Effect of Agricultural Social Services on Green Production of Natural Rubber: Evidence from Hainan, China

Abstract

It is vital to concentrate on conserving the ecological environment and promoting production efficiency in the process of planting natural rubber. Agricultural social services (ASSs) play an essential role in helping rubber farmers to achieve green production. Based on the survey data of 552 natural rubber farmers in Hainan Province, this study builds an index system regarding socialized services for rubber production from three factors: technology extension services, financial insurance services, and market information services. This research uses the slack-based measure (SBM) model and the Tobit model to analyze the green production efficiency (GPE) and the influences of rubber production socialized services on the rubber growers’ green production efficiency. Our results revealed that (1) the average green productivity of rubber planting in Hainan is approximately 0.41, which means that there is ample space for improvement; (2) ASSs have a significant positive effect on increasing the green rubber production efficiency; and (3) among ASSs, the technical extension service has the most significant effect on improving the green production efficiency of the growers. To further raise GPE of natural rubber, the socialized service system can be strengthened in terms of technology, capital, and information. ASSs have noticeable potential in improving the efficiency of rubber green production while ensuring economic, social, and ecological sustainability.

1. Introduction

As a widely used industrial material, rubber plays an important economic role in many developing countries, such as Malaysia, Thailand, and China [1]. However, the production of rubber is also accompanied by a series of ecological destruction and environmental pollution problems, including a decline in soil fertility, biodiversity, and water resources’ protection capacity in the planting area [2,3,4]. Hainan Island is the largest tropical island in China and is a primary source of natural rubber [5]. During the past few decades, the area of rubber plantations in Hainan has expanded, mainly at the expense of forest and agricultural land, resulting in a reduction in tropical forests [6]. In the context of China’s efforts to achieve a carbon peak by 2030, the Hainan Provincial Government is standardizing the green production practice of rubber through legal and economic measures to effectively solve the ecological and environmental problems caused by rubber planting [7]. However, there are still many obstacles in the actual process of green transformation and development, such as ineffective environmental regulation [8], unreasonable resource allocation [9], and farmers’ limited cognitive ability [10]. In order to overcome these problems, Hainan Province has explored the construction of a socialized service system of the rubber industry, aiming to improve the agricultural green productivity-to achieve the coordinated development of the government, market, and farmers.

Research on the green production efficiency, has primarily focused on rice [11,12], wheat [13], vegetables [14], potatoes [15], and other popular crops, and the rubber green production efficiency has received less attention. A literature study revealed that research on the green production efficiency has included three factors. Firstly, the study of green production measuring techniques to measure green production has mainly used a non-parametric technique combining the data envelopment and exponential methods [16,17]. The second factor is the selection of metrics for measuring green production. In contrast to conventional research institutions for production efficiency, green production efficiency includes resources and the environment in its assessment methodology. Some researchers consider resources and the environment to be the input component in the computation [18,19], but the majority of scholars consider resources and the environment to be the undesirable output factors [20,21]. For other input-output indicators, the total production value is often chosen as the predicted output indicator, while land, labor, applied water, seeds, chemical fertilizer, capital, and machinery are used as input indicators [22,23,24]. The third factor is the study of green productivity’s impacting elements. Such as the agricultural mechanization level, human capital, financial capacity [25,26], regional economic disparities, technological innovation, infrastructure investment [27,28], etc. This study was conducted by the aforementioned researchers on the calculation of green productivity, variable selection, and impact mechanism serves as the foundation and methodology for estimating the green rubber production efficiency.

However, few scholars are interested in the link between social services and green production efficiency, and the mechanism behind the influence of social services on the green production efficiency. China has adopted a productivity-driven agricultural policy system to feed the world’s largest population and developed the largest agricultural service system [29]. Some works found that ASSs have a strong impact on the efficiency of production. Chen (2022) found that agricultural technology extension and financial insurance showed the most significant relationship with the production efficiency of farmers [30]. Zang (2022) found that farmers’ characteristics, biophysical conditions, and attributes of the community could influence the process of smallholders’ participation in agricultural socialized services [31]. With the global spread of the COVID-19 pandemic, production in all industries has been severely impacted [32]. Some studies have found that digital technology is playing an increasingly important role as a driver in industrial revitalization [33,34,35]. Shen (2022) found that internet popularization and information technology promote green growth in Chinese provincial agriculture [36]. The above studies found that technical extension, market information, and financial and insurance services play an increasingly important role in the agricultural production process. With the accelerated development of agricultural socialization services, the influence mechanism of agricultural socialization services on green agricultural production should be considered as much as possible in future research.

We focus on the green production efficiency of natural rubber in the green production of tropical cash crops. Currently, the green production efficiency (GPE) of short-term food crops, such as rice and wheat, has been explored, but GPE of long-term commercial crops, such as natural rubber, requires further investigation. This study chose natural rubber, an economic crop in tropical regions, as its subject for the following reasons. First, there are relatively few studies on the green productivity of economic crops. As a tropical commercial crop, natural rubber production and planting procedures vary from those of rice and other short-term crops. In addition, rubber belongs to the category of trees, and the carbon storage capacity and carbon sink function of rubber forests are greater than those of short-term crops, such as rice and wheat [37]. Third, with the strengthening of the socialized service system, it is playing a vital role in promoting green production. The socialized service may boost the green productivity of natural rubber, which will contribute to the green growth of the natural rubber sector.

