The Effect of Agriculture Insurance on Agricultural Carbon Emissions in China: The Mediation Role of Low-Carbon Technology Innovation

Abstract

Global greenhouse gas emissions are increasing, with carbon dioxide being the most prominent. It is urgent to address and resolve the carbon emissions problem. This study investigates the mediating mechanism of agricultural insurance and low-carbon technology innovation on agricultural carbon emissions. We employed a two-way fixed effect panel model with data from 30 provinces in China from 2001–2019 to validate our hypotheses. The results demonstrate that (1) agricultural insurance can play an effective role in reducing agricultural carbon emissions, and (2) an indirect effect of agricultural insurance development on agricultural carbon emissions through low-carbon technology innovation exists. These results indicate that agricultural insurance could suppress agricultural carbon emissions indirectly through low-carbon technology innovation, thus preventing the acceleration of the greenhouse effect. This study further analyzed regional differences and discovered that the suppression effect of agricultural insurance on agricultural carbon emissions is more significant in the eastern regions and non-main grain-producing areas of China. Therefore, the analysis implies that promoting the development of agricultural insurance to encourage low-carbon technology innovation is crucial to accelerate the process of “carbon peak and neutrality”, especially for the eastern regions and non-main grain-producing areas of China.

1. Introduction

The increase in greenhouse gas emissions, such as carbon emissions, leads to unusual global temperatures. It has been a topic discussed by various global economies for a long time. In China, carbon emissions have increased more than three times from 3.214 billion tons in 2000 to 10.435 billion tons in 2019. In particular, carbon emissions from the agricultural sector are one of the major causes of China’s increasing greenhouse gas emissions. The carbon emissions of China’s agricultural sector take up 17% of the country’s total and are continuing to rise at an average rate of 1.46% [1,2]. China issued several national policies to reduce the intensity of greenhouse gas emissions from agriculture since the 13th five-year plan proposed in the year 2016 [3]. These policies encourage Chinese farmers to stop using the traditional method of planting and switch to a low-carbon method.

When farmers tend to adopt alternative production methods, such as the low-carbon method of planting, they will face a variety of risks such as yield reduction and income losses [4]. Therefore, farmers’ willingness to switch agricultural production methods relates to their attitudes toward risk-taking and their risk preferences [5,6]. According to Visser et al. [7], agricultural insurance would impact technology uptake if farmers’ risk preferences tended toward loss aversion. In China, agricultural insurance is treated as a tool to serve the development of agricultural development and to provide protection for agricultural risk management [8]. Taking into consideration the fact that agriculture insurance is one method to manage risks by safeguarding farmers’ economic losses beyond their control, an effective way to encourage a low-carbon method of planting for farmers is to promote agricultural insurance [9].

Previous studies denoted that insurance market development spurs CO₂ emissions since it stimulates economic growth and energy consumption [10]. However, as mentioned, agricultural insurance is a risk management tool for farmers that may increase farmers’ willingness to adopt new production methods. Li and Yu [11] investigated the impact of agricultural insurance on agricultural carbon welfare performance, which was calculated based on agricultural carbon emissions. Mu et al. [12] explored the impact of agricultural insurance on agricultural carbon emissions. However, their results were contrary to the findings of Li and Yu [11]. In addition, agricultural insurance may potentially lead to the adoption of more efficient and environmentally friendly agriculture technologies by stabilizing the incomes of farmers [4,7]. Li et al. [4] also found that agricultural green total factor productivity can be improved with an increase in agricultural insurance and thus significantly promote the sustainable development of agriculture. Furthermore, some studies observed that green technology innovations can reduce carbon emissions [13]. In light of these facts, we believe that agricultural insurance development would reduce carbon emissions. However, to the best of our knowledge, no studies have investigated the role of low-carbon technology innovation on the relationship between agricultural insurance and agricultural carbon emissions based on a mediating framework, which can be used to examine how a variable (e.g., agricultural insurance) influences another (e.g., agricultural carbon emissions). This study tends to bridge this gap by employing a two-way fixed effect panel model with data from 30 provinces in China to figure out the mediation mechanism.

According to Somaini and Wolak [14] and Kropko and Kubinec [15], the two-way fixed-effect model includes both individual fixed effects, which refer to the effect of “not changing with time, but changing with the individual”, and the time fixed effect, which is the effect of “not changing with the individual but changing with time”. This paper uses a two-way fixed-effect model based on the following three points of the model [16,17]. First, the model assumes that individual effects are related to explanatory variables, and each individual has a non-random intercept term. Second, the mainstream estimation strategy is “two-way fixed effect + cluster robust standard error”. Third, the results can be unanimously estimated.

Our results, which remained consistent with several robustness tests, can be summarized as follows. Firstly, we found a significant negative impact of agricultural insurance on agricultural carbon emissions from three estimated coefficients of agricultural insurance development indicators. This result was consistent with the findings of Mu et al. [12] but contrary to those of Li and Yu [11]. Secondly, we found that low-carbon technology innovation played a mediating role in the link between agricultural insurance and agricultural carbon emissions. The development of agricultural insurance has a significant positive effect on low-carbon technology innovation in agriculture, and low-carbon technology innovation has a significant negative effect on agricultural carbon emissions. Thirdly, we also explored the impact of agricultural insurance on carbon emissions in different areas of China. The results indicated that agricultural insurance negatively impacts carbon emissions in eastern regions and non-main grain-producing regions. Finally, we found that most of the control variables, the expansion of the crop sowing scale, the economic development of agriculture, forestry, animal husbandry, and fishery, the income of rural residents, the agricultural mechanization level, natural disasters, and the industrial structure will increase agricultural carbon emissions, while forest coverage will reduce carbon emissions.

The contributions of this study are demonstrated from various perspectives. The main contribution is that it provides more empirical evidence for the relationship between agricultural insurance and carbon emissions. Although limited existing research investigated the impact of agricultural insurance on carbon emissions [11,12], these studies had opposing results. Additionally, Mu et al. [12] applied a dynamic DID model, which suffered from an inappropriate control group, lead time effect, and lagged effect [18]. Li and Yu [11] employed agricultural carbon welfare performance rather than agricultural carbon emissions as the independent variable. Therefore, this study employs a two-way fixed effect panel model and calculates the carbon emissions that come from the agricultural sector to investigate the potential effect of agricultural insurance on carbon emissions. The second contribution is that this study explores the possible mediating role of low-carbon technology innovation between insurance and CO₂ emissions. Previous researchers investigated the direct effects of agricultural insurance on carbon emissions [11,12], the direct effects of insurance on technology adoption [4,7], and the direct effects of technology innovation on carbon emissions [13]. However, they did not consider the mediating role of technology in the relationship between insurance and carbon emissions. Therefore, this study constructs a mediating framework that treats low-carbon technology innovation as a mediator between agricultural insurance and carbon emissions. Third, we investigate the direct relationship between agricultural insurance and carbon emissions in China’s different regions, which are divided based on their levels of economic development and the kinds of natural resources they possess. Empirical evidence can be provided to governments as practical advice for claiming different policies in different regions.

The remainder of this study is organized as follows. The next section is the literature review. The research model and hypotheses development are shown in Section 3. Variables measurement, the data, and the methodology are presented in Section 4. Section 5 presents empirical results, robust and endogenous test results, and further discussion. Section 6 gives the conclusion and provides practical insights.

2. Literature Review

2.1. Agricultural Carbon Emission

According to Wang et al. [6], the direct or indirect greenhouse gas emissions produced from the use of fertilizers, pesticides, and fossil fuels as well as waste treatment in agricultural production can be used to measure agricultural carbon emissions. The latest statistics show that the second largest emitter of greenhouse gas emissions is agriculture and that greenhouse gas emissions from agriculture account for 21% of the total greenhouse gas emissions around the world [19]. Previous studies proposed several different ways to measure agricultural carbon emissions. For example, agricultural carbon emissions can be calculated through the energy consumed by pesticides, fertilizers, irrigation, and seed cultivation [20]. Agricultural carbon emissions can also be measured based on agricultural land use, livestock and poultry, and rice cultivation in China [6,21].

Agricultural production activities are characteristics of a long cycle and high dependence on the natural climate. There are many factors that can influence the process of agricultural production. Therefore, the influencing factors of agricultural carbon emissions are diverse. In general, we can identify two kinds of influencing factors that would impact agricultural carbon emissions [2]. First, previous studies investigated several external factors that affect agricultural carbon emissions, such as innovation capacities, carbon taxes, and employment policies. Second, there are also some internal factors, such as high-standard farmland construction policies and agricultural production methods [22]. In particular, the primary factor that influences agricultural carbon emissions is the method of land production.

