultural productive services affects both agricultural inputs and outputs through promoting technology progress, changing cropping structure, and optimizing factor input structure. The above pathways have significant spatial spillover effects.
This paper contributes to the existing research from the following three aspects. First, we examine the heterogeneity of agricultural productive services affecting agricultural environmental efficiency based on accurately measuring agricultural environmental efficiency, expanding not only the study of the effects of agricultural productive services, but also the analysis of the factors influencing agricultural environmental efficiency. Second, we develop an analytical framework for the impact of agricultural productive services on agricultural environmental efficiency, namely that agricultural productive services affect agricultural environmental efficiency through three aspects, i.e., inputs, outputs, and environmental factors. Third, we propose the main mechanisms through which agricultural productive services affect agricultural environmental efficiency at a theoretical level and test empirically through the causal steps approach for the mediating effect test.
The remainder of this paper is structured as follows: the “Methodology and Data” section describes analysis framework, empirical methods, and the nature of data. Econometric results and discussion are presented in the “Results” and “Discussion” sections, and the “Conclusions” section sets out the main conclusions and some policy implications.
2. Methodology and Data
2.1. Analysis Framework of Agricultural Productive Services Affecting Agricultural Environmental Efficiency
In order to reveal the relationship between agricultural productive services and agricultural environmental efficiency in depth, we attempt to explore the influencing pathways of agricultural productive services on agricultural environmental efficiency from four aspects.
Firstly, agricultural environmental efficiency is influenced by agricultural productive services through agricultural technology progress. Most farmers in China face high labor costs and credit constraints, making it challenging to purchase advanced and costly agricultural machinery and choose the reduction scheme for agricultural chemical materials. Agricultural productive services are the best solutions to these problems by reducing the cost of purchasing costly agricultural equipment and lowering the threshold for adopting advanced agricultural technology. On the one hand, the agricultural technology progress, which contains advanced farming technology and advanced energy technology, conduces to improve energy efficiency and directly reduces carbon emissions from energy consumption. On the other hand, agricultural technology progress can indirectly reduce carbon emissions by optimizing the structure of agricultural energy and fertilizer consumption. For example, the growing use of renewable energy can reduce the carbon emission intensity per unit of energy, and the increasing use of new biological fertilizers can increase the carbon sequestration capacity of soil.
Secondly, agricultural environmental efficiency is influenced by agricultural productive services through agricultural planting structure adjustment. Agricultural productive services help to accelerate the large-scale development of agriculture, promote the proportion of grain sown area [
19], and achieve the agglomeration of planting varieties. On the one hand, research shows that, compared with horticultural crops and cash crops, grain crops require the least input, including labor and fertilizer [
19,
20]. Consequently, the increase in the proportion of grain crops sown has led to a decrease in the demand for agricultural chemical materials (i.e., pesticides, agricultural films, and fertilizers) per unit area. On the other hand, grain crops are inclined to be produced by large machinery, resulting in an increased demand for diesel fuel consumption. Furthermore, grain crops generally grow more organic matter (i.e., fruit and straw) than cash crops such as vegetables and flowers, and thus have a more substantial carbon sink effect. Therefore, the bigger the cultivated area of grain crops, the higher agricultural environmental efficiency [
20].
Thirdly, agricultural environmental efficiency is influenced by agricultural productive services through factor allocation optimization. Agricultural productive services are accompanied by the spatial flow of mechanical resources and the transmission of beneficial agricultural information, and thus realize the effective allocation of regional resources [
12]. The popularization of agricultural productive services is accompanied by a shift of agricultural labor to non-agricultural sectors, so the outflow of agricultural labor will strengthen the scarcity of labor as a primary production factor, and then improve the willingness of farmers to use other capitals as a substitute for labor in agricultural production. The decreasing price of machinery relative to labor has led to a decrease in labor input intensity in farm production and an increase in fertilizers, pesticides, and other agricultural chemicals per unit area. Taking fertilizers for example, the outflow of agricultural labor may reduce the frequency of fertilization, but increase the amount of a single fertilization. In addition, the increased use of machinery requires more fuel consumption and brings more carbon emissions, leading to changes in agricultural environmental efficiency. Agricultural productive services contribute to increased specialization in all segments of the agricultural industry chain. The gains from specialization that arise from the division of labor promote the efficiency of agricultural production and realize economies of scale. At the regional level, because of the formation and expansion of productive service markets, agricultural production factors can fully flow and be exchanged between different agriculture operators with the help of agricultural productive services. In other words, agricultural productive services have realized the change of varying factor combinations. The improvement of resource utilization rate eases the excessive use of agricultural chemical resources [
21], and thus leads to the change in carbon emissions.