The objective of our work was to explore the impact of the socialization service of rubber production on the green production efficiency of rubber plantations. This work consists of four sections: the first section analyzes the influence mechanism of the socialization service on natural rubber green production at the theoretical level. The second section describes the data source, the used econometric model, and the chosen variables. In the third section, the slack-based measure (SBM) is used to determine the present green efficiency of natural rubber planting in Hainan, followed by an analysis of the variables influencing the green production efficiency and a test of the model’s robustness. Finally, we discuss and summarize the results of this study. We hope that this article will enrich the research on the green production efficiency of natural rubber and be adapted to more tropical cash crops, such as coconut and palm.

2. Theoretical Analysis

Recently, the Chinese government has accelerated the socialized agricultural service system to help small farmers modernize and meet the ever-increasing requirements of food and fiber requirements [30]. As an important link in agricultural production, agricultural social services not only promote the improvement of agricultural productivity but are also gradually changing the farmer’s production behavior [38]. In China, socialized agricultural services are currently provided mainly by the government’s public agricultural extension department [39]. They mainly include production, financial, and information services.

This study classified the social services received by rubber farmers into three categories: (1) technology extension services, including new rubber-cutting technology, green pest control technology, soil testing formula fertilization technology, and intercrop technology. Most rubber farmers still adopt traditional rubber production technology, resulting in high pollution. Recently, the government agricultural science and technology department increased the promotion of rubber green production technology, hoping to improve the rubber production efficiency and reduce the input of chemical products. Factor inputs in the rubber production process include two kinds of factor inputs. The first resource factor input includes labor, land, and capital. The second environmental factor input includes pesticides, fertilizers, and fuel. Farmers can promote the rational use of pesticides, fertilizers, fuel, and other environmental factors by adopting green production technology services in socialized services; (2) Financial insurance services. Green insurance has been introduced to advance green development in recent years, and some useful results have been achieved [40]. The existing rubber financial insurance services in Hainan include rubber income price insurance, rubber windstorm insurance, and rubber futures insurance. This is an important method for improving the green productivity of natural rubber by providing rubber to farmers with financial subsidies and guiding farmers to choose environmentally friendly green production behaviors. The financial insurance reduces risk and promotes the production motivation of farmers; (3) Market information services. Rubber production farmers are provided with information directly related to green production, including information on how to use soil formula fertilization to achieve precise fertilization, how to use intercrop production to effectively protect the quality of cultivated land, and how to use green pest control to reduce the use of chemical pesticides and other green input use instructions. The more information about green rubber production provided, the higher the farmers’ understanding of green production. The green production efficiency of rubber farmers will be improved.

Since the purpose of this work was to determine whether socialization services can increase the green productivity of rubber farmers, the influences of socialization services are explained using the two dimensions of input and output. On the one hand, social services can reduce the amount of agricultural chemicals such as fertilizers, pesticides, fuel, and other inputs, which can lead to the transformation of traditional rubber production to green production, an improvement of the production yield, an increase in farmers’ production income, and an increase in the desired output, which in turn helps to improve green productivity. On the other hand, social services can control the amount of chemical pollution emissions, reduce environmental pollution, reduce non-desired output, and thus increase agricultural green productivity. The mechanism of this impact is shown in Figure 1.

3. Materials and Methods

3.1. Study Area

The study area of this paper is six central natural rubber-growing counties in Hainan Province (Figure 2). For the past 20 years, the natural rubber industry in Hainan Province has been dramatically promoted. However, the rapid expansion of natural rubber planting has resulted in a number of environmental and ecological problems, such as a reduction in tropical area, eutrophication of surface water, and nutrient imbalance of land [41]. In order to alleviate the negative impact of the unreasonable production process on the regional ecological environment, the government has accelerated the construction of socialized services for rubber production to help rubber farmers reduce their use of pesticides and fertilizers, improve green production technology, and promote the improvement of the green production efficiency. In 2020, Hainan’s rubber planting area was 526,900 hectares, with an output of 330,800 tons (Statistical Bureau of Hainan Province. Hainan Statistical Yearbook; China Statistical Publishing House: Beijing, China, 2021). Rubber plantation has become a dominant vegetation type in Hainan [42]. The top 4 cities in Hainan’s natural rubber area are Danzhou, Qiongzhong, Baisha, and Chengmai, respectively. Among them, Danzhou in western Hainan has the largest rubber planting area of 84,800 hectares (Figure 3). Due to the high similarity in the production of natural rubber, the production materials and production techniques used by the growers are basically the same. Therefore, it is feasible to measure GPE of rubber growers. Therefore, we selected six major natural-rubber-producing counties in Hainan as the research area, namely the western region (Danzhou City and Chengmai County), the central region (Baisha County and Qiongzhong County), and the eastern region (Qionghai City and Wanning County). This accounts for 61.27% of the total area of natural rubber planted in Hainan, which is representative for studying the influence of social services on the green production efficiency of rubber plantation farmers.

