Will Digital Financial Inclusion Increase Chinese Farmers’ Willingness to Adopt Agricultural Technology?

Abstract

Studies consider the impact of financial support on agricultural technology adoption, but do not consider the role of the rapidly evolving Digital Financial Inclusion (DFI). This study analyzes the impact of DFI on farmers’ willingness to adopt agricultural technology (WTAAT) using data from the China Labor-force Dynamics Survey and the Digital Financial Inclusion Index of Peking University, China. The results show that DFI significantly increases farmers’ WTAAT, consistent with the results of robustness tests. Moreover, the analysis of moderating effects shows that the contribution of DFI to WTAAT increases with the level of financial market development. Finally, WTAAT is affected by DFI development among farmers who receive government subsidies, participate in production technology training, and have no local non-agricultural economy. Therefore, we propose policy recommendations for developing DFI in rural areas, improving the financial market environment, and increasing subsidies and technical training. Our study provides some empirical evidence for exploring the field of agricultural technology adoption from the perspective of DFI and also provides new ideas for combining the digital transformation of finance with sustainable agricultural development, enriching the development of research in this field, which may also provide policy insights for the development of agricultural modernization in China and other countries.

1. Introduction

Agricultural technological progress is a decisive force in promoting agricultural development and modernization. Farmers are the main body of agricultural production and the adopters of technology. Only when these technologies are accepted by farmers and applied to agricultural production can such technology be genuinely transformed into productive forces, thus improving crop production efficiency, and promoting farmers’ income increase, production development, and agricultural modernization. China has long been committed to promoting agricultural technology, and an increasing number of farmers are realizing its importance in production. According to the data of the China Labor-force Dynamics Survey (CLDS) in 2016, 55.32% of the interviewed farmers were willing to adopt newly promoted agricultural technology. However, the implementation effect of the agricultural technology extension policy has not been as effective as expected [1]; the proportion of farmers willing to adopt new agricultural technologies is not high, and there is still much room for improvement compared with the goal of agricultural modernization. Farmers’ application of science and technology in agricultural production modernization in China is insufficient [2,3,4] and needs to be improved imperatively.

Farmers cannot adopt agricultural technology without financial support. Financial development affects farmers’ credit ability and is crucial for agricultural technology adoption [5,6]. However, existing research in China is limited to the role of traditional financial products and services provided by financial institutions [7,8], ignoring the impact of rapidly developing DFI on agricultural technology adoption. This study bridges this gap to analyze agricultural technology adoption from the perspective of DFI. With the continuous improvement in China’s digital infrastructure and the wide application of digital technology, mobile internet, and big data, digital financial inclusion (DFI) has developed rapidly. From 2011 to 2020, the median value of the DFI index in various provinces increased from 33.6 to 334.8, with an average annual growth rate of 29.1% [9], achieving a leap-forward development. However, regions and provinces have gaps in their development of DFI, showing a trend from high to low in the eastern, central, and western regions [9]. Nonetheless, with the rapid development of DFI in the central and western regions, the gap between these and the eastern regions is shrinking. In fact, China’s developed digital finance has played a very important role in China’s fight against the coronavirus disease 2019 (COVID-19) pandemic, including alleviating the impact of the epidemic on the pandemic [10]. The health code promotes the recovery of economic vitality in China while doing a good job in pandemic prevention. The big data technology behind it has contributed 0.5–0.75 percentage points to China’s GDP growth during the epidemic [11]. It is enough to see that DFI has become an indispensable factor in China’s development. With the popularity of DFI, rural residents can also enjoy the digital dividend of DFI. Online lending companies (including P2P, online small loans, etc.) provide financing services to farmers and rural small and micro enterprises [12]. Alibaba can provide financial services based on farmers’ online transaction data on Taobao as a basis for credit evaluation, so farmers can obtain unsecured and secured loans [13], and the MYbank established by Ant Group provides financial services to rural areas based on network data resources [14]. It can be seen that with the development and popularization of DFI in China, rural residents can also fully enjoy the benefits brought by DFI. Compared with traditional financial inclusion, DFI can simplify the approval process, lower the financing threshold, push beyond time and space constraints, broaden the scope of services, and improve the availability of financial services [15,16,17], thereby easing the financial constraints faced by farmers in terms of adopting agricultural technology. However, farmers’ education level is relatively low, their information technology and financial knowledge are lacking, and they are faced with the obstacle of the digital divide. Whether DFI can be used to promote the adoption of agricultural technology must be earnestly addressed in the current rural revitalization process.

We aimed to analyze the impact of the development of DFI on farmers’ willingness to adopt agricultural technology (WTAAT) using the CLDS of Sun Yat-sen University in 2016 and the data of the China DFI Index in 2015. We further intended to demonstrate that the development of DFI can significantly improve farmers’ WTAAT and that a high level of regional financial market development can significantly increase the positive impact of DFI on the WTAAT. Compared with previous studies that only considered the adoption of agricultural technology from the traditional financial perspective [7,8], this study renders the following marginal contributions. First, DFI can reflect financial inclusion, provide new opportunities for financial development with the help of digital technology, and provide low-cost and targeted services for adopting agricultural technology. Therefore, from the perspective of the combination of digital technology and finance, the analytical perspective of the adoption of agricultural technology can be expanded, and the research content in this field can be enriched. Second, we examined the moderating effect of regional financial market development level on the relation between DFI and WTAAT, which can deepen the understanding of the relationship between the two. Our findings on the WTAAT from the perspective of DFI can provide the micro-basis for government departments to improve relevant policies for agricultural technology promotion and some reference for DFI institutions to develop appropriate financial products and services.

Although existing studies have considered the role of finance in the adoption of agricultural technologies, the role of DFI in the adoption of agricultural technologies has not been explored. However, the importance of DFI in agricultural development has become increasingly prominent. This study can enrich the relationship between DFI and WTAAT, provide a micro-foundation for the importance of agricultural technology to production, and provide specific empirical evidence and theoretical basis for future analysis of agricultural modernization and sustainable development from the perspective of DFI.

2. Literature Review and Research Hypotheses

2.1. Literature Review

Regarding the factors that affect farmers’ adoption of agricultural technology, relevant research mainly analyzes three aspects: resource endowment, cognition of new technology, and fiscal and financial support. First, resource endowment is the fundamental factor affecting the adoption of agricultural technology, including physical and human capital. Physical capital includes household income and arable land area. The higher the household income, the more willing farmers are to use agricultural technologies to achieve large-scale and industrialized development [8]. The larger the scale of arable land, the more favorable it is for farmers to choose new technologies. Indeed, smaller arable land and fragmentation will inhibit the adoption of agricultural technologies [18]. Human capital covers education, health, technical training, professional production households, and working experience. The level of education will affect the behavior of farmers in adopting sustainable agricultural technologies [19]. Farmers with low levels of education are less able to accept new knowledge, have a strong risk aversion [20], and have a weak capacity for knowledge [21], which is not conducive to the adoption of agricultural technologies. Physical health can help farmers effectively overcome problems caused by insufficient physical labor and increase farmers’ adoption of agricultural technologies [22]. Technical training can promote the exchange of information and experience between farmers and improve their confidence in new technologies; through training, farmers can acquire new technical knowledge, understand the economic value of new technologies, eliminate the psychological rejection of new technologies, and quickly master and adopt new technologies [18,23]. Professional households in agricultural production have rich experience, strong information acquisition ability, a high degree of specialization, and a high technical level [20,22,24] and are more likely to accept new agricultural technologies. At present, the academic community has not reached a consensus on the impact of migrant workers on the adoption of agricultural technologies. Some scholars believe that migrant workers will change the focus of employment and reduce the emphasis on agriculture, which is not conducive to the adoption of agricultural technologies [25]. Other scholars believe that going out to work will improve farmers’ ability to adopt agricultural technologies [26].

