Does E-Commerce Construction Boost Farmers’ Incomes? Evidence from China

Abstract

Elevating farmers’ incomes is crucial for ensuring socio-economic stability, yet the current stagnation in income growth and the expanding divide between urban and rural incomes present formidable challenges. E-commerce offers a transformative solution by bridging these disparities and fostering sustainable growth. Utilizing county-level data from 2000 to 2020 in China, particularly focusing on “The Pilot Counties of Introducing E-commerce to Rural Areas”, this paper explores the impact of e-commerce construction on farmers’ income growth using the multiple difference method. The findings reveal that e-commerce development exerts a significant positive impact on the enhancement of farmers’ incomes. Mechanistic analyses indicate that the driving effect of the comprehensive demonstration county policy of e-commerce in rural areas on farmers’ incomes is mainly realized through the channels of expanding market accessibility, promoting the advanced transformation of industrial structure, and optimizing the conditions of information infrastructure. Moreover, the efficacy of e-commerce policies in income augmentation is more pronounced in regions characterized by higher levels of human capital and substantial scale effects. This research offers valuable insights for continuously and effectively implementing the “Comprehensive Demonstration of E-commerce in Rural Areas” policy, which are crucial for exploring ways and mechanisms to boost farmers’ incomes in developing countries.

1. Introduction

In the current era of globalization, enhancing the income levels of farmers has emerged as a pivotal concern for the international community. Farmers, as primary producers representing vital social stratum worldwide, play a crucial role in addressing some of the most pressing global challenges, including food security, poverty alleviation, and economic growth. Despite their significance, farmers in numerous nations continue to face relatively low income levels, a situation that not only impinges upon their quality of life but also hampers the modernization of agriculture and the sustainable advancement of the global economy. For instance, as India was poised to join the ranks of the trillion-dollar economies, its agricultural sector, which forms the backbone of the national economy, was grappling with severe adversities: the suicide rates among farmers were alarmingly high, and their per capita incomes were experiencing a consistent decline. This scenario underscores the urgency of identifying and implementing efficacious strategies to bolster farmers’ earnings [1]. Similarly, despite the Brazilian government’s attempts to support smallholder farmers in supplying agricultural products to schools through the creation of institutional markets and non-competitive bidding mechanisms, the actual participation of farmers remains limited due to the complexity of the decision-making process. The situation in Africa is not encouraging either, as although existing technologies can significantly boost smallholder yields, agricultural incomes are still constrained by land area limitations that make it difficult to lift people out of the international poverty line of USD 1.90 per day [2]. It can be seen that in most developing countries, the intrinsic drive of and capacity for farmers to achieve sustainable income growth remain underexploited, thereby impeding social harmony and stability. This observation prompts a critical inquiry: Is there an alternative approach to accelerate income growth among farmers? If such an avenue exists, what mechanisms would it employ to spur this growth, potentially transforming the agricultural landscape?

In the twenty-first century, the accelerated growth and integration of big data technologies have not only signified a paradigm shift in trading patterns but have also emerged as potent mechanisms to confront and overcome present-day challenges. Prior to the advent of e-commerce, conventional trading practices grappled with elevated thresholds and expenditures, stemming from intricate inventory and logistics frameworks, manifold distribution layers, and prevalent information asymmetry. The indirect nature of merchant–consumer interactions erstwhile made precision in forecasting consumer predilections an elusive goal. Nonetheless, the advent of e-commerce has radically altered this landscape. Propelled by the forces of digital innovation, state-of-the-art technologies encompassing the Internet of Things (IoT), big data analytics, cloud computing, and artificial intelligence have gained extensive application within the realm of cross-border e-commerce. These technological advances are pivotal as they markedly diminish trade barriers, refine operational processes, and curtail associated costs. Moreover, they have remarkably augmented merchants’ capacities in discerning and apprehending consumer inclinations, thereby refining their predictive acumen and optimizing alignment with consumer demands [3]. The transformative influence of e-commerce has given rise to an entirely novel trade ecosystem where the nexus between supply and demand is rendered more immediate and efficacious. This evolution extends beyond mere technological refinement, recalibrating trade’s structure, strategy, and operational dimensions and steering global commerce towards a horizon characterized by enhanced openness, streamlined convenience, and heightened intelligence.

Globally, the rise of e-commerce has emerged as an inevitable trend, leading nations to allocate resources towards fostering its establishment and progress. In Europe, to promote the sustainable development of e-commerce, many countries have formulated e-commerce development strategies. In 2005, the European Union launched the Digital Market Strategy, which aims to eliminate digital barriers and obstacles, construct a unified digital marketplace, and promote the development of cross-border e-commerce. The United Kingdom and Germany also introduced the E-Commerce Regulation and Digital Strategy 2025, and France joined the EU’s One Stop Shop (OSS) mechanism, which simplifies tax declaration procedures for cross-border transactions. In Asia, the Malaysian government put forward a 10-year digital economy blueprint—“MyDigital”. The South Korean government has launched the “U-Korea” development strategy to enhance the optimization and popularization of online trading platforms based on communication infrastructure. Similar e-commerce expansion plans have recently been announced in other developing countries with large rural populations, such as Egypt, India, and Vietnam [4]. Rural e-commerce is also emerging in other developing countries. The evolution of e-commerce is poised to furnish rural communities with avenues for economic advancement, a phenomenon duly acknowledged by scholars and governmental bodies. The development of rural e-commerce is widely deemed as an effective way to narrow the urban–rural gap and achieve rural revitalization [5].

Along with the wave of e-commerce construction in countries around the world, the Chinese government has likewise embarked on a concerted effort to bolster farmers’ income and catalyze rural economic advancement through the targeted promotion of e-commerce within rural areas. As a result, “the Pilot Counties of Introducing E-commerce to Rural Areas” project, a national project that focuses on funding rural areas to develop e-commerce, was started in stricken areas in 2014. The burgeoning impact of digital technology on all walks of life is increasingly evident, and e-commerce, a pivotal facet of the digital economy, is poised to unlock new vistas for rural development, potentially acting as a catalyst in reshaping traditional paradigms. In terms of information costs, China’s rural areas are often seen trailing in economic progression, where limited market accessibility remains a principal bottleneck hampering the growth of remote regions [6]. Although the governmental augmentation of investment in transportation infrastructure has yielded benefits for county economies, it inadvertently accelerates resource diversion from underdeveloped areas to developed areas [7,8], further widening the urban–rural income disparity. E-commerce emerges as a novel impetus for rural revitalization, extending market outreach, curtailing transactional expenses, energizing rural market entities, and unlocking intrinsic growth capacities, which are effects that transcend those of singular infrastructure investments. From the perspective of industrial structure enhancement, the income levels in non-agricultural sectors typically surpass those in agriculture, with digital infrastructure playing a more pronounced role in fostering innovation within non-agricultural industries. The widespread implementation of e-commerce policies in rural realms is likely to spur the flourishing of diverse sectors, including catering, communications, and construction. This paradigm shift challenges the agriculture-centric economic fabric of rural areas, optimally utilizing surplus labor, thereby amplifying farmers’ incomes. Regarding infrastructure development, the inadequacy of rural infrastructure constitutes a significant constraint on the progress of rural communities. The advent of e-commerce can, to a certain extent, ameliorate local ICT infrastructures, orchestrating a synergy of various resource types, infusing new vigor and efficiency into other production factors.

