Impact of Rural E-Commerce on Farmers’ Income and Income Gap

Abstract

Rural e-commerce, as a new form of digital economy, is of great significance in promoting the income of rural households and realizing common prosperity. Based on the 2021 China Rural Revitalization Survey (CRRS), the impact of rural e-commerce on farmers’ income and the intra-rural income gap is explored using quantile regression. The results found that (1) rural e-commerce can effectively promote the level of farm household income and alleviate the intra-rural income disparity, with this finding still holding after addressing the potential endogeneity problem and conducting robust-type tests. (2) Rural e-commerce has the most pronounced income-generating effect on low-income households in the southern region; participation in rural e-commerce has a more “center-expanding” effect on households in the secondary education and high material capital groups. (3) Saving production and operation costs, shortening the product circulation chain, and improving access to information are channels through which rural e-commerce affects households’ income. It is recommended to actively promote the deep integration and development of rural e-commerce in rural areas, establish a sound mechanism for bridging the “digital divide”, encourage e-commerce “leaders” to build a perfect industrial chain, and guide smallholders to integrate into the rural e-commerce industrial chain to enjoy the digital dividend.

1. Introduction

Mutual enrichment is an essential requirement of socialism and an important feature of Chinese-style modernization. The report of the 20th Party Congress points out that “Chinese-style modernization is a modernization of mutual enrichment for all people” and that “efforts will be made to promote common wealth for all people”. Secretary Xi Jinping has stressed that, in order to better meet the growing needs of the people for a better life, the promotion of the common prosperity of all people must be made the focus of efforts to seek happiness for the people. In recent years, with the rapid growth of the rural economy, the level of per capita disposable income in rural areas has been rising, but the income gap within the rural population has exceeded the income gap between urban and rural areas and has become the main source of the year-on-year widening of the income gap in recent years. According to the Statistical Abstract of China, the multiplier difference between high-income and low-income rural households has shown a slight fluctuation over the past 10 years, but the overall trend is widening, with the multiplier difference widening from 7.41 in 2013 to 9.17 in 2022, which is much higher than the multiplier difference between high- and low-income urban households of 6.32. It can be seen that the most arduous and burdensome task in promoting common prosperity still lies in the countryside. Accelerating the entry of rural residents into the middle-income group and narrowing the income gap within the countryside is the only way to promote the common prosperity of farmers and rural inhabitants at this stage.

Agricultural income is an important source of income for rural families, so solving the problem of agricultural marketing is crucial to the strategy of rural revitalization. At the present stage, the problem of imbalance and insufficiency of development in China is more prominent in rural areas, and one of the most important manifestations of this is the coexistence of a phased oversupply of agricultural products and an oversupply of demand. The in-depth integration of the digital economy with the real economy in rural areas can realize the optimization of the efficiency of production, distribution, and consumption in rural areas and become a new engine to promote the economic growth of China’s rural areas. The traditional mode of circulation of agricultural products mainly takes large-scale wholesale farms of agricultural products as the pivot and reaches consumers through the links of acquisition at the place of origin, intermediate transportation, wholesale at the place of sale, and retail at the terminal. The excessively long industrial chain causes the cost of agricultural products to increase and even causes frequent food safety problems. At the same time, the imbalance between the supply and demand of agricultural products presents new characteristics, the market demand for product sales can not be conveyed to producers in a timely and efficient manner, and the phenomenon of product stagnation and loss is frequent [1]. Rural e-commerce is an important platform for the digital economy to empower the development of agriculture and rural areas, and its unique cross-regional and extensive characteristics have broken the previous time and space constraints, providing a connection point for the coordinated development of industries in rural areas and the accurate and efficient docking of agricultural products to the market. The “Internet +” era has led to the rapid development of the e-commerce economy. Since Central Document No. 1 in 2018 first proposed the digital countryside strategy, the Internet, as the core technology of networking, informatization, and digitization, has greatly facilitated the extension of the e-commerce economy in the agricultural industry, broadened the channels of modern agricultural production and sales, and provided a richer pathway to promote the development of a new type of agricultural economy, providing farmers with employment, income generation, and wealth [2]. With the wide application of digital technology in agriculture, rural e-commerce provides a new breakthrough for solving the problems of circulating and marketing agricultural products. In 2023, the national rural e-tail sales amounted to CNY 2.5 trillion, a year-on-year increase of 12.9 percent, and the e-tail sales of agricultural products amounted to CNY 587.03 billion, a year-on-year increase of 12.5 percent. The development of rural e-commerce has fundamentally changed the way agricultural products are sold, which not only promotes the flow of market information and improves market transparency, but also improves the ability of small farmers to interface with new channels and technologies, which in turn enables them to obtain higher returns in the supply chain of agricultural products [3,4]. Rural e-commerce not only breaks the spatial distance between producers and consumers and saves circulation links, but also promotes the docking of small farmers with big markets, opening up channels for consumers to buy agricultural products [5]. As a kind of sunrise industry and green industry, rural e-commerce is changing the traditional industrial pattern with rapid development speed and becoming a new engine and new power to promote the development of agriculture and rural areas [6]. It is worth noting that although the development of rural e-commerce has changed the way of production and life in rural areas, it has also caused an imbalance in the distribution of the digital dividend [7]. There is a serious problem of “hollowing out” in rural areas, where there is economic backwardness, poor infrastructure, a shortage of capital, and little employment information. Therefore, most of the young laborers choose to go out to work, and the farmers who stay in the countryside are mostly the elderly, who possess a low cultural quality level and poor acceptance of modern information technology, network marketing, and other new things. There are also some farmers who are reluctant to participate in rural e-commerce because of their low digital literacy and lack of knowledge and skills related to e-commerce sales channels. It can be seen that the root cause of the digital divide lies in the difference between the acceptance ability of farmers and their digital literacy when new technologies emerge, i.e., the advantages of the “rural elite” group make them become the “leaders” in adopting e-commerce in rural areas, compared with the disadvantaged groups. In contrast, the disadvantaged groups are seriously hindered from enjoying the digital dividend due to their lack of sexual ability. How the rise of rural e-commerce affects the income inequality brought about by the digital divide is a scientific question that deserves in-depth exploration.

This study found that the number of publications related to rural e-commerce increased rapidly from 2015 onwards by searching the recent literature on rural e-commerce in the CNKI (China National Knowledge Infrastructure) database and the Web of Science website. By collating the existing literature, it was found that at present, academics are mostly concerned with the income-generating effect of rural e-commerce and its impact on the urban–rural income gap, and there are few studies that analyze the impact of rural e-commerce on the income gap within the rural household in depth. In addition, the findings of these studies have not yet been unified. Some scholars believe that the “digital dividend” brought by the development of rural e-commerce is only embodied in a few “rural elite” groups compared to the lower digital literacy of rural households, because of their unfulfilled willingness to participate in e-commerce, meaning that they are unable to enjoy the dividends of rural e-commerce, thus widening the gap between the rich and the poor. In contrast, less digitally literate farmers are unable to enjoy the dividends of rural e-commerce because they are not willing to participate in e-commerce, thus widening the gap between the rich and the poor [8,9]. For example, using OLS regression and DID models to analyze micro-panel data from fixed observation points in rural China from 2010–2017, Fang Shile et al. found that rural e-commerce boosted rural household incomes while widening the wealth gap between high-income and low-income households [8]. Some scholars have also argued that the development of rural e-commerce reduces market information asymmetry, which contributes to the expansion of the sales scale of agricultural products on online platforms, which in turn promotes the employment and entrepreneurship of rural residents and reduces the internal income gap in the countryside [2,10]. For example, Fang Ying et al. and Liu, M et al. point out that rural e-commerce has changed the product structure of e-commerce transactions, the scale of agricultural output, and increased the subabundance of agricultural sales platforms, and that the number of online sales platforms and the scale of production of core specialty agricultural products have a significant income-generating effect on farm households [2,11]. Qin Fang et al. used OLS regression models to analyze data from the 2017 China Household Finance Survey and found that the popularization of rural e-commerce can boost farm household income by enhancing the level of entrepreneurship, increasing non-farm employment, and increasing the probability of land transfer [10]. Then, how does the development of rural e-commerce affect rural household income under the digital village strategy, and does it help to reduce the internal income gap in rural areas?

