Digital Inclusive Finance, Rural Loan Availability, and Urban–Rural Income Gap: Evidence from China

Abstract

Reducing the urban–rural income gap stands as a pivotal factor in attaining sustainable economic development. Policymakers and researchers have focused on whether digital inclusive finance can narrow the urban–rural income gap. Utilizing provincial-level panel data from 31 regions in China, this paper empirically tests the impact of digital inclusive finance on the urban–rural income gap across different areas of China, specifically analyzing the mediation effect of rural loan availability. The findings indicate that (1) the expansion of digital inclusive finance can enhance the inclusiveness of financial services in rural areas and narrow the urban–rural income gap; (2) the effect of digital inclusive finance on the urban–rural income gap exhibits regional heterogeneity; (3) the rural loan availability has a mediation effect on the urban–rural income gap, but its growth will weaken the narrowing effect of digital inclusive finance on urban–rural income gap; and (4) the reduction in the urban–rural income gap attributable to digital inclusive finance exhibits a nonlinear relationship with the level of urbanization. This paper recommends guiding the digital transformation of rural financial institutions and enhancing farmers’ financial literacy to increase the availability of loans in rural areas. Furthermore, implementing region-specific policies and regulations could effectively narrow the urban–rural income gap.

1. Introduction

The 2030 Agenda for Sustainable Development proposed by the United Nations sets out the goal of ending poverty in all its forms everywhere. The Sustainable Development Goals (SDGs) are firmly dedicated to fostering sustainable, inclusive, and resilient economic growth, as well as promoting shared global prosperity. However, the SDG Progress Report (2024) shows less optimistic figures due to the pandemic and other factors. The report indicates that, in 2023, an alarming 241 million workers worldwide continued to endure extreme poverty [1]. The eradication of poverty hinges on boosting the income levels of residents residing in deprived areas. In 2020, China announced the accomplishment of its objective in the battle against poverty, asserting the complete eradication of absolute poverty within its borders. Nevertheless, the urban–rural income gap persists as a crucial concern, stemming from variations in regional economic structures.

The progression of the urban–rural income gap in China can be traced through several stages. From the founding of New China until the reform and opening-up period, the nation adhered to a planned economic system. This approach, combined with the centralized control of agricultural product distribution through unified purchasing and marketing policies, facilitated rapid industrialization by leveraging agricultural surpluses [2]. During this period, as China also retained control over income distribution, there was no significant widening of the urban–rural income gap. Following the initiation of economic reforms and the opening-up policy in 1978, which brought about shifts in developmental goals and the adoption of a market-based economic system, the urban–rural income gap became increasingly volatile. The early reform period saw the introduction of the household responsibility system, which significantly enhanced agricultural productivity and efficiency [3], narrowing the urban–rural income gap from the early 1980s to the early 2000s. However, as urban development regained prominence, the household registration system restricted the flow of rural labor into urban centers, contributing to a gradual widening of the income gap. The financial crisis of 2007–2008 hit the then export-orientated China hard, and the government launched a massive RMB 4 trillion stimulus package in November of the same year [4]. After 2008, China’s urban–rural income gap began to show a trend of gradual narrowing. Since the beginning of the century, the income gap has consistently shown a persistent upward trend in percentage growth. China has implemented many policies. Examples include the Urban–Rural Integrated (URI) development strategy and the Targeted Poverty Alleviation (TPA) program. Although the policies have been effective, the urban–rural income gap remains an obstacle to China’s sustainable economic development [5,6].

In recent years, the global disparity between the wealthy and the impoverished has markedly widened, impacting social progress and intensifying disparities in income distribution. McKinnon and Shaw pioneered the integration of financial factors into discussions on the income gap, thereby establishing the basis for a follow-up study. Numerous studies indicate that financial development can alleviate poverty in a country or region by stimulating household final consumption expenditure [7], facilitating financial sector reform, and cultivating economic growth [8,9]. In contrast, some researchers hypothesize that financial development can be detrimental to poverty reduction. Arestis et al. argue that, given the nature of the market, financial liberalization disproportionately favors groups that already possess economic resources and occupy advantageous positions [10], prompting the question of whether such development can genuinely alleviate poverty among the poor. Furthermore, financial deepening and structural inequality pose significant challenges in narrowing the income gap. Beyond these opposing viewpoints, certain scholars have posited a non-linear correlation between financial development and poverty rates. An early study found that the income distribution is independent [11]. However, a dynamic model incorporating a fixed fee (representing a wealth threshold level) for financial market access demonstrated the existence of Kuznets’ “inverted U-shaped curve”. This means the income gap initially expands but then contracts as economic development progresses [12]. A study of Chinese counties revealed a notable expansion of the urban–rural income gap in areas where the financial depth quartile falls below 20%. In contrast, the correlation becomes insignificant in areas with a financial depth quartile between 20% and 70%, and the gap narrows significantly in areas where it exceeds 70%. These findings further demonstrate the existence of the above non-linear relationship [13].

Inclusive finance centers on delivering affordable and effective financial services to all demographics and groups requiring such services, according to principles of equal opportunity and commercial sustainability. Omar et al., in a global study including 116 developing countries, determined that inclusive finance correlates with poverty and income gap reduction. Their research suggests that it facilitates the use of formal financial services among groups situated in marginalized areas, a trend that strengthens social well-being and lessens the income gap [14]. Governments globally are increasingly recognizing inclusive finance as a cornerstone of economic and social progress, integrating it as a core element of national poverty alleviation policies and programs. In 2013, China formally introduced “inclusive finance development” as a national strategy, significantly contributing to the country’s progress over the subsequent decade. Domestic researchers have observed that inclusive finance is associated with improvements in farmers’ education levels, generating an effect on agricultural productivity. Rural economies are moving towards a more sustainable path [15].

As economic conditions continue to improve, digital technology and inclusive finance are becoming increasingly integrated. Digital inclusive finance plays a crucial role in promoting employment, alleviating poverty, and addressing other socioeconomic challenges. Since China implemented the “National Big Data Strategy” in 2015, technology has been instrumental in driving digital transformation. Through mobile payments and other digital technologies, digital finance enables the expansion of traditional financial services into remote and rural regions. The development and widespread adoption of digital payment systems effectively narrow financial transaction costs [16] while simultaneously increasing the availability of borrowing and financing options for households [17,18]. In this process, farmers’ innovation and entrepreneurship are stimulated, and the innovation capacity of rural areas is enhanced [19]. Moreover, the application of communication technology and digital currency has positively influenced how households manage and utilize their financial assets [20,21]. Li et al. conducted a study using online shopping, digital payments, online credit, online financing product purchases, and business insurance as mediating variables. The results indicate that digital inclusive finance can effectively promote an improvement in the household consumption structure [22].

The development of digital inclusive finance has strengthened the breadth of coverage of financial services in rural areas, especially the rural loan availability. In a United Nations report, Demirguc-kunt et al. hypothesized that inclusive finance could empower impoverished populations to access loans or borrow funds, enabling them to accumulate assets, cultivate personal credit, and, finally, create pathways toward more secure futures [23]. Jeanneney et al. analyzed panel data from developing countries over 30 years and concluded that financial development benefits low-income people [24]. This positive effect is attributed to improvements in the ease of conducting transactions, savings, and securing loans. Xu et al. argue that digital inclusive finance alleviates the cost challenges faced by farmers in the traditional financial system, allows loans to flow more freely into the countryside, and provides capital support for activities such as rural land transfers [25]. However, Seven et al. argued that people living in poverty have not experienced the expected benefits due to inadequate financial services [26].

