How Does Internet Use Promote Farmer Entrepreneurship: Evidence from Rural China

Abstract

Entrepreneurship and innovation are important driving forces for economic sustainable development. Despite the rapid popularity of the Internet in rural areas, whether—and if so, how—the Internet use may affect farmer entrepreneurship remains a key research gap. This paper studies the impact of the use of the Internet on farmer entrepreneurship and its mechanism by using Probit model, the Karlson–Holm–Breen (KHB) method and China Family Panel Studies (CFPS) dataset from 2014 to 2018. It is found that: (1) Use of the Internet has a positive impact on farmer entrepreneurship, and this result remains robust after addressing endogeneity. (2) Necessity entrepreneurship is more likely to be affected by use of the Internet than opportunity entrepreneurship. (3) Use of the Internet can significantly influence necessity entrepreneurship by affecting farmers’ risk attitude, social capital and information acquisition, while opportunity entrepreneurship is not affected by these mediating effects. (4) Among three mediating effects, the effect of social capital accounts for the largest contribution to the impact of the use of the Internet on farmer entrepreneurship. Our empirical findings could provide theoretical references for policies or reforms intended to promote entrepreneurship in rural regions.

1. Introduction

Entrepreneurship provides continuous impetus for economic growth [1,2,3]. Stimulating people’s willingness for entrepreneurship and unleashing their entrepreneurial potential are of great significance to sustainable economic development. Therefore, understanding the influencing factors and mechanism of entrepreneurship has drawn increasing attention from scholars and practitioners. Existing works have studied the drivers of entrepreneurship activity from different areas, such as institutional factors [4,5], financial factors [6,7], social resources [8,9] and individual characteristics [10].

Internet-related technology changes have a profound impact on entrepreneurship activity, as well as economic development [11,12]. The Internet provides entrepreneurs with more information about business opportunities and business models, reducing information asymmetry and potential risks [13]. Entrepreneurial activities are much less likely to be restricted by physical distance with the help of the Internet and e-commerce [14,15]. Internet use can facilitate finance of entrepreneurial activity through online crowdfunding [16]. Studies revealed that the influence of Internet utilization on entrepreneurship is more significant in rural regions than in urban regions [17,18].

To cope with the dual pressures of slowing economic growth and severe employment situations, the Chinese government has successively proposed “Mass Entrepreneurship, Mass Innovation” and “Rural Revitalization Strategy” to encourage people across the country, including people in rural areas, to carry out entrepreneurial activity. Farmer entrepreneurship is an important approach to increase their income and improve their living standards [18]. Moreover, farmer entrepreneurship can increase employment, alleviate poverty, and revitalize the rural economy [19,20,21]. Farmers’ entrepreneurial decision-making and entrepreneurial activity have been the research hotpots [22,23,24,25,26]. Meantime, the number of netizens in China has reached 989 million and the network coverage has reached 70.4% by the end of 2020 [27]. However, little attention has been paid to the influence of Internet use on farmer entrepreneurship in rural regions. Several works have investigated the association between the Internet use and farmer entrepreneurship, suggesting that the Internet has a significant and positive impact on entrepreneurship, especially in rural areas [27,28]. Promoting farmer entrepreneurship is of great importance for developing countries that have a large rural population. This paper therefore aims to provide quantitative empirical evidence on how use of the Internet impacts farmer entrepreneurship in rural China using data of a nationwide household survey.

Typical sustainable entrepreneurship is characterized by low-carbon and green, entrepreneurial activity in rural areas and has been advocated by both scholars and practitioners. Sustainable entrepreneurship also emphasizes high-quality and responsible entrepreneurship. While focusing on quantitative changes in entrepreneurship, structural or qualitative changes in entrepreneurship in rural areas should not be overlooked. Based on the entrepreneurial motivation, Global Entrepreneurship Monitor (GEM) divides entrepreneurship types into necessity entrepreneurship (individuals have to engage in small-scale entrepreneurial activity, owing to the lack of other better options, to meet survival needs), and opportunity entrepreneurship (individuals want to carry out large-scale entrepreneurial activity to create more opportunities and wealth) [12,27,29,30]. This paper attempts to probe the difference impact of Internet use on necessity entrepreneurship and opportunity entrepreneurship in rural China, which may provide enlightenment on the development of sustainable entrepreneurship in rural regions.

Existing studies have revealed the positive effect of the Internet on entrepreneurship and farmer entrepreneurship [17,18,27,28]; however, the mechanism of the effect remains unclear, and the precise mechanism of the Internet’s effect on farmers’ necessity entrepreneurship and opportunity entrepreneurship is even more ambiguous. Therefore, the aim of this paper is: (1) To explore whether and how use of the Internet affects farmer entrepreneurship, and three main mechanisms (risk attitude, social capital and information acquisition) are considered. (2) To investigate whether there are heterogeneous effects of Internet utilization on farmers’ necessity entrepreneurship and opportunity entrepreneurship. This study could enrich the literature on farmer entrepreneurship in terms of method and content by empirically analyzing the effect of Internet utilization on farmer entrepreneurship using the Probit model, the KHB method and cross-country micro-level survey data. The findings could provide governments with a better understanding of how to promote farmer entrepreneurship. Thus, policymakers can give targeted assistance to necessity entrepreneurs and opportunity entrepreneurs, which would contribute to economic sustainable development in rural regions.