In view of this, this study aims to deeply analyze the impact and mechanism of rural e-commerce development on the household income of rural households and the internal income gap in rural areas from a micro perspective. In turn, it reveals the significance of the application of the digital economy and the three rural areas in promoting the economic development of rural areas and the sustainable increase of farm household income. The marginal contributions of this paper are mainly the following three points. First, the existing literature mostly analyzes the income-generating effect of the e-commerce adoption behavior of rural households from a micro perspective or uses panel data to analyze rural e-commerce to study the urban–rural income gap. Only a few items of literature have studied the impact of rural e-commerce development on the internal income gap of rural residents from the micro level. Therefore, this study takes farm household income and the internal rural income gap as the research object and adopts the riverboat two-step method to analyze the mechanism of rural e-commerce’s impact. This analytical method avoids the possible endogeneity problem of mediating variables in the previous three-step mediation effect method, which can realize the effective expansion of rural e-commerce on rural household income and income gap-related research. Second, this study uses the sample data obtained from the 2021 China Rural Revitalization Comprehensive Survey Database on randomly selected farm households in 50 counties in 10 provinces, including Guangdong, Zhejiang, Guizhou, etc., to explore the relationship between rural e-commerce and the internal income gap in the countryside and the mechanism of influence, which, to a certain extent, overcomes the limitations of the selection of provincial-level county macro-data for the sample data of the study of rural e-commerce in the national region. Third, unlike previous studies, this study divides the sample into north and south regions according to the Qinling-Huaihe River line and explores the heterogeneity of the impact of rural e-commerce development on intra-rural income disparity in terms of various household capitals, which is of great reference significance as it provides ideas that can be learned from rural e-commerce to fully empower the rural revitalization strategy.

2. Theoretical Analysis and Hypotheses

2.1. Rural e-Commerce and Farm Household Income

According to the theory of farmer behavior, the pursuit of maximizing one’s own effects is the baton of the farmer’s behavior as a rational economic agent. The “rational small farmer” school of thought, represented by Schultz, argues that small farmers are rational individuals who seek optimal efficiency in the same way as capitalist entrepreneurs. Therefore, as a rational economic subject, when farmers choose to sell their agricultural products in this way, their behavioral decisions will also take into account the constraints of a variety of internal and external conditions and determine the sales channels and methods of their agricultural products according to the principle of profit maximization. However, in the traditional circulation of agricultural products, farmers often accept the price of products based on past experience or passively, and farmers rely on the traditional middleman mode of selling agricultural products, meaning that the contradiction between “small farmers” and the “big market” still exists [12].

The essence of the rise of rural e-commerce lies in the breakthrough of spatial and temporal constraints, improved market transparency, focused buyers and sellers on a virtual platform, and weakened market information asymmetry, so that farmers’ access to information is strengthened and farmers regain the right to ordering and pricing. At the same time, the status of farmers in the industrial chain has been enhanced, which helps to motivate farmers to explore potential markets and expand the scale of production to obtain more income. Through the e-commerce platform, farmers can independently understand market demand and price trends, so as to make more reasonable production and sales decisions. In addition, the development of rural e-commerce has significantly compressed the intermediate links in the traditional distribution of agricultural products, enabling farmers and consumers to directly interact. With the help of Internet platforms, farmers are able to directly participate in the entire transaction process from pre-consultation, purchase, and electronic payment to post-sale feedback, regaining the right to ordering and pricing and reducing the risk of the middleman earning a price difference in sales. Rural e-commerce has the advantages of transcending time and space limitations, low comprehensive costs, novel transaction methods, etc. Both migrant workers and farmers engaged in traditional agricultural production can learn Internet skills and related operational knowledge through e-commerce platforms, which is conducive to improving the level of human capital of farmers, and they can also participate in production and sales through the information network. This shows that “Internet—farmers” has become a new way to solve family business problems [13,14]. Based on the above analysis, this study proposes:

Hypothesis 1. 

Rural e-commerce development has a positive and significant impact on farm household income.

2.2. Rural e-Commerce and Intra-Rural Income Disparities

The formation of intra-rural income disparities is the result of a combination of multidimensional differentiation. Some scholars believe that the differences in the capital endowment of rural households are the direct cause of the phenomenon of differences in the income-generating effect of rural e-commerce, and that the heterogeneity of the e-commerce participation behavior of rural households is a direct factor leading to the expansion of the income gap among rural households [15,16]. All of the above studies measure the impact of rural e-commerce on intra-rural income disparity from the perspective of e-commerce participation behavior. Specifically, rural e-commerce platforms provide farmers with online channels for the sale of agricultural products, breaking the geographical limitations of traditional sales methods. At the same time, they can provide farmers with timely and accurate market information in order to make timely adjustments to production plans and sales strategies. It is undeniable that the capital endowment of farmers is one of the main factors leading to the differences in the sharing of digital dividends among farmers, and that the capital differences have a direct impact on production capacity, decision-making behavior, market participation, and the level of household income of farmers, and that farmers participating in e-commerce receive a higher return on their agricultural income than non-e-commerce households.

Along with the improvement of Internet infrastructure and the sinking of e-commerce platforms, the improvement of information technology construction and the upgrading of industrial structures in rural areas follow. On the one hand, the development of rural e-commerce has led to the development of rural logistics, express delivery, and other industries, opening the way for agricultural products to increase and industrial products to decrease. In the context of the reality that part-time farmers are common in China, rural e-commerce can drive more farmers to participate in the rural e-commerce industry chain and promote the employment of farmers in the vicinity to increase their income. On the other hand, e-commerce in the countryside has attracted more social capital investment in rural areas. The investment of capital, technology, and human resources in the countryside not only helps to improve the quality of agricultural products and increase the value added but also promotes the upgrading of the industrial structure and the diversification of traditional agricultural production. The processing of agricultural products, rural tourism, etc., helps to raise the level of income of the farmers’ families. At the same time, the increase in the sales of agricultural products and the broadening of sales channels has, in turn, promoted the process of large-scale agricultural operation and land transfer, providing a solution to the problem of “who will farm the land” caused by the hollowing out of the countryside and integrating the use of idle land resources in rural areas through large-scale operation, which not only reduces the cost of production and improves the efficiency of land output but also allows farmers to obtain stable property income and promotes the improvement of farm household income. As a result, farmers can also obtain stable property income, which promotes an increase in the level of their household income. In addition, through the e-commerce platform, farmers can come into contact with more knowledge and new production management methods, and the enhancement of human capital can help to increase the income level of farmers, thus reducing the internal income gap in rural areas.

Hypothesis 2. 

The development of rural e-commerce has a positive and significant effect on reducing the internal rural income gap.

3. Research Design

3.1. Data Sources

The data in this article come from the China Rural Revitalization Survey (CRRS), a comprehensive survey of agricultural production, rural development, farmers’ livelihoods, and social well-being initiated by the Institute of Rural Development of the Chinese Academy of Social Sciences (CASS) in 2020. The data were drawn by a combination of stratified and random sampling, with 10 sample provinces drawn from the eastern, central, western, and northeastern regions in the proportion of one-third of the number of provinces in the subregion, including 10 provinces in Guangdong, Zhejiang, Shandong, and Anhui. Secondly, the project team carried out equidistant grouping based on GDP per capita, i.e., all counties (cities and districts) were divided into five groups according to the level of GDP per capita, and one county (city and district) and three townships were randomly selected from each group. Administrative villages within the grouping were divided into two categories, namely “better” and “worse”, and one village was randomly selected according to its situation of economic development. According to the economic development, administrative villages in the subgroup were divided into “better” and “worse” categories and one village was randomly selected. Finally, 12–14 rural households were randomly selected from the roster of rural households in the administrative villages using the equidistant sampling method to carry out the research.

