农业数字化与农民收入增长——来自中国的证据

Abstract

Agricultural digitalization has become an important driver of rural revitalization and agricultural modernization. Using a balanced panel of 30 Chinese provinces from 2011 to 2022, this study constructs a multidimensional Agricultural Digitalization Index covering digital infrastructure, enabling conditions, digitalized production and operation, and digital services. We estimate baseline two-way fixed-effects models and conduct complementary analyses to address endogeneity, test nonlinearities, and examine the conditioning role of urbanization. The results show that agricultural digitalization is associated with higher per capita disposable income of rural residents, and the finding is robust to alternative index construction, lag structures, and sample adjustments. Threshold estimates reveal stage-dependent returns: marginal income gains are larger at lower levels of digital development and remain positive but weaker beyond the estimated threshold. Human capital exhibits a reinforcing threshold, with markedly stronger income gains once rural education surpasses a critical level. The effects are larger in central and western provinces, economically less developed regions, and major grain-producing areas, and urbanization further amplifies the income effect, underscoring the role of complementary market and service environments. Income-component results indicate that gains are concentrated in wage and operating income. Overall, agricultural digitalization can raise farmers' income, but the magnitude of benefits depends on development stage, human-capital constraints, and local market connectivity.

1 Introduction

The rapid advancement of digital technologies, including the Internet of Things (IoT), artificial intelligence (AI), big data analytics and blockchain, has fundamentally reshaped agricultural production systems and global agri-food value chains (; ). A growing body of research indicates that digital agriculture has become an important driver of agricultural sustainability by improving resource allocation efficiency, enhancing risk management and strengthening resilience to climate and market shocks (; ). In developing countries where smallholders constitute the backbone of the agricultural economy, digital tools can reduce information frictions, lower transaction costs and expand market access, thereby improving household welfare and income (; ). In response to these global trends, China has accelerated its rural digital transformation through policies such as the Digital Village Strategy, accompanied by extensive investment in digital infrastructure, smart agriculture and digital agricultural services (; ; ). However, the agricultural sector remains dominated by small-scale farmers, and substantial disparities in economic development, human capital and digital infrastructure persist across regions, which are closely related to regional disparities in digital infrastructure and rural development (; ; ). At the same time, raising farmers' income is a central policy objective for advancing rural revitalization and achieving common prosperity in China. Against this policy background, whether agricultural digitalization (AD) can effectively promote farmers' income has become a critical research question. It is equally important to understand under what conditions this effect may be strengthened or weakened. In this study, agricultural digitalization (AD) refers to the sector-specific digital transformation of agriculture, encompassing digital infrastructure, enabling environments, digitalized production, digital management, and digital services. Unlike the broader concept of the digital economy, AD focuses exclusively on the agricultural sector. Rather than capturing micro-level adoption by individual farmers, we conceptualize AD as a province-level development process reflecting the overall digital capacity and application intensity within agriculture. This definition forms the theoretical basis for constructing the multi-dimensional AD index used in the empirical analysis.

Although research on digital transformation has expanded rapidly, empirical studies that directly examine agricultural digitalization (AD) remain relatively limited (; ; ). Existing studies on China often rely on broad proxies such as internet penetration, broadband coverage, or composite indices of digital development, rather than directly measuring digital applications and services embedded in agricultural production, management, and marketing (). This inconsistency between the conceptual focus on AD and the use of broad digital-development indicators may obscure the sector-specific mechanisms through which digital technologies affect farmers' income in rural areas (; ). In particular, failing to distinguish agriculture-oriented digital infrastructure, platform services, and rural service systems from broader digital development may hinder an accurate assessment of how digital technologies reshape production modes, transaction patterns, and income structures in the agricultural sector (). There is therefore a clear need to construct a sector-specific and multidimensional Agricultural Digitalization Index tailored to the Chinese context and to use it as the core explanatory variable in empirical analysis, while also examining the channels and conditions through which its income effects may vary.

Beyond the average effect, the income impact of agricultural digitalization is likely to be conditional and nonlinear. From a theoretical perspective, agricultural digitalization can influence farmers' income through multiple channels, including improving production decisions, expanding employment opportunities, facilitating financial inclusion, and enhancing access to agricultural and public services (; ; ). However, the income gains generated by agricultural digitalization are unlikely to evolve linearly along the development path. Related studies on digital development suggest that its effects on income distribution and urban–rural disparities may become nonlinear once certain development stages are reached (; ; ). In the early stages of digital development, income gains may be constrained by infrastructural gaps, limited service coverage, and adoption barriers (). As digital infrastructure and agricultural applications become more widespread, network externalities, learning effects, and business model innovation may strengthen the income-enhancing effects of agricultural digitalization (; ). At higher development levels, diminishing marginal returns may emerge as basic digital services become saturated and new forms of digital divide begin to appear (; ). These stage-based dynamics suggest that the income effects of agricultural digitalization may exhibit threshold characteristics that require explicit empirical testing. Human capital further shapes these effects because farmers need basic education and digital literacy to access, understand, and effectively use digital tools and services (; ; ; ). If rural human capital is insufficient, the benefits of agricultural digitalization may be significantly constrained or may accrue mainly to better-equipped households; once education levels surpass a critical point, however, the income-enhancing effects of AD may become markedly stronger (). Urbanization also constitutes a key structural condition that may alter the relationship between agricultural digitalization and farmers' income. Related studies suggest that digital development can reshape urban–rural income relations, urbanization processes, and urban–rural integration by reallocating labor, capital, and information across space (; ; ; ; ). In this sense, higher levels of urbanization may provide a more favorable external environment for agricultural digitalization to generate stable and sustained income gains.