The paper is organized as follows. The next section, Section 2, constructs arguments for hypotheses as well as providing data, variables and methods. Section 3 reports the results of the statistical analysis. Section 4 discusses the study findings, and Section 5 concludes.

2. Materials and Methods

2.1. Theoretical Framework and Hypotheses

2.1.1. The Analysis of the Internet Use and Farmer Entrepreneurship

The relationship between the technological environment and the development of entrepreneurship has been drawing attention from scholars [31,32]. Access to information and communication technology (ICT) which includes the Internet, mobile-phone and fixed-broadband, was proven to have positive influence on early stage entrepreneurial activity [31]. The use of ICT could broaden individual’s horizons, bring new ideas and opportunities, and thus improving the likelihood of entrepreneurship [32]. Being able to use a computer at home helps to increase individuals’ entrepreneurial willingness [33]. One study revealed that broadband infrastructure boosts entrepreneurship rates more than highways and railways [34]. Broadband access has a significant impact on the location decision of entrepreneurial firms in rural areas [35]. Researchers also found that broadband speed has a more significant impact on rural and agricultural entrepreneurial activity than cities [36]. The first hypothesis is proposed as follows:

Hypothesis 1 (H1).

Use of the Internet can promote farmer entrepreneurship.

2.1.2. Analysis of Risk Attitude, the Internet Use and Farmer Entrepreneurship

Entrepreneurship involves risks, including financial risks related to possible income losses and bankruptcy, as well as non-financial risks related to personal failure. The relationship between entrepreneurship and risk attitude has been widely discussed in the related literature [37,38,39]. Study showed that risk attitudes have a non-linear effect on the possibility of becoming an entrepreneur, and risk-neutral individuals are more likely to become entrepreneurs than risk-averse and risk-seeking individuals [40]. Entrepreneurs’ risk perception has a significant positive impact on their financial risk tolerance, risk propensity and entrepreneurial openness [41]. Risk-taking attitude may differ within the group of entrepreneurs [42]. Although sustainable entrepreneurs are more likely to fear personal failure than average entrepreneurs, there is no significant difference between them in risk attitude or perceived financial risk [43]. For individuals with different entrepreneurial motivations, studies indicated that opportunity entrepreneurs are more willing to take risks than necessity entrepreneurs, while opportunity entrepreneurs and wage workers only have marginal differences in risk attitudes [40,44,45]. Therefore, the following three hypotheses should apply:

Hypothesis 2 (H2).

The mediating role of risk attitude is significant for the impact of the use of the Internet on farmer entrepreneurship.

Hypothesis 2a (H2a).

The mediating role of risk attitude is significant for the impact of the use of the Internet on necessity entrepreneurship.

Hypothesis 2b (H2b).

The mediating role of risk attitude is significant for the impact of the use of the Internet on opportunity entrepreneurship.

2.1.3. Analysis of Social Capital, the Internet Use and Farmer Entrepreneurship

The latest studies have been emphasizing the importance of the influence of social capital on individuals’ entrepreneurial intention and behavior [46,47,48,49,50,51]. Empirical research in rural Pakistan indicated that social capital can promote entrepreneurial intentions [49]. Individuals who share social networks with entrepreneurs are more likely to conduct entrepreneurial activity in agriculture industry [50]. Study showed that social capital has a direct and positive effect on Iranian farmers’ entrepreneurial behavior [51]. Interpersonal communication is an essential way for individuals to develop and maintain social networks, and finally to obtain more social and economic resources. Use of the Internet can develop individuals’ social networks by increasing communication with others, and enlarge social networks without time or space constraints [52,53]. Evidence suggested that ICT utilization helps to expand individual’s social networks, and thus positively influence his/her entrepreneurial activity [24]. Based on all the above findings, the hypotheses are as follows:

Hypothesis 3 (H3).

Social capital plays a mediating role in the impact of use of the Internet on farmer entrepreneurship.

Hypothesis 3a (H3a).

Social capital plays a mediating role in the impact of use of the Internet on necessity entrepreneurship.

Hypothesis 3b (H3b).

Social capital plays a mediating role in the impact of use of the Internet on opportunity entrepreneurship.