3.2. Description of Variables

Explained variables: to overcome the advantages due to household size in terms of total household income, the main explanatory variable used in the econometric analysis of this paper is the annual per capita income of farming households in 2019, which can reflect the overall household income. In order to reduce the heteroskedasticity interference of the income variable, this paper transforms the income variable by natural logarithm.

Explanatory variables: the core research object of this paper is whether farmers participate in e-commerce. Using the question “whether your family operates products traded through the network” as a sample variable, if the answer is “yes”, the value will be assigned to 1; otherwise, it will be zero.

Control variables: In order to alleviate the omitted variable bias, this paper refers to the existing literature to add a series of control variables that may affect the household income of farmers, which mainly contains three levels, namely individual characteristics, household characteristics, and village characteristics. Among them, individual characteristics include the respondent’s age, the age-squared term, and education level; household characteristics include household size, the average age of household members, whether there is a village cadre in the household, and the highest education level of household members; and village characteristics include village transport conditions, distance between the village and county government, village e-commerce infrastructure, and the village’s economic conditions.

This paper focuses on the relationship between farmers’ e-commerce participation behavior and farmers’ household income, deals with the relevant e-commerce, income level, and control variables, removes missing values and outliers, shrinks the upper and lower 0.5% of the total household income, and finally obtains 2910 samples of farmers. Descriptive statistics of specific variables are shown in

Table 1. Descriptive statistics of key variables.

Variables	Define	Observations	Mean	Variance
Ecom	Participation in e-commerce: yes = 1, no = 0	2910	9.382609	1.062979
Lnperincome	Per capita net income of farm household (CNY), taken in logarithms	2910	1.067354	0.2506771
Age	Age	2910	55.54639	11.09463
Agesquare	Age squared divided by 100	2910	32.0845	12.42666
Edu	Educational attainment: no schooling = 1; primary school = 2; junior high school = 3; senior high school = 4; junior college = 5; vocational high school and technical school = 6; specialist = 7; undergraduate = 8; postgraduate = 9	2910	2.762887	1.078939
Famsize	Number of family members	2910	3.99244	1.54522
Mage	Average age of household members	2910	43.24033	13.36191
Maxserv	Whether there are village cadres in the household: Yes = 1, No = 0	2910	1.724055	1.417784
Maxedu	Highest level of education of household members: No schooling = 1; Primary school = 2; Junior high school = 3; High school = 4; Secondary school = 5; Vocational high school = 6; Specialty = 7; Undergraduate school = 8; Postgraduate school = 9	2910	4.635395	2.162698
Transport	Village traffic conditions: whether the roads between the village and the group are hardened roads or not	2910	1.057045	0.2319678
Ecomconditions	Village e-commerce infrastructure: whether there is an e-commerce service station or product resale point in the village	2910	1.471478	0.4992716
Distance	Distance from the village committee to the county government (kilometers)	2910	23.78271	17.36853
Lngdp	Village economic conditions: per capita disposable income of the village in 2019 (take logarithm)	2910	9.432598	0.5431103

.

3.3. Model Setting

In order to test Hypothesis 1, i.e., to determine whether participation in e-commerce has an income-generating effect, this paper constructs a model, as shown below:

$${Lnperincome}_i=cons+\alpha {Ecom}_i+\beta {Controls}_i+{\epsilon }_i$$

(1)

In Equation (1), ${Lnperincome}_i$ represents the level of per capita income of farm household i and ${Ecom}_i$ represents the e-commerce participation of farm household i. If farm household i participates in e-commerce, then $Ecom=1$; otherwise, $Ecom=0$; ${Controls}_i$ is a set of control variables and ${\epsilon }_i$ is a random perturbation term. If α passes the significance test and is positive, it means that the participation of farming households in e-commerce helps to promote the level of household income, and Hypothesis 1 is proved.

Second, in order to further explore the pattern of the impact of participation in e-commerce on samples with different income levels, this paper selects a quantile regression model to determine whether rural e-commerce helps to reduce the income gap within farm households. The quantile regression model is an extension of the ordinary least squares method, which regresses the dependent variable Y on the independent variable X according to the different divisions of the dependent variable Y into points, and is able to describe the full picture of the conditional distribution of the dependent variable more comprehensively. Compared with OLS, the quantile regressions had different regression coefficients at different quartiles. Therefore, quantile regression assumes that each explanatory variable has a distribution over the values of the explanatory variables, which can be expressed as a series of quantiles, and the estimation results are more robust to outliers and help to explain the impact of participation in rural e-commerce on farm households at different income levels. The quantile regression model is shown below:

$${Quant}_{\theta }\left({Lnperincome}_i|{Ecom}_i\right)=cons+{\alpha }_{\theta }{Ecom}_{i\theta }+{\beta }_{\theta }{Controls}_{i\theta }+{\epsilon }_{i\theta }$$

(2)

In the Formula (2), $\theta$ denotes different points, ${Quant}_{\theta }\left({Lnperincome}_i|{Ecom}_i\right)$ denotes the per capita income level of the farming family in the $\theta$ quartile, and the other symbols have the same meanings as in Formula (1).

4. Empirical Results and Analyses

4.1. Benchmark Regression Results

This study used a benchmark regression to estimate the impact of farmers’ participation in e-commerce on their household income, and the regression results are shown in

Table 2. Rural e-commerce and farm household income: basic regression.

Variables	Lnperincome	Lnperincome	Lnperincome	Lnperincome
	(1)	(2)	(3)	(4)
Ecom	0.700 ***	0.582 ***	0.568 ***	0.516 ***
	(0.0837)	(0.082)	(0.081)	(0.080)
Gender		0.154 **	0.083	0.061
		(0.076)	(0.075)	(0.074)
Age		0.030 **	0.015	0.014
		(0.013)	(0.013)	(0.013)
Agesquare		−0.041 ***	−0.027 **	−0.026 **
		(0.012)	(0.012)	(0.011)
Edu		0.100 ***	0.055 ***	0.041**
		(0.019)	(0.019)	(0.019)
Famsize			−0.111 ***	−0.105 ***
			(0.017)	(0.016)
Mage			−0.003	−0.003
			(0.002)	(0.002)
Maxserv			0.055 ***	0.046 ***
			(0.013)	(0.013)
Maxedu			0.060 ***	0.047 ***
			(0.010)	(0.010)
Transport				−0.218 ***
				(0.081)
Ecomconditions				0.060
				(0.037)
Distance				0.001
				(0.001)
Lngdp				0.354 ***
				(0.035)
Constant	0.342888	8.550 ***	9.292 ***	6.272 ***
	(0.0201)	(0.382)	(0.393)	(0.522)
Observations	2910	2910	2910	2910
R-squared	0.023	0.073	0.102	0.139

Note: **, and *** indicate significance at the 5 percent, and 1 percent levels, respectively, and values in parentheses are standard errors.

