Breaking the chains of poverty: examining the influence of smartphone usage on multidimensional poverty in rural settings

China eradicated absolute poverty by the end of 2020, shifting policy focus to relative and multidimensional poverty. Prior work emphasizes that poverty spans education, health, living standards, income, and labor, aligning with the Alkire-Foster (A-F) framework. Concurrently, smartphone penetration in rural China has surged, with most households owning a device, suggesting potential for digital tools to influence livelihoods. The study asks whether and how smartphone usage alleviates multidimensional poverty among rural households that have recently exited income poverty, and whether social capital mediates this relationship. The authors hypothesize: (H1) smartphone usage reduces multidimensional poverty, but its effect diminishes as poverty identification becomes more stringent; (H2) social capital mediates the impact of smartphones on multidimensional poverty.

Evidence is mixed on the role of smartphones/ICTs in poverty reduction. Proponents find smartphones expand market information, reduce transaction costs, and increase incomes and non-farm employment, entrepreneurship, and well-being, while enhancing social capital (e.g., Ma and Wang 2020; Ma et al. 2018; Aker et al. 2016; Zeng et al. 2023; Zhuo et al. 2023). Social capital—networks, trust, participation, and support—has been linked to lower multidimensional poverty and alleviation of financial constraints (Liu et al. 2021; Xiong et al. 2021; Wang et al. 2023). Conversely, critics warn smartphone use may exacerbate the digital divide, limiting gains for poorer or less digitally literate groups, particularly in underdeveloped rural areas (Acılar 2011; Tayo et al. 2016; Deichmann et al. 2016). These contrasting findings underscore the need to examine mechanisms—especially social capital—through which smartphones may influence multidimensional poverty.

Study setting and data: Jiangxi Province, a major agricultural and key poverty alleviation region in China. Fieldwork was conducted in five counties/districts (Anyi, Jinxian, Nanchang County, Wanli District, Xinjian District) in July–August 2020. Using stratified sampling, eight villages per county/district were selected, and ten households per village that had overcome income poverty were randomly sampled. Trained interviewers administered questionnaires and conducted structured interviews on demographics, household capital, and livelihood strategies. Of 400 collected surveys, 382 valid responses remained after excluding missing data (95.5% validity). Variables: Dependent variable is multidimensional poverty measured via an A-F MPI constructed across five dimensions with ten indicators: Health (medical expenses; health insurance), Education (years of education), Income (per capita disposable income below 40% of national rural median in 2019: 5755.6 yuan), Living standards (electricity, cooking fuel, floor, assets, per capita housing area <12 m²), Labor (labor force share <1/3 of household size). Equal weighting was applied across dimensions and indicators. Households with deprivation score ≥1/3 were identified as multidimensionally poor, and were further categorized by MPI into VMPI (0≤M<1/3), GMPI (1/3≤M<2/3), and EMPI (2/3≤M≤1). Key independent variable: smartphone usage (1 if household uses smartphones; 0 otherwise). Mediating variable: social capital index constructed via entropy method from four dimensions—social networks (number of relatives who are cadres/public officials), social trust (5-point trust in government), social participation (membership in associations/cooperatives), and social support (number who would help at weddings/funerals). Controls: Individual characteristics of household head (gender, age, education, marital status); household characteristics (household size, number of laborers, total household income [log]). Analytical strategy: The A-F method (dual cutoff) identified deprivations and calculated H (headcount), A (intensity), and M=H×A. Dimensional contributions were decomposed. The impact of smartphone usage on VMPI, GMPI, and EMPI was estimated using OLS with robust standard errors, first without and then with controls. Mediation analysis assessed the role of social capital using a three-equation framework, with significance tested via bootstrap (1,000 replications; percentile and bias-corrected intervals). Robustness used computer usage as an alternative ICT proxy.

- Contribution of dimensions: Education (37.8%), labor force (29.7%), and health (20.4%) were the largest contributors to multidimensional poverty; income (11.3%) and living standards (0.9%) contributed less (Table 6). - MPI dynamics by threshold: As the poverty cutoff k increased, H and M declined while A rose; overall M followed an inverted U-shaped pattern (rising then falling). For smartphone users, H reached 0 by k=0.7; for non-users, by k=0.9 (Table 5). - Smartphone effects (with controls): Smartphone usage reduced VMPI by 57.6%, GMPI by 52.6%, and EMPI by 5% (all significant; diminishing effect under stricter poverty definitions) (Table 7). - Unidimensional incidence: Highest deprivations were in years of education (87.2%), medical expenses (76.8%), and labor (64.4%); income 24.3%; health insurance 14.1%; living standard indicators 0.9–3.4% (Table 2). - Mediation by social capital: Social capital partially mediated smartphone effects on VMPI (14.09% of total effect) and GMPI (20.84%), and fully mediated the effect on EMPI (91.67%). In models including social capital, the direct smartphone-EMPI link became insignificant while social capital remained significant (Tables 8–10). - Robustness: Replacing smartphone with computer usage yielded consistent reductions in VMPI and GMPI and increased social capital; mediation via social capital persisted (Table 9).

Findings confirm that smartphones alleviate multidimensional poverty among rural households transitioning out of income poverty, primarily by lowering information costs, enabling access to markets, services, and employment, and enhancing social capital. The effect size diminishes as stricter multidimensional poverty thresholds are applied, reflecting that extreme deprivation involves multiple, deeper constraints where smartphones alone are insufficient. Social capital is a key pathway: smartphones foster networks, trust, participation, and support, which in turn reduce poverty—especially for extreme poverty where the effect is fully mediated. The dimensional analysis highlights structural deficits in education, labor capacity, and health as central drivers, indicating that complementary investments in human capital and employability are needed alongside digital inclusion to sustain poverty reduction.

The study develops and tests a framework linking smartphone usage to multidimensional poverty reduction through social capital. Using 382 rural households in Jiangxi Province, it constructs an MPI across five dimensions (education, health, income, living standards, labor) and shows: (1) smartphones significantly reduce VMPI, GMPI, and EMPI, with attenuating effects under stricter poverty definitions; (2) education, labor, and health are the main contributors to multidimensional poverty; (3) social capital is the principal mechanism, fully mediating the effect for EMPI and partially for VMPI and GMPI. Contributions include extending MPI to include income and labor dimensions tailored to post-poverty households, constructing a multi-faceted social capital index, and validating results via robustness and bootstrap mediation tests. Future research should integrate additional contextual indicators (e.g., industrial development, transport), consider psychological dimensions of poverty, and focus on vulnerable subgroups (adolescents, women, the elderly) to refine targeted interventions.

- Geographic and contextual scope: The sample is limited to five counties/districts in Jiangxi Province; local industrial structure and transport conditions likely affect poverty dynamics and generalizability. - Measurement scope: Despite five dimensions, other pertinent factors (e.g., industry access, infrastructure quality) were not incorporated; additional field indicators could improve measurement. - Mediators and mechanisms: While social capital is central, psychological factors and other mediators were not modeled. - Population heterogeneity: The study does not disaggregate effects for particularly vulnerable groups (adolescents, women, elderly), who may experience poverty differently. - Digital divide considerations: Variations in digital literacy and access could moderate effects but were not explicitly modeled.