2.1.4. Analysis of Information Acquisition, the Internet Use and Farmer Entrepreneurship

The Internet is an essential tool for individuals to engage in online learning, online business and online entertainment, and to obtain a wide variety of information which helps them prepare to start their own business [27]. Use of the Internet has a strong informative effect for entrepreneurs, particularly for farmer entrepreneurs, as it can broaden channels for them to acquire knowledge and skills, enhance the efficiency of entrepreneurial learning [54,55]. A large amount of business-related information on the Internet helps entrepreneurs effectively identify entrepreneurial opportunities, reduce transaction costs and realize efficient allocation of resources [27,56]. Moreover, the large social networks and diverse entertainment offered by the Internet have a significant impact on their ability to generate and implement entrepreneurial ideas in the pre-entrepreneurial phase [57,58,59]. The impact of the Internet’s information acquisition functions through online learning, online business and online entertainment on entrepreneurial activity is worth identifying. The related hypotheses are proposed as follows:

Hypothesis 4 (H4).

The mediating role of information acquisition is significant in the effect of Internet use on farmer entrepreneurship.

Hypothesis 4a (H4a).

The mediating role of information acquisition is significant in the effect of Internet use on necessity entrepreneurship.

Hypothesis 4b (H4b).

The mediating role of information acquisition is significant in the effect of Internet use on opportunity entrepreneurship.

The logic behind the formulation of these above-mentioned hypotheses is: First of all, it is necessary to confirm whether the use of the Internet will affect farmer entrepreneurship. If so, then it is worth analyzing how the use of the Internet affects farmer entrepreneurship, farmers necessity entrepreneurship and farmer opportunity entrepreneurship according to the research findings of the existing literature. The logical graph of hypotheses proposed in this paper is shown in Figure 1.

2.2. Methods

2.2.1. Data Source

The data used in this paper are mainly from the China Family Panel Studies (CFPS) conducted by the Institute of Social Science Survey (ISSS) of Peking University. The CFPS project is a study with biannual surveys, collecting detailed information on households. It covers 25 provinces of China. The information collected includes demographic characteristics, social relations, economic activities, income condition, Internet user behavior, etc. As a longitudinal tracking survey, the data provided by CFPS have become the most important data source for studying Chinese families and have advantages in exploring the impact mechanism of Internet use on farmer entrepreneurship.

Hukou is a population registration system in China, and each individual’s Hukou has a specific relationship with the area in which he/she resides [27]. Therefore, hukou status is used to distinguish rural and urban residences in this paper. The sample used in this paper includes data from 29,358 rural households.

2.2.2. Probit Model and KHB Method

The dependent variable (farmer entrepreneurship) in this article is a binary dummy variable, and the traditional regression model often has the problem that the predicted value exceeds the range of the dummy variable when dealing with such variables, and the Probit model is widely used to process such problems to make the regression result in the normal range. Therefore, the Probit model was chosen to analyze the impact of Internet use on farmer entrepreneurship. The benchmark regression model is set as follows:

$$pr\left(y_i=1\right)=\phi \left({\alpha }_1{\mathit{Internet}}_i+\beta x_i+{\epsilon }_i\right)$$

(1)

where $y_i$ indicates whether family member i in a rural household engages in self-employed business. ${\mathit{Internet}}_i$ indicates whether family member i in a rural household access to Internet. $x_i$ is the control variable including variables of individual characteristics, variables of household characteristics, region fixed effect, time fixed effect and constant term. ${\alpha }_i$ and β are the coefficients to be estimated, where β is the vector form and ${\epsilon }_i$ is the error term.

In order to explore whether Internet use has an impact on farmer entrepreneurship through risk attitude, social capital and information acquisition, the KHB method is selected to decompose the three types of mediating effects. Currently, the Sobel method and Bootstrap method are widely used to test the mediation effect besides the traditional stepwise test method of regression coefficient, but none of these methods are fully applicable to cases in which a non-linear mediation effect model also contains multidimensional mediation variables. The traditional mediation effect test method does not meet the research well with the Probit model, as it is not a traditional linear probability model, and five mediation variables are selected. Compared with other methods of mediating effect calculation, the KHB method has a wider range of applications, which can be used for the mediating effect analysis of non-linear probability models and the case of multi-dimensional mediator variables [60]. Therefore, the KHB method is selected in this paper.

2.2.3. Describe of Variables

There are five categories of variables used in this paper, as shown in

Table 1. The definition and description of variables.