3.2. Collected Data

The data was obtained from a field survey of natural rubber growers in Hainan Province from October to December 2021. The survey was implemented in six cities in Hainan Province, including Baisha County, Chengmai County, Danzhou City, Qionghai City, Qiongzhong County, and Wanning City (Figure 2). The survey was conducted by team members of the natural rubber industry economic research post through face-to-face interviews with rubber-growing farmers in Hainan. Questionnaires were distributed through stratified sampling and random sampling. The selection of survey areas followed the sampling requirements of unbiased, random, and consistent sampling. First, considering the representativeness of rubber cultivation, six cities and counties in Hainan where natural rubber is grown were selected. Then, three townships were randomly selected from each city and county. Finally, 30 farmers growing natural rubber were randomly selected from each township for field interviews. The survey questionnaire mainly covered technical extension services, financial and insurance services, market information services, and the personal and family characteristics of farmers in rubber plantation production and social services. In order to ensure the authenticity and reliability of the survey data, we conducted uniform training for the team members before the research and reviewed a large amount of literature on the green production of natural rubber. At the same time, we also held in-depth discussions with experts in related fields and local government departments and continuously optimized the survey questionnaire according to the experts’ opinions. A total of 580 questionnaires were distributed in this study. After checking invalid questionnaires, 552 valid questionnaires were obtained, with an efficiency rate of 95.17%.

3.3. Methods

3.3.1. Slack-Based Measure (SBM)

In the rubber production process, in addition to expected outputs such as income, there are also undesired outputs such as carbon dioxide emissions. In this work, the SBM-undesirable model was chosen to measure the green efficiency of rubber planting, and the undesired output of carbon emissions was adapted to this model (Equation (1). It is assumed that the i-th region has m inputs in rubber production in the t-th year (2021 in this work). Each rubber farmer is regarded as a production decision unit (DMU), and multiple DMUs in the base period are used to construct the GPE of the 2021 year. It is assumed that there are DMUs recorded as ${\mathrm{DMU}}_{\mathrm{j}}$ (j = 1, 2, 3,…, z). Moreover, each DMU has m input; denoted as ${ \mathrm{x}}_{\mathrm{i}}$ (i = 1, 2,3, …, m); a total of n outputs, ${\mathrm{n}}_1$ is the expected output as${ \mathrm{y}}_{\mathrm{r}}$ (r = 1, 2, 3, …,${\mathrm{n}}_1$), ${\mathrm{n}}_2$ is the undesired output as ${\mathrm{y}}_{\mathrm{t}}$ (t = 1, 2, 3,…, ${\mathrm{n}}_2$), and n = ${\mathrm{n}}_1$ + ${\mathrm{n}}_2$. Based on the above definition, the SBM model can be expressed as:

$$DM{U}_j\{\begin{array}{c}{\mathsf{\rho }}^{*}=\min \frac{1-\frac{1}{\mathrm{m}}{\sum }_{\mathrm{i}=1}^{\mathrm{m}}\left(\frac{{\mathrm{S}}_{\mathrm{i}0}^{-}}{{\mathrm{x}}_{\mathrm{i}0}}\right)}{1+\frac{1}{\mathrm{n}1+\mathrm{n}2}[{\sum }_{\mathrm{r}=1}^{\mathrm{n}1}\left(\frac{{\mathrm{S}}_{\mathrm{r}0}^{\mathrm{g}}}{{\mathrm{y}}_{\mathrm{r}0}^{\mathrm{g}}}\right)+{\sum }_{\mathrm{t}=1}^{\mathrm{n}2}\left(\frac{{\mathrm{S}}_{\mathrm{t}0}^{\mathrm{b}}}{{\mathrm{y}}_{\mathrm{t}0}^{\mathrm{b}}}\right)]}\\ \mathrm{S}.\mathrm{t}.\{\begin{array}{c}{\mathrm{x}}_0=\mathrm{X}{\mathsf{\lambda }}_{\mathrm{j}}+{\mathrm{S}}^{-}\\ {\mathrm{y}}_0^{\mathrm{g}}={\mathrm{Y}}^{\mathrm{g}}{\mathsf{\lambda }}_{\mathrm{j}}-{\mathrm{S}}^{\mathrm{g}}\\ {\mathrm{y}}_0^{\mathrm{b}}={\mathrm{Y}}^{\mathrm{b}}{\mathsf{\lambda }}_{\mathrm{j}}+{\mathrm{S}}^{\mathrm{b}}\\ {\mathsf{\lambda }}_{\mathrm{j}}=1\\ {\mathrm{S}}^{-},{\mathrm{S}}^{\mathrm{g}},{\mathrm{S}}^{\mathrm{b}},{\mathsf{\lambda }}_{\mathrm{j}}\ge 0\end{array}\end{array}$$

(1)