Fourthly, agricultural environmental efficiency is influenced by agricultural productive services through spatial spillover. From a regional perspective, the higher the level of agricultural productive services in a region (i.e., the higher proportion of persons engaged in productive services and the larger market for agricultural productive services), the greater the empowering effect on agricultural producers in the neighborhood. Relying on technological innovation, technology diffusion, specialized division of labor, collaboration, and the similarity of agricultural production resource endowment conditions in adjacent regions, agricultural machinery, which has apparent diffusion and spillover, forms the agricultural agglomeration effect and enhances the network connection effect in neighboring areas.
Based on the above analysis, we establish an analysis framework (shown in
Figure 1) as follows.
2.2. Empirical Models
2.2.1. Benchmark Model
To test the relationship between agricultural productive services and agricultural environmental efficiency, the ordinary least squares regression with fixed-effect panel model is employed:
where
represents the agricultural environmental efficiency of province
i at year
t,
denotes the level of agricultural productive services,
refers to control variables;
is the intercept term,
are the estimation coefficients of explanatory variable and control variables,
and
represent the fixed effects of province and year, and
is a stochastic error term.
2.2.2. Causal Steps Approach for Mediating Effect Test
To explore the pathways of agricultural productive services affecting agricultural environmental efficiency, based on the test method of Baron and Kenny (1986) [
22], the causal steps approach for mediating effect test is set as follows:
where
represents the mediating variables,
are the intercept terms, and
are the estimation coefficients of corresponding variables. Equation (2) is used to test the effect of the independent variable on the mediating variables, and Equation (3) is used to test the effect of the independent variable on the dependent variable after introducing mediating variables. If the regression coefficient on agricultural productive services decreases or becomes insignificant, it indicates that the impact of agricultural productive services on agricultural environmental efficiency comes partly or entirely through the pathway of mediating variable.
2.2.3. Spatial Econometric Model
We employ the Spatial Dubin Model with fixed effects to verify the spatial spillover effect of agricultural productive services on agricultural environmental efficiency:
where
W represents the spatial weight matrix, and geographical distance matrix is adopted in the paper.
is the first-order lag coefficient of dependent variable,
is the spatial correlation coefficient, and
are the estimated coefficients of each explanatory variable.
2.2.4. Endogeneity and Two-Stage Least Squares
Biases might remain in empirical models because of two types of endogenous problems. The first is the omitted variable problem, i.e., some variables that affect both agricultural productive services and agricultural environmental efficiency are omitted from the regression model. The second is reverse causality, i.e., agricultural environmental efficiency might affect agricultural productive services. For example, provinces with high agricultural environmental efficiency level are usually regions with the high level of factor endowments. These regions, due to the high level of regional development and high price relative to labor, will adapt to local agricultural development requirements through large-scale adoption of agricultural productive services, which in turn contribute to the improvement of their agricultural productive services. This means that the core independent variable and the dependent variable suffer from reverse causality [
23].
To solve the problem of possible omitted variables, we employ a fixed-effect model with panel data to control the unobservable effects at provincial and time level, while adding as many control variables as possible in the regression analysis based on existing studies, such as rural human capital and production risk. To solve the problem of reverse causality, we introduce a suitable instrumental variable and employ two-stage least squares method to alleviate endogenous problems. The first stage is used to obtain prediction to replace endogenous variables, and the second stage is used to draw the final estimation result.