Second, perception of new technology is an inherent factor in farmers’ decision to adopt it or not. The higher their awareness of new technologies, the greater the probability of adoption [27,28]. Farmers’ judgment on the role of technology directly affects their decision-making behavior in adopting technology, and their cognition of technical and economic benefits plays a significant role in promoting the adoption of farming technology [29]. The higher the farmers’ awareness of the ecological and environmental effects of conservation tillage technologies, the stronger their willingness to adopt such technology. For example, in water-saving irrigation technology, the higher their awareness of the technical effect, the higher the probability of farmers adopting it [30]. However, farmers’ perceptions of new technologies are not necessarily accurate. Chu et al. [20] believed that farmers’ misconceptions of the economic benefits and risks of environmentally friendly agricultural technologies could reduce the possibility of farmers adopting such technology. In addition to perceptions of the role of new technologies, risk perceptions are also crucial influencing factors. Risk perception of new technology will affect farmers’ collection and learning of new technology information, affecting their decision to adopt new technology [31]. Most farmers only attempt to adopt new technologies when they believe that there are few technical risks [29]. Farmers of different ages also have different risk perceptions of long-term technologies. The older they are, the more likely they are to avoid risks and treat new technologies more cautiously [22]. Among farmers with better economic conditions, the negative impact of technological risk perception on the adoption of new technologies is weak [18].

Lastly, fiscal and financial support is an important factor influencing the adoption of agricultural technologies. Government subsidies can directly alleviate the cost pressure of adopting new technologies and also encourage farmers to adopt new technologies [29]. New technologies coexist with high benefits, risks, and costs. Farmers have less self-owned funds, low credit levels, weak anti-risk capabilities, and low loan willingness [32]. Given their lack of sufficient capital, investment and financial support can solve the financial dilemma of farmers adopting new technologies. Providing farmers with risk-guaranteed credit funds and easing the dual constraints of funds and risks can encourage farmers to boldly adopt new technologies and improve agricultural productivity [7,8].

Financial support is crucial to alleviate the financial difficulties faced by farmers in adopting agricultural technology, and existing research provide valuable insights into these. Although some studies have paid attention to the role of financial support, they have analyzed the impact of traditional financial products and services provided by banks and insurance institutions on the adoption of agricultural technology and are lacking in the analysis of the rapidly developing DFI products and services role. Given that traditional finance is easily hindered by time and space factors, it is difficult for financial support to reach farmers, thus affecting farmers’ WTAAT. However, DFI can significantly reduce the time and space constraints, such that the role of financial support in lowering the credit threshold and easing the financing constraints can benefit farmers and improve their financial difficulties in adopting agricultural technologies. Therefore, we analyzed farmers’ WTAAT from the perspective of DFI.

2.2. Effects of DFI on WTAAT

DFI is the cross-border integration of financial inclusion and digital technology. Relying on digital technology to provide direct and convenient financial services and using information technology to innovate financial products, improve the availability of financial services, and meet the financial needs of low-income groups excluded from traditional financial services fully reflect the meaning of financial inclusion. Therefore, DFI has developed rapidly with the wide application of digital technology; specifically in rural areas, agricultural production is facing new opportunities and specific challenges with this development. Because of the rapid innovation and development of DFI, higher requirements are also placed on the financial ecological environment. However, the relatively lagging financial ecological environment in rural areas cannot sufficiently match the innovative products and services of DFI, which makes the financial market in rural areas have great operational risks and hidden dangers in risk supervision [17]. At the same time, farmers’ awareness of DFI is uneven, and innovative DFI products and services may be beyond their acceptance, magnifying hidden risks [17]. Nevertheless, DFI is a form of financial inclusion development. Its purpose is to create better financial inclusion and help establish financial risk diversification and compensation mechanisms [33], provide financial products and services with equal opportunities for vulnerable groups, and overcome the limitations of objective factors for low-income groups to access financial services, and improve financial accessibility [33,34].

Agricultural technology is the science and technology used in agricultural production and is the first driving force for its development. There are many types of agricultural technologies in China, and most existing studies focus on sustainability and resource conservation types of agricultural technologies [27,35,36]. The willingness to adopt agricultural technologies is influenced by the availability of funds. Its impact is mainly reflected in three aspects. First, DFI can lower farmers’ financing threshold and ease credit constraints. According to modern contract theory and credit rationing theory, information asymmetry may lead to problems of adverse selection, moral hazard, and high transaction costs, thus causing credit rationing [37]. The financial exclusion theory reveals that traditional financial services exclude the existence of low-income groups [38]. Therefore, traditional financial institutions are unable to provide sufficient financial products and services to farmers [39] and require them to provide collateral to obtain credit services. Farmers are either short of mortgage collateral or faced with the risk of bankruptcy after the loss of mortgage collateral, thereby leading to low loan willingness. Instead, farmers are compelled to choose traditional production methods [40]. With the help of network information technology, DFI enables the efficient analysis of the credit status of the borrower, assessment of the credit rating of the borrower, and smooth lending of funds, without the borrower providing collateral [41]. Thus, the borrowing threshold is lowered, the availability of financial services improved, and the financial constraints faced by farmers in using agricultural technologies eased. Second, DFI can break through the limitation of time and space and simplify the loan approval process. Traditional financial services require farmers to apply for loans from financial institutions in person, and financial institutions have fewer outlets and cumbersome approval procedures, which makes farmers stay away. Using DFI can effectively avoid rent-seeking costs by simplifying the procedures of farmers’ loans through digital approval business [17], making the cost of financial products and services more likely to be within the range that farmers can afford and thus encouraging farmers to borrow money and alleviating the financial pressure of investing in agricultural technology. Third, DFI can provide accurate financial products and services. With the help of digital technology, DFI collects soft information from farmers’ networks, integrates scattered information into standardized and usable financial data, and through information technology, classifies data, identifies farmers’ borrowing information when adopting agricultural technology, and realizes an accurate profile [42,43]. Thus, targeted financial products suitable for farmers can be developed and they can avail of reasonable and effective financial services.