Variables	Definitions of Variables	2014		2016		2018
		Mean	Std.	Mean	Std.	Mean	Std.
Dependent variable
Farmer entrepreneurship	Family members engage in self-employed business = 1, otherwise = 0	0.092	0.287	0.104	0.306	0.098	0.298
Necessity entrepreneurship	The number of self-employed businesses = 1	0.086	0.281	0.098	0.298	0.092	0.289
Opportunity entrepreneurship	The number of self-employed businesses ≥ 2	0.007	0.086	0.008	0.088	0.008	0.092
Independent variable
Internet use	Access to Internet = 1, otherwise = 0	0.168	0.374	0.317	0.466	0.443	0.497
Control variable
Health	Self-rated health status as good and above = 1, otherwise = 0	0.650	0.477	0.628	0.483	0.669	0.471
Age	Age of householder	48.428	14.264	48.823	14.831	49.139	15.172
Gender	Male = 1, female = 0	0.541	0.498	0.526	0.499	0.537	0.499
Education	Dividing 8 levels as self-rated education status, primary or below = 1, doctor = 8	2.216	1.094	2.273	1.120	2.817	1.382
Marital status	Married = 1, otherwise = 0	0.852	0.355	0.833	0.373	0.822	0.383
Family size	The number of family members	3.896	1.905	3.879	1.980	3.722	1.994
Family income	Total annual household income	4.355	7.794	6.313	19.388	7.058	13.314
Household deposit	Household deposit	1.870	5.493	2.793	9.056	3.432	11.752
Region of residence	East = 1, midland = 2, west = 3	1.910	0.843	1.916	0.846	1.920	0.849
Instrumental variable
Regional average Internet utilization	Regional average Internet utilization in different provinces	47.234	11.762	52.538	10.967	53.200	9.744
Mediator variable
Risk attitude	Owning financial products = 1, otherwise = 0	0.012	0.109	0.016	0.126	0.020	0.138
Social capital	Gift expenditure and communication expenditure each account for 50% weight	0.746	0.772	0.810	1.047	0.872	0.963
Online learning	Self-rated online learning status as almost everyday = 1, otherwise = 0	0.179	0.384	0.178	0.383	0.168	0.374
Online business	Self-rated online business status as almost everyday = 1, otherwise = 0	0.020	0.138	0.186	0.135	0.016	0.126
Online entertainment	Self-rated online entertainment status as almost everyday = 1, otherwise = 0	0.130	0.336	0.129	0.335	0.058	0.234

.

(1)

Dependent variable.

Based on the question “Whether the family members engage in self-employed business” in CFPS [7,61], the dependent variable “farmer entrepreneurship” is defined. If the answer is yes, the value of the variable is one; otherwise, it is zero. The number of self-employed businesses is used as the criterion for dividing necessity entrepreneurship and opportunity entrepreneurship (the question is “The number of self-employed businesses that the family members manage” [61]); it is classed as opportunity entrepreneurship if the number is over one. Otherwise, it is necessity entrepreneurship.

(2)

Independent variable.

The independent variable is the Internet use. It is defined based on the question “Do you use mobile devices to access the Internet?” [61]. If the answer is yes, the value of the variable is one; otherwise, it is zero.

(3)

Control variable.

The control variables are divided into two categories: variables of individual characteristics such as health situation, age, gender, education and marital status come from the Personal Questionnaire in CFPS. Additionally, variables of household characteristics such as family size, family income, household deposit and region of residence come from the Family Economy Questionnaire in CFPS.

(4)

Instrumental variable.

The instrumental variable is the regional average Internet utilization in each province over the years. The data for 2014 and 2016 come from the Statistical report on the State of the Internet in China released by the China Internet Network Information Center. The data for 2018 come from the Internet Development Report released by the China News.

(5)

Mediator variable.

Based on the question “Does your family own any financial products?” in CFPS [61,62], the mediator variable “risk attitude” is defined. If the answer is yes, the value of variable is one; otherwise, the value is zero. Chinese farmers have a relatively low level of understanding of financial products, as well as trust in rural financial institutions. Most farmers are wary of buying financial products because they believe it could increase liabilities and lead to property losses. Farmers’ attitudes and behavior in purchasing financial products can reflect their risk attitude to some extent. Therefore, whether farmers buy financial products is selected as the source of the variable “risk attitude”.

Based on the question “The total amount of money family spent on gifts for social relations” and “The total amount of money family spent on transportation (i.e., bicycle, electric bicycle) and communication tools (mobile phone, etc)” in CFPS [61,63], the mediator variable “social capital” is defined, and the two expenditures each account for 50% of social capital. Social networks in rural China are mainly based on family relationships and friendships, and one of the important means of communication and bonding between relatives and friends is to exchange gifts during festivals or weddings and funerals. Moreover, farmers often have to spend transportation and communication costs on social activities due to the low population density and inconvenient transportation in rural regions. Therefore, gift expenditure and communication expenditure were selected as the sources of variable “social capital”.

Information acquisition includes online learning, online business and online entertainment. Based on the question “The frequency of using Internet to study”, “The frequency of using Internet to do commercial related activities” and “The frequency of using Internet to entertain” in CFPS [61], the variables of online learning, online business and online entertainment are defined, respectively. These variables are divided into two levels of self-rated status. If the answer is almost every day, the value is one. For other answers, the value is zero.

(6)

Statistics description of variables.