2.2. Agricultural Insurance

In China, policy-based agricultural insurance was implemented in 2007 to effectively reduce agricultural disaster losses. According to different kinds of agriculture, agricultural insurance can be divided into breeding insurance and crop insurance. In 2022, China’s agricultural insurance premium income reached 121.9 billion yuan, with a year-on-year growth of about 25%, continuing to maintain a high growth rate. Just like any other insurance, agricultural insurance policies are offered by insurance companies to customers when premiums cover all of their costs, which contain two components: the indemnities that have to be paid out to cover losses and the costs the companies incur when delivering and managing the policies [23]. The agricultural insurance premium that farmers must pay lowers their willingness to purchase it. Therefore, in China, governments usually provide subsidies to participating farmers (as premium subsidies) and insurance companies (as administrative cost subsidies) to improve the farmers’ willingness to purchase agricultural insurance [24]. In addition to reducing economic losses and protecting agricultural production, previous studies denoted that agricultural insurance helps reduce pesticide use and can lead to the adoption of agricultural green technologies [4,25]. Therefore, we believe that agricultural insurance would produce an impact on the environment.

The impact of insurance on the environment has been discussed in recent years. Altarhouni et al. [10] explored the effect of insurance market development on environmental degradation and discovered that insurance market development contributes to an increase in CO₂ emissions. Similarly, Appiah-Otoo and Acheampong [26] found that the expansion of the insurance industry stimulated CO₂ emissions in BRICS countries. According to Altarhouni et al. [10], insurance market development may cause changes to CO₂ emissions in two ways: economic growth and energy consumption. Specifically, the development of the insurance market would help the accumulation of production capital since it allows investors to diversify their investments, which increases the chances of high-productivity investments. Subsequently, it would enhance the liquidity positions in the market and encourage economic growth rates. Consequently, with the growth of the economy, more CO₂ emissions would be stimulated at higher levels [27]. Additionally, insurance market development would impact investors’ investment behaviors through the monitoring mechanism of insurance companies. Investors would attempt to increase the purchase rate of new technologies and equipment to promote the productive potential of their investment. Consequently, insurance development should lead to higher levels of energy consumption and CO₂ emissions.

Regarding agricultural insurance, however, increasing agricultural producers’ understanding of agricultural insurance would help promote the development of sustainable agriculture. Nilwala and Jayarathna [28] argued that agricultural insurance can encourage the use of organic methods of farming, which can reduce the effects of global warming. Compared with other kinds of insurance products, agricultural insurance would influence CO₂ emissions differently since it falls under the category of green insurance, which is designed to promote green development [29,30,31]. Ahmed et al. [32] denoted that agricultural insurance not only motivates farmers to be more accepting of the use of environmentally friendly production technology but also encourages farmers to reduce the use of chemicals that potentially pollute the environment.

Some researchers utilized the index of agricultural green total factor productivity, which measures the input-output efficiency in the process of agricultural production and considers the utilization efficiency of agricultural resources and environmental pollution, to explore its relationship with agricultural insurance [25,32]. For example, Fang et al. [25] found that the stimulation effect of crop insurance on agricultural green total factor productivity increases with the expanded use of agricultural green technologies and an increase in operational scales. These results indicate that agricultural insurance helps improve the input-output efficiency in the agricultural production process and reduce environmental pollution. However, agricultural green total factor productivity is a complex index rather than an index that is calculated just based on carbon emissions from the agricultural production process. Mu et al. [11,12] and Li and Yu [11] explored the relationship between agricultural insurance and carbon emissions in China. Li and Yu [11] denoted a positive relationship between the development of agricultural insurance and agricultural carbon welfare performance, which was calculated based on agricultural carbon emissions. They also found that the interaction between agricultural insurance and the Internet plays a role in reducing carbon emissions. Therefore, they believe that there is a synergistic governance effect of “Internet + Agricultural Insurance” on agricultural green development. On the contrary, Mu et al. [12] applied a multistage dynamic DID model to explore the effect of policy-oriented agricultural insurance on green agricultural development. They observed that the implementation of policy-oriented agricultural insurance has a significant positive impact on reducing agricultural carbon emissions. Obviously, the results of these two studies are opposing, indicating that investigating the impact of agricultural insurance on agricultural carbon emissions is necessary. We provide a brief review of the related literature in Table 1.

2.3. Low-Carbon Technology

Low-carbon technology is beneficial for the development of a low-carbon economy, which incorporates both an increase in carbon absorption and a decrease in carbon emissions. Scholars have developed several measuring methods to measure the intensity of low-carbon technology innovation within a country. The types of statistics that are used in these measuring methods can generally be summarized as the number of research and development (R&D) inputs, the number of scientific publications, and the number of patents [33,34,35,36]. The number of research and development (R&D) inputs is commonly used to measure the level of technological innovation in enterprises or countries. Financialization and digital transformation are confirmed effects of increasing R&D inputs [37,38,39]. As for the number of scientific publications, Fagerberg et al. [35] believed that it basically reflects the research level of technology in the initial stage of development. In addition, the number of patents, which was initially adopted by Lanjouw and Mody [33] to measure low-carbon technology innovation, is widely used in studies related to the climate environment. Its advantage lies in the fact that the method is essentially a collection of statistics for various types of technologies in production. That is to say, it is possible to quantify the specifics of technological innovation in a particular field. This paper utilizes the number of patents to measure low-carbon technology innovation.

In this study, we focus on low-carbon technology that is applied during the agricultural production process. According to Guo et al. [40], agricultural low-carbon technologies refer to technologies that can reduce agricultural carbon emissions during the process of agricultural production, such as biochar technology, cropping systems, and the in-use of agricultural waste. One of the important methods to stimulate the agricultural economy as well as reduce CO₂ emissions is to apply agricultural low-carbon technologies [41]. For example, the application of advanced crop soil nutrient management, water-saving irrigation technologies, and low-carbon energy technologies could decrease carbon emissions without affecting the efficiency of food production [42,43]. Previous studies identified the drivers of low-carbon technology adoption in agriculture. First, some studies found that the characteristics of farmers, such as age, gender, and education level, impacted their willingness to adopt low-carbon technology [42,44]. Second, researchers denoted that farmers tended to adopt low-carbon technologies when they could receive reasonable government subsidies [45]. Third, other factors, such as social media participation, information networks, market incentives, and credit have been proven to significantly impact the adoption of low-carbon technologies in agriculture [46,47].

3. Research Model and Hypotheses Development

3.1. Research Model

The research model is shown in Figure 1 and mainly focuses on the acting path and mediating mechanisms of low-carbon technology innovation between agricultural insurance and agricultural carbon emissions.

3.2. Research Hypotheses

3.2.1. The Relationship between Agricultural Insurance and Agricultural Carbon Emissions

Following the research of Mu et al. [12], we believe the development of agricultural insurance would help restrain agricultural carbon emissions. This study summarizes three reasons why agricultural insurance may help control the increase in agricultural carbon emissions. Firstly, agricultural insurance is an insurance tool that is used to transfer agricultural risks and compensate farmers who contribute to green development but suffer economic losses, thus prompting victims to resume normal operations quickly. Under these conditions, farmers may have more motivation to adopt environmentally friendly production behaviors, such as employing biological pesticides rather than chemical pesticides, because they will not have to worry about the potential losses that come with helping the growth of green agriculture. Secondly, agricultural insurance can decrease the impact of smog, thunderstorms, and other natural disasters on agricultural productivity. Farmers’ capability to deal with the risks associated with agricultural production can be improved. The ability to withstand hazards might inspire farmers to use environmentally friendly agricultural technology to boost their operations’ productivity significantly. Finally, agricultural insurance helps farmers who suffer economic losses during agricultural production obtain economic compensation promptly, thereby stabilizing the social and economic order and reducing the burden of the government to help victims. Consequently, agricultural insurance would encourage the government to adopt and promote more aggressive environmental policies that will benefit the environment. In summary, this paper suggests the following hypothesis.

Hypothesis 1 (H1). 

The development of agricultural insurance helps control the increase in CO₂ emissions.

3.2.2. The Mediating Role of Low-Carbon Technology Innovation between Agricultural Insurance and Agricultural Carbon Emissions

Recently, some findings regarding the impact of low-carbon technology on environmental issues have emerged. Lin [48] argued that low-carbon technology innovation is a leading force for sustainable urban development since these technologies focus on energy saving and emissions reduction. Khattak et al. [49] adopted the statistics from the first quarter of 1990 to the fourth quarter of 2018 for G7 countries to study green technology innovation and CO₂ emissions, and their results reflected that there is an inverse periodicity between the two. Yii and Geetha [50] conducted the Granger Causality test to explore the relationship between technology innovation and CO₂ emissions in Malaysia. Their results showed that technology innovation negatively impacted carbon dioxide emissions in the short term, while a long-term relationship was not found in their test. Du et al. [51] tested the effect of green technology innovation on CO₂ emissions, and in general, green technology innovation did not significantly contribute to a reduction in CO₂ emissions for economies with relatively low income levels, while the reduction effect was significant for economies with relatively high income levels. Liguo et al. [52] used dynamic least-squares methods to discover that technologies related to the production, distribution, or transmission of marine energy innovations contribute to the reduction in CO₂ emissions. Therefore, we hypothesize that low-carbon technology innovation in agriculture would help to reduce agricultural carbon emissions in this study.