This shows that DFI’s low cost, convenience, and availability provide new opportunities for farmers to obtain funds to adopt agricultural technologies. The peasant behavior theory of the rational small peasant school [44] indicates that farmers, as rational economic people, are driven by interests in their decision-making behavior, which is in line with their own optimization. Therefore, in the face of new agricultural technologies, to improve agricultural production efficiency, increase agricultural output and raise income levels, farmers will seize the opportunities provided by DFI and overcome the dilemma of insufficient self-owned funds to realize the adoption of agricultural technologies. Based on the above analysis, we proposed the following hypothesis.

Hypothesis 1 (H1).

The higher the level of development of DFI, the stronger the willingness of farmers to adopt agricultural technologies.

2.3. Moderating Effect of Financial Market Development Level

Although DFI has the advantages of low cost and broad coverage with the help of digital technology, thereby channeling more financial resources and services to rural areas [15] and increasing farmers’ WTAAT, the development of DFI cannot completely break away from the existing financial market foundation. In areas with a high level of financial market development, the following characteristics or advantages contribute to the role of DFI. First, the financial resources are abundant, and the knowledge and technology level of employees is high [45,46], providing talent support for the development of DFI [45], enabling DFI to serve agricultural production by taking advantage of digital technology, allowing farmers to enjoy DFI products and services smoothly, and promoting the adoption of agricultural technology. Second, rich experience and technology in risk control [46] provide a solid foundation for dealing with the financial risks in DFI. Thus, farmers are provided with risk-guaranteed credit funds to invest in agricultural technologies, improving their WTAAT. Third, farmers know more about financial products and services, have higher financial literacy, and are more likely to accept and use the products and services of DFI [15], which can help enhance farmers’ WTAAT in production. In contrast, in areas with a low level of financial market development, financial resources are relatively limited, highly skilled professionals are scarce, the risk management level is low, and the credit scale is small. The financial resources provided by DFI cannot be effectively allocated [46], and it cannot perform with the low financial literacy of farmers. As such, its impact on the WTAAT would be relatively limited. Therefore, we hypothesized as follows.

Hypothesis 2 (H2).

The level of financial market development significantly moderates the relationship between DFI and the WTAAT. Specifically, the level of financial market development enhances the impact of DFI on the WTAAT.

3. Empirical Design and Descriptive Statistics

3.1. Data Sources and Processing

The data used in this study were mainly drawn from the China Labor-force Dynamics Survey (CLDS), a nationally representative household questionnaire survey conducted biennially by the Centre for Social Survey at Sun Yat-sen University. The CLDS is open and licensed. This survey adopts a multi-stage stratified probability proportionate to size sampling (PPS) method to interview households across 29 provinces in China (excluding Hong Kong, Macau, Taiwan, Tibet, and Hainan). CLDS has detailed information on demographic characteristics, socioeconomics, housing conditions, and community context. In addition, we applied the DFI index of China, constructed by the Peking University Institute of Digital Finance and the fintech company Alibaba Ant Financial Service. The DFI captures the three dimensions of coverage breadth, usage depth, and digitization level. As the explained variable of this study was selected from the questionnaire of CLDS2016, and considering that the CLDS2016 questionnaire surveyed the sample in 2015, we selected the total index of China’s DFI at the prefecture (or city) level in 2015 to match the questionnaire of CLDS2016. Specifically, we processed the data as follows. First, because our explained variable was farmers’ WTAAT, we limited the sample to rural areas. Second, given our focus on farmers, we included only interviewees with agricultural registered permanent residence. Third, according to our aim of analyzing the adoption of agricultural technology, we screened the respondents’ working place, industry type, and unit type, and included only those working in their own village, other villages in their own town, and other towns in counties or districts (excluding counties or districts) and engaged in agriculture, forestry, animal husbandry, and fishery production. The final effective sample size was 3890.

3.2. Variable Description

3.2.1. Explained Variable

WTAAT. Studies mainly set up dummy variables to measure the adoption of agricultural technology [7,47]. We used the CLDS2016 individual questionnaire question “Compared with other villagers, I will actively adopt newly promoted agricultural technology” to measure farmers’ WTAAT. The item was rated on a five-point scale, where 1 = strongly disagree and 5 = strongly agree. Higher scores indicated stronger willingness. The average value of this variable was 3.598 (see

Table 1. Definition of variables and descriptive statistical analysis results.

Variable Types	Symbol	Variable Name	Implication	Mean Value	Standard Deviation
Explained variable	WTAAT	Willingness to adopt agricultural technology	Strongly disagree = 1, Do not agree = 2, Neither agree nor disagree = 3, Quite agree = 4, Strongly agree = 5	3.598	1.008
Core explanatory variable	DFI	Digital financial inclusion	Digital financial inclusion index	166.249	16.505
Moderator	Fi-Mk	Financial market development level	Proportion of farmers’ loan balance to GDP (%)	0.130	0.086
Control variable	Gen	Gender	Male = 1, female = 0	0.488	0.500
	Age	Age	Respondent’s age (years)	52.977	11.539
	Age2	Age squared term	Respondent’s age squared term	2939.653	1199.979
	Edu	Education level	No schooling = 0, Primary school = 6, Junior high school = 9, High school = 12, Junior college = 15, undergraduate = 16, Graduate = 19	6.139	3.579
	Health	Health status	Very unhealthy = 1, Moderately unhealthy = 2, General = 3, Healthy = 4, Very healthy = 5	3.393	1.015
	Work-Exp	Working experience	Yes = 1, No = 0	0.144	0.351
	Per-Land	Per capita cultivated land area	Total area of household arable land/total household population	2.227	3.373
	Ag-Pro	Specialized agricultural production household	Yes = 1, No = 0	0.120	0.325
	Gov-Subsidy	Agricultural production and operation are subsidized by the government	Yes = 1, No = 0	0.600	0.490
	Ag-Trn	Whether farmers have received agricultural production technology training	Yes = 1, No = 0	0.742	0.437
	Tot-Income	Total household income	Logarithm of total household income	9.687	1.608
	Non-Ag-Econ	Non-agricultural economy of the village	Yes = 1, No = 0	0.150	0.357

). On average, farmers in the sample had a relatively high WTAAT.

3.2.2. Core Explanatory Variable

DFI. Research generally selects the DFI index published by the Digital Finance Research Center of Peking University to measure DFI [48,49,50]. We also used the DFI index of each city to measure the development level of DFI in Chinese cities. The average value of the index was 166.249 (see Table 1), indicating that the average development level of DFI in the sample was relatively low, leaving much room for improvement.

3.2.3. Moderator

Financial market development level (fin-mk). Studies have used the ratio of the loan balance of financial institutions to GDP to represent the level of financial development [51]. Given our focus on the financial development level of the rural market, we adopted the proportion of farmers’ loan balance to GDP in 2015, compiled from the Almanac of China’s Finance and Banking in 2016, to represent the development level of the local rural financial market. The value ranged from 0.036 to 0.336, with an average value of 0.130 (see Table 1).

