How Does Digital Financial Inclusion Affect Rural Land Transfer? Evidence from China

Abstract

Farmers’ land transfer practices optimize the allocation of agricultural resources by transferring them to more efficient operators. This enhances agricultural productivity and advances rural revitalization. However, due to the lack of financial institution outlets in rural areas, the availability of financial services in rural areas is limited, which in turn hinders the transfer of rural land. This study examines the impact of digital financial inclusion, characterized by the deep integration of internet technology and financial services, on farmers’ land transfer behavior in China. The study uses data from the China Family Panel Studies (2012–2022) and provincial digital financial inclusion data. The results show that digital financial inclusion significantly promotes rural land transfer-out. The mechanisms reveal two pathways: (1) digital financial inclusion expands non-agricultural entrepreneurship by easing credit constraints and reducing reliance on land livelihoods; (2) it increases participation in commercial insurance, mitigating risks of land abandonment. Heterogeneity analysis reveals stronger effects in eastern China and among educated households. Theoretically, the study identifies the dual role of financial technology in reshaping rural land markets through credit access and risk management. Practically, it reveals how DFI influences land transfer behavior, providing a basis for the government to formulate policies that combine the two, ultimately enhancing the production capacity, operational efficiency, and market competitiveness of smallholder farmers. The findings offer global insights for developing countries that are leveraging digital finance to activate rural land markets and achieve digital financial inclusion.

1. Introduction

Land serves as a fundamental input and core asset in agricultural production [1], playing an irreplaceable strategic role in national food security and rural economic development. In rural communities, land is not only foundational to farmers’ livelihoods but also a critical vehicle for asset appreciation and participation in market-driven resource allocation [2]. Rural land transfer denotes the process wherein farm households maintain ownership and contractual rights, yet convey the management rights of their contracted cultivated land to new-type agricultural operation entities via market-oriented contractual agreements. International evidence consistently shows that rural land transfers—by reallocating resources to more productive operators [3,4]—significantly enhance agricultural productivity, mitigate the constraints of land fragmentation [5], and create dual income streams for smallholders through lease payments and off-farm employment, with empirical studies confirming substantial household income gains [6,7]. More importantly, once farmers transfer their land management rights, they no longer need to cultivate their own small plots of land, thereby freeing themselves from agricultural production and gaining the time and energy to engage in local non-agricultural work or seek employment elsewhere. This releases labor and corrects misallocation of resources, driving the transformation of the rural economy from subsistence agriculture to commercialized operations [8]. This process creates a three-dimensional value chain of efficiency improvements, income growth, and structural transformations [9]. Consequently, by regulating and guiding the transfer of rural land use rights, scattered small plots of land can be consolidated and entrusted to more capable entities for management, thereby achieving a more reasonable and efficient farm scale. This has become the core approach to unlocking agricultural potential and optimizing resource allocation. Global experience shows that the consolidation of scattered plots of land through rural land transfers has generally promoted the scale and technological application of agricultural production, significantly improved resource utilization efficiency and output levels, and effectively supported the development of agricultural modernization [5]. Diverse approaches demonstrate this, ranging from market-driven consolidation models in Europe and the U.S. to coordinated transfer mechanisms facilitated by organized entities, such as Japan’s JA (Japan Agricultural Cooperatives) and South Korea’s agricultural cooperatives, and China’s landmark “Three Rights Separation” reform. This Chinese reform upholds rural land collective ownership while separating ownership, contract, and management rights [10,11].

However, rural land transfers face severe constraints due to limited financial access. Remote areas, with their scarcity of financial institutions, experience chronic deficiencies in formal financial services. As financial inclusion is essential for rural revitalization and agricultural modernization, overcoming these barriers is imperative. Digital financial inclusion (DFI) leverages digital technologies to fundamentally restructure financial service models. By innovating products and optimizing processes, it overcomes traditional limitations, enhancing the reach and efficiency of financial services to mitigate imbalances and inadequacies in financial development. Its evolution constitutes a critical pathway for empowering rural development through finance [12]. Its core strengths—extensive geographic reach, lower operational costs, and efficient service delivery—significantly expand financial coverage and accessibility. These strengths effectively reach previously excluded populations, demonstrating unparalleled inclusivity. Digital technology enhances information flow and transaction transparency [13], substantially reducing providers’ operational and risk management costs and enabling farmers to access more affordable credit. Furthermore, integrated information services on DFI platforms help bridge the urban–rural information gap context-specific design features such as multilingual interfaces, voice-assisted navigation, and community-based “digital mentors” programs. For example, according to the World Bank’s 2025 report titled Digital Financial Inclusion in India: Evidence from the PMJDY Program, combining DFI access with basic digital literacy courses increases platform utilization by 42% among rural women. There are significant cross-national variations in DFI implementation. Developed countries, leveraging mature digital infrastructure and robust agricultural financial systems, can effectively provide technological and financial support for the scaling of farm operations. Taking the United States as an example, digital tools such as agricultural insurance work together to address challenges like diseconomies of scale and risk concentration that inevitably arise after farm consolidation. This powerfully propels the scaling process of agricultural operations post-consolidation. Many developing nations, however, still primarily rely on basic services such as mobile payments to achieve financial inclusion. These services have limited application in deepening rural land transfer markets. Leveraging its massive internet user base and strong policy support, China has achieved rapid advancements in mobile payments and digital credit. These developments make China a compelling research setting for examining how DFI enables rural land transfer.

Grounded in the Chinese context and based on a systematic literature review, this study develops an integrated theoretical and empirical analytical framework. The study aims to deeply examine the impact of DFI on farmers’ land transfer decisions and its underlying mechanisms, thereby elucidating the theoretical linkage between the two. Theoretically, the research clarifies the core concepts of DFI and rural land transfer and employs visualization techniques to depict their current developmental status. Subsequently, it constructs a theoretical model that proposes a core hypothesis. DFI significantly positively affects farmers’ land transfer behavior, primarily mediated through two channels: promoting non-agricultural entrepreneurship and enhancing participation in commercial insurance. Empirically, the study integrates macro and micro data to create a combined panel dataset. Utilizing Linear Probability Models (LPMs) and Logit models for baseline regression analysis, the study rigorously tests the causal relationship between DFI and farmers’ land transfer decisions. Furthermore, mediation effect models empirically identify and confirm the significant mediating roles of non-agricultural entrepreneurship and commercial insurance participation in the pathway through which DFI influences rural land transfers. Furthermore, through subsample regression analyses at both the regional and household levels, the article examines the heterogeneity in the effects of digital inclusive finance on rural land transfer.

This study makes significant advances in theoretical depth, mechanism identification, and methodological application, with core contributions manifested in the following aspects:

First, although existing research has extensively examined either the drivers of DFI development or the determinants of rural land transfer, the literature systematically investigating the direct facilitative effect of DFI on farmers’ land transfer decisions and its underlying mechanisms remains scarce. Previous studies predominantly rely on macro-level regional data for correlational analysis, failing to capture individual behaviors at the micro level. Pioneering a micro-level household perspective, this research innovatively integrates large-scale household panel data (CFPS) with regional DFI indices to construct a hybrid panel dataset. This approach precisely identifies and quantifies DFI’s impact on individual farmers’ land-out decisions, effectively overcoming aggregation bias inherent in macro-level studies and addressing the critical gap in micro-level causal evidence.

Second, regarding mechanism revelation, this study transcends the limitations of the conventional literature, which often focuses narrowly on direct financial functions like “enhanced payment convenience” or “improved basic credit access.” Instead, it innovatively proposes and rigorously validates two deeper, transformative transmission pathways: “off-farm entrepreneurship incentives” and “increased commercial insurance participation.” This finding demonstrates that DFI not only improves immediate financial accessibility but, more importantly, fundamentally reshapes farmers’ livelihood strategies and reduces their dependence on land by enabling entrepreneurship and strengthening risk protection capabilities. Consequently, it provides a novel and more explanatory theoretical framework for understanding DFI’s pivotal role in activating land factor markets.

Finally, deeply rooted in China’s institutional context of rural financial system reform and the “separation of land ownership rights, contract rights, and management rights”, this study’s policy significance extends beyond empirically confirming DFI’s efficacy in activating land markets and supporting agricultural-scale operations. Leveraging precise insights from micro-level mechanisms and spatial effects, it proposes a systematic, actionable, and differentiated policy framework—offering a Chinese approach to overcoming current land transfer bottlenecks.

This study focuses on investigating the impact of DFI on rural land transfer and its underlying mechanisms. The specific arrangement is structured as follows: First, it establishes the strategic significance of land transfer for agricultural and rural modernization and rural revitalization, introduces the connection between DFI and rural land transfer practices, and outlines key research questions and potential innovative contributions. The Section 2 systematically synthesizes and critiques existing research on both rural land transfer and DFI. It identifies gaps, revealing their potential theoretical linkages. The Section 3 defines core concepts and employs detailed data analysis to depict the actual landscape of DFI development and rural land transfer in China. The Section 4 integrates relevant theories to analyze the theoretical pathways and transmission mechanisms through which DFI influences rural land transfer and derives specific research hypotheses. The Section 5 details data sources and processing procedures and constructs the econometric model for analysis. The Section 6 presents baseline findings, comprehensively addresses endogeneity using instrumental variables, and analyzes robustness checks. The Section 7 employs mediation effect models to validate the proposed causal pathways. The Section 8 examines variations in effects across different regions and household types. The Section 9 analyzes spillover effects of digital financial inclusion on land transfer. The Section 10 summarizes key findings and derives actionable policy recommendations grounded in empirical evidence and real-world contexts.

