Does digital financial inclusion reduce the risk of returning to poverty? Evidence from China

doi:10.21203/rs.3.rs-2114509/v1

Download PDF

Research Article

Does digital financial inclusion reduce the risk of returning to poverty? Evidence from China

https://doi.org/10.21203/rs.3.rs-2114509/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 04 Apr, 2023

Read the published version in International Journal of Finance & Economics →

Version 1

posted

You are reading this latest preprint version

Digital financial inclusion has become an important way to reduce poverty and prevent poverty return; however, few studies examine the relationship between digital financial inclusion measurement with poverty return governance. Based on data from the 2017 China Household Financial Survey, we construct a digital financial inclusion indicator for micro-households, and explore its impact on the risk of households returning to poverty and its mechanism of action. Our findings suggest that digital financial inclusion can reduce the risk of Chinese families returning to poverty, and that it has heterogeneous effects on families and regions with different characteristics. The main function is to improve household income level by promoting entrepreneurship and employment, and to improve risk resistance by enhancing household financial market participation and household asset allocation. Further analysis shows that digital financial inclusion has structural effects, nonlinear effects, and substitution effects with private lending in poverty governance. This paper has implications for understanding and improving the poverty governance effectiveness of digital financial inclusion.

JEL Codes: D14; I32; O33

Digital Financial Inclusion

Risk of Returning to Poverty

Poverty Vulnerability

During the course of financial poverty alleviation, poverty governance has gradually changed from traditional finance to a combination of traditional finance and digital finance (Yu and Wang, 2021). Financial inclusion has always been regarded as an important tool to alleviate poverty. Based on the principle of equal opportunity, it emphasizes the breadth and extension of financial services and aims to provide adequate financial support to the poor and other vulnerable groups. Characterized by both digitization and inclusiveness, digital financial inclusion has become an important path to reduce poverty and prevent poverty return (Jeanneney and Kpodar, 2011; Neaime and Gaysset, 2018; Seven and Coskun, 2016). Although China built a moderately prosperous society, in an all-round way, in 2020, poverty-stricken households and marginal households return to poverty frequently: This has become a new feature of poverty governance after comprehensive poverty alleviation in 2020 and a new problem that restricts China’s social development (Yu, 2019). China held the G20 Summit and proposed digital financial inclusion in 2016: “By penetrating geographical space, reducing service costs and optimizing information exchange, the service threshold of inclusive finance become much lower”, which pushes financial services to lower-income groups and makes it an important driving force for poverty control and prevention(Chen and Yuan, 2021). In recent years, more scholars have paid attention to the contribution of digital inclusive finance to poverty reduction (Allen et al., 2013; Lal, 2018; Pu et al., 2019; Yu and Wang, 2021; Zhu and Wang, 2017); however, studies on the impact of digital financial inclusion on poverty reduction have been restricted to the macro level, lacking thinking from the micro perspective. In addition, none of these studies delve into its value of poverty control, especially in curbing the return to poverty.

Therefore, it is of great theoretical and practical significance to discuss the relationship between digital financial inclusion and the vulnerability of families to poverty. From this perspective, we consider whether financial inclusion can reduce the probability of returning to poverty by providing financial services and risk management tools for those at potential risk of returning to poverty. Further, through what mechanism does it come into play? Will it bring “elite captured” or “universal benefit” for families with different characteristics? Is it a “timely assistance” or “icing on the cake” effect for disadvantaged groups? Will it be a “digital divide” or a “digital dividend” in terms of the containment effect of returning to poverty? Discussion of these issues is of great importance for China’s poverty governance in the “post-2020 era of poverty alleviation” and provides reference to other developing countries seeking to formulate public finance poverty alleviation policies.

To answer these questions, we used CHFS data in 2017 to construct the household digital financial inclusion indicator at the micro level, and discussed the effect of digital financial inclusion on preventing the risk of returning to poverty. The heterogeneity of the effect of digital financial inclusion on the risk of returning to poverty was explored based on different family endowments and the degree of development of digital finance. We then further investigated the function mechanism, structural effect, nonlinear effect, and relationship with informal finance of digital financial inclusion on the risk of returning to poverty.

The main contributions of this paper are reflected in the following three aspects:

First, from the perspective of research content, this paper studies the role and effect of digital financial inclusion in poverty governance from the perspective of the risk of returning to poverty. When exploring the relationship between financial inclusion and poverty governance, extant literature mainly focuses on the effect of financial inclusion on economic poverty and poverty reduction in a single income dimension (Allen et al., 2013; Lal, 2018; Pu et al., 2019; Seven and Coskun, 2016; Yu and Wang, 2021; Zhu and Wang, 2017). Few studies have examined the impact of financial inclusion or digital financial inclusion on the risk of returning to poverty, and even fewer studies discuss it from the micro-level perspective (Yin and Zhang, 2020). This paper expands the boundary of research on the effect of financial inclusion on poverty management, especially the risk of returning to poverty, and provides a new perspective for understanding the effectiveness of digital financial inclusion for poverty management.

Second, from the perspective of research methods, different from the existing analysis based on macro panel data (Amidžic et al., 2014; Beck et al., 2007; Park and Mercado, 2016; Sarma, 2008), this paper not only considers the reality of economic development at a more subtle level, but also selects some factors that influence household financial level; furthermore, it adopts the “Euclidean Distance Method” to construct a digital financial inclusion index at the household level. Its advantage is that it is easy to calculate, and it does not impose different weights for each dimension (Sarma, 2008, 2016). Additionally, this paper measures the risk of returning to poverty based on the improved poverty vulnerability framework. Most previous studies construct poverty vulnerability indicators based on objective factors such as family endowment, population structure, and human capital (Bjorn, 2004; Giuseppina et al., 2016). However, the subjective factors pertinent to families cannot be ignored when measuring the risk of returning to poverty. If a family lacks internal motivation, it will often fall into poverty again once it is no longer receiving help, regardless of how much help it initially receives from others (Fu et al., 2020; Liu et al., 2020). Therefore, this paper constructs the endogenous motivation index of households to escape poverty and incorporates it into the traditional poverty vulnerability framework to calculate the risk of households returning to poverty.

Third, from the perspective of research content, this paper attempts to examine the influence of digital financial inclusion on the risk of family poverty from multiple perspectives. To begin with, this paper constructs the micro household financial inclusion index to explore the relationship between the two. Then, heterogeneity analysis is conducted from the perspectives of family endowment and regional digital economy development to verify the beneficial and inclusive characteristics of digital financial inclusion, which enriches the literature on the evaluation of inclusive finance systems. In addition, from the perspective of household income and asset allocation optimization, this paper discusses how digital financial inclusion affects poverty vulnerability of micro households. Furthermore, we discuss the structural effect and nonlinear effect of digital inclusive finance on the risk of returning to poverty, as well as its relationship with informal finance (private lending), which enriches the depth of knowledge on the effect of digital financial inclusion on poverty governance.

This paper is organized as follows. Section 2 comprises the literature review. Section 3 provides theoretical analysis and presents the research hypothesis. Section 4 describes the method for the construction of our financial inclusion indicator and measurement of other indicators. Section 5 presents the paper’s main empirical results. Section 6 concludes with policy implications.

In this section, we review and clarify the definition of financial inclusion as well as the role and impact of traditional financial inclusion and digital financial inclusion in poverty governance.

2.1 Definition and measurement of financial inclusion

There are different definitions of financial inclusion in existing literature. The earliest studies defined it from the perspective of financial exclusion. Leyshon and Thrift (1995) were the first to define financial exclusion as those processes that prevent poor and disadvantaged social groups from entering the financial system, focusing on geographic access to financial services, particularly bank branches. They argued that financial exclusion widens geographic differences in income levels and economic development. The 2014 Global Financial Development Report (World Bank, 2014) identified four main forms of financial exclusion, divided into voluntary exclusion and involuntary exclusion. Voluntary exclusion refers to the voluntary choice of certain people or businesses not to use certain financial services because of the lack of prospects for the project or for cultural or religious reasons. Involuntary exclusion refers to the exclusion of those who do not have enough income or cannot obtain financial services because of objective factors such as high loan risks and inadequate financial markets. Sarma (2008) directly defined the concept of financial inclusion, which refers to the access, availability, and convenience of the formal financial system. The concept of financial inclusion can be broadly defined as an economic state in which individuals and firms are not denied access to basic financial services for reasons other than efficiency standards. In recent years, international organizations or institutions have also directly defined digital financial inclusion. Financial inclusion was first clearly proposed by the United Nations at the 2005 “International Year of Microcredit” Conference, which defined it as “a financial system that can effectively and comprehensively provide services to all social classes and groups”. The European Commission (2008) defined financial exclusion as “a process whereby people encounter difficulties accessing and using financial services and products in the mainstream market that are appropriate to their needs and enable them to lead a normal social life in the society to which they belong to”. According to the Development Plan for Promoting Inclusive Finance (2016–2020) issued by The State Council of China in 2016, financial inclusion refers to the provision of appropriate and effective financial services at an affordable cost to all social strata and groups in need of financial services, based on equal opportunity requirements and the principle of business sustainability.

Combining the various definitions above, it can be concluded that financial inclusion should have the following characteristics: inclusiveness and policy support, especially for farmers, micro enterprises, low-income groups, and other special groups such as the disabled and the elderly. In recent years, the digital economy has created development opportunities for traditional financial inclusion, and digital financial inclusion has emerged as a result. Digital financial inclusion puts more emphasis on digitalization, as a result of which it can be said that digital financial inclusion is the combination and application of traditional financial inclusion and modern digital technology, and therefore still essentially belongs under the category of financial inclusion.