The presence of insurance markets can have a behavior-altering stimulation effect on technology adoption decisions [53]. A significant number of health economics studies support the idea that health insurance significantly impacts medical innovation and implementation [54,55]. Research has also been conducted on some insurance products belonging to agricultural insurance. Giné and Yang [56] found that weather insurance reduces loan utilization when bundled with loans that support the adoption of new agricultural technologies. Carter et al. [57] theoretically investigated the impact of index insurance on the adoption of agricultural technologies. On the one hand, they found that risk directly hinders technology adoption and that farmers are reluctant to invest their savings in technological innovation because they need to use those savings to cope with potential income shortages. On the other hand, the associated risks among different farmers, such as weather risks, indirectly create portfolio construction problems for microfinance and other potential loan receivers, further increasing the cost of credit to the small farming sector and hindering the adoption of technology. Therefore, weather insurance, which can be used to eliminate risks, can theoretically increase farmers’ willingness to adopt new technologies. Miao et al. [58] found that agricultural insurance influences farmers’ demand for innovation, whereas innovators foresee this effect and respond by adjusting their agricultural technology innovation activities.

This paper suggests that agricultural insurance affects low-carbon technology innovation in agriculture for three main reasons. Firstly, risk avoidance and transfer are the basic functions of insurance. Agricultural insurance would reduce farmers’ concerns about the risks brought by the adoption of low-carbon technologies since it can avoid and transfer the natural, technological, and market risks in agricultural production and distribution [4,59,60,61]. Secondly, the compensation system of agricultural insurance can compensate for the loss of income caused by risky incidents, even if this loss is caused by the adoption of low-carbon technologies [4,60]. Finally, an agricultural insurance policy issued by the government, such as raising the subsidy standard for farmers who adopt environmental-friendly production patterns, can maintain agricultural producers’ motivation to participate in low-carbon production activities, thus safeguarding the steady development of agriculture [62]. Overall, agricultural insurance not only protects agricultural producers’ incentives to produce but also enhances their willingness to use low-carbon technologies, which ultimately leads innovators to increase their efforts in low-carbon technology innovation.

In summary, agricultural insurance may reduce CO₂ emissions by prompting low-carbon technology innovation. Accordingly, the following hypothesis is proposed.

Hypothesis 2 (H2). 

The mediating role of low-carbon technology innovation between agricultural insurance and agricultural carbon emissions is significant.

4. Data and Methodology

4.1. Data Source and Design

4.1.1. Data Source

This paper examines and evaluates the effect of agricultural insurance development on agricultural carbon emissions by selecting panel data from 30 provinces in China from 2001 to 2019. The data on agricultural carbon emissions (i.e., dependent variable) and the control variables, including economic levels of agriculture, forestry, animal husbandry and fishery, planting scale, farmers’ income level, industrial structure, level of natural hazards, forest carbon concentration capacity, and agricultural mechanization level, were obtained from the “China Statistical Yearbook” [63]. Additionally, the data on agricultural insurance premium income, agricultural insurance breadth, and agricultural insurance density were obtained from the “China Insurance Statistical Yearbook” [64]. The data of the number of low-carbon technology patent applications were obtained from the Incopat database [65].

4.1.2. Dependent Variable

The dependent variable is agricultural carbon emissions. Its estimation formula was constructed as below:

(1)$carbo{n}_{it}={{\displaystyle \sum }}_{j=1}^n{Y}_{ijt}={{\displaystyle \sum }}_{j=1}^n{R}_{ijt}{X}_j$

where $carbo{n}_{it}$ is the agricultural carbon emissions in the province i at the year t, $Y_{ijt}$ is the carbon emissions from the jth carbon source in the province i at the year t, $R_{ijt}$ is the use of the jth carbon source in the province i at the year t, and $X_j$ is the emission coefficient corresponding to the jth carbon source.

4.1.3. Independent Variable

The independent variable, which is the level of agricultural insurance development, was mainly measured by the agricultural insurance premium income that was collected from the “China Insurance Statistical Yearbook”. Additionally, to test the robustness of the model, the agricultural insurance breadth and agricultural insurance density were also collected as measures of the level of agricultural insurance development. The former refers to the insurance premium income per unit of crop area, and the latter represents the unit insurance premium income of the rural population, which is the ratio of agricultural insurance premium income to crop sown area and rural population, respectively.

4.1.4. Mediation Variable

The mediation variable is the level of low-carbon technology innovation, which can be represented by the knowledge stock of low-carbon technology innovation. The raw data of low-carbon technology patents in the fields of agriculture, forestry, animal husbandry, and fishery in each province of China were obtained from the Incopat database. Then, the corresponding knowledge stock was obtained by referring to Popp [66], who constructed the knowledge stock by using the patent data as follows:

(2)$K_{i,t}={\displaystyle \sum }_{s=0}^{\infty }{e}^{-{\beta }_{1}s}\left(1-e^{-{\beta }_2\left(s+1\right)}\right)PA{T}_{i,s}$

where $K_{i,t}$ represents the knowledge stock of the province i in the year t;${\beta }_1$ is the obsolescence rate of knowledge, ${\beta }_2$ represents the diffusion rate; s represents the number of years before the current year; and $PA{T}_{i,s}$ represents the number of low-carbon technology patent applications in the province i in the year t. According to Wei and Yang [67], who calculated the knowledge stock in China using Equation (2), ${\beta }_1$ is set at 0.36, and ${\beta }_2$ is set at 0.03.

4.1.5. Summary Statistics

Linear interpolation method was used to fill in the missing individual data. Additionally, the real variables involved were logarithmically treated in order to reduce the absolute differences between the data, avoid the influence of individual extreme values, and satisfy the assumptions of the classical linear model. The abbreviations and descriptive statistics of each variable are summarized in Table 2.

4.2. Methodology

4.2.1. Two-Way Fixed Effects Model

In this paper, based on the general steps of estimation and analysis of static panel data, after doing some test work on the data, we propose a two-way fixed effects model for regression analysis and establish the following regression model to study the relationship between the independent variables and the dependent variables.

(3)$lncarbo{n}_{it}={\alpha }_0+{\alpha }_{1}lnsuranc{e}_{it}+\sum {\alpha }_{i}Control{s}_{it}+{\sigma }_i+yea{r}_i+{\epsilon }_{it}$

To test the mediating effect of low-carbon technology innovation, which is measured by the knowledge stock of low-carbon technology innovation, the following model is also developed.

(4)$lncarbont{i}_{it}={\gamma }_0+{\gamma }_{1}lnsuranc{e}_{it}+{\sigma }_i+yea{r}_i+{\epsilon }_{it}$

(5)$lncarbo{n}_{it}={\theta }_0+{\theta }_{1}lnsuranc{e}_{it}+{\theta }_{2}lncarbont{i}_{it}+\sum {\theta }_{i}Control{s}_{it}+{\sigma }_i+yea{r}_i+{\epsilon }_{it}$

Within the three models above, i and t represent provinces (i = 1, 2, 3, …, 30) and years (t = 2001, 2002, …, 2019), and ${\epsilon }_{it}$ represents error terms that vary with individuals and time. lncarbon and lnsurance represent the agricultural carbon emissions and agricultural insurance development levels in different provinces. lncarbonti is the mediating variable, which represents the level of low-carbon technology innovation; Controls are the control variables. $\alpha , \beta , \gamma$ and $\theta$ are parameters to be estimated. ${\sigma }_i$ is the province fixed effect, and $yea{r}_i$ is the time fixed effect.

4.2.2. Model Selection

The individual fixed effects model and random effects model are commonly used static panel data models. The F test and LM test can identify individual fixed effects and random effects. The results of the F test and LM test are shown in Table 3. It can be seen that both individual fixed effects and random effects exist since all statistics are significant.

The modified Wald test is used to test the existence of groupwise heteroscedasticity, and Wooldridge test is used to test the existence of autocorrelation. The test results are shown in Table 4. The modified Wald test and the Wooldridge test both reject the null hypothesis since these statistics are significant at a 1% level. That is, the problems of groupwise heteroscedasticity and autocorrelation exist in all three models. Therefore, the robust Hausman test is used to further examine the individual fixed effects and random effects since the traditional Hausman test is invalid when groupwise heteroscedasticity and autocorrelation exist. The results of the robust Hausman test are in Table 5.

The three results of the Hausman test all reject the null hypothesis, indicating that the individual fixed effects model is optimal. Additionally, this study further tests whether we should include time fixed effects in the model by generating time dummy variables. The results for each time dummy variable are significant, indicating that the null hypothesis of “controlling for individual fixed effects only” is rejected, and a two-way fixed effects model should be chosen. Therefore, the effects of both the province and time dimensions should be controlled simultaneously.