We selected rural e-commerce policies within the Chinese context to further extend and enrich the existing research. China is not only one of the fastest-growing countries in the world in terms of e-commerce, but its construction of digital infrastructure, especially its extension to remote rural areas, which is still uncommon in many developing countries, makes China an ideal case study to examine how e-commerce affects farmers’ income growth. In this regard, based on “the Pilot Counties of Introducing E-commerce to Rural Areas” project, this study delves into the mechanisms through which e-commerce catalyzes income growth among farmers, aiming to contribute Chinese insights and methodologies to the international endeavor of poverty alleviation. This paper’s marginal contributions are distinguished in three pivotal respects: Firstly, whereas preceding inquiries have predominantly concentrated on augmenting agricultural productivity via technological advancements to stabilize food income and mitigate poverty, they fall short of effectively addressing the prevailing conundrum. Our study provides a successful experience of rural e-commerce in China, which provides models and experiences that can be used to fundamentally change the dilemma of farmers’ difficulties in increasing their incomes. Secondly, our study highlights the specific pathways through which e-commerce policies impact farmers’ incomes and analyzes the heterogeneous roles of human capital and regional scale in e-commerce building for farmers’ incomes in more detail to gain insights into the conditions under which e-commerce building works better. Lastly, as far as the external validity of our findings is concerned, we believe that the case of China is representative of many developing countries that, like China, are faced with the facts that farmers’ incomes have begun to enter a phase of slow growth, that the growth rate of incomes has been declining for several years, and that the difficulty of farmers in increasing their incomes and maintaining income imbalance has seriously affected the equity of their economic growth as well as social harmony and stability. Analyzing China’s case therefore not only offers a mirror reflecting the shared predicaments of the developing world but also furnishes invaluable policy references and inspiration for other nations striving to ameliorate the livelihoods of farmers through e-commerce initiatives.

2. Literature Review

Regarding the literature on influencing farmers’ income growth, existing studies have mainly explored the impacts of digital finance, Internet technology, agricultural mechanization services, and rural infrastructure development on farmers’ income from a macro perspective. For instance, Li [9], through an analysis of sample data spanning 30 provinces in mainland China from 2011 to 2019 and utilizing the spatial Durbin model alongside panel threshold model methodologies, demonstrated that digital financial inclusion exerts a substantial positive influence on the augmentation of farmers’ incomes. Similarly, Khan et al. [10], by employing a dataset comprising 580 wheat growers from four districts within the Khyber Pakhtunkhwa Province (KPK) of Pakistan, uncovered that the utilization of mobile phone and Internet technology significantly elevates farmers’ income levels. Furthermore, Sang et al. [11], by leveraging data sourced from the publicly accessible China Labor Force Dynamics Survey (CLFDS), ascertained that the provision of agricultural mechanization services markedly boosts the incomes of rural households. Additionally, Kam et al. [12], through the application of a spatially weighted regression analysis on Bangladeshi data, elucidated the pivotal role of rural infrastructure development in fostering non-agricultural employment opportunities and enhancing agricultural incomes both within and across neighboring regions.

The economic impact of e-commerce can be bifurcated into two main aspects: macro and micro impacts. Initially, the macroeconomic consequences of e-commerce, particularly regarding carbon emissions and Green Total Factor Productivity (GTFP), are scrutinized. Illustratively, Wang et al. [13] discerned a significant diminution in carbon emissions to be attributable to e-commerce pilot policies, as evidenced by city-level panel data across China from 2006 to 2016. Concurrently, Cao et al. [14] employed a multi-period Difference-in-Differences (DID) methodology, revealing an average increment of approximately 1.24% in urban GTFP post-adoption as a pilot city. This increment underscores the efficacy of the National E-commerce Demonstration City (NEDC) pilot policy in enhancing urban GTFP. Subsequently, the microeconomic dimension explores the repercussions of e-commerce ascendance on consumer and household behaviors. Through the application of Pearson’s correlation coefficient and the Lasso model, Liu et al. [15] elucidated the transformative effect of e-commerce on consumer behavior and the valuation of retail spaces. Further extending the analysis, Jiang et al. [16], utilizing data from the China Family Panel Studies (CFPSs) spanning 2014 to 2018, illustrated that e-commerce harbors the potential to mitigate household consumption disparity and bolster the purchasing capacity of lower-income families.

There are indeed some studies on the impact of rural e-commerce which are naturally closer to ours. Chen et al. [17] employed a two-stage predictor variable substitution methodology to substantiate the impact of cooperative membership on the embracement of e-commerce within rural households. Their findings elucidate that both the adoption of e-commerce and cooperative affiliation contribute to augmenting household income. To explore this further, Huang et al. [18] leveraged the push–pull–mooring (PPM) framework to delineate the operational dynamics of e-commerce in village settings as a pivotal element influencing returning residents to resume their professional activities domestically. Moreover, through an insightful case analysis of a Bottom-of-the-Pyramid (BoP) community situated in rural China, Gao et al. [19] delineated a sustainable trajectory initiated from a nascent business model, proceeding through intensive learning phases, and culminating in the establishment of a localized e-commerce ecosystem.

To sum up, the current body of research predominantly addresses the topic of farmers’ income enhancement at a macroscopic level, with a paucity of studies employing microdata for examination. There exists a notable void in empirical investigations elucidating the nexus between e-commerce engagement and the augmentation of farmers’ incomes. Notwithstanding the analyses presented in works such as [17,18,19], which scrutinize the implications of rural e-commerce from divergent vantage points, a common limitation across these studies is their oversight of a detailed dissection of the mechanisms underpinning income growth. This paper substantiates the favorable ramifications of e-commerce policies on agricultural earnings, advocating a novel micro-level approach to address income disparities among farmers while delving deeper into the economic effects of e-commerce. Moreover, the majority of existing studies merely skim the surface of digital empowerment and policy implications and are devoid of granular, focused examinations. By employing a Difference-in-Differences methodology, this research delineates the stimulative influence of e-commerce on rural comprehensive demonstration counties on farmers’ incomes and meticulously deconstructs this impact across three dimensions: information costs, industry structure, and information infrastructure. Furthermore, the discourse on the overarching influence of e-commerce on agricultural income, particularly its variant effects under disparate conditions, remains underexplored. This paper illuminates the differential roles of human capital and county scale in the interplay between e-commerce and farmer earnings. Cultivating a nuanced comprehension of the prevailing landscape of rural e-commerce will facilitate its optimal harnessing as an innovative economic paradigm.

3. Theoretical Analysis and Research Hypothesis

Changes in the employment structure of farm households are important influences on the growth of farmers’ incomes. Moreover, the larger the population that is engaged in non-agricultural activities, the faster the growth of farmers’ incomes. W. Arthur Lewis, an illustrious American economist and the recipient of the 1979 Nobel Prize in Economics, introduced the concept of a “dual structure” in his seminal 1954 paper, “Economic Development with Unlimited Supplies of Labour”. This theory elucidates that in the context of developing countries, where agricultural sector labor productivity typically falls short of its industrial counterpart, the redirection of a segment of the agricultural workforce into the industrial sector can markedly ameliorate the overarching societal productivity. This reallocation implies that as farmers migrate from agricultural to non-agricultural roles, their labor efficiency experiences a substantial uplift, thereby fostering an increase in societal wealth and, consequently, enhancing farmers’ income levels. Simultaneously, the theoretical framework established by Joseph Schumpeter, particularly his theory of technological innovation, emphasizes the essence of “innovation” as a novel recombination of production factors and conditions facilitated by the adoption of unprecedented production techniques and the cultivation of new markets. It provides theoretical support for the role of e-commerce, a new mode of production, in changing the employment structure, reducing information costs, and expanding employment opportunities. It not only provides more employment options for farmers, but also effectively promotes the growth of farmers’ income by optimizing resource allocation and improving production efficiency.

