1 Introduction

The importance of greenhouse gas (GHG) emission reduction is increasing daily with the changing climate. At the 2021 Leaders’ Summit on Climate, the Chinese government announced that China’s GHG emissions would be strictly controlled and reaffirmed China’s pledges to peak its emissions by 2030 and reach carbon neutrality by 2060. This announcement could help significantly slow global warming. Agricultural production is a prominent source of GHGs, and its total emissions account for 17% of the overall emissions in China but only 7% in the US and 11% globally [1]. Therefore, it is more important for China to reduce agricultural GHG emissions than for other countries.

However, China is a large agricultural producer as well as a developing country and is highly dependent on agriculture; GHG emissions reduction has been a major challenge in Chinese agricultural development [2]. In recent years, GHG emissions caused by agricultural chemical products, fossil fuel used by machinery, paddy irrigation, and livestock breeding have remained at a high level [3]; furthermore, the rapid development of Chinese agriculture is a leading cause of the acceleration in global warming. Considering China’s pledges to GHG reduction, it is important to pay attention to agricultural GHG emission reduction in China. Therefore, this study aims to calculate agricultural GHG emissions and use difference-in-differences (DID) analysis to estimate the effect of agricultural GHG emissions reduction policies.

2 Literature review

2.1 Greenhouse gas emission sources in agriculture

Traditional farming and livestock breeding produce a large amount of GHGs in their production processes. According to data from the National Development and Reform Commission [4], the use of chemical fertilizers and their volatilization in the field, as well as the management of livestock wastes, account for 39.1% and 6.5% of all GHG emissions from agriculture, respectively (Fig. 1). Considering agricultural production activities in China, nine GHG emissions are included in Table 1 as sources of agricultural GHG emissions.

Fig. 1
figure 1

Overview of China’s greenhouse gas (GHG) emissions (2016)

Full size image
Table 1 Emission factors of agricultural sources
Full size table

2.2 Agricultural GHG emissions reduction approach

Energy conservation and emission reduction actions have been advanced worldwide, and agriculture is one of the major sectors of GHG emission reduction to which these actions have been applied [5]. Among the existing studies on agricultural GHG emissions reduction, two approaches have been taken to reduce GHG emissions. One approach is the policy method. For example, the treatment effect of the Japanese Farmland, Water and Environmental Conservation Improvement Scheme (CIS) was measured based on Shiga Prefecture [6] and concluded that CIS is instrumental in conserving farmland and reducing agricultural GHG emissions in Shiga Prefecture. Moreover, payment for environmental/ecosystem service (PES), also a policy method, is commonly adopted by countries around the world for GHG emissions reduction and energy savings [7, 8]. It provides payments from the government to users of low-carbon environmental services and income compensation to people for adopting low-carbon environmental behaviors. The Japanese government adopted PES while promoting the cooperation system between the cultivation and livestock sectors [9]; in Japan, this method is called the 3-R (Recycling, Resource, Repeat) model.

The other approach is energy supply. Existing studies have indicated that higher agricultural economic growth and higher agricultural energy consumption might increase agricultural GHG emissions, but a sufficient energy supply for the development of the agricultural economy can reduce agricultural GHG emissions [10, 11]. Moreover, the ML index has been used to analyze the dynamic changes in China’s agricultural carbon emission performance (ACEP) and found that China’s ACEP increased by 1.5% from 2009 to 2019 [12].

Research on agricultural GHG emissions has been conducted on government actions, sources of carbon emissions, and the relationship between agricultural carbon emissions and economic growth. However, there is still a lack of research on the effect of agricultural GHG emissions reduction policies in China. This paper concludes with a theoretical mechanism of how policies can reduce agricultural GHG emissions, contributing to a better understanding of agricultural GHG emissions reduction policies in China.

2.3 Agricultural GHG emissions reduction policies in China

Over the past 40 years of opening up in China, the agricultural economy has rapidly grown, and agriculture must continue to develop in China. However, the growth of agricultural production will lead to greater energy consumption and more GHG emissions. Moreover, some environmental problems have increasingly appeared, such as land pollution, soil hardening, and pollution due to chemical fertilizers [13]. The middle and lower reaches of the Yangtze River Basin are main production areas of grain in China and play an important role in ensuring food security in China. This region includes provinces such as Hubei, Hunan, Zhejiang, and Jiangsu. However, this region is also a primary source of agricultural GHG emissions in China. In addition, its environmental problems have become more serious, and some issues, such as the overuse of agricultural resources, soil quality decline, and water pollution, urgently need to be solved [14].