5. Estimated Results

5.1. Agricultural Insurance, Low-Carbon Technology Innovation, and Agricultural Carbon Emissions

This section focuses on the impact of agricultural insurance on agricultural carbon emissions and the mediating role of low-carbon technology innovation. We first investigate the direct effect of agricultural insurance on agricultural carbon emissions. Equation (3) is used to explore this direct effect. Agricultural insurance premium income, agricultural insurance density, and agricultural insurance breadth are used as indicators of the level of agricultural insurance development. The stepwise regression results of the equations are reported in Table 6. The estimated coefficients of the level of agricultural insurance development (i.e., insurance premium, insurance breadth, and insurance density) on agricultural carbon emissions are all significant at the 1% level; thus, Hypothesis 1 is supported. These results indicate that agricultural insurance can play an effective role in reducing agricultural carbon emissions. As mentioned above, agricultural insurance stimulates farmers to adopt environmentally friendly production behaviors by transferring agricultural risks and compensating farmers who suffer economic losses. Agricultural insurance improves farmers’ capability to deal with the risks associated with agricultural production. Under these conditions, farmers tend to adopt low-carbon technology in agriculture without worrying about potential economic loss. We also test the effect of agricultural insurance on low-carbon technology later.

As for control variables, the following results were obtained.

• (1). The expansion of the crop sowing scale will increase agricultural carbon emissions. In the agricultural production process, the scale of agricultural operation will make the overall operating cost lower. However, the expansion of crop planting areas will directly lead to an increase in the application rates of chemicals. Additionally, it will make the diffusing of chemicals such as fertilizers and pesticides more difficult, thus reducing fertilizer efficiency and indirectly expanding the use of chemicals. Therefore, it is reasonable to attain a positive significant result regarding the effect of planting scale on total agricultural carbon emissions.

• (2). The economic levels of agriculture, forestry, animal husbandry, and fishery and the income level of farmers both significantly increase the agricultural carbon emissions. According to a previous study on the Environmental Kuznets Curve (EKC) theory, the pollution level of a region will rise with the rise of income and economic level and will fall with the rise of income when the economic level reaches a certain state [27]. In this study, the empirical results show that there is a significant positive relationship between agricultural economic level and agricultural carbon emissions. Therefore, we can conclude that the agricultural carbon emissions in China are at the stage of increasing with the agricultural economic level, demonstrating that the agricultural economic development in China at this stage is not enough to restrain carbon emissions.

• (3). The positive effect of agricultural mechanization level on agricultural carbon emissions is significant. With other factors unchanged, the increase in the total power of agricultural machinery will lead to an increase in energy inputs such as agricultural diesel, which will affect the total carbon emissions.

• (4). Natural disasters have a significant positive effect on agricultural carbon emissions. Crop damage leads directly to crop yield reduction, and forest damage leads to a decrease in carbon sink capacity. These effects make crops’ restraining effects on agricultural carbon emissions weaker, thus reflecting the regression results of the opposite relationship.

• (5). The industrial structure, expressed as the share of value added in the secondary sector [68], has a significant positive effect on agricultural carbon emissions, indicating that an increase in the share of the secondary sector will increase the carbon emissions of agriculture.

• (6). Forest cover represents the inherent carbon concentration capacity of the agricultural sector, which can play a certain inhibitory role in agricultural carbon emissions. The negative coefficient in Table 6 proves this point of view.

Equations (4) and (5) have been estimated to investigate the mediating role of low-carbon technology innovation between agricultural insurance and agricultural carbon emissions. The first three columns in Table 7 show the regression results, and low-carbon technology innovation is treated as a dependent variable after the stepwise inclusion of agricultural insurance and control variables. The regression coefficients of agricultural insurance development on low-carbon technology innovation are all significant at the level of 5% and do not change with the inclusion of the control variables, indicating that agricultural insurance development significantly improves low-carbon technology innovation. We can examine the indirect effect of agricultural insurance development on agricultural carbon emissions through low-carbon technology innovation according to the estimated coefficients in the third and fourth columns in Table 7. The development of agricultural insurance has a significant positive effect on low-carbon technology innovation in agriculture since the estimated coefficient of agricultural insurance development on technology innovation is positively significant at the level of 5% in the third column. Additionally, the regression coefficients of low-carbon technology innovation and agricultural insurance development on agricultural carbon emissions are both negatively significant at the level of 1% in the fourth column. These results indicate that an indirect effect of agricultural insurance development on agricultural carbon emissions through low-carbon technology innovation exists. Hypothesis 2 is preliminarily supported.

In addition, this study also applies the bootstrap method to test the significance of indirect and direct effects and estimate the strength of these effects. The bootstrap method was conducted by resampling when we test whether the product of the coefficients between the dependent variable and the independent variable is zero, as well as whether the product of the coefficients between the independent variable and the mediating variable is zero. The corresponding null hypothesis is that the product of the coefficients equal zero.

From the results in Table 8, the indirect effect of agricultural insurance development to carbon emissions mediated by low-carbon technology was significant (ind_eff = −0.002, p = 0.033), accounting for 11% (−0.002/(−0.016 + −0.002)) of the total effect (i.e., the total effect equals the sum of indirect effects and direct effects). Therefore, we conclude that the Hypothesis 2 is supported, indicating that low-carbon technology significantly mediates the relationship between agricultural insurance development and agricultural carbon emissions. Additionally, after we held the low-carbon technology and other covariate constants, the direct effect of agricultural insurance development on agricultural carbon emissions was also significant (dir_eff = −0.016, p = 0.001) accounting for 89% (−0.016/(−0.016 + −0.002)) of the total effect, indicating a partial mediation. According to these results, the mediating role of low-carbon technology between agricultural insurance and agricultural carbon emissions has been confirmed.

As rational, economic people, farmers’ goal is to earn profits and avoid risks [12]. Thus, they would not choose low-carbon technology during the agricultural production process since this technology may hinder the improvement of agricultural production efficiency and be expensive. However, agricultural insurance is a tool that can be used to transfer risks during the agricultural production process [8]. Therefore, farmers tend to adopt low-carbon technology if they purchase agricultural insurance. Eventually, agricultural carbon emissions decrease due to the adoption of low-carbon technology. In addition, our results indicate that the relationship between agricultural insurance and agricultural carbon emissions is partially mediated by low-carbon technology. That means farmers who purchase agricultural insurance tend to adopt environmentally friendly production methods, even if they may not use low-carbon technology in agriculture. For example, agricultural carbon emissions decrease due to the decreasing use of pesticides, fertilizers, and energy.

5.2. Addressing Endogeneity: Instrumental Variable Method

The regression results indicate that dependent variables and independent variables are correlated. However, the models may have a potential endogeneity problem since the decrease of agricultural carbon emissions may increase the market demand for agricultural insurance, which in turn leads to an increase in agricultural insurance premium income. For this reason, the two-stage (2sls) least-squares method is applied to regress the Equation (3) again by selecting the one-period lags of the independent variables as the instrumental variables. The regression results show that the instrumental variables KP-LM’s test results are 15.233, 14.124, and 13.440, corresponding to p values of 0.0001, 0.0002, and 0.0002, all of which strongly reject the null hypothesis of unidentifiability; the KP-F’s test values are 199.552, 102.104, and 85.126, all of which are much larger than the 10% critical values of 16.3, indicating the rejection of the weak instrumental variable’s null hypothesis; accordingly, the p-values of the Hansen J statistical test are 0.2151, 0.1502, and 0.1957, thus not rejecting the null hypothesis of exogenous instrumental variables. In summary, it can be seen that the instrumental variables were selected reasonably and effectively.

Compared to Table 6, the decreased effects of agricultural insurance on agricultural carbon emissions in Table 9 are still significant after adding the instrumental variables. The absolute values of the regression coefficients in Table 9 are bigger than those in Table 6, indicating that the estimation results are still reasonable and reliable after considering the existence of endogeneity. Hypothesis 1 is verified again.

5.3. Agricultural Insurance, Low-Carbon Technology Innovation, and Agricultural Carbon Emissions in Different Areas

Considering the differences in economic development and natural resources, China is divided into eastern, western, central, and northeastern regions. Similarly, there are main grain-producing areas and non-main grain-producing areas, according to the differences in agricultural production conditions and agricultural policy support. This paper explores the heterogeneity of the impact of agricultural insurance premium income on agricultural carbon emissions under different divisions. Table 10 shows the regression results.

According to the first and second column results in Table 10, the development of agricultural insurance does not play a role in reducing agricultural carbon emissions in the main grain-producing areas, while its role is significant in the non-main grain-producing areas. Compared with the main grain-producing areas, the non-main grain-producing areas have a disadvantageous ecological environment and do not have a strong risk resistance ability. Therefore, agricultural insurance would stimulate and enhance farmers’ enthusiasm to invest in agricultural production since one of the important functions of insurance is risk prevention. In other words, agricultural insurance plays a restraining role in agricultural carbon emissions.

Looking at the results in the third to sixth columns in Table 10, the effect of agricultural insurance development on agricultural carbon emissions is significantly negative in the eastern provinces. This effect is negative but not significant in the central regions, while it is positive and insignificant in the western and northeastern regions. The eastern regions of China are relatively economically developed and have promoted the development of agricultural insurance relatively early so that farmers’ awareness of agricultural insurance has been strengthened. Farmers have taken the initiative to adjust their agricultural production methods or adopt low-carbon and environmentally friendly production technologies because they are aware that agricultural insurance can enhance post-disaster recovery ability and ensure a stable income. In contrast, agricultural operators in the central regions lack knowledge of insurance products, and their income levels are generally low, so it is difficult for them to purchase agricultural insurance to prevent risks. Therefore, it becomes difficult for agricultural insurance to play a role in carbon emissions reduction. As for the western and northeastern regions, the recent promotion of agricultural insurance can stimulate farmers’ willingness to participate in agricultural production. The development of agricultural insurance in the western area has been boosted since the claim of The Great Western Development Strategy announced by the government in the year 1999. However, those in the western area would not produce in an environmentally friendly way since the development level of their agriculture is low. Thus, agricultural insurance would raise carbon emissions rather than restrain carbon emissions.