It can be found from Table 1 that: (1) The amount of farmer entrepreneurship has shown an overall upward trend, with an increase of 6% in 2018 compared with 2014. (2) In contrast, the amount of necessity entrepreneurship in rural Chinese households is much greater than that of opportunity entrepreneurship. Additionally, the average amount of necessity entrepreneurship in 2018 was 0.092, which is 11.5 times the amount of opportunity entrepreneurship (0.008). (3) The Internet use among Chinese farmers has shown a significant upward trend from 2014 to 2018; the mean value of Internet use by farmers in 2018 is 2.637 times that shown in the data from 2014.

3. Results

3.1. Results of Benchmark Regression Model

The panel model was constructed using the CFPS data in 2014, 2016 and 2018. Using OLS model to estimate the impact of the Internet use on farmer entrepreneurship, the mean VIF value of all the variables is 1.25, and the VIF value of all the variables is far less than 10, indicating that there is no collinearity problem among the variables. Based on the model (1), the Probit model is used to estimate the total effect of Internet use on farmer entrepreneurship.

Table 2. Results of benchmark regression.

	(1) The Whole Sample	(2) Necessity Entrepreneurship	(3) Opportunity Entrepreneurship
Internet use	0.063 *** (0.006)	0.058 *** (0.006)	0.009 *** (0.002)
Age	−0.004 (0.001)	−0.005 (0.001)	0.000 (0.003)
Education	0.020 *** (0.002)	0.014 *** (0.002)	0.002 *** (0.001)
Marital status	0.038 *** (0.007)	0.038 *** (0.006)	0.001 (0.002)
Gender	−0.002 (0.004)	−0.001 (0.004)	−0.000 (0.001)
Health	0.010 ** (0.005)	0.008 *** (0.005)	0.004 *** (0.002)
Household deposit	0.001 *** (0.000)	0.001 *** (0.000)	0.000 * (0.000)
Family size	0.012 *** (0.001)	0.011 *** (0.001)	0.002 *** (0.000)
Family income	0.009 *** (0.001)	0.007 *** (0.001)	0.005 *** (0.001)
Region fixed effect	Yes	Yes	Yes
Time fixed effect	Yes	Yes	Yes
N	29,358	29,150	26,676

Note: Standard errors are in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1. Regression coefficients are marginal effects.

shows the results of benchmark regression. It indicates that Internet use promotes farmer entrepreneurship at 1% significance level for all samples, and farmers with Internet access are 6.3 percent more likely to engage in entrepreneurship than those without. This validates Hypothesis 1, agreeing that the Internet use positively effects farmer entrepreneurship in China. Internet use can also increase the probability of necessity entrepreneurship and opportunity entrepreneurship. In contrast, the influence of Internet use on necessity entrepreneurship (5.8 percent) is much greater than the impact on opportunity entrepreneurship (0.9 percent). This suggests that Internet use affects farmers with varying degrees of influence in terms of necessity entrepreneurship and opportunity entrepreneurship.

3.2. Results of Endogeneity Test

In order to accurately estimate the possible endogeneity problems in the Probit model, the average Internet use of every province in China is selected as an instrumental variable. The logic behind this is that the average Internet use largely reflects residents’ acceptance of new things such as the Internet, and the average network usage in a region is more representative than the network usage of individuals. Therefore, there is basis to consider that regional average Internet utilization has a strong exogeneity to the entrepreneurial behavior of farmers.

Table 3. The result of instrumental variable regression.

	(1) The Whole Sample	(2) Necessity Entrepreneurship	(3) Opportunity Entrepreneurship
Internet use	0.015 *** (0.002)	0.016 *** (0.002)	0.016 *** (0.002)
Control variables	yes	yes	yes
N	21,012	20,872	19,012
p value of the instrumental variable	0.000	0.000	0.000
p value of Wald exogeneity test	0.000	0.000	0.497

Note: Standard errors are in parentheses; *** p < 0.01. Regression coefficients are marginal effects. The control variables include variables of individual characteristic, variables of household characteristic, region fixed effect and time fixed effect.

shows the result of instrumental variable regression. The coefficient of the first-stage regression of the independent variable “Internet” is significantly positive at the 1% level for the whole entrepreneurship samples, necessity entrepreneurship sample and opportunity entrepreneurship sample. This shows that the regional average Internet use has a positive impact on individual Internet utilization, and the instrumental variables are strongly correlated.

The Wald exogeneity test rejects the null hypothesis that there is no endogeneity of the independent variable “Internet use” for the whole samples, but necessity entrepreneurship sample accepts the null hypothesis. This shows that there is a significant endogenous problem in the necessity entrepreneurship sample. The impact of Internet utilization on the necessity entrepreneurship is still positive at the 1% level after correcting the endogenous problem, which further indicates that Internet use plays a significant positive role in necessity entrepreneurship. Additionally, opportunity entrepreneurship does not have a significant endogenous problem; regional average Internet use does not significantly affect opportunity entrepreneurship.

3.3. Results of Mediating Effect Test

The impact of Internet use on farmers entrepreneurship through the mediator variables is called the mediating effect. The direct impact of Internet use on entrepreneurship without the influence of other variables is called the direct effect. The sum of the mediating effect and the direct effect is the total effect of Internet use on farmer entrepreneurship, as shown in Figure 2.