In 2013, to better achieve energy conservation goals, reduce GHG emissions in agriculture and address environmental pollution in the middle and lower basins of the Yangtze River, the Ministry of Agriculture announced a policy to decrease agricultural GHG; Zhejiang Province was chosen to be the first low-carbon agriculture pilot province in 2015. According to the statement of the Zhejiang government in 2015, specialized GHG emissions reduction policies were conducted in Zhejiang Province by reducing the use of chemical fertilizers in agriculture, improving the efficiency of using chemical fertilizers and pesticides, popularizing low-emission agricultural machinery, and realizing resource utilization and harmless treatment of livestock and poultry waste. These policies, implemented in Zhejiang Province, are designed to reduce agricultural GHG emissions and establish a low-carbon circular agricultural production system (Fig. 2).

Fig. 2
figure 2

The low-carbon circular agriculture model in Zhejiang Province

Full size image

Financial funds are used to subsidize and reward low-carbon agricultural production behaviors such as the collection and treatment of manure, the use of energy-saving agricultural machinery, the application of manure to fields, and the recycling of plastic agricultural film. Meanwhile, publicity that targets farmers is strengthened. Various methods are widely used to enhance policy publicity, such as special lectures and visits to farmers in the countryside by technical and subsidy personnel, which popularize the concept of low-carbon agriculture and create a favorable atmosphere regarding public opinion.

In addition to providing a series of subsidies and support measures, the government reformed the agricultural production system on nationally owned farms and model farms through administrative orders and replaced agricultural machinery that does not meet carbon emission standards. Excluding national-owned farms, the government ordered private farms to rectify their production behaviors that did not meet low-carbon agricultural production standards and gave exhortations and inducements to individual farmers to help agricultural producers change to the low-carbon and circular agricultural production system. Specific policy measures are shown in Table 2.

Table 2 Supportive measures and regulation mechanisms of pilot policies
Full size table

The sources of agricultural GHG emissions mainly consist of agricultural production behaviors such as agricultural irrigation, the use of chemical fertilizers, agricultural machinery, livestock production, animal manure management, and the rice-growing process [15]. In the agricultural production process, the reduction of GHG emissions is highly related to the sources of agricultural production behaviors. In other words, through the implementation of policies, production behaviors can be changed, and green production behaviors can be introduced to reduce GHG emissions during agricultural production efficiency. Other policies, such as China’s carbon trading pilot, have been proven to decrease regional carbon emissions [16]. Therefore, the implementation of the policies in Zhejiang Province can effectively reduce agricultural GHG emissions.

Hypothesis

Agricultural GHG emissions reduction policies which are implemented in Zhejiang Province, can change farmers’ production behaviors and reduce agricultural GHG emissions (Fig. 3).

Fig. 3
figure 3

Hypothesis of the effects of policies on agricultural GHG emissions

Full size image

The effect of pilot policies will have a positive impact on farmers’ production behaviors, according to government measures. The policies use subsidies and regulation measures to achieve overall energy conservation and agricultural GHG emission reduction by reducing the use of common chemical fertilizers, promoting the popularization of energy-efficient agricultural machinery, encouraging the recycling of plastic agricultural film, and reducing direct emissions from livestock breeding.

3 Methods and data

3.1 Calculation of agricultural GHG emissions and data

Considering that there are many influencing factors that affect the volume of agricultural carbon emissions in a certain region, a scientific and reasonable evaluation of the treatment effect of the pilot policies needs to eliminate the influence of other factors.

Since the reduction policies in Zhejiang Province mainly focus on reducing agricultural GHG emissions by improving the cooperation of cultivation and livestock, the popularization of energy-saving agricultural machinery, the recycling of agricultural plastic film, and the reduction of chemical fertilizers, these policies can affect emission sources and farmers’ production behaviors, such as the use of chemical fertilizers and agricultural machinery. Due to data availability and considering the relationship between policies and agricultural GHG emissions, we estimate agricultural GHG emissions based on sources or activities in Table 3.