2.3. Variables
2.3.1. Dependent Variable: Agricultural Environmental Efficiency
Following Zhu et al. [
7], we adopt the multi-output stochastic frontier analysis method, which is based on the output-oriented distance function, to obtain agricultural environmental efficiency. In terms of the production function form, we introduce a time-varying parameter model to estimate elastic changes across time accurately. Drawing on the traditional literature [
24,
25], input variables and output variables are selected to calculate agricultural environmental efficiency. The main input variables of agricultural production are labor, machinery, fertilizer, land and fuel, which are measured by the number of employees (in millions), the total power of planting machinery (in million kilowatts), the sum of the gross weight of various fertilizers (in million tons), the sown area (in million hectares) reflecting the actual utilization of the cultivated land, and diesel oil (in million tons) in the planting industry, respectively. We employ the gross value of output (in million CNY) and net carbon sinks (in million tons of CO
2-equivalent) in the planting industry as the output variables because agriculture has the attribute of net carbon sinks. The calculation formula, coefficient of agricultural carbon sinks, and carbon emissions are based on the research of Zhu et al. (2022) [
26].
The descriptive statistics on agricultural input and output variables on provincial level are shown in
Table 1, and the changing trends of agricultural inputs and outputs on national level in China from 2004 to 2019 are shown in
Figure 2.
2.3.2. Independent Variable: Agricultural Productive Services
From the point of view of the demand side (i.e., the objects of agricultural productive services), the sown area and the number of farmers adopting productive services are the ideal indexes to measure the level of productive services; from the point of view of the supply side, the supplier (i.e., the number of agricultural machinery service organizations and agricultural machinery professional service households or organizations) and the power of agricultural machinery for productive services are reasonable indicators. Limited by data availability, we employ the number of people engaged in agricultural productive services per unit area to measure the level of agricultural productive services.
2.3.3. Instrumental Variable
To solve the problem of endogeneity, the usual approach is to introduce a suitable instrumental variable [
23]. Following Zhu et al. [
7], this paper adopts the ratio of road mileage to farmland area in the region as an instrumental variable for agricultural productive services. As the transportation infrastructure, roads do not directly impact agri-cultural output value and net carbon sinks, but they can improve the degree of agricultural productive services via improving road conditions and reducing transportation costs for agricultural machinery.
2.3.4. Mediating Variables
Based on the analysis in the “Literature Review” section, technology progress, planting structure, and input structure are selected to study the influencing mechanism of agricultural productive services affecting agricultural environmental efficiency. Agricultural machinery, which contains advanced planting and harvesting techniques, is accompanied by the transfer and diffusion of management techniques in its service process. Consequently, technology progress is characterized by the ratio of the total power of machinery to farmland area. There are significant differences between cash crops and grain crops in mechanical demand and agricultural chemical materials input [
5], so the planting structure is represented by the ratio of grain sowing area to crop sowing area. The main agricultural inputs include machinery, farmland, labor, and chemical fertilizer. Considering the relevance of machinery and the stability of farmland, the ratio of fertilizer to labor per unit area is used to characterize the input structure.
2.3.5. Control Variables
Agricultural operation scale, planting industry development level, rural human capital, production risk, regional economic development level, part-time employment of labor, and urban-rural income gap are controlled in our models. Agricultural operation scale (in hectare per household), which is an essential factor affecting the consumption of farm machinery and chemical materials [
11], is represented by the ratio of farmland to rural households, and its square term is introduced to examine the possible threshold of land. Based on the viewpoint of comparative advantage, the development level of the planting industry is measured by the share of the planting industry in agriculture. As a factor with strong positive externalities, rural human capital (in years) is calculated as the average length of schooling of the rural population. As farmers will increase fertilizer and pesticide inputs to avoid risks [
3,
27], production risk is selected as the control variable, which refers to the ratio of the affected area to the total sown area of crops. Regional economic development level (in CNY per person), which affects carbon emissions intensity [
1], is expressed by the ratio of a region’s GDP to its population. As an essential factor affecting the capital inputs (i.e., machine, chemical fertilizer, and pesticides) in agricultural production [
11], the part-time employment of labor is measured as the proportion of wage income of rural residents. The urban-rural income gap, which leads to the flow of the labor to cities [
28], is represented by the ratio of disposable income of urban residents to rural residents. In addition, due to the complexity of agricultural production, agricultural environmental efficiency will be affected by many unobserved factors, such as social culture and natural conditions [
29], so provincial and year fixed effects are controlled too.