. In order to ensure the robustness of model estimation, this study sequentially included the individual characteristics of the head of the household, the characteristics of the farmer’s family, and the characteristics of the village into the benchmark regression model, and the results of the benchmark regression of Equation (1) are demonstrated in Table 2. (1) With the addition of farmers’ household characteristics and other control variables at the end of the column, the regression results show that the behavior of participation in e-commerce is significant at the 1% level and the coefficient value is 0.700. The results in columns (2)–(4) indicate that the income-enhancing effect of participation in e-commerce is significantly positive. The above results indicate that farmers’ participation in e-commerce can effectively promote their household income level, and Hypothesis 1 of this paper is verified. In terms of control variables, the effect of the square term of age on the income of farm households is significantly negative, indicating that along with the increase in age, the level of income of farm households decreases. The coefficients of education level and the highest education level of the household are significantly positive, and higher human capital is one of the key factors contributing to the increase in the income of farming households. The coefficient of village economic conditions is significantly positive, and a relatively developed village economy is associated with higher levels of farm household income.

4.2. Endogeneity Treatment

Although the baseline regression in the previous section incorporates household characteristics and other control variables, there are still other potential factors that may affect household income and measurement error may cause the random disturbance term to incorporate the unobserved portion of the explanatory variables, thus creating an endogeneity problem. Therefore, this study adopts the instrumental variable method to correct the possible endogeneity of the baseline regression, which requires that the instrumental variable only acts on the dependent variable through the endogenous variable. Farmers’ e-commerce participation behavior may be affected by the spillover effect of the e-commerce participation behavior of other farmers in the surrounding area, but this spillover effect will not have a direct impact on farmers’ household income, so this paper chooses the “proportion of other e-commerce households in the same village” as the instrumental variable regression results, as shown in

Table 3. Rural e-commerce and farm household income: linear model 2SLS.

Variables	Lnperincome
	Phase I	Phase II
Ecom		1.077 *** (0.475)
Instrumental variables IV	0.441 *** (0.070)
Control variables	Controlled	Controlled
F-statistic	110.53
DWH test χ2	6.68
p-value	0.0097
Observations	2910

Note: *** indicates significance at the 1 percent level, and values in parentheses are standard errors.

. From the regression results of the first stage, it can be seen that the proportion of other e-commerce households in the same village is highly correlated with whether farmers participate in e-commerce or not, and the F-value is 110.528, which is greater than the critical empirical value of 10. For robustness, the preferred information maximum likelihood method, which is more insensitive to weak instrumental variables, is used, and the estimation results show that the coefficient estimates are very close to the 2SLS, which side-steps the rejection of the weak instrumental variable hypothesis. The results of the Durbin–Wu–Hausman test (DWH test) to test for endogeneity are shown in the table below.

This study further adopts the propensity score matching method for the correction of selectivity bias and verifies the stability of the matching results, and the results are shown below.

Table 4. Rural e-commerce and farm household income: average treatment effects.

Matching Method	Intervention Group	Control Group	Mean Treatment Effect
One-to-one matching	10.042	9.746	0.295 ***
k-nearest-neighbor matching	10.040	9.741	0.300 ***
Radius Matching	10.040	9.458	0.582 ***
Kernel Matching	10.040	9.466	0.574 ***

Note: *** indicates significance at the 1 percent level, and values in parentheses are standard errors.

shows the average treatment effect of rural e-commerce on income, measured by the one-to-one matching and k-nearest-neighbor matching methods, respectively, and the results verify the previous conclusions and show that the estimation results have a high degree of consistency, indicating that the findings of this paper are highly consistent and robust, i.e., the participation of farmers in e-commerce has a significant promotion effect on their household income.

4.3. Robustness Test

In order to test the robustness of the above results, this paper takes the approach of replacing the variables and replacing the model for the robustness test. Firstly, the core independent variable is replaced with the online sales (Sales) of farm households in 2019. The results in

Table 5. Robustness test: substitution of variables.

Matching Method	Lnperincome			lnPerOperation
Sales	0.0288 *** (0.00515)	0.0292 *** (0.00507)	0.0272 *** (0.00498)
Ecom				1.065 *** (0.151)	1.038 *** (0.150)	0.970 *** (0.148)
Head of Household Controls	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Household Controls		Controlled	Controlled		Controlled	Controlled
Village Controls			Controlled			Controlled
Constant	8.800 *** (0.367)	8.800 *** (0.375)	8.800 *** (0.513)	8.915 *** (0.735)	9.715 *** (0.760)	6.529 *** (1.020)
Observations	2910	2910	2910	1770	1770	1770
R-squared	10.040	9.466	0.574 ***	0.079	0.079	0.079

Note: *** indicates significance at the 1 percent level, and values in parentheses are standard errors.

show that there is no significant change in the sign and significance level of the core independent variables, and the results after replacing the core independent variables are consistent with the results of the benchmark model. Secondly, since the income from e-commerce business is operating income, this paper replaces the dependent variable with the per capita operating income (lnPerOperation) of the farming household for the regression. The regression results show that the sign and significance of the key variables are consistent with the previously concluded benchmark model. The above analysis shows that the estimation results in this paper are robust and reliable.

Second, this paper adopts the endogenous switching regression (ESR) model for robustness testing. The results in

Table 6. Robustness test: replacement model.

Type of Farm Household	Treatment Effect
	ATT	ATU
Participating e-commerce farmers	10.042	9.746
Non-participating e-commerce farmers	10.040	9.466

show that the per capita household income of e-commerce farmers will decrease by 53.09% when they are not involved in e-commerce, considering the counterfactual assumptions. When non-e-commerce farmers participate in e-commerce, the per capita household income will increase by 17.48%. This suggests that both e-commerce and non-e-commerce households are able to promote household income levels through participation in rural e-commerce, consistent with the findings of the previous benchmark regression model. The above analysis shows that the estimation results of this paper are robust and reliable.

4.4. Quantile Regression

From the previous estimation results, it can be seen that rural e-commerce has a significant impact on the improvement of household income level but the baseline regression model is unable to reflect the distribution pattern of the impact of each explanatory variable on different income levels. Therefore, this study adopts the quantile regression method in order to comprehensively and deeply analyze the distribution dynamics of rural households with different income levels. The quantile regression method can further interpret how farmers’ e-commerce participation behavior affects the median and quartile of farmers’ household income on the basis of the benchmark regression. The quantile regression results, as shown in

Table 7. Rural e-commerce and farm household income: baseline regression and quantile regression results.

Variables	Lnperincome
	10%	30%	50%	70%	90%	Lnperincome
Ecom	0.742 ***	0.505 ***	0.415 ***	0.443 ***	0.518 ***	0.516 ***
	(−0.188)	(−0.109)	(−0.0909)	(−0.0747)	(−0.134)	−0.0744
Gender	−0.0785	0.0515	0.175 **	0.141 **	0.0762	0.0606
	(−0.175)	(−0.101)	(−0.0842)	(−0.0693)	(−0.125)	−0.0802
Age	−0.00535	0.0103	0.00599	0.0168	0.0454 **	0.014
	(−0.0301)	(−0.0174)	(−0.0145)	(−0.012)	(−0.0215)	−0.0115
Agesquare	−0.0162	−0.0243	−0.0161	−0.0231 **	−0.0519 ***	−0.0262 **
	(−0.0269)	(−0.0156)	(−0.013)	(−0.0107)	(−0.0192)	−0.0102
Edu	−0.00117	0.0405	0.0519 **	0.0524 ***	0.0593 *	0.0412 **
	(−0.0453)	(−0.0262)	(−0.0218)	(−0.018)	(−0.0323)	−0.0194
Famsize	−0.145 ***	−0.125 ***	−0.109 ***	−0.0929 ***	−0.104 ***	−0.105 ***
	(−0.0383)	(−0.0222)	(−0.0185)	(−0.0152)	(−0.0274)	−0.0166
Mage	0.00233	−0.00562*	−0.00497*	−0.00759 ***	−0.00651	−0.00328
	(−0.0057)	(−0.0033)	(−0.00275)	(−0.00226)	(−0.00407)	−0.00226
Maxserv	0.0771 **	0.0588 ***	0.0471 ***	0.0380 ***	0.0262	0.0462 ***
	(−0.031)	(−0.018)	(−0.015)	(−0.0123)	(−0.0222)	−0.0127
Maxedu	0.0338	0.0405 ***	0.0509 ***	0.0519 ***	0.0629 ***	0.0466 ***
	(−0.023)	(−0.0133)	(−0.0111)	(−0.00913)	(−0.0164)	−0.01
Transport	−0.172	−0.264 **	−0.185**	−0.148 *	−0.144	−0.218 ***
	(−0.19)	(−0.11)	(−0.0918)	(−0.0755)	(−0.136)	−0.0825
Ecomconditions	−12	0.0857 *	0.0821*	0.0982 ***	0.114 *	0.0596
	(−0.0871)	(−0.0504)	(−0.042)	(−0.0346)	(−0.0622)	−0.0373
Distance	0.000957	0.000245	0.00124	0.00261 **	0.000807	0.00144
	(−0.00256)	(−0.00148)	(−0.00124)	(−0.00102)	(−0.00183)	−0.00105
Lngdp	0.266 ***	0.325 ***	0.370 ***	0.392 ***	0.433 ***	0.354 ***
	(−0.0834)	(−0.0483)	(−0.0403)	(−0.0331)	(−0.0596)	−0.039
Constant	6.878 ***	6.498 ***	6.210 ***	6.028 ***	5.614 ***	6.272 ***
	(−1.229)	(−0.712)	(−0.593)	(−0.488)	(−0.878)	−0.512
Observations	2910	2910	2910	2910	2910	2910
R-squared						0.139