Table 3 Calculation method of agricultural GHG emissions and emission coefficients
Full size table

This paper uses the equation of agricultural GHG emissions developed by the IPCC [17]:

$$E=\Sigma {E}_{i}=\Sigma {T}_{i}\times {e}_{i},$$
(1)

\(E\) is the total amount of agricultural GHG emissions, \({E}_{i}\) are the emissions of various GHG emission sources, \(i\) is the emission source type, \({T}_{i}\) are the specific values of GHG emission sources, and \({e}_{i}\) are the coefficients of various GHG emission sources. This study uses the emissions coefficient from the IPCC and Japan’s Ministry of the Environment [18]. The IPCC recommends using default emission factors (Tier 1) for countries with limited data, such as most developing countries [19]. Considering the lack of statistics on agricultural GHG emissions in China, it is recommended that the Tier 1 method for emission factors be used in this study.

We used data for the period 2011–2020 for the empirical analysis. The data were taken from the 2012–2021 Zhejiang agricultural statistical yearbooks and Hubei agricultural statistical yearbooks, and the agricultural GHG emissions in each city were calculated for each year from 2011 to 2020. In addition, Table 3 lists all of the data used in the measurement of agricultural GHG emissions. Table 4 provides the cities of data sources in this study.

Table 4 Data sources
Full size table

3.2 DID analysis

DID analysis is a common method used to evaluate policy effects and is now widely used in various studies of economics and management. The impact of energy policies on household energy consumption in the UK was examined using DID analysis [20]. DID analysis was also used to explore the impact of China’s carbon trading rights pilot policy on the amount of total GHG emissions by specific economic sectors [21]. Therefore, referring to relevant policy effect research articles, combined with the actual situation of Zhejiang Province’s implementation of agricultural GHG reduction policies, this study also uses DID analysis to conduct research.

In 2015, Zhejiang Province became the first low-carbon agricultural pilot province in China. Cities in Zhejiang Province are taken as the treatment group, and cities in Hubei Province that are not participating in agricultural GHG reduction policies are taken as the control group.

In the selection of control and treatment groups, the DID analysis requires the control group to be comparable to the treatment group regarding relevant characteristics. Hubei Province has topographical and climatic conditions similar to those of Zhejiang Province. Like Zhejiang Province, Hubei Province consists of mainly paddy field agriculture; it also belongs to the middle and lower reaches of the Yangtze River basin; therefore, the natural agricultural conditions and production patterns of the two provinces are relatively consistent. Although Zhejiang Province and Hubei Province have differences in economy and education levels, the province to be taken as the control group can only be selected in the middle and lower reaches of the Yangtze River basin, considering that the purpose of the pilot policies is to reduce carbon emissions and to protect the agricultural ecological environment in this region. The other provinces in the region, such as Hunan and Jiangsu Provinces, have climatic and geographic conditions and patterns of agricultural production that are completely different from those of Zhejiang. For example, the topography of Hunan Province is dominated by mountains and hills, while Zhejiang Province is dominated by an agricultural system of plains and paddy fields. At the same time, the geo-climatic conditions and patterns of agricultural production in the southern part of Jiangsu Province are similar to those of Zhejiang Province, but the share of wheat and livestock in the agricultural production of north-central Jiangsu Province is quite high, which is significantly different from that of Zhejiang Province. Considering that agricultural production in Hubei Province, like Zhejiang Province, is dominated by paddy fields and has similar geo-climatic conditions, Hubei Province was selected as the control group in this study.

Based on the panel data of the 20 cities in the treatment group (Zhejiang Province) and control group (Hubei Province) from 2011 to 2020, the agricultural GHG emissions of each city were estimated. To sort out the treatment effects of the pilot policies, this research adopted DID analysis to assess the effect of the pilot policies on GHG emissions in the two provinces. Since the policies started in 2015, the policy’s effect would only appear in 2016; hence, the years from 2011 to 2015 were selected as the pre-policy period, and the years from 2016 to 2020 were selected as the post-policy period. The regression equation was constructed as follows:

$${lnE}_{it}={\beta }_{0}+{\beta }_{1}{T}_{it}+{\beta }_{2}{R}_{it}+{\beta }_{3}{T}_{it}{R}_{it}+{\varepsilon }_{it},$$
(2)

\({lnE}_{it}\) are the log-transform of agricultural GHG emissions; \(i\) and \(t\) denote the city number and the t-th year, respectively. \({\beta }_{1}\) denotes the coefficient of the time effect on agricultural GHG emissions, and \({T}_{it}\) is a time dummy variable that is assigned a value of 0 before the implementation of the pilot policies and a value of 1 after the implementation. \({\beta }_{2}\) denotes the coefficient of province dummy variables on agricultural GHG emissions, and \({R}_{it}\) is a province dummy variable that is assigned a value of 1 if the city is in Zhejiang Province; otherwise, it is 0. \({\beta }_{3}\) denotes the coefficient of interaction between time and province on agricultural GHG emissions. \({T}_{it}{R}_{it}\) is the interaction term of the time dummy variable and the province dummy variable, and \({\varepsilon }_{it}\) is the random perturbation. Therefore, the DID analysis first uses the dummy variable \({R}_{it}\) to divide the treatment group that received the policy impact and the control group that was not affected by the policies; it then uses the dummy variable \({T}_{it}\) to capture the change in \({E}_{it}\) in the treatment and control groups and finally reflects the trend difference between the two changes.

