*Article* **Influencing Factors and Path Analysis of Sustainable Agricultural Mechanization: Econometric Evidence from Hubei, China**

**Zhi Li 1,2, Ming Zhu <sup>1</sup> , Huang Huang <sup>1</sup> , Yu Yi 3,\* and Jingyi Fu <sup>4</sup>**


**Abstract:** The importance of supporting agricultural mechanization in agri-food supply chains to achieve agricultural and rural development has been comprehensively recognized. There has been a surge in the attention given to Sustainable Agricultural Mechanization (SAM) in the context of developing countries. However, it is important to address the major challenge of studying the important factors and the influencing path of SAM. As a representative province of China's agricultural development, Hubei has developed significantly in terms of agricultural mechanization in the past 20 years. Therefore, using a literature review, representative field survey data, and statistical analytical approaches, 28 relevant factors related to SAM were extracted, and the main influencing factors of SAM were determined by building an integrative conceptual framework and using the corresponding structural equation model based on partial least squares (PLS-SEM). The relationships and influencing paths between the factors were analyzed, and a confirmatory measurement model and a structural model of the effects on sustainable agricultural mechanization were constructed. The results show that (1) the PLS-SEM model fits the experimental data well and can effectively reflect the relationships among factors in this complex system; (2) within the factors influencing the development level of SAM in Hubei, China, the economic factors have the greatest weight, whereas government policy factors are the core elements promoting development, and environmental factors are the most noteworthy outcome factors; and (3) economic and policy factors play a very obvious role in promoting SAM through the influencing paths of agricultural production and agricultural machinery production and sales. Ultimately, corresponding suggestions have been put forward for decisions regarding the implementation of SAM for similar countries and regions.

**Keywords:** sustainable; agricultural mechanization; structural equation model (SEM); partial least square (PLS); affecting factors; agri-food supply chain

#### **1. Introduction**

Mechanization is a crucial input for agricultural crop production and one that historically has been neglected in the context of developing countries [1], especially in sub-Saharan Africa, Southeast Asia, South Asia, and Latin America. Mechanization contributes significantly to the development of food supply chains through improved agricultural practices for increased production and enhanced food security. It eases and reduces hard labor, relieves labor shortage, and improves the productivity and timeliness of agricultural operations [2,3].

The issue of Sustainable Agricultural Mechanization (SAM) has received considerable critical attention in recent years. The research of SAM continues the typical paradigm

**Citation:** Li, Z.; Zhu, M.; Huang, H.; Yi, Y.; Fu, J. Influencing Factors and Path Analysis of Sustainable Agricultural Mechanization: Econometric Evidence from Hubei, China. *Sustainability* **2022**, *14*, 4518. https://doi.org/10.3390/ su14084518

Academic Editor: Aaron K. Hoshide

Received: 7 March 2022 Accepted: 8 April 2022 Published: 11 April 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

of sustainable agriculture [4], wherein SAM can be described as mechanization that is economically viable, environmentally sensitive, and socially acceptable [3]. The United Nations (UN) Food and Agriculture Organization (FAO) SAM website noted that sustainable mechanization is important, as farmers who have access to improved agricultural tools and powered technologies can shift from subsistence farming to more market-oriented farming, making the agricultural sector more attractive to rural youth [2]. SAM can improve the efficient use of resources, enhance market access, and contribute to mitigating climate-related hazards, as it has the potential to render producing, processing, and marketing activities and functions more efficient, economically feasible, socially acceptable, and environmentally friendly [5]. As the effects of climate change and natural resource depletion become more visible, sustainable mechanization has adopted Conservation Agriculture principles, and the "Save and Grow" paradigm-which aims to protect the soil, use less energy, and encourage more efficient and precise use of inputs-will be essential to maintain and sustainably improve food production and distribution. Analyses of SAM have to account for not just the technical, economical, and engineering aspects, but also the linkages and inter-dependencies with other sectors, such as social, environmental, cultural, and policy aspects, and consider their role when contributing to the sustainable development of the food and agriculture sector. Overcoming the environmental and social challenges of today is not an isolated action but is part of a comprehensive view of agriculture that considers efficiency and ecology [2].

As a representative of developing countries, the agricultural mechanization of China has made remarkable achievements in the past 20 years. The comprehensive mechanization rate of crop cultivation and harvesting in China has risen from 45.8% in 2008 to 71.3% in 2020, an average annual increase of about 2%. However, with the slowing of economic growth and the "The New Normal" of agriculture, the development of agricultural mechanization has recently faced many unsustainable problems [6]. The challenges to agricultural machinery and equipment, production technology, and professional and technical personnel are structural shortages; issues with public service funding and an insufficient effective supply of social service systems and policy support increase pressure on agricultural resources, the ecological environment, and the cost of agricultural machinery [7]. Therefore, this present research focuses on the role of the influencing factors, the development paths, and the development mode of SAM.

The purpose of this study was to take Hubei, a typical region of China, as an example for empirical analysis to estimate the effect of various factors and the development paths of SAM in an integrated analytic framework. The results will enable us to understand the mutual influence of SAM on agriculture, society, the economy, and the environment and can be used to help policymakers and project implementers of agricultural machinery purchase subsidy policies and further formulate and implement their policies' strategy and development path, thus promoting steady and efficient improvements in SAM.

#### **2. Literature Review and Conceptual Framework**

#### *2.1. Literature Review*

In line with new efforts and opportunities to promote mechanization, there is a growing body of empirical research on the topic of SAM. Research on adaptation to SAM is diverse but mainly focuses on two aspects: (1) the relationship between SAM and economic, environmental, and social sustainability and policy factors; (2) the influencing factors of mechanization development and effective implementation. These two aspects complement each other.

As a sub-element of sustainable development of agriculture, SAM is bound to interact with multiple systems. Some scholars have tried to explore the agronomic, environmental, and socioeconomic effects of mechanization, thereby revealing linkages and trade-offs. For example, the economy has a driving effect on mechanization, which is a direct requirement for improving agricultural output; mechanization is bound to have an impact on the environment, and the machinery industry can promote mechanization [8–12]. Some

research has given a voice to the rural population in Africa regarding mechanization and allowed researchers to identify causal impact chains [13]. Other scholars have researched and analyzed the effects of policy formulation. Governments must create an enabling environment to allow the multiple dimensions of SAM to develop. This policy environment includes mechanization policy instruments, including appropriate short-term subsidies for purchasing and leasing equipment [14,15], and law [16]. Sustainability requires the mechanization pathways promoted through policies to be thought through carefully. Formulating adaptation strategies or frameworks are the most common means used by governments to carry out SAM actions, which can guide countries or regions [17]. According to different national conditions, some countries have issued national promotion policies or laws to guide practice, while others have issued action plans that match the strategies. Some studies have also empirically analyzed the relationships of agricultural mechanization with agricultural carbon emissions [18–20], green agricultural transformation [21–23], a low-carbon economy, and food safety [24]. Table 1 lists various agricultural sustainabilityand SAM-related policies introduced by developing countries in the past two decades.


**Table 1.** Agricultural sustainability/SAM related policies.

With the continuous deepening of SAM research, scholars have begun to pay attention to the influencing factors of SAM. Few previous studies have looked at the potential effects of mechanization empirically but rather have mostly focused on yields and labor alone [13]. However, the factors involved in SAM are likely to be more complex. However, because of the differences in the research objects, research perspectives, or sample selection, the conclusions of the different studies are different. In China, the research related to SAM can be roughly divided into three categories: (1) qualitative policy analysis [7]; (2) mechanization as a sub-element of agricultural sustainability [10,25]; and (3) a discussion of factors related to SAM, including the environment [11,21,22], agricultural carbon emissions [19,20,23], mechanization level [14,26,27], agricultural machinery industry [18], etc. However, there is a lack of quantitative and systematic research on SAM in China.

This study focused on the interactions among SAM factors and undertook an overall and systematic quantitative empirical study to make up for the shortcomings in the existing literature. At the same time, an analysis system covering education and training, science and technology, and other influencing factors was constructed, which expanded the scope of influencing factors and the path of research by including the ecological environment in the influencing factors of agricultural mechanization. This part of the research is an important complement to the existing literature on SAM. These two aspects are the important innovation points of this study, which are different from those in previous studies.