3.2.4. Other Variables

Drawing on the research of Zou et al. [19,22,29], combined with the CLDS questionnaire, we included the following control variables: individual-, family-, and village-level characteristics of the respondents. Individual characteristics included gender (gen), age (age), age square (age²), education level (edu), health status (health), and working experience (work-exp). Family-level features included per capita cultivated land area (per-land), whether the family is agricultural production professional (ag-pro), whether the agricultural production and operation are subsidized by the government (gov-subsidy), whether farmers have received agricultural production technology training (ag-trn), and total household income (tot-income). The village-level features included the non-agricultural economy of the village (non-ag-econ). Descriptive statistics of control variables are shown in Table 1.

3.3. Model Specifications

3.3.1. Baseline Model

The explained variable “WTAAT” in our study is an ordinal variable. When the model is set correctly, there is no significant difference between the results of the ordinary least squares (OLS) estimation method and those of the standard ordered Probit model or standard ordered Logit model [52]. Thus, we adopted the OLS model as the baseline model:

$$WTAA{T}_i={\alpha }_0+{\alpha }_{1}DF{I}_i+{\alpha }_{2}Controls+C_i+{\mu }_i$$

(1)

In Equation (1), WTAAT_i indicates the willingness to adopt agricultural technology; DFI_i represents digital financial inclusion; Controls is a set of control variables, including individual, family, and village characteristics; C_i represents city fixed effect, and ${\mu }_i$ represents the random error term. If the H1 holds, then the coefficient ${\alpha }_1$ of the variable DFI_i should be significantly greater than zero.

3.3.2. Moderating Effect Model

To explore whether the relation between DFI and the WTAAT will be affected by the level of financial market development, that is, to verify H2, we established the following moderating effect model:

$$\begin{gathered} WTAA{T}_i={\gamma }_0+{\gamma }_{1}DF{I}_i+{\gamma }_{2}fin-m{k}_i+{\gamma }_{3}DF{I}_i\times fin-m{k}_i \\ +{\gamma }_{4}Controls+C_i+{ϵ}_{i } \end{gathered}$$

(2)

In Equation (2), fin-mk_i is the moderator, indicating the level of financial market development; DFI_i × fin-mk_i is the interaction term, that is, the multiplication term of the DFI index and level of financial market development;${ϵ}_{i }$ represents the random error term, and the rest of the variables are the same as those in the baseline model. If H2 holds, then the coefficient ${\gamma }_3$ of the variable DFI_i × fin-mk_i should be significant and greater than zero. As the development level of the financial market, that is, the proportion of farmers’ loan balances to GDP, is a continuous variable, according to the approach of Wen et al. [53], we centralized and then regressed the explanatory variable DFI and the moderator variable financial market development level.

3.3.3. Endogenous Processing Model

Due to possible endogeneity problem in the baseline model, we established the following endogenous processing model:

$$I{V}_i={\beta }_0+{\beta }_{1}DF{I}_i+{\beta }_{2}Controls+C_i+{\epsilon }_i$$

(3)

$$WTAA{T}_i={\delta }_0+{\delta }_1\widehat{I{V}_i}+{\delta }_{2}Controls+C_i+{\theta }_i$$

(4)

In Equation (3), IV_i is the instrumental variable (IV), indicating the number of Internet users per capita at the prefecture-level city level; ${\epsilon }_{i }$ represents the random error term, and the rest of the variables are the same as those in the baseline model. Through the OLS method, the fitted value $\widehat{I{V}_i}$ of IV_i is obtained. In this equation, IV_i is related to ${\epsilon }_{i }$, and $\widehat{I{V}_i}$ is not related to ${\epsilon }_{i }$. Equation (3) is the first-stage result of the two-stage estimation (2SLS) of the IV, which aims to test the correlation between the IV and DFI. If the correlation is satisfied, then the coefficient ${\beta }_1$ of the variable DFI_i should be significant and greater than zero.

In Equation (4), $\widehat{I{V}_i}$ is the fitted value of IV_i; ${\theta }_{i }$ represents the random error term, and the rest of the variables are the same as those in the baseline model. Equation (4) is the second-stage result of the two-stage estimation (2SLS) of the IV, which aims to demonstrate the reliability of the results of the estimation results.

4. Results

4.1. Results of the Baseline Model

In order to solve the possible heteroscedasticity problem, our standard error adopts the heteroscedasticity robust standard error, thus ensuring the accuracy and stability of the baseline model. The estimated results of the baseline model are shown in

Table 2. Baseline model and moderating effect estimation results.

Column	1	2	3	4	5
Variables	WTAAT	WTAAT	WTAAT	WTAAT	WTAAT
DFI	0.003 ***	0.035 ***	0.037 ***	0.037 ***	0.105 ***
	(0.001)	(0.009)	(0.010)	(0.010)	(0.033)
Fin-Mk					0.429 ***
					(0.153)
DFI * Fin-Mk					0.014 ***
					(0.005)
Gen		0.133 ***	0.137 ***	0.136 ***	0.136 ***
		(0.035)	(0.035)	(0.035)	(0.035)
Age		0.024 ***	0.024 ***	0.025 ***	0.025 ***
		(0.009)	(0.009)	(0.009)	(0.009)
Age2		−0.0002 ***	−0.0002 ***	−0.0002 ***	−0.0002 ***
		(0.0001)	(0.0001)	(0.0001)	(0.0001)
Edu		0.030 ***	0.028 ***	0.029 ***	0.029 ***
		(0.005)	(0.005)	(0.005)	(0.005)
Health		0.035 *	0.032 *	0.035 **	0.035 **
		(0.018)	(0.018)	(0.018)	(0.018)
Work-Exp		0.087 *	0.082 *	0.084 *	0.084 *
		(0.046)	(0.046)	(0.046)	(0.046)
Per-Land			0.002	0.001	0.001
			(0.005)	(0.005)	(0.005)
Ag-Pro			0.199 ***	0.196 ***	0.196 ***
			(0.055)	(0.055)	(0.055)
Gov-Subsidy			0.085 **	0.085 **	0.085 **
			(0.037)	(0.037)	(0.037)
Ag-Trn			−0.056	−0.022	−0.022
			(0.067)	(0.068)	(0.068)
Tot-Income			0.005	0.007	0.007
			(0.001)	(0.011)	(0.011)
Non-Ag-Econ				−0.289 ***	−0.289 ***
				(0.071)	(0.071)
Constant	3.124 ***	−3.044 *	−3.482 **	−3.573 **	3.857 ***
	(0.168)	(1.598)	(1.683)	(1.684)	(0.514)
City fixed effects	no	yes	Yes	yes	yes
Observations	3890	3890	3890	3890	3890
Adj. R2	0.002	0.103	0.107	0.111	0.111

Notes: Heteroscedasticity robust standard errors are shown in parentheses; ***, **, and * represent p < 0.01, p < 0.05, and p < 0.1, respectively. Since there is no unit treatment for the square term of age in the literature, our estimation result of this variable retains four decimal places; the same below.