In addition, the construction of the financial inclusion index is also a core contribution of this paper. Measuring the development level of financial inclusion involves multiple indicators of different dimensions, and the design of a complete and accurate index system for inclusive finance is also being gradually explored(Sarma, 2008; Amidžic et al., 2014; Park and Mercado, 2018; Yang and Fu, 2019). At present, financial inclusion evaluation indices can be macro-level, such as national, regional, or provincial financial inclusion indices; or micro-level, such as the community and household financial inclusion indices; however, neither of them specifically emphasize digitalization. At the macro level, several studies have discussed different levels of financial inclusion development in advanced and emerging economies. Among the early evaluation measures of financial inclusion, Sarma (2008) adopted the distance-based method that was used by the United Nations Development Program (UNDP) to calculate well-known development indexes such as the Human Development Index (HDI) and to construct the financial inclusion index of Asian countries from the respective of Bank Penetration Rate (BPR), Banking Service Availability (BSA), and Banking System Utilization (BSU). Amidžic et al. (2014) derived a new comprehensive index of financial inclusion at the country level by weighting the Financial Access Survey (FAS) database of the International Monetary Fund (IMF) with the method of factor analysis. Park and Mercado (2018) construct financial inclusion indicators at the country level and evaluate various macroeconomic factors affecting the degree of financial inclusion in 176 economies.

Digital finance is a frontier innovation on a global scale; however, there are few data available for research, and the development of inclusive finance in China is relatively slow compared with that in developed western economies. Therefore, the evaluation of inclusive finance is immature. The China Digital Inclusive Finance Development Index jointly compiled by the Digital Finance Research Center of Peking University and the Ant Financial Group is and widely accepted and used. Based on the huge data system of Ant Financial’s account transaction records, they depicted the development of digital finance in China based on different dimensions, proposed a measurement method for the development level of inclusive finance at the provincial level, and built a regional digital financial inclusion index. Yang and Fu (2019) constructed a rural digital financial inclusion index at the provincial level in China and tested its poverty-reduction effect based on the data of China Family Panel Study (CFPS). However, these evaluations were performed at the macro level, and there are few studies focusing on the micro level. Furthermore, Zhang and Yin (2018) used the factor analysis method to construct the financial inclusion index at the village level based on the three dimensions of penetration, usability, and satisfaction of financial services. Yin et al. (2019) constructed the financial inclusion index system from the perspective of the supply side and the demand side, and synthesized the household financial inclusion index by principal component analysis.

2.2 The impact of financial inclusion on poverty governance

Financial inclusion has been regarded as an important mechanism to alleviate social poverty. Its original intention is to support vulnerable groups, help them seize production opportunities, and improve human capital to increase income and eliminate poverty and inequality. Although existing studies have failed to form a unified statement on the functional impact of financial inclusion on poverty governance, the mainstream view is that financial inclusion has a positive influence on poverty reduction. Allen et al. (2013) pointed out that commercial banks could help improve the access of the impoverished to financial services in Kenya. Zhu and Wang (2017) found that although the development of financial inclusion is beneficial in terms of increasing the income of rural residents, its role in poverty reduction and income increase varies by income group based on data from counties in China. For example, the role of rural residents in poor counties in poverty reduction and income increase is significantly smaller than that in non-poor counties. Conversely, Seven and Coskun (2016) argued that the financial development of banks and stock markets in emerging countries fails to benefit the poorest social class. Although financial development promotes economic growth, banks and stock markets do not play an important role in poverty reduction.

In recent years, digital financial inclusion has begun to break through spatial and geographical restrictions, improving the financial availability of the poor and significantly reducing family poverty. Yu and Wang (2021) believed that although digital financial inclusion plays a significant role in narrowing the income gap between urban and rural areas and improving unbalanced development, there is an unbalanced effect between regions. Chen and Zhao (2021) found that digital finance significantly inhibits absolute and relative poverty of rural households in China, and its poverty-reduction effect mainly alleviates poverty by easing credit and information constraints and expanding social networks.

Existing studies may have several shortcomings. First, there is no consensus on the poverty-reduction effect of financial inclusion, and there is little research on the links between poverty governance, poverty vulnerability, and curbing the risk of returning to poverty. Second, although digital technology as a support tool can effectively compensate for the deficiencies in efficiency and quality of financial services of traditional financial institutions, the evaluation of digital financial inclusion system: in particular, the construction of the evaluation system at the micro level, needs to be further improved. Therefore, this paper aims to construct an evaluation system of household digital financial inclusion at the micro level, and further explore the influence of digital inclusive finance development on poverty governance, especially the risk of returning to poverty, as well as the mechanism underlying its effect.

By combining financial poverty reduction theory and the development of digital financial inclusion in China, we summarize the mechanism underlying the effect of digital financial inclusion on the risk of returning to poverty among families. The effect of digital financial inclusion on the risk of family poverty return is mainly manifested in promoting entrepreneurship and employment to increase income and optimizing asset allocation to improve risk resistance.

For households that have been removed from poverty, income failure is the main cause of their return to poverty; future risk shocks such as high incidence of natural disasters, accidents, disability, serious illness, and unstable employment are more specific causes (Hao, 2019). The former is relative to the current income status of the family, while the latter mainly reflects the ability of the family to resist unknown risks, and is more related to the level of family wealth or assets.

We argue that the first channel for digital financial inclusion to exerts its effects in reducing the risk of family poverty return is to promote the increase of family income, and its influence is mainly reflected in the following two aspects: First, digital financial inclusion affects the profits of individual businesses by promoting residents’ entrepreneurship and improving family income. Capital constraint is an important factor affecting the survival and development of entrepreneurs and enterprises (Bianchi, 2010; Bruton et al., 2010). According to the characteristics of inclusiveness and policy support of digital financial inclusion, digital financial inclusion can directly expand the coverage and improve the availability of finance to individuals, which reduces the credit constraint and threshold of entrepreneurship for entrepreneurs (Dutta and Meierrieks, 2021). In addition, through digital technology, digital financial inclusion supervises financing risks, reduces financing costs caused by information asymmetry, and effectively solves the problem of financing difficulties and high financing costs (Gomber et al., 2017; Singh, 2012). Second, digital financial inclusion can increase household income by promoting employment. The establishment and development of enterprises create jobs, and non-entrepreneurial families can increase their household income by obtaining employment opportunities (Ayyagari et al., 2014; Bruhn and Love, 2014). Moreover, digital financial inclusion can ease the financial constraints of the labor force, provide financial support for the accumulation of human capital and payment of transfer costs, and reduce the difficulty in finding employment (Mugo and Kilonzo, 2017).

Therefore, we propose the first hypothesis:

H1: Digital financial inclusion can reduce the risk of returning to poverty by promoting the establishment and development of enterprises, improving employment, and thus increasing the income of families.

In addition, the second influence channel is via promoting the optimization of household asset allocation and improving households’ resistance to future risk shocks. Financial literacy is a major factor influencing household investment decisions and diversification. Existing studies have shown that financial literacy affects family asset allocation and investment decisions. The higher the financial literacy of family members, the stronger their awareness of future risk aversion, and the stronger their willingness to invest financial products (Bernheim and Garrett, 2003; Luigi and Jappelli, 2008). This is mainly reflected in that digital financial inclusion accelerates the popularization of financial knowledge, and the increase in financial knowledge reserves will, in turn, promote households’ participation in the financial market; in particular, the proportion of households’ risk assets will increase significantly (Lu et al., 2021; Wang, 2009); furthermore, residents’ risk tolerance and financial literacy may be further improved in the process of daily use of digital financial products and gradually understanding the products, thereby stimulating residents’ decentralized investment (Fernandes et al., 2014; Jappelli and Padula, 2013; Liao et al., 2017).

Accordingly, we propose the second hypothesis:

H2: Digital financial inclusion can reduce the risk of returning to poverty by improving household financial literacy and financial market participation, optimizing household asset allocation, and thus improving risk resistance.

4.1 Model construction

To verify the impact of digital financial inclusion on the risk of returning to poverty, we applied a Probit model as follows:

$\Pr (Ris{k_i}=1{\text{| }}IFI,{X_i})={\alpha _0}+{\alpha _1}IF{I_i}+{\sum \alpha _i}{X_i}+{\mu _i}$ (1)

where the explained variable is $Ris{k_i}$, and $Ris{k_i}=1$ represents the families with the risk of returning to poverty under a specific standard. $IF{I_i}$ represents the index of household digital financial inclusion. ${X_i}$ denotes a series of control variables, including individual, family, and province-level characteristics. The term ${\mu _i}$ is a random disturbance term. We applied the coefficient ${\alpha _1}$ of $IF{I_i}$to verify the influence of digital financial inclusion; a significantly negative value of ${\alpha _1}$ indicates that the household digital financial inclusion index is significantly negatively correlated with the risk of returning to poverty. Conversely, a significantly positive value of ${\alpha _1}$ indicates that the household digital financial inclusion index is not conducive to reducing the risk of returning to poverty.

4.2 Data

For empirical research, we used data from the CHFS, which is a nationwide sample survey project conducted by the China Household Finance Survey and Research Center of Southwestern University of Finance and Economics. The project aims to collect relevant information, at the micro level, on household finance nationwide through scientific sampling, including demographic characteristics and information on employment, assets and liabilities, income and consumption, social security and insurance, and subjective attitudes. To date, four rounds of the survey have been carried out. Based on the needs of our research, we selected data from the fourth round of the China Household Finance Survey in 2017. The survey samples covered 29 provinces, 355 counties, and 1,428 communities across the country, with a sample size of 40,011 households. In addition, we also controlled the regional economic and demographic characteristics variables collected from the National Bureau of Statistics.