#### *2.2. Analysis of the Influencing Factors of SAM*

There is a wide range of factors affecting SAM. Each country has different land conditions, planting bases, and climate backgrounds, and there are great differences in the mechanization process. Therefore, research on the development mode mechanization and appropriate strategies should ensure the application of mechanization theory at the decomposition level. The types of strategies needed to promote the development of SAM must account for the conditions of specific sites, each of the factors and the mechanisms, and the extent to which these influence SAM will vary from country to country, potentially even within countries. According to investigation and research, literature reviews, and policy analyses, combined with the actual situation in different regions, it can be concluded that the factors affecting the SAM include those summarized in Table 2. Of course, one should not ignore that there are potentially several adverse propositions that have emerged from using agricultural mechanization, such as "mechanization leads to unemployment" or "smallholders cannot benefit from mechanization" (particularly in developing countries) [3,28]. Such topics also can affect policies and programs regarding mechanization.

**Table 2.** Influencing factors of SAM.


From the analysis above, it can be seen that the factors affecting the sustainable development of agricultural mechanization mainly include economics, society, the population and labor force, agricultural production, land resources, industrial technology development, education, the energy environment, ecology, and policies and regulations. There are many corresponding component indicators with each aspect. The relationships are also more complicated, and the mutually influencing relationship paths are often not clear, so traditional methods of research are more difficult. Therefore, this article focused on the use of structured statistical research methods to comprehensively and quantitatively analyze the relationships among the influencing factors of agricultural mechanization and the path and intensity, as well as to quantitatively verify the conclusions of the qualitative analysis. According to the analysis above and index-selection principles, 28 representative indicators were finally selected from the different categories (socioeconomic, environmental, production and land resource, agricultural machinery industry and technology, agricultural mechanization status and policy support, etc.).

#### **3. Materials and Methods**

#### *3.1. Research Area and Data Sources*

The regional area of Hubei Province (185,900 km<sup>2</sup> ) is equivalent to that of a mediumsized developing country, such as Uganda, Ghana, or Cambodia. The terrain includes plains, hills, mountains, and lakes. There are various agricultural planting operations, and they have been dominated by small farmers and small business owners for a long time. The development strategy of SAM is highly typical of quite a few developing countries. The original data of Hubei Province collected in this article came from China Statistical Yearbook, China Agricultural Machinery Industry Yearbook, Hubei Statistical Yearbook, Hubei Rural Statistical Yearbook, and some field investigations. For some of the missing data and unreasonable data, we estimated the missing values through mean replacement and regression interpolation, then completed data preprocessing and finally obtained 392 valid data for the 28 measurement indicators used in this article. The descriptive statistical results of indicators data are shown in Table A1 (see in Appendix A).

To eliminate the effects of the different orders of magnitude and dimensions of different variables, the data of all variables were standardized. The method used for standardization of the variables was the Min-Max standardization method [14]. That is, all variables were transformed linearly. If MinX and MaxX are the minimum and maximum values of variable X, after standardization, X' = (X − MinX)/(MaxX − MinX). It is also difficult to deal with the complexity of SAM via traditional methods. Furthermore, in this study, there were several latent influencing variables (latent variables) of practical significance for agricultural mechanization, and there were also several different observation variables or manifest variables for each latent variable, which may have also affected other latent variables. These can be influenced by the internal and external relationships of SAM within the model, and it was necessary to evaluate the influencing relationships and size from different aspects. The six aspects of the influencing factors can be regarded as latent variables, and the influencing factors themselves can be regarded as manifest variables. This article established the latent variables as economic and population factors (EP), agricultural production (AP), the agricultural mechanization development level (AMDL), the agricultural machinery industry and agricultural technology (AMIAT), policies (P), and the environment (E). The final results are shown in Table 3.


#### **Table 3.** Impact factors of SAM.

#### *3.2. Basic Hypotheses*

Our research set the first-level indicators, divided their corresponding explicit variables, and established the causal relationships among latent variables. Since the assignment of indicators and the setting process of causality are subjective and referential, the set construction and adjustment process relied on the overall assumptions of the model described in the following hypotheses:

**Hypothesis 1.** *The correlation between the latent variable and its corresponding explicit variable can be expressed by linear equations; the latent variables do not cross each other in the theoretical sense.*

**Hypothesis 2.** *According to the actual meaning of the selected indicators, the selected latent variables are directly related to each other, and they may have indirect secondary path effects through other latent variables.*

According to the influencing factors and the related relationships analyzed in the literature review, the following assumptions were put forward: The impact of agricultural mechanization and agricultural economic development has a strong two-way positive effect. Conversely, to promote the development of agricultural mechanization, capital investment is indispensable. At the same time, the development of the agricultural machinery industry is an important basic guarantee for the sustainable development of agricultural mechanization. The development of agricultural mechanization and the agricultural machinery

industry complement each other. In addition, making the input of agricultural machinery produce real profits and increasing wealth by using production machinery is the only way to encourage agricultural machinery users to further invest and expand production [38]. Thus, based on the above view, we hypothesize:

**Hypothesis 2a.** *The agricultural mechanization development level (AMDL) is affected by economic factors, agricultural production, policy, and the agricultural machinery industry and agricultural technology (AMIAT) factor.*

The healthy development of agricultural mechanization can directly increase the output and efficiency of agricultural workers, directly increase labor income, and stimulate the overall development of the agricultural economy [9]. Many studies have also discussed the impact of policies and the agricultural machinery industry on the growth of agricultural products [6–8,12]. Thus, based on the above view, we hypothesize:

#### **Hypothesis 2b.** *Economic factors, policy factors, the AMDL, and AMIAT can promote agricultural production.*

With their rapid development, modern science and technology have become widely used in agricultural production, including high-tech informatization and intelligent agricultural machinery used in innovative crop production methods. Improved machinery operation capabilities are used to implement precision production operations, saving labor while improving efficiency. It is no doubt that science and technology play a key role in the development of modern agricultural mechanization. At the same time, the agricultural machinery industry must strive to improve its innovation and investment, which is an important new growth point to realize the development of SAM for developing countries [39]. Moreover, the implementation of China's Agricultural Mechanization Promotion Law in 2004 and the subsidy policy for the purchase of agricultural machinery in 1998 played significant roles in improving the agricultural machinery industry and agricultural mechanization [14]. Thus, based on the above view, we hypothesize:

#### **Hypothesis 2c.** *AMIAT is positively correlated with economy and policy.*

Physical limits to land and water availability within ecosystems are often worsened by climate change. By including SAM in its projects, FAO promotes conservation agriculture practices that contribute to soil conservation and water use efficiency [2]. The development of SAM must be organically combined with energy-conservation technology, emission control, and ecological protection, and must strive to achieve harmonious coexistence between human activities and nature. To advocate for a green economy [11,19,20], there needs to be active promotion by the government. We are sure that the change in the environment must be the product of a comprehensive effect [16]. Thus, based on the above view, we hypothesize:

### **Hypothesis 2d.** *Environmental factors are affected by economic factors, agricultural production, AMIAT, AMDL, and policy factors at the same time.*

Relevant national policies and regulations can ensure that capital investment and subsidy policies can effectively reduce purchasing costs so that they can effectively promote the sound and rapid development of agricultural mechanization and the agricultural machinery industry, which is one of the main ways to effectively promote the popularization and extension of agricultural machinery [7]. Meanwhile, policies and regulations can also manage and coordinate various development goals and promote the balanced development of society. Therefore, national policies provide strong support and guarantee the sustainable development of agricultural mechanization. Thus, based on the above view, we hypothesize:

#### **Hypothesis 2e.** *Economic factors are affected by policy factors.*

Based on these assumptions, this research first established an initial path graph structure in a fully connected form and then continuously made corrections based on the analysis results to create the final improved model. In the establishment of the measurement model, the corresponding relationships and influencing paths between the observed variables and latent variables were set according to the actual meaning of the indicators. All the indicators were then matched to the latent variables to achieve a causal equilibrium. The initial hypothesis structure is shown in Figure 1. *Sustainability* **2022**, *14*, x FOR PEER REVIEW 9 of 21

**Figure 1.** Hypothetical structural equation framework of SAM. **Figure 1.** Hypothetical structural equation framework of SAM.