While the existing research has made significant progress in understanding digital inclusive finance and the urban–rural income gap, there is still a scarcity of research on the indirect role of rural loan availability. In addition, there is insufficient empirical evidence on specific measures to reduce income inequality in developing countries. Understanding the specific transmission paths is important for policymakers. To fill the gap in the field, our study evaluates four key issues concerning digital inclusive finance and the urban–rural income gap. First, does digital inclusive finance directly narrow the urban–rural income gap? Second, is there heterogeneity in the above effects? Third, does rural loan availability mediate between digital inclusive finance and the urban–rural income gap? Fourth, is the relationship between digital inclusive finance and the urban–rural income gap linearly correlated with urbanization? We empirically test these relationships to provide micro-level data to support theoretical studies.

Our study has several major contributions and innovations. First, this study helps to better clarify the relationship between digital inclusive finance and the urban–rural income gap in China. Second, the discovery of the indicator of rural loan availability is innovative. Specifically, while rural loan availability increases significantly with the growth of digital inclusive finance, it simultaneously reduces the effectiveness of narrowing the urban–rural income gap. While it is partly an intermediary in the transmission mechanism, policymakers should consider the resulting use of rural loan funds. Furthermore, our findings enrich the literature on the rural–urban income gap in China. They provide empirical references for other developing countries to better achieve poverty reduction and the goal of sustainable economic development.

The remainder of this paper is structured as follows. Section 2 elaborates on the theoretical background and the hypotheses based on a review of the academic literature. Subsequently, Section 3 presents the data, models, and methods used in this paper. Section 4 presents and discusses the empirical results. Finally, Section 5 concludes the implications of our findings for policymakers and academics. The research framework is represented in Figure 1.

2. Literature Review and Hypotheses

2.1. Theoretical Review: Financial Exclusion and Inclusive Finance

Leyshon et al. pioneered the definition of financial exclusion: the exclusion of poor and vulnerable social groups from the financial system [27]. Financial exclusion widens regional differences in economic development and inequality. Kempson and Whyley have since expanded it to include six aspects of financial exclusion: geographic exclusion, appraisal exclusion, conditionality exclusion, price exclusion, marketing exclusion, and self-exclusion.

Financial exclusion is relevant to the development of inclusive finance. Research on financial exclusion initially focused on financial services in developed countries such as the United Kingdom and the United States. However, over time, the theory of financial exclusion has gradually been extended to developing countries. Poor financial infrastructure and high financing costs in developing countries may exacerbate financial exclusion. Koku P S. argues that financial deregulation allows low-income people to access credit and financial services at a lower cost [28]. As a result of the long history of financial inhibition and exclusion in rural China, rural households cannot enjoy the financial services to which they are entitled. The development of inclusive finance can effectively resolve financial exclusion. In recent years, inclusive finance has begun to achieve digital transformation. Will digital technology lower the costs of access to financial services for low-income people? Or is the exclusion phenomenon more serious? In this regard, different scholars have given different insights. Fernández-Olit et al. are neutral about the impact of digital advances [29]. Adeoye et al. mention in past studies that AI and technological innovations can provide an alternative path for developing economies in mitigating financial exclusion and advancing inclusive finance [30]. In the following section, we will examine this issue specifically in China.

2.2. Empirical Review

2.2.1. Digital Inclusive Finance and the Urban–Rural Income Gap

Advancements in digital technology have effectively bridged the physical divide between urban and rural areas while enhancing service efficiency. Directly, the robust expansion of digital inclusive finance has broadened the reach and availability of financial services. This availability plays a crucial role in preventing vulnerable low-income people from falling into poverty and contributing to overall poverty reduction [31]. Song pioneered the research of digital inclusive finance and its impact on the urban–rural income gap in China. She advocated for a restructured developmental approach in China that cultivates a collaborative digital inclusive financial ecosystem capable of effectively narrowing the urban–rural income gap by facilitating inclusive financial services and access to them [32].

Indirectly, digital inclusive finance also increases farmers’ incomes by mitigating financial exclusion [33]. First, employment opportunities have increased, and farmers’ creative and entrepreneurial vigor has been stimulated. Hao et al. analyzed the study from two perspectives: wage income and property income inequality. They argued that digital technology reduces income inequality by promoting household entrepreneurship at a higher rate and frequency [34]. In their research, Zou and Zhang et al. posited that digital inclusive finance holds the potential to foster shared prosperity by narrowing the digital divide and empowering individuals to engage in innovative and entrepreneurial activities [35,36]. Second, companies are undergoing digital transformation, leading to lower financing costs. Technological innovation plays a crucial role in this process [37,38]. Third, rapid infrastructure development has increased access to markets and basic services. Suhrab et al. conducted a study using the BRICS countries as an example. Their research indicates that digital inclusive finance can reduce income inequality in the BRICS countries by facilitating more cost-effective transactions [39]. In addition, other scholars studying developing economies besides China have come to similar conclusions. For example, the development of digital inclusive finance in Cannes has been accompanied by financial stability and improved financial literacy among the population, thus contributing to increased local incomes and sustainable economic development [40].

Drawing on previous research, Mo et al. further analyzed this relationship. Through their intermediary effect mechanism test, they hypothesized that digital inclusive finance stimulates industrial expansion and generates more farmer employment prospects, thereby boosting their income and shrinking the urban–rural income gap [41]. Zhao et al. approached the issue from the primary distribution and redistribution perspectives. They found that primary distribution mediated the narrowing of the urban–rural income gap, while redistribution did not [42]. Other studies have highlighted the spatial impact of digital inclusive finance on income disparity through factors such as industrial upgrading and educational advancement [43]. Wan et al., using a panel data sample that included 31 provinces, concluded that digital inclusive finance could shrink the urban–rural income gap nationally. Their research also indicated “low–low” as well as “high–high” spatial clustering characteristics [44].

Simultaneously, some researchers argue that the development of digital inclusive finance may widen the gap in people’s access to financial services. The development of digital technology relies on Internet technology, it may leave underdeveloped and poor areas at a disadvantage. This is because the residents of these areas believe that the transaction costs of digital finance are too expensive for them. Consequently, these low-income people are likely unable to enjoy its benefits and instead rely on traditional financial transaction methods [45,46].

The existing literature suggests that the impact of digital inclusive finance on the urban–rural income gap is complex. This study is not only of theoretical significance but also has implications for other developing economies. Drawing from the literature review conducted above, we formulate the subsequent hypotheses:

Hypothesis 1.

Digital inclusive finance effectively narrows the urban–rural income gap.

Hypothesis 2.

The impact of digital inclusive finance on the urban–rural income gap exhibits regional heterogeneity.

2.2.2. Digital Inclusive Finance and Rural Loan Availability

Digital inclusive finance not only narrows information asymmetry but also expands the reach of financial services. This increased accessibility enhances the availability of financial services and funding opportunities for rural residents, a phenomenon closely tied to the rapid progress of internet technology. However, China currently has few formal financial institutions offering rural loans. More than half of the farmers who received loans indicated that the loan amount was not sufficient to meet their financial needs [47]. Many rural areas in China suffer from challenges such as low agricultural incomes, limited financing options, a sparse network of financial institutions, and geographical remoteness. Even studies claim that the banking sector in rural areas of China faces institutional financial exclusion. Although geographically accessible, financial services are inaccessible [48]. These factors have significantly narrowed the reach of financial services and resulted in reduced access to credit for rural residents.

However, the rapid development of digital inclusive finance in China has effectively bridged the gap between rural communities and financial institutions. Many scholars presented their views on whether this shift benefits rural loans in China. From a supply perspective, Sheng believed that fintech effectively facilitated the supply of credit to SMEs in the banking sector [49]. Xie et al., in their empirical study utilizing data from rural bank loans across 30 Chinese provincial administrative regions, discovered that digital finance encourages banks to expand loan provisions in rural areas. This is achieved by reducing operational costs and potential risks, finally increasing relative profitability. Their findings indicate that digital inclusive finance can increase rural loan availability [50]. Sun et al., employing China’s rural microeconomic data and a county-level digital inclusive finance index, conducted a Probit model analysis on farmers’ access to loans. Their study indicates that digital inclusive finance enhances farmers’ access to formal credit, effectively addressing the issue of financial exclusion in rural areas [51].

