The Role of Digital Finance Embedded in Green Agricultural Development: Evidence from Agribusiness Enterprises in China

Abstract

Digital inclusive agriculture refers to an agricultural development model that integrates various digital technologies into the agricultural production process, aiming to deliver benefits for all stakeholders throughout the agricultural value chain. This paper draws on the ecological symbiosis theory, embeds the concepts of digital finance and social responsibility into the goal of green development in agriculture, selects 395 agribusiness enterprises in China from 2013 to 2022 as the analysis sample, and examines the impact by adopting an improved weighted least squares (WLS) fixed effects model. Results show that digital finance has a significant effect on the quantity and quality of green innovation in agribusiness enterprises, and good social responsibility performance can enhance the innovation promotion effect of digital finance. Heterogeneity analysis reveals that agribusiness belonging to the processing and distribution type, located in the eastern region, and in the growth stage benefit more clearly. This paper provides theoretical references and practical guidelines for solving agricultural financing problems, boosting their green innovation capacity in the digital age. It is of great practical significance for realizing the green symbiotic ecology of responsible agricultural industry, promoting the win–win situation between enterprises and society, and the high-quality development of agriculture.

1. Introduction

Rural areas are a vast world for entrepreneurship and innovation, and agribusiness is an important force for serving agriculture, rural areas, farmers, and realizing “rural revitalization”, so it is of practical significance to study how to promote the green innovation of agribusiness for the development of low-carbon agriculture. As a large agricultural country, China is striving to promote the transformation and upgrading of traditional agriculture. However, due to the weakness of traditional agricultural production, the technological innovation of agriculture-related enterprises is deeply affected by the natural environment, difficult to prevent and control risks, high input costs, and there is a certain financial exclusion problem in the key links of breeding technology innovation, product research, and so on [1,2], which has always made it difficult to bring significant results to the transformation and upgrading of agriculture.

In recent years, with the booming development of the digital economy, the creation of digital finance has provided a new path for financing innovation in agribusiness. Compared with the traditional financial system, digital finance is inclusive and conducive to helping disadvantaged enterprises solve financing problems [3]. Through the application of emerging digital technology and big data information, digital finance improves the information transparency of the financial market, expands the coverage breadth, depth of use, and degree of support services [4], and can better serve rural areas and lower the financial threshold [5]. Agricultural enterprises using big data financial technology to plan high-quality financing schemes can reduce the cost of information acquisition and risk identification [6] and provide stable capital flow for green innovation. Therefore, it is of great significance to explore whether digital finance can promote the substantial green innovation performance of agriculture-related enterprises to promote digital agriculture and low-carbon development.

At the same time, according to the social network embeddedness of management, economies do not exist in isolation but are dependent on political, cultural, religious, and social relations [7,8]. Due to their natural characteristics of being embedded in the local vernacular environment, agribusinesses have complex and multiple intrinsic social, economic, and cultural connections with local communities and residents and, in many cases, are called “community enterprises” [4]. The author’s field research in Heilongjiang and Inner Mongolia, which are located in China’s major agricultural production bases, has shown that the social responsibility performance of China’s agribusinesses is distinctive. For example, the factory site of the Jiansanjiang Branch of the Beidahuang Agricultural Group itself is located in the countryside. In 2023, the company provides education investment, medical assistance, and even housing repair and other livelihood support for local workers and communities, which promotes the employment of more than 20,000 rural people, promotes the renewal of municipal infrastructure in the towns around the farm, and plays a social integration function. China’s agriculture-related enterprises have improved the living environment of the surrounding communities and promoted the local social and economic development through social responsibility behavior, and the enterprises have also won the protection of the surrounding communities for their farms [3].

From this perspective, agribusiness not only supports but also coexists with local communities through the fulfillment of social responsibility. Then, in the context of digital finance’s role in financing innovation for agriculture-related enterprises, how does the fulfillment of these social responsibilities influence outcomes? Additionally, variations in industrial characteristics and regional development conditions may contribute to heterogeneous impacts. Existing research on the impact of digital finance on enterprise green innovation mostly focuses on the whole enterprise [2,9,10], with a lack of research on agricultural enterprises. Most scholars have mentioned the financing constraints of agriculture-related enterprises, but only a few studies have begun to pay attention to the impact of rural financial development on agricultural technological innovation [1,11], and the impact on the social responsibility performance of agriculture-related enterprises in the context of digital economy development still needs further analysis.

In summary, to address these knowledge gaps, we have two objectives. Objective 1: From the perspective of ecological symbiosis and social embeddedness of management, this paper integrates Peking University’s Digital Financial Index and the long panel data of Chinese listed agro-related enterprises from 2013 to 2022. It aims to investigate the influence of rural digital finance on the green innovation performance of these enterprises and assess its correlation with their social responsibility performance. Objective 2: To conduct decomposition and control analyses that distinguish between industrial chain location heterogeneity, geographical location heterogeneity, and individual growth stage heterogeneity; further simulating the evolving trends of economic and environmental benefits across different ecological units within the symbiotic system. Compared to the existing literature, the added value of this paper is mainly in the following areas:

(1) Innovation in research perspective. This paper introduces the theory of symbiosis into the study of agribusiness management and agricultural economy and argues that the symbiotic relationship between agribusiness and surrounding communities can better explain the role and effect of agribusiness social responsibility and complement the symbiosis value study of agribusiness contact with various stakeholders. Meanwhile, based on the current background of the integration of digital technology and traditional financial services, the impact of digital finance is investigated, and research on the antecedents of green innovation in agricultural enterprises is extended.

(2) Innovation in research methods. We collected 10-yearlong panel data of 395 Chinese agribusinesses for econometric modeling and applied improved WLS estimation, the Tobit model, group regression, and the IV-2SLS methods to more accurately and effectively simulate the impact of digital financial development on the two dimensions of quality and quantity of green innovation, as well as the moderating effect of social responsibility performance. It overcomes the shortcomings of the single research method.

(3) The research samples are more detailed. This paper divides the industrial chain location difference, geographical location difference, and individual growth stage difference to explore the heterogeneity of the utility difference and the reasons. It provides theoretical and empirical support for agricultural enterprises of different business types, regions, and development levels to use digital finance to enhance green innovation capabilities, create multiple value symbiosis of society, economy, and environment, and achieve a more inclusive, resilient, and sustainable agricultural system.

The structure of this paper is as follows: 1. The Introduction explains the research background and the shortcomings of the existing research and leads to the main research purpose and research value of this paper. 2. In the Theoretical Analysis and Hypothesis Development section, the logical deduction is made according to the symbiosis theory. The internal relationship and mechanism of each factor are analyzed, and the research model is formed. 3. In the Methodology section, samples, data, research methods, and models are explained. 4. In the Results and Discussion section, the research results are presented and specifically analyzed. 5. The Conclusion and Future Studies section summarizes the research findings, discusses the significance of the research, and suggests policy suggestions and future research directions.

2. Theoretical Analysis and Hypothesis Development

2.1. Digital Finance and Green Innovation in Agribusiness

Green technological innovation requires sufficient financial support and a stable financial system to achieve better results [12]. As vulnerable enterprises, agricultural enterprises face greater risks in green innovation investment and higher uncertainty in green innovation output [9]. The current market risks are particularly random, and the traditional financial system has obvious shortcomings in supporting the innovation of agriculture-related enterprises [3,13,14,15], which makes it difficult for agriculture-related enterprises to obtain the support of financial institutions to achieve high-level green innovation.