6. Concluding Remarks

In this paper, we collected panel data from 30 Chinese provinces from 2001–2019 to study the mechanism of how agricultural insurance reduces agricultural carbon emissions. Two-way fixed effect models were used to determine the impact of agricultural insurance on low-carbon technological innovation and agricultural carbon emissions in this study. We tested the indirect effect of agricultural insurance on carbon emissions through low-carbon technology innovation. Additionally, instrumental variables were included to address the endogeneity problem, and regression analyses on grouped data were conducted to analyze regional heterogeneity. Our conclusions are as follows:

Firstly, the estimated coefficients between the dependent variable and the independent variable are significantly negative, indicating that the development of agricultural insurance can significantly reduce agricultural carbon emissions. This result is consistent with the findings of Mu et al. [12] but contrary to those of Li and Yu [11]. Additionally, the indirect effect results show that agricultural insurance can achieve carbon emissions reduction through the indirect path of promoting the innovation of low-carbon technologies.

Secondly, the grouped regression results show that there is significant regional heterogeneity. Specifically, the reduction effect of agricultural insurance on agricultural carbon emissions is significant in the eastern regions, while there is no significant reduction effect in other regions. Additionally, the development of agricultural insurance in non-main grain-producing areas has a significant reduction effect on carbon emissions, while main grain-producing regions do not.

Finally, most of the control variables, the expansion of the crop sowing scale, the economic development of agriculture, forestry, animal husbandry, and fishery, the income of rural residents, the agricultural mechanization level, natural disasters, and the industrial structure will increase agricultural carbon emissions, while forest coverage will reduce carbon emissions.

In summary, this paper confirms that the promotion effect of agricultural insurance on carbon emissions is mediated by low-carbon technology, which is conducive to paying attention to the development of agricultural insurance and low-carbon technology. Our conclusions indicate that it is necessary to promote the development of agricultural insurance and fully boost technological progress and efficiency improvement. In addition, our results support the idea that farmers who purchase agricultural insurance tend to apply low-carbon technology during the agricultural production process, thus reducing agricultural carbon emissions. Since agricultural insurance belongs to green finance, our results could be applied to the relationship between green finance, low-carbon technology, and carbon emissions, indicating that the development of green finance could reduce carbon emissions through the adoption of low-carbon technology.

6.1. Managerial Implications

Our results indicate that agricultural insurance can directly or indirectly play a role in agricultural carbon emission reduction, suggesting several practical managerial implications as follows.

First, the government should adopt multi-faceted and multi-level measures to accelerate the process of popularizing agricultural insurance and continuously enhance the utilization efficiency of agricultural insurance. On the one hand, the government should solve problems at the supply level, such as the high cost of signing transactions at the early stage and the cumbersome procedures of claim settlement at the later stage, to help agricultural insurance companies improve their operational efficiency and effectiveness. On the other hand, the government should also improve the shortage at the demand level by popularizing the knowledge of agricultural insurance nationwide, especially in the western and northeastern regions and the main grain-producing areas. Farmers’ awareness of agricultural risk prevention would increase due to the popularization, which in turn would stimulate their willingness to purchase agricultural insurance. Consequently, agricultural insurance plays a role in their protection and helps offset the disadvantages in agricultural infrastructure, farmers’ production skills, and natural environment endowments.

Second, the government should introduce talents and technologies to solve the shortcoming of low-carbon products and adopt policies to stimulate the innovation enthusiasm of the relevant departments. For example, the government could combine agricultural insurance, which provides risk protection, and green technology to guide agricultural operators to adopt environmentally friendly and low-carbon production technology. The government could also raise the subsidy standard of agricultural insurance for farmers who adopt environmental-friendly production patterns. By doing so, farmers’ concerns about the risks of adopting low-carbon technologies would be reduced. Then, farmers’ willingness to apply low-carbon technologies would increase, and agricultural carbon emissions would decrease. Additionally, companies that research and develop low-carbon technology would be eager to make innovations since the farmers’ demand for low-carbon technology would increase. Therefore, the government could accomplish the goal of reducing carbon emissions by effectively promoting the transformation of low-carbon technology products.

Third, the degree of impact of agricultural insurance on carbon emissions in different regions is significantly different. The reduction effect of agricultural insurance on agricultural carbon emissions is significant in the eastern regions and the non-main grain-producing areas, while this effect is insignificant in the western and northeastern regions and the main grain-producing areas. Therefore, the government should focus on regional characteristics when they formulate policies to promote green agriculture through agricultural insurance and low-carbon technology in different regions.

6.2. Limitations and Future Directions

This study provides substantial evidence about the roles of agricultural insurance and low-carbon technology in reducing agricultural carbon emissions. However, our study still has some issues that will be left for further research. First, our study utilizes samples of China; however, other countries or regions with different agricultural development patterns and different development levels of agricultural insurance may possess different effects on carbon emissions. Future works could investigate a broader sample from different countries or regions.

Second, we only collect the data of low-carbon technology and treat it as a mediator. However, relevant studies denoted that agricultural chemicals, such as pesticides and fertilizers, also contribute to agricultural carbon emissions [40]. Additionally, farmers who purchase agricultural insurance tend to reduce their consumption of agricultural chemicals [12,69]. Therefore, future research could introduce agricultural chemicals into the mediation framework developed by this study to test the mediation role of agricultural chemicals between agricultural insurance and agricultural carbon emissions.

Third, in this study, we carry out empirical analysis through macro data rather than from a micro perspective. However, the characteristics of farmers, such as age, gender, and education level, may also influence the willingness of farmers to purchase agricultural insurance and adopt green agricultural production methods. The influence mechanism among agricultural insurance, low-carbon technology, and agricultural carbon emissions could be investigated from a micro perspective in the future research.

Despite these limitations, we believe that our research helps fill some important gaps in the understanding of green agricultural development, provides empirical evidence to the relevant literature, and suggests useful directions for future research.

Author Contributions

Conceptualization, S.-j.J. and F.X.; Methodology, L.W.; Software, L.W.; Formal analysis, S.-j.J.; Investigation, L.W.; Writing—original draft, L.W. and F.X.; Writing—review & editing, S.-j.J.; Supervision, F.X.; Funding acquisition, S.-j.J. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Acknowledgments

The authors thank the editors and reviewers for their helpful comments.

Conflicts of Interest

The authors declare no conflict of interest.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figure and Tables

Figure 1. Research Model.

Table 1

A brief review of the related literature.

Study	Methodology	Results (+/−)	Indirect Effect (Yes/No)
The impact of agricultural insurance on agricultural carbon emissions
Li and Yu [11]	Panel model	+	No
Mu et al. [12]	DID model	−	No
The impact of agricultural insurance on agricultural green total factor productivity
Li et al. [4]	Panel model	+	No
Fang et al. [25]	Panel model	+	No
Ahmed et al. [32]	Panel model	+	No

Notes: The “+” in the table represents that the impact of agricultural insurance on agricultural carbon emissions is positive, and the impact of agricultural insurance on agricultural green total factor productivity is positive. The “−” represents that the impact of agricultural insurance on agricultural carbon emissions is negative, and the impact of agricultural insurance on agricultural green total factor productivity is negative.

Table 2

The abbreviations and summary statistics for all variables.

Variables	Abbreviation	Meaning of Variables/Calculation Method	Unit	Mean	S. D.	Min.	Max.
Agricultural carbon emissions	lncarbon	Total agricultural carbon emissions, logarithmic	Tons/million	14.40	1.01	11.63	15.98
Level of agricultural insurance development	lninsur	Agricultural insurance premium income, logarithmic	Millions of Yuan	4.75	2.57	0.00	8.80
Agricultural insurance breadth	insurgd	Insurance premium income per unit of crop area, logarithmic	Yuan/hectare	3.67	2.29	0.00	9.12
Agricultural insurance density	insurmd	Insurance premium income per unit of rural population, logarithmic	Yuan/person	3.25	2.21	0.00	6.31
Level of low-carbon technology innovation	lncarbonti	Knowledge stock of low-carbon technology innovation	— —	2.32	1.53	0.00	6.05
Industrial structure	secondst	Ratio of value added in the secondary sector to gross regional product	— —	0.42	0.08	0.16	0.62
Economic level of agriculture, forestry and fisheries	lnngdp	Regional gross agricultural, forestry and fishery products, logarithmic	billions	7.30	1.07	4.14	9.17
Planting scale	lnbzmj	Total area sown to crops, logarithmic	Thousand hectares	15.08	1.09	11.39	16.52
Income level of farmers	lnnmcf	Average income of rural residents, logarithmic	Yuan	8.71	0.71	7.25	10.41
Level of natural hazard	lnzai	Damage area, logarithmic	1000 Ha	6.41	1.51	0.00	8.90
Forest carbon concentration capacity	senlin	Forest coverage rate	%	30.08	17.68	2.94	66.80
Level of agricultural mechanization	nyjxh	Ratio of total agricultural machinery power to rural population	Kw/person	1.32	0.74	0.09	4.33

Table 3

Results of F test and LM test.