Taken together, these considerations suggest that the income effects of agricultural digitalization may vary systematically with development stage, human capital, and local market conditions. Against this background, this study constructs a provincial-level Agricultural Digitalization Index for China and empirically examines the impact of AD on farmers' income using panel data for 30 provincial-level regions in China from 2011 to 2022. A two-way fixed-effects model is employed as the baseline framework, and potential endogeneity is addressed through instrumental-variable estimation. Furthermore, panel threshold regressions are used to identify possible development-stage thresholds of AD and human capital thresholds, while moderation analysis is applied to investigate how urbanization shapes the income effects of AD. Specifically, this study addresses four core research questions: (1) Does AD significantly increase farmers' income? (2) Does the income effect of AD exhibit stage-based threshold characteristics along the digital development path? (3) Does rural human capital also present a threshold that conditions the income-enhancing effect of AD? (4) Does urbanization moderate the relationship between AD and farmers' income? In addition, the empirical analysis further examines whether the income effects of AD differ across eastern, central, and western regions with distinct economic structures, industrial foundations, and resource endowments. This extension provides additional evidence on possible digital divide effects.

This paper makes three main contributions. First, by developing a sector-specific Agricultural Digitalization Index, this study provides a more suitable empirical basis for examining how agricultural digitalization is associated with farmers' income growth. Moving beyond broad digital-economy proxies helps better identify the sector-specific channels through which digital technologies affect rural income. Second, the paper shows that the income effect of agricultural digitalization is conditional rather than uniform. By examining development-stage thresholds, the threshold role of human capital and the moderating role of urbanization, the study clarifies the conditions under which agricultural digitalization can be translated more effectively into income growth. Third, the paper further reveals substantial heterogeneity across regional and functional contexts, offering additional evidence on the differentiated pathways through which agricultural digitalization supports rural income growth. The remainder of this paper is structured as follows. Section 2 reviews the related literature. Section 3 develops the theoretical framework and research hypotheses. Section 4 describes the data and variable construction. Section 5 outlines the empirical methodology. Section 6 reports the empirical results, including the baseline estimates, robustness checks, endogeneity tests, threshold effects, heterogeneity analysis, moderating effects and further analysis. Section 7 discusses the main findings. Section 8 concludes with policy implications.

2 Literature review

2.1 Agricultural digitalization and farmers' income

Agricultural digitalization is commonly understood as the embedding of digital technologies such as sensors, the Internet of Things (IoT), big data analytics, and artificial intelligence into agricultural production, management, and value chains, thereby enabling data-driven decisions at both farm and system levels (; ). These technologies support more precise input use, real-time monitoring, and predictive decision-making, which can improve productivity, reduce risk, and enhance environmental performance across the agri-food system (; ).

A growing body of empirical research has examined whether agricultural digitalization and related digital technology adoption can translate into higher household income. At the micro level, studies for low- and middle-income countries find that digital agriculture solutions and platforms can improve smallholders' access to price information, weather forecasts, extension services and buyers, thereby increasing farm revenues and, in some cases, off-farm earnings (; ). Recent evidence from Vietnam and ethnic minority regions of China suggests that digital technology adoption significantly raises farm and total household income, particularly in remote rural areas with limited traditional market access (; ). Other work highlights that digital infrastructure, including broadband and e-commerce logistics, facilitates the expansion of farm scale and market-oriented production, which can further enhance income levels ().

However, these benefits are not automatic or evenly distributed. Some studies emphasize that digital agriculture may reduce the demand for certain types of manual labor and restructure employment within the rural economy, with ambiguous implications for low-skilled workers (). In addition, heterogeneous access to digital tools and services can generate a “digital dividend” for connected farmers while reinforcing a “digital divide” for others, especially where infrastructure, finance and skills are unevenly distributed (). Recent evidence also shows that digital applications can significantly improve family farm income, with stronger income-enhancing effects observed for smaller-scale and lower-income farms, further supporting the income-improving potential of agricultural digitalization (). Overall, the literature suggests that agricultural digitalization has considerable potential to raise farmers' income, but its realized effects depend strongly on local development conditions and complementary factors such as infrastructure, finance, and human capital.