As of December 2022, statistics from China’s social science network show that digital finance, with the help of the “long-tail effect” of the Internet and information technology, has been applied to nearly 200 million farmers, 70 million small and micro-enterprises and merchants in China, so that the tail group of people suffering from financial constraints in the market can also enjoy efficient and convenient services. Therefore, this paper argues that the emergence of rural digital finance can improve the green innovation performance of agribusiness, mainly from the following three aspects:

First, improve the interaction between supply and demand [16] and broaden the channels of innovative cooperation [17]. The digitalization of the financial system has changed the traditional service model, broken down barriers to entry, and expanded the scope of services. By alleviating the degree of financial mismatch and strengthening the role of the market in the allocation of resource factors, it can significantly increase the availability and scale of formal credit for farmers [11], which in turn drives enterprise technological innovation [18]. Farming-related enterprises can rely on big data and information technology to capture multi-dimensional information on financing sources, find financial support for their own green innovation, and alleviate the financing constraints required for innovation [10]. Financial institutions can also quickly capture timely and effective market information and objective data, accurately match agricultural enterprises in need of loans, and set interest rates in line with the current situation of the agricultural market.

Second, reduce financing costs and improve innovation efficiency. In the traditional financial system, most of the assets of agricultural enterprises do not meet the conditions for mortgage loans and are strongly affected by the natural environment and market price fluctuations. To compensate for potential risks, financial institutions usually raise interest rates and increase guarantee requirements, resulting in higher financing costs for agricultural enterprises [11]. Digital finance reduces the front-end transaction costs of traditional finance for agriculture-related enterprises [19]. Through the online platform for information organization and financing transactions, the time cost and debt financing cost required for agriculture-related enterprises to capture effective financing information are greatly reduced. By optimizing the allocation of financial resources and improving the efficiency of enterprise innovation financing, it has a significant role in promoting the green innovation of enterprises [9]. At the same time, digital technology also provides timely and efficient back-end technical support for the subsequent innovation of agriculture-related enterprises [19], and the overall green innovation cost is lower.

Thirdly, it guarantees financing security and reduces the risk of innovation. Digital finance makes the financing reception information transparent through massive information processing, avoids the occurrence of moral hazard behavior caused by the information asymmetry between farmers and financial institutions [17], and helps to control the risk and ensure the robustness of the transaction. In turn, it strengthens the foundation of innovation, enhances the innovation willingness and innovation ability of the main body of innovation, and has a facilitating effect on the transformation of enterprise innovation results [16]. Meanwhile, in the Guiding Opinions on Promoting the Healthy Development of Internet Finance, jointly issued by the People’s Bank of China, the Ministry of Industry and Information Technology, the China Banking Regulatory Commission, and 10 other ministries and commissions, the regulation of data governance and risk management is clearly emphasized. It is pointed out that the stability and security of the Internet and digital financial services are ensured under the complex and volatile characteristics of digital finance [20]. This undoubtedly provides a good institutional environment and external security for agriculture-related enterprises to leverage digital finance for green innovation.

In summary, this paper argues that digital finance can help agriculture-related enterprises improve the quality and efficiency of green innovation by broadening innovation cooperation channels, improving innovation efficiency and reducing innovation risks. Therefore, this paper puts forward the following hypotheses:

H1: 

All other things being equal, digital finance is conducive to enhancing the green innovation performance of an agribusiness.

H1a: 

Digital finance is able to enhance the quality level of green innovation of agribusiness enterprises in all aspects.

H1b: 

Digital finance is able to enhance the quantitative level of green innovation of agribusiness enterprises in all aspects.

2.2. Digital Finance, Agribusiness Social Responsibility, and Green Innovation

The concept of symbiosis originated in biology and was first proposed by German scientist H. de Barv in 1878, which originally referred to the phenomenon of organisms of different species living together [21]. However, the “Prisoner’s Dilemma” in economics, which demonstrates how individual rationality in competition can lead to collective irrationality, and the “team philosophy” advocated in management both essentially emphasize that individuals need to consider issues from the perspective of “reciprocal symbiosis”.

For agribusinesses, the promotion of agricultural production, the guarantee of food security, the protection of farmers’ interests, the transmission of agricultural technology, and other socially responsible behaviors have a certain degree of compatibility with the “pluralistic interaction and harmonious symbiosis” emphasized in the theory of symbiosis. On the one hand, it can make farmers and agribusinesses with a common vision evolve and develop together; on the other hand, it can realize sustainable symbiosis with nature and society. Therefore, this paper draws on the theory of symbiosis to construct the following responsible agricultural industry digital green symbiosis model (see Figure 1). It is argued that the performance of social responsibility based on the concept of symbiosis can enhance the facilitating effect of digital finance on the green innovation performance of agribusiness, which mainly works in the following three aspects:

First of all, from the perspective of symbiosis units, agribusiness enterprises have a large number of stakeholders, and the relationship between them has obvious circularity and radiality. Through social responsibility behaviors such as poverty alleviation, employment of the local population, and transmission of agricultural technology, agribusiness enterprises establish and maintain good symbiotic network relationships with different stakeholders, which are conducive to the exchange and distribution of material, information, and energy among the symbiotic units in the system. For example, enterprises that actively fulfill social responsibility disclose information in a more timely and comprehensive manner, are more likely to obtain positive responses from customers, and can strengthen the political connection with the government and communicate more adequately with upstream and downstream customers in the supply chain, third-party logistics, and financial institutions [22]. Agricultural enterprises that actively fulfill their social responsibility have an easier time obtaining financial support through agricultural supply chain finance [17]. Employees in a good CSR atmosphere will enhance the sense of occupational safety and corporate identity, which will help improve corporate green innovation performance [23].

Second, from the perspective of the symbiosis model, digital finance has a greater ability to acquire and disseminate information. The establishment of digital pathways allows for a smoother flow of information among the various subjects of social responsibility. Big data technology can make the performance of social responsibility, which is difficult to assess quantitatively, more measurable through diversified information carriers and realization means. Based on the non-linear, symbiotic, dynamic cycle characteristics of the symbiotic unit through coupling, nesting, co-evolution, and gradually from the “intermittent partiality symbiosis” to “continuous reciprocal symbiosis”, and ultimately form a digital, responsible green innovation “super system” of agriculture-related enterprises [24]. And the higher the quality of the social responsibility relationship established in the symbiosis model, the stronger the system symbiosis, the more it helps to improve the level of green innovation and symbiosis value of agriculture-related enterprises. At the same time, agriculture-related enterprises can contact more levels of technological innovation resources for quality integration in the symbiotic interface, reduce adverse selection and moral risk behavior, use R&D funds more rationally, and improve green innovation output and efficiency.

Finally, from the perspective of the symbiotic environment, all the external factors of the symbiotic unit constitute the symbiotic environment. In terms of the “hard environment”, China has formulated a series of policy documents to promote the development of agricultural digitization since 2015, with the basic logic of “developing digital infrastructure--promoting the application of digital technology—establishing an agricultural big data system” [25]. Digitalization has become an important indicator for the central government to investigate in the construction of modern agricultural projects and the declaration of modern agricultural subsidies. At the same time, the Food and Agriculture for Sustainable Transformation Initiative (FAST), launched by the United Nations Climate Change Conference, is also aimed at leading agriculture-related enterprises to fulfill their social responsibilities, utilize scientific and technological innovation, and practice the concept of sustainable development. In terms of “soft environment”, technological innovation by agribusiness enterprises requires the co-management of system, technology, organization, and culture. A good image of social responsibility can help agribusiness enterprises effectively realize social capital accumulation [26], improve their chances of obtaining financial credit [27], and promote green innovation [22] through long-term trust construction, interaction, and the construction of information channels.

