The poverty alleviation effect of transfer payments: evidence from China

Poverty has long been a formidable obstacle hindering rapid economic progress worldwide. Particularly for low-income rural residents, poverty poses challenges in fulfilling fundamental necessities such as water, housing, and transportation. Reducing poverty is essential for enhancing residents' well-being and health outcomes. Consequently, “no poverty” has become a primary objective within sustainable development goals. While economic growth is crucial for alleviating poverty, it alone cannot eradicate this issue entirely; macro-control policies are important complements. China initiated major poverty alleviation efforts around 2012. World Bank data show that the number of poor people (US$5.50/day, 2011 PPP) in China fell from 223.8 million in 2019 to 186 million in 2021; using US$3.20/day standards, around 50 million people were lifted out of poverty from 2016 to 2019, and 98.99 million rural poor were lifted out of poverty by end-2020. Rural areas are central to poverty alleviation strategies, and fiscal transfers are a main tool in China, including the new rural endowment insurance, rural minimum living security, new rural cooperative medical system, and agricultural subsidies. This paper asks: (1) Can government transfer payments effectively eradicate rural poverty? (2) Do China’s transfer payment policies have adverse effects? (3) Have transfer payments improved living conditions for impoverished rural families? The paper outlines theory and hypotheses, research design, empirical results and discussion, and conclusions with policy recommendations.

Debate persists on the impact of transfer payments on poverty. Some studies argue transfers increase non-labor income and reduce labor supply by raising the shadow price of household labor, potentially offsetting gains and even lowering residents’ income (Van den Berg & Cuong, 2011; Clark & Lee, 2008; Habimana et al., 2021; Song & Xie, 2020). Others find transfers can improve healthcare access, reduce medical impoverishment risk, enhance labor efficiency and income, and improve education and human capital (Shmueli et al., 2008; Aryeetey et al., 2016; Qin et al., 2021; Fang & Zhou, 2020; Chiapa et al., 2012; Gutiérrez et al., 2019; Hofmarcher, 2021; Campillo & García, 2022). Measuring poverty solely by income may be biased; consumption can better reflect living standards (Short et al., 1998; Liu et al., 2023). Based on this, the authors propose: Hypothesis 1: Transfer payments contribute to an increase in income among rural residents in China. Hypothesis 2: Transfer payments enhance total consumption expenditure and elevate living standards among rural residents, with priority on basic needs (e.g., healthcare due to disease prevalence).

The study employs counterfactual causal inference via propensity score matching (PSM) to estimate the effect of government transfer payments on rural poverty outcomes and mitigate omitted variable bias and sample self-selection. Outcomes are modeled as: lny = α + β p + γ x + u, E(u)=0, where y denotes income or consumption outcomes (often in logs), p indicates receipt of transfer payments (1=treatment, 0=control), and x is a vector of covariates. A probit model estimates the propensity score P(X)=Pr{exp1=1|X}. Matching on the common support yields the average treatment effect on the treated (ATT): ATT = E(Y1|p=1) − E(Y0|p=1), with the counterfactual E(Y0|p=1) obtained via matching. Data: China Labor-force Dynamic Survey (CLDS) 2016 and CLDS 2018 from the Center for Social Survey (Sun Yat-sen University). The analysis focuses on rural households; final 2016 sample size n=4525 (treatment=553; control=3972). Treatment is defined from survey items indicating receipt of unemployment benefits, social assistance funds, or basic living allowances. Covariates capture household head characteristics (age, marital status, education), family size, living conditions and infrastructure (tap water, uncontaminated water, natural gas, power interruptions, productive electricity, internet, phone), assets (car, tractor, agricultural implements, livestock), debt, agricultural cost, agricultural subsidy access, and other production/employment factors. Outcomes include total income and its components (agricultural, wage, operating, property), and consumption expenditures (total, food, healthcare, education). Balance and common support tests confirm good matching quality with low post-match bias and sufficient overlap; graphical diagnostics support standardized bias near zero. Nearest neighbor matching is presented; kernel and radius matching yield consistent results (provided in appendix). Robustness: Results are replicated using CLDS 2018 with the same variables and procedures. Additional analyses include effects on the wage-labor ratio and Engel’s coefficient. Further robustness uses regression with the logarithm of transfer payments as the key regressor, controlling for the same covariates.