#### *3.3. Statistical Modeling Methods*

*3.3. Statistical Modeling Methods*  Traditional statistical analysis methods, such as linear regression and principal component analysis, cannot effectively deal with these latent variables, but they can be studied with the help of structural equations. Structural equation modeling (SEM) is a systematic analysis method that integrates factor analysis and path analysis. SEM has the advantages of simultaneously processing multiple dependent variables, allowing independent variables and dependent variables to contain measurement errors; estimating factor structures and factor relationships; and estimating the fitting degree of the whole model [14]. It uses Traditional statistical analysis methods, such as linear regression and principal component analysis, cannot effectively deal with these latent variables, but they can be studied with the help of structural equations. Structural equation modeling (SEM) is a systematic analysis method that integrates factor analysis and path analysis. SEM has the advantages of simultaneously processing multiple dependent variables, allowing independent variables and dependent variables to contain measurement errors; estimating factor structures and factor relationships; and estimating the fitting degree of the whole model [14]. It uses a structure of linear equations to deal with the relationships between manifest variables and latent variables and the relationships between latent variables.

a structure of linear equations to deal with the relationships between manifest variables and latent variables and the relationships between latent variables. SEM can be divided into two types according to the nature and relationship characteristics of the variables. One is the measurement model, and the other is the structural model, which uses a similar path-analysis method to establish the structural relationships SEM can be divided into two types according to the nature and relationship characteristics of the variables. One is the measurement model, and the other is the structural model, which uses a similar path-analysis method to establish the structural relationships between latent variables. The following equations show the specific forms of the measurement model and the structural model. The measurement model is [40]:

between latent variables. The following equations show the specific forms of the measure-

In Formula (1), x is a ×1 dimensional vector formed by exogenous manifest variables, is an ×1 dimensional vector formed by exogenous latent variables, Λ௫ is a × dimensional load matrix, and is a ×1 dimensional vector composed of measurement errors. In Formula (2), y is a ×1 dimensional vector formed by exogenous manifest variables, is an ×1 dimensional vector formed by exogenous latent variables, Λ௬ is a × dimensional load matrix, and is a ×1 dimensional

x=Λ௫+ (1)

y = Λ௬+ (2)

vector composed of measurement errors.

The structural model is:

ment model and the structural model. The measurement model is [40]:

$$\mathbf{x} = \Lambda\_{\mathbf{x}} \mathfrak{f} + \delta \tag{1}$$

$$\mathbf{y} = \Lambda\_y \boldsymbol{\eta} + \boldsymbol{\varepsilon} \tag{2}$$

In Formula (1), x is a *p* × 1 dimensional vector formed by *p* exogenous manifest variables, *ξ* is an *m* × 1 dimensional vector formed by *m* exogenous latent variables, Λ*<sup>x</sup>* is a *p* × *m* dimensional load matrix, and *δ* is a *p* × 1 dimensional vector composed of *p* measurement errors. In Formula (2), y is a *q* × 1 dimensional vector formed by *q* exogenous manifest variables, *η* is an *n* × 1 dimensional vector formed by *n* exogenous latent variables, Λ*<sup>y</sup>* is a *q* × *n* dimensional load matrix, and *ε* is a *q* × 1 dimensional vector composed of *q* measurement errors.

The structural model is:

$$
\eta = \mathbf{B}\eta + \Gamma \mathfrak{F} + \zeta \tag{3}
$$

where B is an *n* × *n* dimensional correlation coefficient matrix, which is used to reflect the relationships among the various endogenous latent variables; Γ is an *n* × *m* dimensional correlation coefficient matrix, which is used to reflect the relationship between the exogenous latent variable *ξ* and the endogenous latent variable *η*; ζ is an *n* × 1 dimensional vector composed of interpretation errors. The correlation coefficient is a standardized path coefficient. This path coefficient is used to measure the degree of correlation between two variables and is generally used to indicate reliability, under the premise that when the significance of the path coefficient is larger, the indicator has a greater impact [41].

In a realistic structural model, the variables may be both dependent variables and independent variables, and there may be not only direct but also indirect relationships among the variables. Some causal variables will affect the result variable through one or more intermediate variables, which are called indirect effects. The path coefficient of the indirect path is the product of the direct path coefficient involved in each path. The meaning of the total effect is the sum of the result variables affected by the causal variables, which is expressed as the sum of the direct effects and indirect effects, which can be used to verify the rationality of the hypothetical effect through path analysis.

Due to the number of samples collected, the maximum likelihood estimation method was not used, but the partial least squares (PLSs) method based on nonparametric estimation was used, as it has no strict assumptions about the sample size and sample distribution. Using the PLSs method to solve the SEM can avoid the situation where the model cannot be recognized because of a non-positive definite covariance matrix, and the method is more extensive. In summary, the study finally conducted an empirical analysis of the factors affecting the SAM by using the PLS-SEM method.

#### **4. Results of the Case Study**

In this study, Smart-PLS3 [42] was used to estimate Equations (1)–(3) based on the standardized data. Model evaluation and testing used multiple test statistical indicators to carry out correlation reliability and validity tests, such as Cronbach's alpha coefficient (Cronbach's α) and composite reliability. Hair et al. stated that it is acceptable for Cronbach's α to be greater than 0.7 for verification purposes [43].

#### *4.1. Model Specification Tests*

The test results of the reliability and the goodness of fit of the model are shown in Table 4. It can be seen from Table 4 that the model passed the reliability and validity test. The outer loading of the measurement model and the path coefficient of the structural model were calculated by the PLSs method. Then, the bootstrapping method was used to test and evaluate the estimated results of the two coefficients, as shown in Tables A2 and A3.


It can be seen from Table A2 that the loading of each path in the measurement model

**Table 4.** Test result of reliability and goodness of model.

It can be seen from Table A2 that the loading of each path in the measurement model passed the significance test. The general test passing standard is that the significance level is 0.05, and the t-statistic is greater than 1.96. The parameter estimation results of the structural model are shown in Table A3. In Table A3, the estimated values of the direct path coefficients of four paths did not pass the significance test. As these direct relationships were not supported by the test results, these four paths were excluded from the model. Generally speaking, the path coefficient relates to the number, type, and nature of the observed variables corresponding to the latent variables. An insignificant path coefficient in the internal model does not prove that there is no causal relationship between these latent variables. The current variables and model settings were not enough to prove their relationship, and other latent variables can be used as intermediaries to supplement the path. Since the model passed the best test, the significant initial variables used when setting the measurement model did not change, but the causal relationships between the latent variables in the structural model were adjusted, and several insignificant paths were removed. The model was then tested again. The result was that in the revised structural model, all of the path coefficients also passed the significance test, and therefore, the model is desirable. The structural equation model obtained after the final adjustment is referred to as Model B, as shown in Figure 2. passed the significance test. The general test passing standard is that the significance level is 0.05, and the t‐statistic is greater than 1.96. The parameter estimation results of the struc‐ tural model are shown in Table A3. In Table A3, the estimated values of the direct path coefficients of four paths did not pass the significance test. As these direct relationships were not supported by the test results, these four paths were excluded from the model. Generally speaking, the path coefficient relates to the number, type, and nature of the observed variables corresponding to the latent variables. An insignificant path coefficient in the internal model does not prove that there is no causal relationship between these latent variables. The current variables and model settings were not enough to prove their relationship, and other latent variables can be used as intermediaries to supplement the path. Since the model passed the best test, the significant initial variables used when set‐ ting the measurement model did not change, but the causal relationships between the la‐ tent variables in the structural model were adjusted, and several insignificant paths were removed. The model was then tested again. The result was that in the revised structural model, all of the path coefficients also passed the significance test, and therefore, the model is desirable. The structural equation model obtained after the final adjustment is referred to as Model B, as shown in Figure 2.

**Figure 2.** Model B of the influencing factors of SAM in Hubei. **Figure 2.** Model B of the influencing factors of SAM in Hubei.

*4.2. Results of the First‐Order PLS‐SEM Model*

(SAM).

The calculation results of measurement Model B are shown in Table A4. The factor loadings in Table A4 indicate that most of the indicators have a higher explanatory degree,

measurement model's interpretation ability is good. The negative numbers reflect nega‐ tive correlations between the indicator and Sustainable Agricultural Mechanization

#### *4.2. Results of the First-Order PLS-SEM Model*

The calculation results of measurement Model B are shown in Table A4. The factor loadings in Table A4 indicate that most of the indicators have a higher explanatory degree, reflecting that the selection of indicators is more representative and indicating that the measurement model's interpretation ability is good. The negative numbers reflect negative correlations between the indicator and Sustainable Agricultural Mechanization (SAM).