The dummy variables can be set to divide the sample into four groups, namely, Zhejiang Province before the policy implementation (\({T}_{it}\) = 0, \({R}_{it}\) = 1), Zhejiang Province after the policy implementation (\({T}_{it}\) = 1, \({R}_{it}\) = 1), Hubei Province before the policy implementation (\({T}_{it}\) = 0, \({R}_{it}\) = 0), and Hubei Province after the policy implementation (\({T}_{it}\) = 1, \({R}_{it}\) = 0). The specific differences are shown in Table 5. According to the results in Table 5, the treatment effect of the policies in the pilot province is \({\beta }_{3}\), and the sign of \({\beta }_{3}\) should be significantly negative if the pilot policies helped reduce the agricultural GHG emissions in Zhejiang Province. However, only the effects of time \({T}_{it}\), province \({R}_{it}\) and the time-province interaction term \({T}_{it}{R}_{it}\) are considered in regression Eq. (2) above. Agricultural GHG emissions are significantly influenced by rural population size, per capita wealth, and agricultural technology level [22].

Table 5 The estimation principle of the DID analysis
Full size table

To improve the accuracy of the model estimation results, the following variables are introduced into the DID analysis as control variables in this study:

  1. (1)

    Rural employment: The impact of rural population size on agricultural GHG emissions is represented using rural employment.

  2. (2)

    Net income per capita: The impact of per capita wealth on agricultural GHG emissions is represented using the net income per capita of the agricultural population.

  3. (3)

    Average power of agricultural machinery: The impact of the level of agricultural technology on agricultural GHG emissions is represented using the average power of agricultural machinery.

The final DID regression analysis is constructed as follows:

$${lnE}_{it}={\beta }_{0}+{\beta }_{1}{T}_{it}+{\beta }_{2}{R}_{it}+{\beta }_{3}{T}_{it}{R}_{it}+{\beta }_{4}{lnEmployment}_{it}+{\beta }_{5}ln{Income}_{it}+{\beta }_{6}{lnMachinery}_{it}+{\varepsilon }_{it}$$
(3)

3.3 Applicability test of the DID analysis

Reasonable treatment and control groups are required to satisfy a series of prerequisites and assumptions for the DID analysis. The first is to satisfy the assumption of randomness; that is, the pilot areas for policy implementation are randomly selected. The second is that the agricultural GHG emissions of Zhejiang and Hubei Provinces should have a common trend before the implementation of the pilot policies.

3.3.1 Random selection of the pilot policies province assumption

According to the background of the pilot policies, the purposes of these low-carbon and circular agriculture pilot policies are to simultaneously promote cooperation between cultivation and livestock in the main rice production areas around the Middle and Lower Yangtze River Basin, reduce agricultural GHG emissions, and popularize the circular agriculture production system. The selection of pilot province can be considered random as long as the related indicators of agricultural development in Zhejiang Province (e.g., agricultural output, rural employment) and the dependent variable in this study, agricultural GHG emissions, are not determinants of Zhejiang Province becoming a pilot province.

To verify the randomness of the pilot province selection, this study collected pre-policy implementation data related to agricultural GHG emissions, rural employment, and the gross output value of agricultural practices from major provinces in the Middle and Lower Yangtze River Basin.

As shown in Table 6, Zhejiang Province does not have the largest rural employment among the major provinces in the middle and lower Yangtze River Basin; the output value of agricultural practices and agricultural GHG emissions are neither the highest nor the lowest. Therefore, it is clear that the selection of the pilot province was not based on these variables. Consequently, the selection of the pilot province is considered to satisfy the randomness assumption.