Bollaert et al. argued that the rapid development of digital technology has fueled the growth of fintech lending. Businesses and investors have thus improved their access to financial loans [52]. Phan et al. examined the performance of loan availability in underdeveloped areas using rural areas of Vietnam and Thailand as examples. The results of the study, however, led to opposite conclusions. The advancement of Internet technology has exacerbated information asymmetry in rural regions, consequently impeding farmers’ access to loan financing [53]. The authoritative international literature examining digital inclusive finance and rural loan availability is scarce and primarily focuses on developed country perspectives. China’s rural economy has a special structure, making our study necessary. Drawing from the literature review conducted above, we formulate the subsequent hypothesis:

Hypothesis 3.

Digital inclusive finance promotes rural loan availability.

2.2.3. Rural Loan Availability and the Urban–Rural Income Gap

Few existing studies have specifically examined the relationship between rural loan availability and the urban–rural income gap. The revenue structure of financial institutions leads to a concentration of their branches in more economically developed regions. Therefore, most rural areas suffer from challenges from geographic isolation and underdeveloped production practices. Compared to their urban counterparts, most rural residents experience significant difficulties securing financing, with limited access to the loans necessary for production and business operations.

The expansion of digital inclusive finance not only directly contributes to an increase in rural wage income but also indirectly enhances disposable income by improving rural loan availability. Some studies have argued that rural loan availability positively affects this process. Using the number of bank branches to measure the accessibility of financial services, Mookerjee et al. analyzed income inequality globally across countries with varying degrees of economic progress. The results suggest that increasing the number of bank branches and lowering barriers to bank access can mitigate income inequality [54]. Yu et al. discovered that augmenting rural credit allocations can mitigate financing limitations. This, in turn, leads to a return of labor and increased labor productivity [55]. Yang et al. conducted a similar study, and they found that capital can indirectly raise the income of rural residents by increasing non-farm employment [56].

However, does this necessarily mean that the urban–rural income gap will narrow? This conclusion has yet to be verified. The increase in the availability of financial services will bring about not only an increase in rural loan availability but also a shift in the consumption structure of rural residents [57]. This paper contends that rural financial availability serves as a vital indicator of the progress of digital inclusive finance in the rural financial sector. However, the growth in the volume of rural loans, while indicating improved financial availability, can also contribute to an increase in the non-property income of rural residents. Expenditures on non-productive consumer goods can lead to capital outflow, indirectly impacting both disposable income and the overall economic well-being of residents. Therefore, we argue that rural loan availability indirectly affects the urban–rural income gap and include it as a mediating variable in the empirical analysis. Based on the aforementioned theoretical research, this paper puts forward the following hypothesis:

Hypothesis 4.

The rural loan availability will weaken the narrowing effect of digital inclusive finance on the urban–rural income gap.

Through theoretical and empirical evidence, we analyzed the relationship between digital financial inclusion, rural loan availability, and the urban–rural income gap. Digital inclusive finance increases agricultural production and rural entrepreneurship by providing more accessible financial services. This increases farmers’ sources of income, thereby reducing poverty. In addition, digital inclusive finance and China’s fiscal policies to support agriculture complement each other, effectively narrowing the urban–rural income gap. Secondly, digital inclusive finance has dramatically increased rural loan availability by lowering transaction costs and service thresholds. Traditional financial institutions tend to allocate limited financial resources to urban areas, leading to persistent financial exclusion. In comparison, digital inclusive finance expands finance to rural areas through digital technology. This has increased farmers’ access to financial services and products, increasing rural loan availability. Finally, our research delved into the transmission mechanism linking digital inclusive finance with the urban–rural income gap, specifically through the lens of rural loan availability. The results show that rural loan availability has a partial mediation effect in the influence mechanism. Rural loan availability weakens the impact of digital inclusive finance on narrowing the rural–urban income gap. Although little literature demonstrates this phenomenon, this is not contrary to the current reality in China. Despite an increase in the number of loans, rural funds are not necessarily being utilized effectively. Rural areas suffer from poor infrastructure, farmers’ financial literacy requires enhancement, and rural households exhibit weak money management skills. These issues can result in the inefficient use of rural loan funds, which, instead of supporting local economic development, often flow back to the cities. Consequently, the potential of digital inclusive finance to narrow the urban–rural income gap is diminished.

3. Methodology

3.1. Data Sources

The data in this study are mainly from the China Rural Financial Services Report, the CSMAR Database, and the Digital Inclusive Finance Index, released by the Financial Research Center of Peking University. Rural loan availability before 2015 is based on shares published in the China Rural Financial Services Report; after 2015, odd-year data are based on year-on-year growth rates from the report published every two years. The Digital Inclusive Finance Index before 2010 was directly adopted from the data in the report; data after 2010 were obtained by emailing the original authors of the data compilation team. In addition, data for the remaining control variables were mainly obtained from the official website of the National Bureau of Statistics of China. Indicators such as urban and rural population counts, disposable income levels of urban and rural residents, gross domestic product, education levels, and innovation capacity were obtained from the National Bureau of Statistics of China and the websites of local statistical bureaus. Supplementary and missing data were mainly sourced from local statistical yearbooks and their accompanying bulletins.

3.2. Variable Measurement

The definition and calculation of the variables in this study are shown in

Table 1. Definition of variables and measurement methods.

	Variable	Symbol	Measurement Method
explained variable	Urban–Rural Income Gap	Theil	calculation of the Theil Index
explanatory variable	Digital Inclusive Finance Index	Duf	total Index/100
mediating variable	Rural Loan Availability	Al	year-on-year growth rate projection
control variables	Level of Economic Development	Lngdp	logarithmic GDP
	Foreign Trade Dependence	Foreign	total import/export/GDP
	Rationalization of Industrial Structure	Rat	rationalization of industrial structure inverse Tel Index
	Optimization of Industrial Structure	Insh	Feng Suling [58]—advanced industrial structure
	Urbanization Level	Urb	urban population/total population
	Education Level	Edu	number of college students per capita/total population
	Innovation Capacity	Innov	patents per capita/total population

:

3.2.1. Explained Variable: Urban–Rural Income Gap

The urban–rural income gap index, calculated as the ratio of urban residents’ disposable income to the per capita net income of rural residents, has often been measured utilizing the Gini coefficient in prior research. Given that the composition of income is influenced by various factors, including shifts in the proportion of urban and rural populations, this paper adopts the methods employed by Wang et al. [59]. This approach integrates demographic factors into the measurement of the urban–rural income gap by using the following formula:

$${Theil}_{i,t}={\displaystyle \sum _{j=1}^2(\frac{I_{ij,t}}{I_{i,t}})}\ln ({\displaystyle \frac{I_{ij,t}}{I_{i,t}}}/{\displaystyle \frac{P_{ij,t}}{P_{i,t}}})$$

(1)

where $I_{i,t}$ represents the income of residents in year t. In this paper, i = 1 denotes urban, and i = 2 expresses rural in data processing. It represents the sum of the income of urban and rural residents in year t, $P_{i,t}$ describes the number of urban or rural populations in year t, and P_t refers to the total number of urban and rural populations in year t. The larger the Theil is, the wider the urban–rural income gap is.

3.2.2. Explanatory Variable: Digital Inclusive Finance Index

Drawing on the methodology of Guo et al. [60], our study utilizes the “Peking University Digital Inclusive Finance Index” to measure the development of digital inclusive finance. Our study supplements the original calculation method to account for more recent data not included in the report. Given the strong correlation between the sub-indices with the overall digital inclusive finance index, which could potentially impact the reliability and accuracy of our analysis, this paper focuses solely on the overall index in the subsequent calculations.