According to measurement Model B, the latent variables of the original value of agricultural machinery, GDP, and agricultural machinery profit were the three most important economic factors, indicating that the overall economic environment and the economic conditions of the agricultural machinery industry are important economic factors of SAM. The agricultural machinery price index is negatively related to the economic factors, indicating that the higher the price of agricultural machinery, the lower the market for agricultural machinery, in line with the actual situation. The relationship between the SAM and the number of people in the labor force is also negatively related, consistent with theoretical analysis. It is also worth noting that the degree of education (the proportion of the population educated in junior high school) is about 0.78. Mechanization is related to the quality of workers, but with the popularization of education, the degree of relevance of the impact of this indicator is not particularly sensitive. The correlations of the three indicators of agricultural production are high, which indicates that the benefits of agricultural mechanization are obvious from the statistical point of view. It is noted that the correlations between several indicators of the agricultural machinery industry and scientific and technological factors are also strong, indicating that the contribution of scientific and technological input to SAM is increasing, and the previous qualitative analysis is verified.

In the indicator corresponding to the latent variable "agricultural level", most of the indicators' factor loadings are relatively large. The correlation coefficient of the number of households with agricultural machinery is about 0.76, which does not show a high correlation, indicating that SAM has slowly reflected the trend of increasing through quality and intensive development, rather than a simple absolute increase in quantity, which may also be reflected by the two negatively related indicators of the number of agricultural machinery service organizations and the number of agricultural technology extension agencies. The results for social services and the use of agricultural machinery technology to promote education suggest that the increase in agricultural mechanization in itself is not through popularization in terms of head counts but has been a gradual process of information dissemination regarding the precision and characteristics of SAM. The results of the structural model are shown in Table 5. The influencing factors are presented in Table 6, including the total effect of the change after considering the indirect effects.

**Table 5.** Causality and path coefficients of latent variables in SEM.


Note: Economic and population factors (EP), agricultural production (AP), the agricultural mechanization development level (AMDL), and the agricultural machinery industry and agricultural technology (AMIAT).


**Table 6.** Total effect of SAM in Model B.

Note: Economic and population factors (EP), agricultural production (AP), the agricultural mechanization development level (AMDL), and the agricultural machinery industry and agricultural technology (AMIAT).

The path coefficients in Tables 5 and 6 describe whether there is a causal relationship between a pair of latent variables. If the path coefficient is small, the causal relationship reflected by that path may not exist. Of course, the paths in Table 5 only show the direct effects between the latent variable, that is, the direct impact of the cause variable on the result variable. Table 6 reflects the combination of direct and indirect effects of some of the latent variables.

#### *4.3. Results of Hypothesis Testing*

The results of the PLS-SEM analysis showed that agricultural mechanization is directly related to the six dimensions of economy, population, agricultural production, the agricultural machinery industry and agricultural technology, the environment, and policies. It can be seen from Figure 3 that the level of Agricultural Mechanization Development Level (AMDL) is obviously promoted by the economic, population, and policy factors. The economic and population factors are the most critical factors affecting SAM. The effect of the agricultural machinery industry and agricultural technology (AMIAT) on SAM is not statistically obvious.

Secondly, the effects of policy on agricultural machinery (mainly referring to the agricultural machinery purchase subsidy policy) are not significant, but the path regression coefficient of the overall impact of the effect is 0.9725, and that is a strong positive correlation. In terms of agricultural production, the positive effect of economic factors on agricultural production input and output is obvious. The direct effect of policy factors on agricultural production is negatively correlated, but the total effect is still a relatively large positive correlation. The effects of the agricultural machinery industry and agricultural science and technology on agricultural production were also quantified in this article, showing a certain degree of persuasive power. The impact of agricultural mechanization on agricultural production is not significant. In terms of economic and population factors, apart from the strong influence of policy factors, no other path of influence is significant. The factors of the agricultural machinery industry and agricultural science and technology are similar to agricultural production factors: economic factors and policy factors are significantly affected by these factors. Finally, among the directly related factors, the two factors with the greatest effect on the environment are the AMDL and AMIAT. The path regression coefficient of the impact of agricultural mechanization on the environment is 0.7745, and it shows a strong positive correlation. Although this is not very high, a certain significant correlation has been shown, reflecting the increasing degree of the environmental impact of SAM. The coefficient of the direct impact of the agricultural machinery industry and agricultural science and technology on the environment is not particularly high, but the overall effect is reduced because the impact of agricultural production itself has a relatively strong negative correlation, indicating that when agricultural output is higher, the environmental impact will improve. There is a certain harmony between the two factors. Statistically speaking, this improvement (more than 40%) should be given sufficient attention.

AMIAT → Environment 0.2512 5.1996

Policy → AP 0.9496 162.2230 Policy → AMIAT 0.9612 146.2765 Policy → AMDL 0.9725 343.0872 Policy → Environment 0.9787 378.2150 Note: Economic and population factors (EP), agricultural production (AP), the agricultural mechanization development level (AMDL), and the agricultural machinery industry and agricultural tech-

The path coefficients in Tables 5 and 6 describe whether there is a causal relationship between a pair of latent variables. If the path coefficient is small, the causal relationship reflected by that path may not exist. Of course, the paths in Table 5 only show the direct effects between the latent variable, that is, the direct impact of the cause variable on the result variable. Table 6 reflects the combination of direct and indirect effects of some of

The results of the PLS-SEM analysis showed that agricultural mechanization is directly related to the six dimensions of economy, population, agricultural production, the agricultural machinery industry and agricultural technology, the environment, and policies. It can be seen from Figure 3 that the level of Agricultural Mechanization Development Level (AMDL) is obviously promoted by the economic, population, and policy factors. The economic and population factors are the most critical factors affecting SAM. The effect of the agricultural machinery industry and agricultural technology (AMIAT) on

**Figure 3.** Radar chart of the total effect of the latent variables. **Figure 3.** Radar chart of the total effect of the latent variables.

#### *4.4. Discussion*

nology (AMIAT).

the latent variables.

*4.3. Results of Hypothesis Testing* 

SAM is not statistically obvious.


Because of the many lakes in large areas, mechanization is only found in a few areas, and only a few popular crops are grown. There is still a significant gap between Hubei and the provinces and regions with a high degree of SAM, such as Heilongjiang, Henan, and Jiangsu Provinces.

(4) Environmental factors are influenced by several other factors comprehensively. The past period of high economic growth has been accompanied by high pollution and high consumption issues, which can be seen very clearly here. Adjusting the relative balance of agricultural development and the ecological environment, which must be a non-negotiable part of SAM, requires attention. At the same time, our results also show that through the improvement of agricultural machinery technology and agricultural science and technology input, accompanied by the effective improvement of agricultural output mode and processes, the environment can also play a significant role in achieving the sustainable green development of agriculture [11].

So far, we have clarified the relationships, influence paths, and intensity of the interactions among several factors affecting SAM, analyzed the factors influencing SAM within the overall and structural relationships, and clearly showed the mechanisms and quantity of the internal influencing relationships. Among these key elements of SAM, the total effect of the two aspects of policy and economic factors is consistent with the actual situation [7], the relationship between the level of agricultural mechanization and agricultural production is worth exploring, and the environment is the result of comprehensive action. In addition, we also suggest that we should focus on improving the system of laws and regulations regarding agricultural mechanization and that we should standardize the production, sales, use, and service of agricultural machinery. New studies should be pursued to actively explore policies and measures to promote the development of agricultural mechanization, strengthen administrative laws regarding agricultural machinery, and perfect the operation mechanism of regulatory supervision of agricultural machinery. Multiple measures should be used to strengthen public legislation and education and to enhance the legislative awareness and ideas of the public to meet the objective requirements of building a modern agricultural system and sustainable development.

#### **5. Conclusions**

Although some studies have investigated the factors affecting agricultural mechanization in China, relatively few have involved systematic and structured econometric research. Based on historical data and the status quo of the development of agricultural mechanization in Hubei, this study used a partial least squares–structural equation model (PLS-SEM) framework to conduct a reasonable modeling analysis based on 28 measurement indicators and determined the relationships and paths of the factors affecting Sustainable Agricultural Mechanization (SAM) in Hubei. The measurement results provided solid support for most of our hypotheses and effectively verified and supplemented the corresponding qualitative research: SAM is directly affected by economic effects to a high degree, and environmental factors are comprehensively influenced by several other factors. In addition, some influencing relationships are presented in the form of quantitative results for the first time, such as the total effect of policy on the agricultural mechanization level and the path coefficient of the impact of agricultural mechanization on the environment.