3.3.1. The Mediating Effect of Risk Attitude

Individuals’ perception of risk attitude is a key issue affecting entrepreneurial behavior. Farmers generally dare to take risks. It is necessary to investigate whether Internet use can promote farmer entrepreneurship by influencing their risk attitude.

Table 4. The mediating effect of risk attitude.

	(1) The Whole Sample	(2) Necessity Entrepreneurship	(3) Opportunity Entrepreneurship
The total effect	0.483 *** (0.026)	0.470 *** (0.026)	0.421 *** (0.070)
The direct effect	0.310 *** (0.034)	0.298 *** (0.035)	0.370 *** (0.093)
The mediating effect	0.173 *** (0.025)	0.171 *** (0.025)	0.050 (0.068)

Note: Standard errors are in parentheses; *** p < 0.01. Regression coefficients are marginal effects.

verifies the mediating effect of risk attitude on the impact of Internet use on farmer entrepreneurship. Column (1) contains the overall regression results, and columns (2) and (3) are the categorical regression results of necessity entrepreneurship and opportunity entrepreneurship. The mediating effect value of the whole sample and necessity entrepreneurship sample are significantly positive at the level of 1%, suggesting that Hypothesis 2 and Hypothesis 2a are validated, and risk attitude plays a mediating role in the impact of Internet use on farmer entrepreneurship and necessity entrepreneurship in rural China. The mediating effect value of opportunity entrepreneurship sample is not positive, and Hypothesis 2b is not validated. Moreover, the direct effect of the Internet use on farmer entrepreneurship and necessity entrepreneurship is greater than the mediating effect of risk attitude on entrepreneurship.

3.3.2. The Mediating Effect of Social Capital

Entrepreneurial behavior is closely related to individuals’ social relations. The social capital of the entrepreneurial family is the embodiment of this relationship, and the Internet enhances social capital by providing a convenient social platform.

Table 5. The mediating effect of social capital.

	(1) The Whole Sample	(2) Necessity Entrepreneurship	(3) Opportunity Entrepreneurship
The total effect	0.500 *** (0.025)	0.486 *** (0.025)	0.449 *** (0.066)
The direct effect	0.357 *** (0.033)	0.342 *** (0.033)	0.408 *** (0.089)
The mediating effect	0.143 *** (0.024)	0.144 *** (0.024)	0.040 (0.066)

Note: Standard errors are in parentheses; *** p < 0.01. Regression coefficients are marginal effects.

verifies the mediating effect of social capital in relation to the impact of the Internet use on farmer entrepreneurship. The mediating effect value of the whole sample and necessity entrepreneurship sample are significantly positive at the level of 1%, reflecting that there is a mediating role of social capital in the influence of the Internet use on farmer entrepreneurship and necessity entrepreneurship (Hypothesis 3 and Hypothesis 3a are validated). On the contrary, the mediating role of social capital is not pronounced in farmer opportunity entrepreneurship (Hypothesis 3b is not validated). Similar to the situation with risk attitude, the direct effect of Internet use on farmer entrepreneurship and necessity entrepreneurship is much greater than the mediating effect of social capital on entrepreneurship.

3.3.3. The Mediating Effect of Information Acquisition

Individuals’ entrepreneurial behavior depends, in part, on the amount of information they gain. It is necessary to examine whether farmer entrepreneurship is promoted by the greater information acquisition provided by Internet use. Online learning, online business, and online entertainment are used as proxy variables for information acquisition in this paper.

Table 6. The mediating effect of information acquisition.

Category of Information Acquisition		(1) The Whole Sample	(2) Necessity Entrepreneurship	(3) Opportunity Entrepreneurship
Online learning	The total effect	0.500 *** (0.025)	0.486 *** (0.025)	0.447 *** (0.066)
	The direct effect	0.362 *** (0.042)	0.345 *** (0.043)	0.443 *** (0.110)
	The mediating effect	0.138 *** (0.036)	0.141 *** (0.036)	0.004 (0.096)
Online business	The total effect	0.495 *** (0.025)	0.483 *** (0.025)	0.392 *** (0.069)
	The direct effect	0.344 *** (0.033)	0.333 ** (0.033)	0.369 *** (0.091)
	The mediating effect	0.151 *** (0.024)	0.150 *** (0.024)	0.023 (0.067)
Online entertainment	The total effect	0.499 *** (0.025)	0.486 *** (0.025)	0.447 *** (0.066)
	The direct effect	0.338 *** (0.036)	0.324 *** (0.036)	0.364 *** (0.096)
	The mediating effect	0.161 *** (0.027)	0.162 *** (0.028)	0.082 (0.074)

Note: Standard errors are in parentheses; *** p < 0.01, ** p < 0.05. Regression coefficients are marginal effects.

verifies the mediating effect of information acquisition in the impact of Internet use on farmer entrepreneurship. The mediating effect value of the whole sample and necessity entrepreneurship sample are significantly positive at the level of 1%, proving that Internet use can promote farmer entrepreneurship and necessity entrepreneurship by enriching the information acquisition activities of online learning, online business and online entertainment (Hypothesis 4 and hypothesis 4a are validated). The mediating role of information acquisition on farmer opportunity entrepreneurship (Hypothesis 4b) is not significant according to the results.