2. Literature Review

2.1. Effect of Digital Financial Inclusion on Rural Development

The theoretical underpinnings of DFI stem from classical research on financial development and economic growth. Early scholarship established that the financial system promotes economic growth by optimizing resource allocation and managing risk through its core functions [14,15]. These functions enhance economic efficiency by alleviating liquidity constraints and smoothing consumption patterns [16,17]. From a corporate perspective, the alleviation of financing constraints lies at the core intersection between finance and the macroeconomy: it serves as the critical transmission hub through which the financial accelerator effect is realized [18], amplifying the impact of shocks on the macroeconomy; and it is also the fundamental microfoundation upon which the development of the financial system relies to enhance macroeconomic performance (such as growth and efficiency) [19], as the optimization of financial functions ultimately depends on improving the corporate financing environment and promoting effective investment.

Within this theoretical framework, inclusive finance’s evolution exhibits a distinct technology-driven character. The positive association between traditional financial inclusion and economic growth [20] is further amplified by digital technology [21]. A fundamental breakthrough lies in the transformation of credit assessment models, which substitute traditional hard metrics with soft information derived from behavioral data [22,23], and in the establishment of dynamic risk assessment frameworks [24]. This paradigm shift has directly improved the financing environment for small and microenterprises [25], lowering barriers to entrepreneurship and stimulating innovation [26], ultimately forming an economic growth transmission chain initiated by digital payment tools and mediated by entrepreneurial activities [27].

The significant promotional effect of digital finance on consumption lies in its ability to effectively alleviate the financial constraints faced by households. This is manifested through two complementary pathways: first, digital finance enhances intertemporal resource allocation capabilities, thereby releasing consumption demand that was previously suppressed due to liquidity constraints [28]. Second, the widespread adoption of digital payment methods significantly reduces operational complexity and psychological friction costs in transaction processes [29]. The alleviation of credit constraints activates latent consumption capacity, while the reduction of transaction frictions enhances the convenience and willingness to consume. These two mechanisms work synergistically, ultimately significantly enhancing households’ sensitivity and adaptability to economic environment changes such as income fluctuations at the micro level, thereby increasing consumption elasticity [30].

As the digital economy advances, the intrinsic connections between digital technology and rural land transfers, as well as the mechanisms through which DFI interacts with challenges in agriculture, rural areas, and farmers, are increasingly becoming cutting-edge research priorities. The existing literature widely agrees that inclusive finance serves as a pivotal institutional instrument for driving rural inclusive growth [31], with its core mechanism focused on effectively overcoming market failures, alleviating the adverse impacts of financial exclusion on low-income populations. Consequently, it contributes to reduced income disparities and more balanced regional development. Specifically, enhancing financial inclusion can effectively alleviate farmers’ credit constraints, promote productive investment, benefit low-income groups, and ultimately contribute to structural improvements in income inequality [32,33,34]. From a macro perspective, the deepening of financial institutions and diversifying financing channels significantly impact achieving poverty reduction goals, mitigating urban–rural development imbalances, and driving rural economic growth [35].

As the digital upgrade of traditional inclusive finance, the core advantage of DFI lies in its significantly more efficient operational mechanisms. It leverages mobile internet and big data technologies to overcome physical limitations and reach groups that traditional finance struggles to serve, anytime and anywhere; it uses alternative data and algorithmic models to achieve automated, precise risk control, significantly lowering credit thresholds and costs; and it simplifies processes through digital platforms, making the marginal cost of serving a massive user base approach zero. These fundamental changes collectively form the foundation of DFI’s efficient operations. Empirical research indicates that DFI not only directly promotes the convergence of urban–rural income gaps [36], but also exerts its influence through multiple indirect pathways. These include driving aggregate economic growth; substantially reducing transaction costs and collateral requirements for rural residents accessing financial services [37]; and promoting human capital accumulation by improving access to resources such as education and skills training. At the micro level, digital payment tools reshape farmers’ consumption patterns by reducing price sensitivity and the psychological discomfort of paying [29], enabling them to make smoother intertemporal consumption decisions. Within agricultural production, DFI is widely recognized as a key enabler for overcoming persistent financing constraints that hinder agricultural development and support agricultural industrialization and modernization [38].

However, given DFI’s nascent stage of development, some studies caution that its short-term impact on rural economic transformation may be limited due to multiple “hidden constraints” such as insufficient technological penetration depth, structural disparities in digital literacy—manifesting as age-based skill gaps, gendered access barriers, and linguistic limitations in user interface design—and uneven infrastructure coverage [39].

2.2. Factors Influencing Rural Land Transfer

Current academic research has made considerable progress in identifying the drivers of rural land transfer. This research has revealed that rural land transfer decisions are systematically shaped by a four-dimensional framework that encompasses property rights institutions, resource endowments, household characteristics, and behavioral psychology [40,41,42].

Land titling and registration programs significantly increase farmers’ willingness to participate in land rental markets by clarifying property boundaries and strengthening legal safeguards. This effect has been widely documented in land system reforms across developing countries [43]. Gender-specific analyses further indicate that enhanced property rights security has a particularly strong positive influence on female-headed households. This demonstrates the potential of institutional design to empower vulnerable groups [40]. China’s agricultural property rights reforms also validate the activating effect of institutional safeguards on the rural land transfer market, notably through reducing contract enforcement risks [44].

Due to land scarcity, farmers in economically disadvantaged regions often passively engage in informal rental markets [45]. In flat areas, decisions about transferring rural land are primarily driven by land quality and geographical location, whereas transaction costs emerge as the core constraint in hilly and mountainous regions [41]. This spatial heterogeneity is evident in the complex rural land transfer patterns observed in mountainous areas, where geographical factors influence contract design through mechanisms like transportation accessibility and disaster risk exposure mechanisms [46]. Additionally, the degree of land fragmentation significantly impacts transfer propensity [47].

Family life cycles and livelihood transitions substantially shape land transfer decisions. The aging of household heads and increased off-farm employment significantly raise the probability of rural land transfer. This intergenerational succession effect is particularly strong in agricultural transition economies [42]. Research in China’s ethnic minority regions reveals a negative correlation between household labor force size and rural land transfer willingness, while income diversification serves as a key enabling factor [48]. Against the backdrop of rural–urban migration, improved urban integration of migrant workers systematically promotes rural land transfer. Economic integration primarily drives decision-making, while social integration facilitates the expansion of transaction networks [49].

Departing from traditional approaches, behavioral and psychological factors are demonstrating a significant influence on rural land transfer choices. Substantial imitation effects operate within local networks, with lower-income groups exhibiting more pronounced conformity tendencies [50]. Emotional attachment to land is manifested through multidimensional mechanisms in transfer decisions: land dependence and rootedness inhibit land outflow, while land satisfaction exerts a positive moderating influence. These psychological responses vary significantly across generational cohorts [51]. Furthermore, agricultural income expectations, as dynamic predictive indicators, demonstrate significantly greater explanatory power than static demographic variables in analyzing influences on rural land transfer decisions [44].

2.3. Spatial Spillover Effects of Rural Land Circulation and Digital Financial Inclusion

Notably, digital financial inclusion inherently possesses strong spatial attributes and spillover potential. DFI leverages digital technologies to overcome geographical and temporal barriers, facilitating cross-regional resource flows and efficient data sharing, which significantly enhances interregional interactions [52]. Substantial empirical evidence demonstrates that DFI development generates not only local impacts but also significant spatial spillovers. DFI influences surrounding regions by promoting cross-regional labor mobility, technology diffusion, and commodity trade, thereby accelerating the spread of knowledge and experience [53]. Specifically, DFI significantly boosts farmer income while generating positive spatial spillovers [54,55]. Furthermore, positive spatial spillovers exist for rural revitalization [56], high-quality agricultural development [57], agricultural socialized services, industrial revitalization [58], and agricultural production clustering [59]. Additionally, DFI demonstrates negative spatial spillovers (i.e., poverty reduction effects) in alleviating rural relative poverty [60]. Some studies even suggest that DFI’s spatial spillover effect on neighboring regions’ farmer income growth may exceed its local direct impact [55].

The spatial spillover effect is increasingly recognized by scholars as a critical factor influencing rural land circulation. Theoretical research indicates that individual decision-making is not entirely independent but exhibits significant spatial dependence. The aggregation of micro-level behaviors and spatial interactions among individuals shapes macro-level decision patterns with socialized characteristics [61]. This spatial interdependence implies that observed choices in one region, such as land circulation decisions, often demonstrate similarity to choices in neighboring regions [62,63]. Within China’s rural context, farmer decision-making, including land circulation, frequently manifests “herd behavior,” closely tied to households’ capacity to access, interpret, and utilize information. Empirical studies further confirm significant spatial agglomeration and spatial spillover effects in China’s regional economic activities, including rural land circulation [64].