Our finding relies on data we collected. Different data-collection methods, data facticity, and limitations of data may result in greater deviation from our results and the interpretation of our results to formulate our conclusions. Therefore, future research that could enrich our understanding of China's SAM could potentially proceed with longer-term empirical research. Meanwhile, the internet and big data technology can be used to monitor SAM in real time to reflect instantaneous developments and changes in SAM in response to different factors. These represent improvements that future studies can be undertaken to develop a more in-depth understanding of other economic, policy, and environmental factors impacting the adoption of Sustainable Agricultural Mechanization by producers in China and beyond.

**Author Contributions:** Conceptualization, Z.L. and M.Z.; methodology, Z.L., M.Z. and H.H.; software, Z.L. and J.F.; validation Z.L.; formal analysis, Z.L.; investigation, Z.L. and H.H.; resources, Z.L.; data curation, Z.L. and J.F.; writing—original draft preparation, Z.L.; writing—review and editing, Z.L. and Y.Y.; visualization, Z.L.; supervision, Z.L. and Y.Y.; project administration, Y.Y.; funding acquisition, Z.L. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was supported by the National Natural Science Foundation of China (71503095), a guiding project of humanities and social sciences of the Hubei Provincial Department of Education (15G028).

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** The data used in this study are available from the authors on reasonable request.

**Acknowledgments:** We appreciate the advice of the reviewers very much.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A**

**Table A1.** Descriptive statistics of the variables.



**Table A2.** Factor load estimation results of the measurement model.


Note: (1) Economic and population factors (EP), agricultural production (AP), the agricultural mechanization development level (AMDL), the agricultural machinery industry and agricultural technology (AMIAT), environment (E), and policy (P). (2) \*\*\* indicates a significance level of 1%, \*\* indicates a significance level of 5%. (3) STERR indicates standard error.


**Table A3.** Estimation of path coefficients for structural model.

Note: (1) Economic and population factors (EP), agricultural production (AP), the agricultural mechanization development level (AMDL), and the agricultural machinery industry and agricultural technology (AMIAT). (2) \*\*\* indicates a significance level of 1%, \*\* indicates a significance level of 5%, and \* indicates a significance level of 10%, **–** indicates failed *t*-test. (3) STERR indicates standard error.

**Table A4.** Factor load of observed variables in measurement Model B.


Note: Economic and population factors (EP), agricultural production (AP), the agricultural mechanization development level (AMDL), and the agricultural machinery industry and agricultural technology (AMIAT).

### **References**


## *Article* **The Impact of Government Subsidies on Technological Innovation in Agribusiness: The Case for China**

**Liping Wu <sup>1</sup> , Kai Hu <sup>1</sup> , Oleksii Lyulyov 2,3, Tetyana Pimonenko 2,3,\* and Ishfaq Hamid <sup>4</sup>**


<sup>4</sup> MICA, The School of Ideas, Ahmedabad 380058, India

**\*** Correspondence: tetyana\_pimonenko@econ.sumdu.edu.ua

**Abstract:** With the implementation of the rural revitalization strategy and the promotion of agricultural and rural modernization, the subsidies enjoyed by agricultural enterprises in China are increasing. As a result, the effectiveness of government subsidies for the technological innovation of agricultural enterprises has attracted more and more attention. Based on the perspectives of the whole industry chain of agriculture, forestry, animal husbandry, fisheries, and of processing, manufacturing, circulation, and service, this paper takes the listed agricultural companies from 2007 to 2019 as a research sample and empirically tests the effects and mechanisms of government subsidies on the technological innovation of agricultural enterprises. The study applies the fixed effect and intermediary effect models. The findings show that government subsidies potentially encourage agricultural enterprises to grow more successfully. Moreover, R&D expenditure is essential for enterprise technological innovation and leads to an intermediate impact. At the same time, government subsidies for the technological innovation of agricultural enterprises have a certain heterogeneity between different industries, state-owned enterprises and non-state-owned enterprises, and large enterprises and small and medium-sized enterprises. Therefore, this study argues that the government should continue to raise subsidies. In addition, the subsidies should be "different from enterprise to enterprise", and government subsidy funds should be better supervised to foster agricultural technological innovation properly.

**Keywords:** industry chain; government grants; technological innovation in agricultural enterprises; R&D investment

### **1. Introduction**

The Chinese government accepted the agricultural and rural modernization plan during the 14th Five-Year Plan period (2021–2025) [1]. In consideration of this, innovation was outlined as the core force for agricultural and rural modernization. The innovations in agricultural development are directed at improving the production of agricultural goods. The innovations in rural development allow the improvement of the production of agricultural goods and the education, health, and social infrastructure of rural areas.

Therefore, agricultural enterprises face several types of risks, such as environmental risks and operations risks [2–4]. In addition, agricultural enterprises face the issue of a lack of financing for the implementation of innovations [5–11]. Consequently, this limits the development of agricultural enterprises. Government subsidies in the form of financial aid have been implemented for a long time in China to modernize agricultural and rural development. In this case, government subsidies for agriculture and rural development may be defined as investments [12–18]. Past studies [14,19–21] outline that government subsidies could guide and motivate enterprises to increase R&D investment to implement technological innovation activities. At the same time, the inefficiency of government subsidies could be caused by the adverse selection of the innovation activities of enterprises

**Citation:** Wu, L.; Hu, K.; Lyulyov, O.; Pimonenko, T.; Hamid, I. The Impact of Government Subsidies on Technological Innovation in Agribusiness: The Case for China. *Sustainability* **2022**, *14*, 14003. https://doi.org/10.3390/ su142114003

Academic Editors: Aaron K. Hoshide and Francesco Caputo

Received: 15 September 2022 Accepted: 24 October 2022 Published: 27 October 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

for subsidies [14]. Adverse selection results in asymmetric information on available options for government subsidies.

Consequently, it could provoke inequalities and gaps in a company's innovation development and cause a decline in their long-term competitiveness [15]. Past research [16] has proven that information asymmetry between the government and enterprises causes subsidies to have a reverse effect. This could limit the achievement of indicated goals in the plan for agricultural and rural modernization during the 14th Five-Year Plan period (2021–2025) [1]. Thus, it is justifiable to analyze how government subsidies affect the technological innovation of agricultural enterprises and their mechanisms of action. It should be noted that in the ongoing economic open system theory [22], the development of all sectors, including agriculture and rural development, should be analyzed in connection with each other. Thus, agriculture is increasingly closely linked to the secondary and tertiary industries and fails to scientifically reflect the value of the whole industrial chain, such as production, processing, circulation, the service of agriculture, forestry, animal husbandry, and fisheries.

This paper focuses on analyzing the impact of government subsidies on agricultural enterprises' technological innovation from the whole industry chain perspective. Such samples allow the modelling of agricultural enterprises' whole and individual behavior. In addition, they allocate and measure the statistical effects that could not be determined based on the data of the individual enterprises. Regarding the standard of the National Bureau of Statistics' "Statistical Classification of Agriculture and Related Industries (2020)" (Order No. 32 of the National Bureau of Statistics) [23], agricultural enterprises are defined as all economic activities formed in the production, processing, manufacturing, service and other links of agriculture, forestry, animal husbandry, and fisheries, as well as relevant enterprises in the secondary and tertiary industries.

Our research aims to fill the following scientific gaps: (1) to develop a methodology to check the link between government subsidies and the technological innovation of agricultural enterprises; (2) to analyze agriculture from the whole industrial chain, and extend the scope of agricultural enterprises to agriculture, forestry, animal husbandry, fisheries production, processing, manufacturing, circulation, service, and other industries; and (3) to develop a methodology to check whether research and development could extend the innovation among agricultural enterprises. The remainder of this paper is divided into the following sections: Section 2 presents empirical evidence from the literature; Section 3 discusses the methodology and data; Section 4 analyzes the findings; and Section 5 considers conclusions and policy implications.