3.1. The Market Connectivity Effect

In the wave of information technology in the twenty-first century, the rapid development of information technology and the widespread popularity of the Internet have brought unprecedented changes to human society. In this transformation, the introduction and implementation of e-commerce in rural areas’ policies has played a key role for rural residents in the urban–rural digital divide. The rise of e-commerce has opened up unprecedented market opportunities for farmers by significantly reducing information asymmetry and drastically lowering transaction costs, as noted by Jensen [20,21] and Shimamoto [22]. The Internet has facilitated the dissolution of barriers across commodity and factor markets, markedly reducing information asymmetry within the social division of labor and empowering farmers with enhanced access to external information beyond geographical limitations. This expanded market accessibility augments employment possibilities and avenues for farmers. For instance, the introduction of a “pop-up” product, accelerated by rapid information dissemination, incites other producers to swiftly engage with the market. This dynamic stimulates market activity and affords rural laborers opportunities to elevate their income levels.

This transformation not only facilitates farmers’ access to precise market data, liberating them from the historical limitations of being passive price takers [23], but also empowers them to synchronize information across spatial and temporal divides, substantially broadening their market reach [24,25]. This accessibility aids regions that were previously constrained by lesser market potential, diminishing the impact of geographical barriers on the economic progression of remote areas and infusing fresh momentum into balanced and synergistic economic development. With the gradual expansion in the market scale, farmers can not only achieve a personal level of production and income, but they can also drive the development of local related industries, which may even give rise to the phenomenon of e-commerce-oriented industrial agglomeration. This agglomeration effect not only improves the synergistic efficiency of the supply chain but also forms a strong supply-side economy of scale, which injects new vitality into the sustained growth of the regional economy [26].

Hypothesis 1.

The implementation of the comprehensive demonstration county policy on e-commerce in rural areas effectively broadens market accessibility through the mitigation of information search costs, thereby empowering farmers to engage with and capitalize on a broader spectrum of business opportunities, culminating in income augmentation.

3.2. The Effect of Industrial Structure Upgrading

With the advancement of e-commerce, on the one hand, by promoting the development of rural e-commerce represented by Taobao villages, the use of e-commerce in the promotion and sale of agricultural products not only enhances the revenue generated from agricultural production, but can also aid the traditional industry in achieving a digital division of labor, fostering an ecosystem characterized by knowledge sharing and elemental synergies. This, in turn, elevates the productivity of the traditional sector, thereby providing a robust stimulus for the sustainable development of the rural economy.

On the other hand, scholars assert that the integration of digital technology with traditional industries can facilitate the comprehensive and efficient utilization of data resources [27], fostering the comprehensive development of the industrial chain and catalyzing the transition of the industrial economy from labor-intensive to technology-intensive modalities [28]. This transformation of the industrial structure is often paralleled by a diversification of employment opportunities, thereby affording rural residents with a broader array of development avenues and choice spaces. The steady expansion of e-commerce not only spurs an increase in local labor demand but also propels the further extension of the pertinent industrial chain. More specifically, it encompasses the development of diverse e-commerce segments, encompassing express delivery, warehousing, packaging, training, and the like, and the flourishing of these sectors generates a substantial number of non-agricultural job opportunities for rural surplus labor, steering them towards a more diversified and high value-added career trajectory [29]. The escalating proportion of output value within the tertiary industry is typically accompanied by a shift of rural labor towards non-agricultural sectors, which not only augments their income sources but also elevates their living standards. In comparison to traditional agricultural pursuits, these nascent employment opportunities offer them higher income levels and broader development horizons, thereby narrowing the income disparity between urban and rural regions.

Hypothesis 2.

The implementation of the policy of e-commerce into rural demonstration counties can effectively facilitate the optimization and upgrading of the rural industrial structure, injecting fresh vitality into the rural economy while bolstering non-agricultural employment opportunities and diversifying farmers’ income streams.

3.3. The Driving Effect of ICT Infrastructure

Information and communication technologies (ICTs) present a vast array of prospects for developing countries. They do not merely expedite the expansion of social and economic networks and enhance access to knowledge and information, but also spawn novel services and employment avenues, thereby emerging as a catalyst for socio-economic transformation [30]. Indeed, with the continual enhancement of ICT infrastructure, the significance of these technologies in augmenting agricultural incomes is progressively magnified. Elevating the caliber of ICT infrastructure within a locality effectively bridges the digital access chasm, thereby contributing to the amplification of agricultural incomes. The positive moderating influence that ICT infrastructure exerts in bolstering agricultural earnings underscores its critical importance within development strategies across developing countries [31].

ICT empowers farmers to acquire pivotal market insights and streamline their ingress to both factor and product markets, thereby elevating agricultural production efficiency and augmenting farmers’ earnings [32]. Through the lens of the information economics theory, the refinement and enhancement of information infrastructure precipitate a decrement in information dissemination costs and a steady advancement in interoperability and sharing capacities. This favorable progression furnishes robust support for the mobility and conglomeration of technology among other critical factors, concurrently breathing fresh dynamism and momentum into economic expansion. From an industrial viewpoint, contemporary infrastructure assumes an indispensable role in technology-dense sectors. It markedly amplifies the technological efficacy of nascent industries via the optimization of the innovation continuum and the digital confluence of production elements such as labor, capital, land, knowledge, and technology. In comparison to conventional infrastructure, the added value contribution of novel infrastructure is especially pronounced in sectors characterized by high levels of technology, including information transmission, software and IT services, alongside scientific research and technical services. The provision of ICT support avails rural locales the opportunity to metamorphose their economic landscapes into entities characterized by higher value addition and technological underpinnings. Such a transformation heralds a structural shift within the rural economy, proffering farmers an array of superior-quality employment prospects and income sources, thereby substantially furthering the development and flourishing of rural regions.

Hypothesis 3.

ICT infrastructure affords farmers seamless accessibility, utilization, and dissemination of information, significantly bolstering their employability and engagement in economic endeavors and thus contributing to the augmentation of farmers’ income growth.

4. Research Design

4.1. Sample and Data

The primary data sources for this article are outlined as follows: First, the list of national-level e-commerce in rural areas and comprehensive demonstration counties from the years 2014 to 2020 is obtained from the official portal of the Ministry of Commerce (MOFCOM), People’s Republic of China. Subsequently, the dataset delineates the per capita disposable income of rural inhabitants within each county in the period from 2000 to 2020 derived from the county-level annual database in the CEINET database and the CSMAR Economic and Financial Research database. Moreover, an assortment of control variables pertinent to the county level—encompassing the gross national product, demographic figures, administrative area dimensions, primary industry value addition, secondary industry value addition, public fiscal expenditures, and year-end financial institution loan balances—was obtained from the China County Economic Statistical Yearbook. Among them, data on the number of mobile phone subscribers in each region are from the China Urban Statistical Yearbook. The data on years of schooling per laborer come from the Rural Fixed Observation Points. (The Rural Fixed Observation Points is a nationwide rural survey approved by the Secretariat of the Central Committee of the Communist Party of China (CPC) in 1984 and is organized and guided by the Central Policy Research Office (CPRO) and the Ministry of Agriculture of the People’s Republic of China (MOA), and it has been tracked continuously for 28 years (1986–2013). Since 2003, the survey has used two levels of questionnaires for rural households and household members, more comprehensively reflecting the production, investment, consumption, employment, and other activities of rural households and their family members across the country, providing a possibility and a good database for a more comprehensive selection of variables in this paper.) Finally, based on the above databases, we constructed unbalanced panel data for 2419 county-level administrative regions in 30 provinces, autonomous regions, and municipalities directly under the Central Government from 2000 to 2020. Given the specificity of poor counties in terms of economic development and policy treatment, we identified 638 poor counties from the above samples based on the list of state-level poverty-stricken counties promulgated by the Poverty Alleviation and Development Office of the State Council in 2012, and we conducted a more in-depth analysis of this special sample in the following text. Among them, Shanghai, Hong Kong, Macao, and Taiwan are not included in the sample of this article, as the corresponding per capita disposable income data of rural residents in these regions for the corresponding years are not included in the above databases.