. Column 1 shows a univariate relation that only considers DFI and the WTAAT; the results are significantly positive at the 1% level. Column 2, adding the control variable at the individual level and the city fixed effect controls, shows significantly positive results at the 1% level. Column 3 adds household-level control variables on the basis of column 2 and the city fixed effect controls. The results showed that the coefficient of DFI was still significantly positive at the 1% level. Column 4, adding all control variables, shows that the coefficient of the core explanatory variable (DFI) is 0.037, which is still significant at the 1% level. In terms of economic significance, the estimated coefficient of DFI suggests that a one standard deviation increase in DFI leads to a 60.58% (=(0.037 × 16.505)/1.008) increase in the WTAAT. Thus, DFI can help increase the WTAAT, supporting H1.

Among the control variables, gender had a significant positive impact on WTAAT: men were more willing than women. The coefficients of age and the square of the age were both significant at the 1% level. The coefficient of age was positive, the coefficient of the square of the age was negative, and the absolute value of the age coefficient was greater than that of the coefficient of the square of the age, indicating that the influence of farmers’ age on the WTAAT presented an inverted-U curve relation: when farmers are younger, their WTAAT will increase with age. However, after reaching a certain age, with the further increase in age, farmers’ WTAAT decreases.

The effect of education on the WTAAT was also significantly positive. The higher the education level of farmers, the stronger their ability to accept new things, and the more willing they were to adopt agricultural technology. Health status likewise had a significant impact on farmers’ adoption of agricultural technology. The higher the health level of farmers, the more they could ensure sufficient physical labor support in the process of agricultural technology investment, and the more willing they were to adopt agricultural technology.

The experience of going outside to work also had a significant positive impact on the WTAAT, consistent with the conclusion of Zhong et al. [54] but different from the conclusion of Zou et al. [19]. The possible reason for the discrepancy is that the sample in our work included people engaged in agricultural production. Farmers pay more attention to agriculture, and they are more willing to adopt agricultural technology after their experience of going out to work broadens their horizons and expands their capabilities and funds. Meanwhile, professional production households showed a significant positive impact on the WTAAT. These households have rich production experience and pay more attention to production input, resulting in stronger WTAAT. Government subsidies also had a significant positive impact on the WTAAT. They can reduce the cost of adopting technologies, making farmers more inclined to adopt agricultural technologies.

The per capita arable land area had no significant effect on the WTAAT, which is consistent with the conclusion of Khanna [55] but inconsistent with those of Tang et al. [18]. The reason may be the difference in technology and topography, which leads to the difference in the influence intensity of cultivated land area. Similarly, the coefficient of the variable of production technology training was not significant, in contrast to the conclusion of Mao et al. [23]. The possible reason is that most farmers have a lower education level and require a prolonged duration to digest and absorb the content of technical training. Consequently, it takes a certain amount of time for farmers to decide to adopt agricultural technology after receiving production technology training. In other words, production technology training is a slow variable, and it is difficult to be identified in a short period of time. Likewise, the total household income had no significant effect on the WTAAT, which is inconsistent with the conclusion of Li et al. [29]. Our conclusion is reasonable to a certain extent because the increase in non-agricultural income does not always promote the adoption of agricultural technology. Factors such as the degree of agricultural specialization of farmers [23], the degree of emphasis farmers place on agricultural production [19], and the type of agricultural technology [56] may hinder the adoption of agricultural technologies [19], impacting total household income on agricultural technology adoption. Meanwhile, the presence of a non-agricultural economy in the village had a significant negative impact on the WTAAT, which may be because a non-agricultural economy represents the openness and economic development level of the village to a certain extent. If there are secondary and tertiary industries in rural areas, the probability of farmers engaging in non-agricultural production will be greatly enhanced, and thus, the opportunity cost of engaging in agricultural production will be greater, resulting in a decrease in their WTAAT.

4.2. Endogenous Problem and Robustness Test

4.2.1. Endogenous Problems

Limited by the questionnaire items and matching data, issues were expected for omitted variables and measurement errors, which can lead to the endogeneity problem in the baseline model. Moreover, because the higher the willingness of farmers to adopt agricultural technology, the more willing they may be to accept DFI, a new financial tool, there may be a reverse causality between DFI and the WTAAT. However, the questionnaire does not include information on individuals’ use of DFI, which is city-level rather than individual-level data, which helps mitigate the impact of reverse causality. This study used the IV method to solve any endogeneity problems that the above problems may cause. To address the endogeneity problem, we referred to the research of Xie and Zhu [50] and adopted the number of Internet users per capita at the prefecture-level city level as an IV of DFI. The data comes from the “China City Statistical Yearbook 2016”. We selected this IV for two main reasons: first, the number of Internet users per capita and DFI are closely related, because DFI is a new financial model based on the internet, and its rapid development benefits from the extensive penetration of the Internet. The close relation between the two satisfies the correlation conditions of IV. Second, there is a negligible relationship between farmers’ WTAAT and the number of Internet users per capita at the regional level, thus satisfying the exogenous condition of IV.

Columns 1 and 2 of

Table 3. Two-stage estimation results of instrumental variables.

Column	1	2
Reg. Stages	First Stage	Second Stage
Variables	DFI	WTAAT
Predicted Value of the DFI		0.005 **
		(0.002)
IV	129.227 ***
	(4.038)
Gen	−0.320	0.124 ***
	(0.302)	(0.035)
Age	0.049	0.032 ***
	(0.076)	(0.009)
Age2	0.0002	−0.0003 ***
	(0.001)	(0.0001)
Edu	0.055	0.034 ***
	(0.043)	(0.005)
Health	−0.093	0.037 **
	(0.134)	(0.017)
Work-Exp	−0.152	0.076 *
	(0.400)	(0.045)
Per-Land	−0.031	0.011 **
	(0.044)	(0.004)
Ag-Pro	−1.544 **	0.237 ***
	(0.466)	(0.049)
Gov-Subsidy	1.010 ***	0.090 ***
	(0.295)	(0.033)
Ag-Trn	−0.045	0.042
	(0.335)	(0.040)
Tot-Income	0.105	0.014
	(0.109)	(0.010)
Non-Ag-Econ	−1.591 ***	−0.099 **
	(0.489)	(0.046)
Constant	151.528 ***	1.243 ***
	(2.494)	(0.380)
Regional fixed effects	yes	yes
Cragg–Wald F statistic		5970.072
10% max IV size		(16.38)
Observations	3890	3890
Adj. R2	0.727	0.043

Notes: Heteroscedasticity robust standard errors are shown in parentheses; ***, **, and * represent p < 0.01, p < 0.05, and p < 0.1, respectively.

give the two-stage estimation results and test results of the rationality of the IV, respectively. First, we tested the correlation between the IV and DFI (i.e., the results of the first stage in the two-stage least square estimation). As shown in column 1 in Table 3, the coefficient of the IV is significantly positive at the 1% level, satisfying the correlation requirements for instrumental variables. Second, the value of the Cragg–Wald F statistic is much larger than the critical value at the 10% level of the Stock–Yogo weak identification test in parentheses, indicating the absence of a weak IV problem. Therefore, this IV had good properties. The second-stage regression results in the IV regression were consistent with those results of the baseline model, indicating the reliability of the baseline estimation results.

