The effects of China's poverty eradication program on sustainability and inequality

The paper examines how China’s decade-long poverty eradication program (2010–2020) interacted with other SDGs, particularly environmental sustainability and equitable access to public services. While SDG1 targets ending poverty, interactions and potential tradeoffs with goals on health, education, water, energy, cities, climate, and terrestrial ecosystems remain uncertain. The authors focus on 832 officially designated poverty-stricken counties, asking: (1) Did sustainability in ecological/environmental dimensions and public services improve during poverty eradication? (2) Did inequality between poverty-stricken and non-poverty-stricken counties narrow? (3) How are these changes related to government investment structure, and what investment adjustments could reduce future inequalities? They hypothesize that multidimensional government investments (ecological restoration, clean energy, education, health, housing, transportation) can jointly enhance sustainability and reduce inequality, but that unbalanced regional or sectoral investment may widen gaps in higher education and high-quality health care.

Prior work documents China’s post-1978 growth lifting hundreds of millions from poverty but with environmental degradation and regional disparities in public services (Lu et al. 2019; Zhao et al. 2022b). Large-scale ecological programs (Grain for Green, Natural Forest Conservation, Grassland Protection) improved ecosystems (Ouyang et al. 2016; Bryan et al. 2018), though subsidy-based approaches had limited durable poverty impacts and did not adequately improve public services (Liu et al. 2008; Miao and Li 2023). Studies assessing SDG progress at provincial scales exist (Xu et al. 2020), but county-level analyses for designated poverty counties and explicit inequality decomposition between poverty and non-poverty regions were lacking. The Theil index is widely used to decompose total, within-, and between-region inequality (Akita 2003; Xian and Chen 2022), offering advantages over the Gini for subregional analyses. The authors also reference potential SDG tradeoffs (Fu et al. 2019) and interactions (Wu et al. 2022), and broader literature on environmental and service access inequalities (Bluhm et al. 2022; Rammelt et al. 2023; Sanders et al. 2023), motivating their multidimensional inequality indicators and gap measures.

Study scope: 832 poverty-stricken counties (official list issued in 2014) versus non-poverty counties in China, comparing 2010 and 2020 to capture changes over the poverty eradication decade.

Indicators and metrics: Seventeen metrics were aggregated into nine sustainability indicators aligned with multiple SDGs (3, 4, 6, 7, 11, 13, 15): clean air (PM2.5), clean surface water (NH3-N), climate change (CO2 emissions per capita), ecological resources (NPP), basic education (primary education buildings/facilities per million people), higher education (vocational and higher education facilities per million), basic health (rural clinics/anti-epidemic stations; drug stores per million), high-quality health (general hospital buildings/facilities per million), and housing & transportation (housing quality, water, energy use, electricity, internet, road, bus access). Sustainability scores (0–1) were computed per metric at county level using min–max normalization: negative metrics via S = (max−x)/(max−min) and positive metrics via S = (x−min)/(max−min). County indicator scores are averages of their constituent metric scores.

Inequality measurement: One-stage Theil decomposition quantified overall (Tp), within-region (TWR), and between-region (TBR) inequality for each sustainability metric/indicator, with regions defined as poverty-stricken vs non-poverty counties (population-weighted). A complementary gap indicator measured relative performance of poverty counties against the mean of non-poverty counties for each metric: Gap = k × (x − mean_nonpoverty)/mean_nonpoverty, where k = −1 for negative metrics and +1 for positive metrics; values near 0 denote equality, negative indicates lag.

Projections: A business-as-usual (BAU) scenario projected metrics for 2030 and 2040 by applying the 2010–2020 mean change ratios to each metric for poverty counties and to the means for non-poverty counties; sustainability and gap indicators were recomputed with fixed bounds.

Investment analysis and scenarios: Provincial-level annual investments (2010–2020) in forest/grassland restoration (livelihood projects of forestry, forestry industry development, ecological construction/protection), rural clean energy (solar, small hydropower, biogas), and public services/infrastructure (housing security, transportation, education, health) were compiled from statistical yearbooks. Pearson correlations related cumulative investments to provincial poverty incidence, share of poverty counties, and GDP. Two linear investment scenarios were simulated: (1) increase per-capita investment in poverty counties to at least match non-poverty levels in 2020 for fields showing negative and declining inequality indicators; (2) estimate per-capita increases required to eliminate inequality in the next decade using linear relationships between investment and the targeted metrics.

Data sources: Statistical yearbooks and rural poverty monitoring (housing, water, cooking fuel, electricity, internet, highway, bus); county population from National County Statistical Yearbook; POI data for counts of education and health facilities (RESDC) normalized per million population; environmental data: PM2.5 (TAP database; 1 km), NH3-N (national monitoring network interpolated to 1 km), CO2 per capita (CEADs; county totals for 2010, 2017 divided by population; Tibet estimated via GDP-based factors), NPP (MODIS MOD17A3, 500 m). GIS processing in ArcGIS 10. Supplementary materials document bounds, processing, and additional methods.