In summary, existing studies have conducted extensive empirical analyses of the factors influencing rural land transfers and examined the positive effects of DFIs on farmers’ income growth, entrepreneurial activity, consumption patterns, and risk management capabilities. As a pivotal institutional mechanism for addressing rural financial repression—a systemic condition in developing economies where government policies (e.g., interest rate controls, directed credit programs, and underdeveloped rural banking infrastructure) artificially suppress market-driven financial allocation, thereby perpetuating credit scarcity for agricultural producers and smallholders [25,38]—the core pathways through which DFI promotes inclusive growth, by alleviating credit constraints, reducing transaction costs, and empowering human capital development, have received preliminary empirical validation.

However, specialized and systematic research on the direct impact of DFI on rural land transfer systems, along with its underlying mechanisms, remains relatively scarce. The existing literature tends to focus on either macro-level policy discourses or restricts analysis to localized effects of individual financial instruments on specific land transfer modalities. Comprehensive theoretical frameworks and rigorous empirical deconstructions of these dynamics are notably absent.

First, this study conceptualizes DFI as an integrated empowerment system to systematically examine its multidimensional pathways of influence, including alleviating credit constraints, reducing transaction costs, enhancing information symmetry, optimizing risk management, and strengthening capabilities of market participants, on the scale, efficiency, contractual models, and equity outcomes of rural land transfers. Second, through rigorous empirical testing, we aim to unravel the intrinsic causal chains connecting DFI to land transfer outcomes. This includes investigating how DFI reshapes non-agricultural entrepreneurial decision-making processes and, consequently, influences land transfer behaviors among rural stakeholders.

3. Background and Current Status

3.1. Digital Financial Inclusion

3.1.1. The Concept of Digital Financial Inclusion

The United Nations formally introduced inclusive finance in 2005 during the “International Year of Microcredit.” Its core objective is to establish universally accessible financial ecosystems. This approach aims to eliminate price and non-price barriers by strengthening institutional frameworks to ensure equitable access to financial services for diverse populations and to provide tailored credit support to traditionally underserved groups. Since 2006, China has advanced inclusive finance through microcredit institution establishment, dedicated “inclusive finance departments within financial organizations, and pilot rural “two rights” mortgage programs. Nevertheless, traditional finance exhibits inherent limitations in serving micro-enterprises and low-income populations. These groups’ key characteristics—including geographic dispersion, small operational scale, information asymmetry, and collateral inadequacy—result in elevated service costs, ineffective risk mitigation, and compromised commercial viability [65].

The convergence of internet infrastructure and artificial intelligence has fundamentally restructured inclusive finance delivery mechanisms. As defined by the G20 High-Level Principles for Digital Financial Inclusion (2016), DFI encompasses all initiatives that advance inclusive finance objectives through digital financial services, marking its evolution into a digitally enhanced phase. DFI operates through synergistic collaboration between traditional financial institutions and technology firms, leveraging digital innovations to transform business models in payment systems, financing mechanisms, and investment platforms.

This paradigm overcomes traditional constraints through three dimensions: First, service carrier innovation eliminates physical barriers via digital payments and online lending. Second, technological enablers use social media and e-commerce data streams to create credit assessment frameworks, significantly improving risk management. Finally, ecosystem collaboration through platforms like Taobao and WeChat enhances user engagement and transactional transparency. Empirical studies confirm that this technological transformation reduces operational costs and credit risks, and addresses traditional service gaps through big data analytics. Consequently, it improves financial service coverage, accessibility, and satisfaction, effectively transitioning inclusive finance from policy concept to sustainable commercial practice [13].

3.1.2. Historical Process of Digital Financial Inclusion

China’s development of DFI is an evolutionary process characterized by deepening integration between internet technology and financial services. This process has progressed through three distinct historical phases, which are defined by technological and institutional milestones.

The initial phase (late 1990s–2004) featured the digital transformation of traditional financial institutions. In 1996, the Bank of China launched the nation’s first banking website. Then, in 1998, the Bank of China and China Merchants Bank introduced the first online banking services. These initiatives prompted commercial banks to transition core operations, including offline transactions and regulatory compliance, to digital platforms. This shift significantly reduced operational costs while enhancing efficiency. During this stage, technology functioned solely as a supporting tool for existing financial models, with technology companies limited to providing technical services rather than acting as financial entities. This established the paradigm of “technology enabling without business disruption” [65].

The acceleration phase (2004–2013) was characterized by innovation breakthroughs driven primarily by financial technology companies. The creation of Alipay, a third-party payment platform, in 2004 signaled the start of technology firms providing inclusive financial services, such as mobile payments, through their proprietary ecosystems. The introduction of Yu’e Bao, an online money market fund, in 2013 fueled explosive market growth and significantly accelerated the development of digital finance [65]. During this period, internet giants leveraged their platform advantages to fundamentally reshape service scenarios, with mobile payments and online lending emerging as central vehicles for extending inclusive financial services.

The ongoing integration phase (2013–present) demonstrates collaborative evolution among various stakeholders. Traditional financial institutions achieved profound transformation through independent innovation and cross-sector collaboration. The Bank of China pioneered new digital business models by launching its open platform in 2013, while the subsequent establishment of internet banks in 2014, notably China Minsheng Bank’s Direct Banking unit, significantly accelerated the comprehensive digitization of banking operations. Meanwhile, technology companies have empowered financial institutions by supplying them with artificial intelligence and big data technologies, while also leveraging ecosystem data to build credit risk management systems for their financial subsidiaries. This synergy ultimately established an interdependent ecosystem where traditional banks, technology firms, and internet platforms continuously expand the scope of inclusive services including digital payments, insurance, and investment through technological innovations such as blockchain [37].

3.1.3. Current Situation of Digital Financial Inclusion

In recent years, DFI has achieved transformative development, propelled by infrastructure initiatives, including the Broadband Village program and 5G base station deployment, as well as accelerated digital technology iteration and innovation. Figure 1 illustrates this development through provincial and municipal DFI index averages and growth rates. Overall, all indices demonstrated sustained growth from 2011 to 2023. The provincial average surged from 40 in 2011 to 220.01 in 2015 while municipal averages rose from 49.4 to 170.23 during the same period. These figures represent increases of more than threefold. Subsequently between 2015 and 2023, DFI development moderated, with provincial and municipal indices advancing at approximately 8.5% annually, ultimately reaching 393.7 for provincial and 297.67 for municipal averages by 2023.

A regional analysis reveals progressive development across China’s four major regions, though tiered disparities persist. As depicted in Figure 2, all regions followed similar growth trajectories. However, the eastern region consistently led in DFI, while the western region lagged behind. In 2023, the eastern region dominated with an index of 428.14, followed by central, western, and northeastern regions in descending order. Notably, the western region, despite having the lowest baseline, achieved nearly twelvefold growth, significantly exceeding the eastern region’s rate. This suggests DFI policies disproportionately benefit underdeveloped areas, contributing to the mitigation of the regional gap.

As illustrated in Figure 3, county-level sub-indices measuring coverage breadth, usage depth, and digitization level revealed distinct evolutionary patterns. From 2014 to 2016, all indicators demonstrated parallel and steadily increasing trends. Significant divergence among them materialized after 2016. Following a period of rapid expansion since 2017, the coverage breadth of China’s DFI approached the upper limit of population scale, constraining further expansion potential and entering a period of relative stability. Concurrently, the sector’s developmental focus shifted decisively toward deepening application penetration and service integration, evidenced by sustained enhancements in usage depth and digitization level. After 2018, the growth of digitization slowed, indicating that the application of digital technologies in inclusive financial services required adjustment and more innovative models. Notably, the recent property value downturn in China, which has led to lower real estate values and contributed to slowing economic growth, may have implications for DFI and rural economies. A weaker real estate market could potentially reduce collateral values for loans, affecting financial institutions’ lending behaviors and possibly increasing the demand for alternative, digital-based financial services in rural areas. Moreover, as the broader economy slows, there might be a greater emphasis on leveraging DFI to stimulate rural economic activities and offset some of the negative impacts from the real estate sector. In contrast, DFI usage depth demonstrated stronger and more resilient growth. Though it temporarily stagnated in 2018 due to macroeconomic policy adjustments, it quickly recovered after this period. By 2023, the usage depth index had climbed to 151, significantly surpassing both coverage breadth and digitization-level metrics and establishing itself as the core driving force propelling the deepening development of DFI.

3.2. Rural Land Transfer

3.2.1. The Concept of Rural Land Transfer

Fundamentally, rural land transfer constitutes a market-based reallocation of land rights. Broadly defined, it encompasses ownership transfers, usage rights transfers, and functional conversions. When defined more narrowly, it focuses specifically on transfers of usage rights. Within China’s public ownership framework, the Constitution and Land Management Law explicitly assign land ownership to the state or collectives. Agricultural land specifically denotes farmland, forestland, grassland, and other land legally designated for agricultural use that is collectively or state-owned. Farmers legally hold only usage rights, establishing a statutory dual structure of ownership and usage rights. To safeguard agricultural security and farmers’ interests, the state has institutionalized a collective ownership system with household contracts, which permits the transfer of usage rights while maintaining immutable ownership.