2.4. Data
From the time perspective, agricultural mechanization and agricultural productive services have developed rapidly in China since 2004, and the data of agricultural productive services have been included in the national statistical yearbook since then. Therefore, the data employed in this study are the province-by-year panel of 30 provinces of mainland China from 2004 to 2019, and Tibet, Hong Kong, Macao, and Taiwan are excluded due to data unavailability. The annual data for each province associated with agricultural production are obtained from the annual China Rural Statistical Yearbook and China Agricultural Machinery Industry Yearbook. In addition, price variables are deflated according to the price level in 2004.
The descriptive statistics on variables are shown in
Table 2. The differences of agricultural environmental efficiency and agricultural productive services among different provinces are presented in
Figure 3 and
Figure 4, showing that there are great differences between samples, which are suitable for regression analysis.
5. Conclusions
Based on the panel data of 30 provinces in China from 2004 to 2019, we adopt a multi-output stochastic frontier analysis method to measure agricultural environmental efficiency based on net carbon sink, then explore the effects and mechanisms of agricultural productive services on agricultural environmental efficiency using OLS with fixed-effect panel data, two-stage least squares with instrumental variable, causal steps approach, and a spatial econometric method. The main conclusions are as follows:
Firstly, agricultural productive services have a significant contribution to agricultural environmental efficiency. Agricultural productive services enhance agricultural productivity as well as the agricultural environment through inputs and outputs, and then improve the efficiency of the agricultural environment. This result still holds after using instrumental variables to deal with endogeneity, changing the measurement of dependent and independent variables, and subdividing the sample. Secondly, the mechanism pathways of agricultural productive services affecting agricultural environmental efficiency are mainly reflected in technology progress, planting structure adjustment, factor allocation optimization, and spatial spillover. Advanced agricultural technology and management measures can improve agricultural productivity and energy efficiency, and optimize the energy consumption structure. A reasonable planting structure is conducive to reducing agricultural chemical inputs, improving factor utilization efficiency, and realizing economies of scale. The rational allocation of factors is the fundamental guarantee for the role of agricultural input resources such as labor, farmland, and chemical fertilizer. Agricultural agglomeration and different crop maturity periods make cross-regional services of agricultural productive services possible, thus generating spatial spillover effects. Thirdly, the impact of agricultural productive services on agricultural environmental efficiency is more significant in the regions with low agricultural productive services level because agricultural productive services, as a factor input in agricultural production, conform to the law of diminishing marginal returns.
Given the above evidence and arguments, we draw some policy implications.
Firstly, policy attention should be paid to improving the level of agricultural productive services continuously according to the actual situation in different regions. For regions with low level of agricultural producer services due to terrain factors or crop structure factors, the marginal contribution of agricultural producer services is more pronounced. It is valuable to promote appropriate agricultural productive services in these areas to promote the quality of agricultural development and environmental efficiency. For regions with a high agricultural producer services level, such as main grain-producing areas, the marginal contribution of agricultural producer services is relatively small. The critical measure is to update or replace the existing productive service equipment, and promote the use of agricultural productive services with more advanced technology and higher efficiency. Secondly, efforts should be made to strengthen information diffusion and regional cooperation to realize the positive spatial spillover effect of agricultural productive services. For regions with similar resource endowments and similar crop types, information cooperation is used to guide and promote the cross-regional operation of agricultural productive services, to achieve rational flow and scientific allocation of agricultural machinery. Thirdly, it is necessary to improve the operating conditions of agricultural productive services, such as appropriately increasing the investment in transportation infrastructure, guiding the centralized farmland transfer, and realizing regional agglomeration of crop planting. Improving transportation infrastructure can provide primary conditions and convenience for agricultural machinery operations. Farmland transfer and crop agglomeration are conducive to realizing economies of land scale and service scale in agricultural production. The scale of expansion is beneficial to cultivating the agricultural productive services market and introducing small farmers to modern agriculture.
There are also some potential limitations in this paper. First, in terms of the research scope, this study was conducted at the provincial level. Future research can be conducted at the county level, which potentially has richer data and more accurate results. Second, in terms of the research depth, as agricultural productive services exist in all aspects of agricultural production, such as tillage, sowing, irrigation, fertilization, and harvesting, a possible research direction is to profoundly investigate the impact of different stages of productive services on agricultural environmental efficiency.