Note: *, **, and *** indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively, and values in parentheses are standard errors.

, show that the impact of rural e-commerce on farm household families with different income levels shows some variability. Specifically, the impact of different explanatory variables or the same explanatory variable on the income of farm household families at different income quartiles is different.

In this study, different income levels are represented by divisions based on quartile sizes: 0.1, 0.3, 0.5, 0.7, and 0.9 quartiles represent the income levels of five farming households: low-income, lower-middle-income, middle-income, higher-middle-income, and high-income, respectively, and the self-help sampling method (with 400 repetitions of the sample) is used to calculate the standard errors, weakening the unknown disturbances of the error term and increasing the validity of the estimation. Table 7 shows the quantile regression results for different income levels of farm households. The regression results show that rural e-commerce has a significant quantile effect on the level of household income of farm households, and all of them show a positive effect and pass the test of significance at the 1 percent level. It shows that rural e-commerce has a significant income-generating effect for farm households at all income levels, with the highest coefficient of 0.742 for the e-commerce participation behavior of farm households with low income levels and the lowest coefficient of 0.415 for the e-commerce participation behavior of households with middle income levels. Overall, the coefficients of e-commerce participation behaviors show a “U-shaped” trend at different quartiles, indicating that the impact of rural e-commerce on rural household incomes shows a gradual weakening and then a gradual strengthening trend. Specifically, for rural households with a middle income level or below, the effect of participation in e-commerce on their household income level decreases as their income rises, while for rural households with a higher-middle or high income level, the effect of their income-enhancing behavior becomes more pronounced as their income level rises. In summary, it can be seen that rural e-commerce has a more significant impact on farm households with low income levels, an average impact on lower-middle, higher-middle, and high incomes, and the smallest impact on middle-income farm household families. In addition, among the control variables, household size and the village’s economic conditions have a significant effect on the impact of farm household families at different income levels. In summary, rural e-commerce has income distribution effects and helps to reduce the internal income gap in rural areas, and Hypothesis 2 of this paper is proven.

5. Further Analyses

5.1. Heterogeneity Test

The previous analysis has confirmed that rural e-commerce has a significant income-generating effect, so this part will analyze the heterogeneity from two levels: regional heterogeneity and farm household heterogeneity.

5.1.1. Based on Farm Household Heterogeneity

In terms of farm household heterogeneity, this study will analyze the four dimensions of human capital, physical capital, social capital, and financial capital.

(1)

Based on human capital

Studies have found that the quantity and quality of the labor force is the major determinant of household income in farming households and that an increase in the level of human capital has a positive effect on increasing the income of rural residents [17,18]. Referring to the studies of Yin Zhichao et al., Qin Fang, Luo Qianfeng, etc., this study adopts the highest level of education of family households to measure the human capital stock of farm households [9,11,19]. Specifically, the farm household families are divided into three categories: a low-education group (junior high school and below), a middle-education group (high school and above but below specialized), and a high-education group (university specialized and above). At the same time, the variable of the household’s highest level of education is excluded from the regression model, and the quantile regression results are shown in

Table 8. Heterogeneity analysis of income-generating effects of rural e-commerce: human capital of farm households.

	Variables	Lnperincome
		10%	30%	50%	70%	90%	Lnperincome
Lower-Education Group	Ecom	0.820 **	0.735 ***	0.461 **	0.557 ***	0.339	0.644 ***
		(−0.327)	(−0.23)	(−0.189)	(−0.151)	(−0.276)	(−0.162)
	Controls	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
	Constant	6.858 ***	6.722 ***	6.489 ***	6.323 ***	7.092 ***	6.368 ***
		(−1.616)	(−1.133)	(−0.934)	(−0.745)	(−1.361)	(−0.8)
	Observations	1214	1214	1214	1214	1214	1214
	R-squared						0.118
Middle-Education Group	Ecom	0.285	0.404 **	0.348 **	0.226 *	0.398 **	0.404 ***
		(−0.298)	(−0.182)	(−0.167)	(−0.129)	(−0.202)	(−0.139)
	Controls	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
	Constant	9.056 ***	8.766 ***	6.850 ***	7.032 ***	6.723 ***	7.429 ***
		(−2.277)	(−1.392)	(−1.274)	(−0.988)	(−1.545)	(−1.064)
	Observations	808	808	808	808	808	808
	R-squared						0.099
Higher-Education Group	Ecom	0.701 **	0.493 ***	0.439 ***	0.484 ***	0.593 ***	0.540 ***
		(−0.348)	(−0.178)	(−0.149)	(−0.147)	(−0.196)	(−0.126)
	Controls	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
	Constant	4.860 *	6.358 ***	6.730 ***	6.611 ***	5.940 ***	6.682 ***
		(−2.866)	(−1.466)	(−1.224)	(−1.205)	(−1.615)	(−1.035)
	Observations	888	888	888	888	888	888
	R-squared						0.155

Note: *, **, and *** indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively, and values in parentheses are standard errors.

. According to the results of the quantile analysis in Table 8, rural e-commerce significantly contributes to the household income of farmers in the low-, medium-, and high-education groups. Analyzing the quartile regression of different education levels, we found that the regression coefficients of rural e-commerce for both the low- and high-education groups show a U-shaped trend of “decreasing first and then increasing later”, and the effect of increasing income on low-income farming households is the most obvious (β = 0.820, p < 0.05), showing that rural e-commerce has a more obvious effect of “raising the low” for low-income farming households; β = 0.701, p < 0.05. It can be seen that the rural electric power of low-income farm families produced by the “lower” effect is more obvious. In contrast, the income-generating effect of rural e-commerce on middle-income and high-income households in the secondary education group is more significant than on low-income households, and the income-generating effect of e-commerce in the secondary education group has a stronger “middle-expansion” effect.

(2)

Based on Physical Capital

Physical capital is one of the factors affecting farm household income and income inequality. Therefore, this paper refers to the study of Zhao Lijuan et al., which uses the household productive land indicator to study the relationship between physical capital and the effect of rural e-commerce on income generation [20]. Specifically, using the household sown crop area as a proxy variable for physical capital, farmers were divided into two groups based on their 2019 household sown crop area, with those below the median being the low-physical-capital group and those above the median being the high-physical-capital group, and the results of the regression are shown in

Table 9. Heterogeneity analysis of income-generating effects of rural e-commerce: farmers’ household physical capital.