Due to the strategic importance of cultivated land and the unique characteristics of its transfer market, this study focuses on transfers of cultivated land management rights, which households obtain at no cost under the household contract responsibility system. It explicitly excludes agricultural lands with different ownership structures such as forest or orchard lands, as well as non-agricultural lands including residential plots. Under China’s institutional innovation of separating collective ownership rights, household contract rights, and land management rights, this research defines rural land transfer exclusively as transactions where contracted households transfer farmland management rights to new agricultural operators through market-based instruments like subleasing or leasing within legally prescribed timeframes. To address the structural imbalance in land markets characterized by persistent excess demand and constrained supply, the analysis concentrates on rural land transfer-out behavior: farmers’ economic decisions to retain contract rights while transferring management rights to third parties such as professional farming entities or agricultural enterprises. This conceptualization captures the market-based allocation of land as a production factor while establishing the legal-conceptual foundation for empirically testing how DFI influences supply-side behavioral mechanisms.

3.2.2. Historical Progression of Rural Land Transfer

• Phase I: Institutional Constraints and Practical Breakthroughs (1950–1987)

During the era of China’s planned economy (1953–1992), the transfer of rural land faced stringent legal restrictions. The Constitution (1982) explicitly prohibited rural land transfers, reinforced by the National Rural Work Conference Minutes banning contracted rural land transfers. Although this policy was designed to safeguard food security, it inadvertently spurred institutional innovation. The household contracting system functionally separated land ownership from usage rights [66]. Collective organizations retained legal ownership while farmers gained actual control over the land. This duality of legal ownership and actual control created space for informal land cultivation practices in coastal regions. Despite legal prohibitions, practical demands for rural land transfer began challenging institutional boundaries [67].

• Phase II: Legalization and Conflict Resolution (1988–2013)

The 1988 Constitutional Amendment, which abolished the system of rural land use right transfer, marked China’s initiation of a transition toward regulated marketization. This period exhibited dual developmental trajectories. Institutionally, the Rural Land Contract Law (2002) first granted legal recognition to transfer methods including subleasing and leasing. Meanwhile, the Property Rights Law (2007) explicitly classified contractual rights as usufructuary property rights, fundamentally transforming contractual entitlements into property-based rights. Practically, agricultural tax burdens in the 1990s induced coercive transfer models like government-led centralized subleasing, prompting central authorities to reaffirm voluntary, compensated transfer principles [67].

• Phase III: Property Rights Structuring and Efficiency Leap (2014 to present)

In 2014, China’s property rights separation reform institutionalized the “three-rights separation” framework [67], reconfiguring rural land transfer functions. This framework designates collective ownership as the guarantor of institutional stability, assigns social security functions to household contracting rights, and empowers land operating rights to release factor value. The Revised Rural Land Contract Law (2018) achieved systemic breakthroughs by endowing operating rights with financial attributes including mortgages and equity conversions, thereby establishing formalized factor capitalization pathways. Current reforms strategically transition from scale expansion to quality enhancement, targeting three objectives: comprehensive land trusteeship system coverage in key agricultural zones; enhanced cross-regional allocation efficiency through digital platforms; and smallholder integration into modern production systems via shareholding cooperatives [68]. This equilibrium between market efficiency and social protection offers developing economies a referential model for balancing land system equity and efficiency.

3.2.3. Current Status of Rural Land Transfer

According to data from China’s Ministry of Agriculture and Rural Affairs (MARA), we find China’s rural land transfer market has demonstrated substantial expansion and deepening over the past two decades. From 2005 to 2022, the total area of transferred farmland under household contracts exhibited sustained rapid growth, increasing from under 10 million mu to nearly 60 million mu (Figure 4). Concurrently, the rural land transfer rate—calculated as transferred area divided by the total contracted farmland area—rose steadily from single-digit levels in 2005, surpassing 35% by 2016 and maintaining relative stability since then. This trajectory signals accelerated optimization of agricultural land allocation and signifies structural transformation beyond traditional small-scale, fragmented farming models.

According to data from China’s Ministry of Agriculture and Rural Affairs (MARA), rural land transfer modalities exhibit distinct compositional patterns. In 2023, household leasing or subletting dominated, constituting 88.87% of the total transferred area and growing at a rate of 3.57%, which exceeded the overall transfer growth rate of 2.62%. This confirms short-term leases as the prevailing market mechanism. Shareholding cooperatives accounted for 5.09% of transfers and maintained positive growth at 0.58%, despite operating on a smaller scale. This indicates ongoing experimentation with property rights reform. Other transfer methods represented 6.04% but contracted substantially by 8.15% during 2023. These metrics collectively demonstrate high market concentration in leasing or subletting arrangements, reflecting strong preference for flexible, low-transaction-cost mechanisms.

Regarding entities assuming rural land transfer, the data from MARA paints a picture of diversity with uneven distribution. Traditional smallholders remained the largest recipients in 2023, receiving 46.49% of transferred land. However, new agricultural operators demonstrated stronger growth momentum: professional farmer cooperatives absorbed 20.73% while family farms received 15.80%. Collectively, they accounted for over 36%. Notably, their respective land inflow growth rates reached 4.87% and 5.94%—significantly surpassing the smallholder growth rate of 1.94%—evidencing steady progress in agricultural organization and scaling. Agricultural enterprises acquired another 10.19% of transferred land. This coexistence structure illustrates operational diversity and scaling potential during China’s agricultural transition.

4. Hypotheses Development

DFI represents a deep integration of fintech and inclusive finance principles, fundamentally transforming the logic of rural land resource allocation. This approach leverages mobile payments, internet banking, and blockchain technologies to overcome geographic barriers and credit assessment limitations inherent in traditional financial services [12,69]. By delivering cost-effective financial solutions to small-scale entities historically excluded from formal financial systems, this model enhances operational efficiency while exerting systemic influence on rural land transfer markets. Through agricultural big data integration, DFI precisely identifies farmers’ financing needs during transitions to non-farm employment and agricultural scaling operations. This targeting provides vital capital supporting rural land transfers, enabling market-driven reallocation of fragmented land resources.

Regarding financial inclusion, DFI addresses traditional institutions’ service gaps in rural areas. Physical branch scarcity and limited risk assessment methods have historically subjected farmers to severe credit constraints. Digital solutions counter this through remote identity verification, multidimensional credit profiling, and real-time transaction monitoring, significantly improving financial accessibility [70]. This enhanced accessibility generates dual effects: First, it directly enables productivity-enhancing investments in transferred land parcels. Second, it strengthens farmers’ risk tolerance in rural land transfer markets by alleviating liquidity constraints [71]. Importantly, improved financial accessibility interacts synergistically with information environment optimization. Rural land transfers constitute contractual processes where information asymmetry in decentralized smallholder economies has long impeded market development through high transaction costs [72]. DFI platforms substantially reduce search and matching costs for both suppliers and demanders in rural land transfer through efficient information aggregation and dissemination mechanisms [73], consequently increasing transaction probability. Based on the above analysis, this study proposes the following research hypotheses:

Hypothesis 1. 

The development of DFI can promote rural land transfer.

As previously established, DFI’s core value lies in delivering cost-effective financial services to historically excluded farmers through digital technologies. This innovative model activates non-agricultural entrepreneurship through dual pathways [74]: First, by integrating agricultural big data to build precision risk models, DFI significantly alleviates entrepreneurial credit constraints [27]. Digital credit assessment mitigates traditional lenders’ collateral-based restrictions, enabling entrepreneurs to overcome initial capital barriers. Second, mobile-enabled information networks dismantle rural “information silos.” DFI is claimed to enhance financial literacy through its integration of accessible educational tools, real-time information dissemination, and behavioral reinforcement mechanisms. Digital platforms embed interactive resources like budgeting apps and investment simulators, enabling self-paced learning of financial concepts such as compound interest. Simultaneously, DFI reduces information asymmetry by delivering instant market updates and product comparisons, fostering informed decision-making aligned with behavioral economics principles. Additionally, features like transaction tracking, automated savings goals, and repayment reminders cultivate financially responsible habits, with empirical evidence linking frequent DFI engagement to improved credit management. By transforming passive knowledge into active skills, DFI bridges theoretical literacy and practical application. Entrepreneurial policies, market intelligence, and financial literacy now reach farmers efficiently via digital platforms, drastically reducing information search costs [75,76]. This optimized information environment cultivates commercial awareness while enhancing opportunity recognition capabilities [77,78], empowering farmers to transition from passive adaptation to proactive market engagement. Significantly, entrepreneurial activities require substantial labor reallocation to non-agricultural sectors, reducing agricultural labor input. The resulting income substitution effect diminishes survival dependency on landholdings. Moreover, DFI’s proliferation stimulates regional economic growth, generating non-farm employment that accelerates labor migration [79]. These structural transformations reposition land as a tradable production factor.

Hypothesis 2. 

The development of DFI can increase the probability of farmers engaging in non-agricultural entrepreneurship, thereby promoting rural land transfers.