#### **2. Literature Review**

#### *2.1. The Relationship between Government Subsidies and Technological Innovation in Agricultural Enterprises*

Past research shows that there has been no consensus on the effect of government subsidies on enterprise technology innovation. A few main views constitute these findings. Government subsidies could incentivize enterprises to innovate technologically [3,22–41]. The late economist Kenneth Arrow [24] suggested that the technological innovation of enterprises has a spillover effect. Moreover, the free-riding behavior of other enterprises has seriously hit the enthusiasm of enterprises for independent innovation. This has provoked an insufficient supply of technological innovation. Subsequent research [25–27] has confirmed that government subsidies positively impact companies' performances. One of these studies [25] analyzed 158 listed energy companies in China. In this case, government subsidies for technological innovations negatively impacted company performance in the short term. At the same time, a positive effect was shown in the long term. Other researchers [26] demonstrated that Chinese government subsidies stimulate innovations in environmental management. However, these types of subsidies did not encourage the rapid growth of technological innovations. It should be noted that carbon-free technological

innovation could enhance the performance of companies [27]. However, the effect could be different depending on the time and efficiency of management.

Government subsidies can directly provide financial support to agricultural enterprises. As part of the profits of enterprises, government subsidies directly increase enterprises' funds, alleviate the shortage of funds available to agricultural enterprises, improve their enthusiasm to innovate, and solve the spillovers of innovative results. They also reduce the risk caused by the uncertainty of innovation and encourage enterprises to increase R&D investment in technological innovation [28,33]. Secondly, government subsidies send a positive signal of government recognition and reduce the information asymmetry between enterprises and external investors. An enterprise that enjoys this subsidy shows that the government recognizes their development. This proves that the enterprise has strong R&D innovation ability, good innovation projects, and is more willing and capable of technological innovation [3,28–33]. At the same time, scholars [32,33] confirm that government subsidies should be implemented at all levels, from companies to individuals. In this case, government subsidies could positively impact agriculture.

Government subsidies can improve the ability of enterprises to access resources. They can also improve the ability of enterprises to obtain resources by supplementing their innovation resources and enhancing their recruitment of talented workers. Agribusinesses receiving government subsidies send positive signals of good relations with the government, indicating they have sufficient government resources. The government provides an invisible guarantee for agricultural enterprises to make up for the natural weakness of agriculture and attracts banks, venture capitalists, etc., to increase investment. Furthermore, it also increases the attractiveness of enterprises to prospective employees, which improves the overall level of Research and Development (R&D) personnel, and enhances the technological innovation capabilities of enterprises [34].

Past research [35] on strategic emerging industry enterprises found that the impact of financial incentive policies on innovation conforms to an inverted U-shape. In this case, the scholars confirmed that government subsidies stimulate innovation to a certain point, after which efficiency declines. Thus, the government should control monitoring systems for government subsidies. Other research [36] found no significant positive impact of government subsidies on private R&D for small- and medium-sized firms. In summary, agricultural government subsidies increase funds for production and investment, release positive signals to attract more external financing and outstanding human capital, improve the ability of companies to obtain resources, and promote technological innovation for agricultural producers. Thus, we propose our first research hypothesis:

#### **Hypothesis 1 (H1).** *Government subsidies can promote technological innovation in agribusiness*.

#### *2.2. Mechanisms of Government Subsidies for Technological Innovation in Agricultural Enterprises*

Enterprises with effective Research and Development (R&D) generate technical knowledge, have certain externalities, are easily learned or reproduced, and suffer from market failure. At the same time, the investment, risk, and uncertainty of these activities provokes issues for enterprises, especially agricultural enterprises, in obtaining funds from the capital markets. Nevertheless, based on the importance of R&D and the solutions to market failures, the government should promote enterprises to carry out such innovation [42].

Based on data from Chinese companies, past research [42–44] finds that government subsidies have an incentive effect on companies' R&D activities. Government subsidies can support agricultural enterprises in increasing investment in three ways: by reducing the cost of R&D, reducing the uncertainty of these types of projects, and dispersing subsequent risks. Thus, government subsidies reduce R&D costs. According to the theory of externalities, the externalities of R&D activities lead to the spillover of knowledge, which to a certain extent discourages the enthusiasm of enterprises involved in such research. Government subsidies, as part of corporate profits, reduce the marginal cost of enterprise R&D and then stimulate agricultural enterprises to increase investment. Furthermore, government

subsidies reduce uncertainty about projects [45]. This increases the market demand for a project's results and improves the expected return of the enterprise [45]. At the same time, it can also attract more qualified personnel to participate in projects, reducing their uncertainty. Finally, government subsidies can diversify R&D risks. The government provides subsidies and shares information about the project, which can attract external investors and incentivize them to join, reducing the risk of failure for enterprises to a certain extent [46,47].

According to the new economic growth theory, human capital and investment are important factors in promoting economic growth and technological innovation [42]. Enterprises, through R&D activities, improve the stock of human capital and promote enterprise innovation [41]. R&D activities are the most direct source of technological innovation. Enterprises increase investment in activities, generate new knowledge and information, and directly promote technological innovation. Furthermore, this increase in investment enables enterprises to use existing external knowledge better, enhance their knowledge stock, and indirectly promote their innovation capabilities [46]. Thus, past studies [48,49] emphasize that providing an effective R&D policy allows the development of additional advantages. This could be due to the implementation of transborder strategies on knowledge sharing, geographical changes in research developments and innovations, and the international fragmentation of research activities. It has been demonstrated that competitiveness depends on innovative activities [50]. At the same time, lack of labor and financial resources are the biggest limitations to investing in R&D.

Therefore, an increase in investment in R&D can promote technological innovation. Thus, it can be concluded that government subsidies encourage agricultural enterprises to increase investments by reducing the cost of R&D and project uncertainty, as well as helping to disperse production risk. Therefore, we propose our second hypothesis:

**Hypothesis 2 (H2).** *Government subsidies encourage both investment and technological innovation*.

#### **3. Materials and Methods**

#### *3.1. Sample Selection and Data Sources*

Taking the A-share (representing publicly listed Chinese companies that trade on Chinese stock exchanges, such as the Shenzhen and Shanghai Stock Exchanges) of listed agricultural companies from 2007 to 2019 as a research sample, this paper no longer limits agriculture to traditional agriculture, forestry, animal husbandry, and fisheries. Instead, it extends it to the perspective of the whole industry chain to the production, processing, manufacturing, circulation, and service of agriculture, forestry, livestock, and fisheries. Drawing from the practice of [44] and referring to the standards of the Statistical Classification of Agriculture and Related Industries (2020) (Order No. 32 of the National Bureau of Statistics) issued by the National Bureau of Statistics and the Guidelines for the Classification of Listed Companies (Revised in 2012) issued by the China Securities Regulatory Commission, agriculture-related industries include agriculture, forestry (A02), animal husbandry (A01 and A03), fisheries (A04), and services related to these natural resource-based industries (A05). Processing and manufacturing in these industries includes food processing (C13) and the manufacture of food (C14), fertilizers and pesticides (C26), and agricultural machinery (C35). Listed agricultural companies are involved in agriculture, forestry, animal husbandry, and fisheries, as well as enterprises in the secondary and tertiary input sectors whose products are essential for firms within these natural resource-based industries. We narrowed down our sample to 177 listed agricultural companies from 194 after removing 17 companies with serious financial risks. Our non-balanced panel data consisted of 2301 enterprises from these 177 companies. The enterprise patent data used in this article come from the China Research Data Service Platform (CNRDS database) [51]. Some of the missing data were provided by searching on the patent website of the State Intellectual Property Office [52]. The screening of listed agricultural companies was mainly based on analyzing enterprises' main business scopes, such as Hexun Network [53] and

Flush Database [54]. The data for the other variables were collected from the CSMAR database [55].

#### *3.2. Variable Settings*

Past studies [45,46] demonstrate that patent applications are one of the incentives for developing and implementing technological innovation at companies. In addition, considering the analytical report of World Intellectual Property Indicators 2021 [56], patents guarantee the authorship protection of innovation. Furthermore, patents allow the obtainment of additional revenue for agricultural companies. Considering this, our research used the patent applications of enterprises as the measure of technological innovation (Patentt+1). Considering the time lag of technological innovation, the technology innovation level of t + 1 was measured by adding 1 logarithm to the number of patent applications in the t-period based on the methods outlined in [45,46]. The t-period starts with a value of 0 zero.