• Overall sustainability improved: average sustainability score in poverty-stricken counties rose from 0.48 to 0.64 (2010–2020). Ecological/environmental sustainability increased 15.2% (0.74 to 0.85); public services increased 74.9% (0.27 to 0.47).
• Declines in specific indicators: climate change sustainability decreased 4.2% (0.892 to 0.855) due to rising CO2 per capita; higher education sustainability decreased 34.2% (0.112 to 0.074).
• Inequality reduced overall: mean Theil index fell from 0.46 to 0.35; between-region (poverty vs non-poverty) share of overall inequality declined from 9.3% (2010) to 7.7% (2020).
• Metric-level changes (poverty counties vs non-poverty): • PM2.5 (µg/m³): 40.6 → 22.9 vs 58.8 → 30.1. • NH3-N in surface water (mg/L): 1.2 → 0.1 vs 1.2 → 0.2. • CO2 per capita (t): 4.7 → 6.0 vs 8.7 → 10.1. • NPP (g C/m²): 506.6 → 544.5 vs 420.5 → 458.7. • Primary education facilities (per million): 102.2 → 175.4 vs 171.2 → 261.0. • Vocational education facilities (per million): 9.7 → 7.1 vs 35.0 → 15.3. • Higher education facilities (per million): 3.8 → 2.5 vs 13.8 → 27.9. • Rural clinics/anti-epidemic stations (per million): 48.0 → 276.0 vs 74.3 → 385.4. • Drug stores (per million): 50.7 → 487.8 vs 163.4 → 1003.5. • General hospital facilities (per million): 49.0 → 92.6 vs 61.4 → 196.3. • Housing/transportation in poverty counties: adobe houses 7.7% → 1.4%; purified tap water 34.3% → 58.6% (non-poverty: 71.5% → 96.9%); firewood use 57.7% → 36.6%; electricity 98.0% → 100%; internet 43.7% → 95.1%; highway 66.2% → 100%; public bus 49.7% → 75.4%.
• Spatial heterogeneity: Western regions (notably Tibetan Plateau, Xinjiang) lagged in improvements; advantages in ecological status weakened and public service gaps widened there.
• County classifications: 227 poverty counties (15% of population; 54.8% of land) both lost environmental advantage and saw increased lags in public services; increased inequality concentrated in southwestern China.
• Projections (BAU to 2040): clean air, water, and ecological resources continue improving with positive gap indicators (advantage for poverty counties); basic education, basic health, housing/transport gradually improve with negative gap indicators approaching zero; climate change indicator slightly decreases (CO2 continues to rise, convergence of per-capita emissions); high-quality health improves but inequality becomes more negative (gap widens); higher education sustainability decreases and inequality gap widens.
• Investment patterns and impacts: Per-capita investment (2010–2020) in forest/grassland restoration was 67.9% higher in poverty counties (US$88.50 vs US$52.70); rural clean energy >3× higher (US$35.80 vs US$11.70). Outcomes included ~5× larger afforestation area, ~20× larger grass planting/improvement area, and nearly 2× small hydropower generation in poverty counties, relative to non-poverty counties. However, per-capita investments in education and health were lower in poverty counties and correlated with higher GDP (education r≈0.442; health r≈0.293), contributing to widening gaps in higher education and high-quality health care.
• Investment scenarios: To prevent further widening (2020–2030), increase investment in poverty counties by 23.2% (higher education) and 5.0% (high-quality health), assuming no increases in non-poverty areas. To eliminate inequality, raise investments by 226.2% (higher education) and 72.0% (high-quality health).

Findings indicate that poverty eradication can be synergistic with environmental protection: substantial improvements in air quality, surface water quality, and ecological resources were achieved alongside poverty reduction, consistent with national pollution control trends. However, continued growth raised CO2 emissions per capita, reducing the climate-change sustainability indicator and underscoring the need to align poverty alleviation with carbon peaking and neutrality goals. Inequality fell overall, but persistent and in some cases widening gaps in higher education and high-quality health care emerged, especially in western regions. The results suggest sectoral and regional investment biases: poverty counties received comparatively higher per-capita investments in ecological restoration and clean energy, which delivered co-benefits for sustainability, but lagged in per-capita investment for education and health, likely reinforcing service disparities. Cross-ministry or central coordination of investments can mitigate SDG tradeoffs and target lagging dimensions, with higher education investment offering potential double dividends for reducing inequality and fostering regional economic growth. The BAU projections highlight where gaps would widen absent policy optimization, guiding prioritization toward education and high-quality healthcare provisioning in poverty-stricken regions.

China’s 2010–2020 poverty eradication program coincided with notable gains in environmental and public service sustainability and an overall reduction in inequality across counties. Yet, gaps vis-à-vis non-poverty counties widened in higher education and high-quality healthcare. Investment structures strongly relate to observed inequalities: per-capita spending favored ecological restoration and clean energy in poverty counties but lagged for education and health. To close gaps, substantial increases in education and health investments in poverty counties are needed—on the order of 226.2% and 72.0%, respectively, under a scenario targeting elimination of inequality. While BAU projections provide a worst-case baseline, future research should incorporate more comprehensive policy scenarios to inform coordinated, sector-balanced strategies that jointly advance SDGs and equity.

The business-as-usual projections are baselines, not predictions; they assume linear continuation of 2010–2020 trends and fixed indicator bounds, which oversimplifies complex dynamics and potential policy shifts. The analysis aggregates changes at two time points (2010, 2020), potentially masking annual dynamics. Linear assumptions linking investment to metric improvements may miss nonlinearities. Data constraints (e.g., incomplete metrics for non-poverty counties in housing/transport) required relying on subsets of indicators. Spatial heterogeneity (e.g., ecological limits in arid/high-altitude regions) may constrain restoration effectiveness. Classification at county level may overlook intra-county poverty in non-poverty areas; household-level data could refine inequality assessments. Nonetheless, concurrent policies (rural revitalization, carbon peaking and neutrality) could alter BAU trajectories, likely improving air quality and narrowing service gaps beyond the baseline.