The deep application of digital technologies has substantially reduced barriers to rural financial service access [33,69]. Specifically, mobile payment systems and online credit facilities enable farmers to conveniently secure land improvement financing. Concurrently, blockchain-powered rural land transfer platforms significantly enhance transactional transparency while mitigating trust deficiencies inherent in traditional models. This optimized financial ecosystem has catalyzed innovative products—including land mortgage loans and transfer performance insurance—while establishing institutional safeguards for large-scale land management. Crucially, advancing financial inclusion is simultaneously restructuring risk management frameworks [30]. The Internet Plus model transcends physical branch limitations, enabling commercial insurance penetration into remote areas via mobile interfaces [75]. Supported by big data analytics, insurers achieve operational cost reduction through precision risk assessment while developing customized products for differentiated protection needs.

Furthermore, widespread digital platform adoption is fundamentally transforming farmers’ risk perceptions. Continuous dissemination of financial literacy through user interaction mechanisms steadily enhances economic decision-making capabilities [80]. Abundant product information and peer reviews dramatically lower information search costs, gradually correcting traditional insurance misconceptions. This cognitive evolution effectively alleviates information asymmetry-induced insurance participation reluctance [81], encouraging proactive use of commercial insurance for livelihood risk management [82]. Notably, land’s dual identity as both economic asset and social safety net has historically positioned it as a core buffer against unemployment, retirement insecurity, and livelihood risks. This protective characteristic frequently constrains rural land transfer decisions [83]. Particularly when non-agricultural employment proves unstable, farmers preferentially retain land as a survival safeguard [50]. As commercial insurance networks mature, their expanding coverage of pension, health, and unemployment risks progressively substitutes land’s protective functions. With pension pressures alleviated by retirement insurance and medical burdens shifted through health coverage, land’s social security utility continuously diminishes. This functional substitution profoundly reshapes risk calculus: insured farmers reduce land-dependent security reliance in favor of generating property income through rural land transfers. Thus, DFI effectively addresses core institutional constraints by establishing an endogenous mechanism where insurance supplants land as protection—creating essential preconditions for market-based agricultural land allocation.

Hypothesis 3. 

The development of DFI can increase farmers’ participation in commercial insurance, thereby promoting rural land transfers.

5. Empirical Strategy

5.1. Data Sources

This study uses longitudinal data from six waves (2012, 2014, 2016, 2018, 2020, and 2022) of the China Family Panel Studies (CFPS) database, which is administered by the Institute of Social Science Survey at Peking University. As a nationally representative household survey, the CFPS comprehensively tracks Chinese households across multiple dimensions including household composition, educational attainment, employment status, income-expenditure structures, asset ownership, agricultural/industrial operations, land use dynamics—with particular attention to rural land transfers—and social security participation. Using a stratified random sampling framework, the survey covers all 31 province-level administrative divisions in China to ensure robust representativeness at the national, provincial, and community levels.

DFI development is measured using the authoritative Peking University Digital Financial Inclusion Index, co-developed by Peking University’s Digital Finance Research Center and Ant Group. The index evaluates DFI through three primary dimensions: coverage breadth, usage depth, and digitization level. These dimensions are further decomposed into 33 granular indicators that span account penetration, payment services, credit utilization, insurance adoption, and mobile service accessibility. Leveraging Ant Group’s real-time transactional data, the index establishes an objective evaluation framework distinguished by its extensive geographical coverage, methodological rigor, and comprehensive indicator selection. Widely recognized as the definitive metric for China’s digital financial inclusion landscape, it provides scientifically valid and empirically robust measurement for scholarly investigation.

We use stata18.0 to process and analyze the data. Data processing followed sequential procedures: First, we retained only households possessing farmland during surveyed years. Second, provincial-level DFI indices were matched to corresponding survey years. Third, observations with missing values or statistical outliers in core variables—particularly rural land transfers, the DFI index, and key covariates—were systematically excluded. Finally, cross-sectional datasets were longitudinally merged to construct an unbalanced panel. The resultant dataset includes 41,755 household-year observations spanning six survey waves: 2012 (The sample size is 7793), 2014 (The sample size is 7702), 2016 (The sample size is 8004), 2018 (The sample size is 7717), 2020 (The sample size is 5615), and 2022 (The sample size is 4924). The data cover households from all 31 province-level divisions across mainland China.

5.2. Empirical Model

To examine the impact of DFI on rural land transfers, this study constructs the following model and establishes a mediation effect model:

$$Land\_Transfe{r}_{ijt}={\alpha }_0+{\alpha }_1\cdot \lg \_Inde{x}_{jt}+{\gamma }_1{X}_{ijt}+{\delta }_1{W}_{jt}+{\nu }_t+{\mu }_i+{\theta }_j+{\epsilon }_{ijt}$$

(1)

$$M_{ijt}={\beta }_0+{\beta }_1\cdot \lg \_Inde{x}_{jt}+{\gamma }_2{X}_{ijt}+{\delta }_2{W}_{jt}+{\nu }_t+{\mu }_i+{\theta }_j+{\epsilon }_{ijt}$$

(2)

$$Land\_Transfe{r}_{ijt}={\lambda }_0+{\lambda }_1\cdot \lg \_Inde{x}_{jt}+{\lambda }_2{M}_{ijt}+{\gamma }_3{X}_{ijt}+{\delta }_3{W}_{jt}+{\nu }_t+{\mu }_i+{\theta }_j+{\epsilon }_{ijt}$$

(3)

In this model, i denotes the household, j denotes the province where the household is located, and t denotes the survey year. Land_Transfer_ijt represents the dependent variable, indicating the land transfer decision of the i-th household in province j in year t. lg_Index_jt represents the core explanatory variable, indicating the level of DFI development in province j in year t. X_ijt are household-level control variables. W_jt are provincial-level control variables. ν_t is the year fixed effect. μ_i is the household fixed effect. θ_j is the provincial fixed effect. And ε_ijt is the random disturbance term. Additionally, M_ijt represents the two mediating variables of interest in this study, including non-agricultural entrepreneurship and participation in commercial insurance. The coefficient α₁ in Equation (1) represents the total effect of DFI on rural land transfer, the coefficient β₁ in Equation (2) represents the effect of DFI on the mediator variable M_ijt, and the coefficient λ₂ in Equation (3) represents the effect of the mediator variable on rural land transfer.

The specific definitions of the relevant variables are provided below.

Land_Transfer: Rural land transfer refers to the process where farm households retain ownership and contractual rights to their contracted cultivated land, yet transfer management rights to new-type agricultural operation entities through market-oriented contractual arrangements. This phenomenon is primarily measured by whether micro-level farm households have participated in land transfer activities. This binary dummy variable measures participation in rural land transfers at the household level. It is assigned a value of 1 if the household engaged in compensated rural land transfer activities during the survey year, as determined by affirmative responses to the China Family Panel Studies (CFPS) Land Module questionnaire item “Have you leased your land to others?” This operational definition explicitly excludes informal rural land transfers among relatives and friends as well as non-agricultural land conversion activities. Cases without such market participation receive a value of 0.

lg_Index: Digital financial inclusion (DFI) utilizes digital technologies to fundamentally transform financial service delivery models. Through product innovation and process optimization, it transcends traditional constraints, expanding the accessibility and efficiency of financial services to address developmental imbalances and gaps. Its progression represents a pivotal mechanism for advancing rural development through financial empowerment. In this study, the Peking University Digital Financial Inclusion Index is employed to measure the developmental level of digital financial inclusion. The original index, developed by Peking University’s Digital Finance Research Center in collaboration with Ant Group, is the Peking University Digital Financial Inclusion Index. The index comprehensively measures regional DFI development across three key dimensions: coverage breadth, usage depth, and digitization degree, measured by mobility and affordability. To mitigate the effects of heteroskedasticity and skewed distribution, the original index values were transformed by adding one and then taking the natural logarithm. The resulting variable reflects the regional penetration level of digital financial services, where a higher value indicates a more advanced level of DFI development in the area.

nonagricultural_entrepreneur: The theoretical framework of this study postulates that nonagricultural entrepreneurship is a key channel through which DFI influences rural land transfers. Non-agricultural entrepreneurship refers to the creation, operation, and expansion of economic enterprises outside the agricultural sector, particularly in rural areas. It encompasses innovative and risk-bearing activities that leverage local resources (e.g., water, woodlands, skilled labor, and infrastructure) to engage in manufacturing, services, trade, tourism, and agro-processing. Farmer land transfer decisions are fundamentally shaped by the reallocation of household labor toward non-agricultural activities, representing a core aspect of joint household decision-making. Household laborers make these non-agricultural reallocation decisions by holistically evaluating their own endowments alongside those of other family members, ultimately aiming to maximize collective household welfare. Consequently, this study incorporates a household-level dummy variable for nonagricultural entrepreneurship. Households engaged in individual or private business ventures are designated as nonagricultural entrepreneurial households and coded as one, with all other households assigned a value of zero.

lg_commercial_ins: Commercial insurance refers to risk management products offered by private-sector insurance companies, distinct from government-mandated social insurance programs. It operates on a voluntary basis where policyholders pay premiums in exchange for financial compensation against specified risks. Purchasing commercial insurance can alleviate psychological constraints on farmers, such as their reliance on land as a primary retirement safeguard and their perception of land as an essential safety net against risks associated with non-agricultural employment. This study employs the continuous variable of commercial insurance expenditure to capture household participation in commercial insurance markets. This measure is derived from the household expenditure module of the CFPS survey. It utilizes the survey item that asks about the total amount spent on commercial insurance purchases during the previous year. To address concerns related to heteroskedasticity, skewed distribution, and the presence of zero expenditures, the original insurance expenditure value undergoes a transformation where one is added before applying the natural logarithm.