2.2 Nonlinear and threshold effects

Beyond average effects, recent studies stress that the impact of agricultural digitalization on income distribution and inequality is often nonlinear and may exhibit threshold characteristics. For China, several papers find that the development of agricultural digitalization affects the urban–rural income or consumption gap in a U-shaped or inverted U-shaped pattern along the digitalization path (; ; ). At low levels of agricultural digitalization, infrastructure constraints, limited coverage, and digital exclusion tend to reinforce existing disparities, leading to widening income and consumption gaps. As coverage and usage expand, inclusive applications and economies of scale enable agricultural digitalization to narrow these gaps, before potential re-widening occurs when high-end services and advanced sectors become concentrated in leading regions (; ).

Related evidence on digital finance and rural agricultural digitalization also points to nonlinear effects. Studies using national panel data for China document that digital finance can initially increase income inequality, particularly where access is biased toward higher-income or urban residents, but may reduce inequality once coverage becomes sufficiently broad and inclusive (; ). At the regional level, agricultural digitalization is found to promote urban–rural integration in a nonlinear way, with clear dual-threshold effects driven by differences in economic development (). At the local level, rural agricultural digitalization significantly affects intra-rural income distribution and has not yet led to inclusive growth in many areas (). These findings suggest that agricultural digitalization may generate different “phases” of impact along the development trajectory and that modeling it with linear specifications may obscure important stage-based dynamics. However, most of this work assesses the digital economy as a whole. More recent evidence on digital rural development further identifies significant stage characteristics, threshold effects, and regional heterogeneity in agricultural performance, reinforcing the need to consider nonlinear effects in the analysis of agricultural digitalization outcomes (). Whether agricultural digitalization displays similar nonlinear patterns in its effects on farmers' income remains largely unexplored.

2.3 Conditioning factors: human capital and urbanization

Human capital is widely recognized as a key determinant of whether farmers can access, understand, and effectively utilize agricultural digital technologies. Empirical studies show that education, digital literacy, and training significantly raise the probability that rural households use digital platforms, adopt agricultural digital technologies, and participate in e-commerce, thereby increasing both agricultural and total household income (; ; ). Evidence from China further indicates that digital literacy has a positive and significant effect on farmers' income, both directly and indirectly through improved information acquisition, technology adoption and participation in digital finance (; ; ).

From a broader perspective, policy reports and analytical work emphasize that strengthening farmers' digital skills and general education is necessary to prevent agricultural digitalization from amplifying existing inequalities and to ensure that rural residents can share in the benefits of digital innovation (; ). Inadequate human capital can create “capability bottlenecks” that limit adoption even when infrastructure is present and may lead to a situation where better-educated farmers capture most of the digital gains, while low-skilled farmers are left behind (; ). Despite this consensus on the complementary role of human capital, relatively few studies have examined whether there exists a threshold level of rural human capital beyond which the income effects of digitalization, particularly agricultural digitalization, become significantly stronger. This represents an important gap in the literature on digitalization and rural development.

Urbanization fundamentally reshapes the environment in which agricultural digitalization diffuses and affects rural incomes. For China, a large literature documents that agricultural digitalization interacts with urban–rural structures, influencing both the level and the spatial pattern of income gaps (; ; ). Several studies show that the impact of agricultural digitalization on the urban–rural income gap displays spatial heterogeneity, with stronger narrowing effects in central and western regions and weaker or ambiguous effects in more developed eastern regions (; ). Rural agricultural digitalization has also been found to play a role in promoting new-type urbanization by reallocating labor and capital and by improving access to digital public services ().

At the same time, research on rural agricultural digitalization and income distribution suggests that digitalization can generate new forms of inequality within rural areas, particularly when digital infrastructure and services are concentrated in peri-urban zones or county seats (). Urban–rural integration studies indicate that agricultural digitalization can facilitate factor mobility and service equalization across space, yet these effects are conditioned by regional fiscal capacity, industrial structure, and institutional arrangements (; ). Despite these insights, most existing work focuses on the overall digital economy or digital finance. Few studies explicitly examine how urbanization moderates the relationship between agricultural digitalization and farmers' income, or how this relationship varies across eastern, central and western regions with distinct industrial bases and resource endowments. Understanding this moderating and heterogeneous pattern is crucial for designing differentiated policies that align agricultural digitalization promotion with regional development strategies.

2.4 Research gaps

Overall, the existing literature provides valuable insights into how agricultural digitalization and the broader digital economy affect rural development, income distribution, and spatial structures. Nevertheless, several important research gaps remain. First, there is a measurement gap. Most empirical studies for China construct indices of the general digital economy, whereas agriculture-specific measur