To construct effective samples, we processed the original data using the following steps. First, we selected the sample data and variables via questionnaires administered to families and individuals and by using a comprehensive non-questionnaire variable database, and then matched them accurately. Specifically, we paged against the uniform personal codes in the personal and family database and then matched against the data for the composite variable. Second, we excluded samples with missing values for core variables.

4.3 Variables

4.3.1 Household Digital Financial Inclusion Index

To design the micro digital financial inclusion indicator system, it is necessary to fully consider the macroscopic reality of China’s economic development and financial exclusion as well as the financial characteristics of Chinese households at the micro level. Based on the main macro financial inclusion index systems and drawing on the practices of Yin et al. (2019) and Sarma (2008, 2016), we used the Euclidean distance method to design the digital financial inclusion index of households. The advantage of this approach is that it is easy to calculate the index, and it does not impose different weights on each dimension.

The first step is to design the index system. Based on the data from the China Household Financial Survey, we selected seven indicators from the three levels of digitalization, usability, and coverage, as shown in the table below.

Table 1

Composite sub-dimension indicators of digital financial inclusion
Dimension	Variable name	Definition
Digitalization	E-payment	Have you ever used electronic payment? 1 = Yes; 0 = No
Digitalization	Internet finance	Have you ever had an Internet financial account? 1 = Yes; 0 = No
Usability	Insurance	Have you ever purchased commercial insurance or social insurance services? 1 = Yes; 0 = No
	Loan	Have you ever had a loan from a bank? 1 = Yes;0 = No
	Credit card	Have you ever had a credit card? 1 = Yes; 0 = No
	Deposit	The number of deposit cards or current passbooks
Coverage	Distance	Distance from home to the nearest bank. (This index is treated logarithmically)

The second step is to standardize digital financial inclusion indicators. To solve the dimension problem of the data index, we used deviation standardization to deal with the original index data. The index of the ${d_i}$ is computed by the following formula:

${d_i}=\frac{{{A_i} - {m_i}}}{{{M_i} - {m_i}}}$ (2)

where ${A_i}$ is the true value of the dimension, ${M_i}$ is the maximum value, and ${m_i}$ is the minimum value. The purpose of the above formula is to standardize in order to ensure that the value interval of each indicator is [0,1]; through standardization, each indicator can meet the unit independence and boundedness. The higher the value of ${d_i}$, the higher the level achieved by this indicator.

The third step is to synthesize a digital financial inclusion index. Referring to the practice of Yin et al. (2019) and Sarma (2008, 2016), and the equal weight assumption and the average Euclidean distance method to sum up sub-indexes, the financial inclusion index is constructed:

$IF{I_{i1}}={{\sqrt {\sum\nolimits_{{i=1}}^{n} {d_{i}^{2}} } } \mathord{\left/ {\vphantom {{\sqrt {\sum\nolimits_{{i=1}}^{n} {d_{i}^{2}} } } {\sqrt n }}} \right. \kern-0pt} {\sqrt n }}$ (3)

$IF{I_{i2}}={{1{\text{-}}\sqrt {\sum\nolimits_{{i=1}}^{n} {d_{i}^{2}} } } \mathord{\left/ {\vphantom {{1{\text{-}}\sqrt {\sum\nolimits_{{i=1}}^{n} {d_{i}^{2}} } } {\sqrt n }}} \right. \kern-0pt} {\sqrt n }}$ (4)

$$IF{I_i}=\frac{{IF{I_{i1}}+IF{I_{i2}}}}{2}$$

$$IFI=\frac{{IF{I_1}+IF{I_2}+IF{I_3}}}{3}$$

where$IF{I_{i1}}$,$IF{I_{i2}}$, and$IF{I_i}$ represent the distance from the actual point to the worst point, the reverse distance to the best point, and the average distance of each indicator under the ${d_i}$ dimension, respectively. $IFI$ is the final household digital financial inclusion index obtained by summing up the average three sub-dimension indexes.

4.3.2 The risk of returning to poverty

According to the World Bank, poverty vulnerability is defined as the possibility of becoming poor or poorer in the future (World Bank, 2001). There are three quantitative dimensions of vulnerability: Vulnerability as Expected Poverty (VEP), Vulnerability as Uninsured Exposure to Risk (VER), and Vulnerability as Low Expected Utility (VEU). Correspondingly, there are three different methods of measuring these vulnerability dimensions (Chaudhuri and Suryahadi, 2002; Gaiha and Ima, 2004; Gaiha and Katsushi, 2008), which respectively use risk sensitivity, gaps in welfare effects, and the probability of falling into poverty to express poverty vulnerability. A higher sensitivity to risk corresponds to a larger gap in welfare effects and a greater probability of falling into poverty corresponds to a higher level of vulnerability.

The risk of returning to poverty refers to the probability that the living standard of a family or individual who has been out of poverty will fall below the poverty line in the future because of the risk hitting, such as diseases, economic fluctuation, natural disasters and so on. Because the two definitions are similar, we considered the measurement of poverty vulnerability in existing studies as a measure of the risk of returning to poverty. We applied the VEP method to estimate the probability of a household’s loss of future welfare resulting from exposure to risk as a measure of poverty vulnerability. The VEP method has two main advantages. First, it can be adapted and used with cross-sectional data, which is necessary given the difficulty in obtaining multi-year data from micro-surveys conducted in rural areas. Second, the likelihood of falling into poverty measured using the VEP method can be objectively compared. The key objective of the VEP method is to predict the probability of falling into poverty during a certain future period using historical data on incomes or consumption and based on the normal distribution form of a given welfare level (Moore, 2001; Ward, 2016). The following equation was used to calculate the VEP:

$$~~{V_{h,t}}=Pr({Y_{h,t+1}} \leqslant G)$$

The above formula expresses the probability of household returning to poverty during period , indicating the probability that the income level${Y_{h,t+1}}$ of household during period $t+1$ is lower than the poverty line,.

Assuming that the level of household income follows a lognormal distribution, income levels ${Y_{h,t}}$are expressed as follows:

$\ln {Y_{h,t}}={\beta _1}\cdot power+{\beta _2}{X_h}+{e_k}$ (8)

where ${X_h}$ is a collection of family feature vectors. It is worth noting that existing research generally includes family demographic and endowment characteristics such as the age, gender, health, and education of the head of the household. In addition, we took into account the family’s endogenous power index of poverty alleviation in the model. The endogenous power of poverty alleviation is the fulfilment of basic needs, including income and rights, use one’s own knowledge or skills to actively link social resources, and convert resources into potential action trends that can sustain poverty alleviation and development (Fu et al., 2020; Liu et al., 2020). If a family’s internal motivation for poverty alleviation is insufficient, even with extensive external aid, once support is no longer received, the family will often fall into poverty again because of internal reasons such as lack of willingness to develop independently or initiative and enthusiasm for poverty alleviation. Drawing on existing research, we select indicators based on four aspects, namely the health level ($health$), education level ($edu$), subjective well-being ($wellbeing$), and attitude of trust towards strangers ($trust$) of the head of household. Using the equal-weight assumption and the method of adding the total sub-indices of the average Euclidean distance method to construct the endogenous driving force of household poverty alleviation ($power$), it is calculated as follows:

$$power={{1{\text{-}}\sqrt {{{(1 - edu)}^2}+{{(1 - health)}^2}+{{(1 - wellbeing)}^2}+{{(1 - trust)}^2}} } \mathord{\left/ {\vphantom {{1{\text{-}}\sqrt {{{(1 - edu)}^2}+{{(1 - health)}^2}+{{(1 - wellbeing)}^2}+{{(1 - trust)}^2}} } {\sqrt 4 }}} \right. \kern-0pt} {\sqrt 4 }}$$

In addition, we controlled other household features including the political status of the household head, household registration, age, gender, health, education, household size, annual household expenditure, and the values of the household’s durable goods and financial assets. $\beta$ is the parameter to be estimated, and ${e_k}$ is the random error.

We assumed that fluctuations in household income could be replaced by the regression residual squared, that is, by future income variance. According to the heteroscedasticity of the cross-sectional data, this variance was determined by the characteristics of the sampled households:

$\sigma _{{e,h}}^{2}=\lambda {X^{\prime}_h}$ (10)

We applied the three-stage feasible generalized least squares method to estimate the expectations of incomes and income variance among the sampled households:

$$E\left[ {\ln {Y_{h,t}}|{{X^{\prime}}_h}} \right]={X^{\prime}_h}\beta$$

$Var\left[ {\ln {Y_{h,t}}|{{X^{\prime}}_h}} \right]={X^{\prime}_h}\lambda$ (12)

Ultimately, the following formula was obtained:

${V_{h,t}}=\Pr ({Y_{h,t+1}} \leqslant G)=\phi [(\ln G - {X^{\prime}_h}\hat {\beta })/\sqrt {{{X^{\prime}}_h}\hat {\lambda }} ]$ (13)

where ${V_{h,t}}$ denotes the probability that household will return to poverty in period , $\Pr (\cdot )$ denotes the probability value, and $\varphi [\cdot ]$ is the positive distribution function. ${X^{\prime}_h}$ represents the set of all family feature vectors of the family, including endogenous dynamics. In this paper, the poverty vulnerability of the family is classified according to whether it is greater than the average poverty vulnerability of the sample in order to assess whether a poverty-stricken family is vulnerable. That is, a family is considered vulnerable if the probability of returning to poverty in the future is greater than or equal to the mean poverty vulnerability. Furthermore, we treated the probability as a zero-one dummy variable, which indicates whether poverty-stricken families are at risk of returning to poverty. We applied the World Bank’s poverty line standard of US$1.9/day as the poverty line, and converted it at the exchange rate of 6.7519 yuan per US dollar (using the 12-month average in 2017). Accordingly, we calculated 4618.29 yuan per annum as the global poverty line.