Control variables: This study’s analysis systematically integrates control variables at multiple levels to rigorously mitigate potential omitted variable bias. Previous research robustly demonstrates the significant influence of household characteristics on rural land transfer behaviors. Given the distinctive nature of agricultural production, where the household constitutes the fundamental production unit, household size directly reflects potential agricultural production capacity. Simultaneously, overall household income levels significantly impact decisions regarding continued engagement in agricultural activities. Furthermore, the household head usually makes decisions about agricultural operations and management. Their gender, age, and educational attainment play critical roles in shaping rural land transfer choices. To bolster the reliability of our findings, this study therefore incorporates key household-level controls: total household size, aggregate household income, and the gender, age, and educational level of the household head. Data for all household-level control variables originate directly from the CFPS questionnaire. Recognizing that rural land transfer decisions represent resource allocation behavior intrinsically embedded within specific institutional contexts, provincial economic scale fundamentally shapes the opportunity structure for non-agricultural employment and entrepreneurship. Concurrently, the intensity of fiscal resource allocation directly influences the quality of rural infrastructure provision, and natural disaster risks significantly constrain farmer land disposition strategies. To precisely isolate the net effect attributable to DFI, this study consequently introduces provincial-level control variables capturing regional gross domestic product, total fiscal budget expenditures, and monetary losses due to natural disasters. Data for all provincial-level controls are sourced exclusively from the officially published China Statistical Yearbook. Following standard practice to normalize distributions, these provincial variables are incorporated into the econometric model after logarithmic transformation. Descriptive statistics for all major variables in the article are shown in

Table 1. Descriptive statistics for variables.

	Variable	Definition	Mean	SD
Dependent variable	Land_Transfer	Whether the rural land has been transferred (1 for yes, 0 for no)	0.183	0.386
Explanatory variable	lg_Index	Logarithm of the Digital Financial Inclusion Index	5.375	0.473
Mechanism variables	lg_commercial_ins	Logarithm of commercial insurance expenditures	1.769	3.324
	nonagricultural_entrepreneur	Whether in non-farm business start-up status (yes takes 1, no takes 0)	0.088	0.284
Household-level control variables	educated	Educational attainment (1 for Bachelor’s degree and above, 0 for less than Bachelor’s degree)	0.030	0.171
	income_p	Per capita household income (in RMB 10,000)	1.400	2.463
	household_size	Number of family members	4.083	1.927
	gender	Gender of household decision maker (1 for males, 0 for females)	0.577	0.494
	decisionmaker_age	Age of household decision maker	51.482	13.235
Province-level control variables	lg_GDP	Logarithm of GDP	10.100	0.766
	lg_budget_exp	Logarithm of fiscal budget expenditures	8.638	0.481
	lg_disaster_loss	Logarithm of direct economic losses due to natural disasters	4.583	0.981

Data source: China Household Survey Tracking Questionnaire, Peking University Digital Financial Inclusion Index, and China Statistical Yearbook.

.

6. Empirical Results

6.1. Baseline Regression Results

Table 2. Impact of digital financial inclusion on rural land transfer.

Land_Transfer	LPM (1)	LPM (2)	LPM (3)	Logit (1)	Logit (2)	Logit (3)
lg_Index	0.081 **	0.109 ***	0.128 ***	1.726 ***	2.017 ***	2.228 ***
	(0.040)	(0.042)	(0.043)	(0.668)	(0.701)	(0.755)
educated		−0.000	−0.000		−0.013	−0.010
		(0.011)	(0.011)		(0.112)	(0.112)
income_p		0.006 ***	0.006 ***		0.049 ***	0.050 ***
		(0.001)	(0.001)		(0.013)	(0.013)
household_size		−0.010 ***	−0.010 ***		−0.102 ***	−0.102 ***
		(0.001)	(0.001)		(0.026)	(0.026)
gender		−0.030 ***	−0.030 ***		−0.272 ***	−0.274 ***
		(0.005)	(0.005)		(0.040)	(0.041)
decisionmaker_age		0.002 ***	0.002 ***		0.012 ***	0.012 ***
		(0.000)	(0.000)		(0.002)	(0.002)
lg_GDP			−0.012			−0.076
			(0.032)			(0.330)
lg_budget_exp			−0.051 *			−0.368
			(0.030)			(0.401)
lg_disaster_loss			−0.002			−0.023
			(0.002)			(0.030)
Household Effects	Yes	Yes	Yes	Yes	Yes	Yes
Province Effects	Yes	Yes	Yes	Yes	Yes	Yes
Year Effects	Yes	Yes	Yes	Yes	Yes	Yes
Observations	41,755	40,339	40,216	41,755	40,339	40,216

Notes: *** p < 0.01, ** p < 0.05, * p < 0.10. Standard errors clustered at province level in parentheses.

systematically reports the stepwise regression results of DFI on farmers’ land transfer decisions in columns (1) to (6). The Logit model coefficients are untransformed; no further special explanation is provided in the text. Regardless of control variable inclusion, the empirical results indicate that the coefficient for the DFI index remains significantly positive and statistically significant at the 5% level. The coefficient value displays a monotonically increasing trend as control variables are added, confirming result robustness. After including all control variables, Column LPM (3) shows that a one-unit increase in the DFI index increases the probability of rural land transfer by an average of 12.8 percentage points. These results strongly demonstrate that DFI significantly impacts farmers’ land transfers and activates the land factor market, thus validating Research Hypothesis 1. Although the Logit model coefficients are untransformed, their original estimates are also highly significantly positive, confirming the robustness of the positive effect.

Regarding control variables, household income significantly promotes rural land transfer, indicating that economic capital accumulation reduces reliance on agricultural production. However, household size expansion inhibits transfer behavior, consistent with labor supply imposing rigid constraints on land disposal. Households with male heads exhibit a lower rural land transfer probability, reflecting agricultural operational stickiness stemming from male physical advantages. Each one-year increase in the household decision-maker’s age raises the transfer probability by 0.2 percentage points, corresponding to declining agricultural labor capacity due to aging. The household head’s educational attainment did not reach statistical significance, suggesting that educational returns have a dual offsetting effect on land decisions, where the increased attractiveness of non-agricultural employment due to improved human capital is balanced by stronger agricultural management capabilities. Among provincial variables, fiscal expenditure shows a weak negative association, suggesting government intervention may crowd out market-based transaction mechanisms.

6.2. Robustness Tests

To systematically evaluate the reliability of the benchmark findings, this study employs multiple robustness approaches. First, we introduce a one-period lagged DFI index to address potential reverse causality between the dependent and explanatory variables. Second, to mitigate bias attributable to their distinct governance structures, we exclude observations from the four centrally administered municipalities—Beijing, Tianjin, Shanghai, and Chongqing. As shown in

Table 3. Robustness checks using lagged explanatory variable and excluding municipalities.

Land_Transfer	Lag_LPM	Lag_Logit	Exclude_LPM	Exclude_Logit
lg_Index_lag	0.076 *	1.520 **
	(0.044)	(0.628)
lg_Index			0.138 ***	2.159 **
			(0.043)	(0.906)
Control_variables	Yes	Yes	Yes	Yes
Province Effects	Yes	Yes	Yes	Yes
Household Effects	Yes	Yes	Yes	Yes
Year Effects	Yes	Yes	Yes	Yes
Observations	33,223	33,223	38,921	38,921

Notes: *** p < 0.01, ** p < 0.05, * p < 0.10. Standard errors clustered at province level in parentheses.

, while controlling for province fixed effects, year fixed effects, household-level characteristics, and provincial macroeconomic factors, both LPM and Logit model estimates demonstrate high robustness. Specifically, the lag specification test reveals that a one-unit increase in the lagged index corresponds to a statistically significant 7.6 percentage point rise in rural land transfer probability based on LPM estimates. The restricted sample test excluding the four centrally administered municipalities shows markedly enhanced statistical significance for the core explanatory variable, with its coefficient increasing to 0.138. These results confirm consistent directionality and stable significance levels across tests, establishing a robust statistical foundation for DFI’s positive impact on rural land transfers.

The third robustness check uses a Probit model to reestimate the core relationship with the original coefficient estimates, without converting the marginal effects. As

Table 4. Probit models.

Land_Transfer	Probit_None	Probit_Family	Probit_All
lg_Index	0.930 **	1.100 ***	1.213 ***
	(0.376)	(0.393)	(0.423)
Control_variables	Yes	Yes	Yes
Province Effects	Yes	Yes	Yes
Household Effects	Yes	Yes	Yes
Year Effects	Yes	Yes	Yes
Observations	41,755	40,339	40,216

Notes: *** p < 0.01, ** p < 0.05. Standard errors clustered at household level in parentheses.

demonstrates, the original coefficients for the DFI index are significantly positive in the baseline specification. Incorporating both household-level and provincial-level controls substantially strengthens these effects, exhibiting identical positive significance and a consistently strengthening pattern mirroring the benchmark regression results in Table 2. Furthermore, Table 4 indicates that key control variables within the Probit model also corroborate the benchmark findings. This consistency shows that the core variable maintains a uniform direction across different estimation models. The observed trends in significance changes and the patterns of control variable influence consistently align, reinforcing the high robustness of DFI’s positive effect on rural land transfers.