The objective function value ρ* satisfies the following condition: 0< ρ* < 1, where λ is an intensity variable. $\sum _{\mathrm{j}}^{\mathrm{z}}{\mathsf{\lambda }}_{\mathrm{j}}$ = 1 means variable returns to scale (VRS) [43]. If this condition is not satisfied, it means that there are constant returns to scale (CRS) [44]. Since CRS is obtained under the premise of a constant return to scale, but the return to scale in reality is not fixed [45], we chose VRS to measure the efficiency. ρ* is the value of GPE, where ${\mathrm{S}}^{-}$, ${\mathrm{S}}^{\mathrm{g}}$, and ${\mathrm{S}}^{\mathrm{b}}$ represent the slack variables of the input variable, expected output, and undesired output, respectively. The optimal solution to the above equation is (λ, ${\mathrm{S}}^{-}$, ${\mathrm{S}}^{\mathrm{g}}$, ${\mathrm{S}}^{\mathrm{b}}$). When ${\mathrm{S}}^{-}$ = ${\mathrm{S}}^{\mathrm{g}}$ = ${\mathrm{S}}^{\mathrm{b}}$ = 0, there is an optimal solution of the function, i.e., ρ* = 1, which represents a fully efficient decision unit; if 0 ≤ ρ* ≤ 1, it means that there is a loss of efficiency in the decision unit, and corresponding improvements can be carried out in the inputs and outputs.

3.3.2. Tobit Model

Since the green production efficiency of rubber calculated by the SBM model takes values in the range of [0, 1], it is a typical “restricted explanatory variable” with truncation at both ends, and if the ordinary least squares (OLS) method is used for parameter estimation, it will lead to inconsistent and biased results. Therefore, the Tobit model was used in this paper to explore the exogenous factors influencing the green production efficiency of natural rubber. Tobin first proposed the Tobit regression model in 1958 for taking values when the dependent variable is continuous but subject to certain restrictions [46]. In this work, the Tobit model was used to analyze the factors influencing the green production efficiency of natural rubber by agricultural social services, mainly because when the green production efficiency of each rubber-growing farmer is calculated using the SBM model, the efficiency values of some farmers may be at the boundary values; if the general regression model is used for the analysis, effective data may be truncated, and the consistency of the estimated values cannot be guaranteed. Therefore, the research in this article used the Tobit model to analyze the influencing factors of the green production efficiency of natural rubber, and then further discover the key factors affecting the green production efficiency of rubber. The development of the model is as follows (Equations (2)):

$${\mathrm{D}}_{\mathrm{Tobit}}\{\begin{array}{c}{\mathrm{y}}_{\mathrm{i}}=\mathsf{\alpha }+{\mathrm{x}}_{\mathrm{i}}\mathsf{\beta }+{\mathrm{u}}_{\mathrm{i}} ({\mathrm{y}}_{\mathrm{i}}>0)\\ {\mathrm{y}}_{\mathrm{i}}=0 \left({\mathrm{y}}_{\mathrm{i}}\le 0\right)\\ {\mathrm{u}}_{\mathrm{i}}\sim \mathrm{N}\left(0,{\mathsf{\sigma }}^2\right)\end{array}$$

(2)

where the explanatory variable ${\mathrm{y}}_{\mathrm{i}}$ is the value of green productivity of natural rubber, ${\mathrm{X}}_{\mathrm{i}}$ is the factor affecting the green productivity of rubber; $\mathsf{\beta }$ is the parameter estimation coefficient;${ \mathsf{\mu }}_{\mathrm{i}}$ is the error term, ${\mathsf{\mu }}_{\mathrm{i}}$∼N (0, ${\mathsf{\sigma }}^2$). The explanatory variable ${\mathrm{X}}_{\mathrm{i}}$ takes the actual observed value, and the explained variable ${\mathrm{y}}_{\mathrm{i}}$ is in a restricted manner. When the value ${\mathrm{y}}_{\mathrm{i}}$ ≥ 0, we take the actual value for the operation; when ${\mathrm{y}}_{\mathrm{i}}$ < 0, we assign the actual value to 0. $\mathsf{\alpha }$ is the fixed effect of the rubber farmer, which is an unknown constant.

3.3.3. Propensity Score Matching (PSM)

We used a propensity score matching (PSM) model to investigate whether social services have an impact on the green production of natural rubber by considering the heterogeneity between the “adoption” and “non-adoption” of agricultural social services by rubber farmers. The reason for adopting the PSM method is that there is a problem of self-selection of the sample as each farmer has different basic conditions for receiving social services or not. If the differences in green productivity due to farmers receiving or not receiving social services are directly compared, this can easily lead to systematic bias. The ideal solution is to control or reduce the interference of non-treatment factors, highlighting the effect of treatment factors [47].Therefore, the use of the PSM model sets the rubber farmers who adopted social services as the experimental group and the farmers who did not adopt them as the control group. The characteristics of the control group farmers were simulated through various matching methods so there was no significant difference between the two groups of farmers after matching except the adoption of socialized services. The PSM model obtained the average treatment effect (ATT) of socialized services on the green productivity of rubber farmers by constructing a counterfactual framework. According to the definition of Rosenbaum [48], the specific model was set as follows (Equations (3)–(5)):

Primarily, regression is carried out on the selection equation of whether rubber farmers adopt socialized services:

$${\mathrm{P}}_{\mathrm{r}}\left({\mathrm{x}}_{\mathrm{i}}\right)={\mathrm{P}}_{\mathrm{r}}\left({\mathrm{D}}_{\mathrm{i}}=1{|\mathrm{x}}_{\mathrm{i}}\right)$$

(3)