In summary, agribusiness social responsibility performance can enhance the relationship between digital finance and agribusiness green innovation by promoting multi-subject interaction and benign interaction, acting on three elements: symbiosis unit, symbiosis model, and symbiosis environment, and thus realizing symbiotic value-added of the whole symbiosis system. Therefore, this paper puts forward the following hypotheses:

H2: 

Agribusiness social responsibility performance can enhance the facilitating effect of digital finance on green innovation performance.

H2a: 

Agribusiness social responsibility performance can enhance the full-spectrum contribution of digital finance to the quality of green innovation.

H2b: 

Agribusiness social responsibility performance can enhance the full-spectrum contribution of digital finance to the quantity of green innovation.

Based on the above theoretical analysis, this paper proposes a responsible digital green symbiosis model for the agricultural industry with reference to the ecological symbiosis system, as shown in Figure 1. In the symbiosis interface similar to nature, symbiotic units at all levels (platform finance, farmers, processing enterprises, agricultural technology application big data) realize the multi-channel flow of material flow, energy flow, responsibility flow, and information flow in the symbiotic environment (soft environment and hard environment). Through the platform connection, a digital-driven, innovation-sharing, responsibility-sharing green symbiotic situation will be formed and ultimately build a benign symbiotic model to promote the rise of the agricultural industry.

3. Methodology

3.1. Data Sources

This paper selects the industry classification of “Agriculture, forestry, animal husbandry and fishery”, “Agriculture and food processing”, “Food manufacturing”, “Alcohol, Beverage and Refined Tea Manufacturing”, “Wood Processing and Wood, Bamboo, Rattan, Palm and Grass Products Industry”, and “Chemical Raw Materials and Chemical Products Manufacturing”, in which the main business is pesticide and fertilizer production as well as agricultural machinery production as the data for the research sample (industry classification according to the China Securities Regulatory Commission industry classification standard, http://www.csrc.gov.cn/csrc/c100103/common_list.shtml, accessed on 8 October 2024). And the original samples were processed as follows: exclude the enterprises marked as Special Treatment (i.e., ST) and Particular Transfer (i.e., PT) by the China Securities Exchange Association, as fluctuations in the data of ST and PT companies may affect the accuracy of the research results; exclude companies listed in 2013 and later; exclude listed companies with gearing ratios exceeding 1 as well as samples with more missing values of variables. In addition, in order to eliminate the effect of extreme values, all the continuous variables involved were Winsorized at the 1% and 99% quantiles. The final result is 2404 sample observations.

The digital finance-related data used in this paper are from the Center for Digital Finance Research (IDF) of Peking University. Corporate green innovation patent data are collected from the database of the State Intellectual Property Office (SIPO) and the CNRDS database. CSR data are from the database of Shanghai Huazheng Index Information Service Co. Other financial data are from corporate annual reports.

3.2. Variable Design

(1) Agribusiness Green Innovation. Agricultural patents can directly and objectively respond to agricultural innovation, and the number of patents has a good time continuity [28]. According to China’s patent law, enterprise innovation patents are categorized into three levels: invention patents, utility model patents, and design patents. In this study, the number of green invention patents and the number of green new type utility patents are used as indicators to measure the green innovation performance of agriculture-related enterprises. The number of green invention patents focuses on measuring the “qualitative” level of green innovation of agriculture-related enterprises (lnGreen_qua), and the number of green utility model patents focuses on measuring the “quantitative” level of green innovation of agriculture-related enterprises (lnGreen_num). Due to the existence of the number of zero cases, and in order to solve the problem of right-skewed distribution of data, the empirical analysis of this indicator is logarithmic treatment after adding 1.

(2) Digital Finance. Referring to existing studies [20], this paper selects the Digital Inclusive Finance Index compiled by the Digital Finance Research Center of Peking University for analysis. The index is constructed based on Alibaba Ant Financial Service’s big data, including 31 provinces, 337 cities above the prefecture level, and nearly 2800 counties in China’s mainland, with a wide coverage [29]. This paper mainly selects provincial-level data and further incorporates the secondary indicators: breadth of digital finance coverage (lnBre), depth of digital finance use (lnDep), and degree of digitization (lnDig) into the study to examine their impact on the “quality” and “quantity” of green innovation in agribusiness. In view of the relatively large size of the data, the index was also logarithmized after adding 1.

(3) Agribusiness Social Responsibility. Sino-Securities Index Information Service (Shanghai) Co., Ltd. is a third-party independent company specializing in index and indexing investment; corporate environment, society, and governance (ESG) performance rating; and consulting services. Its ESG evaluation system comprehensively evaluates the level of corporate social responsibility and sustainable development ability of listed companies from three dimensions of environment, society, and corporate governance, including three primary indicators, 14 secondary indicators, 26 tertiary indicators, and more than 130 underlying data indicators, which is the longest retrospective evaluation system in the Chinese market at present, and the data are fairer and more authoritative. In this paper, we use the total annual ESG scores of enterprises in the Huazheng ESG report to measure the social responsibility performance of agribusiness enterprises. Again, considering the magnitude issue, the score is processed by adding 1 to take the logarithm.

(4) Control Variables. In order to minimize the impact of omitted variables on the results, this study sets control variables which include the following: agricultural research investment intensity (RDI) refers to the ratio of the amount of R&D investment in agribusiness enterprises to their operating revenues. Enterprise maturity (Age) is the natural logarithm of the number of years of establishment. Enterprise growth (Growth) refers to the growth rate of main business revenue of agribusiness enterprises. Shareholding concentration (Share) refers to the ratio of the number of shares held by the first largest shareholder to the total number of shares. Gearing ratio (Lev) refers to the ratio of total liabilities to total assets. Utilization efficiency of corporate assets (Roa) refers to the ratio of net profit to total assets. Corporate financing environment (WW_index) is the calculated WW index of corporate-level financing constraint values. Enterprise size (Size) is the natural logarithm of the annual total assets of agriculture-related enterprises. In addition, this paper sets and fixes province (PROV), industry segment (IND), and year (YEAR) dummy variables.

3.3. Model Construction

In order to study the impact of digital finance on the level of green innovation in agribusiness, model (1) is constructed.

lnGreen_qua_it/lnGreen_num_it = α₀ + α₁lnDf_it + α₂RDI_it + α₃Age_it + α₄Growth_it + α₅Share_it + α₆Lev_it + α₇Roa_it + α₈WW_index_it + α₉Size_it + ∑PROV + ∑IND + ∑YEAR + ε

(1)

In order to investigate the moderating effect of social responsibility performance on the relationship between digital finance and green innovation in agribusiness, model (2) is constructed.

lnGreen_qua_it/lnGreen_num_it = α₀ + α₁lnDf_it + α₂lnCSR_it + α₃Df_it*CSR_it + α₄RDI_it + α₅Age_it + α₆Growth_it + α₇Share_it + α₈Lev_it + α₉Roa_it + α₁₀WW_index_it + α₁₁Size_it + ∑PROV + ∑IND + ∑YEAR + ε

(2)

In the model, lnGreen_qua_i,t and lnGreen_num_i,t reflect the green innovation output level of agribusiness i at time t, which is examined from two dimensions of quality and quantity. lnDf_i,t represents the digital financial level of agribusiness i at time t; lnCSR_i,t represents the social responsibility performance of agribusiness i at time t. The remaining variables in the model are the control variables chosen in this paper. ∑PROV represents the province fixed effect; ∑IND represents the industry fixed effect; ∑Year represents the year fixed effect. ε represents the random error term of the model.

Model (1) mainly tests hypotheses H1a and H1b. If α₁ > 0, it means that digital finance promotes the green technology innovation performance of agricultural enterprises; model (2) mainly tests hypotheses H2a and H2b. If α₁ > 0 and α₃ > 0, it means that the CSR performance of agricultural enterprises strengthens the relationship between digital finance and green innovation of agricultural enterprises.