3.2.3. Mediating Variable: Rural Loan Availability

Based on the literature review, we argue that digital inclusive financial development positively impacts rural loan availability. Therefore, we draw on Asad’s research methodology and select rural loan availability as the mediating variable for mediation effect analysis [61]. This variable is calculated by integrating statistical data from the CSMAR database with the calculation methodology and percentage ratios presented in the China Rural Financial Services Report. Data points from 2015 to 2022 are derived from the year-on-year growth rate provided in the China Rural Financial Services Report to calculate the preceding year’s value. Data points prior to 2015 are calculated based on the percentage ratios outlined in the report.

3.2.4. Control Variables

This paper identifies several key control variables: level of economic development, foreign trade dependence, rationalization of industrial structure, optimization of industrial structure, urbanization level, education level, and innovation capacity. The following variables are of particular interest:

Liang [62] suggests that foreign trade may contribute to urban–rural income inequality over the short term. Our study utilizes data obtained from the National Bureau of Statistics website, employing the ratio of total imports and exports to gross domestic product (GDP) as a measure of foreign trade dependence. To address potential unit inconsistencies with other variables, the original values are converted to RMB utilizing the annual average exchange rate, with the original values reported in billions of dollars.

This paper, referencing the work of Feng [58] and Gan [63], measures the effect of industrial structure through two indicators: industrial structure rationalization and industrial structure optimization. Industrial structure rationalization is defined as the process of aligning industrial structure with the prevailing level of economic development. This involves the allocation of production factors both between and in industries.

$${Rat}_{i,t}={\displaystyle \frac{1}{{Theil}_{i,t}}}={\displaystyle \frac{1}{{\displaystyle \sum _{j=1}^2(\frac{I_{ij,t}}{I_{i,t}})\ln (\frac{I_{ij,t}}{I_{i,t}}/\frac{P_{ij,t}}{P_{i,t}})}}}$$

(2)

The optimization of industrial structure centers on maximizing existing factors of production in the parameters of established industrial productivity to achieve superior economic efficiency. Drawing on the work of Feng [58] and expanding upon the traditional approach of utilizing only the output value ratio of secondary and tertiary industries as an indicator of an advanced industrial structure, this paper incorporates the output value of the primary industry. This results in the following formula, where i denotes the province, t signifies the year, and all other values remain consistent with previous definitions:

$${Insh}_{i,t}={\displaystyle \sum _{j=1}^3\frac{I_{ij,t}}{I_{i,t}}}*j$$

(3)

This paper carries out an empirical study of 31 Chinese provinces from 2011 to 2022, yielding 372 balanced panel observations. Raw data for our study were collected from sources including the National Bureau of Statistics website and the CSMAR database. These data were cleaned and processed to align with the specific methodological and statistical requirements of the research. As indicated through descriptive statistical analysis presented in

Table 2. Descriptive statistics analysis.

Variables	Mean	SD	Min	Max	Range
Theil	0.087	0.039	0.017	0.202	0.185
Duf	2.429	1.076	0.162	4.607	4.445
Al	8.712	1.069	3.404	11.09	7.687
Lngdp	9.776	1.009	6.416	11.77	5.353
Foreign	0.266	0.279	0.008	1.464	1.456
Rat	12.49	16.20	1.312	133.7	132.3
Insh	2.401	0.122	2.132	2.836	0.704
Urb	0.592	0.130	0.227	0.896	0.669
Edu	0.021	0.006	0.008	0.044	0.036
Innov	0.001	0.001	0.000	0.005	0.005

Data source: The National Bureau of Statistics of China, Peking University Digital Inclusive Finance Research Center and stata17 analysis results.

, the $Theil$ exhibits a mean value of 0.087. The maximum observed value for this gap is 0.202, while the minimum value stands at 0.017. These findings highlight the unbalanced nature of China’s urban and rural economic development, represented by a significant income gap between these areas. In addition, the $Duf$ demonstrates significant regional heterogeneity, with a difference of 4.445 between its maximum and minimum values. This index presents a mean value of 2.429 and a standard deviation of 1.076. Such difference highlights the uneven development of digital inclusive finance across different regions in China.

3.3. Baseline Model

To test the above research hypotheses, this paper first constructs the following two-way fixed effects model. By controlling for time and individual characteristics, our study analyses the impact of digital inclusive finance on the urban–rural income gap:

$$\begin{array}{l}{Theil}_{i,t}={\beta }_0+{\beta }_1{Duf}_{i,t}+{\beta }_{2}Lngd{p}_{i,t}+{\beta }_{3}Foreig{n}_{i,t}+{\beta }_{4}Ins{h}_{i,t}+\\ {\beta }_{5}Ra{t}_{i,t}+{\beta }_{6}Ur{b}_{i,t}+{\beta }_{7}Ed{u}_{i,t}+{\beta }_{8}Inno{v}_{i,t}+i.year+i.province+{\epsilon }_{i,t}\end{array}$$

(4)

Based on previous assumptions, it is reasonable to hypothesize that rural loan availability exerts a certain degree of influence on the shifts observed in the urban–rural income gap. This paper utilizes rural loan availability to test the mediation effect of digital inclusive finance on the urban–rural income gap. In this paper, the following models are constructed to carry out the next step of this study:

$$Thei{l}_{i,t}={\alpha }_0+{\alpha }_{1}Du{f}_{i,t}+{\alpha }_2{Z}_{i,t}+i.year+i.province+{\epsilon }_{i,t}$$

(5)

$$A{l}_{i,t}={\theta }_0+{\theta }_1{Duf}_{i,t}+{\theta }_2{Z}_{i,t}+i.year+i.province+{\epsilon }_{i,t}$$

(6)

$$Thei{l}_{i,t}={\gamma }_0+{\gamma }_{1}A{l}_{i,t}+{\gamma }_2{Duf}_{i,t}+{\gamma }_3{Z}_{i,t}+i.year+i.province+{\epsilon }_{i,t}$$

(7)

where ${\alpha }_1$ represents the total effect, ${\theta }_1*{\gamma }_1$ expresses the indirect effect, and ${\gamma }_2$ indicates the direct effect, $Thei{l}_{i,t}$ denotes the degree of urban–rural income gap in the ith province in year t (which is represented by the Theil index in this paper), and $Du{f}_{i,t}$ refers to the degree of digital inclusive finance in the ith province in year t. $A{l}_{i,t}$ illustrates the number of agricultural loans in the ith province in year t, which is utilized as a mediator variable in this paper to denote the rural loan availability as a way to develop the analysis. $Lngd{p}_{i,t}$, $Foreig{n}_{i,t}$, $Ins{h}_{i,t}$, $Ra{t}_{i,t}$, $Ur{b}_{i,t}$, $Ed{u}_{i,t}$, and $Inno{v}_{i,t}$ all stand for other control variables, which are represented in aggregate by $Z_{i,t}$. β depicts the parameter to be estimated, i.year illustrates the time effect of year t, i.province signifies the individual effect of province i, and ${\epsilon }_{i,t}$ describes a random error term representing a set of unobservable factors.

4. Results and Discussion

4.1. Benchmark Regression

This paper employed a benchmark regression analysis on panel data, including 31 Chinese regions from 2011 to 2022. Firstly, we conducted a mixed regression analysis, which was performed by integrating panel data and performing a least squares regression. Column (1) in

Table 3. Results of regression estimation of digital inclusive finance on the urban–rural income gap.