4.3.3 Control variables

Following the practice of existing literature and based on the availability of data, this paper focused on the risk of returning to poverty for the explained variable, and selected control variables from three levels: individual, family, and region. At the individual level, we selected whether the head of household was currently employed ($employ$), joined the Chinese Communist Party ($party$), and owned a smartphone ($smartphone$). At the family level, the the number of family members who are part of the labor force ($labor$) is selected to measure the family labor situation, whether the family accepts government subsidies ($subsidy$) to measure the family’s economic difficulty, and how much transfer expenditure has been paid to relatives and friends in the past year ($tranexpense$) to represent social networks. The household economic situation is measured by total household consumption ($lntotal\_comsup$). At the regional level, we selected the urbanization rate of the province where the household is located ($cityrate$) to measure the urbanization process in the region. The ratio of financial product to total product ($finance\_gdp$) is used to measure the level of financial development. The economic development level of the region is measured by the logarithm of GDP per capita ($\ln gdp\_per$).

The meanings and descriptive statistics of the main variables are shown in Table 2 below.

Table 2

Descriptive statistics of variables
Variable	Mean	Std. Dev.	Min	Max
Risk	0.161	0.368	0	1
IFI	0.174	0.057	0	0.667
employ	0.628	0.483	0	1
party	0.46	0.498	0	1
smartphone	0.596	0.491	0	1
labor	1.573	1.194	0	8
subsidy	0.228	0.419	0	1
tranexpense	0.749	0.434	0	1
lntotal consump	10.623	0.894	6.602	14.885
cityrate	0.612	0.101	0.433	0.916
finance gdp	0.078	0.029	0.048	0.177
lngdp_per	10.972	0.405	10.241	11.832

5.1 Regression analysis: The influence of digital financial inclusion on the risk of returning to poverty

To verify the influence of digital financial inclusion on the risk of returning to poverty, we first constructed the digital financial inclusion index of Chinese families from a micro perspective, and then tested model (1) by using the probit model. In column (1) of Table 1, without any control variables, the coefficient of $IFI$ is -6.08, which is significant at the significance level of 1%. The control variable at the individual, family and regional level are gradually added in column (2)-(4). In column (4), the coefficient of $IFI$is -3.918, and its standard deviation is 0.76, which is significant at the 1% level, indicating that there is a significant negative relationship between household digital financial inclusion and the risk of returning to poverty. This also shows that a one standard deviation increase in the digital financial inclusion index will significantly reduce the risk of households returning to poverty by 8.09%. Wang and Fu (2021) and Suri and Jack (2016) also otained the similar results.

The estimated results in Table 3 also reveal that there are other important factors affecting the risk of rural households returning to poverty. These factors include the working status and political identity of household heads, degree of family digitization, family social network, economic status, and degree of regional urbanization. According to the regression results of these control variables, having a party member as the head of a household can significantly reduce the risk of a family returning to poverty. In terms of family characteristics, as emphasized by existing studies (Bjorn et al., 2004), the higher the degree of family digitization, the more developed the social network, and the larger the number of family workers, the greater the reduction in the risk of family returning to poverty. At the regional level, the higher the urbanization rate of the family’s location, the greater the reduction of the risk of returning to poverty, which is also consistent with existing research (Yang and Fu, 2019; Zhu and Wang, 2017).

Table 3

Regression results
	(1)	(2)	(3)	(4)
	Risk	Risk	Risk	Risk
IFI	-6.08***	-4.419***	-3.685***	-3.918***
	(0.602)	(0.621)	(0.744)	(0.76)
employ		-0.553***	0.022	0.021
		(0.053)	(0.081)	(0.081)
party1		-0.138**	-0.142**	-0.119**
		(0.054)	(0.059)	(0.06)
smartphone		-0.545***	-0.207***	-0.205***
		(0.064)	(0.071)	(0.072)
labor			-0.355***	-0.365***
			(0.039)	(0.039)
subsidy			0.134**	0.122**
			(0.059)	(0.059)
tranexpense			-0.273***	-0.255***
			(0.058)	(0.059)
lntotal_consump			-0.692***	-0.693***
			(0.04)	(0.041)
cityrate				-1.518***
				(0.543)
finance_gdp				3.296
				(2.071)
lngdp_per				0.001
				(0.105)
_cons	0.308***	0.566***	7.401***	8.093***
	(0.1)	(0.106)	(0.41)	(1.086)
Observations	3156	3156	3156	3156
Pseudo R²	0.04	0.099	0.265	0.268
Note: Values in brackets are t-stat. The values in parentheses are standard deviations; , , ** indicate the level of significance of 10%, 5%, and 1%, respectively. Data are calculated by authors using Stata16.

5.2 Robustness analysis

5.2.1 Substitution of the poverty line standard

China’s poverty alleviation standard is a comprehensive standard. The national income standard is the annual income per capita of farmers at constant prices of 2,300 yuan in 2011. According to the price index, the poverty alleviation standard for poverty-stricken households in 2020 was an annual income of about 4,000 yuan. Therefore, we selected 4,000 yuan as the poverty line standard ($Risk4000$) for the robustness test. The first column of Table 4 shows that the coefficient of $IFI$ remains significantly negative, indicating that when a poverty standard of 4,000 yuan is applied, the digital financial inclusion significantly reduces the risk of a return to poverty, which is consistent with the previous regression results.

5.2.2 Substitution of the criteria for measuring poverty vulnerability

In the benchmark regression, we classified the vulnerability of poverty-stricken households according to whether it is greater than the mean of poverty vulnerability of the sample and process it into a 0–1 variable, which enables assessment of whether there is a risk of returning to poverty based on the sample itself. The World Bank has defined and measured relevant standards and thresholds for vulnerability to poverty, according to which 29% is the mild vulnerability threshold and 50% is the moderate vulnerability threshold. We judged whether a family that has been lifted out of poverty is vulnerable or not with the slight vulnerability value of 29% and the moderate vulnerability threshold value of 50%, respectively. That is, if the probability of falling into poverty in the future is greater than or equal to 29% or 50%, the family that has been lifted out of poverty is considered vulnerable. To further conduct robustness tests, we processed them into 0–1 variables, denoted as $Risk29$ and $Risk50$, respectively, to indicate whether the poverty-stricken families are at risk of returning to poverty. The results are shown in columns (2)–(3) of Table 4. The coefficients of $IFI$are significantly negative at the significance level of 1%, indicating that under the criteria of mild poverty vulnerability and moderate poverty vulnerability, household digital financial inclusion can significantly reduce the risk of families returning to poverty. Based on the above analysis, the negative correlation between household digital financial inclusion and the risk of returning to poverty has been verified again, which means that the conclusion of this paper is relatively robust.

5.2.3 Substitution of the method to construct the digital financial inclusion index

There is no unique way to synthesize the digital financial inclusion index. We further refer to the methods of Sarma (2008) and Park and Mercado (2018) to synthesize the digital financial inclusion index, with the formula below:

$IF{I_{}}^{\prime }={{1{\text{-}}\sqrt {\sum\nolimits_{{i=1}}^{n} {(1 - {d_i}} {)^2}} } \mathord{\left/ {\vphantom {{1{\text{-}}\sqrt {\sum\nolimits_{{i=1}}^{n} {(1 - {d_i}} {)^2}} } {\sqrt n }}} \right. \kern-0pt} {\sqrt n }}$ (14)

where, ${d_i}$ refers to the value of the standardized index of each dimension. is the total number of indicators. A total of seven indicators are selected (as shown in Table 1) from three levels of digitalization, usability, and coverage. Therefore, $n{\text{=}}7$ in this paper.

We substituted the newly synthesized digital financial inclusion index into model (1) for regression. From the regression results in column (4) of Table 4, the coefficient of $IF{I_{}}^{\prime }$ is -2.239, significantly at the significance level of 1%, suggesting that household digital financial inclusion can significantly reduce the risk of households returning to poverty.

5.2.4 Substitution of the regression method

We compared the robustness of our basic conclusions obtained using different estimation methods. Specifically, without processing the vulnerability to poverty ($Vul$) into 0–1 variables, we directly estimated by OLS method. As shown in column (5) of Table 4, the coefficient of $IFI$is -0.178, which is significant at the significance level of 1%, and the result reveals that the main conclusion is essentially consistent with the benchmark regression while using different estimation methods.

Table 4

Results of the robustness test
	(1)	(2)	(3)	(4)	(5)
	Risk4000	Risk29	Risk50	Risk	Vul
IFI	-3.388***	-5.14***	-7.783***		-0.178***
	(0.731)	(1.285)	(2.182)		(0.034)
IFI’				-2.239***
				(0.583)
Individual control variables	Yes	Yes	Yes	Yes	Yes
Family control variables	Yes	Yes	Yes	Yes	Yes
Region control variables	Yes	Yes	Yes	Yes	Yes
_cons	7.981***	6.881***	3.621	8.292***	0.675***
	(1.091)	(1.863)	(3.149)	(0.841)	(0.072)
Observations	3081	3156	3156	6422	3156
PseudoR² /(R²)	0.258	0.299	0.329	0.291	0.257
Note: Values in brackets are t-stat. The values in parentheses are standard deviations; , , ** indicate the level of significance of 10%, 5%, and 1%, respectively. Data are calculated by authors using Stata16.