Data processing and estimation in this paper are mainly performed using Stata 15.0 (Stata Corp., College Station, TX, USA). Firstly, according to the results of the Hausman test, the p-values of models (1) and (2) are both significantly 0 under the 1% condition, so the fixed effect model should be selected for estimation. Second, a preliminary statistical test is performed using ordinary least squares (OLS). According to White’s test, since p > chi2 of models (1) and (2) is equal to 0 and the null hypothesis of stability is rejected, the regression results are highly likely to be confounded by heteroskedasticity. According to existing studies, the influence of potential residual correlation and heteroskedasticity on the significance inference of the estimated coefficients can be eliminated by applying robust standard errors or weighted least squares (WLS) [30]. Therefore, in this study, the inverse of the residual square fit of the regression is chosen as the weight for WLS estimation. At the same time, the standard errors of the regression results are clustered at the firm level, and the t-values of all regressions are adjusted by the cluster of the firm dimension.

4. Results and Discussion

4.1. Descriptive Statistics Analysis

Table 1. Descriptive statistics.

Variables	Obs	Mean	Std. Dev.	Min	Max
lnGreen_qua	2404	1.169	1.492	0	4.097
lnGreen_num	2404	1.517	1.365	0	3.296
lnDf	2404	5.687	0.398	4.277	6.134
lnBre	2404	5.607	0.354	3.479	6.122
lnDep	2404	5.664	0.336	4.536	6.235
lnDig	2404	5.902	0.399	3.384	6.147
lnCSR	2404	4.304	0.176	3.923	4.499
RDI	2404	2.310	1.853	0	14.480
Age	2404	2.983	0.287	1.791	3.850
Growth	2404	0.209	1.426	−0.825	58.749
Share	2404	34.865	14.262	4.160	89.091
Lev	2404	0.389	0.190	0.019	1.905
ROA	2404	0.046	0.085	−1.395	0.526
WW_index	2404	−0.992	0.080	−3.022	−0.082
Size	2404	22.258	1.103	19.940	26.616

reports the descriptive statistics of the main variables. The maximum value of the quality level of green innovation of enterprises, lnGreen_qua, is 4.097, and the maximum value of the quantity level, lnGreen_num, is 3.296, and the minimum value of both is 0, which indicates that there is a large gap in the green innovation output capacity among agribusiness enterprises. The mean value is 1.169/1.517, both of which are at a low level, indicating that the green innovation development of agribusiness enterprises should be given sufficient attention. The mean of lnDf, the level of digital finance development, is 5.687, and the standard deviation is 0.398, indicating that the overall development of digital finance in the provinces where the sample enterprises are located is good, but there are large differences between regions.

4.2. Regression Analysis and Hypothesis Tests

4.2.1. Digital Finance and Agribusiness Green Innovation

The fixed effect regression method was used in this paper to estimate the model (1), and the test results are shown in

Table 2. Impact of digital finance on the quality and quantity of green innovation in agribusinesses.

Variables	lnGreen_qua				lnGreen_num
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
lnDf	0.286 ***				0.448 ***
	(2.97)				(6.33)
lnBre		0.271 ***				0.450 ***
		(2.87)				(6.65)
lnDep			0.437 **				0.264 ***
			(2.27)				(5.54)
lnDig				0.140 **				0.116
				(2.29)				(0.77)
RDI	0.026 ***	0.026 ***	0.026 ***	0.025 ***	0.105 ***	0.106 ***	0.104 ***	0.108 ***
	(3.89)	(3.84)	(3.87)	(3.81)	(2.95)	(3.11)	(2.84)	(2.45)
Age	−0.005	−0.005	−0.004	−0.005	0.008	0.009	0.008	0.009
	(−0.61)	(−0.65)	(−0.59)	(−0.63)	(0.41)	(0.50)	(0.34)	(0.55)
Growth	0.025 *	0.026 *	0.026 *	0.026 *	0.024 *	0.024 *	0.024 *	0.024 *
	(1.66)	(1.80)	(1.81)	(1.83)	(1.81)	(1.74)	(1.82)	(1.79)
Share	−0.008	−0.008	−0.008	−0.008	−0.006 ***	−0.006 ***	−0.005 ***	−0.005 **
	(−1.57)	(−1.55)	(−1.49)	(−1.46)	(−2.68)	(−2.72)	(−2.61)	(−2.32)
Lev	−0.158 ***	−0.157 ***	−0.159 ***	−0.161 ***	−0.168 ***	−0.165 **	−0.169 **	−0.171 **
	(−3.03)	(−3.00)	(−3.05)	(−3.09)	(−2.60)	(−2.55)	(−2.63)	(−2.65)
Roa	0.533 *	0.531 ***	0.536 ***	0.542 ***	0.356 ***	0.351 ***	0.360 ***	0.364 ***
	(4.12)	(4.11)	(4.14)	(4.18)	(3.93)	(3.87)	(3.98)	(3.96)
WW_index	−0.122	−0.128	−0.119	−0.128	−0.132	−0.141	−0.122	−0.148
	(−0.68)	(−0.72)	(−0.67)	(−0.71)	(−0.94)	(−1.00)	(−0.87)	(−1.06)
Size	0.063 *	0.063 ***	0.063 ***	0.063 ***	0.057 ***	0.057 ***	0.057 ***	0.056 ***
	(6.06)	(6.06)	(6.02)	(6.02)	(6.75)	(6.75)	(6.73)	(6.65)
PROV	YES	YES	YES	YES	YES	YES	YES	YES
IND	YES	YES	YES	YES	YES	YES	YES	YES
YEAR	YES	YES	YES	YES	YES	YES	YES	YES
Constant	0.425	0.341	−0.378	1.429	1.557 ***	1.557 ***	0.510	−0.105
	(0.49)	(0.40)	(−0.53)	(1.12)	(2.87)	(3.03)	(1.04)	(−0.11)
Observations	2404	2404	2404	2404	2404	2404	2404	2404
R-squared	0.297	0.297	0.296	0.296	0.403	0.404	0.401	0.392
F test	0	0	0	0	0	0	0	0
Adj_R2	0.280	0.280	0.279	0.280	0.386	0.388	0.384	0.375
F	26.23	25.73	30.29	21.02	53.90	55.61	52.46	50.10

Robust t-statistics in parentheses. *** p < 0.01, ** p < 0.05, and * p < 0.1; the same as below.

. It can be seen that the embeddedness of digital finance in agribusiness enterprises has a positive contribution to both the quality and quantity dimensions of their green innovation output and passes the 1% significance level test (β = 0.286, p < 0.01; β = 0.448, p < 0.01). Models (2)–(4) and (6)–(8) report the three structural variance sub-subjects of embeddedness in the digital finance dimensions on the green innovation output of agribusiness firms, respectively. It can be seen that the breadth of coverage and depth of use of digital finance embedded in enterprises have a driving effect on the quality and quantity of green innovation of agribusiness enterprises, but the effect of the degree of digitization on the quantity of green innovation of enterprises is not significant (β = 0.116, p > 0.1). This suggests that the development of digital finance can drive the agricultural enterprises embedded in it to enhance their green innovation capacity and improve the “quality and quantity” of green innovation output, which is mainly realized through the expansion of the breadth of digital financial embeddedness and the increase of the depth of embeddedness.

The reason may be that the dimensions of embedding depth and embedding breadth better reflect the mitigating effect of digital inclusive finance on the financing constraints of rural areas as well as agribusinesses, thus stimulating the enhancement of the green innovation efficiency and capability of agribusinesses. Hypothesis 1 of this study is partially verified.