This study further implements a placebo test to eliminate the effects of unobserved confounding factors and strengthen confidence in the core causal inference. The test randomly generates simulated values for the DFI index while preserving all control variables and the original model specification, performing 500 repeated sampling regressions. Figure 5 illustrates that the distribution of coefficients from these simulations approximates a normal distribution centered on zero, closely matching the reference normal curve. Importantly, the actual coefficient estimate from the benchmark regression decisively breaches the boundaries of this simulated distribution, lying far beyond its mean. This outcome indicates that none of the 500 randomly generated coefficients approached the magnitude of the true effect. These results confirm that DFI’s positive impact on rural land transfers reflects a statistically significant economic relationship rather than random chance, thus affirming the robustness of the benchmark findings.

6.3. Endogeneity Addressing

Theoretically, DFI functions as a macro-level financial environment variable shaped predominantly by regional policy orientations and technological diffusion mechanisms. This inherent characteristic precludes significant reverse causality with micro-level farmer land transfer decisions, thereby mitigating reverse causality bias at its source. Due to data availability constraints, this study employs the Peking University DFI Index, recognized as China’s most comprehensive and widely adopted measure, to address measurement error. While inherent measurement error remains unavoidable, the statistically significant benchmark regression and robustness test results collectively indicate it does not constitute a primary source of bias affecting the core findings.

To rigorously address endogeneity stemming from omitted variables, the 2010 internet penetration rate serves as an exogenous instrumental variable. This selection satisfies the three essential econometric criteria: relevance, exogeneity, and the exclusion restriction. Specifically, internet infrastructure is a prerequisite for the development of digital finance, as evidenced by the first-stage regression results, which confirm its significant positive correlation with the DFI Index. Furthermore, the historical timing of the 2010 data ensures temporal precedence, effectively insulating the instrument from influence by sample-period farmer behavior. Critically, the variable plausibly affects rural land transfers exclusively through digital finance channels, satisfying the exclusion restriction. The Kleibergen–Paap test rigorously confirms instrument validity, rejecting concerns regarding instrument weakness.

This analysis uses a two-stage least squares approach within the Linear Probability Model (LPM) framework. The results presented in

Table 5. Instrumental variable regression.

	Lg_Index	Land_Transfer
internet_2010	0.007 ***
	(0.001)
lg_Index		0.116 ***
		(0.019)
Control_variables	Yes	Yes
Household Effects	Yes	Yes
Province Effects	Yes	Yes
Year Effects	Yes	Yes
Kleibergen–Paap rk LM statistic	6.286
Kleibergen–Paap rk Wald F statistic	32.414
Observations	41,623	40,216

Notes: *** p < 0.01. Standard errors clustered at household level in parentheses.

demonstrate a strong first-stage relationship: the 2010 internet penetration rate exhibits a significantly positive impact on DFI. This confirms that provinces with higher historical internet penetration achieved greater subsequent digital finance development. The second-stage results establish a statistically significant positive coefficient for DFI at the 1 percent level, confirming its causal role in promoting rural land transfers. Notably, this causal effect estimate of 0.128 aligns closely with the benchmark LPM result, substantially reinforcing confidence in the core conclusions.

7. Mechanism Analysis

The regression results in the third column of Table 2 confirm a significantly positive coefficient for DFI in Equation (1). Building on this, we estimate Equations (2) and (3). The significance of both coefficient β₁ in Equation (2) and coefficient λ₂ in Equation (3) demonstrates that DFI affects rural land transfer through the mediating variables. A significantly positive coefficient λ₁ alongside this indicates partial mediation, while its insignificance would indicate full mediation.

Table 6. Mechanism analysis results.

	Lg_Commercial_ins (Step1)	Land_Transfer (Step2)	Nonagricultural_Entrepreneur (Step1)	Land_Transfer (Step2)
lg_Index	1.263 ***	0.205 ***	0.084 **	0.195 ***
	(0.428)	(0.060)	(0.038)	(0.060)
lg_commercial_ins		0.004 ***
		(0.001)
nonagricultural_entrepreneur				0.084 ***
				(0.008)
Control_variables	Yes	Yes	Yes	Yes
Household Effects	Yes	Yes	Yes	Yes
Province Effects	Yes	Yes	Yes	Yes
Year Effects	Yes	Yes	Yes	Yes
Observations	39,878	39,878	40,216	40,216

Notes: *** p < 0.01, ** p < 0.05. Standard errors clustered at household level in parentheses.

shows the results of the mediation mechanism test for the impact of DFI on rural land transfer. First-stage regressions demonstrate that an increase of one unit in the DFI index significantly raises farmers’ participation in commercial insurance by 1.263 units and increases the probability of non-agricultural entrepreneurship by 8.40 percentage points. In second-stage regressions controlling for both core explanatory and mediating variables, the direct effect of DFI on rural land transfer remains significantly positive. Coefficients for both commercial insurance participation and non-agricultural entrepreneurship achieve statistical significance. These results confirm that both variables partially mediate the pathway from DFI to rural land transfer, aligning with theoretical expectations.

The results of the integrated mechanism test indicate that DFI promotes rural land transfer through non-agricultural entrepreneurship and commercial insurance participation. First, DFI empowers small business development and creates non-agricultural jobs by providing farmers with digital credit support and payment infrastructure for non-agricultural entrepreneurship. This stable non-agricultural income reduces farmers’ dependence on land cultivation, thereby increasing their likelihood of transferring land out. These findings support Research Hypothesis Two. Second, DFI significantly improves farmers’ access to commercial insurance and enhances their risk perception and management capabilities by fostering internet-based insurance. As commercial insurance supplants land’s traditional role as a safety net, farmers’ willingness to transfer land rises markedly, validating Research Hypothesis Three.

8. Heterogeneity Analysis

Having established the robust promotional effect of DFI on rural land transfer and its underlying mechanisms, and considering the well-documented regional disparities in China’s development, this study further investigates the spatial heterogeneity and economic gradient characteristics of this effect. The heterogeneity analysis presented in

Table 7. Regional heterogeneity.

Land_Transfer	Southeast	Northwest	High_GDP	Low_GDP
lg_Index	0.108 **	0.021	0.209 ***	−0.137
	(0.047)	(2.311)	(0.067)	(0.087)
Control_variables	Yes	Yes	Yes	Yes
Household FE	Yes	Yes	Yes	Yes
Province FE	Yes	Yes	Yes	Yes
Year FE	Yes	Yes	Yes	Yes
Observations	35,509	6246	20,463	21,160

Notes: *** p < 0.01, ** p < 0.05. Standard errors clustered at province level in parentheses.

reveals significant regional divergence in the impact of DFI on rural land transfer. Grouped regressions based on the Hu Line show that DFI development significantly and positively affects rural land transfer in provinces southeast of the line. In contrast, the coefficient for provinces northwest of the line is 0.021 and is not statistically significant. This pattern primarily stems from differences in regional foundational conditions. Provinces in the southeast benefit from dense digital infrastructure and mature rural land transfer markets, which systematically support the risk management functions and resource allocation efficiency of digital financial tools. Conversely, northwestern provinces are subject to dual constraints: a digital access gap and underdeveloped land transaction systems. These constraints offset the benefits of financial technology due to high transaction costs. Furthermore, the disparity in human capital endowment exacerbates this differentiation. Farmers in the southeast have a higher average level of education than their counterparts in the northwest, enabling them to leverage financial services more effectively to develop non-agricultural entrepreneurial capabilities. This enhanced capability, in turn, facilitates the reallocation of land resources.

Along the economic development dimension, DFI exhibits a strong positive impact on rural land transfer in high-GDP regions, whereas a statistically insignificant negative trend is observed in low-GDP regions. This divergence arises because developed regions possess robust industrial ecosystems and well-established property rights protection. These factors allow DFI to more easily increase land liquidity by empowering entrepreneurs and through the insurance substitution channel. In contrast, underdeveloped regions suffer from a scarcity of non-agricultural employment opportunities and inadequate rural land transfer mechanisms. In these contexts, the adoption of financial technology may inadvertently reinforce farmers’ reliance on land for its traditional security function.

This study further investigates the heterogeneous impact along the household income gradient. The results reveal that the promotional effect of DFI on rural land transfers exhibits a significant inverted U-shaped distribution pattern. As presented in

Table 8. Household heterogeneity.

Land_Transfer	Low_Income	Middle_Income	High_Income
lg_Index	0.156 **	0.213 ***	0.196 *
	(0.064)	(0.076)	(0.104)
Control_variables	Yes	Yes	Yes
Household FE	Yes	Yes	Yes
Province FE	Yes	Yes	Yes
Year FE	Yes	Yes	Yes
Observations	13,433	13,420	13,363

Notes: *** p < 0.01, ** p < 0.05, * p < 0.10. Standard errors clustered at province level in parentheses.