5.3 Endogeneity

We draw on the method of Sarma (2016) to synthesize a digital financial inclusion index as the status of household digital financial inclusion; however, this may suffer from endogeneity problems. First, although we attempted to include all variables that may affect the risk of a household returning to poverty, there could still be some variables that were missing in the equation but nevertheless impacted on the risk of returning to poverty and were associated with the digital financial inclusion, such as government-implemented subsidy policies for emerging industries or informal financial development at the macro level. Second, the interviewees may have concealed some information to maintain their personal privacy, which could have led to errors in the measurement of variables. Third, the development of digital financial inclusion may have a reverse causal relationship with the risk of returning to poverty. That is, the higher the risk of returning to poverty for household, the more difficult the poverty situation, and the greater the possibility of being excluded from traditional finance or digital inclusive finance, thereby reducing the possibility of using financial services and falling into a vicious circle of poverty. Conversely, the better the family’s economic conditions, the more likely the family is to participate in the emerging digital financial inclusion business, which in turn drives the development of regional digital financial inclusion.

Therefore, we tested for endogeneity problems by including more control variables and using instrumental variables. First, we considered the issue of family self-selection, that is, whether the family uses digital financial inclusion as a conscious choice based on its own resource endowment. If family members pay attention to financial news and obtain more information about digital inclusive finance consultation and services, they will have more opportunities to make financial choices, which may change the family’s economic situation and affect the risk of family falling back into poverty. Therefore, we added the control variable of financial literacy “whether we pay attention to financial news or financial information” ($information$) to perform regression of model (1). In column (1) of Table 5, the coefficient of $IFI$ remains significantly negative and consistent with the conclusion of the benchmark regression result. The coefficient of $information$is significantly negative at the significance level of 10%, indicating that attention of householders to finance can reduce the family’s risk of returning to poverty. The reason may be that the higher their level of financial knowledge, the more confident decision-makers are in their own judgment, and they can easily avoid most risks so that they are more confident in asset allocation. Furthermore, good financial literacy helps residents obtain relevant information from newspapers, Internet and other media, actively pay attention to and understand relevant policies, and thus participate in the financial market (Jappelli and Padula, 2013). Therefore, existing inclusive financial resources can be used scientifically and reasonably to improve family economic conditions (Van Rooij et al., 2011).

Table 5

The results of endogeneity analysis
	(1)	(2)	(3)
	Risk	IFI	Risk
IFI	-3.875***
	(0.76)
information	-0.26*
	(0.14)
Distance*IFI		0 .0004***	-2.395***
		(52.82 )	(0.883)
Individual control variables	Yes	Yes	Yes
Family control variables	Yes	Yes	Yes
Region control variables	Yes	Yes	Yes
_cons	8.1392***	-0.711***	8.048***
	(1.092)	(0.029)	(1.08)
Observations	3156	3156	3156
Pseudo R²	0.269	0.5342	0.5265
Note: Values in brackets are t-stat. The values in parentheses are standard deviations; , , ** indicate the level of significance of 10%, 5%, and 1%, respectively. Data are calculated by authors using Stata16.

In addition, we used the instrumental variable method to control the potential endogeneity problems and re-estimate the relationship between digital financial inclusion and the risk of returning to poverty. IV-Probit model is an effective method to test the endogeneity of the probit model. Referring to the practice of He et al. (2020), we adopted the product of the spherical distance from the province where the city is located to Hangzhou and the digital financial inclusion index ($Distance*IFI$) as an instrumental variable. The main reasons are as follows: although digital financial inclusion is mainly realized through the Internet, its rapid development is still limited by many factors such as geographical considerations and time. The development of digital financial inclusion in neighboring areas is increasingly similar, and Hangzhou City of Zhejiang Province is the financial service center of diffusion. Therefore, “the distance between the province where the family address is located and Zhejiang Province” is related to the development level of digital inclusive finance, and the geographical location factor is a pure exogenous variable that is not affected by any subjective factors and is not directly related to the economic situation of the family or to other families and individual characteristics. Therefore, the instrumental variable satisfies the characteristics of exogeneity and correlation.

Column (2) in Table 5 presents the regression result of the first stage, with a coefficient of 0.0004 and significant at the significance level of 1%, indicating that the farther away from the digital financial development center, the lower the development level of digital inclusive finance. Column (3) in Table 5 lists the second-stage regression results estimated by instrumental variables. First, the value of the F statistic of the weak instrumental variable test in the first stage is 325.58, which is far greater than the 10% critical value level, indicating that the instrumental variable selected in this paper is significantly effective. Second, after the instrumental variable test, the coefficient of $IFI$is -2.395, significantly at the significance level of 5%. Additionally, the coefficient of IV-Probit regression is higher than the benchmark regression coefficient, indicating that the benchmark probit regression has downward deviation, which shows that the measurement error is the main reason for the coefficient difference. On the whole, instrumental variable estimation shows that digital financial inclusion can still significantly reduce household poverty and vulnerability, and the impact coefficient value increases, which indicates that if endogenous issues are ignored, the impact of digital financial inclusion on the occurrence of poverty will be underestimated.

5.4 Heterogeneity analysis

The influence of digital financial inclusion on the risk of a family’s return to poverty has already been demonstrated. However, the question of whether there is any difference in the influence of relationships on families or regions with different characteristics arises. Therefore, we divided the overall sample using different standards, and explored the heterogeneity of digital financial inclusion in relation to the risk of a family returning to poverty from three perspectives: human capital, vulnerable groups, and the developmental status of digital economy in each region.

5.4.1 “Elite capture” or “universal benefit”: Heterogeneity analysis of human capital

Education level is an important human capital and a crucial factor affecting individual’s investment decision and asset allocation. Therefore, it is necessary to further explore whether there is heterogeneity in the effect of digital inclusive finance on the risk of returning to poverty from the perspective of individual human capital. Drawing on the usual practice of existing literature, we first chose the number of years of education of the household head to measure the household human capital. According to the new growth theory of human capital, the improvement of education level means that the accumulation of human capital is increased, and individuals’ ability to accept knowledge or technological progress and material capital is enhanced accordingly, thus bringing about the improvement of productivity and income level. Specifically, based on the nine-year compulsory education in China as the cut-off point, we divided the years of education of household heads into two groups, namely below high school and above high school, to perform regression of model (1). The results in columns (1)–(2) of Table 6 shown that financial inclusion has a role in reducing the risk of returning to poverty for families with different human capital endowments. From this perspective, the development of digital financial inclusion does not have elite capture in poverty management. However, in absolute terms, digital financial inclusion has different effects on the higher-education group and the lower-education group. The more educated the head of the household, the more effective digital financial inclusion is in curbing the return to poverty. The reason may be that highly educated people tend to have higher financial literacy (Luigi and Jappelli, 2008); furthermore, they are likely to embrace emerging financial products such as digital finance and Internet finance with a more positive attitude (Nasri and Charfeddine, 2012; Polatoglu and Ekin, 2001). This also indicates that the Chinese government should pay special attention to the improvement of family human capital in the formulation of poverty control policies.

5.4.2 “Timely assistance” or “icing on the cake”: Heterogeneity analysis of vulnerable groups

The original intention of developing digital financial inclusion was to provide financial services to all sectors of society, especially the low-income and disadvantaged groups, in order to achieve the inclusive growth of finance. If the poor and vulnerable groups are truly benefited, digital inclusive finance will play a “timely assistant” role in helping these disadvantaged groups. If digital financial inclusion only plays a role in non-vulnerable groups, the “pro-poor effect” of digital financial inclusion is not sufficient, and only serves as the “icing on the cake”. Therefore, the question arises of whether the development of digital financial inclusion has achieved the goal of being “pro-poor”.

Accordingly, we classified vulnerable groups based on “Is the household selected as a China’s Targeted Poverty Alleviation (TPA) beneficiary? 1 = yes, 0 = no”, and performed regression on model (1). It can be seen from the regression results in columns (3)–(4) in Table 6 that the coefficient of $IFI$ is significantly negative, indicating that while digital financial inclusion is conducive to reducing the risk of families returning to poverty regardless of whether they are poor or not, it has a greater effect on reducing poverty vulnerability of non-poor households than that of poor households from the perspective of absolute value. That is, digital financial inclusion plays more of a “icing on the cake” role in “benefiting the poor”. The reasons may be that poor families are mostly distributed at the edge of cities and towns or in vast rural areas and in China’s poor areas and the “last mile” of network infrastructure is still not fully developed. The “digital divide” results in digital inclusive finance having different effects on the risk of returning to poverty in different poor areas and non-poor areas (Chen et al., 2010). At the micro level, poor families have lower advantages in terms of digitalization, human capital, and other aspects than non-poor families, and have lower awareness, explanation, or participation in inclusive financial services (Jeanneney and Kpodar, 2011). Therefore, “knowledge gap” also hinders the impact of digital financial inclusion on the risk of family poverty return.