	Variables	Lnperincome
		10%	30%	50%	70%	90%	Lnperincome
Low-Physical-Capital Group	Ecom	0.666 ***	0.518 ***	0.383 ***	0.404 ***	0.544 ***	0.511 ***
		(−0.226)	(−0.136)	(−0.102)	(−0.101)	(−0.159)	(−0.1)
	Controls	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
	Constant	6.640 ***	5.721 ***	5.178 ***	4.924 ***	4.850 ***	5.570 ***
		(−1.635)	(−0.985)	(−0.739)	(−0.733)	(−1.148)	(−0.723)
	Observations	1454	1454	1454	1454	1454	1454
	R-squared						0.189
High-physical-capital group	Ecom	0.708 **	0.555 ***	0.503 ***	0.519 ***	0.488 *	0.531 ***
		(−0.312)	(−0.191)	(−0.153)	(−0.122)	(−0.258)	(−0.134)
	Controls	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
	Constant	7.317 ***	7.311 ***	7.945 ***	8.114 ***	7.966 ***	7.300 ***
		(−1.804)	(−1.104)	(−0.884)	(−0.703)	(−1.492)	(−0.775)
	Observations	1456	1456	1456	1456	1456	1456
	R-squared						0.096

Note: *, **, and *** indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively, and values in parentheses are standard errors.

. The quantile regression results show that the e-commerce participation coefficients of both the low- and high-material-capital groups show a “U-shaped” trend of decreasing and then increasing, and the e-commerce participation coefficients of the low-income group are higher than those of the high-income group, so it can be seen that rural e-commerce has the most significant income-generating effect on low-income households (β = 0.666; β = 0.708, p < 0.05; 0.666, p < 0.01; β = 0.708, p < 0.05). It is worth mentioning that low-income households in the high-physical-capital group’s participation in e-commerce contribute significantly more to income enhancement than high-income households (β = 0.708, p < 0.01; β = 0.0.488, p < 0.1), which shows that rural e-commerce narrows the intra-rural income gap among different income groups.

(3)

Based on social capital

Many scholars have pointed out that the social relationship is an important factor affecting the income of farming households [21,22,23]. Therefore, this paper refers to Yang Yi’s study and chooses the indicator of “how many relatives and friends do you have who can borrow money” to measure the social capital of farming households [24]. Farm households were categorized into two groups based on their social ties in 2019, with those below the median being the low-social-capital group and those above the median being the high-social-capital group, and the quantile regression results are shown in

Table 10. Heterogeneity analysis of the income-generating effects of rural e-commerce: farmers’ household social capital.

	Variables	Lnperincome
		10%	30%	50%	70%	90%	Lnperincome
Low-social-capital group	Ecom	0.500 *	0.286	0.383 **	0.256 *	0.408 *	0.339 **
		(−0.261)	(−0.196)	(−0.177)	(−0.131)	(−0.238)	(−0.143)
	Controls	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
	Constant	6.922 ***	6.268 ***	6.348 ***	5.751 ***	6.982 ***	6.108 ***
		(−1.379)	(−1.035)	(−0.934)	(−0.694)	(−1.258)	(−0.755)
	Observations	1427	1427	1427	1427	1427	1427
	R-squared						0.076
High-social-capital group	Ecom	0.500 **	0.492 ***	0.394 ***	0.469 ***	0.450 ***	0.511 ***
		(−0.194)	(−0.118)	(−0.098)	(−0.108)	(−0.163)	(−0.095)
	Controls	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
	Constant	7.486 ***	7.917 ***	7.268 ***	6.615 ***	6.838 ***	7.236 ***
		(−1.47)	(−0.894)	(−0.742)	(−0.818)	(−1.231)	(−0.719)
	Observations	1483	1483	1483	1483	1483	1483
	R-squared						0.155

Note: *, **, and *** indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively, and values in parentheses are standard errors.

. The regression results show that compared with the high-social-capital group, the income enhancement effect of rural e-commerce on the low-social-capital group is not significant, except for middle-income farm household families, which may be due to the fact that firstly, this paper only retains the respondent’s information and data during data cleansing, and some of the information filled in by the respondents may only represent the individual’s social relationship, which reduces the impact of the social relationship on the household income level. Secondly, the farm household’s choice of participation in e-commerce is easily influenced by the e-commerce participation of other people around them, and farm households with fewer social relationships weaken the spillover effect of the e-commerce participation of people around them to a certain extent. As for the high-social-capital group, the coefficient of participation in e-commerce for farm families with different income levels shows a U-shaped potential in the range of 0.394–0.5, indicating that the income-boosting effect of participation in e-commerce in the high-social-capital group is not significantly different for families with different income levels. Overall, rural e-commerce has a more significant income-boosting effect on farm households in the low-income group, regardless of whether the high- or low-social-capital group is involved.

(4)

Distinguishing financial capital

The accumulation of financial capital is conducive to improving the degree of coordination between the various capitals of the household, promoting farmers to make different behavioral decisions and thus raising the level of household income [25]. Therefore, this study refers to the study of Luo Qianfeng [14] and chooses the indicator of “the amount of loan applied for and actually obtained by your household in 2019” to measure the financial capital of farming households, classifying farming households into two groups according to the amount of loans obtained in 2019. Those below the median are the low-financial-capital group, and those above the median are the high-financial-capital group. The regression results are shown in

Table 11. Heterogeneity analysis of income-generating effects of rural e-commerce: financial capital of farm households.

	Variables	Lnperincome
		10%	30%	50%	70%	90%	Lnperincome
Low-financial-capital group	Ecom	0.426	0.116	0.495 *	0.216	−0.0231	0.255
		(−0.521)	(−0.334)	(−0.266)	(−0.211)	(−0.326)	(−0.223)
	Controls	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
	Constant	−0.367	5.465 **	4.847 **	6.397 ***	4.483 *	3.787 **
		(−4.124)	(−2.638)	(−2.102)	(−1.668)	(−2.575)	(−1.762)
	Observations	404	404	404	404	404	404
	R-squared						0.106
High-financial-capital group	Ecom	0.724 ***	0.541 ***	0.426 ***	0.489 ***	0.528 ***	0.542 ***
		(−0.193)	(−0.114)	(−0.096)	(−0.087)	(−0.154)	(−0.086)
	Controls	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
	Constant	7.912 ***	7.092 ***	6.542 ***	5.992 ***	6.253 ***	6.546 ***
		(−1.233)	(−0.727)	(−0.616)	(−0.553)	(−0.985)	(−0.548)
	Observations	2506	2506	2506	2506	2506	2506
	R-squared						0.148

Note: *, **, and *** indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively, and values in parentheses are standard errors.

. As can be seen from the regression results, the income-generating effect of rural e-commerce is not significant for farm households with low financial capital, and the coefficient of e-commerce participation at the 0.9 quantile is negative, indicating that insufficient financial capital stock inhibits the income-generating effect of participation in e-commerce for farm households with high levels of income, which is in line with the previous conclusion. For high-financial-capital households, the coefficient of e-commerce participation still shows a trend of “decreasing and then increasing” with the increase in the income level of farming households, and the overall trend is decreasing. The effect of rural e-commerce on low-income households is more pronounced than on high-income households (β = 0.724, p < 0.01), confirming Hypothesis 2.