4.2. Variable Measurement

Explained variables. The rural disposable income (LnRuralDI), in line with Zhang [33], is the natural logarithm of the per capita disposable income of rural residents.

Core explanatory variables. One core explanatory variable is the cross-multiplier term of e-commerce into rural demonstration county policy and the implementation time (e-commerce). The e-commerce into Rural Comprehensive Demonstration County program is gradually being implemented. As such, the policy variable of e-commerce into Rural Comprehensive Demonstration County Pilot is defined as 1 in the year that a county starts piloting e-commerce into rural areas and subsequent years, and it is defined as 0 otherwise. This creates a double difference between the treatment and control groups and before and after the policy pilots.

Control variables. The control variables in this paper include the level of regional economic development (LnperGDP), measured by the logarithm of local per capita GDP; the proportion of the secondary industry (SecInd), expressed as the ratio of the value of the secondary industry added to the regional GDP; the proportion of the tertiary industry (TerInd), expressed as the ratio of the value of the tertiary industry added to the regional GDP; the level of education (Edu), expressed as the proportion of the number of students enrolled in general middle schools to the total population; the level of social welfare (Welfare), expressed as the logarithm of the number of beds in various social welfare adoptive units; the scale of non-agricultural employment (NFP), expressed as the number of people employed in the secondary and tertiary industries in the county summed up and taken as the logarithm; and the level of savings (Deposit), expressed as the logarithm of the balance of residents’ savings deposits. In the heterogeneity analysis, this paper also uses human capital (HCL), measured using labor-per-capita years of schooling (years), with the relevant data obtained from rural fixed observation point data.

Other variables. In the robustness test, the logarithm of the number of mobile phone subscribers in the prefecture-level city where the county is located (LnMOBILE) is selected to reflect the popularity of mobile phones; the logarithm of the number of fixed-line phone subscribers is selected to measure the level of ICT infrastructure (ICTI); and the Digital Inclusive Finance Index reflects digital financial inclusion. (The Digital Inclusive Finance Index is published by the Internet Finance Research Centre of Peking University, which is a comprehensive index portraying the development level of digital finance according to the breadth index, depth index, and digitization degree of the business of Internet financial enterprise indicators and the degree of digitization. The total index that comprehensively portrays the level of digital financial development.) The total power of agricultural machinery is selected to measure the total power of machinery (Machine). The area of administrative area (ACR), the resident population of the area (POP), and the degree of industrial scale (ISD) reflect the geographic characteristics of the county economy; among them, the logarithmic value of the number of units of industrial enterprises above the large scale is used to measure the degree of industrial scale.

4.3. Model Construction and Data Description

In this study, the implementation of the rural e-commerce comprehensive demonstration policy is evaluated as an exogenous shock employing the Difference-in-Differences (DID) approach. Given the policy’s staggered introduction across various timelines, this investigation leverages the multi-period DID methodology to refine the precision in discerning the policy’s impact on the income growth of farmers. This method can effectively control the interference of other potential influencing factors by comparing the differences before and after the implementation of the policy as well as between different policy implementation periods to estimate the policy effect more accurately. The benchmark regression model is shown in Formula (1):

$$\begin{array}{c}{\mathit{LnRuralDI}}_{i,t}={\beta }_0+{\beta }_{1}ecommerc{e}_{i,t}+{\beta }_c{Z}_{i,t}+{\mu }_i+{\delta }_t+{\epsilon }_{i,t}\end{array}$$

(1)

In this formulation, ${LnRuralDI}_{i,t}$ represents the logarithm of the per capita disposable income of rural residents in county i during period t, while ${ecommerce}_{i,t}$ acts as a dummy variable reflecting the policy experiment. Specifically, ${ecommerce}_{i,t}={treatment}_i\times {post}_t$, where the dummy variable ${treatment}_i$ denotes the selected demonstration counties, and it is assigned a value of 1 for designated counties and 0 otherwise; the time dummy variable ${post}_t$ is allocated a value of 0 preceding the policy’s implementation and 1 subsequent to it. The regression coefficient ${\beta }_1$ for ${ecommerce}_{i,t}$, highlighting the income augmentation effects attributable to the comprehensive demonstration policy of integrating e-commerce into rural locales, is the focus of attention in this paper. $Z_{i,t}$ represent a series of control variables that will affect the changes in farmers’ income; ${\mu }_i$ denotes county fixed effects, which are used to control the characteristics of the county level that do not change over time. ${\delta }_t$ denotes time fixed effects in order to control for other exogenous events that may exist during the sample period. ${\epsilon }_{i,t}$ denotes a random perturbation term. Lastly, to address potential correlations among farmers residing within the same county, this study employs robust standard errors, which are clustered at the county level.

This paper presents the descriptive statistics of the main variables, as shown in

Table 1. Descriptive statistics of key variables.

Variables	Obs	Mean	Std. dev.	Min	Max
${\mathit{LnRuralDI}}_{i,t}$	39,817	8.6670	0.7918	6.2106	10.6946
${\mathit{ecommerce}}_{i,t}$	50,946	0.0887	0.2843	0	1
LnperGDP	22,257	9.8742	0.9973	6.5027	13.0435
Seclnd	29,291	0.4078	0.1625	0.0062	0.9780
Terlnd	20,386	0.3454	0.1103	0.0528	0.9876
Edu	26,685	0.0519	0.0187	0.0002	0.2069
Welfare	22,736	6.5564	1.3410	0.6931	9.9422
NFP	16,520	13.8714	1.3014	10.0236	17.9133
Deposit	13,537	13.5397	1.0484	5.3033	17.0977
HCL	1705	7.0751	0.9567	2.2589	12.3158

.

5. Empirical Results and Analysis

5.1. Baseline Regression Results

Table 2. Baseline results.