Table 6 Agricultural GHG emissions and related indicators of agricultural development of major provinces in the middle and low Yangtze River Basin, 2011–2015
Full size table

3.3.2 Common trend assumption

The common trend assumption is intended to demonstrate whether the trends of agricultural GHG emissions in the control and treatment groups were similar before the implementation of the pilot policies. To assess the heterogeneity in emission trends across the cities, Figs. 4 and 5 show the trends of cities in the two groups. As shown in Fig. 6, the average agricultural GHG emissions by the city of the two provinces do differ after the implementation of the policies; however, their trends are similar before the implementation. Furthermore, we used the first-order difference value of agricultural GHG emissions of each city from 2011 to 2015 in the two groups as the dependent variable and used R (Cities in Zhejiang = 1; Cities in Hubei = 0) as the only independent variable to test the trend of agricultural GHG emissions of each city before 2016.

Fig. 4
figure 4

Changes in agricultural GHG emissions of treatment group by city (2011–2020)

Full size image
Fig. 5
figure 5

Changes in agricultural GHG emissions of the control group by city (2011–2020)

Full size image
Fig. 6
figure 6

Changes in average agricultural GHG emissions in the 2 Groups (2011–2020). Note: GHG emissions in Figs. 4, 5, and 6 were calculated by the author based on the data of each emission source

Full size image

From the results of the fixed effects model in Table 7, the effect of R on the first-order difference value of agricultural GHG emissions of each city is negative but not statistically significant. Therefore, it can be concluded that this study's treatment and control groups are appropriately selected, and the common trend assumption of the DID analysis is valid.

Table 7 Regression results of the first-order difference value of agricultural GHG emissions before 2016 based on the fixed effects model
Full size table

3.3.3 The pilot policies in Zhejiang Province have no direct effect on the agricultural GHG emissions of Hubei Province

The effect of the pilot policies in Zhejiang Province on agricultural GHG emissions in Hubei Province may be reflected in the fact that Hubei Province has also started to implement similar policies to reduce agricultural GHG emissions inspired and motivated by Zhejiang Province. However, the in-depth analysis indicated that the pilot policies in Zhejiang Province are unlikely to affect the agricultural GHG emissions of Hubei Province.

First, although the agricultural models of Zhejiang and Hubei Provinces are both paddy field agriculture, the farmers’ income level and mechanization level in the two provinces differ significantly; therefore, it is difficult for Hubei Province to implement similar policies at the same time. Second, Zhejiang Province is also located in the Yangtze River Delta region, which has national solid financial support and location advantages compared to Hubei Province; therefore, the agricultural production system of Zhejiang Province can be redesigned rapidly. However, the current conditions of Hubei Province make it challenging to achieve a comprehensive transformation of low-carbon agricultural production in a short period.

We also conducted a statistical test on whether the pilot policies in Zhejiang Province directly affected the agricultural GHG emissions in Hubei Province by using panel data from 2011 to 2020 for each city in Hubei Province. We used agricultural GHG emissions in Hubei Province as the dependent variable, with a time dummy variable T representing the pilot policies (2011–2015 = 0; 2016–2020 = 1) and controlled for a total of three control variables of rural employment, net income per capita, and average power of agricultural machinery in Hubei Province. The fixed-effects regression results are shown in Table 8. The regression results show that the pilot policies did not significantly affect the agricultural GHG emissions in Hubei Province.

Table 8 Regression results of agricultural GHG emissions in Hubei Province based on the fixed effects model
Full size table

4 Results

4.1 Statistical results

In this study, the following were selected as indicators to calculate the agricultural GHG emission \({E}_{it}\) in 10 cities of Zhejiang Province and 10 cities of Hubei Province during 2011–2020; the consumption of chemical fertilizers used; agricultural plastic film; the annual number of hogs, cattle, and buffalo reared; effective irrigated area; and used agricultural diesel oil (Tables 9, 10).

Table 9 Definitions of variables
Full size table
Table 10 Descriptive statistics of the variables in Eq. (3)
Full size table

During calculation, the calculation results are statistically converted to CO2 to facilitate the aggregation of the GHG emissions results. Rural employment (\(\mathit{ln} \, Employment\)), net income per capita (\(\mathit{ln} \, Income\)), and average power of agricultural machinery (\(\mathit{ln} \, Machinery\)) are also included in the DID analysis as control variables.

Hubei Province is the main agricultural production area in the middle and lower reaches of the Yangtze River, with a large, irrigated area and a large scale of agricultural and livestock production. The agricultural model of Zhejiang Province mainly focuses on paddy fields while simultaneously developing agricultural product distribution and trade. The scale of animal husbandry and livestock production in Zhejiang Province is lower than that in Hubei Province. Simultaneously, Zhejiang Province has been vigorously developing high-tech agriculture, tourism agriculture, and other modern agricultural forms to enhance the value of its agricultural products.