4.2.2. Impact of Agribusiness Social Responsibility Performance

In order to test the impact effect of agribusiness CSR performance, this study conducts a moderated effects test based on model (2), and the specific regression results are shown in

Table 3. Impact of agribusiness social responsibility performance on the relationship between digital finance and green innovation quality.

Variables	lnGreen_qua
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
lnDf	0.203 ***	0.203 ***
	(3.09)	(3.12)
lnCSR	0.265 *	0.258	0.262 *	0.263	0.258 *	0.258	0.245	0.183
	(1.69)	(1.51)	(1.66)	(1.55)	(1.65)	(1.58)	(1.59)	(0.96)
Df*CSR		0.186 **
		(2.14)
lnBre			0.265 ***	0.265 ***
			(2.96)	(2.90)
Bre*CSR				0.107 **
				(2.04)
lnDep					0.351 **	0.351 **
					(2.43)	(2.42)
Dep*CSR						0.191 **
						(2.02)
lnDig							0.145 **	0.146 **
							(2.35)	(2.36)
Dig*CSR								0.157
								(0.72)
RDI	0.026 ***	0.026 ***	0.025 ***	0.025 ***	0.026 ***	0.026 ***	0.025 ***	0.025 ***
	(3.83)	(3.84)	(3.78)	(3.78)	(3.82)	(3.82)	(3.75)	(3.76)
Age	−0.004	−0.004	−0.005	−0.005	−0.004	−0.004	−0.004	−0.004
	(−0.56)	(−0.56)	(−0.61)	(−0.61)	(−0.54)	(−0.54)	(−0.58)	(−0.57)
Growth	0.025 *	0.025 *	0.025 *	0.025 *	0.025 *	0.025 *	0.025 *	0.025 *
	(1.69)	(1.71)	(1.67)	(1.67)	(1.69)	(1.70)	(1.72)	(1.70)
Share	−0.008	−0.008	−0.008	−0.008	−0.007	−0.007	−0.007	−0.007
	(−1.42)	(−1.42)	(−1.43)	(−1.42)	(−1.37)	(−1.37)	(−1.33)	(−1.35)
Lev	−0.179 ***	−0.179 ***	−0.177 ***	−0.177 ***	−0.180 ***	−0.180 ***	−0.181 ***	−0.179 ***
	(−3.30)	(−3.29)	(−3.26)	(−3.26)	(−3.31)	(−3.31)	(−3.32)	(−3.30)
Roa	0.520 ***	0.520 ***	0.518 ***	0.518 ***	0.524 ***	0.524 ***	0.530 ***	0.527 ***
	(4.12)	(4.11)	(4.10)	(4.10)	(4.14)	(4.14)	(4.18)	(4.15)
WW_index	−0.103	−0.103	−0.110	−0.110	−0.100	−0.100	−0.111	−0.108
	(−0.58)	(−0.58)	(−0.62)	(−0.62)	(−0.56)	(−0.56)	(−0.62)	(−0.60)
Size	0.060 ***	0.060 ***	0.060 ***	0.060 ***	0.059 ***	0.059 ***	0.060 ***	0.059 ***
	(5.50)	(5.52)	(5.51)	(5.54)	(5.47)	(5.47)	(5.49)	(5.49)
PROV	YES	YES	YES	YES	YES	YES	YES	YES
IND	YES	YES	YES	YES	YES	YES	YES	YES
YEAR	YES	YES	YES	YES	YES	YES	YES	YES
Constant	−1.188 **	−1.189 **	−1.165 **	−1.165 *	−1.162 *	−1.162 *	−1.135 *	−1.141 *
	(−2.00)	(−2.00)	(−1.96)	(−1.96)	(−1.95)	(−1.95)	(−1.88)	(−1.88)
Observations	2404	2404	2404	2404	2404	2404	2404	2404
R-squared	0.299	0.299	0.299	0.299	0.297	0.297	0.298	0.298
F test	0	0	0	0	0	0	0	0
Adj_R2	0.284	0.284	0.285	0.281	0.280	0.279	0.280	0.280
F	32.76	30.41	28.86	28.79	39.07	38.45	23.67	20.34

Robust t-statistics in parentheses. *** p < 0.01, ** p < 0.05, and * p < 0.1; the same as below.

and

Table 4. Impact of agribusiness social responsibility performance on the relationship between digital finance and green innovation quantities.

Variables	lnGreen_num
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
lnDf	0.436 ***	0.417 ***
	(6.47)	(6.45)
lnCSR	0.242 **	0.269 **	0.241 **	0.263 **	0.240 **	0.267 **	0.198 *	0.241 **
	(2.20)	(2.32)	(2.19)	(2.29)	(2.17)	(2.31)	(1.79)	(2.02)
Df*CSR		0.332 ***
		(2.59)
lnBre			0.463 ***	0.462 ***
			(6.78)	(6.78)
Bre*CSR				0.313 ***
				(2.71)
lnDep					0.274 ***	0.273 ***
					(5.71)	(5.70)
Dep*CSR						0.249
						(0.90)
lnDig							0.127	0.120
							(0.84)	(0.79)
Dig*CSR								0.727
								(0.87)
RDI	0.106 ***	0.106 ***	0.106 ***	0.106 ***	0.105 ***	0.105 ***	0.108 ***	0.108 ***
	(3.05)	(3.07)	(3.22)	(3.23)	(2.93)	(2.96)	(3.54)	(3.56)
Age	0.008	0.008	0.009	0.009	0.008	0.008	0.009	0.009
	(0.35)	(0.35)	(0.44)	(0.45)	(0.27)	(0.27)	(0.50)	(0.52)
Growth	0.023	0.023	0.023	0.023	0.023	0.023	0.023	0.023
	(1.50)	(1.37)	(1.43)	(1.31)	(1.52)	(1.33)	(1.54)	(1.55)
Share	−0.005 **	−0.005 **	−0.005 **	−0.005 **	−0.005 **	−0.005 **	−0.005 **	−0.005 **
	(−2.47)	(−2.46)	(−2.52)	(−2.50)	(−2.40)	(−2.39)	(−2.14)	(−2.12)
Lev	−0.187 **	−0.188 ***	−0.184 **	−0.185 **	−0.188 ***	−0.190 ***	−0.186 **	−0.187 **
	(−2.58)	(−2.60)	(−2.52)	(−2.53)	(−2.60)	(−2.64)	(−2.56)	(−2.57)
Roa	0.344 ***	0.345 ***	0.339 ***	0.340 ***	0.349 ***	0.350 ***	0.355 ***	0.356 ***
	(3.79)	(3.81)	(3.73)	(3.75)	(3.85)	(3.87)	(3.85)	(3.87)
WW_index	−0.115	−0.116	−0.124	−0.126	−0.105	−0.105	−0.134	−0.136
	(−0.81)	(−0.81)	(−0.87)	(−0.88)	(−0.74)	(−0.74)	(−0.95)	(−0.96)
Size	0.054 ***	0.054 ***	0.054 ***	0.054 ***	0.054 ***	0.054 ***	0.054 ***	0.054 ***
	(6.38)	(6.39)	(6.39)	(6.39)	(6.37)	(6.37)	(6.35)	(6.35)
PROV	YES	YES	YES	YES	YES	YES	YES	YES
IND	YES	YES	YES	YES	YES	YES	YES	YES
YEAR	YES	YES	YES	YES	YES	YES	YES	YES
Constant	−0.980 **	−0.977 **	−0.957 **	−0.955 **	−0.974 **	−0.972 **	−0.777 *	−0.773 *
	(−2.19)	(−2.18)	(−2.14)	(−2.13)	(−2.19)	(−2.19)	(−1.76)	(−1.75)
Observations	2404	2404	2404	2404	2404	2404	2404	2404
R-squared	0.405	0.405	0.406	0.407	0.403	0.403	0.393	0.394
F test	0	0	0	0	0	0	0	0
Adj_R2	0.387	0.387	0.389	0.389	0.385	0.385	0.376	0.375
F	77.72	77.60	85.34	84.06	68.15	65.61	49.31	47.92

Robust t-statistics in parentheses. *** p < 0.01, ** p < 0.05, and * p < 0.1; the same as below.