	Equation (3)	Equation (4)	Equation (5)
F test	135.40 ***(0.00)	22.52 ***(0.00)	146.54 ***(0.00)
LM test	3117.79 ***(0.00)	977.68 ***(0.00)	3132.13 ***(0.00)

Notes: *** refer to the significance at the 1% level; p-values are in the brackets.

Table 4

Results of modified Wald test and Wooldridge test.

	Equation (3)	Equation (4)	Equation (5)
Modified Wald test	1385.02 ***(0.00)	2555.26 ***(0.00)	1224.49 ***(0.00)
Wooldridge test	180.12 ***(0.00)	860.41 ***(0.00)	182.95 ***(0.00)

Notes: *** refer to the significance at the 1% level; p-values are in the brackets.

Table 5

Results of robust Hausman test.

	Equation (3)	Equation (4)	Equation (5)
Robust Hausman test	36.28 ***(0.00)	17.18 ***(0.02)	51.14 ***(0.00)
Effects	Individual fixed effect	Individual fixed effect	Individual fixed effect

Notes: *** refer to the significance at the 1% level; p-values are in the brackets.

Table 6

Results of stepwise regression.

Variable	lncarbon
	(1)	(2)	(3)
lninsur	−0.01 ***(0.00)
lnsurmd		−0.01 ***(0.00)
lnsurgd			−0.01 ***(0.00)
lnbzmj	0.29 ***(0.04)	0.29 ***(0.04)	0.28 ***(0.04)
lnngdp	0.28 ***(0.03)	0.27 ***(0.03)	0.28 ***(0.03)
lnnmcf	0.35 ***(0.09)	0.36 ***(0.09)	0.35 ***(0.09)
senlin	−0.01 ***(0.00)	−0.01 ***(0.00)	−0.01 ***(0.00)
lnzai	0.01 ***(0.00)	0.01 ***(0.00)	0.01 ***(0.00)
nyjxh	0.07 ***(0.01)	0.07 ***(0.01)	0.06 ***(0.01)
secondst	0.45 ***(0.11)	0.39 ***(0.11)	0.40 ***(0.11)
constant	4.93 ***(0.74)	4.96 ***(0.74)	5.24 ***(0.74)
year	controlled	controlled	controlled
$\sigma$	controlled	controlled	controlled
N	570	570	570
R2	0.82	0.82	0.82
Adjusted R2	0.80	0.80	0.80

Notes: The table reports the regression results from the OLS method with year and province fixed effects; *** refer to the significance at the 1% level; standard errors are in the brackets.

Table 7

Results of indirect effects.

Variable	lncarbonti			lncarbon
	(1)	(2)	(3)	(4)
lninsur	0.0454 **(0.0194)	0.0346 *(0.0194)	0.0430 **(0.0192)	−0.0164 ***(0.0046)
lnbzmj	−0.3637 ***(0.1241)	−0.7309 ***(0.1619)	−0.5899 ***(0.1943)	0.2642 ***(0.0464)
lnngdp		0.0800(0.1285)	0.0942(0.1297)	0.2937 ***(0.0307)
lnnmcf		1.3837 ***(0.3596)	1.1987 ***(0.3846)	0.4211 ***(0.0918)
senlin			0.0191 **(0.0075)	−0.0071 ***(0.0018)
lnzai			−0.0647 **(0.0254)	0.0132 **(0.0060)
nyjxh			0.0431(0.0587)	0.0765 ***(0.0139)
secondst			−1.0452 **(0.4858)	0.3988 ***(0.1154)
lncarbonti				−0.0557 ***(0.0104)
constant	5.7196 ***(1.8763)	0.0092(2.7808)	−0.4384(3.0720)	4.9130 ***(0.7264)
year	controlled	controlled	controlled	controlled
$\sigma$	controlled	controlled	controlled	controlled
N	570	570	570	570
R2	0.9379	0.9402	0.9426	0.8341
Adjusted R2	0.9321	0.9343	0.9364	0.8160

Notes: The table reports the regression results from the OLS method with year and province fixed effects; *, **, and *** refer to the significance at the 10%, 5%, and 1% levels, respectively; standard errors are in the brackets.

Table 8

Results of the bootstrap method.

	Coefficients	Bootstrap Std. Err.	p Value	95% Conf. Interval
ind_eff	−0.002	0.001	0.033	−0.004	−0.001
dir_eff	−0.016	0.005	0.001	−0.026	−0.006

Notes: “ind_eff” is the indirect effect coefficient and “dir_eff” is the direct effect coefficient.

Table 9

Results of stepwise regression, including the instrumental variables.

Variable	lncarbon
	(1)	(2)	(3)
lninsur	−0.02 *(0.01)
lnsurmd		−0.03 *(0.01)
lnsurgd			−0.03 *(0.01)
control variables	controlled	controlled	controlled
year	controlled	controlled	controlled
$\sigma$	controlled	controlled	controlled
N	540	540	540
R2	0.8173	0.8134	0.8144
Adjusted R2	0.7966	0.7922	0.7933

Notes: The table reports the regression results from the OLS method with year and province fixed effects, including the instrumental variables; * refer to the significance at the 10%, level; standard errors are in the brackets.

Table 10

Results of sub-samples.

Variable	lncarbon
	(1)	(2)	(3)	(4)	(5)	(6)
	Main Grain-Producing Area	Non-Main Grain-Producing Area	Eastern Region	Central Region	Western Region	Northeast Region
lninsur	0.0067(0.0052)	−0.0161 **(0.0063)	−0.0477 ***(0.0109)	−0.0029(0.0055)	0.0022(0.0063)	0.0081(0.0073)
lnbzmj	1.2023 ***(0.0844)	0.2833 ***(0.0572)	0.1812**(0.0882)	0.5132 ***(0.0891)	0.1154(0.1477)	0.3947 ***(0.0895)
lnngdp	−0.0192(0.0592)	0.2270 ***(0.0478)	0.3669 ***(0.0898)	0.0139(0.0811)	−0.1413 **(0.0633)	0.0579(0.0520)
lnnmcf	0.4025 ***(0.1185)	0.2409 **(0.1156)	0.5128 ***(0.1903)	1.0141 ***(0.2797)	0.0316(0.1476)	0.1704(0.1604)
senlin	−0.0115 ***(0.0034)	−0.0073 ***(0.0019)	−0.0024(0.0041)	−0.0137(0.0084)	−0.0049(0.0031)	−0.0084 ***(0.0020)
lnzai	0.0041(0.0085)	0.0207 ***(0.0068)	0.0099(0.0093)	−0.0056(0.0097)	−0.0055(0.0099)	0.0410 ***(0.0106)
nyjxh	0.0726 ***(0.0170)	0.1821 ***(0.0251)	0.0185(0.0315)	0.0607 **(0.0233)	0.0253(0.0160)	0.1268 ***(0.0239)
secondst	0.4729 ***(0.1335)	0.5262 **(0.2339)	1.1708 **(0.4587)	0.4139 *(0.2082)	−0.7013 ***(0.1542)	−0.6513 ***(0.2486)
constant	−7.1904 ***(1.4129)	6.0245 ***(0.9093)	4.3577 **(1.8518)	6.2918 ***(1.6495)	14.1092 ***(2.7254)	−1.2337(2.5652)
year	controlled	controlled	controlled	controlled	controlled	controlled
$\sigma$	controlled	controlled	controlled	controlled	controlled	controlled
N	247	323	190	209	114	57
R2	0.9122	0.8738	0.7341	0.9424	0.9452	0.9943
AdjustedR2	0.8962	0.8548	0.6737	0.9304	0.9244	0.9886

Notes: The table reports the regression results from the OLS method with year and province fixed effects; *, ** and *** refers to the significance at the 10%, 5%, and 1% levels, respectively; standard errors are in the brackets.

References

1. Huang, X.; Xu, X.; Wang, Q.; Zhang, L.; Chen, L. Assessment of agricultural carbon emissions and their spatiotemporal changes in China, 1997-2016. Int. J. Environ. Res. Public Health; 2019; 16, 3105. [DOI: https://dx.doi.org/10.3390/ijerph16173105] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31455022]

2. Zhu, Y.; Huo, C. The impact of agricultural production efficiency on agricultural carbon emissions in China. Energies; 2022; 15, 4464. [DOI: https://dx.doi.org/10.3390/en15124464]

3. Yang, H.; Wang, X.; Bin, P. Agriculture carbon-emission reduction and changing factors behind agricultural eco-efficiency growth in China. J. Clean. Prod.; 2022; 334, 130193. [DOI: https://dx.doi.org/10.1016/j.jclepro.2021.130193]

4. Li, H.; Tang, M.; Cao, A.; Guo, L. Assessing the relationship between air pollution, agricultural insurance, and agricultural green total factor productivity: Evidence from China. Environ. Sci. Pollut. Res.; 2022; 29, pp. 78381-78395. [DOI: https://dx.doi.org/10.1007/s11356-022-21287-7] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35689771]

5. Xue, Y.; Cao, S.; Xu, Z.; Jin, T.; Jia, T.; Yan, C. Status and trends in application of technology to prevent plastic film residual pollution. J. Agro-Environ. Sci.; 2017; 36, pp. 1595-1600.