3.3.4. Contribution of Mediating Effects

While judging the mediating effects of the above-mentioned risk attitude, social capital and information acquisition in the impact of the Internet use on farmer entrepreneurship, it is necessary to further analyze the contribution of these three types of mediating effects on farmer entrepreneurship.

Table 7. The contribution of the mediating effects.

	(1)	(2)	(3)	(4)
	The Whole Sample		Necessity Entrepreneurship
	Absolute Proportions	Relative Proportions	Absolute Proportions	Relative Proportions
Social capital	10.66%	50.54%	10.17%	51.64%
Online learning	5.35%	25.36%	4.60%	23.35%
Online entertainment	2.62%	12.40%	3.12%	15.85%
Online business	2.44%	11.54%	1.77%	8.96%
Risk attitude	0.03%	0.16%	0.04%	0.20%
The total mediating effects	21.10	100%	19.70	100%
N	19,231		19,101

shows the comparison of the contribution of the mediating effects calculated by the KHB method.

Columns (1) and (2) show the contribution of risk attitude, social capital and information acquisition to the whole sample. Overall, in absolute proportions, the contributions of social capital, online learning, online entertainment, online business and risk attitude to Internet use to promote farmer entrepreneurship are 10.66%, 5.35%, 2.62%, 2.44% and 0.03%, respectively. From a relative scale, the contribution of the mediating effect of social capital accounts for the largest proportion, reaching 50.24%. The subsequent is the contribution of the mediating effect of information acquisition as online learning, online entertainment and online business accounts for 25.36%, 12.40% and 11.54%, respectively. The mediating effect of risk attitude has the lowest contribution, whose relative proportion is only 0.16%.

Columns (3) and (4) illustrate the contribution of the five mediator variables to the necessity entrepreneurship sample. The mediating effect contribution of social capital accounts for the largest proportion, reaching 51.64%. Following this is online learning, online entertainment and online business. The contribution proportion of risk attitude is the lowest, with a relative proportion of only 0.20%.

Since the mediating effect value of opportunity entrepreneurship is not significant, the contribution analysis is not performed.

4. Discussion

Through Probit model regression, an endogeneity test and a mediation effect test, it is found that Internet use has a significant impact on farmer entrepreneurship in rural China, which is consistent with the conclusions of Conroy and Low [22], Barnett et al. [24], Huang et al. [26], Tan and Li [27] and Robert [33]. Moreover, the influence of Internet use on necessity entrepreneurship is much greater than that of opportunity entrepreneurship, which aligns with the conclusion of Tan and Li [27]. These findings suggest that with the increase in the Internet use in rural China, the willingness of farmers to start a business has also increased significantly, and necessity entrepreneurship has been affected more than opportunity entrepreneurship. According to the analysis results of the mediating effect, the reasons for the findings are geographical, economic, educational, etc. People in rural areas have relatively little communication with others, the information they obtained is limited and their risk attitude is relatively conservative. More importantly, they have very little understanding of entrepreneurial activity, and naturally, they struggle to conceptualize entrepreneurship. Due to widespread Internet use in rural regions, people interact with others online and learn online; their social networks have been expanded. More knowledge including the mode, benefits and risks of entrepreneurial activity has been obtained, and therefore their entrepreneurial willingness has improved. For necessity entrepreneurs, when they obtain information related to entrepreneurship through the Internet, their entrepreneurial willingness will significantly improve because they have no other better choice to meet their survival needs. For opportunity entrepreneurs, the impact of social networks, entrepreneurial information and knowledge brought by the Internet on their entrepreneurial intention is limited, because this is their first attempt at entrepreneurial activity.

The results of the mediation effect analysis show that the mediating effects of social capital, risk attitude and information acquisition in the impact of the Internet use on the whole sample of entrepreneurship and necessity entrepreneurship are significant, which is similar to the finding of Barnett et al. [24] and Tan and Li [27], while the mediating effect on opportunity entrepreneurship is not validated. This indicates that for farmers considering starting a business to meet their survival needs, Internet use will positively affect their entrepreneurial decision-making by enriching their social capital [24,64], reducing their risk aversion [45] and broadening their access to knowledge, entertainment and business information [24,27]. In contrast, farmers motivated by personal interest in pursuing a business opportunity generally have lower risk aversion [44], more information access channels and social capitals than other farmers, so their entrepreneurial willingness is much less affected by Internet use.