. The results of columns (2), (4), and (6) in Table 3 show that the coefficients of the cross-multiplier term, Df*CSR, pass the test of significance at the 5% level (β = 0.186, p < 0.05; β = 0.107, p < 0.05; β = 0.191, p < 0.05); and the results of columns (2) and (4) in Table 4 show that the coefficients of the cross-multiplier term Df*CSR passed the test of significance at the 1% level (β = 0.332, p < 0.01; β = 0.313, p < 0.01). Social responsibility performance enhances the relationship between digital finance and some of its sub-dimensions and green innovation output of agribusinesses, and hypothesis H2 is partially tested.

This suggests that agricultural enterprises can build and maintain a good symbiotic network relationship by integrating a sense of social responsibility when dealing with various stakeholders in their daily agricultural production and operation activities. By improving the heterogeneous strategic resources such as reputation, social capital, and customer response, which are difficult to obtain through market transactions in the symbiotic network [24,26], the symbiotic units at all levels are encouraged to shift towards more positive innovation output.

4.2.3. Robustness Tests

In order to verify the robustness of the baseline regression and moderated effects regression results, this paper first replaces the estimation model, adopts the Tobit model for testing, and refers to the method of Tang Song et al. [18] to consider the higher-order joint fixed effects of “industry*year”, and the specific regression results are shown in

Table 5. Robustness test (1).

Variables	lnGreen_qua			lnGreen_num
	(1)	(2)	(3)	(4)	(5)	(6)
lnDf	0.320 ***	0.289 ***	0.280 ***	0.602 ***	0.553 ***	0.530 ***
	(3.56)	(3.22)	(3.51)	(7.78)	(7.86)	(7.82)
lnCSR		0.320 **	0.301 *		0.320 ***	0.353 ***
		(2.24)	(1.86)		(2.81)	(2.86)
Df*CSR			0.153 ***			0.373 **
			(4.48)			(2.48)
Controls	YES	YES	YES	YES	YES	YES
Constant	0.882	−3.148 **	−3.157 **	2.678 ***	−1.248 ***	−1.219 ***
	(1.07)	(−2.40)	(−2.36)	(3.82)	(−3.08)	(−3.03)
IND	YES	YES	YES	YES	YES	YES
YEAR	YES	YES	YES	YES	YES	YES
IND*YEAR	YES	YES	YES	YES	YES	YES
N	2404	2404	2404	2404	2404	2404

Robust t-statistics in parentheses. *** p < 0.01, ** p < 0.05, and * p < 0.1.

. Through models (1) and (3) and (4) and (6), it can be seen that the estimated coefficients of lnDf are significant at the 1% level, and the estimated coefficients of the cross-multiplier term Df*CSR are still significantly positive, indicating that the role of digital finance in enhancing the green innovation performance of agribusiness enterprises and positively promoting the performance of social responsibility still remains robust (β = 0.153, p < 0.01; β = 0.373, p < 0.05).

Secondly, this paper further controls the regional marketization process (MI), regional economic development level (Pgdp), and regional agriculture, forestry, animal husbandry, and fishery development level (Agdp) [3]. Meanwhile, the variable measurement is replaced, and the digital finance index at the municipal level is used as a substitute index for robustness analysis, and the specific results are shown in

Table 6. Robustness test (2).

Variables	lnGreen_qua			lnGreen_num
	(1)	(2)	(3)	(4)	(5)	(6)
lnDf	0.293 ***	0.210 **	0.206 **	0.438 ***	0.421 ***	0.390 ***
	(2.99)	(2.56)	(2.58)	(6.85)	(6.92)	(6.92)
lnCSR		0.262 *	0.251 *		0.253 **	0.284 **
		(1.85)	(1.69)		(2.42)	(2.51)
Df*CSR			0.193 **			0.311 **
			(2.38)			(2.43)
Controls	YES	YES	YES	YES	YES	YES
Constant	0.701	−2.202 *	−2.250 *	1.553	−0.846 *	−0.832 *
	(0.96)	(−1.82)	(−1.89)	(1.38)	(−1.81)	(−1.81)
CITY	YES	YES	YES	YES	YES	YES
IND	YES	YES	YES	YES	YES	YES
YEAR	YES	YES	YES	YES	YES	YES
Observations	2404	2404	2404	2404	2404	2404
R-squared	0.292	0.288	0.290	0.434	0.439	0.439
F test	0	0	0	0	0	0
Adj_R2	0.274	0.272	0.272	0.420	0.422	0.422
F	33.57	38.74	37.05	51.56	66.71	67.09

Robust t-statistics in parentheses. *** p < 0.01, ** p < 0.05, and * p < 0.1.

. According to the results in the table, there is no significant difference in the main estimation results, and the positive effect of digital finance on the agribusiness enterprises’ green innovation performance remains significant, and the results of the moderating effect of the performance of social responsibility are also in line with the above.

Finally, in order to avoid potential endogeneity problems as much as possible, such as the correlation between digital finance and the innovation performance of agribusinesses that may be jointly influenced by other factors at the same time, this paper uses the two-stage instrumental variable method (IV-2SLS) to monitor. Referring to the study of Xie X.L. et al. (2018) [31], this paper uses Internet penetration (IP) as an instrumental variable. Generally speaking, a higher Internet penetration rate means a newer financial industry, financial products, and financial service system, which will prompt agribusinesses to be more easily embedded in digital financial development in their business processes, but exogenously it will not directly affect individual green innovation output. The instrumental variable test results show that the Kleibergen–Paap rk LM statistic is 329.51 (p_val of 0.000), rejecting the original hypothesis of non-identifiability. The Kleibergen–Paaprk Wald F statistic is 247.59, which is greater than the 10% Stock–Yogo criterion (17.83), and the weak instrumental variable risk is low. The regression results, as shown in

Table 7. Robustness test (3).

Variables	(1)	(2)	(3)	(4)
	First Stage	Second Stage	First Stage	Second Stage
	lnDf	lnGreen_qua	lnDf	lnGreen_num
IP	0.098 ***		0.098 ***
	(3.41)		(3.41)
lnDf		0.186 ***		0.279 ***
		(5.31)		(9.02)
Controls	Control	Control	Control	Control
PROV/IND/YEAR	Control	Control	Control	Control
Constant	−1.525 ***	0.803 **	−1.525 ***	1.294 ***
	(2.74)	(2.31)	(2.74)	(5.76)
Observations	2404	2404	2404	2404
R-squared	0.453	0.247	0.453	0.321

Robust t-statistics in parentheses. *** p < 0.01, and ** p < 0.05.

, show that Internet penetration (IP) is significantly associated with the level of digital finance, which still positively contributes to the green innovation output of agribusiness and passes the 1% significance level test. This indicates that the conclusions of the main hypothesis of this paper remain robust after the endogeneity problem is mitigated by adding instrumental variables for the two-stage least squares regression.