5.4.3 “Digital divide” or “digital dividend”: Heterogeneity analysis of the financial environment

The uneven distribution of financial resources in China is relatively prominent. Financial institutions are not only more concentrated in central towns, county suburbs, and other areas with relatively developed economic transportation and relatively dense populations, but indirectly lead to relatively serious traditional financial exclusion of relatively poor families (Jeanneney and Kpodar, 2011). In theory, digital finance can overcome the dependence of traditional inclusive finance on physical outlets and maximize the supply of digital financial services by relying on information technologies such as the Internet and data communications. Therefore, does the degree of financial environment development have a heterogeneous impact on the risk of households returning to poverty?

To be specific, we made three different classifications. First, “the distance between the nearest bank and the house” is taken as a substitute variable of the financial environment. If the distance between the family community and the nearest bank is greater than the mean, it is defined as “far from the bank and the financial environment is poor”, and the value is assigned as 1, and 0 otherwise. Group regression was performed based on the level of financial development in the region where the family is located. Column (5)–(6) of Table 6 show that the coefficients of $IFI$ is significantly negative under the significance level of 1% in both subsamples with differing financial development. This shows that although the development of digital inclusive finance has an overall effect on reducing the risk of households returning to poverty, regions with a well-developed financial environment have obvious location advantages and a greater containment effect on returning to poverty. In addition, we classified China’s economic location development based on differences in geographical location from a macro perspective. The eastern region is the birthplace of digital inclusive finance and represents the advanced region of digital technology innovation. However, the situation is similar in many provinces in the central and western regions, where innovation in the field of digital technology lags behind; therefore, the central and western regions are included. Specifically, the provinces where the samples are located are classified as eastern regions with better financial environment development, including Beijing, Tianjin, Hebei, Liaoning, Shanghai, Zhejiang, Jiangsu, Fujian, Shandong, Guangdong, and Hainan. And financial environment development in general central and western provinces (Inner Mongolia, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang, Shanxi, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hunan, Hubei). Columns (7)–(8) of Table 6 still yield consistent results. While it is worth noting that in columns (9)–(10) of Table 6, when measuring the financial development environment according to the size of the synthetic digital financial inclusion index, the overall sample is divided into the top 25% of the financial environment with good development and the rest of the financial environment with general development. In the samples with well-developed financial environment, the development of digital financial inclusion has a significant effect on mitigating the risk of returning to poverty, while in the remaining samples, the coefficient of digital financial inclusion is not significant. These results show that the effect of digital financial inclusion on the risk of returning to poverty is closely related to the development environment of finance and digital economy. In conclusion, there are obvious regional differences in the impact of digital inclusive finance development on the risk of returning to poverty among families in China. The development of digital financial inclusion has a stronger moderating effect on the risk of families’ returning to poverty in regions with developed digital economies; while that is relatively weak in the backward regions with lower development. It is also evident that an environment with a favorable financial environment is better in terms of financial infrastructure, service level, and financial products than a location with a less developed financial environment. Therefore, the location with different development of digital financial inclusion has different effects on the risk of families returning to poverty (Chen and Zhao, 2021; Yu and Wang, 2021).

Table 6

Heterogeneity analysis
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)
	High school or above	High school or below	Poverty-stricken family	Nonpoverty-stricken family	Less than average distance	More than average distance	Eastern Region	Central and western regions	The top 25%	The remaining 75%
	Risk	Risk	Risk	Risk	Risk	Risk	Risk	Risk	Risk	Risk
IFI	-4.812*	-3.881***	-3.302***	-4.108***	-4.755***	-2.677**	-4.917***	-3.536***	-8.664***	-0.597
	(2.704)	(0.962)	(1.154)	(1.002)	(0.898)	(1.316)	(1.385)	(0.898)	(2.252)	(0.878)
Individual control variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Family control variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Region control variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
_cons	1.804	8.861***	9.131***	7.308***	7.572***	8.881***	4.234**	9.743***	10.233***	7.243***
	(4.965)	(1.333)	(2.107)	(1.29)	(1.357)	(1.803)	(2.014)	(1.915)	(1.872)	(1.33)
Observations	250	1966	817	2339	2114	1042	890	2266	734	2422
Pseudo R²	0.222	0.26	0.24	0.269	0.267	0.278	0.275	0.274	0.267	0.245
Note: Values in brackets are t-stat. The values in parentheses are standard deviations; , , ** indicate the level of significance of 10%, 5%, and 1%, respectively. Data are calculated by authors using Stata16.

5.5 The functional mechanisms of digital financial inclusion

We theoretically analyzed the mechanism of digital financial inclusion in reducing the risk of returning to poverty through promoting entrepreneurship and employment to increase household income and optimizing household asset allocation to improve risk resistance level. To further explore the mechanism of digital financial inclusion, we constructed models for testing.

Residents’ entrepreneurship is measured according to “whether they are engaged in industrial and commercial operations”, represented by $business{\text{=}}1$. That is, family entrepreneurship is regarded as 1, otherwise it equals 0, which is substituted into model (1) as an explained variable for regression. As can be seen from the results in column (1) of Table 7, digital financial inclusion has a significant influence on promoting family entrepreneurship. Capital is crucial for entrepreneurs, and liquidity constraints tend to exclude the underfunded. Digital financial inclusion improves traditional finance and inclusive finance, and uses digital technology to popularize financial services more widely, so that as many individuals as possible can enjoy financial convenience (He and Li, 2019; Xie et al., 2020). Therefore, digital financial inclusion accelerates the flow of capital and promotes entrepreneurship. Meanwhile, the development of enterprises will undoubtedly provide extensive employment opportunities; therefore, we further explored whether digital financial inclusion can stem the slide back into poverty by providing off-farm employment and employed total family income ($total\_income$) and the logarithm of salary income ($wage$) as explained variables to conduct model (1). In the columns (2)–(3) of Table 7, total family income ($total\_income$) and salary income ($wage$) coefficient is positive under the 5% and 1% significance levels, which reflects that digital financial inclusion can increase employment opportunities and reduce the risk of returning to poverty for the families.

Second, another mechanism of digital inclusive finance for reducing the risk of returning to poverty is to optimize household asset allocation and improve families’ future risk resistance level. Specifically, the proportion of the value of household financial assets in total assets ($asset$) is used to represent the household asset allocation level for model 1 regression test. The results are shown in column (4) of Table 7. The coefficient of financial asset proportion ($asset$) is 0.263, significantly at the significance level of 1%. This suggests that digital financial inclusion has a significant positive impact on optimizing household asset allocation. Digital financial inclusion reduces the risk of returning to poverty by optimizing household asset allocation and improving household risk resistance, which is consistent with the conclusions of existing research (Fernandes et al., 2014; Liao et al., 2017).

Table 7

Results of Mechanism Analysis
	(1)	(2)	(3)	(4)
	business	total_income	wage	asset
IFI	1.834***	1.013**	1.063***	0.263***
	(0.56)	(0.422)	(0.403)	(0.043)
Individual control variables	Yes	Yes	Yes	Yes
Family control variables	Yes	Yes	Yes	Yes
Region control variables	Yes	Yes	Yes	Yes
_cons	-7.025***	2.972***	3.78***	0.588***
	(1.319)	(0.708)	(0.871)	(0.087)
Observations	3936	3838	1610	3747
R-squared	0.115	0.26	0.156	0.024
Note: Values in brackets are t-stat. The values in parentheses are standard deviations; , , ** indicate the level of significance of 10%, 5%, and 1%, respectively. Data are calculated by authors using Stata16.

5.6 Further analysis

5.6.1 Structural analysis of digital inclusive finance on the risk of returning to poverty

Analysis of the effect of the overall development of digital financial inclusion on the risk of returning to poverty may not be comprehensive. We used different sub-dimensions of digital financial inclusion to study its impact on family poverty. The reason is that the development of digital financial inclusion is multi-faceted and multi-dimensional, which may be reflected in the increase in the number of electronic accounts based on the Internet, the in-depth use of Internet financial services, and effective reduction of the cost and threshold of digital financial services. Therefore, we deemed it necessary to explore the structural impact of each sub-dimension of digital financial inclusion on the risk of household returning to poverty. Specifically, we substituted the sub-indicators of its secondary dimensions into model (1) for corresponding regression to explore the effects and differences of different dimensions of digital financial inclusion. Results are shown in column (1)–(3); usability ($IF{I_u}$) significantly reduced the risk of returning to poverty. The effect of digitalization ($IF{I_d}$) and coverage ($IF{I_c}$) on reducing the risk of falling back into poverty were not significant. The reasons may be that financial infrastructure facilitates the preconditions for the sustainable development of digital financial inclusion. The usability index is measured by the available digital financial services. Digital finance is integrated into all aspects of family daily life, such as education, medical care, insurance, and consumption, to achieve an increase in the types of digital financial services, thereby fullfilling household financial service needs. However,similar to those in developing countries, there are serious regional financial exclusion problems in China, such as obvious differences in digital financial infrastructure, unequal financial knowledge between urban and rural residents, and low level of digital development, which seriously restrict the effectiveness of digital financial inclusion, especially, digitalization and coverage, in curbing poverty.(Ficawoyi and Kevin, 2016).