Government grants were the explanatory variable we evaluated. There are large differences in the amount of government subsidies distributed based on the size of a natural resource-based enterprise. In order to narrow the absolute difference between the data, the logarithm of the government subsidies received by the company in the current year was taken to measure the explanatory variables. Based on other scholars' work on enterprise technological innovation, our research used six control variables that may affect the technological innovation of agricultural enterprises, such as enterprise size, age, asset– liability ratio, growth potential, proportion of fixed assets, concentration of equity, and salary incentives (Table 1). In order to analyze the impact mechanism of government subsidies on technological innovation, we defined investment as an intermediary variable using the logarithm of the company's investment in the current year.


**Table 1.** Description of the variables and the calculation formula.

#### *3.3. Model Settings*

In order to analyze the impact of government subsidies on the technological innovation of agricultural enterprises, we used a basic econometric model specified as:

$$\text{Paten}\_{\text{it}+1} = \alpha\_0 + \alpha\_1 \text{SUB}\_{\text{it}} + \beta \text{CV}\_{\text{it}} + \sum \text{Year} + \sum \text{Ind} + \varepsilon\_{\text{it}} \tag{1}$$

where Patentit+1—technological innovation in the Company<sup>0</sup> s t+1 period; α0—denotes the constant term; SUBit—the government subsidy of the company's t period; CVit—the control variable matrix; εit—the residual term; i and t—the enterprises and years; and Year and Ind—the fixed effect of the year and industry, respectively.

The two-way fixed-effect model [57] is applied to decrease the impact of the macroeconomic environment and the nature of the industry. However, R&D investment is introduced as the intermediary variable to identify the mechanisms of government subsidies for the technological innovation of agricultural enterprises. Therefore, the following Ordinary Least Square (OLS) econometric models are set up based on model (1) r using methods from [58,59] in order to analyze the intermediary effect of R&D investment.

$$\text{RD}\_{\text{it}} = \alpha\_0 + \alpha\_1 \text{SUB}\_{\text{it}} + \beta \text{CV}\_{\text{it}} + \sum \text{Year} + \sum \text{Ind} + \varepsilon\_{\text{it}} \tag{2}$$

$$\text{Paten}\_{\text{it}+1} = \alpha\_0 + \alpha\_1 \text{SUB}\_{\text{it}} + \alpha\_2 \text{R\&D}\_{\text{it}} + \beta \text{CV}\_{\text{it}} + \sum \text{Year} + \sum \text{Ind} + \varepsilon\_{\text{it}} \tag{3}$$

where RDit equals the R&D for company i during time period t; SUBit is the government subsidy of the company's t period; and CVit is the control variable matrix with εit as residual error of the model. Year and Ind are the fixed effect of the year and industry, respectively.

#### **4. Results**

#### *4.1. Descriptive Statistics and the Correlation Analysis*

The descriptive statistical results of the variables signify that the average number of patent applications is 1.6146, the median is 0.6931, and the maximum and minimum values are 7.3671 and 0 with a standard deviation of 1.3992 (Table 2). Thus, the vast majority of listed agricultural companies have technological innovations but vary greatly. In addition, the average value of government subsidies is 16.3633, the median is 16.4341, the maximum and minimum values are 20.7799 and 8.9227, respectively, and the standard deviation is 1.5357. This suggests that the government subsidies enjoyed by listed agricultural companies are more balanced, but specific differences exist.

**Table 2.** Descriptive statistical results of the variables.


The correlation analysis of the variables is shown in Table 3. Thus, the correlation coefficient between the current government subsidy (SUB) and the next phase of patent applications is 0.423 at the 1% level of significance. The correlation coefficient between the SUB and the intermediary variable for R&D input is significant with a value of 0.384. The correlation coefficient (r) denoting a positive association between R&D and the next phase of patent applications is 0.574, which is also significant at the 1% confidence level. Among the control variable, enterprise size and age are significant and positively correlate with the number of next patent applications.


**Table 3.** Variable correlation analysis.

Note: \*\*, and \*\*\* are significant at the 10%, 5%, and 1% levels.

However, equity concentration is significant and negatively correlated (−0.112) with the number of next patent applications. Executive compensation correlates significantly with the number of next patent applications at the 1% significance level with a positive r equal to 0.534. There is a significant correlation between the main variables and further multiple regressions. The absolute value of the correlation coefficient between the main variables is less than 0.5, indicating no limited multicollinearity. Multicollinearity or high degrees of association (r > 0.7) between independent variables is problematic since the OLS regression model assumes "independent" impacts of independent variables specified in the model on the dependent variable. Multicollinearity distorts the parameter estimates in the OLS model rendering inferences gleaned from the model results potentially inaccurate.

#### *4.2. Regression Analysis Results*

The regression results from empirical tests on the impact of government subsidies on technological innovation in agribusiness using model (1) are shown in Table 4. After the number of patent applications in the current period plus one to take the logarithm and lag one period as the explanatory variable, the enterprise-level variables and the annual and industry fixed effects are gradually controlled. Additionally, the regression coefficient of government subsidies is significantly positive at the 1% confidence level. The findings from column (4) of Table 4 suggest that under the two-way fixed effect of control years and industries, the regression coefficient of SUB is 0.221. The change in government subsidies in the current period is 1%, and the average change in the number of patent applications of enterprises in the next year is 0.221%. This implies that government subsidies promote agricultural innovation, which validates our first hypothesis. Among the other control variables, the regression coefficients of enterprise size, asset–liability ratio, and executive compensation are significantly positive. This indicates that growth in scale results in an increasing level of debt. Furthermore, increases in executive compensation are conducive to increasing patent applications and technological innovation. The regression coefficients of enterprise age and equity concentration are significantly negative. This suggests that the longer the company is established, the higher the equity concentration, the fewer the number of patent applications, and the lower the level of technological innovation.


**Table 4.** Return results of the impact of government subsidies on technological innovation in agricultural enterprises.

Note: \*, \*\*, and \*\*\* are significant at the 10%, 5%, and 1% levels.

#### *4.3. Analysis of the Intermediary Affect Test Results*

Empirical testing has verified that government subsidies can promote technological innovation in agribusiness. According to the previous analysis, government subsidies may affect the technological innovation of enterprises by influencing their R&D investment. According to [58], the empirical test is carried out through models (1) and (3), and whether the R&D investment plays an intermediary role according to the regression coefficient and significance level of government subsidies and R&D investment.

Column (1) of Table 5 shows the regression results of model (1). The regression coefficient of government grants is 0.221, which is significant at the 1% confidence level. This implies that the basic variable government grant significantly positively affects the number of patent applications for the interpreted variable. Column (2) shows the regression results of model (2), and the regression coefficient of government subsidy is 0.201, which is also significant at the 1% level. Thus, government subsidies appear to have a significant impact on investment in R&D.

Column (3) in Table 5 summarizes the regression results for model (3). The regression coefficient of government subsidy after adding the intermediary variable R&D investment is still significant, but the coefficient drops from 0.221 to 0.212. This indicates that the positive effect of government subsidies on the number of patent applications is partially absorbed by the R&D investment of the intermediary variable. Thus, R&D investment plays a part in the intermediary effect. The proportion of the intermediary effect to the total effect is 27.56%. Moreover, the government subsidy acts on the level of technological innovation of the enterprise by influencing such investment of the enterprise. Therefore, our second hypothesis is also validated.


**Table 5.** Test of the intermediary effect of government subsidies affecting the technological innovation in agricultural enterprises.

Note: \*\*\* is significant at the 10%, 5%, and 1% levels.

#### *4.4. Analysis of Heterogeneity*

In order to investigate the heterogeneity of the samples, this paper conducts empirical tests according to the industry, the nature of the enterprise, and the size of the enterprise. Our research analyzes the production, processing, manufacturing, circulation, and service of agriculture, forestry, animal husbandry, and fisheries from the perspective of the whole industrial chain. The nature of the enterprise is according to whether the actual controller of the enterprise is a government department at all levels. If so, it is a state-owned enterprise; otherwise, it is a non-state-owned enterprise. The size of enterprise is divided into large, small, and medium-sized enterprises. The core criteria are the operating income of the enterprise in the current year. If it exceeds RMB200 million, it is a large enterprise; otherwise, it is a small or medium-sized enterprise.