5.1.2. Analysis Based on Regional Heterogeneity

Unbalanced regional economic development has been the focus of academic attention over the years, but scholars have paid more attention to the East–West gap, involving fewer North–South issues. Since the emergence of rural e-commerce, there has been a clear difference between “strong in the south and weak in the north”. By the end of 2021, there were 3590 Taobao villages in the southern region and only 1019 Taobao villages in the northern region in the sample area. First of all, in the southern region, the level of economic development of the southeastern coastal region is in the country’s leading position. Although the economic foundation of the southwest region is relatively weak, the overall industrial development of the northern region is still relatively backward. Secondly, the natural geography of the southern region is also significantly better than the northern region. The southern region has a favorable climate, adequate heat, and abundant precipitation, and these conditions provide a good foundation for the growth of a variety of crops. Therefore, this study further divides the national sample into two groups according to regions, with the southern region being south of the Qinling-Huaihe River line and the northern region being north of the line, to compare the impact of rural e-commerce on the level of household income of rural households at the regional level and the trends shown by the interquartile regression. From the results in

Table 12. Analysis of regional heterogeneity in rural income-generating effects.

	Variables	Lnperincome
		10%	30%	50%	70%	90%	Lnperincome
Southern Region	Ecom	0.984 ***	0.422 ***	0.462 ***	0.534 ***	0.590 ***	0.574 ***
		(−0.245)	(−0.146)	(−0.121)	(−0.099)	(−0.145)	(−0.11)
	Controls	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
	Constant	3.537 **	3.088 ***	4.514 ***	4.137 ***	2.679 ***	3.950 ***
		(−1.718)	(−1.022)	(−0.847)	(−0.69)	(−1.019)	(−0.772)
	Observations		1345	1345	1345	1345	1345
	R-squared						0.174
Northern Region	Ecom	0.422 *	0.454 ***	0.385 ***	0.415 ***	0.389 *	0.404 ***
		(−0.228)	(−0.162)	(−0.141)	(−0.118)	(−0.223)	(−0.118)
	Controls	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
	Constant	9.091 ***	8.588 ***	8.963 ***	9.889 ***	10.44 ***	8.806 ***
		(−1.421)	(−1.01)	(−0.879)	(−0.733)	(−1.388)	(−0.734)
	Observations		1565	1565	1565	1565	1565
	R-squared						0.124

Note: *, **, and *** indicate significance at the 10 percent, 5 percent, and 1 percent levels, respectively, and values in parentheses are standard errors.

, it can be seen that in both the southern and northern regions, rural e-commerce has a significant effect on the income level of rural households, which indicates that rural e-commerce can achieve a significant income-generating effect throughout the country. Further analyzing the quantile regression results for different regions, it is found that the regression coefficients in the southern region show a trend of “decreasing and then increasing”, with the most obvious effect on the income of low-income farming households (β = 0.984, p < 0.01). The southern region is economically active, with a strong focus on innovation and entrepreneurship and many business opportunities, while farmers in the southern region are more receptive to new business models and technologies. Low-income farmers may choose to participate in e-commerce to achieve a significant increase in their household income levels, but the income-generating effects of e-commerce participation may be slightly weakened as their income level rises. In contrast, rural e-commerce in the northern region showed no significant difference in the household income of farmers at different quartiles, with coefficients fluctuating in the range of 0.385–0.454. Overall, rural e-commerce helps to alleviate the intra-rural income gap.

5.2. Analysis of Impact Mechanisms

5.2.1. Weakening Market Information Asymmetry

The ability to obtain information represents the sensitivity of farmers to market information. Accurate access to market supply and demand, prices, and other information can enable farmers to more accurately judge the price trends of agricultural products and market demand before making appropriate decisions. The theory of information asymmetry points out that the distribution of information in the market between the two parties in the transaction is uneven, and the advantageous party is often able to quickly and accurately obtain the information it needs. In the rural market, farmers are often at a disadvantage because of the lack of market information, and this information asymmetry is one of the most important factors restricting the development of the rural economy and the improvement of the income level of farmers. The rise of rural e-commerce has broken the “information gap” between smallholder producers and consumers in large markets [26]. From the production side, the development of e-commerce platforms has enabled farmers to more accurately grasp key information such as product market demand and price trends, so as to adjust planting structures and sales strategies in a timely manner and improve the added value and competitiveness of agricultural products. From the sales side, the e-commerce platform, as an information-sharing platform, breaks through geographical constraints and effectively connects producers and consumers, and both buyers and sellers can quickly and accurately find matching information, effectively saving the cost of searching for information and the cost of time and enhancing the efficiency of product sales. In addition to this, short video platforms for livestreaming goods can visually present agricultural products in front of consumers, so that consumers understand in detail the process of the product, from production to sales, which will help to increase the purchasing power and repurchase rate of consumers. In addition, the development of the digital economy in rural areas has also promoted the construction of Internet infrastructure and the improvement of related platform systems in rural areas. On the one hand, the government and relevant departments have increased investment in the construction of Internet infrastructure in rural areas to ensure full network coverage, so that farmers can have fast and stable access to the Internet, providing a fast and convenient channel for timely access to market information. On the other hand, the improvement of the agricultural information service platform integrates agricultural information resources including price, demand, policy, etc., and at the same time carries out relevant training for groups of farmers, which helps to improve their ability to obtain information. It can be seen that the digital dividend brought about by rural e-commerce development reduces the internal income gap arising from unequal opportunities in rural areas and helps to promote income growth among farm households. The existing literature generally agrees that the improvement of information literacy in farm households has a significant income-enhancing effect. Information asymmetry has always been an obstacle between farmers and the market. The information transmission channels are not smooth, making it difficult for farmers to obtain accurate information in a timely manner. All these factors lead to the high risk of slow sales and losses faced by farmers. The improvement of information acquisition efficiency can directly reduce the information acquisition cost to farmers, prompting farmers to make timely and efficient agricultural production and management decisions based on real-time information, thus improving agricultural production efficiency, promoting the enhancement of farmers’ household income, and narrowing the income gap within the countryside. Tang points out that differences in farmers’ information acquisition ability will directly lead to differences in their sales behavior, which in turn affects the level of farmers’ household income. In addition to this, it has been pointed out that improved access to information can increase the level of household income through the social capital of farmers [27]. Drawing on Luo Qianfeng’s study, this study adopts the question “Please rank the top three daily average hours of mobile phone function use” to measure whether farmers will enhance their human capital level through the Internet [14]. According to the results in column (1) of

Table 13. Rural income-generating effects and farm household income: analysis of mechanisms.

Variables	Information	Link	Cost
Ecom	0.122 ***	−0.181 ***	0.214 ***
	(−0.021)	(−0.029)	(−0.035)
Controls	Controlled	Controlled	Controlled
Constant	−0.158	2.194 ***	0.273
	(−0.134)	(−0.216)	(−0.221)
Observations	2910	1520	2832
R-squared	0.064	0.042	0.124

Note: *** indicates significance at the 1 percent level, and values in parentheses are standard errors.

, the coefficient of e-commerce adoption is 0.122, which is significant at the 1% level, indicating that rural e-commerce can help farmers improve their information acquisition ability, weaken information asymmetry, and then obtain a higher level of income.