Table 8

The results of further analysis
	(1)	(2)	(3)	(4)	(5)
	Risk	Risk	Risk	Risk	Risk
IFI_d	-0.286
	(0.237)
IFI_u		-1.507***
		(0.256)
IFIc			0.353
			(3.006)
IFI				-9.178***	-4.126***
				(1.479)	(0.777)
IFI²				14.360***
				(3.202)
privatelending*IFI					1.207***
					(0.419)
Individual control variables	Yes	Yes	Yes	Yes	Yes
Family control variables	Yes	Yes	Yes	Yes	Yes
Region control variables	Yes	Yes	Yes	Yes	Yes
_cons	7.944***	8.365***	14.836***	8.972***	8.164***
	(0.84)	(1.089)	(1.542)	(1.085)	(1.08)
Observations	6422	3156	6422	3156	3156
Pseudo R²	0.289	0.269	0.289	0.27	0.269
Note: Values in brackets are t-stat. The values in parentheses are standard deviations; , , ** indicate the level of significance of 10%, 5%, and 1%, respectively. Data are calculated by authors using Stata16.

5.6.2 Nonlinear effects of digital financial inclusion and risk of returning to poverty

Our study found that digital inclusive finance has a significant moderating effect on the risk of returning to poverty. To further investigate whether the impact of digital inclusive finance on the risk of returning to poverty has nonlinear characteristics, we established the following model for empirical testing:

$Risk={\beta _0}+{\beta _1}IFI+{\beta _2}IF{I^2}+\sum {{\beta _i}} {X_i}+{\mu _{_{i}}}$ (15)

As can be seen in column (4) of Table 8, the coefficient of $IFI$ is -9.092 and the second item is 14.142, both of which have passed the 1% significance level test. This indicates that the development of digital financial inclusion has a non-linear effect on the risk of returning to poverty of Chinese families, specifically, an U-shaped curve, with an inflection point of 0.77. That is, when the digital financial inclusion index is lower than 0.321, digital financial inclusion can significantly reduce the risk of returning to poverty; but when the digital financial inclusion index is higher than 0.321, the development of digital financial inclusion is not conducive to reducing the risk of returning to poverty. The range of financial inclusion index measured in this paper is [0,0.667], and its three-quartile is 0.177. Therefore, the basic conclusion of this paper is still valid. The reasons behind this may be as follows: First, data form the basic element of digital finance. As a new type of production factor, data have the advantages of being replicable, easy to share, and immune to physical wear. With the improvement of digital financial inclusion, the scope of the digital market and users has been gradually expanded, and the accessibility and liquidity of data elements have increased, thus forming economies of scale and scope. The impact of the development of digital inclusive finance on reducing the risk of returning to poverty has been highlighted (Zhang and Yang, 2019). Second, digital technology has not changed the nature of finance. Because of the high risk of finance itself (the larger and more complex the financial system, the worse its stability), digital inclusive finance, technological risk, and network risk are superimposed on each other to create new risks. In addition, people’s risk awareness, digital supervision, online lending platforms, and other digital financial models do not match, which has a significant spillover effect on exacerbating the risks of the banking system and increasing financial uncertainty, and has a negative impact on poverty governance. Furthermore, digital financial inclusion will shift its development path and its positive influence on poverty governance when it reaches a certain scale. Based on the above analysis, although the development of digital inclusive finance can mitigate the risk of returning to poverty, the increase of its scale and level to a certain threshold may also inhibit its effect on poverty control. These complex factors determine that the relationship between digital financial inclusion development and poverty governance may not be linear.

5.6.3 The relationship between digital financial inclusion and private lending: “complementarity” or “substitution”?

In the development process of traditional finance, digital inclusive finance has compensated for the deficiency of traditional finance due to development problems such as unbalanced development of basic financial facilities, unsound development of rural financial credit market, and limited policy-based financial service capacity. However, in the process of the development of informal finance, on the basis of social capital ties, folk lending has the advantages of flexible operation, easy information accessibility, and low transaction cost, which formal financial transactions do not have. Therefore, digital inclusive finance and private lending play an uncertain role in alleviating the risk of returning to poverty. Specifically, families with high social capital may prefer private loans to avoid risk: because they have relatively sufficient resources, the cost of obtaining private loans is lower. Thus, private lending can replace the role of digital financial inclusion in reducing the risk of returning to poverty. For families lacking in social resources, the costs of digital financial inclusion channel may be even lower compared with private lending; therefore, these families tend to choose the formal digital inclusive finance service. Based on the above analysis, we consider whether there is a substitution relationship between digital inclusive finance of formal finance and private lending of informal finance in alleviating the risk of family returning to poverty. To this end, we introduce the cross-product term of “have you ever participated in private lending” ($IFI\cdot privatelending$) and digital financial inclusion index ($IFI$) to construct the following model:

$Risk={\beta _0}+{\beta _1}IFI+{\beta _2}IFI\cdot privatelending~+\sum {{\beta _i}} {X_i}+{\mu _{_{i}}}$ (16)

In column (5) of Table 8, the coefficient of $IFI\cdot privatelending$ is 1.206, and the coefficient of$IFI$is − 4.126, both of which are significant at the significance level of 1%. This shows that although private lending can meet the demand for household financial services to a certain extent, it squeezes out some families’ demand for inclusive financial services through formal financial channels, that is, private lending and digital financial inclusion can substitute for reducing the risk of returning to poverty. Accordingly, it also reflects that the development of digital financial inclusion can reduce residents’ dependence on their social network, which provides conditions for improving the inequality of opportunity.

In this paper, we adopted data from China Household Finance Survey (CHFS) to construct a micro-level household digital financial inclusion index, and empirically tested the influence of digital financial inclusion on the risk of Chinese households returning to poverty. Further, we explored its mechanism of action and performed further analysis.

Through empirical analysis, four conclusions are drawn: First, digital financial inclusion has a significant effect on reducing the risk of returning to poverty. Second, heterogeneity analysis suggested that digital financial inclusion has different degrees of restraint on the risk of returning to poverty for families with different household endowments and the development characteristics of the digital economy region in which they are located. That is, digital financial inclusion has a more obvious effect on curbing the risk of returning to poverty for households with high human capital, non-poor households, and households located in places with a high level of digital economy development. Third, two mechanisms for digital financial inclusion to reduce the risk of returning to poverty are found, namely, increase household income by promoting entrepreneurship and non-agricultural employment and improve risk resistance by improving household financial literacy and market participation, and optimizing household asset allocation. Fourth, the further analysis found that the usability of digital financial inclusion has an obvious effect on alleviating the risk of returning to poverty at this stage, although the effect of digitalization and coverage on alleviating the risk of returning to poverty has not yet been prominent. Moreover, digital financial inclusion has a significant nonlinear effect on the risk of returning to poverty, and there is a substitution effect with private lending on the risk of returning to poverty.

Based on the above conclusions, we put forward the following recommendations: First, more technical and innovation support should be given to digital financial inclusion; the potential of digital service should be explored; and the information integration, opening, and sharing of grassroots financial service demand groups should be accelerated. Second, in view of the heterogeneous effect and mechanism of digital inclusive finance on the risk of returning to poverty, the government should increase support for areas where the development of the digital economy is relatively lagging, and increase the coverage of the infrastructure for the development of digital finance. In addition, the role of science and technology should be brought into play to empower the popularization of financial knowledge and improve the literacy of residents in applying digital inclusive finance. The government should guide financial supervision departments, and financial institutions and other social entities should actively participate in the teaching of financial knowledge and skills, rational consumption, and other financial knowledge.

Funding information

This paper is supported by National Natural Science Foundation of China (Grant No. 72003049). National Social Science Fund of China (Grant No.18BTJ011). Humanities and Social Science Fund of Ministry of Education of China (Grant No. 20YJC790191). National Statistical Science Research Project of National Bureau of statistics of China: (Grant No. 2020LY101).

Disclosure statement

No potential conflict of intersect was reported by the authors.