Variable	(1)	(2)	(3)	(4)	(5)
	OLS		RE	FE	FE2
Duf	−0.026 ***	−0.001	−0.002 **	−0.002 ***	−0.007 ***
	(−9.07)	(0.46)	(−1.96)	(−2.58)	(−2.71)
Al	−0.001	0.008 ***	0.000	0.000	−0.000
	(−0.71)	(4.78)	(0.71)	(0.28)	(−0.17)
Lngdp		−0.084 ***	−0.054 ***	−0.052 ***	−0.025 ***
		(−15.94)	(−20.08)	(−19.35)	(−4.91)
Foreign		−2.650	3.700	6.910	6.740
		(−0.51)	(0.73)	(1.27)	(1.17)
Rat		−0.001 ***	−0.000	−0.000	0.000
		(−5.99)	(−0.77)	(−0.54)	(0.93)
Insh		0.078 ***	0.029 ***	0.035 ***	0.059 ***
		(4.74)	(3.20)	(3.79)	(4.51)
Urb		0.000 ***	2.570	−2.290	3.350
		(4.64)	(0.10)	(−0.79)	(1.40)
Edu		−0.000 ***	−0.000 ***	−0.000 ***	−0.000 ***
		(−3.94)	(−7.63)	(−8.23)	(−6.39)
Innov		0.001 ***	0.001 ***	0.001 ***	0.001 ***
		(4.46)	(9.49)	(9.58)	(9.96)
Constant	0.237 ***	0.746 ***	0.622 ***	0.595 ***	0.264 ***
	(13.30)	(15.05)	(21.35)	(20.21)	(3.87)
Sample size	372	372	372	372	372
adj. R2	0.209	0.710	0.571	0.527	0.828
Rho			0.950	0.970	0.986

Notes: *** p < 0.01, ** p < 0.05.

shows that the coefficient without including control variables is −0.026 with a p-value of 0.000. This suggests that digital inclusive finance has a narrowing effect on the urban–rural income gap at the 1% significance level. We added control variables in column (2), showing that the effect of digital inclusive finance on reducing the urban–rural income gap still exists, but the results are not significant. So, we compared the results of different model choices. We included a random effects model to estimate whether individuals are present in the form of random effects. The results are still significant. However, combining the results of the three Rho comparisons, we believe that the fixed effects model is still superior. In addition, we used the Hausman test to complement the model results further. The test results indicate a p-value of 0.000, so the original hypothesis was rejected, and it was concluded that the fixed effects model should be used. Column (3) of the table shows the random effects model estimation results. While the coefficient is −0.002 and significant at the 1% significance level, its adjusted R² is 0.571, and our exploration of the results remains open to further optimization. Columns (4) and (5) of Table 3 show the regression results of the individual and two-way fixed effects models, respectively. The Rho values are 0.970 and 0.986. Individual effects account for a more significant percentage of the total error variance, so we should select the fixed effects model. In addition, the two-way fixed effects model is superior to the individual fixed effects model. The correlation coefficient in column (5) is −0.007 and significant at the 1% level, indicating that the development of digital inclusive finance can significantly reduce the urban–rural income gap and enhance the coordination and sustainability of economic development. Hypothesis 1 is verified.

4.2. Analysis of Robustness

In this section, we test the robustness of the regression results by combining time-period regressions, excluding exceptional sample sizes, and substituting explanatory variables. The concept of digital inclusive finance was proposed for the first time at the G20 summit held in Hangzhou, China, in September 2016. We suspect that the level of digital inclusive finance in China is bound to be different around 2016. Columns (1) and (2) of

Table 4. Robustness test results.

Variable	(1)	(2)	(3)	(4)	(5)
	Theil	Theil	Theil	Theil	Theil
Duf	−0.006 ***	−0.020 ***	−0.005 **
	(−2.73)	(−2.86)	(−1.91)
Dep				−0.007 ***
				(−3.81)
Cov					−0.004 ***
					(−3.65)
Lngdp	−0.340 ***	−0.324 ***	−0.023 ***	−0.290 ***	−0.023 ***
	(−4.39)	(−4.63)	(−4.60)	(−5.90)	(−4.54)
Foreign	−1.060	8.470	2.020	7.450	6.010
	(−1.23)	(0.83)	(0.33)	(1.31)	(1.06)
Rat	0.000	−0.000	0.000	0.000	0.000
	(1.00)	(−0.68)	(1.41)	(0.65)	(1.13)
Insh	0.004	0.086 ***	0.061 ***	0.053 ***	0.058 ***
	(0.22)	(5.96)	(4.61)	(4.04)	(4.47)
Urb	0.000 ***	−5.890 **	4.880 **	3.940 *	3.630
	(5.57)	(−2.47)	(2.07)	(1.65)	(1.52)
Edu	−0.000 **	−0.000 ***	−0.000 ***	−0.000 ***	−0.000 ***
	(−2.36)	(−8.09)	(−3.35)	(−6.79)	(−6.38)
Innov	0.000 **	0.000 ***	0.000 ***	0.001 ***	0.001 ***
	(2.52)	(7.53)	(5.69)	(10.55)	(9.85)
Constant	0.488 ***	0.352 ***	0.237 ***	0.325 ***	0.233 ***
	(4.79)	(5.26)	3.53	(4.68)	(3.44)
Sample size	155	217	324	372	372
adj. R2	0.5322	0.0921	0.4182	0.2321	0.1348

Note: *** p < 0.01, ** p < 0.05, * p < 0.1.

indicate that digital inclusive finance significantly reduces the urban–rural income gap in both the before and after 2016 time periods. Comparing the two, we find that the coefficient of −0.02 after 2016 is larger in absolute value than before and more significant at the 1% level. This suggests that the formalization of the concept of digital inclusive finance has contributed to its development. In addition, considering the actual development situation in China, the economic development, industrial structure, and policy support of Chinese municipalities may differ from those of ordinary provinces. Therefore, we considered conducting a regression to test the robustness of the results after excluding the four municipalities directly under the central government. The results, as shown in column (3) of Table 4, show that the effect of digital inclusive finance on the urban–rural income gap is still significant at the 5% level. Both of the above methods prove that our results are robust. Finally, we considered regressing the explanatory variable Digital Inclusive Finance Total Index (Duf) by replacing it with Digital Inclusive Finance Depth of Impact (Dep) and Breadth of Coverage (Cov). The results in column (4) show that each unit increase in Dep impacts −0.007 units on the urban–rural income gap. Similarly, the column (5) results show a significant negative correlation between Cov and the urban–rural income gap. After replacing the explanatory variables, the results remain significant, which is further evidence of the robustness of our results.

4.3. Analysis of Endogeneity

In the empirical analysis process, there may be an unavoidable endogeneity problem. To alleviate the endogeneity problem caused by omitted explanatory variables, two-way causality, and measurement error, we used one-period lagged digital inclusive finance level (IV1) and digital inclusive finance coverage breadth (IV2 (IV2 has a strong correlation with the endogenous explanatory variable (Duf) of this study and can effectively explain its changes. In addition, IV2 is also consistent with the actual development of the Chinese economy and is not directly related to the urban–rural income gap. This satisfies the requirements for IV selection to a certain extent)) as instrumental variables in this section to conduct two-stage least squares (2SLS) regression. As can be seen in

Table 5. Endogeneity tests of 2SLS regression.

Variable	(1)	(2)	(3)	(4)
	Duf	Theil	Duf	Theil
	Phase I	Phase II	Phase I	Phase II
Duf		−0.083 ***		−0.036 ***
		(−3.23)		(−2.91)
IV1	0.418 ***
	(30.11)
IV2			0.362 ***
			(31.25)
Lngdp	0.093 ***	−0.061 ***	0.109 ***	−0.072 ***
	(10.65)	(−7.29)	(7.16)	(−10.58)
Foreign	0.000	0.000	0.000 ***	0.000
	(0.33)	(1.37)	(3.27)	(1.33)
Rat	0.001 ***	−0.000 ***	0.001 ***	−0.001 ***
	(5.52)	(−3.73)	(4.27)	(−4.54)
Insh	0.037 *	0.061 ***	0.091 **	0.060 ***
	(1.66)	(3.52)	(2.20)	(3.74)
Urb	−0.000 ***	0.000 **	−0.000	0.000 ***
	(−3.30)	(2.55)	(−1.42)	(2.88)
Edu	0.000 ***	−0.000	0.000 ***	−0.000 *
	(3.02)	(−1.28)	(2.64)	(−1.68)
Innov	−0.000	0.001	−0.002 ***	0.001 *
	(−0.44)	(1.37)	(−2.82)	(1.81)
Anderson canon. corr. LM statistic	251.612 ***		273.260 ***
Cragg–Donald Wald F statistic	906.374 [16.380]		976.387 [16.380]
Constant	2.004 ***	0.982 ***	1.070 ***	0.848 ***
	(29.51)	(14.19)	(8.21)	(17.58)
Observations	341	341	372	372
R-squared		0.695		0.712

Note: *** p < 0.01, ** p < 0.05, * p < 0.1.