	(1)	(2)	(3)	(4)
	${\mathit{LnRuralDI}}_{i,t}$	${\mathit{LnRuralDI}}_{i,t}$	${\mathit{LnRuralDI}}_{i,t}$	${\mathit{LnRuralDI}}_{i,t}$
${\mathit{ecommerce}}_{i,t}$	0.0619 *** (0.0058)	0.09865 *** (0.0092)	0.0937 *** (0.0054)	0.0247 *** (0.0032)
LnperGDP		0.5640 *** (0.0085)	0.7638 *** (0.0087)	0.1631 *** (0.0101)
Seclnd		0.8603 *** (0.0451)	1.3850 *** (0.0406)	0.1486 *** (0.0290)
Terlnd		0.1246 *** (0.0263)	0.1551 *** (0.0192)	0.0326 *** (0.0086)
Edu		−2.5406 *** (0.3016)	−2.1901 *** (0.3050)	0.3127 (0.2873)
Welfare		0.0419 *** (0.0038)	0.01782 *** (0.0024)	−0.0013 (0.0016)
NFP		−0.1193 *** (0.0058)	−0.0690 *** (0.0073)	0.0142 *** (0.0052)
Deposit		0.0062 *** (0.0007)	0.0032 *** (0.0005)	0.0018 * (0.0009)
Constant	8.6608 ***	3.6306 ***	0.4593 **	7.0290 ***
Year-fixed effects	YES	NO	NO	YES
County-fixed effects	YES	NO	YES	YES
N	39787	12433	12406	12406
R-squared	0.9802	0.8369	0.9658	0.9879

The standard errors are in parentheses, * p < 0.10, ** p < 0.05, *** p < 0.01. The standard errors are calculated using robust standard errors for county-level clustering.

presents the outcomes of the foundational regression analysis assessing the effects of e-commerce policy implementations in rural areas on farmers’ incomes. To ascertain the robustness of these regression findings, this study incrementally incorporates control variables and adjusts for fixed effects. As anticipated, the estimated coefficients of e-commerce into rural policy are all positive and significant at the 1 percent level, regardless of the form of model setting. The results robustly demonstrate that the implementation of e-commerce into rural policy has a significant positive effect on enhancing farmers’ incomes, aligning with the principal theoretical proposition of this research and affirming the income-augmenting capacity of e-commerce.

Among the control variables, the estimated coefficients of LnperGDP, Seclnd, Terlnd, NFP, and Deposit are positive and significant, suggesting that the degree of local economic development, industrial structure, and the scale of non-agricultural employment have certain impacts on farmers’ incomes. Notably, the coefficient for Edu is positive, albeit not statistically significant, hinting that the role of current human capital investment in increasing farmers’ income has not yet been fully utilized. This underscores the necessity to amplify public expenditure on education to bolster farmers’ earnings more effectively. The negative coefficient of Welfare, although its result is not significant, suggests that the impact of the level of social welfare on farmers’ income may be relatively weak, or there may be a certain degree of an inhibitory effect. This could be attributed to deficiencies or a suboptimal execution of the social welfare framework; the potential impact of the social welfare system on farmers’ incomes should be fully taken into account, and corresponding measures should be taken to fulfill its due role.

5.2. Parallel Trend Test

The Difference-in-Differences (DID) methodology based on exogenous event shocks can effectively address the endogeneity problem of causal identification, but the model is founded on the premise that the treatment and control groups need to satisfy the parallel trend assumption, i.e., in the absence of policy shocks, the explanatory variables have the same trend in the treatment and control groups. If the parallel trend is not satisfied, event shocks may be endogenous, leading to biased estimation results. To ensure the reliability and robustness of the identification results, this study rigorously tests the parallel trend stipulation inherent to the DID approach, employing the analytical framework of Formula (2).

$$\begin{array}{c}{\mathit{LnRuralDI}}_{i,t}={\sigma }_0+{\displaystyle \sum _{s=-4}^5}{\sigma }_s{D}_{it}+{\beta }_c{Z}_{i,t}+{\mu }_i+{\delta }_t+{\epsilon }_{i,t}\end{array}$$

(2)

In Formula (2), $D_{it}$ represents a set of dummy variables, which are assigned a value of 1 for county i where the e-commerce policy in rural areas was enacted in year t, and 0 otherwise. ${\sigma }_s$ represents the estimated coefficient. This variable serves to capture the differential temporal trends between “pilot counties” (those implementing the policy) and “non-pilot counties” (those not implementing the policy). When the value of S is 0, it denotes the year of implementation of the e-commerce policy in rural areas. If S is a negative value, it indicates the number of years prior to the policy’s implementation. Conversely, if S is a positive value, it represents the number of years after the policy’s implementation. The intercept term is denoted by ${\sigma }_0$, while the remaining variables align with those outlined in Formula (2).

Considering that there are less data for the first 4 years and the last 5 years of the policy implementation, this study aggregates data from the first 4 years into period −4 and the data from the final five years into period 5. In addition, this paper takes the current period of e-commerce in rural areas’ policy implementation as the base period. Figure 1 showcases the outcome estimations derived from Formula (2). As depicted in Figure 1, the estimated coefficients for each time window are insignificant before the policy implementation, which implies that the economic trends in the pilot and non-pilot counties were similar before the policy shock, with no significant systematic differences. This observation substantiates the parallel trend hypothesis, a fundamental presupposition of the Difference-in-Differences (DID) approach, which posits that, in the absence of a policy intervention, the projected paths of change for both the treatment and control groups should be congruent. Subsequent to the policy’s enactment, the coefficient estimates become significantly positive, which aligns with the exogeneity prerequisite and confirms that our research sample meets the parallel trends criterion. Such results are indicative of a beneficial influence from the deployment of e-commerce policies within rural areas on the economic dynamism of the pilot counties. This finding not only verifies our research hypothesis, but also further confirms the effectiveness of the e-commerce into rural areas policy in promoting the growth of farmers’ incomes.

5.3. Placebo Test

Another concern about the findings of this study is that statistical significance is likely to stem from some random factor. To address this concern, this paper refers to an existing treatment [34] to determine whether the farmers’ income-generating effect of e-commerce in rural policy is caused by other unobserved omitted variables. In this paper, we first mismatch the information on the policy implemented in the county and then delimit the experimental and control group samples. We repeat this random process 100 times, run 100 regressions, and compare the t-statistic of the pilot policy of returning to the village in the 100 regressions with the t-statistic of the policy of e-commerce into rural demonstration counties in the baseline regression. As illustrated in Figure 2, all of the e-commerce demonstration county policy statistics in the 100 regressions generated are smaller than the t-statistic for the e-commerce in rural areas policy in the benchmark regression (10.13). Specifically, out of under 100 randomly generated policy shocks, only 0.60% of the random policies have a significant positive impact on local farmers’ incomes, and thus, the probability of committing a “pseudo-error” in this paper is very low, thus eliminating the possibility of pseudo-regression. This further suggests that e-commerce policies in rural areas have a relatively robust effect on the growth of county farmers’ income and promote local farmers’ income growth.

5.4. Robustness Checks

5.4.1. Excluding the Interference of Other Policies

To avoid the possibility that other policies would affect farmers’ incomes during the sample period and cause bias in the results of the baseline estimation, this study, by collecting and combing through the literature, identifies two pilot policies that may affect farmers’ employment during the sample period. They are the National E-commerce Demonstration City policy and the Information to Village and Household Project policy. To accurately assess the effect of e-commerce on rural areas’ policies and to exclude the potential impact of other pilot policies, this paper includes dummy variables for these two policies in the baseline regression model. Specifically, CITY represents the national-level e-commerce demonstration city policy dummy variable. If CITY is equal to 1, it means that the county is recognized as a national-level e-commerce demonstration city in year t and later; otherwise, it is 0. INF stands for the information to village and household project policy dummy variable. If INF is equal to 1, this indicates that the county is selected as a pilot county for information to the village and household in year t and later, and when it is equal to 0, the opposite is true. A subsequent analysis, controlling for these policy variables, yielded results that are consistent with the initial baseline findings, affirming the robustness of this study’s conclusions. This consistency underscores the assertion that the employment effects attributed to the rural e-commerce policy are not confounded by other pilot initiatives. The detailed outcomes of this refined analysis are presented in

Table 3. Robustness checks.