, the regression coefficient for the low-income group is 0.156 and statistically significant at the 10% level. The coefficient for the middle-income group climbs to 0.213 and achieves significance at the 5% level. In contrast, the coefficient for the high-income group retreats to 0.196, remaining significant at the 10% level. This divergent pattern originates from variations in structural constraint across income strata. Middle-income households, benefiting from moderate capital accumulation capacity and risk tolerance, demonstrate the highest effectiveness in transforming digital financial services into capital for non-agricultural entrepreneurship. This efficient conversion significantly facilitates the release of land resources. Low-income households, despite possessing strong entrepreneurial aspirations, confront dual barriers: deficient digital skills and inadequate social safety nets. These constraints prevent financial resources from surpassing the threshold for productive investment. Meanwhile, high-income households exhibit diminishing marginal sensitivity to inclusive finance. This reduced responsiveness stems from land assets constituting a smaller share of their overall portfolio and their established channels for non-agricultural income.

9. Further Analysis: Spatial Spillover Effects

The spatial spillover effects of digital financial inclusion on rural land transfer fundamentally stem from the spatial externality of financial technologies and cross-regional mobility of production factors. The network effects of digital financial infrastructure transcend administrative boundaries, enabling the digital financial capabilities of neighboring provinces to directly influence local markets through technological diffusion [52]. The cross-regional compatibility of technologies such as mobile payments and blockchain-based notarization reduces connectivity costs for interprovincial land information platforms, fostering synergistic technological benefits. Simultaneously, digital credit platforms restructure capital flow pathways, attracting investments from adjacent provinces into local land transfer markets and triggering cross-jurisdictional reallocation of production factors. Crucially, provincial policies exhibit spatial interdependence, as regional digital finance initiatives generate multiplicative effects through institutional emulation. Consequently, digital financial inclusion development not only impacts land transfer behavior within a province but also radiates to neighboring regions, accelerating market integration. But traditional models neglecting this spatial linkage systematically underestimate digital finance’s policy efficacy. Research on this phenomenon reveals new mechanisms through which digital finance influences land resource allocation and provides critical policy insights for dismantling administrative barriers and establishing a unified national land market.

This study constructs the provincial land transfer rate (Land_T) based on six waves of panel data (2012, 2014, 2016, 2018, 2020, 2022) covering 41,755 rural households across China’s 31 provinces. The computation involves aggregating micro-level land transfer decisions at the province-year level, specifically calculating the percentage of households engaged in land transfer relative to each province’s total surveyed households annually. This procedure ultimately yielded a balanced provincial panel dataset comprising 140 valid observations. This study adheres to the Queen contiguity criterion to construct the spatial weights matrix, enabling the derivation of the core explanatory variable—the spatially lagged digital inclusive finance index (W_lg_Index). The empirical framework employs the SDM that incorporates year fixed effects to control for macroeconomic cyclical shocks while introducing provincial control variables including per capita GDP and fiscal expenditure to effectively mitigate omitted variable bias.

Regression results from the SDM in

Table 9. Regression results: spatial spillover effects.

	Land_T	Land_T
W_lg_Index	0.414 *	0.670 ***
	(0.238)	(0.251)
Control Variables	No	Yes
Spatial Fixed Effects	Yes	Yes
Year Fixed Effects	Yes	Yes
R-squared	0.175	0.237
Observations	140	140

Notes: *** p < 0.01, * p < 0.10. Standard errors clustered at province level in parentheses.

demonstrate robust spatial spillover effects. After controlling for key covariates, the core explanatory variable W_lg_Index shows a statistically significant coefficient of 0.670 at the 1 percent level. The regression results from the Spatial Durbin Model (SDM) provide compelling evidence of significant spatial spillover effects in the relationship between digital financial inclusion and rural land transfer. Therefore, the development of digital financial services in a province not only influences its own land transfer behavior but also exerts a notable impact on neighboring provinces. This spatial dimension must be integrated into policy frameworks to harness the full potential of financial technologies in promoting rural land market development.

10. Conclusions and Implications

10.1. Conclusions

This study examines the impact of DFI on rural land transfers using theoretical and empirical approaches. Building on a review of the relevant literature and the construction of a theoretical framework, we utilize pooled panel data formed by matching six waves (2012, 2014, 2016, 2018, 2020, 2022) of the China Family Panel Studies (CFPS) with corresponding provincial DFI indices. Using Linear Probability Models (LPMs), Logit models, and mediation effect models, we conduct empirical tests to thoroughly analyze the influence of DFI on farmers’ decisions to transfer out land and its underlying mechanisms. Additionally, this study examines the heterogeneous nature of this impact across regions, GDP levels, and household income tiers. The principal findings are as follows:

First, DFI significantly promotes rural land transfers. Benchmark regression results show that an increase of one unit in the DFI index raises the probability of rural land transfer by an average of 12.8 percentage points. This effect is statistically significant at the 1% level. Consistency across robustness checks and instrumental variable estimations confirms the reliability of this finding.

Second, non-agricultural entrepreneurship and commercial insurance participation act as significant mediating channels through which DFI influences rural land transfers. The development of DFI enhances farmers’ propensity for non-agricultural entrepreneurship and commercial insurance uptake. This effectively substitutes for the traditional role of land in providing economic security, employment, and old-age support, significantly incentivizing farmers to transfer out land.

Third, the impact of DFI on rural land transfers varies significantly across regional, economic development, and household income dimensions. Geographically, a positive effect is observed in areas southeast of the Hu Line, while the impact proves statistically insignificant in regions northwest of the line. Economically, DFI strongly promotes rural land transfers in regions with high GDP, whereas in regions with low GDP, its effect manifests as a non-significant negative trend. Regarding household income, although DFI significantly encourages rural land transfers across all income groups, the magnitude of this promotional effect varies considerably among low-income, middle-income, and high-income households. Middle-income households, characterized by balanced capital accumulation capacity and moderate risk tolerance, exhibit the greatest efficacy in converting digital financial services into capital for non-agricultural enterprises, thereby substantially accelerating land resource liberation. Low-income households, despite strong entrepreneurial motivation, face dual barriers of inadequate digital literacy and insufficient social protection, which impede financial resources from crossing the productive investment threshold. Meanwhile, high-income households demonstrate diminishing marginal responsiveness to inclusive finance, as land assets constitute a smaller proportion of their wealth portfolio and they possess established non-agricultural income channels.

Finally, digital inclusive finance demonstrates a significant positive spatial spillover effect on rural land transfer. After controlling for key covariates, a one-unit increase in the digital financial development level of neighboring provinces leads to a 0.67 percentage point rise in the local province’s land transfer rate. This suggests that traditional models underestimate policy effectiveness by overlooking inter-provincial linkages, underscoring the necessity of employing a spatial econometric framework.

10.2. Implications

Based on the research findings, this study makes the following recommendations:

First, institutional coordination between the DFI and rural land system reforms should be deepened to enhance the efficiency of land resource allocation through diversified financial innovations. Local governments can foster tripartite collaboration among government bodies, financial institutions, and farmers. This collaboration should encourage financial institutions to develop tailored credit and insurance products addressing farmers’ diverse needs, particularly addressing startup capital constraints and social security gaps following rural land transfers. Exploring pledge financing models using anticipated rural land transfer income rights could enable farmers to leverage digital land contracting certificates for entrepreneurial funding. Concurrently, developing protective insurance products covering multiple risks during the transition to non-agricultural employment is essential. At the national level, establishing a dedicated risk compensation mechanism to support county-level inclusive agricultural lending is recommended. This mechanism should incentivize financial institutions to integrate rural land transfer platform data into risk models to optimize credit strategies dynamically, ultimately creating a virtuous cycle where financial resources activate land assets and industrial upgrading fuels rural development.

Second, strengthen the targeted support capacity of DFI for non-agricultural employment and entrepreneurship to systematically supplant land’s traditional security functions. Provincial governments should integrate employment services with financial resources to establish integrated smart service platforms. These platforms should offer farmers transferring land comprehensive support, including skills training, job matching, and entrepreneurial financing. It is crucial to implement a tiered credit policy for entrepreneurs that offers preferential loan terms to farmers who complete rural land transfers and pass project evaluations. Developing mobile applications for migrant workers to facilitate precise job matching and streamlined access to microcredit is also necessary. Simultaneously, promoting insurance innovation for land income conversion pension products allows farmers to convert a portion of transfer income into long-term security benefits. Providing differentiated government subsidies for low-income participants is vital to fundamentally eroding farmers’ dependence on land-based security.

Third, implement tiered capacity-building initiatives and prioritize infrastructure development to overcome region-specific bottlenecks. To enhance human capital, local governments should partner with financial institutions to launch village-level financial literacy programs. Utilizing farmer-centric training systems and intelligent tutoring tools should prioritize financial risk education and digital skill development for less-educated groups. A complementary step is developing user-friendly inclusive finance terminals embedded with risk control modules. In terms of infrastructure, the deployment of next-generation communication networks in underdeveloped regions is a top priority. Promoting mobile service terminals addresses financial access issues in areas lacking reliable power and communication coverage. It is imperative to reduce digital access barriers through terminal subsidies and preferential tariffs, as well as to establish emergency communication safeguards for challenging terrains. Concurrently, advancing integrated land consolidation and linked development of specialized industrial parks in lagging counties promotes high-quality local employment for the transitioning workforce.