The formula ${\mathrm{D}}_{\mathrm{i}}$ = 1 indicates that the farmers have adopted socialized services; ${\mathrm{P}}_{\mathrm{r}}\left({\mathrm{x}}_{\mathrm{i}}\right)$ illustrates the conditional probability of farmers adopting socialized services; ${\mathrm{x}}_{\mathrm{i}}$ is the selected matching variable. The Logit formula is used to calculate the propensity score value of farmers adopting socialized services:

$${\mathrm{P}}_{\mathrm{r}}\left({\mathrm{x}}_{\mathrm{i}}\right)={\mathrm{P}}_{\mathrm{r}}\left({\mathrm{D}}_{\mathrm{i}}=1{|\mathrm{x}}_{\mathrm{i}}\right)=\frac{\exp \left({\mathsf{\beta }\mathrm{x}}_{\mathrm{i}}\right)}{1+\exp \left({\mathsf{\beta }\mathrm{x}}_{\mathrm{i}}\right)}$$

(4)

After obtaining the propensity score, the matching results are obtained according to different matching methods. The ATT expression of the impact of socialized services on GPE is:

$$\mathrm{ATT}=\mathrm{E}\left[{\mathrm{Y}}_1^{\mathrm{i}}-{\mathrm{Y}}_0^{\mathrm{i}}{|\mathrm{D}}_{\mathrm{i}}=1\right]=\mathrm{E}\left[{\mathrm{Y}}_1^{\mathrm{i}}{|\mathrm{D}}_{\mathrm{i}}=1\right]-\mathrm{E}[{\mathrm{Y}}_0^{\mathrm{i}}|{\mathrm{D}}_{\mathrm{i}}=1]$$

(5)

where ${\mathrm{Y}}_1^{\mathrm{i}}$ demonstrates the GPE of farmers who have adopted social services. ${\mathrm{Y}}_0^{\mathrm{i}}$ represents the green productivity of farmers who have not adopted social services. ATT is the net effect of the treatment group.

3.4. Variables

3.4.1. Input Variables and Output Variables

Based on natural rubber as the research object, to measure the efficiency of production of green, the evaluation index system of natural rubber’s green production was constructed. At the same time, considering the rubber production practical situation and data availability, the rubber planting area and the amount of labor employment, fertilizers, pesticides, and agricultural diesel gasoline were chosen as indicators of the production input, with rubber income as the expected output indicator. The carbon dioxide emission of rubber plants was taken as the undesired output index [49]. To achieve the goal of controlling greenhouse gas emissions, China has acceded to the United Nations Framework Convention on Climate Change (UNFCCC), which includes a reduction in carbon dioxide emissions as a binding target in its long-term plans for national economic and social development [50]. For the calculation of carbon emissions from rubber production, according to our actual research, in the calculation process, three indexes of fertilizer, fuel oil, and pesticide input by rubber farmers in the process of rubber planting were selected as carbon sources to calculate the carbon emissions. For the measurement of carbon emissions, we refer to the parameters in Carbon Dioxide Information Analysis Centre and China Emission Accounts and Datasets (

Table 1. Major carbon sources and carbon emission factors.

Carbon Emission Sources	Emission Factors	Reference Sources
Chemical fertilizer	0.8956 kg/kg	Carbon Dioxide Information Analysis Centre (CDIAC)
Pesticides	4.9341 kg/kg	Carbon Dioxide Information Analysis Centre (CDIAC)
Fuel	0.5927 kg/kg	China Emission Accounts and Datasets (CEADs)

). The calculation of the carbon emission in this work adopted the method of multiplying the corresponding index by the carbon emission coefficient. The calculation formula is as follows:

$$\mathrm{P}={\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{P}}_{\mathrm{i}}={\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{P}}_{\mathrm{i}}\times {\mathsf{\sigma }}_{\mathrm{i}}$$

(6)

where P represents the total amount of carbon emissions from agricultural land use; ${\mathrm{P}}_{\mathrm{i}}$ is the carbon emissions of various carbon sources; ${\mathrm{T}}_{\mathrm{i}}$ is the quantity of various carbon sources; and ${\mathsf{\sigma }}_{\mathrm{i}}$ is the carbon emission coefficient of various carbon sources. The final green efficiency index system of rubber production is shown in

Table 2. The green efficiency index system of rubber production.

Items	Variables	Variable Description	Max	Min	Average	Standard
Input Variables	Labor input	Working days of rubber production	145.00	30.00	88.59	22.51
	Land input	Rubber planting area (mu)	90.00	3.00	22.61	15.18
	Tool input	Input of rubber production tools (yuan)	3900.00	100.00	757.71	652.84
	Fertilizer input	Fertilizer usage in rubber plantation (kg)	3200.00	10.00	474.32	479.38
	Pesticide input	Pesticide usage in rubber plantation(kg)	107.14	0.89	14.47	16.16
Expected output variable	Planting income	Total income from rubber plantation (yuan)	168,500.00	3000.00	27,031.05	24,805.82
Unexpected output variable	CO2 emissions	Fertilizer carbon emissions (kg)	2865.92	8.59	424.80	429.33
		Pesticide carbon emissions (kg)	528.65	4.40	71.41	79.74
		Fuel carbon emissions (kg)	222.26	1.48	43.66	42.03

.