${\mathbf{LnRuralDI}}_{\mathbf{i},\mathbf{t}}$	$\mathbf{Exclusion} \mathbf{of} \mathbf{Other} \mathbf{Policies}$	$\mathbf{MOBILE}$	Removal of Special Sample	Restricted Sample: Samples from 2010 to 2020	Nei-1	Mahalanobis Distance Matching
	(1)	(2)	(3)	(4)	(5)	(6)
${\mathit{ecommerce}}_{i,t}$	0.0245 *** (0.0032)	0.0246 *** (0.0032)	0.0092 ** (0.0044)	0.0260 *** (0.0025)	0.0222 *** (0.0033)	0.0219 *** (0.0033)
$\mathrm{CITY}$	−0.0227 ** (0.0094)
$\mathrm{INF}$	−0.0004 (0.0098)
Log(MOBILE)		0.0001 (0.0004)
Controls	YES	YES	YES	YES	YES	YES
Year-fixed effects	YES	YES	YES	YES	YES	YES
County-fixed effects	YES	YES	YES	YES	YES	YES
R-squared	0.9879	0.9879	0.9881	0.9873	0.9882	0.9882
N	12,406	12,406	8559	11,129	11,670	11,696

The standard errors are in parentheses; ** p < 0.05, and *** p < 0.01. The standard errors are calculated using robust standard errors for county-level clustering.

, column (1).

Moreover, in light of the pervasive adoption of mobile communication devices, particularly mobile phones in rural domains, which ostensibly influences farmers’ incomes, there arises a conjecture that the policy variable may inadvertently encapsulate confounding factors tied to mobile phone usage rather than the direct effects of the policy itself. To address this concern, this study integrates the logarithmic metric of mobile phone subscribers within the pertinent county (Log(MOBILE)) into the regression equation prior to the analysis. The findings delineated in Table 3 (2) affirm that the coefficient associated with e-commerce demonstration counties retains its significant positivity even after adjusting for the mobile phone usage variable, Log(MOBILE).

5.4.2. Eliminating the Influence of Special Samples

The present study’s sample encompasses 638 impoverished counties. Considering the notable disparities in economic development status and policy incentives between impoverished and non-impoverished counties, the inclusion of these specific samples may introduce bias into the regression results. To safeguard the robustness and reliability of our findings, we excluded the 638 impoverished county samples from the regression analysis and conducted the regressions solely based on data from non-impoverished counties. The regression results, after excluding the special samples, are presented in column (1) in

Table 4. E-commerce demo counties’ impact on farmers’ incomes.

${\mathit{LnRuralDI}}_{\mathit{i},\mathit{t}}$	$\left(1\right)$	(2)	(3)	(4)	(5)
${\mathit{ecommerce}}_{i,t}$	0.0222 *** (0.0045)	0.0025 (0.0116)	0.0178 *** (0.0024)	0.0609 *** (0.0129)	0.0077 (0.0068)
${\mathit{MAK}}_{it}$	0.2110 *** (0.0434)
${\mathit{ecommerce}}_{i,t}\times {\mathit{MAK}}_{it}$	−0.0413 *** (0.0086)
$\mathrm{Ind}$		0.0845 * (0.0462)
${\mathit{ecommerce}}_{i,t}\times \mathrm{Ind}$		0.0504 ** (0.0250)
$\mathrm{ICTI}$			0.0061 ** (0.0029)
${\mathit{ecommerce}}_{i,t}\times \mathrm{ICTI}$			−0.0155 *** (0.0024)
Machine				0.0538 *** (0.0060)
${\mathit{ecommerce}}_{i,t}\times$ Machine				−0.0128 *** (0.0035)
DIF					0.0013 *** (0.0001)
${\mathit{ecommerce}}_{i,t}\times$ DIF					0.0001 (0.0001)
Control variables	YES	YES	YES	YES	YES
Year-fixed effects	YES	YES	YES	YES	YES
County-fixed effects	YES	YES	YES	YES	YES
R-squared	0.9880	0.9877	0.9880	0.9879	0.9928
N	12,406	12,347	12,318	10,587	7171

The standard errors are in parentheses; * p < 0.10, ** p < 0.05, and *** p < 0.01. The standard errors were calculated using robust standard errors for county-level clustering.

. Notably, the interaction term pertaining to the pilot policy remains significant with a positive estimated coefficient, indicating the robustness of our research conclusions.

5.4.3. Reducing the Time Span of the Sample

The temporal scope of this study spans from 2000 to 2020, with the pivotal introduction of rural e-commerce policies occurring in 2014, raising concerns that the pre-policy phase may excessively extend. To ensure analytical robustness, this study narrows the sampling frame to the decade of 2010 to 2020. The ensuing findings presented in column (4) in Table 3 affirm the substantial impact of rural e-commerce policies on the growth of farmers’ incomes, thereby reinforcing the robustness of this study’s conclusions.

5.4.4. PSM-DID

In this study, the Propensity Score Matching–Difference-in-Differences (PSM-DID) methodology is employed to secure robust findings. The initial step involves estimating the propensity score via a Logit model, articulated as per the formula below:

$$\begin{array}{c}\mathrm{logit}\left({\mathit{treatment}}_i=1\right)={\beta }_0+{\beta }_c{Z}_{i,t}+\gamma X_{i,t}+{\mu }_i+{\delta }_t+{\epsilon }_{i,t}\end{array}$$

(3)

${\mathit{treatment}}_i$ represents a binary variable identifying membership within the experimental group; $Z_{i,t}$ serves as a control variable for the multiple difference method model above; and $X_{i,t}$ encompasses additional determinants influencing a district’s selection for policy enactment. Drawing on the study presented in [35], $X_{i,t}$ contains the area variable (ACR), the population variable (POP), and the industrial scale degree (ISD), which reflect the local economic and geographical characteristics.

Upon conducting logit regression, propensity scores are derived, facilitating sample pairing through various matching techniques such as Nearest Neighbor, kernel, and radius matching. Nevertheless, the efficacy of the Propensity Score Matching–Difference-in-Differences (PSM-DID) approach hinges on the successful completion of the “balance test” and “common support test” to affirm matching integrity. Consequently, this study elects to employ both the Nearest Neighbor-1 matching technique and the Mahalanobis distance matching method, concurrently executing the aforementioned tests. After eliminating the samples that contravene the common support hypothesis, this research re-evaluates the policy’s income augmentation effect via the Difference-in-Differences methodology. Regression analyses, as delineated in columns (5)–(6) in Table 3, indicate that the effect of the rural e-commerce policy on income is consistently positive irrespective of the matching method employed—be it Nearest Neighbor-1 or the Mahalanobis distance. Hence, the findings of this study underscore the robustness of the model’s estimations.

6. Mechanism Testing and Heterogeneity Analysis

6.1. Market Connectivity Effect

One of the mechanisms through which e-commerce development contributes to regional economic growth is the reduction in trade costs, thus helping regions with low market potential to connect to markets. Such regions are often beset by challenges, including inadequate information dissemination and constrained distribution networks, culminating in elevated information and transaction costs that hamper economic progression. E-commerce platforms enable enterprises and local inhabitants in these areas to engage directly with broader markets, thereby streamlining access to market intelligence and sales opportunities, expanding their customer base [36] and, to a degree, offsetting their inherent market limitations. Consequently, the development of e-commerce should have a greater impact on areas with a smaller local population (smaller market) and a remote geographical location (smaller foreign market). Based on this idea, this paper designs Formula (4) for testing under the framework of the above identification strategy.

$$\begin{array}{c}{\mathit{LnRuralDI}}_{i,t}={\beta }_0+{\beta }_1{\mathit{ecommerce}}_{i,t}+{\beta }_2{\mathit{ecommerce}}_{i,t}\times {\mathit{MAK}}_{it}+{\beta }_c{Z}_{i,t}+{\mu }_i+{\delta }_t+{\epsilon }_{i,t}\end{array}$$

(4)

In this study, market potential is conceptualized as ${\mathit{MAK}}_{it}$, signifying the market potential of county $i$ at the conclusion of year $t$. Herein, ${\mathit{GDP}}_{it}$ symbolizes the gross regional product of county $i$ at the end of year $t$, which is subjected to logarithmic transformation, while $d_{ij}$ represents the linear distance between county $i$ and the central market at county $t$. It is specified as follows:

$${\mathit{MAK}}_{it}=\sum _{j=1}^N\frac{{\mathit{GDP}}_{it}}{d_{ij}}$$

The results derived from the estimation of Formula (4), as presented in Table 4(1), reveal a statistically significant negative coefficient for the interaction term between market potential and e-commerce demonstration counties. This finding suggests that e-commerce serves to reduce the constraints imposed by population size and geographic distance on the expansion of farmers’ incomes. It fosters an enhanced flow of information and optimizes resource allocation, thereby broadening farmers’ market access, diminishing transaction costs, and elevating sales efficiency.