, both IVs pass the unidentifiable test and the weak instrumental variable test. Columns (1) and (2) in Table 5 show the results of two-stage regressions using one-stage lagged digital inclusive finance (Duf), respectively. IV1 is significantly positively correlated with the explanatory variables, and the estimated coefficient on the core explanatory variable is −0.0825 after eliminating the endogeneity problem using 2SLS, and it acts as a significant reducer of the urban–rural income gap at the 1% level. Columns (3) and (4) show the results of the two-stage regression using the breadth of digital inclusive financial coverage (Cov). IV2 is significantly positively correlated with the explanatory variable, the estimated coefficient of the core explanatory variable is −0.0363 after eliminating the endogeneity problem using 2SLS, and it acts as a significantly narrower of the urban–rural income gap at the 1% level. All these results demonstrate the validity of the selected instrumental variables and the alleviation of the endogeneity problem.

4.4. Analysis of Heterogeneity

We analyzed heterogeneity through both regional and population size heterogeneity. Considering the developmental differences among China’s regions, with the eastern region showing higher economic progress and digital inclusiveness than the relatively less developed central and western regions, this paper conducted regression analyses in four major regions. A sample of 31 regions in China was classified into east, central, west, and northeast regions based on economic and geographic factors (

Table 6. Spatial regional divisions in China.

Area	Provinces
Eastern (area = 1)	Beijing, Tianjin, Hebei, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, Hainan
Central (area = 2)	Shanxi, Anhui, Jiangxi, Henan, Hubei, Hunan
Western (area = 3)	Inner Mongolia, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Xizang, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang
Northeast (area = 4)	Liaoning, Jilin, Heilongjiang

Source: the National Bureau of Statistics of China.

).

According to the results in

Table 7. Analysis of regional heterogeneity in the urban–rural income gap.

Variable	(1)	(2)	(3)	(4)
	Theil	Theil	Theil	Theil
Duf	−0.019 ***	−0.103 **	−0.000	−0.006
	(−3.20)	(−2.04)	(−0.02)	(−1.34)
Lngdp	−0.049 ***	−0.016	−0.031 *	0.026 *
	(−7.83)	(−0.81)	(−1.65)	(1.87)
Foreign	0.000 ***	0.000	−0.000	0.000
	(3.76)	(0.09)	(−0.87)	(1.17)
Rat	−0.000 *	−0.001	−0.001 ***	−0.000 ***
	(−1.69)	(−1.00)	(−5.09)	(−10.29)
Insh	−0.057 **	0.159 *	0.024	−0.011
	(−2.19)	(1.93)	(0.65)	(−0.85)
Urb	−0.000 **	−0.000	0.000	0.001 ***
	(−2.19)	(−0.01)	(1.63)	(4.22)
Edu	−0.000 ***	−0.000	0.000	0.000
	(−5.29)	(−0.75)	(0.78)	(1.60)
Innov	0.001 ***	0.001	0.001	−0.001
	(3.87)	(0.93)	(1.22)	(−1.56)
Constant	0.824 ***	0.281	0.414 *	−0.170
	(9.69)	(1.61)	(1.88)	(−1.54)
Observations	120	72	144	36
R-squared	0.6452	0.8751	0.6459	0.9960

Note: *** p < 0.01, ** p < 0.05, * p < 0.1.

, digital inclusive finance still shows a narrowing effect on the urban–rural income gap in China’s east, central, west, and northeast regions. Hypothesis 1 is again verified. In particular, column (1) indicates that the coefficient of the eastern region is −0.019 and significant at the 1% level. It indicates that digital inclusive finance significantly narrows the urban–rural income gap in the eastern region. Column (2) of Table 7 represents the regression results for the Central region. The coefficient is −0.103 and is significant at the 5% level. However, the impact of digital inclusive finance is weaker in the central region than in the eastern region. Columns (3) and (4) of Table 7 present the regression results for the western and northeastern regions, respectively. Both are not significant, although they still show the narrowing effect of digital inclusive finance on the urban–rural income gap. This suggests that the impact of digital inclusive finance on the urban–rural income gap is heterogeneous across different regions in China, and Hypothesis 2 is verified. We speculate that the reason for this result may be the existence of financial exclusion. Residents in some regions may be unable to access appropriate financial services due to their remote location, low economic development, and financial market regulation. The inability of digital technology to extend to these areas has led to the problem of urban–rural income disparity remaining severe. Reducing the urban–rural income gap and alleviating financial exclusion are issues that digital inclusive finance is committed to addressing.

To further explore the heterogeneity in the impact of digital inclusive finance on the urban–rural income gap, we reclassified the sample according to the number of permanent residents. According to the classification method of Major Figures on 2020 Population Census of China, we divided the 31 regions into four classes according to the population size: more than 100 million people, 50 million to 100 million people, 10 million to 50 million people, and less than 10 million people, as shown in

Table 8. Population size divisions in China.

Population Size	Provinces
More than 100 million (area = 5)	Shandong, Guangdong
50 million to 100 million (area = 6)	Henan, Sichuan, Jiangsu, Hebei, Hunan, Anhui, Hubei, Guangxi, Zhejiang
10 million to 50 million (area = 7)	Jiangxi, Shaanxi, Heilongjiang, Fujian, Liaoning, Shanxi, Jilin, Guizhou, Chongqing, Yunnan, Shanghai, Beijing, Tianjin, Hainan, Gansu, Xinjiang, Inner Mongolia
Less than 10 million (area = 8)	Xizang, Qinghai, Ningxia

Source: the National Bureau of Statistics of China.

below:

Regions with different population sizes may differ in terms of market size, labor supply, and technological innovation capacity. This may affect the relationship between digital inclusive finance and the urban–rural income gap. So, based on the above division, we conducted group regressions for four regions with different population sizes in

Table 9. Analysis of population size heterogeneity in the urban–rural income gap.

Variable	(1)	(2)	(3)	(4)
	Theil	Theil	Theil	Theil
Duf	−0.003 *	−0.011	−0.001	−0.054 ***
	(−12.48)	(−1.43)	(−0.20)	(−14.04)
Lngdp	−0.025 **	−0.033 **	−0.035 *	0.131 **
	(−13.99)	(−2.38)	(−1.90)	(6.37)
Foreign	−0.000 ***	−0.000	−0.000 ***	0.000
	(−242.05)	(−0.23)	(−2.98)	(0.14)
Rat	−0.001 ***	−0.000	0.000	0.002 *
	(−110.82)	(−0.05)	(1.25)	(2.98)
Insh	−0.120 **	0.145	0.039	−0.051
	(−31.11)	(1.58)	(1.04)	(−1.42)
Urb	−0.000 *	0.000	−0.000	−0.000
	(−9.07)	(0.09)	(−0.95)	(−2.70)
Edu	0.000 **	−0.000	−0.000	−0.001
	(20.76)	(−1.52)	(−0.81)	(−1.11)
Innov	0.000 **	0.001 **	0.001 ***	−0.000
	(44.95)	(2.38)	(5.21)	(−1.02)
Constant	0.665 **	0.169	0.402	−0.886 *
	(23.60)	(0.97)	(1.66)	(−3.86)
Observations	24	108	204	36
R-squared	0.999	0.960	0.942	0.991

Note: *** p < 0.01, ** p < 0.05, * p < 0.1.