In terms of production indices, the highest annual income of the 552 rubber planting families is 168,500 yuan, the lowest is 3000 yuan, and the average annual income of the sample farmers is 27,031.05 yuan. In terms of input indicators, the yearly labor input of the family with the lowest labor input is 30 days and the average annual labor input value of the cultivators is 88 days. The minimum planting area for rubber is 3 mu, the maximum is 90 mu, and the average planting area of the sample farmers is 22.61 mu. The sample farmers’ average chemical fertilizer input is 474.32 kg, average pesticide input is 14.47 kg, and average fuel input is 73.67 kg. From the perspective of carbon emissions, the use of chemical fertilizers in the rubber planting process is the largest source of carbon emissions.

3.4.2. Description of Variables Affecting Rubber Green Productivity

Core variables: Combined with the previous theoretical analysis, this work selected technology promotion services, production subsidy services, and market information services as the core explanatory variables. This survey was based on core variables and asked rubber farmers: “Have you received technical promotion services related to rubber production?”, “Have you received financial insurance services related to rubber production?”, and “Have you obtained market information services related to rubber production?”. If the surveyed grower answered “Yes”, the variable was assigned a value of 1. If the grower answered “No”, the variable was assigned a value of 0.

Control variables: In order to focus on the impact of ASSs on the green productivity of rubber growers, it was necessary to control other factors that may affect the green productivity of growers. This work selected the characteristics involving age, gender, health status, education level, planting scale, social identity, risk appetite, household income, and labor type as control variables. The variables affecting the green productivity of rubber farmers are shown in

Table 3. Description of the variables influencing green productivity.

Items	Variables	Definition	Average	Standard
Core explanatory variables	Financial insurance services	Have you received financial insurance services for rubber production? yes = 1, no = 0	0.68	0.46
	Market information service	Have you accepted market information services for rubber production? yes = 1, no = 0	0.60	0.49
	Technical extension services	Have you received technical extension services for rubber production? yes = 1, no = 0	0.52	0.50
Control variables	Age	Respondent’s age (years)	51.41	9.25
	Gender	Male = 1, Female = 0	0.90	0.30
	Health status	1 = poor, 2 = average, 3 = healthy	2.88	0.36
	Educational level	1 = elementary school; 2 = junior high school; 3 = high school; 4 = college or above	2.03	0.88
	Social identity	Whether served as a village cadre	0.28	0.45
	Planting scale	Rubber planting land area (mu)	22.61	15.18
	Labor type	1 = Agricultural; 2 = Part-time; 3 = Non-agricultural	1.51	0.84
	Household income	Farming household rubber income (ten thousand yuan)	2.70	2.48
	Risk appetite	1 = risk averse, 2 = risk neutral, 3 = risk like	1.46	0.78

.

4. Results

4.1. Green Production Efficiency for Rubber Farmers

Based on field research data, DEAP.2.0 software was used to calculate the green production efficiency of natural rubber growers. The calculation results show that the current average green productivity of natural rubber planting in Hainan Province is 0.41, and there is still much room for improvement. In order to preliminarily analyze the differences in green productivity, according to the level of green productivity, we divided the productivity as follows: I (green productivity value below 0.25); II (green productivity value between 0.25 and 0.50); III (green productivity value between 0.50 and 0.75); and IV (green productivity value above 0.75). In order to further analyze the impact of socialized services on green productivity, the samples were grouped according to the number of farmers using socialized services and divided into a “not receiving services”, “receiving one service”, “receiving two services”, and “receiving three services” (see

Table 4. Types of socialized services used and the distribution of green productivity.

Green Productivity	No Social Service	Received One Service	Received Two Services	Received Three Services
	Number (57)	Number (153)	Number (186)	Number (156)
Ⅰ: 0–0.25	38	74	60	15
Ⅱ: 0.25–0.50	15	58	79	78
Ⅲ: 0.50–0.75	1	11	17	22
Ⅳ: 0.75–1.0	3	10	30	41
Average value	0.27	0.32	0.42	0.54

). Through this grouping, it was found that the farmers who adopt more socialized services obtain a higher green productivity. The average green productivity of rubber farmers in the “receiving three services” was 0.54. The average green productivity of farmers in the “receiving two services” was 0.42. The average green productivity of the households in the “receiving one service” was 0.32 while the average green productivity of the “not receiving services” households was 0.27. The green productivity distribution of the number of socialized services used by planting farmers in rubber production preliminarily determined that ASSs have a positive effect on improving the green productivity of natural rubber.

Through Stata14.0 statistical analysis software, we conducted Tobit regression on the impact of social services on rubber green productivity. In order to improve the accuracy of the estimated results and to avoid multicollinearity, different socialized services based on the incorporation of control variables were gradually introduced into the model. The corresponding model estimated results were model 1 to model 5 (

Table 5. Tobit regression model estimated results.