5.2.2. Shorten the Product Circulation Link

The circulation system of agricultural products is a series of complex economic activities from the pre-production end to the post-consumption end of the transfer process. On the one hand, a highly efficient agricultural product circulation system will reduce the value of agricultural products in the circulation of losses and information asymmetry under the probability of being weakened. The high cost and high market risk of the multi-link circulation model will directly lead to constant price increases in the agricultural products in the process, and the profit margins available to farmers will constantly be compressed. In contrast, the short-channel circulation model can directly realize production and marketing docking across the intermediary by the producer directly to the consumer, weakening the impact of market information asymmetry on the income of farmers, so that farmers can firmly grasp their profit in their own hands. E-commerce is a modern circulation model and its wide application in rural areas promotes the process of upstream agricultural products. The e-commerce platform, from the origin of direct sales, the online order, and the offline delivery of the form, not only avoids the risk of the intermediate on the price difference but also reduces the cost of circulation of the product. These cost reductions can be directly converted into an increase in the income of the farm family, and thus reduce the internal income gap in rural areas. On the other hand, the development of rural e-commerce has promoted the improvement of the cold chain logistics system. Agricultural products are fresh products, which are highly susceptible to loss and deterioration in the traditional circulation chain, resulting in the sale of agricultural products being limited to the neighboring areas. With the sinking of e-commerce, the cold chain logistics infrastructure in rural areas has been continuously improved, which significantly reduces the loss and deterioration rate of agricultural products in circulation. Under the protection of the cold chain logistics system, the sales channels and sales radius of agricultural products can be expanded to the whole country, and the sales volume is significantly increased. The development of cold chain logistics is of great significance in safeguarding the quality of agricultural products, improving their market competitiveness, reducing the risk of loss of agricultural products in the circulation chain, and significantly increasing the production and business income of farmers. The existing literature also generally agrees that the distribution chain of agricultural products significantly affects the household income of farm households. For example, Zeng Huimin et al. found that an increase in the efficiency of agricultural product distribution has a significant positive effect on farm household income [28]. Referring to the mechanism test method of Jiang Ting, if the development of rural e-commerce helps to shorten the circulation link of agricultural products, it can be side-stepped to show that the shortening of the circulation link is an important influence mechanism of rural e-commerce development to promote the increase in income of farmers’ households [29]. For this reason, this paper draws on the study of Song Ying et al., which used the question “What are the main intermediate stages that your products go through?” to measure the distribution stages that farmers need to go through in order to sell their agricultural products [30]. Column (2) of Table 13 demonstrates the impact of rural e-commerce on the distribution chain of agricultural products. The coefficient of Ecom is −0.181 and is significant at the 1% level, indicating that rural e-commerce contributes to the shortening of the distribution chain of the products, while short-channel distribution contributes to the lowering of the cost of the distribution, and contributes to the increase in the income of farming households at the level of the household.

5.2.3. Promote Agricultural Cost-Saving and Efficiency

The cost of production and operation is the combination of all factor inputs of farmers in the process of agricultural production, and the reduction of operating costs is one of the most important paths for the improvement of farmers’ income. Rural e-commerce optimizes the allocation of resources in rural areas by means of information technology, provides farmers with intensive and large-scale procurement and logistics services, improves production efficiency, and thus reduces agricultural production costs. On the one hand, as consumers, small farmers usually face the problems of limited purchasing channels for production materials and high purchasing costs, which lead to excessive cost inputs at the pre-production end. Through the e-commerce platform, farmers can skip the intermediary and go directly to the manufacturers to purchase agricultural production materials at wholesale prices. Centralized purchasing at preferential prices effectively saves production and operation costs. In addition to this, farmers can still choose to establish cooperative relationships with e-commerce enterprises, co-operatives, and so on, with enterprises and co-operatives providing farmers with production materials, technical guidance, and other support at the early production stages and then carrying out joint marketing activities at a later stage in order to reduce production and operating costs and marketing costs, promoting an increase in the level of income of farmers. On the other hand, rural e-commerce in the countryside promotes the construction of warehousing facilities and logistics channels in rural areas, improves the efficiency of resource utilization, reduces the production risks of small farmers due to their own limited factors, and reduces the business burden of small farmers, thus enhancing their production and business income. Relevant scholars also generally agree that high production and operation costs are an important factor restricting small farmers from increasing their incomes. For example, Sun Huachen et al. point out that Internet deepening is an important factor in optimizing the structure of production and business costs for farmers, raising the level of their income, and reducing the rural income gap [31]. Therefore, this study adopts the question “Do you buy agricultural products online” to measure whether farmers will buy production materials through online channels. Column (3) of Table 13 demonstrates the effect of rural e-commerce on whether farmers will choose the online method of purchasing agricultural materials. The coefficient of Ecom is 0.214 and is significant at the 1% level, indicating that rural e-commerce can significantly increase the probability that farmers will purchase production materials through online channels, reducing the cost of agricultural production, which in turn contributes to the increase in farmers’ income.

6. Conclusions and Insights

Rural e-commerce has become a driving force in promoting agricultural and rural development, increasing farm household income, and realizing common prosperity. Using data from a random sample of 2910 sample farm households from the 2021 China Rural Revitalization Comprehensive Survey Database, this study explores the impact of rural e-commerce development on farm household incomes and intra-rural income gaps using the micro-unit of the farm household as the object of study. The main findings are as follows. On the whole, rural e-commerce has the most significant income-generating effect on the income of rural households, and this conclusion remains valid after considering endogeneity and conducting several robustness tests. Among them, the most significant income-enhancing effect is on low-income farm household families, i.e., rural e-commerce has an income distribution effect, which helps to alleviate the phenomenon of intra-rural income inequality and reduce the intra-rural income gap. From the point of view of objective regional differences, rural e-commerce has the most significant effect on raising the income of low-income families in the southern region, and it has a more “low life” effect. From the point of view of family capital heterogeneity, the income-generating effect of rural e-commerce is most obvious for middle-income families in the education group and the high-physical-capital group, with a more “middle-expansion” effect, and the income-generating effect is most significant for low-income families in the high-financial-capital group, which helps to narrow the gap between rural areas. Further analysis found that the income-generating effect of rural e-commerce mainly comes from the improvement of farmers’ information acquisition ability, the shortening of product circulation links, and the reduction of production and operation costs, i.e., farmers purchasing agricultural products through online channels in the production stage in order to reduce production inputs, and then effectively connecting with consumers through the e-commerce platform in the later stage, which weakened the asymmetric nature of market information, reduced the circulation cost of the intermediary links, and effectively improved the farmers’ family income level.

Rural e-commerce empowers farmers to increase their family income and promotes common prosperity by providing theoretical support and a practical basis. Based on the above conclusions, this paper puts forward the following three recommendations.

First, e-commerce publicity and training for rural households should be strengthened, and a sound mechanism for narrowing the “digital divide” should be established. The real-time monitoring of its effects should be carried out. Improving the digital literacy of rural households is an important indicator of rural revitalization, and cross-sectoral collaboration is needed to implement policies into effect. The training of rural e-commerce-related skills should be strengthened to stimulate the endogenous momentum of economic development in rural areas. Rural e-commerce talent should be nurtured and incentive policies should be implemented for talent revitalization, so as to promote rural e-commerce as a stable and long-term means of increasing farmers’ incomes.

Second, establish and improve the supervision mechanism to promote the deep integration of rural e-commerce into rural areas, strengthen the popularization and publicity of e-commerce knowledge, enhance the promotion and application of e-commerce in rural areas, especially in the northern region, encourage leading enterprises, farmers’ cooperatives, and other new business entities to rely on e-commerce platforms, build and improve industrial chains according to their own strengths, and, in this way, drive the surrounding farming groups to increase their incomes, jointly pushing forward the rural e-commerce empowering rural revitalization. In addition, in conjunction with the diversified information service needs of rural areas, a big data platform for rural informatization has been constructed to expand the business scope of the online platform, strengthen its service functions, and build a comprehensive information service platform system for agriculture.

Thirdly, financial support should be increased. On the one hand, it has vigorously promoted the construction of Internet infrastructure and logistics and warehousing facilities in rural areas to provide a convenient and effective platform for farmers to access rural e-commerce, and to provide a guarantee for digitization in rural areas. On the other hand, improve the policy incentive system, focus on the advantageous resources tilted to rural areas, actively introduce capital, technology, finance, talent, and other resources to the countryside, cultivate rural e-commerce “leaders”, give full play to the leading and radiation-driven role of highly digitally literate people, and encourage them to establish cooperative relationships with traditional small farmers, especially low-income groups. It also guides traditional small farmers to embed themselves in the rural e-commerce industry chain to enjoy the digital dividend together, and realizes mutual benefits and win-win results through resource sharing and complementary advantages.