4.1.1 Period 2011–2015

As shown in Fig. 6, the overall trend of agricultural GHG emissions in Zhejiang Province and Hubei Province from 2011 to the end of 2015 indicated an upward trend. The specific reason for this is that the Chinese government promulgated a series of agricultural policies in approximately 2010, which greatly improved farmers’ enthusiasm for agricultural production, resulting in more capital, machinery, land, and other factors of production being invested in agricultural production. The high consumption of agricultural materials, frequent use of land, and participation in livestock farming activities led to the increasing trend of agriculture GHG emissions.

More specifically, Zhejiang and Hubei provinces responded to the national policies on the countryside, agricultural production, and farmers during the 2011 to 2015 period by increasing investment in agricultural production in addition to promoting urbanization. In both provinces, used fertilizer, used agricultural plastic film, effective irrigated area, used diesel oil, and average power of agricultural machinery generally showed upward trends during the 2011–2015 period; similarly, livestock and poultry output remained at a high level.

Regarding rural employment, Zhejiang Province showed a decreasing trend, while Hubei Province showed an overall decrease but a high and low fluctuation between 2011 and 2015. The net income per capita of farmers in the two provinces has been increasing every year, which shows that the two provinces were able to increase farmers’ incomes and agricultural productivity by increasing investment in agriculture and upgrading agricultural technology while ensuring the urbanization process. With high agricultural investment, agricultural GHG emissions showed an upward trend in both provinces during this period.

4.1.2 Period 2016–2020

As seen in Fig. 5, the agricultural GHG emissions of Zhejiang Province have been decreasing every year since 2016 compared with the emissions of Hubei Province, which still showed a trend of slow growth until 2020. Moreover, Fig. 6 shows that the average agricultural GHG emissions in Zhejiang Province from 2011 to 2015 were 66.404 million tons and 58.224 million tons from 2016 to 2020; therefore, the GHG emissions after the implementation of the policies were reduced by 8.18 million tons. The average agricultural GHG emissions in Hubei Province were 73.842 million tons during 2011–2015 and 81.95 million tons during 2016–2020, showing an increase of 8.108 million tons after 2015.

Regarding fertilizers and pesticides, Zhejiang Province has promoted farmers’ participation in agricultural waste composting and recycling by developing low-carbon and circular agriculture and introducing a series of subsidy policies and relevant laws and regulations; these efforts have effectively encouraged the operation system of composting and agricultural waste and the use of organic fertilizers. From 2016 to 2020, the use of chemical fertilizers and pesticides in Zhejiang Province was rapidly reduced, and the environmental problems caused by fertilizers and pesticides significantly improved. Although Hubei Province is not a pilot area for the policies, the government of Hubei Province has taken a series of corresponding measures to implement the low-carbon and circular agriculture practice as it has become one of the main strategies of agricultural development in China. Therefore, during 2016–2020, the use of fertilizers and pesticides in Hubei Province declined each year; however, the rate of decline was lower than that in Zhejiang Province, and the growth of agricultural GHG emissions was not significantly curbed.

Regarding livestock and poultry, due to the sporadic outbreak of swine and chicken fevers, livestock and poultry breeding in Hubei and Zhejiang Provinces have been affected to a significant extent. According to relevant statistics, livestock and poultry stocks in Zhejiang Province are decreasing yearly, while Hubei Province still shows a trend of fluctuating changes. At the same time, Zhejiang Province has successfully implemented low-carbon agriculture practices and a cooperative system between cultivation and livestock, effectively controlling the agricultural GHG emissions from livestock and poultry breeding activities.

In the paddy field sector, the use of agricultural plastic films in the two provinces has been increasing since 2010, mainly due to the desire of the two provinces to promote the use of agricultural plastic films to increase growth and farmers’ incomes. As a convenient weed suppression and yield enhancement technology, agricultural plastic film plays an important role in China’s agricultural production and is not only less costly but can serve as a response to natural disasters [23].