6. Wang, Y.; He, K.; Zhang, J.; Chang, H. Environmental knowledge, risk attitude, and households’ willingness to accept compensation for the application of degradable agricultural mulch film: Evidence from rural China. Sci. Total Environ.; 2020; 744, 140616. [DOI: https://dx.doi.org/10.1016/j.scitotenv.2020.140616]

7. Visser, M.; Jumare, H.; Brick, K. Risk preferences and poverty traps in the uptake of credit and insurance amongst small-scale farmers in South Africa. J. Econ. Behav. Organ.; 2020; 180, pp. 826-836. [DOI: https://dx.doi.org/10.1016/j.jebo.2019.05.007]

8. Tuo, G. How to promote agricultural insurance in China?. China Agric. Econ. Rev.; 2016; 8, pp. 194-205. [DOI: https://dx.doi.org/10.1108/CAER-01-2016-0016]

9. Ali, W.; Abdulai, A.; Mishra, A.K. Recent advances in the analyses of demand for agricultural insurance in developing and emerging countries. Annu. Rev. Resour. Econ.; 2020; 12, pp. 411-430. [DOI: https://dx.doi.org/10.1146/annurev-resource-110119-025306]

10. Altarhouni, A.; Danju, D.; Samour, A. Insurance market development, energy consumption, and Turkey’s CO₂ emissions. New perspectives from a bootstrap ARDL test. Energies; 2021; 14, 7830. [DOI: https://dx.doi.org/10.3390/en14237830]

11. Li, S.; Yu, W. The dual-driven impact of “internet+ agricultural insurance” on the agricultural carbon welfare performance in China. Pol. J. Environ. Stud.; 2022; 31, pp. 2183-2196. [DOI: https://dx.doi.org/10.15244/pjoes/141803]

12. Mu, L.; Wang, Y.; Mao, H. Does agricultural insurance drive variations in carbon emissions in China? Evidence from a quasi-experiment. Pol. J. Environ. Stud.; 2023; 32, pp. 653-665. [DOI: https://dx.doi.org/10.15244/pjoes/155869] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35393951]

13. Lin, B.; Ma, R. Green technology innovations, urban innovation environment and CO₂ emission reduction in China: Fresh evidence from a partially linear functional-coefficient panel model. Technol. Forecast. Soc. Change; 2022; 176, 121434. [DOI: https://dx.doi.org/10.1016/j.techfore.2021.121434]

14. Somaini, P.; Wolak, F.A. An algorithm to estimate the two-way fixed effects model. J. Econom. Methods; 2016; 5, pp. 143-152. [DOI: https://dx.doi.org/10.1515/jem-2014-0008]

15. Kropko, J.; Kubinec, R. Interpretation and identification of within-unit and cross-sectional variation in panel data models. PloS ONE; 2020; 15, e0231349. [DOI: https://dx.doi.org/10.1371/journal.pone.0231349]

16. Goodnight, J.H. Tests of hypotheses in fixed effects linear models. Commun. Stat. Theory Methods; 1980; 9, pp. 167-180. [DOI: https://dx.doi.org/10.1080/03610928008827869]

17. Borenstein, M.; Hedges, L.V.; Higgins, J.P.; Rothstein, H.R. A basic introduction to fixed-effect and random-effects models for meta-analysis. Res. Synth. Methods; 2010; 1, pp. 97-111. [DOI: https://dx.doi.org/10.1002/jrsm.12]

18. Saeed, S.; Moodie, E.E.; Strumpf, E.C.; Klein, M.B. Evaluating the impact of health policies: Using a difference-in-differences approach. Int. J. Public Health; 2019; 64, pp. 637-642. [DOI: https://dx.doi.org/10.1007/s00038-018-1195-2]

19. Liu, X.; Zhang, S.; Bae, J. The nexus of renewable energy-agriculture-environment in BRICS. Appl. Energy; 2017; 204, pp. 489-496. [DOI: https://dx.doi.org/10.1016/j.apenergy.2017.07.077]

20. Nurse, J.; Basher, D.; Bone, A.; Bird, W. An ecological approach to promoting population mental health and well-being—A response to the challenge of climate change. Perspect. Public Health; 2010; 130, pp. 27-33. [DOI: https://dx.doi.org/10.1177/1757913909355221]

21. Ma, J.; Chen, S. Does land transfer have an impact on land use efficiency? A case study on rural China. Natl. Account. Rev.; 2022; 4, pp. 112-134. [DOI: https://dx.doi.org/10.3934/NAR.2022007]

22. Zhu, J.; Wang, M.; Zhang, C. Impact of high-standard basic farmland construction policies on agricultural eco-efficiency: Case of China. Natl. Account. Rev.; 2022; 4, pp. 147-166. [DOI: https://dx.doi.org/10.3934/NAR.2022009]

23. Smith, V.H.; Glauber, J.W. Agricultural insurance in developed countries: Where have we been and where are we going?. Appl. Econ. Perspect. Policy; 2012; 34, pp. 363-390. [DOI: https://dx.doi.org/10.1093/aepp/pps029]

24. Wang, M.; Shi, P.; Ye, T.; Liu, M.; Zhou, M. Agriculture insurance in China: History, experience, and lessons learned. Int. J. Disaster Risk Sci.; 2011; 2, pp. 10-22. [DOI: https://dx.doi.org/10.1007/s13753-011-0007-6]

25. Fang, L.; Hu, R.; Mao, H.; Chen, S. How crop insurance influences agricultural green total factor productivity: Evidence from Chinese farmers. J. Clean. Prod.; 2021; 321, 128977. [DOI: https://dx.doi.org/10.1016/j.jclepro.2021.128977]

26. Appiah-Otoo, I.; Acheampong, A.O. Does insurance sector development improve environmental quality? Evidence from BRICS. Environ. Sci. Pollut. Res.; 2021; 28, pp. 29432-29444. [DOI: https://dx.doi.org/10.1007/s11356-021-12760-w]

27. Mardani, A.; Streimikiene, D.; Cavallaro, F.; Loganathan, N.; Khoshnoudi, M. Carbon dioxide (CO₂) emissions and economic growth: A systematic review of two decades of research from 1995 to 2017. Sci. Total Environ.; 2019; 649, pp. 31-49. [DOI: https://dx.doi.org/10.1016/j.scitotenv.2018.08.229]

28. Nilwala, W.M.; Jayarathna, R.S.L. Agriculture insurance for sustainable development in Sri Lanka Evidence from Kurunegala district. Proceedings of the 15th International Conference on Business Management; Colombo, Sri Lanka, 6 December 2018.

29. Baerenklau, K.A. Some simulation results for a green insurance mechanism. J. Agric. Resour. Econ.; 2005; 30, pp. 94-108.

30. Mills, E. The greening of insurance. Science; 2012; 338, pp. 1424-1425. [DOI: https://dx.doi.org/10.1126/science.1229351]

31. Zona, R.; Roll, K.; Law, Z. Sustainable/Green Insurance Products. Casualty Actuarial Society E-Forum; Casualty Actuarial Society: Arlington, VA, USA, 2014.

32. Ahmed, N.; Hamid, Z.; Mahboob, F.; Rehman, K.U.; Ali, M.S.E.; Senkus, P.; Skrzypek, A. Causal linkage among agricultural insurance, air pollution, and agricultural green total factor productivity in United States: Pairwise Granger causality approach. Agriculture; 2022; 12, 1320. [DOI: https://dx.doi.org/10.3390/agriculture12091320]

33. Lanjouw, J.O.; Mody, A. Innovation and the international diffusion of environmentally responsive technology. Res. Policy; 1996; 25, pp. 549-571. [DOI: https://dx.doi.org/10.1016/0048-7333(95)00853-5]

34. Gallagher, K.S.; Anadon, L.D.; Kempener, R.; Wilson, C. Trends in investments in global energy research, development, and demonstration. Wiley Interdiscip. Rev. Clim. Chang.; 2011; 2, pp. 373-396. [DOI: https://dx.doi.org/10.1002/wcc.112]

35. Fagerberg, J.; Srholec, M. National innovation systems, capabilities and economic development. Res. Policy; 2008; 37, pp. 1417-1435. [DOI: https://dx.doi.org/10.1016/j.respol.2008.06.003]

36. Doranova, A.; Costa, I.; Duysters, G. Knowledge base determinants of technology sourcing in clean development mechanism projects. Energy Policy; 2010; 38, pp. 5550-5559. [DOI: https://dx.doi.org/10.1016/j.enpol.2010.04.055]

37. Li, T.; Li, X.; Albitar, K. Threshold effects of financialization on enterprise R&D innovation: A comparison research on heterogeneity. Quant. Financ. Econ.; 2021; 5, pp. 496-515.