. Column (1) is for regions with a population size of more than 100 million, and the results indicate that the coefficient is −0.003 and significant at the 10% level. Columns (2) and (3) are regions with populations between 10 million and 100 million. The results show that digital inclusive finance can reduce the urban–rural income gap but not significantly. Column (4) is for areas with a population size of less than 10 million. The coefficient is −0.054 and significant at the 1% level. Overall, the regression results grouped according to population size still confirm that digital inclusive finance narrows the urban–rural income gap but with regional differences. Hypothesis 2 was again tested.

4.5. Analysis of the Mediation Effect

In order to study the role of rural loan availability in the mechanism of digital inclusive finance affecting the urban–rural income gap, our study used three methods, namely, three-step regression and the Bootstrap test to assess the mediating effect. The test results are presented in

Table 10. Results of the three-step regression test.

Variable	(1)	(2)	(3)
	Theil	Al	Theil
Al			−0.000
			(−0.17)
Duf	−0.007 ***	1.165 ***	−0.007 ***
	(−2.81)	(3.89)	(−2.71)
Lngdp	−0.025 ***	1.183 **	−0.025 ***
	(−4.97)	(2.09)	(−4.91)
Foreign	0.000	0.000	0.000
	(1.17)	(0.84)	(1.17)
Rat	0.059 ***	0.017 ***	0.000
	(4.52)	(4.56)	(0.93)
Insh	0.000	1.214	0.059 ***
	(0.92)	(0.82)	(4.51)
Urb	0.000	0.000	0.000
	(1.39)	(1.48)	(1.40)
Edu	−0.000 ***	0.002	−0.000 ***
	(−6.41)	(0.62)	(−6.39)
Innov	0.001 ***	0.008	0.001 ***
	(9.98)	(0.64)	(9.96)
Constant	0.265 ***	−11.713	0.264 ***
	(3.91)	(−1.52)	(3.87)
Observations	372	372	372
R-squared	0.928	0.451	0.928

Note: *** p < 0.01, ** p < 0.05.

and

Table 11. Results of the Bootstrap test.

Mediation Effect Type	Z	p > Z	95% Conf. Interval
Indirect effect	2.19	0.028	0.0003635	0.0064404
Direct effect	−0.39	0.693	−0.0044497	0.0066922

.

First, we used the stepwise regression method proposed by Baron and Kenny to conduct regression analyses of the mediating role among the three. Based on the model developed in Section 3.3, the analysis process is as follows. The first step involves testing whether the coefficient ${\alpha }_1$ in the model (5) is significant. Column (1) of Table 10 shows that the coefficient is −0.007 and is significant at the 1 percent level. This indicates that digital inclusive finance has a significant impact on narrowing the urban–rural income gap. The second step tests the coefficients ${\theta }_1$ and ${\gamma }_1$ in models (6) and (7), respectively. If neither coefficient reaches significance, this indicates an absence of a mediation effect. The coefficient of the second column in Table 10 is 1.165 and significant at the 1% level, indicating that digital inclusive finance contributes to the increase in rural loan availability, and Hypothesis 3 is tested. In addition, in model (7), a significant ${\gamma }_2$ indicates a partial mediation effect, while a non-significant indicates a full mediation effect. The third column of regression results in Table 10 also suggests that there may be a partial mediating effect of rural loan availability in the process of reducing the rural–urban income gap with digital inclusive finance.

To further validate the reliability of the regression results, we drew on Bradley Efron’s study and used the Bootstrap test for the follow-up study; see Table 11. The first row indicates the indirect effect results for mediated effects. If the interval includes 0, this signifies the absence of a mediation effect. On the other hand, if the interval excludes 0, it suggests a mediation effect. The Bootstrap test results indicate that the 95% confidence interval of the indirect effect does not include 0. The above test results indicate a partial mediation effect of rural loan availability on the effect of digital inclusive finance on the urban–rural income gap. Hypothesis 4 is verified.

Rural loan availability embodies a debilitating effect. Although there is little literature on this phenomenon, it is consistent with the current context. Based on these findings, this paper proposes several potential explanations. While the growth of digital inclusive finance has increased the availability of rural loans, many problems have arisen. Firstly, rural areas have inefficient use of finance due to problems with infrastructure and institutional set-up. Secondly, rural residents’ financial literacy and asset management capabilities need to be improved. All these reasons may lead to rural capital not effectively supporting local economic development and instead flowing to the cities.

4.6. Analysis of the Threshold Effect

The panel data regression results above confirm that digital inclusive finance has varying impacts on the urban–rural income gap across different regions of China. However, these findings are based on the assumption of linear impacts and do not consider the potential for variation across different stages of development. The threshold effect is when one economic parameter, upon reaching a specific value, triggers a sudden shift in another economic parameter, which leads to other forms of development (structural mutation). Following the approach of Xiong et al., our study utilized the urbanization level as a threshold variable [31]. The Bootstrap method was employed to obtain the test statistic’s significance level, with the measured value’s magnitude representing the basis for assessing the presence of a threshold effect. The results are presented in

Table 12. Results of the threshold effect test.

	F-Statistic	p-Value	10%	5%	1%
Single Threshold Test	62.90	0.0200	37.5292	47.4116	77.1998
Double threshold test	25.21	0.2800	41.3154	52.6123	69.2075

Note: p-values and critical values are the results of 300 repeated samples utilizing the Bootstrap method.

.

As illustrated in Table 12, when employing the urbanization level as the threshold variable, the single threshold effect yields an F-value of 62.9 with a corresponding p-value of 0.02. This indicates that the single threshold effect is significant at the 5% significance level. In contrast, the double threshold effect produces an F-value of 25.21 with a p-value of 0.28, signifying that it does not pass the significance test. Therefore, this paper adopted the single threshold value as the method for the following measurements in our study. In addition, this paper estimated the threshold values, with the results displayed in

Table 13. Results of the threshold estimate.

Model	Threshold Estimate	95% Confidence Interval
Single Threshold	0.4189	[0.4105, 0.4198]

. A single threshold LR diagram was also generated to offer a more intuitive visualization of these results, see Figure 2.

Upon establishing a single threshold value, this paper categorized provinces based on their urbanization level from 2011 to 2022. Provinces were grouped into two intervals: those with an urbanization level less than or equal to 0.4189 (Urb ≤ 0.4189) and those exceeding this threshold (Urb > 0.4189). A panel threshold regression analysis was conducted separately for areas exhibiting relatively low and high levels of urbanization development. The findings are presented in

Table 14. The threshold effect of digital inclusive finance on the urban–rural income gap.

Variable	Parameter Estimation
Lngdp	−0.0253 **
	(0.01)
Foreign	−0.0147 *
	(0.01)
Insh	0.0378 *
	(0.02)
Rat	0.0000 ***
	(2.80)
Urb	−0.1421 **
	(0.05)
Edu	0.0003
	(0.61)
Innov	2.2622
	(1.58)
Duf_1(urb ≤ 0.4189)	−0.0294 ***
	(−0.01)
Duf_1(urb > 0.4189)	−0.0232 ***
	(−0.01)
Constant term	0.328 ***
	(0.11)
Observations	372
R-squared	0.959

Note: *** p < 0.01, ** p < 0.05, * p < 0.1.

.

Table 14 indicates that digital inclusive finance significantly affects both sides of the urbanization threshold, albeit with varying degrees of impact. In areas where urbanization declines below 0.4189, digital inclusive finance yields a marginal effect of 0.0294. In contrast, when the urbanization level surpasses the critical value of 0.4189, the results indicate that while urbanization contributes to reducing the urban–rural income gap through digital inclusive finance, this reducing effect weakens as urbanization rates continue to rise beyond this threshold. Therefore, to counteract the potential for urbanization to exacerbate the urban–rural income gap, governmental bodies should prioritize expanding the implementation and reach of digital inclusive finance.