The group regression results (Table 6) show that from the perspective of the industry, the regression coefficient between the government subsidies for the processing of agriculture, forestry, animal husbandry, and fishery products and the manufacturing industry, the number of manufacturing materials in the manufacturing industry, and the number of patent applications in the next period is significantly positive. At the same time, the regression coefficient between the government subsidies for traditional agriculture, forestry, animal husbandry, and fisheries and the number of patent applications in the next period is not significant. Government subsidies for these natural resource-based industries promote technological innovation by these businesses. At the same time, government subsidies for traditional agriculture, forestry, animal husbandry, and fisheries do not significantly affect enterprises' technological innovation. The reason for this may be that agriculture, forestry, animal husbandry, and fisheries are more susceptible to fluctuations in natural factors and market factors. Therefore, despite government subsidies, these subsidies have not substantially improved enterprises' R&D conditions, and their R&D power is insufficient.

#### *4.5. Robustness Test*

In order to test the robustness of the results, we used the number of patent grants instead of the number of patent applications as the agent variable of technological innovation. The regression results (Table 7) show that the regression coefficient of the SUB is significantly positive at the 1% level, which is consistent with the results in Table 4. This confirms that the regression results of Table 4 are stable. The conclusions of this study have passed the empirical test, have strong explanatory power, and can be used to guide and encourage technological innovation in agricultural enterprises.


**Table 6.** Group regression results by industry, enterprise nature, and size.

Note: \*, \*\*, and \*\*\* are significant at the 10%, 5%, and 1% levels; A: agriculture, forestry, animal husbandry and fisheries; Agr: agriculture, forestry, animal husbandry, and fishery products processing and manufacturing industry; Ag: agriculture, forestry, animal husbandry, and fishery means of production manufacturing industry; St: state-owned enterprises; NSt: non-state-owned enterprises; L: large-lot producer; S/M: medium and small-sized enterprises.

**Table 7.** Summary of Ordinary Least Squares (OLS) regression parameter estimates for technological innovation and research and development.


Note: \*, \*\*, and \*\*\* are significant at the 10%, 5%, and 1% levels.

#### **5. Discussion**

Our model results are consistent with the results of [42,43]. At the same time, the findings underline the necessity of government subsidies for technological innovation in agribusiness in China. Firstly, the study found that government subsidies effectively promote technological innovation in agribusiness. Government subsidies affect the technological innovation of enterprises by influencing their R&D investment; that is, the positive effects of government subsidies on the number of patent applications are partially absorbed by the R&D investment of the intermediary variable. Moreover, R&D investment is an intermediary effect that accounts for 27.56% of the total effect. Thirdly, the effects of government subsidies on the technological innovation of agricultural enterprises have a certain heterogeneity. From an industry perspective, government subsidies for processing agriculture, forestry, animal husbandry, and fishery products and manufacturing promote technological innovation in enterprises. However, government subsidies for traditional agriculture, forestry, animal husbandry, and fisheries do not significantly affect these enterprises' technological innovations. In terms of the nature of the enterprises, government subsidies promote the technological innovation of state-owned and non-state-owned enterprises. Their impact on technological innovation for non-state-owned enterprises is greater than it is for state-owned enterprises. In terms of the size of enterprises, government subsidies promote technological innovation for all sizes of companies. The impact of technological innovation is greater for large enterprises than it is for small and medium-sized enterprises.

The results of this study confirm the assumptions that innovations and digital technologies are the core instruments with which to support the sustainable development of agriculture. These findings are consistent with past research [60–62]. At the same time, innovations and digital technologies require sufficient financial resources from the government subsidies that are available to agricultural companies. However, the government should consider all the effects from innovation projects when making decisions on how to allocate government subsidies to innovative agricultural projects. These subsidized projects can positively and/or negatively impact the environment and society. Past research confirms that innovations in water management can provoke the relocation of local people [63–65]. Other researchers have demonstrated that R&D investments in agriculture positively impact farmers and local communities [66–69]. This suggests that the government should balance agricultural productivity and economic profits with minimizing negative environmental impacts (e.g., soil degradation, water and soil pollution, deforestation, etc.) and promoting societal benefits (e.g., healthy diets, community vibrancy, etc.). The following three policy suggestions are put forward based on the above research conclusions: Firstly, the government should continue to increase subsidies. The rural revitalization strategy needs scientific and technological innovation as a support. The core key to agricultural and rural modernization also depends on scientific and technological innovations, which play a pivotal role in agricultural and rural development. As the main body of technological innovation, agricultural enterprises play an important strategic role in agricultural modernization. Studies have shown that government subsidies effectively promote the technological innovation activities of agricultural enterprises. Moreover, our findings confirm that government subsidies are effective options for stimulating innovation in agricultural enterprises. Therefore, the Chinese government should continue to increase agricultural subsidies, such as direct subsidies, tax incentives, and research and development subsidies. The Chinese government should also account for possible negative externalities of subsidized agriculture, including environmental pollution and the forced relocation of entire communities.

Secondly, government subsidies should "vary from enterprise to enterprise". The impact of government subsidies on the technological innovation of agricultural enterprises varies according to the type of industry, the nature of the enterprise, and its size. Government subsidies have a significant role in promoting technological innovation in the processing and manufacturing of agriculture, forestry, animal husbandry, and fishery products. Their impact on technological innovation for non-state-owned enterprises is greater

than it is for state-owned enterprises. The impact of technological innovation is greater for large enterprises than it is for small and medium-sized enterprises. Therefore, government departments should be divided into categories. The government's limited subsidy resources should be invested in enterprises with strong technological innovation capabilities. Thus, agricultural processing and manufacturing companies need to be supported with high-quality resources to invest in agricultural enterprises with a strong willingness to adopt innovative technologies.

Thirdly, government subsidy funds need to be better supervised. Government subsidies affect the technological innovation of agricultural enterprises through R&D investment. Therefore, the government should strengthen the supervision of the use of subsidy funds and improve the performance of the use of funds. It is possible to establish and improve a monitoring system covering the whole process and the whole chain of fund allocation, implementation, and supervision. It is necessary to analyze the efficiency of government subsidies. At the same time, the focus is on supervising agricultural enterprises with low R&D investment levels and on encouraging enterprises to increase their investment in innovative, sustainable technologies and processes.

The efficiency of government policy for supporting the innovation implementations in agricultural companies should become an instrument for improving the export structure of agriculture and achieving sustainable development goals. Thus, the agricultural sector is a crucial element of food security. This involves the rational use of limited resources and the implementation of green technologies and energy efficiency innovations while mitigating adverse environmental and community impacts.

#### **6. Conclusions**

From the whole industry chain perspective, this paper extended the agricultural scope to the production, processing, manufacturing, circulation, and service of agriculture, forestry, animal husbandry, and fisheries. It empirically tested the effect and influence mechanism of government subsidies on agricultural enterprises' technological innovation by taking the companies listed from 2007 to 2019 as a research sample. We developed Ordinary Least Squares statistical regression models to test these hypotheses.

Despite the valuable findings and practical recommendations, our research has a few limitations. Our analysis focused on China only. At the same time, the globalization and openness of the economy facilitates potential improvements or declines in the competitiveness and sustainability of companies involved in agriculture and agro-forestry. The competitiveness of agricultural businesses also depends on other internal and external factors and should be studied in future investigations. Internal factors include the social responsibility of companies, the education level of managers, technological innovations, etc. External factors include government corruption and quality, sustainable development pathways in the region, geographic characteristics, etc. Innovative agricultural projects that are subsidized by the government can have a wide range of positive and/or negative economic, ecological, and social impacts which warrant further investigation.

**Author Contributions:** Conceptualization, L.W., K.H., O.L., T.P. and I.H.; methodology, L.W., K.H., O.L., T.P. and I.H.; formal analysis, L.W., K.H., O.L., T.P. and I.H.; investigation, L.W., K.H., O.L., T.P. and I.H.; writing—original draft preparation, L.W., K.H., O.L., T.P. and I.H.; writing—review and editing, L.W., K.H., O.L., T.P. and I.H.; visualization, L.W., K.H., O.L., T.P. and I.H. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by a grant from the Ministry of Education and Science of Ukraine, 0121U100468, "Green investing: cointegration model of transmission ESG effects in the chain "green brand of Ukraine—social responsibility of business".

**Institutional Review Board Statement:** Not applicable.

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Not applicable.

#### **Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**