4.2.2. Robustness Test

To examine the robustness of the model estimation results, we conducted tests from three perspectives: replacing core explanatory variables, replacing models, and removing extreme values.

The first is the method of substituting the core explanatory variable. Drawing on the method proposed by Peng and Xu [16], we replaced the core explanatory variable in the model with two sub-indices, namely, digital financial coverage and digital financial depth in the China DFI Index in 2015. We did not replace the core explanatory variable with the sub-indicator of digital degree because digital financial coverage and depth can reflect farmers’ acceptance and use of DFI, whereas digital degree does not directly relate to the financial services farmers receive. The estimated results after replacing the core explanatory variable are shown in columns 1 and 2 of

Table 4. Robustness test.

Column	1	2	3	4	5
Variables	Substitute Core Explanatory Variables		Replace the Model		Eliminate Extreme Values
Digital Financial Coverage	0.015 ***
	(0.004)
Digital Financial Depth		0.067 ***
		(0.018)
DFI			0.040 ***	0.018 ***	0.028 **
			(0.011)	(0.005)	(0.011)
Gen	0.136 ***	0.136 ***	0.163 ***	0.067 ***	0.133 ***
	(0.035)	(0.035)	(0.039)	(0.016)	(0.036)
Age	0.025 ***	0.025 ***	0.029 ***	0.012 ***	0.021 *
	(0.009)	(0.009)	(0.010)	(0.004)	(0.012)
Age2	−0.0002 ***	−0.0002 ***	−0.0003 ***	−0.0001 ***	−0.0002 *
	(0.0001)	(0.0001)	(0.0001)	(0.00004)	(0.0001)
Edu	0.029 ***	0.029 ***	0.033 ***	0.015 ***	0.030 ***
	(0.005)	(0.005)	(0.006)	(0.002)	(0.005)
Health	0.035 **	0.035 **	0.041 **	0.010	0.029
	(0.018)	(0.018)	(0.020)	(0.008)	(0.018)
Work-Exp	0.084 *	0.084 *	0.093 *	0.025	0.053
	(0.046)	(0.046)	(0.053)	(0.023)	(0.048)
Per-Land	0.001	0.001	0.001	0.002	−0.006
	(0.005)	(0.005)	(0.006)	(0.003)	(0.009)
Ag-Pro	0.194 ***	0.196 ***	0.245 ***	0.053 **	0.213 ***
	(0.055)	(0.055)	(0.064)	(0.026)	(0.057)
Gov-Subsidy	0.085 **	0.085 **	0.095 **	0.051 ***	0.092 **
	(0.037)	(0.037)	(0.042)	(0.017)	(0.039)
Ag-Trn	−0.022	−0.022	−0.026	−0.014	−0.039
	(0.068)	(0.068)	(0.076)	(0.030)	(0.069)
Tot-Income	0.007	0.007	0.008	0.010**	0.010
	(0.011)	(0.011)	(0.012)	(0.005)	(0.011)
Non-Ag-Econ	−0.289 ***	−0.289 ***	−0.354 ***	−0.087 **	−0.282 ***
	(0.071)	(0.071)	(0.079)	(0.034)	(0.073)
Constant	0.191	−6.489 ***			−1.851
	(0.703)	(2.464)			(1.952)
City fixed effects	yes	yes	yes	yes	yes
Observations	3890	3890	3890	3890	3657
Adj. R2/Pseudo-R2	0.111	0.111	0.056	0.132	0.105

Notes: Heteroscedasticity robust standard errors are shown in parentheses; ***, **, and * represent p < 0.01, p < 0.05, and p < 0.1, respectively. In the standard ordered Logit estimation results in column 3, the values of cut points 1, 2, 3, and 4 are 5.740 *** (1.770), 6.560 *** (1.772), 7.561 *** (1.774), and 8.713 *** (1.775), respectively.

. Both the coverage and depth of digital finance could significantly improve farmers’ WTAAT, indicating the robustness of our results.

The second is the method to replace the model. We take two approaches. The first is the Ordered Probit model. The sample’s responses to the WTAAT are divided into five levels (from strongly disagree to strongly agree). Thus, we estimated this ordinal variable using the standard Ordered Pobit model (column 3 of Table 4). The results showed that the coefficient of DFI was significantly positive, consistent with the results using the OLS method. The second is the dichotomous Probit model. We divide the WTAAT into 1 (strongly agree/quite agree) and 0 (neither agree nor disagree/do not agree/strongly disagree), and using the Probit model for robustness check (column 4 of Table 4), the results of the marginal effect showed that the direction and significance of the digital financial inclusion coefficient did not change, confirming the robustness of the benchmark results. Simultaneously, we also divide the WTAAT into 1 (strongly agree/quite agree/neither agree nor disagree) and 0 (do not agree/strongly disagree), and using the Probit model for robustness check, and the result was still relatively stable.

The third is to use the method of eliminating extreme values. To reduce the influence of extreme values, we removed data below the 1% quantile and above the 99% quantile for the continuous variables in the core explanatory and control variables. The regression results are shown in column 5 of Table 4. The coefficient direction and significance of DFI did not change, indicating that DFI could significantly improve the WTAAT.

4.3. Moderating Effect Results

Column 5 of Table 2 presents the analysis results of the moderating effect of the level of financial market development on the relation between DFI and the WTAAT. The coefficient of the interaction term between DFI and the financial market development level was significantly positive at the 1% level, indicating that the development level of the local financial market has a significant moderating effect between DFI and the WTAAT (Figure 1). In other words, with the improvement of the development level of the regional financial market, the positive impact of DFI on the WTAAT would be strengthened. The development of the financial market can provide knowledge, technology, and talent support for DFI, offer experience and reference for controlling financial risks, and promote DFI to serve farmers in adopting agricultural technologies. Thus, H2 was verified.

5. Heterogeneity Analysis

5.1. Heterogeneity Analysis Based on Government Subsidies

Farmers, as rational economic actors, pursue the maximization of their own interests. Before deciding to adopt technology, they primarily consider input cost as a factor [57]. If the cost of new agricultural technology is greater than that of the original technology, then it will hinder farmers’ adoption behavior. On the one hand, government subsidies can alleviate the financial pressure caused by high costs, reduce the replacement cost of new and old technologies, and compensate for the expected benefits and additional costs. On the other hand, it can incentivize people to obtain massive outputs with minimal cost [58], which produces more than the expected effects. DFI can provide a low-cost, convenient, and fast supply of funds, and farmers who receive government subsidies are more motivated to use them to meet the financial needs of adopting agricultural technologies. Therefore, the impact of DFI on the farmers’ WTAAT may vary depending on whether farmers receive government subsidies. To verify whether this difference exists, we divided the sample into two groups for estimation according to receipt of government subsidies. The results are shown in columns 1 and 2 of

Table 5. Heterogeneity analysis.