With the advancement of urbanization, rural employment in the two provinces shows a decreasing trend. Furthermore, the two provinces have adopted different agricultural development measures on top of ensuring a certain degree of effectively irrigated area. Zhejiang Province specifically focuses on agricultural processing and trade, the enhancement of the added value of agricultural products, the production of high-quality agricultural raw materials, and the building of agricultural processing industrial parks and trading platforms. These areas of focus ensure the rapid transformation of agricultural production systems by reducing the use of agricultural diesel oil, popularizing a large number of energy-saving agricultural machinery, and reducing GHG emissions in agricultural production processes. Conversely, Hubei Province has adopted an approach that consolidates the main grain and livestock production and has continued to increase investment in agricultural production processes. Although rural employment and agricultural production in Hubei Province have decreased somewhat with urbanization, they remain at a high level overall and the net income per capita of rural areas has significantly improved.

4.2 Analysis of the pilot policy treatment effect

The main content of this study is to test the effect of reduction policies on agricultural GHG emissions through DID analysis. Considering the possible cross-sectional heteroskedasticity and serial correlation problem of the sample data, this study uses feasible generalized least squares (FGLS) to evaluate parameters and takes FGLS as the basic model, which is robust and feasible for this sample. Before using the FGLS model for parameter estimation, the model form was determined by taking a fixed-effects model or a random-effects model by the Horsman test; the P value of the Hausman test was 0.0076. Since the P value was less than 0.05, the fixed-effects model and the pooled OLS regression were also selected as the analysis models in this study. Comparing the results obtained from Models 1, 2, and 3 revealed that there was no significant change in the significance and sign of the coefficient on the policies treatment effect variable, T * R, indicating that the estimation results of FGLS were robust (Table 11).

Table 11 Policy treatment effect on agricultural GHG emissions based on DID analysis
Full size table

The results of the regression analysis (Model 1) show that the estimated value of the policy treatment effect is -0.049 and passes the 5% significance test, indicating that the pilot policies have a significant reduction effect on agricultural GHG emissions in Zhejiang Province. More specifically, after considering variables such as income level, rural employment, and average power of agricultural machinery, the agricultural GHG emissions in Zhejiang Province decreased after 2015, driven by the pilot policies, when compared to Hubei Province. Therefore, it can be concluded that the treatment effect of the pilot policies was found to reduce agricultural GHG emissions after 2015, which effectively promoted energy savings and emission reduction in Zhejiang Province within the agricultural sector. The regression results also show that, in addition to policy factors, rural employment, income level and the average power of agricultural machinery are also important influencing factors of agricultural GHG emissions, with estimated coefficients of 0.319, 0.063 and 0.244 and significance at the 1%, 5% and 1% levels, respectively. This indicates that the expansion of the agricultural production scale, the increase in household income and mechanization will increase agricultural GHG emissions, which in turn will have a hindering effect on the reduction effects of the pilot policies.

4.3 Treatment effect of policies among different cities

When estimating the treatment effect on agricultural GHG emissions, the results in Table 11 only show the treatment effect on all cities in the two provinces. However, the effect differs among the cities due to the level of agricultural development. This study takes the top five cities in the two provinces in terms of the gross output value of agricultural production as the high agricultural performance group and the last five cities as the low agricultural performance group. The basic model was used to estimate the effect of the policies on different cities.

From the results presented in Table 12, the treatment effect is not the same between the two groups. The results show that the pilot policies can reduce agricultural GHG reduction in the high agricultural performance group more efficiently than in the low agricultural performance group. Cities in the high agricultural performance group produce more agricultural carbon emissions due to their higher agricultural output. By implementing the pilot policies, emission factors in the high-performance group, such as the use of fertilizer, can be reduced more than in the low-performance group. Therefore, the effect of implementing pilot policies in these areas will be more efficient. According to Table 12, the results also show that rural employment, income level, and agricultural machinery are factors that are highly related to agricultural GHG emissions in both groups.

Table 12 Policy treatment effect on different groups of cities
Full size table

4.4 Robustness check

To ensure the reliability of the effect of the pilot policies on agricultural GHG emissions, in this part, a placebo test was used to conduct a robustness test. In 2013, the Chinese government announced a policy to reduce agricultural GHG, and the Zhejiang government implemented pilot policies in 2015. Therefore, we construct a dummy variable for implementing the policies in 2013 and use the basic model for empirical analysis. If the effect of the pilot period pilot on agricultural GHG emissions is zero or the result is not significant, it means that the pilot policies started in 2015 can reduce GHG emissions in Zhejiang Province. The test results are shown in Table 13.