4.3. Heterogeneity Analysis

4.3.1. Heterogeneity Based on Industry Chain Location

Considering the different stakeholder groups (symbiotic units) faced by agricultural enterprises located in different subsectors (ecological niches) in the agricultural industrial chain, the same stakeholder group (symbiotic units) also has different needs for different types of enterprises (symbiotic models). This paper divides agricultural enterprises into two types—planting and raising enterprises and processing and distribution enterprises—according to the different positions of agricultural enterprises in industrial clusters [32] and further analyzes the samples of these two types of agricultural enterprises. Among them, the planting and raising agribusiness enterprises (sub-sample 1) include 81 enterprises in category A, agriculture, forestry, animal husbandry, and fishery, while processing and distribution agribusiness enterprises (sub-sample 2) include the remaining 314 enterprises in agro-food processing and other industries. The two subsamples are tested using group regression. The specific regression results are shown in

Table 8. Heterogeneity based on industry chain location.

	lnGreen_qua				lnGreen_num
Variables	Sub-Sample 1		Sub-Sample 2		Sub-Sample 1		Sub-Sample 2
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
lnDf	1.091	1.055	0.275 **	0.273 **	0.627 **	0.613 ***	0.434 ***	0.405 ***
	(1.38)	(1.24)	(2.40)	(2.54)	(2.23)	(2.98)	(3.17)	(3.35)
lnCSR		0.173		0.292 *		0.410		0.355 ***
		(1.53)		(1.80)		(1.08)		(3.01)
Df*CSR		0.561		0.098 **		0.226 *		0.344 ***
		(0.36)		(2.41)		(1.76)		(2.67)
Controls	YES	YES	YES	YES	YES	YES	YES	YES
Constant	0.148	−1.902	0.332	−1.885 **	1.019 ***	−1.156 ***	1.898 ***	−1.544 ***
	(0.96)	(−1.49)	(0.35)	(−2.46)	(2.68)	(−3.46)	(3.28)	(−2.76)
PROV/IND/YEAR	YES	YES	YES	YES	YES	YES	YES	YES
Observations	485	485	1889	1889	485	485	1889	1889
R-squared	0.257	0.259	0.492	0.493	0.362	0.364	0.490	0.494
F test	0	0	0	0	0	0	0	0
Adj_R2	0.172	0.179	0.478	0.479	0.323	0.327	0.473	0.480
F	24.55	24.12	77.25	86.92	28.97	28.19	77.10	86.97

Robust t-statistics in parentheses. *** p < 0.01, ** p < 0.05, and * p < 0.1.

.

The results show that there is a difference in the explanatory power of digital finance on the green innovation performance of agribusiness. Comparing the results of the two sub-samples, it can be found that the variance explaining ability of the processing and distribution sample is significantly larger than that of the plantation companies. The enhancing effect of digital finance on the quality of green innovation works only in the processing category, and the enhancing effect on the quantity of green innovation is more significant in the processing category (chi2(1) = 13.87, Prob > chi2 = 0.0000). The positive impact of social responsibility performance is more pronounced in the sample of processing category firms. This may be due to the fact that planting and raising class agribusiness is directly related to agriculture, forestry, animal husbandry, and fishery industries. In order to reduce the harm of natural risks, planting and raising class-listed companies generally choose semi-industrial chain or whole industrial chain forms of agricultural business organizations to carry out agricultural production and operation activities. In contrast, most of the processing and distribution agribusinesses are in a node in the agricultural industry chain, interconnected and symbiotic with a wider range of external stakeholders, so a good performance of social responsibility can enable the processing and distribution agribusinesses to establish a stronger symbiotic relationship with more symbiotic units in the symbiotic interface, to obtain all kinds of necessary resources through the smooth digital information channel, and to realize the good growth of enterprise green innovation.

4.3.2. Heterogeneity Based on Geographic Location

According to the regional division of the National Bureau of Statistics of China, Chinese provinces can be categorized into the eastern region and the central and western region. In this paper, we further subdivide the provinces where the main business activities of agribusinesses take place and divide the sample into the eastern region group (sub-sample 1) and the central and western region group (sub-sample 2) for regression analysis, respectively. The specific results are shown in

Table 9. Heterogeneity based on industry geographic location.

	lnGreen_qua				lnGreen_num
Variables	Sub-Sample 1		Sub-Sample 2		Sub-Sample 1		Sub-Sample 2
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
lnDf	0.502 ***	0.503 ***	0.351 **	0.282 **	0.818 ***	0.808 ***	0.500 ***	0.453 ***
	(3.43)	(3.41)	(2.20)	(2.01)	(7.53)	(7.48)	(4.83)	(4.68)
lnCSR		0.486 ***		0.240		0.516 ***		0.237 *
		(3.17)		(1.03)		(2.61)		(1.66)
Df*CSR		0.274 **		0.365 *		0.705 ***		0.443 **
		(2.36)		(1.84)		(3.23)		(2.36)
Controls	YES	YES	YES	YES	YES	YES	YES	YES
Constant	1.397	−2.028 **	1.823	−2.346	2.562 ***	−1.045 **	2.895 ***	−1.448 **
	(1.13)	(−2.53)	(1.51)	(−1.54)	(6.00)	(−2.09)	(5.22)	(−2.56)
PROV/IND/YEAR	YES	YES	YES	YES	YES	YES	YES	YES
Observations	1368	1368	1036	1036	1368	1368	1036	1036
R-squared	0.322	0.323	0.330	0.340	0.431	0.433	0.447	0.453
F test	0	0	0	0	0	0	0	0
Adj_R2	0.297	0.297	0.302	0.310	0.407	0.408	0.419	0.423
F	34.55	34.65	34.76	34.97	45.15	45.22	45.23	45.44

Robust t-statistics in parentheses. *** p < 0.01, ** p < 0.05, and * p < 0.1.

.

The results show that the impact of digital finance on the green innovation performance of agribusiness in the eastern region is greater than that of the sample from the central and western regions. The positive contribution of social responsibility performance of the eastern region sample is significantly stronger than that of the central and western regions sample. And the difference passed the SUEST test at the 1% significance level. The reason for this is that the eastern region of China has a higher degree of financial marketization, which can provide better application conditions for the development of digital finance, so the increase in green innovation performance of agribusinesses in the central and western regions is smaller in the case of insufficient incentives for digital financial support. At the same time, agribusiness is sensitive to the external environment, and eastern China is more superior in both knowledge sharing and resource exchange and has a stronger atmosphere for fulfilling social responsibility. This also reflects that the more support agribusinesses are able to obtain from symbiotic units in the interface through social responsibility performance, the more capable they are to carry out green innovation.

4.3.3. Heterogeneity Based on Individual Growth Stages

Based on the theory of the enterprise life cycle, an enterprise is an organization with a class of life state; there are large differences in the production and operation and organizational characteristics of enterprises in different stages of the life cycle, and it is necessary to think about the business strategy in a contingent manner. In this paper, with reference to the study of Liang S.k. et al. (2019) [33], the life cycle of the sample enterprises is divided into the growth period (sub-sample 1) and the maturity period (sub-sample 2) by using the comprehensive score discrimination method to carry out regression, respectively, and the results are shown in

Table 10. Heterogeneity based on individual growth stages.

	lnGreen_qua				lnGreen_num
Variables	Sub-Sample 1		Sub-Sample 2		Sub-Sample 1		Sub-Sample 2
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
lnDf	0.330 ***	0.304 ***	0.264 **	0.244 **	0.328 ***	0.277 ***	0.289 ***	0.262 ***
	(2.84)	(2.75)	(2.05)	(1.95)	(5.42)	(4.36)	(4.54)	(4.41)
lnCSR		0.133		0.283 *		0.176		0.308 **
		(0.45)		(1.94)		(1.29)		(2.32)
Df*CSR		0.314		0.344 **		0.646		0.413 **
		(1.08)		(2.50)		(1.48)		(2.49)
Controls	YES	YES	YES	YES	YES	YES	YES	YES
Constant	0.363	−1.018	0.552	−1.293 **	2.410	−1.955	1.454 **	−1.300 ***
	(1.00)	(−0.44)	(0.66)	(−2.21)	(1.51)	(−1.60)	(2.24)	(−2.85)
PROV/IND/YEAR	YES	YES	YES	YES	YES	YES	YES	YES
Observations	740	740	1124	1124	740	740	1124	1124
R-squared	0.247	0.266	0.302	0.304	0.405	0.407	0.412	0.415
F test	0	0	0	0	0	0	0	0
Adj_R2	0.236	0.258	0.283	0.284	0.366	0.371	0.386	0.388
F	40.97	41.05	45.44	45.60	41.14	41.44	45.37	45.64

Robust t-statistics in parentheses. *** p < 0.01, ** p < 0.05, and * p < 0.1.