Using CLDS 2016 (PSM, nearest neighbor): Income outcomes (Table 4): - Total income ATT −8,071.502 CNY (SE 1,789.075), significant at 1%. - Wage income ATT −4,653.331 CNY (SE 1,361.371), significant at 1%. - Agricultural income ATT −1,215.525 CNY (SE 1,163.302), not significant. - Operating income ATT −892.829 CNY (SE 617.890), not significant. - Property income ATT +83.480 CNY (SE 297.217), not significant. Consumption outcomes (Table 5): - Total consumption ATT −5,670.809 CNY (SE 3,562.517), not significant. - Food expenditure ATT −2,300.661 CNY (SE 832.702), significant at 1%. - Healthcare expenditure ATT +1,286.382 CNY (SE 699.204), significant at 10%. - Education expenditure ATT +370.110 CNY (SE 671.043), not significant. Using CLDS 2018 (PSM): Income outcomes (Table 6): - Total income ATT −12,108.048 CNY (SE 2,430.237), significant at 1%. - Wage income ATT −7,785.765 CNY (SE 1,715.237), significant at 1%. - Agricultural income ATT −2,432.918 CNY (SE 1,477.286), not significant. - Operating income ATT −1,389.632 CNY (SE 1,248.588), not significant. - Property income ATT −344.634 CNY (SE 265.310), not significant. Consumption outcomes (Table 7): - Total consumption ATT +474.368 CNY (SE ~832.702), not significant. - Food expenditure ATT −2,468.343 CNY (SE ~493.936), significant at 1%. - Healthcare expenditure ATT +1,173.278 CNY (SE ~2,123.477), positive (significance not indicated). - Education expenditure ATT −782.508 CNY (SE ~533.834), not significant. Mechanisms and ancillary indicators: - Wage-labor ratio (Table 8): 2016 ATT −0.080 (SE 0.027), 2018 ATT −0.110 (SE 0.031); both significant at 1%, indicating reduced labor supply among recipients. - Engel’s coefficient (Table 9): 2016 ATT −0.117 (SE 0.064), significant at 10%; 2018 ATT −0.093 (SE 0.038), significant at 1%, implying improved living conditions (lower food share). Regression robustness (Table 10; log transfer payments as regressor): - Total income coefficient −0.043 (SE 0.007), 1% significance. - Wage income coefficient −0.107 (SE 0.034), 1% significance. - Food expenditure coefficient −0.046 (SE 0.008), 1% significance. - Healthcare expenditure coefficient +0.078 (SE 0.022), 1% significance. Overall: Transfer payments are associated with lower total and wage incomes (labor supply disincentive), reduced food expenditure, and increased healthcare expenditure; negligible effect on education spending.

The findings invalidate Hypothesis 1: transfer payments do not increase rural households’ income in China and instead are associated with reduced total and wage income, consistent with a labor supply reduction among recipients (confirmed by declines in the wage-labor ratio). Hypothesis 2 is supported in terms of living standards: despite little change in total consumption, transfers are linked to a reallocation of consumption away from food and toward healthcare, and a lower Engel’s coefficient, indicating improved basic living conditions and health protection. The positive association with healthcare expenditure aligns with China’s poverty alleviation focus on enhancing rural medical access and insurance reimbursement, helping address medical impoverishment, a key cause of rural poverty. However, education expenditures do not significantly increase, suggesting limited progress on longer-term human capital accumulation and poverty reduction via education. These results imply that current transfer programs in China primarily provide short-term basic livelihood support and health improvements rather than income augmentation or long-term human capital investment. The labor supply disincentive effect highlights the need for careful program design to mitigate potential adverse incentives while preserving protective functions.

Using CLDS 2016 and 2018 data with PSM and robustness checks, the study concludes: 1) Transfer payments are associated with decreases in total and wage incomes among rural poor households, likely through reduced labor supply, indicating limited effectiveness in raising incomes. 2) Transfers reduce food expenditure and increase healthcare spending; along with lower Engel’s coefficients, this suggests improved basic living security and health conditions, addressing disease-related poverty risks. 3) Transfers have no significant effect on education spending, implying limited long-term impact on poverty avoidance and intergenerational transmission. Policy recommendations: - Target transfers toward households with limited labor capacity to reduce labor supply disincentives among able-bodied recipients. - Complement unconditional transfers with conditional cash transfers aimed at reducing schooling costs (e.g., meals, supplies) to encourage human capital investment. - Continue strengthening rural healthcare access to restore labor capacity and prevent medical impoverishment. - Use multidimensional poverty indicators (including consumption) when evaluating policy impacts, not income alone.

• The analysis focuses on rural households using CLDS 2016 and 2018; findings may not generalize to urban populations or other contexts. - PSM relies on the unconfoundedness assumption given observed covariates; unobserved factors may bias estimates. - The authors note potential reliability issues if common support is insufficient, which could cause sample loss; balance diagnostics suggest good overlap here, but residual concerns remain. - Consumption and income outcomes are based on survey-reported data, subject to measurement error. - The datasets are not publicly available due to confidentiality agreements, limiting external replication. - Education effects are assessed via expenditure, which may not fully capture educational attainment or quality impacts.