Column	1	2	3	4	5	6
Type	No Gov-Subsidy	Gov-Subsidy	No Ag-Trn	Ag-Trn	No Non-Ag-Econ	Non-Ag-Econ
Variables	WTAAT	WTAAT	WTAAT	WTAAT	WTAAT	WTAAT
DFI	−0.020	0.049 ***	0.002	0.040 ***	0.037 ***	−0.016
	(0.0195)	(0.010)	(0.007)	(0.011)	(0.010)	(0.012)
Gen	0.111 **	0.163 ***	0.158 **	0.128 ***	0.128 ***	0.221 **
	(0.0551)	(0.045)	(0.072)	(0.040)	(0.038)	(0.088)
Age	0.007	0.042 ***	0.042 **	0.020 **	0.029 ***	0.009
	(0.0137)	(0.012)	(0.020)	(0.010)	(0.009)	(0.027)
Age2	−0.0001	−0.0004 ***	−0.0004 **	−0.0002 **	−0.0003 ***	−0.0001
	(0.0001)	(0.0001)	(0.0002)	(0.0001)	(0.0001)	(0.0003)
Edu	0.026 ***	0.028 ***	0.034 ***	0.030 ***	0.029 ***	0.030 **
	(0.0088)	(0.007)	(0.011)	(0.006)	(0.006)	(0.014)
Health	0.017	0.046 *	0.050	0.028	0.039 **	0.059
	(0.0274)	(0.024)	(0.034)	(0.021)	(0.020)	(0.046)
Work-Exp	0.193 ***	−0.022	0.072	0.078	0.074	0.055
	(0.0705)	(0.064)	(0.090)	(0.054)	(0.049)	(0.133)
Per-Land	0.003	−0.004	−0.002	0.003	0.002	0.028
	(0.0061)	(0.007)	(0.009)	(0.006)	(0.005)	(0.048)
Ag-Pro	0.257 ***	0.162 **	0.136	0.261 ***	0.233 ***	0.192
	(0.082)	(0.078)	(0.118)	(0.062)	(0.062)	(0.117)
Gov-Subsidy			0.136 *	0.070 *	0.067	0.104
			(0.078)	(0.043)	(0.041)	(0.088)
Ag-Trn	−0.065	−0.040			−0.029	0.603
	(0.145)	(0.080)			(0.073)	(0.453)
Tot-Income	−0.008	0.020	−0.019	0.015	0.013	−0.014
	(0.015)	(0.0178)	(0.021)	(0.013)	(0.012)	(0.027)
Non-Ag-Econ	−0.283 **	−0.285 ***	0.042	−0.335 ***
	(0.111)	(0.102)	(0.344)	(0.080)
Constant	6.608 **	−6.007 ***	2.047 *	−3.927 **	−3.738 **	4.720 **
	(3.143)	(1.622)	(1.203)	(1.788)	(1.693)	(2.104)
City fixed effects	yes	yes	yes	yes	yes	yes
Observations	1554	2336	1002	2888	3308	582
Adj. R2	0.150	0.109	0.107	0.135	0.119	0.084

Notes: Heteroscedasticity robust standard errors are shown in parentheses; ***, **, and * represent p < 0.01, p < 0.05, and p < 0.1, respectively.

. For farmers who do not receive government subsidies, the impact of DFI on the WTAAT is not significant. For those who do, the impact is significantly positive at the 1% level, indicating that with government subsidies, DFI can improve WTAAT.

5.2. Heterogeneity Analysis Based on Production Technology Training

For farmers to adopt agricultural technology, they must understand and accept agricultural technology, along with its expected benefits and risks. On the one hand, agricultural technology training can impart technical knowledge and experience exchange, improve the efficiency of technology promotion, allow farmers to understand and master the characteristics and operational essentials of agricultural technology, and have an intuitive feeling of the expected economic value brought by the use of agricultural technology [20,22]. It can also increase farmers’ confidence in adopting technology [18], making them more likely to take advantage of the financing advantages brought by DFI and adopt agricultural production technology. Farmers who have not received production technology training lack the knowledge and skills to use new agricultural technologies. They may also be unable to estimate the risks and expected returns brought by the use of agricultural technologies, and thus, even in the face of low-cost capital and finance services brought by DFI, they may be reluctant to try new agricultural technologies. Therefore, the impact of DFI on the WTAAT may differ depending on whether farmers receive training in production technologies. To verify this difference, we divided the samples into two groups according to receipt of production technology training for regression estimation. The results are shown in columns 3 and 4 of Table 5. For farmers who have not received production technology training, the impact of DFI on the WTAAT was not significant. For farmers who have, the impact was significantly positive at the 1% level. Therefore, among farmers who have received production technology training, DFI can improve their WTAAT.

5.3. Heterogeneity Analysis Based on Non-Agricultural Economy

Areas with better non-agricultural economic development have a higher level of economic level and openness, and relatively developed industry and commerce [7]. These can increase the possibility of farmers engaging in non-agricultural jobs with higher wages, resulting in the larger opportunity cost of agricultural production [20]. Other consequences are the shift in focus of farmers’ employment [19,59], reduction of their emphasis on agriculture, and higher willingness to invest the funds they find in non-agricultural jobs with higher returns, accompanied by reduced WTAAT. In contrast, farmers in areas with poor non-agricultural economic development would be motivated to adopt agricultural technology because the proportion of agricultural income to farmers’ income is larger, and the emphasis on agriculture is higher [23,24]. Farmers are willing to invest more energy and capital into agricultural production and adopt technology to obtain higher efficiency returns. Therefore, the impact of DFI on the WTAAT may vary according to local non-agricultural economic conditions. To verify, we divided the samples into two groups for regression according to non-agricultural economic status. The results are shown in columns 5 and 6 of Table 5. In the sample with a non-agricultural economy, the impact of DFI on the WTAAT was not significant; in the sample without a non-agricultural economy, the impact was significantly positive at the 1% level. Therefore, in the sample without a non-agricultural economy, DFI could significantly improve farmers’ WTAAT. When the village lacks secondary and tertiary industries, the main economic income of the entire village comes from agriculture, and the emphasis on agriculture is relatively high. Its ideological awareness of developing agricultural production would also be relatively high. On the contrary, farmers who do not pay much attention to agricultural production have a bias against the importance of agricultural production, and the degree of patience in developing agricultural production is not enough [60], which is not conducive to agricultural technology adoption.