Table 13 Placebo test results
Full size table

The interaction terms in Columns (1) and (2) are constructed with 2013 as the policy implementation period, and Column (3) is the result of Table 11 (Model 1), which is based on 2015 as the policy implementation period. In addition, control variables were only added to Column (2). Empirical results found that the interaction terms of the virtual policies as well as the core variable in Columns (1) and (2) were near zero and not significant, and that the coefficient was zero and negligible, so the empirical results in Model 1 are robust and Zhejiang Province started pilot policies in 2015 that can effectively reduce agricultural GHG emissions in the pilot area.

5 Discussion and conclusions

This study examines the impact of agricultural GHG reduction pilot policies in Zhejiang Province and compares the change in agricultural GHG emissions in Zhejiang and Hubei Provinces using DID analysis. Since Zhejiang Province is the first pilot province of low-carbon agriculture in China, studying the effect of the pilot policies is important to reducing agricultural GHG emissions. This study provides theoretical and methodological references for future research on low-carbon agricultural policies in China.

This study found the following conclusions. First, during the period of 2011–2015, influenced by the relevant national policies to encourage agricultural production, the agricultural GHG emissions of each city in the two provinces showed a trend of yearly increases. After 2015, Zhejiang Province took the lead in implementing pilot policies and built a low-carbon and circular agricultural production system and a large number of state-owned demonstration farms to reduce agricultural GHG emissions. Simultaneously, Zhejiang Province has been accelerating its transformation of agricultural production and promoting energy-efficient agricultural machinery to produce high-quality agricultural products while reducing agricultural GHG emissions. In this study, we found that the pilot policies targeting agricultural GHG emissions in Zhejiang Province have made remarkable achievements in reducing emissions. Pilot policies have become a key driving force in accelerating the redesign of the agricultural production system and building low-carbon and circular agriculture.

Second, we analyzed the treatment effect of the pilot policies from the perspective of agricultural energy conservation and emission reduction. It was found that the pilot policies accelerated the reduction of agricultural GHG emissions in Zhejiang Province through massive subsidies and the construction of state-owned farms. Furthermore, the pilot policies are a leading motive behind why Zhejiang Province has achieved energy saving and emission reduction in agriculture.

Third, agricultural GHG emissions, as one of the important outcome indicators of low-carbon agriculture construction, are a key performance indicator (KPI) about which government departments of agriculture are increasingly concerned. Therefore, in the short term, behaviors such as the blind pursuit of KPIs and an unreasonable reduction of agricultural GHG emissions by local governments can be implemented to quickly achieve policy goals. In the short term, agricultural GHG emissions can be effectively reduced by increasing national financial investment and administrative orders; however, excessive administrative orders and forced transformation of agricultural systems are likely to harm farmers’ interests in the process of policy implementation. Therefore, it is crucial to combine incentives such as subsidies with administrative orders to establish a long-term mechanism for reducing agricultural GHG emissions based on ensuring farmers’ interests.

Fourth, according to the empirical results of this paper, the effect of agricultural GHG emissions reduction policies in Zhejiang Province does differ between cities with different levels of agricultural development. Therefore, it is very important to narrow the gap in policy effects between the different cities and gradually improve the reduction policies before they are fully launched in the middle and lower reaches of the Yangtze River Basin.

Our findings have several policy implications for promoting low-carbon agricultural development. First, the environmental damage caused by agricultural carbon emissions is not only in the middle and lower reaches of the Yangtze River Basin. Therefore, it is suggested that reduction policies should be implemented in other regions across China. Second, in the long run, the continuous reduction of agricultural GHG emissions is necessary for achieving carbon neutrality by 2060. Therefore, the Chinese government should concentrate on building a long-term mechanism for reducing agricultural GHG emissions and ensuring farmers’ interests.

Similarly, considering the differences in the economic conditions of various developing countries, making good use of these subsidies and regulation methods can lead to the gradual implementation of agricultural GHG emissions reduction policies at the global scale. The pilot policies implemented in Zhejiang Province can serve as an example of low-carbon agricultural policies in other developing countries where paddy field agriculture is the mainstay.

This study exposes some limits and raises a number of issues for possible further research. First, the marginal abatement cost for each policy cannot be estimated due to the lack of related statistics in China. The GHG emissions calculated in each city are smaller than the actual emissions because some carbon emission source data and coefficients are not available. Second, this research can only examine the impact of pilot policies related to agricultural GHG emissions but cannot assess the specific impact of the policies on farmers during their implementation. With a breakthrough in data and methods, the relevant shortcomings of this research are to be explored in depth by subsequent studies.