Among the mediating effects of social capital, risk attitude and information acquisition on the whole sample of entrepreneurship and necessity entrepreneurship, the mediating effects of social capital account for the largest contribution to farmers’ entrepreneurship, and the effect of risk attitude makes the smallest contribution. In comparison with previous literature, the contribution of these three types of mediating effects has not been evaluated or considered in existing studies on farmer entrepreneurship. The result reflects that: (1) In rural areas with resource constraints, entrepreneurship activity is more dependent on social relationships such as relatives and neighbors to provide material capital and human resources. Internet use can provide more interpersonal communication channels to farmers and broaden their social networks to tackle these constraints, and positively impact their willingness to start their own business. (2) Although the Internet use of Chinese farmers will affect their risk attitude to a certain extent, some major decisions relating to family benefits, such as engaging in entrepreneurial activity, are more likely to be affected by their family members and relatives’ attitudes.

Among the mediating effects of three types of information acquisition on the whole sample of entrepreneurship and necessity entrepreneurship, online learning provides the largest contribution, followed by online entertainment. The contribution of online business is the smallest. This conclusion is similar to the findings of Huang et al. [26] and Tan and Li [27]. The reasons for this result may be: (1) With the intervention of the Internet, farmers have more access to knowledge about entrepreneurship through network in rural areas with a relatively low education level. With the increase in knowledge related to entrepreneurship, farmers’ awareness of details of entrepreneurial activity becomes much clearer, and they are more confident about starting a business successfully. (2) Entertainment activities in rural areas are relatively scarce. Internet use has provided more online entertainment activities, accompanied by social communication and entrepreneurial information for farmers who may never have considered entrepreneurship, which can ignite the idea of entrepreneurship to some extent. (3) Due to the generally low level of education in rural areas, it is difficult for farmers to distinguish the authenticity of some business information on the Internet, as the judgment of the information often requires specific professional knowledge. Therefore, the contribution of the mediating effect of online business on entrepreneurship is relatively lower than other two channels of information acquisition.

5. Conclusions

Based on the data of the China Family Panel Studies (CFPS) in 2014, 2016 and 2018, this paper empirically studies the impact of Internet use on farmer entrepreneurship in rural China. It is found that the development of the Internet and the rise of digitalization in rural areas are conducive to improving the entrepreneurial tendency of farmers, improving the living environment and conditions of farmers and promoting an increase in regional economic income. These positive influences persist after correcting endogenous problems. The mediating effect analysis shows that the Internet use can promote farmers’ willingness to engage in entrepreneurship by broadening social capital, changing risk attitude and promoting access to information.

The results show that: (1) Internet use plays a positive role in farmer entrepreneurship with the popularity of the Internet in rural areas of China. (2) Generally, the entrepreneurial willingness of necessity entrepreneurs is more likely to be affected by Internet use than that of opportunity entrepreneurs. (3) Internet use can significantly influence necessity entrepreneurship by affecting farmers’ social capital, information access and risk attitude, while opportunity entrepreneurship is not affected by these mediating effects. (4) The mediating effect of social capital accounts for the largest contribution to the impact of Internet use on farmer entrepreneurship.

Based on the results of this study, the following suggestions are made: (1) Government departments should continue to strengthen the construction of Internet infrastructure and increase the average Internet-use rate in rural areas, so that farmers can make full use of the Internet to carry out high-quality and sustainable entrepreneurial activity. (2) Government departments need to strengthen the supervision of Internet social platforms and applications so that farmers can participate in active and beneficial online social activities. In the meantime, it is worth trying to set up a special rural Internet social networking platform to provide opportunities for farmers considering entrepreneurship to communicate related topics and find entrepreneurial partners. (3) Government departments should popularize farmer-entrepreneurship-related laws, regulations and assistance policies in rural areas and teach e-commerce and webcast knowledge to farmers who are considering starting a business, encouraging them to turn their willingness into entrepreneurial actions. (4) Government departments should become acquainted with the needs and difficulties of necessity entrepreneurs and opportunity entrepreneurs, respectively, and then provide targeted consultation and help for them to create a better entrepreneurial environment and conditions.

There are still some limitations of this study. First of all, the data in this paper come from CFPS. Due to the certain time lag of data release, this paper fails to obtain the latest data in 2020 to analyze the impact of Internet use on farmer entrepreneurship. Further studies can be conducted using the updated data. Second, due to the limitations of secondary structural data, the number of self-employed businesses is used as the criterion for dividing necessity entrepreneurship and opportunity entrepreneurship, which is feasible to some extent. In the future, other data can be used and combined to define these two variables from the perspective of entrepreneurial motivation. Finally, in terms of mechanism analysis, the influence mechanism of Internet use on farmer entrepreneurship is more complex, involving many factors. This paper preliminarily explores the role of risk attitude, social capital and information acquisition under limited conditions. Future research can be discussed in greater depth, considering the impact of the Internet on farmers’ entrepreneurship through capital acquisition, opportunity identification, market channel expansion, etc.