.

The regression results can be found that the positive effect of digital inclusive finance on the green innovation capacity of agribusiness is more obvious in the growth period of enterprises (chi2(1) = 17.59, Prob > chi2 = 0.0000). The positive facilitating effect of CSR performance is only significant in the maturity sample. This may be due to the fact that digital financial inclusion is more conducive to alleviating the various innovation constraints of agribusinesses in the growth stage and can create more favorable symbiotic conditions for growth stage firms, making it more likely that growth stage firms will achieve green business. However, due to the lower resilience of growth stage enterprises, their social responsibility performance cannot play a positive role in enhancing innovation. In the maturity stage, enterprises gradually form a perfect governance system and business model. Therefore, only when agribusinesses are in the maturity stage, can their social responsibility performance enhance the symbiotic relationship with the industry and the external environment, realize the exchange of resources, and increase the output of green innovation.

4.4. Value Effect Test

Based on the above research, this paper analyzes whether digital financial embedding under the symbiosis perspective can have an impact on the high-quality development of agribusiness through the green innovation effect. Regarding the measurement of the level of high-quality development of enterprises, this paper refers to James and Amil’s research method [34] and uses Levinsohn and Petrin’s measure (referred to as the LP method) to estimate the total factor productivity of agribusiness, Y_i,t = AK^∂L^ß. When deformed, $\mathrm{T}\mathrm{F}\mathrm{P}\_\mathrm{L}\mathrm{P}\mathrm{i},\mathrm{t}={\displaystyle \frac{Y}{K^{\mathsf{\partial }}{L}^{ß}}}$ Where Y is output, measured by operating income; A is total factor productivity (TFP); K is capital input, measured by net fixed assets; L is labor input, measured by the number of persons employed in the enterprise. The logarithms of the above input-output indicators are taken and then estimated. ∂ and ß are the output elasticity coefficients of capital and labor, respectively.

Table 11. Value effect test.

	(1)	(2)	(3)
Variables	TFP_LP	TFP_LP	TFP_LP
lnDf	0.223 ***	0.186 ***	0.197 ***
	(2.94)	(3.05)	(3.83)
lnGreen_qua		0.218 ***
		(2.87)
lnGreen_num			0.205 **
			(2.25)
Controls	YES	YES	YES
PROV	YES	YES	YES
IND	YES	YES	YES
YEAR	YES	YES	YES
Constant	−4.185 ***	−4.197 ***	−4.177 ***
	(−3.64)	(−3.66)	(−3.62)
Observations	2281	2281	2281
R-squared	0.573	0.574	0.573
F test	0	0	0
Adj_R2	0.569	0.569	0.569
F	27.15	27.73	27.13

Robust t-statistics in parentheses. *** p < 0.01, and ** p < 0.05.

reports the empirical results of integrating digital finance, green innovation output, and the total factor productivity of agribusiness. The results show that both digital finance and green innovation performance have significant positive effects on agribusiness total factor productivity. Panel (1) and Panel (2) in the table show that digital finance can improve the total factor productivity of enterprises by improving the quality of green innovation in agriculture-related enterprises (β = 0.186, p < 0.01; β = 0.218, p < 0.01); it can be seen from panel (1) and panel (3) in the table that the amount of green innovation also plays a partial mediating role in the relationship between digital finance and total factor productivity of agriculture-related enterprises (β = 0.197, p < 0.01; β = 0.205, p < 0.05).

The above analysis shows that digital finance effectively empowers agribusinesses to link up with upstream and downstream symbiotic units in total factor linkage, enhances green innovation capacity, and then facilitates the high-quality development of agribusinesses.

5. Conclusions and Future Studies

This paper introduces the ecological symbiosis theory into the study of agribusiness economy, collates the data of Chinese A-share agribusinesses from 2013–2022 and the China Digital Financial Inclusion Development Index released by the Digital Finance Research Center of Peking University, and constructs a digital green symbiosis model of responsible agribusiness industry to analyze it. The empirical results provide a good answer to the research objective mentioned in the introduction, namely, can agribusiness achieve green innovation and high-quality development in the context of digital inclusive finance and social responsibility concept embedding?

The results show that digital financial development has a driving effect on the green innovation output of agribusiness firms at both the quality and quantity levels and that good social responsibility performance can enhance this influence. The analysis of enterprise heterogeneity finds that the impact of digital finance on green innovation of agribusinesses is more significant in processing and distribution agribusinesses, eastern regions, and growth stage enterprises, indicating that the balanced development of digital finance needs to be further promoted. The results of the value effect test show that digital finance realizes the total factor productivity progress of agribusiness and promotes the high-quality development of agriculture by promoting green innovation.

The findings of this paper have the following insights: First, investing in digital financial services can greatly benefit the development of agribusiness. In particular, processing and distribution enterprises, enterprises in the eastern region, and agribusinesses in the growth stage should make full use of the convenience that digital finance brings to their own innovative development and help transform the traditional agricultural economy. Secondly, although agribusiness is a for-profit economic organization, it should essentially be a link embedded in farmers and the market and is the backbone of promoting the development of the three rural areas and realizing rural revitalization. Therefore, agribusinesses should strengthen the awareness of social responsibility for mutual benefit and symbiosis, actively disclose information, establish and maintain good symbiotic network relationships, and build a co-evolutionary and synergistic agricultural system. Finally, green innovation is the development direction of national strategy and the key to enhancing core competitiveness and promoting low-carbon development in agriculture. Agribusinesses should adhere to green development, ensure that corporate technological innovation develops in the direction of greening, raise awareness of low-carbon production and consumption, and superimpose linked and interactive symbiotic relationships of social responsibility in order to realize a win–win situation for both the ecology and the economy.

There are also some limitations in this study: firstly, in terms of sample selection, the data selection of agribusinesses is somewhat one-sided. The constraints faced by unlisted agribusinesses may be more prominent due to the limitations of their business scale and management level. In the future, more samples of agricultural micro and small enterprises from different regions and market environments can be collected for refined research. In addition, the promotion of digital finance and innovation strategy is closely related to the characteristics of enterprise managers and the external environment in which the enterprises are located. Subsequent in-depth testing of its intermediary mechanism and possible boundary effects through the questionnaire survey method and the use of structural equation modeling on the basis of this paper is also a further direction. Finally, there is still much room for future research on symbiotic network systems in the agricultural industry. For example, further research can be conducted on how agribusinesses located in different ecological niches can utilize the social network embeddedness of management for effective interaction and dependence in symbiotic interfaces, as well as how to design and implement effective management strategies to optimize the symbiotic patterns and the quality of relationships of organizations within the interfaces.