6.2. Industrial Structure Upgrading Effect

In this study, the proportion of the tertiary sector’s value added to the GDP is utilized to represent the industrial structure ($\mathrm{Ind}$). The analysis presented in column (2) in Table 4 reveals that the interaction between the integration of e-commerce in rural policy and the industrial structure exerts a significantly positive impact on agricultural income, while the findings pertaining to e-commerce demonstration counties do not exhibit statistical significance. This indicates that with the advanced industrial structure, the same magnitude of e-commerce implementation will bring more substantial economic growth for farmers, and the expansion of e-commerce fuels the transformation of the rural economy and promotes the process of modernization and diversification in the industrial structure. This stimulates a broader development potential for rural e-commerce and creates new non-agricultural jobs, prompting the transfer of surplus rural labor to non-agricultural jobs, which, in turn, increases the non-agricultural employment income of farmers.

6.3. ICT Infrastructure-Driven Effect

E-commerce policies for rural areas usually include a range of measures aimed at promoting rural e-commerce, one of which is the enhancement of rural ICT infrastructure. Such policies might furnish financial, technological, or additional support aimed at augmenting Internet connectivity, communication networks, and ICT services within rural regions.

Column 3 in Table 4 explores the role of ICT infrastructure in facilitating e-commerce in rural areas. The analysis reveals a statistically significant negative coefficient for the interaction between e-commerce policies and rural infrastructure. This finding implies that e-commerce policies play a stronger role in facilitating the growth of farmers’ incomes in environments with weaker infrastructure. The rationale behind this phenomenon could be attributed to e-commerce serving as a novel market conduit in such areas, markedly enhancing farmers’ capacity to tap into broader markets and escalate their product sales. Nevertheless, as infrastructure conditions ameliorate, farmers may increasingly leverage a multitude of channels for market access, thereby somewhat diluting the distinct advantage offered by e-commerce. This, however, does not imply that e-commerce policies become ineffective in well-infrastructure regions; rather, their impact is synergistic with other developmental strategies, collectively contributing to the continuous income growth of farmers. This interpretation is reinforced by the significantly positive coefficients associated with infrastructure, indicating its progressively beneficial influence on farmers’ incomes as infrastructure conditions advance. Consequently, in settings where infrastructure is more developed, the positive contribution of its sophistication to the growth of farmers’ incomes is increasingly significant, and the two work together to create a diversified development dynamic that promotes sustained growth in farmers’ incomes.

6.4. Discussion

In this subsection, we discuss potential alternative mechanisms that could explain our results.

6.4.1. Agricultural Mechanization Level

Some may argue that the development of e-commerce contributes to farmers’ incomes by increasing the level of agricultural mechanization. This is because e-commerce fuels agricultural modernization, scales production, and widens market access for farm produce, thereby urging a shift towards efficient, high-yield mechanized farming techniques. This transition is anticipated to enhance agricultural productivity, curtail labor costs, and foster specialization and intensification in agricultural practices, potentially maximizing economic returns and carving new avenues for income enhancement for farmers. However, the empirical evidence presented in column (4) in Table 4 indicates a discernible negative correlation between the mechanization level and the interaction term of e-commerce demonstration counties. This intimates that, within the confines of this study’s scope and the specificities of the policy implementation context, although escalated mechanization is conventionally viewed as conducive to bolstering agricultural production efficiency and augmenting farmers’ incomes, it may not necessarily serve as the most efficacious strategy to leverage e-commerce as the principal mechanism for enhancing rural incomes under the policy framework examined. Therefore, the findings of this paper cannot be explained by the fact that increasing the mechanization level promotes farmers’ income growth.

6.4.2. Digital Inclusive Finance

The adoption of e-commerce policies is frequently linked with the advancement and refinement of financial services, facilitating farmers’ access to financial assistance. These enhanced financial services play a crucial role in mitigating the financial constraints faced by farmers, thereby augmenting their production capabilities and market competitiveness. This enhancement, in turn, contributes to elevated income levels. Although the optimization and innovation of financial services have positive impacts on farmers’ incomes in theory, the role of this mechanism is not significantly reflected in the empirical results. As shown in column (5) in Table 4, the interaction term between e-commerce demonstration counties and digital financial inclusion does not reach statistical significance. This implies that the direct role of digital financial inclusion in e-commerce development for farmers’ incomes is not significant within the scope of this paper. The lack of penetration of financial services in rural areas and the limited understanding and utilization of financial services by farmers prevent this mechanism from fully explaining the empirical results of this paper.

6.5. Heterogeneity Analysis

6.5.1. Heterogeneity of Human Capital

The pivotal influence of human capital on the proliferation of nascent industries is incontrovertible. Studies have shown that the enhancement of human capital accumulation can not only improve non-agricultural employment opportunities, but also increase income [37]. E-commerce, as an emerging industry, has certain requirements regarding the knowledge and skill levels of participants. In regions with low levels of human capital, there tends to be a lower receptivity towards novelties and a constrained proficiency in leveraging contemporary informational technologies. This situation may limit the income-generating effect of the e-commerce into rural areas policy. Therefore, this paper is divided into a higher education group and a lower education group according to the mean value of years of education, and the values are regressed using Equation (1).

As depicted in

Table 5. Heterogeneity tests.

	High-Education Regions	Low-Education Regions	Larger-Scale Counties	Smaller-Scale Counties
	(1)	(2)	(3)	(4)
${\mathit{ecommerce}}_{i,t}$	0.0247 *** (0.0032)	0.0108 (0.0111)	0.0282 *** (0.0035)	0.0106 (0.0077)
Control variables	YES	YES	YES	YES
Year-fixed effects	YES	YES	YES	YES
County-fixed effects	YES	YES	YES	YES
N	12,160	224	9742	2435
R-squared	0.9879	0.9920	0.9887	0.9850

The standard errors are shown in parentheses; *** p < 0.01. The standard errors were calculated using robust standard errors for county-level clustering.

, the implementation of e-commerce policies markedly elevates farmers’ incomes at a significance level of 1 percent, exerting a positive moderating influence, particularly in regions with superior educational infrastructure. Conversely, in counties characterized by lower educational attainment, this effect does not manifest statistical significance. This observation intimates that districts with higher levels of education are able to utilize e-commerce policies more effectively, which, in turn, significantly contributes to farmers’ income growth. The reason for this may lie in the positive impact of the human capital level on the effectiveness of e-commerce policies. In areas with higher levels of education, farmers usually have better e-commerce skills, broader access to information, and greater market awareness, which, together, contribute to farmers’ ability to grasp e-commerce opportunities, make better use of e-commerce opportunities for sales and promotion, and identify potential business risks, enhancing the effectiveness of e-commerce policies.