Variables	Model 1	Model 2	Model 3	Model 4	Model 5
Financial insurance services	0.0562 ** (0.0235)			0.0562 ** (0.0235)	0.0439 * (0.0234)
Market information services		0.074 2 *** (0.0224)		0.127 *** (0.0253)	0.0619 *** (0.0220)
Technology extension services			0.0700 *** (0.0232)	0.121 *** (0.0258)	0.0578 ** (0.0230)
Village cadre	−0.0242 (0.0256)	−0.0253 (0.0253)	−0.0295 (0.0257)		−0.0328 (0.0255)
Gender	−0.0593 (0.0404)	−0.0605 (0.0396)	−0.0544 (0.0389)		−0.0587 (0.0390)
Age	0.00113 (0.00130)	0.00132 (0.00129)	0.000999 (0.00131)		0.00109 (0.00129)
Labor type	0.0127 (0.0146)	0.0156 (0.0144)	0.00942 (0.0147)		0.0123 (0.0144)
Income	0.0756 *** (0.00730)	0.0727 *** (0.00731)	0.0740 *** (0.00721)		0.0687 *** (0.00734)
Educated level	0.0669 *** (0.0139)	0.0644 *** (0.0139)	0.0630 *** (0.0141)		0.0594 *** (0.0139)
Health status	0.0453 * (0.0233)	0.0536 ** (0.0235)	0.0488 ** (0.0229)		0.0484 ** (0.0234)
Planting scale	−0.00979 *** (0.000982)	−0.00917 *** (0.000975)	−0.00932 *** (0.000959)		−0.00908 *** (0.000970)
Risk appetite	−0.0577 *** (0.0127)	−0.0590 *** (0.0126)	−0.0573 *** (0.0125)		−0.0537 *** (0.0124)
Constant term	0.206 * (0.115)	0.164 (0.115)	0.207 * (0.114)	0.239 *** (0.0226)	0.154 (0.113)
Prob > chi2	0.0000	0.0000	0.0000	0.0000	0.0000
Pseudo R2	0.4657	0.4760	0.4743	0.1411	0.4980

Note: *** p < 0.01, ** p < 0.05, * p < 0.1.

). The variables of financial insurance services, market information services, and technology promotion services were introduced in models 1 to 3. Then, model 4 and model 5 were used for overall regression. Model 4 included the socialized service content without considering the control variables to explore the influences of socialized services on GPE. In contrast, model 5 included all core explanatory and control variables and comprehensively considered the influencing factors of green farmers’ green productivity.

The estimated results of the Tobit regression model are shown in Table 5. From the estimation results of each model, the p values all pass the significance test at the 1% level, indicating that the estimation results of each model are accurate and effective, and the fitting degree is appropriate. From this table, financial and insurance services, market information services, and scientific and technological extension services have positive effects on improving the green productivity of rubber plantation. In addition, from model 5, the income and education level of the growers positively affect the green production efficiency of rubber planting while the planting scale and risk preference negatively affect the green production efficiency of rubber planting. The findings of model 5 indicate that the income, education level, and health status of farmers have a good impact on the green production efficiency of rubber planting, but the planting size and risk preference have a negative impact. The green production efficiency of rubber planting is not significantly affected by whether the farmer had served as a village cadre, gender, age, or part-time job.

4.2. Robustness Test: Propensity Score Matching Method (PSM)

Since some factors can simultaneously affect farmers’ tendency to choose socialized services and green production practices, which will interfere with the causality evaluation of regression models, this work used the propensity score matching method (PSM) in order to obtain more accurate results. PSM makes the observation data as close to the random experimental data as possible by matching resampling to reduce the bias of the observation data, which can effectively solve the selection bias and biased estimation problems. We used the mahalanobis distance matching method, nearest neighbor matching method, radius matching method, and kernel matching method to calculate the matching result (

Table 6. Propensity score matching results.

Service Type	Match Type	Treat	Control	ATT	SE
Financial insurance	before matching	0.601	0.191	0.411	0.017
Services	mahalanobis distance matching	0.601	0.203	0.398 ***	0.019
	nearest neighbor matching	0.557	0.211	0.346 ***	0.019
	radius matching	0.557	0.209	0.348 ***	0.019
	kernel matching	0.557	0.211	0.346 ***	0.019
Average value				0.370
Market information services	before matching	0.601	0.191	0.411	0.017
	mahalanobis distance matching	0.601	0.206	0.395 ***	0.018
	nearest neighbor matching	0.565	0.212	0.354 ***	0.020
	radius matching	0.565	0.211	0.355 ***	0.019
	kernel matching	0.563	0.212	0.350 ***	0.020
Average value				0.373
Technology extension service	before matching	0.601	0.191	0.411	0.017
	mahalanobis distance matching	0.601	0.206	0.395 ***	0.018
	nearest neighbor matching	0.569	0.209	0.359 ***	0.020
	radius matching	0.569	0.211	0.357 ***	0.019
	kernel matching	0.568	0.213	0.354 ***	0.020
Average value				0.375

Note: *** p < 0.01.

).

After matching, the average treatment effect (ATT) value of the adopted service group was significantly positive, indicating that if the endogeneity problem caused by the selection bias is ignored, the promotion effect of financial insurance services, market information services, and technology promotion services on the green productivity of natural rubber will be overestimated. Additionally, from the average ATT value of several matching methods, the average ATT value of technology promotion services is the largest at 0.375, the average ATT value of market information services is 0.373, and the average ATT value of financial insurance services is 0.370. This indicates that the technical extension services of ASSs have a more significant impact on the efficiency of rubber green production than the market information service and financial subsidy service.