5. Conclusions and Policy Recommendations

5.1. Conclusions

With its immense generative capacity, the digital economy has emerged as a critical driver of economic growth and a potent force for narrowing the income gap between urban and rural areas. This study examines the relationship between digital inclusive finance and the urban–rural income gap and focuses on the mediating role of rural loan availability. We use powerful technical tools such as panel data, two-way fixed effects models, heterogeneity analysis, mediation effects analysis and threshold effects analysis to complement and justify the theory and hypotheses. Our study draws the following conclusions and provides essential policy recommendations and practical implications for relevant researchers and policymakers.

First, the development of digital inclusive finance in China has significantly narrowed the urban–rural income gap. A two-way fixed-effects model is selected for this research and analysis following a series of preliminary tests. The results confirm the feasibility of this paper’s hypotheses. Amid the surge in digital development, traditional inclusive finance has been digitally integrated, positively affecting China’s less developed rural areas. This impact is twofold: firstly, it enables rural residents to access financing for their production and living needs, cultivating rural industrial development and increasing disposable income; secondly, it empowers rural residents with mobile payment options and online access to financial services, expanding their financial management channels and shrinking the income gap.

Second, the development of digital inclusive finance in China demonstrates regional heterogeneity. In Section 4.4, we analyze the regional and population size heterogeneity in the impact of digital inclusive finance on the urban–rural income gap. The results show that digital inclusive finance is more effective in reducing the urban–rural income gap in eastern regions and regions with larger populations. These regions have well-developed infrastructures, high levels of policy implementation, and advanced digital technologies. These factors create an enabling environment for the growth of digital inclusive finance, facilitating greater access and utilization of financial services by rural populations, thereby effectively bridging the urban–rural income divide.

Third, rural loan availability in China plays a partly mediating role in narrowing the urban–rural income gap. This paper utilizes the three-step regression test and Bootstrap test to test the mediation effect of rural loan availability. The results show that the development of digital inclusive finance can expand the breadth of financial services coverage and thus increase the availability of rural loans. After adding the mediating variables, the effect of digital inclusive finance on narrowing the urban–rural income gap is still significant but, to some extent, weakened. This suggests that increasing rural loan availability does not mean that rural funds can be effectively utilized. An increase in unproductive expenditures and a reverse flow of funds to the cities may result from widening access to finance. Financial institutions need to be reformed, and farmers’ financial literacy and money management skills must be improved. This finding has implications for institutional reform and policymaking in China.

Fourth, the role of digital inclusive finance in narrowing the urban–rural income gap exhibits nonlinear characteristics. This paper utilizes the urbanization level as the threshold. When urbanization is below the threshold value, digital inclusive finance exhibits a more substantial narrowing effect on the urban–rural income gap. This is attributable to the disadvantages faced by rural residents, both in terms of their financial literacy and the developmental limitations of their regions of residence. However, as urbanization progresses, the effectiveness of digital inclusive finance in narrowing the urban–rural income gap becomes impeded.

Our study contributes to a clearer understanding of the relationship between digital inclusive finance and the urban–rural income gap. However, this study also has the following limitations. Firstly, this study’s results only use China’s provincial-level regions as a sample, which may differ from other developing countries. In the future, we aim to research on a larger scale to provide insights for other countries worldwide. In addition, our mediation effects test results indicate a weakening effect of rural loan availability. An increase in loan availability enhances farmers’ access to finance and may influence their household consumption patterns. Inefficient use of finance and outflows due to increased unproductive consumption could contribute to this result. Our research on the relationship between digital inclusive finance and consumption patterns is limited, and we can add relevant and more in-depth studies in the future.

5.2. Policy Recommendations

The preceding analyses demonstrate the overwhelmingly positive effect of digital inclusive finance on regional economic growth. It has proven crucial in promoting rural revitalization strategies, narrowing the urban–rural income disparity, and improving financial accessibility in rural areas. However, these analyses also highlight challenges, particularly in allocating and utilizing rural loans. To address these challenges, governmental intervention through reasonable policies is essential. This paper offers the following suggestions.

First, guiding rural financial institutions through digital transformation is crucial for increasing rural loan availability. As digitalization and financial services deeply integrate, societal structures progressively shift toward intelligent systems. However, there needs to be more developed digital infrastructure in many rural areas of China to ensure the expansion of digital inclusive finance. Therefore, the government should prioritize increased support for developing rural digital infrastructure. This support should include financial investment, policy development, and technical assistance. In alignment with new goals for quality productivity development, encouraging rural financial institutions to utilize digital platforms and technologies for loan management is critical. Moreover, government departments should collaborate on multifaceted strategies to increase subsidies for rural financial institutions’ agriculture-related support projects. For instance, the central bank could establish differentiated reserve requirement ratios for agricultural loans or offer differentiated refinancing at narrowed interest rates. Simultaneously, the tax department could introduce preferential tax rates for agricultural loan projects and services. Offering appropriate tax incentives and exemptions for agricultural loans would lower the cost of lending for financial institutions, finally increasing loan investment in rural areas.

Second, expanding access to digital inclusive finance can empower rural residents to manage their finances better. The results of the mediating effect test suggested that the existence of low financial literacy, poor infrastructure, and low household money management skills in the rural population led to the low utilization of rural funds. Many financial transactions require digital platforms and skills, considering the current integration of digital technology and financial services. To address this, the government should prioritize digital skills training programs to equip rural residents with the knowledge and abilities needed to understand digital technology. In addition, government agencies should collaborate with financial institutions to organize public awareness campaigns and establish digital platforms dedicated to inclusive finance. These efforts will ensure that rural residents have access to the information and resources necessary to make informed financial decisions. Farmers can effectively leverage rural loans to improve their livelihoods only when they have enhanced financial management skills. Accordingly, this will reduce poverty in rural areas and bridge the income gap between urban and rural populations.

Third, policymakers should implement regionally differentiated strategies to address the unique developmental needs of diverse areas. The empirical findings presented earlier demonstrate that the effects of digital inclusive finance on the urban–rural income gap vary significantly across regions. Therefore, governmental efforts to cultivate the growth of digital inclusive finance should be context-specific. The central and western regions needed to increase investment in capital, technology, infrastructure, and other vital resources, while the northeastern region should promote reform of old traditional industrial zones and foster new industries. Such actions would effectively bridge the “digital divide” between economically advanced regions and their less developed counterparts, whereas, in the eastern regions, the focus should shift towards incentivizing financial institutions to pioneer innovative development models, deliver a more comprehensive array of high-quality financial services, and optimize their governance structures. These measures would cater to the specific requirements of residents and deepen the integration of digital inclusive finance into their lives. Simultaneously, policymakers must prioritize balanced interregional development and connectivity, cultivating a cohesive, regionally driven growth strategy.

Fourth, strengthening the regulation of digital inclusive finance and enhancing the quality of public services is crucial. The threshold effect test shows that the impact of digital inclusive finance on narrowing the urban–rural income gap shows a non-linear relationship with urbanization. This paper hypothesizes that this fluctuation is attributable to the various factors that affect both the implementation of digital inclusive finance and the quality of public services. To ensure compliance and risk control, the government should create a comprehensive regulatory framework for digital inclusive finance that clearly defines the roles and responsibilities of regulators. Regulators should establish adaptable rules for the specific characteristics of digitally inclusive finance, including digital identity verification, data privacy protection, and cybersecurity. Moreover, to maintain market order, they should strengthen their regulatory and enforcement capacities and levy significant penalties for non-compliance. The government, regulators, financial institutions, and technology companies should increase cooperation, share information, experience, and technology, and collaborate to promote developing and regulating digitally inclusive financial services. In terms of mechanism development, strengthening the complaint and supervisory apparatus is essential. A robust complaint channel and supervisory mechanism will increase public trust and satisfaction with public services.