38. Li, T.; Wen, J.; Zeng, D.; Liu, K. Has enterprise digital transformation improved the efficiency of enterprise technological innovation? A case study on Chinese listed companies. Math. Biosci. Eng.; 2022; 19, pp. 12632-12654. [DOI: https://dx.doi.org/10.3934/mbe.2022590]

39. Liu, Y.; Failler, P.; Liu, Z. Impact of Environmental Regulations on Energy Efficiency: A Case Study of China’s Air Pollution Prevention and Control Action Plan. Sustainability; 2022; 14, 3168. [DOI: https://dx.doi.org/10.3390/su14063168]

40. Guo, L.; Zhao, S.; Song, Y.; Tang, M.; Li, H. Green finance, chemical fertilizer use and carbon emissions from agricultural production. Agriculture; 2022; 12, 313. [DOI: https://dx.doi.org/10.3390/agriculture12030313]

41. Liu, Y.; Feng, C. What drives the decoupling between economic growth and energy-related CO₂ emissions in China’s agricultural sector?. Environ. Sci. Pollut. Res.; 2021; 28, pp. 44165-44182. [DOI: https://dx.doi.org/10.1007/s11356-021-13508-2]

42. Norris, P.E.; Batie, S.S. Virginia farmers’ soil conservation decisions: An application of Tobit analysis. J. Agric. Appl. Econ.; 1987; 19, pp. 79-90. [DOI: https://dx.doi.org/10.1017/S0081305200017404]

43. Fan, X.; Zhang, W.; Chen, W.; Chen, B. Land-water-energy nexus in agricultural management for greenhouse gas mitigation. Appl. Energy; 2020; 265, 114796. [DOI: https://dx.doi.org/10.1016/j.apenergy.2020.114796]

44. Liu, Y.; Ruiz-Menjivar, J.; Zhang, L.; Zhang, J.; Swisher, M.E. Technical training and rice farmers’ adoption of low-carbon management practices: The case of soil testing and formulated fertilization technologies in Hubei, China. J. Clean. Prod.; 2019; 226, pp. 454-462. [DOI: https://dx.doi.org/10.1016/j.jclepro.2019.04.026]

45. Liu, L.; Zhu, Y.; Guo, S. The evolutionary game analysis of multiple stakeholders in the low-carbon agricultural innovation diffusion. Complexity; 2020; 2020, 6309545. [DOI: https://dx.doi.org/10.1155/2020/6309545]

46. Perosa, B.; Newton, P.; Carrer, M.J. Access to information affects the adoption of integrated systems by farmers in Brazil. Land Use Policy; 2021; 106, 105459. [DOI: https://dx.doi.org/10.1016/j.landusepol.2021.105459]

47. Yang, Q.; Zhu, Y.; Wang, F. Social media participation, low-carbon agricultural practices, and economic performance of banana farmers in Southern China. Front. Psychol.; 2021; 12, 790808. [DOI: https://dx.doi.org/10.3389/fpsyg.2021.790808] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34975688]

48. Lin, M. The linkage mechanism between urban intelligence and low carbon innovation. Front. Eng. Manag.; 2022; 6, pp. 584-586. [DOI: https://dx.doi.org/10.1007/s42524-019-0076-6]

49. Khattak, S.I.; Ahmad, M.; ul Haq, Z.; Shaofu, G.; Hang, J. On the goals of sustainable production and the conditions of environmental sustainability: Does cyclical innovation in green and sustainable technologies determine carbon dioxide emissions in G-7 economies. Sustain. Prod. Consum.; 2022; 29, pp. 406-420. [DOI: https://dx.doi.org/10.1016/j.spc.2021.10.022]

50. Yii, K.J.; Geetha, C. The nexus between technology innovation and CO₂ emissions in Malaysia: Evidence from Granger causality test. Energy Procedia; 2017; 105, pp. 3118-3124. [DOI: https://dx.doi.org/10.1016/j.egypro.2017.03.654]

51. Du, K.; Li, P.; Yan, Z. Do green technology innovations contribute to carbon dioxide emission reduction? Empirical evidence from patent data. Technol. Forecast. Soc. Chang.; 2019; 146, pp. 297-303. [DOI: https://dx.doi.org/10.1016/j.techfore.2019.06.010]

52. Liguo, X.; Ahmad, M.; Khattak, S.I. Impact of innovation in marine energy generation, distribution, or transmission-related technologies on carbon dioxide emissions in the United States. Renew. Sustain. Energy Rev.; 2022; 159, 112225. [DOI: https://dx.doi.org/10.1016/j.rser.2022.112225]

53. Woodard, J.D.; Pavlista, A.D.; Schnitkey, G.D.; Burgener, P.A.; Ward, K.A. Government Insurance Program Design, Incentive Effects, and Technology Adoption: The Case of Skip-Row Crop Insurance. Am. J. Agric. Econ.; 2012; 94, pp. 823-837. [DOI: https://dx.doi.org/10.1093/ajae/aas018]

54. Acemoglu, D.; Cutler, D.; Finkelstein, A.; Linn, J. Did Medicare induce pharmaceutical innovation?. Am. Econ. Rev.; 2006; 96, pp. 103-107. [DOI: https://dx.doi.org/10.1257/000282806777211766] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29131561]

55. Blume-Kohout, M.E.; Sood, N. Market size and innovation: Effects of medicare part D on pharmaceutical research and development. J. Public Econ.; 2013; 97, pp. 327-336. [DOI: https://dx.doi.org/10.1016/j.jpubeco.2012.10.003] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/23869113]

56. Giné, X.; Yang, D. Insurance, credit, and technology adoption: Field experimental evidence from Malawi. J. Dev. Econ.; 2009; 89, pp. 1-11. [DOI: https://dx.doi.org/10.1016/j.jdeveco.2008.09.007]

57. Carter, M.R.; Cheng, L.; Sarris, A. Where and how index insurance can boost the adoption of improved agricultural technologies. J. Dev. Econ.; 2016; 118, pp. 59-71. [DOI: https://dx.doi.org/10.1016/j.jdeveco.2015.08.008]

58. Miao, R. Climate, insurance and innovation: The case of drought and innovations in drought-tolerant traits in US agriculture. Eur. Rev. Agric. Econ.; 2020; 47, pp. 1826-1860. [DOI: https://dx.doi.org/10.1093/erae/jbaa010]

59. Cai, J. The impact of insurance provision on household production and financial decisions. Am. Econ. J. Econ. Policy; 2016; 8, pp. 44-88. [DOI: https://dx.doi.org/10.1257/pol.20130371]

60. Gunnsteinsson, S. Experimental identification of asymmetric information: Evidence on crop insurance in the Philippines. J. Dev. Econ.; 2020; 144, 102414. [DOI: https://dx.doi.org/10.1016/j.jdeveco.2019.102414]

61. Gu, H.Y.; Wang, C.W. Impacts of the COVID-19 pandemic on vegetable production and countermeasures from an agricultural insurance perspective. J. Integr. Agric.; 2020; 19, pp. 2866-2876. [DOI: https://dx.doi.org/10.1016/S2095-3119(20)63429-3]

62. Gao, Y.; Shu, Y.; Cao, H.; Zhou, S.; Shi, S. Fiscal policy dilemma in resolving agricultural risks: Evidence from China’s agricultural insurance subsidy pilot. Int. J. Environ. Res. Public Health; 2021; 18, 7577. [DOI: https://dx.doi.org/10.3390/ijerph18147577]

63. The Data of “China Statistical Yearbook”. Available online: http://www.stats.gov.cn/tjsj./ndsj/ (accessed on 15 February 2023).

64. The Data of “China Insurance Statistical Yearbook”. Available online: https://www.epsnet.com.cn (accessed on 16 November 2021).

65. The Website of the Incopat Database. Available online: https://www.incopat.com/newLogin (accessed on 31 March 2022).

66. Popp, D.C. The effect of new technology on energy consumption. Resour. Energy Econ.; 2001; 23, pp. 215-239. [DOI: https://dx.doi.org/10.1016/S0928-7655(00)00045-2]

67. Wei, W.X.; Fang, Y. The impact of technological progress on China’s CO₂ emissions. Stat. Res.; 2010; 27, 9.(In Chinese)

68. According to the Definition Given by National Bureau of Statistics in China, Secondary Sectors Refer to the Mining Industry (Excluding Mining Assistance Activities), Manufacturing (Excluding Metal Products, Machinery and Equipment Repair), Electricity, Heat, Gas and Water Production and Supply, and Construction. Available online: http://www.stats.gov.cn/tjsj/tjbz/201301/t20130114_8675.html (accessed on 18 November 2022).

69. Tang, L.; Luo, X. Can agricultural insurance encourage farmers to apply biological pesticides? Evidence from rural China. Food Policy; 2021; 105, 102174. [DOI: https://dx.doi.org/10.1016/j.foodpol.2021.102174]