To test the PSM effect, it is necessary to discuss whether the model satisfies the common support assumption. The common support domain determines the quality of the sample matching calculation in the model, specifically based on the analysis of the overlap degree between the propensity score intervals of the experimental group and the control group. To visually present the matching quality, we drew a kernel density map of the propensity score values to show the density distribution of the score values in the experimental group and the control group before and after matching. Figure 4 shows the kernel density distribution diagram of the model before and after matching. The density distribution of the matched control group and the treatment group is very similar, and the overlap interval of the propensity score interval is large. As shown in Figure 5, most of the observed values are within the common value range, and there are few matching loss samples, so the samples are representative. Therefore, the model satisfies the common support assumption.

5. Discussion

(1) The green production behavior of rubber growers is affected by socialized services. Green production behavior is not only an economic behavior but also a social behavior. According to the assumption of the “economic man” [51], farmers who want to maximize income from production will increase the input of pesticides while ignoring the environmental pollution. If the long-term ecological benefits brought by green rubber production behavior cannot be reasonably weighed against the current economic costs, farmers will not adopt green production methods. Second, green agricultural production technology is more complicated than traditional production technology. New technologies increase the learning cost of farmers, which leads to a specific threshold in the process of farmers adopting green technology [52]. Finally, under the influence of comparative advantages, farmers will input labor factors into non-agricultural fields with higher output. A reduction in the sustainable input into rubber forests may lead to the input behavior not transitioning in the direction of greening. With the intensification of ASSs, the economic cost of farmers’ adoption of green production behavior progressively decreases, and the degree of adoption of green production behaviors will increase.

(2) Rubber growers receiving social services can obtain a higher rubber output. The financial and insurance services, market information services, and technology extension services in the socialization services of rubber production have a positive impact on the green productivity of rubber farmers. The market information service provides rubber farmers with green production information and market price information. Technology extension services have increased yields for rubber farmers. Financial and insurance services have enhanced the enthusiasm of rubber farmers for production. Through further empirical tests, we found that financial insurance services, market information services, and technical extension services have a positive impact on the green production of rubber farmers. Therefore, the socialization service of rubber production, with the goal of green production, is having an obvious effect. In the future, the government should encourage diversified, multi-level, and multi-type service subjects to participate in the construction of the socialization service system of natural rubber green production and promote the green development of the natural rubber industry.

(3) The grower’s income and degree of education positively influence the green production efficiency of rubber planting. Households with a higher income from rubber cultivation pay greater attention to the socialization services of rubber production and are willing to invest more time and money in new green production technologies, so the level of green productivity is higher. Farmers with a higher education level have a stronger ability to learn and accept new things, and master more rubber production technology, market information, and financial knowledge in socialized services. Therefore, these farmers will have a higher green production efficiency. Consequently, improving the overall quality of rubber-planting farmers and expanding farmers’ access to knowledge are crucial to the transition of rubber production to a sustainable practice. The scale of planting and risk preference have a detrimental impact on the green production efficiency of rubber plantations. It is difficult for farmers with large areas of land to change their production methods in a short period of time, which forces them to adopt traditional farming methods, thus resulting in less efficient green production. Moreover, the higher the risk in adopting a green production mode, the less willing farmers are to adopt the green production technology provided in social services, so the green production efficiency is low.

(4) According to the research, it was found that research on the green production efficiency of natural rubber is still in the initial stage. In the future, with the promotion of the concept of “carbon-neutral” green development in agriculture, more research on the green production of natural rubber will appear in the near future. Future research on the green production efficiency of natural rubber should focus on the following aspects: First, more attention needs to be paid to the impact of rubber producers’ behavior and willingness on green production. Second, continuous tracking of farmers is needed to obtain long-term time series data to study the factors affecting the green production efficiency over a long period of time. Third, with the development of digital technology, cooperative research among scholars around the world should be strengthened in the future to study the factors influencing the green production efficiency of cash crops such as rubber from different perspectives.

6. Conclusions

We focused on whether agricultural socialized services can improve the green productivity of rubber planting. The influence mechanism of ASSs on the green productivity of rubber at the theoretical level was analyzed. Utilizing the survey data of 552 rubber growers in 6 cities and 18 townships in Hainan Province, an SBM model of the undesired outputs was constructed to measure the green productivity of natural rubber. Further analysis of the impact of social services on the green productivity of rubber was carried out. (1) The average green productivity of natural rubber in Hainan Province is only 0.41. The green productivity level of Hainan rubber planting is relatively low, and there is still much room for improvement. (2) From the empirical analysis, ASSs have a positive effect on improving the green productivity of rubber planting. Professional services in the field of social services, such as financial insurance services, market information services, and technology promotion services, have a positive impact on the improvement of rubber planters’ green productivity. Moreover, the effect of technology promotion services in socialized services is greater than that of market information services and financial insurance services. (3) Family planting income, health status, and education level positively impact rubber planters’ green productivity while planting scale and risk preference negatively affect the green production efficiency of rubber planting. Both internal and external factors affect green rubber production. Future research should not only focus on internal factors such as farmer quality, management ability, information reception ability, and the technology level but also focus on macroeconomics, policy factors, and organizational structure to propose corresponding strategies to promote green and sustainable development of the natural rubber industry.