6. Discussion, Conclusions, and Suggestions

6.1. Discussion

DFI can significantly increase income, alleviate poverty and promote entrepreneurship [15,61,62]. However, the role of DFI in the adoption of agricultural technologies is underexplored. Agriculture is related to China’s food security, resource security, and ecological security, and its importance is self-evident. However, more people and less water are China’s primary national conditions [63]. Therefore, promoting the sustainable development of agriculture is the inherent requirement of realizing the new road of agricultural modernization in China. To realize the sustainable development of agriculture, we must pay attention to the input of agricultural technology to realize the virtuous circle of resources, environment, and agricultural product supply. The continuous innovation of modern agricultural new technologies can ensure the improvement of the level of resource utilization and the adequate supply of agricultural products, and effectively provide strong technical support for the sustainable development of agriculture [63]. Based on the nationally representative CLDS questionnaire, we found that DFI can significantly improve farmers’ WTAAT, this conclusion remained robust even after considering endogeneity, replacing core explanatory variables and models, and removing extreme values. The result is in line with our expectations, indicating that DFI can indeed promote agricultural technology adoption and play a role in the development of agricultural modernization. This provides new empirical evidence for exploring the field of agricultural technology adoption from a DFI perspective. Although existing studies in China have paid attention to the ability of financial support to promote the adoption of agricultural technologies, their scope of financial support is limited to traditional financial products and services [7,8], ignoring the role of the rapidly developing DFI, which may make their analysis incomplete. Therefore, our findings enrich and complement existing research.

Studies have confirmed that credit availability and financial support have a significant impact on agricultural technology adoption in other developing countries [6,64,65,66], and they also recognize the convenience of financial services that come with digital technologies [67,68]. But there is no empirical analysis of the impact of financial instruments of digital transformation on agricultural technology adoption and production. With the development of information technology in China, internet finance continues to expand the accessibility of financial inclusion. Therefore, the DFI index is constructed to scientifically describe the development status of DFI in China [9]. China DFI is a combination of digital technology and financial inclusion [16,33,61], our findings confirm that digital transformation of financial inclusion is well placed to facilitate agricultural technology adoption, agricultural modernization, and sustainable development. Therefore, our study provides experience for other developing countries to develop agriculture from three aspects. First, they can use digital technology to innovate traditional financial products and achieve digital transformation and innovation in finance. Second, governments and financial institutions should popularize digital financial products and services to farmers, and encourage them to use digital financial tools for agricultural production and adoption of agricultural technologies. Third, it also further provides new ideas for the combination of digital transformation of financial products with agricultural modernization and sustainable development.

The analysis result of moderating effect comprehensively considers the important role played by the development level of the financial market in China in the process of developing DFI. This realistic background is also consistent with the second research hypothesis of this study, that is, the promotion effect of DFI on the WTAAT increased with the improvement in the development level of the financial market. This result reflects the multi-faceted role of traditional financial markets in supporting the development of DFI [15,45,46], and also makes our research more realistic.

The results of the heterogeneity analysis showed that farmers who received government subsidies and participated in production technology training were more willing to use DFI to improve their WTAAT. Such results fully demonstrate the important role of government support policies in agricultural technology adoption in China [7,22,29,58], which is consistent with China’s realistic background [69]. In areas without a non-agricultural economy, DFI can influence WTAAT more significantly. This result shows that farmers’ emphasis on agriculture affects their WTAAT [19,23,24,59], which provides a reference for DFI to develop targeted services in rural areas.

6.2. Conclusions and Suggestions

This study analyzes the impact of DFI on farmers’ willingness to adopt agricultural technology (WTAAT) using data from the China Labor-force Dynamics Survey and the Digital Financial Inclusion Index of Peking University, China. The results show that DFI significantly increases farmers’ WTAAT, consistent with the results of robustness tests. In terms of economic significance, the estimated coefficient of DFI suggests that a one standard deviation increase in DFI leads to a 60.58% (=(0.037 × 16.505)/1.008) increase in the WTAAT. Moreover, the analysis of moderating effects shows that the contribution of DFI to WTAAT increases with the level of financial market development. Finally, WTAAT is affected by DFI development among farmers who receive government subsidies, participate in production technology training, and have no local non-agricultural economy.

We formulated the following suggestions based on the above conclusions. First, the government should continue to promote the development of DFI in rural areas. This initiative involves accelerating the construction of communication and financial infrastructure in rural areas, further reducing network service costs, and making it easier for farmers to obtain and use DFI services. Another step is to broaden the supply channels of DFI and develop DFI products and services suitable for the adoption of agricultural technologies by combining its characteristics and regions. Lastly, this initiative calls for improvement in the level of education, popularization of financial knowledge, expansion of farmers’ financial literacy and ability to resist risks, development of the level of human capital in rural areas, and bridging the education gap and digital gap.

Second, with the development of local financial markets, the impact of DFI on the WTAAT would be significantly enhanced. Therefore, in places with a relatively high level of financial market development, the government should continue to improve the financial market environment to promote the rapid and healthy development of DFI. In places with poor financial market development, efforts should be made to strengthen traditional financial services and contribute a solid foundation and support to DFI development and functioning.

Third, the government should enhance financial support for the adoption of agricultural technologies. Financial support can be a combination of ex-ante and ex-post subsidies. For ex-ante subsidies, the government should flexibly adjust the amount and duration of subsidies, maintain the continuity of the subsidy policy, reduce the cost of adopting agricultural technologies, and form stable subsidy expectations to encourage farmers to adopt agricultural technologies. For ex-post subsidies, a systematic and long-term subsidy policy should be adopted to ensure the stability of the subsidy effect and encourage farmers to adopt agricultural technologies.

Fourth, the government should strengthen agricultural technical training. Authorities should integrate various resources (e.g., state, corporate, and social welfare organizations), expand the investment scale of agricultural technology training, improve the technical ability of trainers, and use various channels to impart agricultural technology knowledge to farmers, all to enhance farmers’ understanding of agricultural technology and its economic benefits. At the same time, the government should promise and ensure that farmers who use agricultural technology are provided with comprehensive scientific guidance and guarantees, improve their technology application level, and reduce the possible risks of technology application.

Lastly, the development of a non-agricultural economy will gradually change the focus of farmers’ employment while activating market vitality in rural areas and reducing the emphasis on agriculture, which is not conducive to the adoption of agricultural technology. Therefore, on the one hand, the government should actively promote moderate-scale operations formed by land transfer and cultivate new agricultural business entities, such as family farms, large professional households, farmers’ cooperatives, and agricultural industrialization enterprises. On the other hand, the authorities should encourage and support non-agricultural development to drive agricultural development, guide and motivate farmers to adopt agricultural technology actively, and improve agricultural production capacity.

The importance of agriculture in China and the reality that per capita water and soil resources are scarce and poorly matched make sustainable agricultural development an issue that must be considered in the process of agricultural modernization. Agricultural technology inputs can achieve a virtuous cycle of resources, environment, and food supply and provide strong technical support for sustainable agricultural development. Our study provides new empirical evidence to explore the field of agricultural technology adoption from the DFI perspective. It also provides new ideas for the combination of digital financial transformation and sustainable agricultural development, enriches the research development in this field, provides a micro basis for government departments to improve policies related to agricultural technology diffusion, and provides some reference for DFI institutions to precisely develop financial products and services. It may also provide policy insights for the digital transformation of financial products and services, modernization, and sustainable development of agriculture in other developing countries.

This study is limited by the items of the questionnaire, the use of cross-sectional data for analysis, and the lack of analysis of the dynamic change process. Therefore, the dynamic analysis of the impact of DFI on the WTAAT is an important research direction in the future.