6.5.2. Heterogeneity of County Scale

In contrast to their smaller counterparts, larger counties boast a higher capacity for population support and employment, exhibiting significant clustering effects in labor, capital, technology, and other key resources. Additionally, these areas often have a more robust industrial base and advanced level of information technology, which are factors that could lead to divergent outcomes in the income-generating impact of e-commerce policies in rural settings. For this reason, this paper is based on the median annual average resident population which will be divided into samples of larger size counties and smaller size counties. The empirical findings illustrated in Table 5 reveal that e-commerce policies significantly bolster farmers’ incomes in larger counties, whereas this effect is less pronounced in smaller counties. This discrepancy may stem from the more comprehensive transport and communication infrastructure in larger counties, which underpins the effective execution of e-commerce strategies. Counties with a relatively weak level of comprehensive development are still unable to fully unleash the driving effect of e-commerce on income growth enhancement even if they are included in the demonstration list due to the economic environment (e.g., production and consumption), and insufficient supporting facilities, such rural areas, could enhance their capacity to exploit e-commerce for income generation by reinforcing infrastructure, nurturing talent, and refining supportive policies, thereby progressively unlocking the potential of e-commerce in augmenting farmers’ incomes.

7. Discussion

7.1. Comparison with Previous Results

The research landscape has yielded several compelling insights into strategies for augmenting farmers’ incomes. For instance, evidence from a study on a governmental initiative aiming to provide every village with broadband Internet access and additional rural services indicates that increased access to computers in rural areas has a positive correlation with enhanced income levels among residents. Furthermore, such advancements are partly attributed to local governmental investments in rural informational infrastructure and logistics [29]. Contract farming is recognized as an instrumental approach for creating fresh market opportunities that can mitigate market inefficiencies that are commonly observed in numerous developing countries. The findings from this investigation highlight contract farming’s pivotal contribution to elevating farm revenues and bolstering the technical efficiency of agricultural production [38]. A separate analysis focusing on the development of a multifunctional agricultural network illustrates how optimizing the ecological benefits through interest alignment within this network can lead to an uplift in the average incomes of farmers [39]. Building on the existing literature, this paper extends the exploration of the influence that e-commerce policies, underpinned by electronic information technology, have on augmenting farmers’ incomes. It provides a detailed microeconomic examination of how e-commerce initiatives at the county level can positively affect farmers’ financial gains. Our research uncovers that when e-commerce policies are implemented in rural regions, they serve to enhance farmers’ productivity by diminishing information costs, fostering the advancement of the local industry, and bolstering information and communication technology (ICT) infrastructures. This trajectory of development is instrumental in driving up farmers’ incomes. The enactment of such policies goes beyond merely spurring on rural economic growth; it delivers tangible advantages to the agricultural community at a grassroots level.

7.2. Research Implications

As highlighted previously, the adoption of e-commerce within rural settings can markedly bolster productivity, which will consequently enhance income levels. The establishment of e-commerce frameworks can further foster income sustainability by offering farmers a broader comprehension of input materials. This enables them to access expanded markets and mitigate the negative impacts arising from information discrepancies. Traditionally, farmers’ income growth was predominantly dependent on the advancement of the tangible economy, a paradigm fraught with constraints pertaining to information access and resource distribution, thereby amplifying the economic vulnerabilities faced by farmers. The introduction of e-commerce has the potential to revolutionize this scenario, facilitating a more strategic allocation of resources such as land, labor, and capital. The Internet serves as a cutting-edge platform for the seamless and swift exchange of goods. Concurrently, the digital mindset fosters tangible economic growth, catalyzing the metamorphosis of local industrial frameworks. Rural regions are progressively coalescing into industrial chains centered on e-commerce, spurring growth across associated sectors such as logistics, warehousing, and packaging. This development not only proliferates alternative non-agricultural employment options for farmers but also amplifies the value-added component of agricultural produce, thereby augmenting farmers’ income levels. Moreover, e-commerce adopts the intrinsic strengths of the physical economy to orchestrate synchronized online and offline progress within rural areas, culminating in the symbiotic integration of virtual and tangible economies. Within a market landscape characterized by heightening competition, such a strategic fusion is pivotal in safeguarding a consistent and sustainable elevation in farmers’ incomes.

8. Conclusions and Policy Recommendations

Drawing on the framework of China’s initiative to introduce e-commerce to rural comprehensive demonstration counties, this study constructs an unbalanced panel dataset encompassing 2419 county-level jurisdictions across 30 provinces, autonomous regions, and municipalities under the Central Government from 2000 to 2020. Utilizing a quasi-natural experimental framework for analysis, the investigation yields systematic empirical insights. The findings underscore that the Rural Comprehensive Demonstration County policy for e-commerce significantly contributes to the enhancement of farmers’ income levels. Mechanistic analyses reveal that the diminution of information costs, the facilitation of local industrial structural refinement, and the bolstering of ICT infrastructure emerge as pivotal mechanisms for poverty alleviation and income augmentation following the integration of e-commerce into rural domains. Moreover, the study identifies that the efficacy of e-commerce initiatives in elevating farmers’ incomes is markedly pronounced in locales characterized by superior human capital endowments and greater county dimensions.

Our conclusions have important practical implications that extend to many policy recommendations. First, the empirical results of this paper verify the effectiveness of the “Comprehensive Demonstration of E-commerce in Rural Areas” policy in raising farmers’ incomes, promoting rural employment, reducing the stagnation of agricultural products, and alleviating poverty. As the fastest-growing emerging economy, China’s development experience is worthy of reference, providing a scientific basis for regions and countries that want to activate rural economic vitality through the development of rural e-commerce. Secondly, the government can strengthen its investment in rural infrastructure and optimize the structure of public education expenditure, especially basic education, to improve human resources and achieve the revitalization of rural talents. In addition, there is variability in the role of e-commerce policies depending on the size of the county, so targeted policy design and implementation can be carried out according to regional characteristics to give full play to the maximum potential of e-commerce in promoting rural income generation and rural revitalization. In summary, bridging the digital divide is essential for ensuring a sustainable increase in farmers’ income.

To facilitate further research, we summarize the limitations of this paper and suggest some possible additional research ideas for the future. Initially, even though we have engaged rigorous checks, including Propensity Score Matching–Difference in Differences (PSM-DID), to attenuate concerns of endogeneity, it is incumbent upon subsequent inquiries to accrue additional empirical substantiation affirming this methodology’s efficacy. Given our existing dataset’s finite scope and the inherent constraints of analytic methodologies at our disposal, there remains a possibility that certain influential factors may have eluded our assessment. Secondly, while our examination is primarily concerned with the ramifications of China’s “E-commerce Demonstration County” policy on agricultural revenues, it omits comparative dissections with analogous digital commerce initiatives unfolded in other sovereignties. This comparative exclusion potentially constricts our ability to thoroughly evaluate the distinctive attributes and potency of China’s strategies and to gauge heterogeneities across diverse milieus. In light of this, prospective research could profit from embedding cross-national comparative analyses that might yield more expansive and insightful discernments. Lastly, with respect to the assemblage of samples and variables, we acknowledge the omission of factors that may carry significance. Notwithstanding our diligent efforts to curate an exemplary sample and pinpoint pivotal variables, we anticipate that future investigations might benefit from augmenting the sample corpus, diversifying the control variables array, and thereby enhancing the overarching rigor and precision of this research.