Allen, F., Carletti, E., Cull, R., Qian, J., Senbet, L., & Valenzuela, P. (2016). 1. Resolving the African Financial Development Gap: Cross-Country Comparisons and a Within-Country Study of Kenya (pp. 13–62). University of Chicago Press.
Amidžic, G., Massara, M. A., & Mialou, A. (2014). Assessing countries’ financial inclusion standing-A new composite index. International Monetary Fund.
Ayyagari, M., Demirguc-Kunt, A., & Maksimovic, V. (2014). Who creates jobs in developing countries?. Small Business Economics, 43(1), 75–99.
Beck, T., Demirguc-Kunt, A., & Levine, R.. (2004). Finance, inequality, and poverty: cross-country evidence. NBER Working Papers.
Bernheim, B.D. & Garrett, D.M. (2003) The effects of financial education in the workplace: evidence from a survey of households. Journal of Public Economics, 87(7–8), 1487–1519.
Bianchi, M. (2010). Credit constraints, entrepreneurial talent, and economic development. Small business economics, 34(1), 93–104.
Björn Gustafsson and Li Shi and Hiroshi Sato. (2004). Can a subjective poverty line be applied to China? Assessing poverty among urban residents in 1999. Journal of International Development, 16(8), 1089–1107.
Bruhn, M., & Love, I. (2014). The real impact of improved access to finance: Evidence from Mexico. The Journal of Finance, 69(3), 1347–1376.
Bruton, G. D., Ahlstrom, D., & Li, H. L. (2010). Institutional theory and entrepreneurship: where are we now and where do we need to move in the future?. Entrepreneurship theory and practice, 34(3), 421–440.
Chaudhuri, S., Jalan, J., & Suryahadi, A. (2002). Assessing Household Vulnerability to Poverty: A Methodology and Estimates for Indonesia, Columbia University Department of Economics Discussion Paper No. 0102 – 52. New York: Columbia University.
Chen, D. Y.; Lin, Z. X.; Lai, F. J. (2010). Crossing the Chasm - Understanding China's Rural Digital Divide. Journal of Global Information Technology Management, 13(2), 4–36.
Chen, W., & Yuan, X. (2021). Financial inclusion in China: an overview. Frontiers of Business Research in China, 15(1), 1–21.
Dutta, N., & Meierrieks, D. (2021). Financial development and entrepreneurship. International Review of Economics & Finance, 73, 114–126.
European Commission (2008). Financial services provision and prevention of financial exclusion. (p. 48). Report by the Director General for Employment, Social Affairs and Equal Opportunities, European Commission.
Fernandes, D., Lynch Jr, J. G., & Netemeyer, R. G. (2014). Financial literacy, financial education, and downstream financial behaviors. Management Science, 60(8), 1861–1883
Ficawoyi Donou-Adonsou and Kevin Sylwester. (2016). Financial development and poverty reduction in developing countries: New evidence from banks and microfinance institutions. Review of Development Finance, 6(1), pp. 82–90.
Fu, A., Zhang, Z., Zheng, J., Yue, T., Lin, Z., Wu, N., & Huang, X. (2020). Qualitative research on the endogenous power mechanism for poverty elimination. Acta Psychologica Sinica, 52(1), 66–80.
Gaiha, R., & Imai*, K. (2004). Vulnerability, shocks and persistence of poverty: estimates for semi-arid rural South India. Oxford Development Studies, 32(2), 261–281.
Gaiha, R., & Imai, K. (2008). Measuring vulnerability and poverty estimates for rural India (No. 2008/40). WIDER Research Paper.
Giuseppina Guagnano and Elisabetta Santarelli and Isabella Santini. (2016). Can Social Capital Affect Subjective Poverty in Europe? An Empirical Analysis Based on a Generalized Ordered Logit Model. Social Indicators Research, 128(2), 881–907.
Gomber, P., Koch, J. A., & Siering, M. (2017). Digital Finance and FinTech: current research and future research directions. Journal of Business Economics, 87(5), 537–580.
Hao A. Research on Risk Points and Countermeasures of Poverty Population Returning to Poverty in Deep Poverty Areas[J]. 2019. International Conference on Emerging Researches in Management, Business, Finance and Economics (ERMBFE 2019)
He Z.Y., Zhang X & Wan G.H.(2020). Digital finance, digital divide and multidimensional poverty. Statistical Research(in Chinense)(10),79–89.
He, J., & Li, Q. (2019). Digital finance and farmers’ entrepreneurship. China Rural Econ, 1, 112–126.
Jakšič, M., & Marinč, M. (2019). Relationship banking and information technology: The role of artificial intelligence and FinTech. Risk Management, 21(1), 1–18.
Jappelli, T. & Padula, M. (2013) Investment in financial literacy. CFS Working Paper: Social Security and Portfolio Choice.
Jappelli, T., & Padula, M. (2013). Investment in financial literacy and saving decisions. Journal of Banking & Finance, 37(8), 2779–2792.
Leyshon, A., & Thrift, N. (1995). Geographies of financial exclusion: financial abandonment in Britain and the United States. Transactions of the Institute of British Geographers, 312–341.
Liao, L., Xiao, J. J., Zhang, W., & Zhou, C. (2017). Financial literacy and risky asset holdings: evidence from China. Accounting & Finance, 57(5), 1383–1415.
Liu, J., Huang, F., Wang, Z., Shuai, C., & Li, J. (2020). Understanding the role of rural poor’s endogenous impetus in poverty reduction: Evidence from China. Sustainability, 12(6), 2487.
Lu, X., Xiao, J., & Wu, Y. (2021). Financial literacy and household asset allocation: Evidence from micro-data in China. Journal of Consumer Affairs, 55(4), 1464–1488
Luigi, G. & Jappelli, T. (2008) Financial literacy and portfolio diversification. European University Institute Economics Working papers No. ECO 2008/31.
Moore, K. (2001). Frameworks for Understanding the inter-generational transmission of poverty and well-being in developing countries. Chronic Poverty Research Centre Working Paper, (8).
Mugo, M., & Kilonzo, E. (2017). Community–Level Impacts of Financial Inclusion in Kenya with Particular Focus on Poverty Eradication and Employment Creation. Central Bank of Kenya,13.
Nasri, W., & Charfeddine, L. (2012). Factors affecting the adoption of Internet banking in Tunisia: An integration theory of acceptance model and theory of planned behavior. The journal of high technology management research, 23(1), 1–14.
Neaime, S., & Gaysset, I. (2018). Financial inclusion and stability in MENA: Evidence from poverty and inequality. Finance Research Letters, 24, 230–237.
Park, C. Y., & MERCADO JR, R. O. G. E. L. I. O. (2018). Financial inclusion, poverty, and income inequality. The Singapore Economic Review, 63(01), 185–206.
Patrick S. Ward. (2016). Transient Poverty, Poverty Dynamics, and Vulnerability to Poverty: An Empirical Analysis Using a Balanced Panel from Rural China. World Development, 541–553.
Polatoglu, V. N., & Ekin, S.(2001). An empirical investigation of the Turkish consumers’ acceptance of Internet banking services. International Journal of Bank Marketing, 19(4), 156–165.
Sarma, M. (2008). Index of financial inclusion (No. 215). Working paper.
Sarma, M. (2016). Measuring financial inclusion using multidimensional data. World Economics, 17(1), 15–40.
Seven, U., & Coskun, Y. (2016). Does financial development reduce income inequality and poverty? Evidence from emerging countries. Emerging Markets Review, 26, 34–63.
Singh, C. (2012). Financial inclusion through information technology with reference to banking sector. Asian Journal of Research in Banking and Finance, 2(10), 12–22.
Suri, T., & Jack, W. (2016). The long-run poverty and gender impacts of mobile money. Science, 354(6317), 1288–1292.
Wang, A. (2009) Interplay of investors' financial knowledge and risk taking. The Journal of Behavioral Finance, 10(4), 204–213.
Wang, X., & Fu, Y. (2021). Digital financial inclusion and vulnerability to poverty: evidence from Chinese rural households. China Agricultural Economic Review, 12(4), 1668.
World Bank Group. (2013). Global financial development report 2014: Financial inclusion (Vol. 2). World Bank Publications.
Xie, W., Wang, T., & Zhao, X. (2020). Does Digital Inclusive Finance Promote Coastal Rural Entrepreneurship? Journal of Coastal Research, 103(SI), 240–245.
Yin Z.C. & Zhang D.H.(2020). Financial inclusion, household poverty and vulnerability. Economics quarterly(in Chinense) (05),153–172.
Yin Z.C., Peng C.Y. & Leon Angela.(2019). The Development and Impact of Household inclusive finance in China. Management World(in Chinense) (02),74–87.
Yin, Z., Gong, X., Guo, P., & Wu, T. (2019). What drives entrepreneurship in digital economy? Evidence from China. Economic Modelling, 82, 66–73.
Yu X. J.. (2019). Analysis of the Influencing Factors of Returning to Poverty in Deep Poverty Areas. 2019International Conference on Modern Education and Economic Management (ICMEEM 2019),407–411.
Yu, N., & Wang, Y. (2021). Can Digital Inclusive Finance Narrow the Chinese Urban–Rural Income Gap? The Perspective of the Regional Urban–Rural Income Structure. Sustainability, 13(11), 6427.
Zhang, M., & Yang, J. (2019). Research on financial technology and inclusive finance development. In 2018 6th International Education, Economics, Social Science, Arts, Sports and Management Engineering Conference (IEESASM 2018) (pp. 66–71). Atlantis Press.
Zhu Y. M., Wang W.(2017). How Does Inclusive Finance Achieve Precise Poverty Alleviation? Journal of Finance and Economics, 43(10), 43–54.

No competing interests reported.

Download PDF

Journal Publication

published 04 Apr, 2023

Read the published version in International Journal of Finance & Economics →

Version 1

posted

You are reading this latest preprint version

Does digital financial inclusion reduce the risk of returning to poverty? Evidence from China

Status:

Journal Publication

Version 1

Abstract

1. Introduction

2. Related Literature

2.1 Definition and measurement of financial inclusion

2.2 The impact of financial inclusion on poverty governance

3. Theoretical Analysis And Research Hypothesis

4. Research Design

4.1 Model construction

4.2 Data

4.3 Variables

4.3.1 Household Digital Financial Inclusion Index

4.3.2 The risk of returning to poverty

4.3.3 Control variables

5. Empirical Analysis

5.1 Regression analysis: The influence of digital financial inclusion on the risk of returning to poverty

5.2 Robustness analysis

5.2.1 Substitution of the poverty line standard

5.2.2 Substitution of the criteria for measuring poverty vulnerability

5.2.3 Substitution of the method to construct the digital financial inclusion index

5.2.4 Substitution of the regression method

5.3 Endogeneity

5.4 Heterogeneity analysis

5.4.1 “Elite capture” or “universal benefit”: Heterogeneity analysis of human capital

5.4.2 “Timely assistance” or “icing on the cake”: Heterogeneity analysis of vulnerable groups

5.4.3 “Digital divide” or “digital dividend”: Heterogeneity analysis of the financial environment

5.5 The functional mechanisms of digital financial inclusion

5.6 Further analysis

5.6.1 Structural analysis of digital inclusive finance on the risk of returning to poverty

5.6.2 Nonlinear effects of digital financial inclusion and risk of returning to poverty

5.6.3 The relationship between digital financial inclusion and private lending: “complementarity” or “substitution”?

6. Conclusions And Recommendations

Declarations

References

Additional Declarations

Status:

Journal Publication

Version 1