Solar photovoltaic interventions have reduced rural poverty in China

China has rapidly reduced poverty alongside profound economic transformation over the past two decades. Since 2013, targeted poverty alleviation has been a national strategy prioritizing precise assistance to poor individuals, households, and communities. Among ten targeted initiatives, the Solar Energy for Poverty Alleviation Program (SEPAP) aims to add over 10 GW of solar capacity by 2020 to benefit over 2 million citizens, primarily via village-level arrays, joint construction arrays, and rooftop systems for poor households. County selection prioritizes sunlight conditions, with economic conditions secondary. Despite extensive deployment, prior work lacks a systematic ex-post quantitative evaluation of SEPAP’s causal impact on rural poverty. This study asks whether targeted PV deployment increases rural per-capita disposable income, how effects evolve over time, and whether impacts differ between national poverty-stricken and non-poverty counties. Using county-level panel data (2013–2016), we estimate the program’s effects and explore mechanisms and heterogeneity.

Prior literature on PV and poverty has largely emphasized rural electrification in off-grid or developing country contexts and in a few developed countries, including reviews on PV-based rural electrification and energy poverty (e.g., Mandelli et al., Chaurey and Kandpal, Rodríguez et al., Rosas-Flores et al.). Studies specific to China’s SEPAP have identified funding constraints as key bottlenecks (Li et al.; Xu et al.; Wu et al.), highlighted industry structure and overcapacity issues (Xue; Zhou and Liu), and examined governance challenges and incentive alignment (Geall et al.). Liao and Fei studied 471 pilot counties focusing on project information and capacity rather than county-level socioeconomic outcomes. Overall, the literature lacks a systematic causal, ex-post evaluation of SEPAP’s impact on poverty reduction at the county level, particularly disentangling effects in national poverty-stricken versus non-poverty counties.

Design and data: The study uses a county-level panel (2013–2016) covering 1142 counties after excluding missing data, including 211 SEPAP pilot counties (representing >52% of GDP among 471 SEPAP counties, distributed across Eastern, Central, Western China) and a control group of non-pilot counties. Key variables include rural per-capita disposable income (outcome), SEPAP treatment status, program exposure duration, industrial structure (secondary industry share), public expenditure-to-revenue ratio (PUBEXINR), agricultural facilities land area, education (secondary school students/population), solar resource (annual sunshine hours), provincial GDP per capita, and a regional marketization index. Solar irradiance is proxied by sunshine hours from >700 meteorological stations (China Meteorological Administration).

Identification: To address nonrandom treatment assignment, the authors implement a difference-in-differences (DID) estimator with county and year fixed effects, comparing pre/post changes in treated versus control counties. They complement DID with propensity score matching (PSM) to construct comparable control groups based on county-level characteristics (per-capita GDP, education, agricultural land area), then estimate DID on the matched sample (PSM-DID). Matching ratios tested include 1:30 to 1:70; the final ratio is 1:70. Covariate balance is reported.

Models: The baseline DID specification is ln(DISINRURAL_it) = α_i + β SEPAP_it + λ_t + γ Z_it + ε_it, where Z_it includes the controls listed above. They also model dynamic effects by interacting SEPAP with exposure duration following King and Jere to capture cumulative impacts over time. Event-study specifications test the parallel trends assumption by estimating leads and lags relative to policy implementation; the year before implementation is the base period. Robustness includes alternative samples (e.g., restricting to national poverty-stricken pilot counties; excluding national poverty counties), adding region-year fixed effects (East, Central, West), and alternative control definitions.

Hypotheses: H1—SEPAP increases rural disposable income. H2—Effects grow with exposure duration (cumulative effect). H3—Impacts are greater in national-level poverty-stricken pilot counties than in non-poverty counties.

Heterogeneity and mechanisms: The study examines heterogeneity by region and economic condition and discusses direct (income distribution from village-owned PV revenues) and indirect (improved electricity access, knowledge/information spillovers, employment) channels through which SEPAP might affect incomes (illustrated in Fig. 3).

• Main DID results (Table 1): SEPAP is associated with a 7.24%–7.52% increase in rural per-capita disposable income (coefficients ~0.0724–0.0725, p<0.01) controlling for county and year fixed effects and standard covariates.
• Dynamic/cumulative effects: The interaction with exposure duration is positive and significant (e.g., coefficient 0.0274, p<0.01), indicating that income gains grow in the two to three years following implementation; a model with additional controls shows 0.0195 (p<0.01).
• PSM-DID on matched sample (Table 2, Panel A): SEPAP effect remains positive and significant at about 2.52%–2.67% (p<0.05) with improved covariate balance (Panel B).
• Alternative specifications: When focusing only on national poverty-stricken SEPAP pilot counties or restricting the dataset, estimated effects drop to roughly 2.6%–2.7%, contrary to H3, possibly due to crowding-out by other poverty funds or increased rent-seeking behavior.
• Adding region-year fixed effects yields an estimated impact of about 7.51%, similar to main results.
• Control variables: The public expenditure-to-revenue ratio (PUBEXINR) shows a significant negative association with rural disposable income; marketization index and solar resource (sunshine hours) are positively associated with income.
• Heterogeneity: Effects are stronger in poorer regions and notable in Eastern China; regional disparities suggest local context (market development, policy environment) mediates outcomes.
• Scale and scope: By end-2018, 15.44 million kW of PV poverty alleviation capacity was allocated nationwide benefiting 2.24 million registered poor households, illustrating substantial program scale (descriptive context).

The findings provide causal evidence that SEPAP increased rural per-capita disposable incomes, directly addressing the research question about PV’s contribution to poverty alleviation in China. The documented cumulative effects support the view that benefits accrue as governance of village-level distribution systems improves, electricity access quality rises, and local capacities develop over time. The negative association between public expenditure ratios and income suggests potential bureaucratic frictions that may impede efficient transfer of PV revenues to target households, aligning with concerns about rent-seeking and fund misallocation.

Heterogeneous impacts by region and economic condition underscore the importance of tailoring interventions to local market development, policy environments, and resource endowments. The study highlights plausible mediating factors—feed-in-tariffs, financing policies, guaranteed dispatch for renewables, and support systems for village cadres—that can shape program effectiveness. The channels identified (direct income distributions from village-owned PV projects and indirect effects via improved electricity services, knowledge, and employment opportunities) situate SEPAP within broader community-development dynamics.

Internationally, the results offer lessons for other developing regions (e.g., Sri Lanka, Bangladesh, Palestine) considering targeted PV for poverty reduction: success depends not just on technology deployment but also on institutional incentives, transparent revenue distribution, and alignment with local economic structures.

This study provides a systematic, ex-post causal evaluation of China’s SEPAP, showing that PV-based targeted interventions increased rural per-capita disposable income by about 7%–8%, with benefits growing in the years following implementation and stronger effects in poorer regions. While PV deployment presents a viable pathway for poverty alleviation, program design and governance are crucial to ensure revenues reach intended beneficiaries and to mitigate potential rent-seeking. Policy recommendations include tailoring support to regional heterogeneity, strengthening financial and technical mechanisms, ensuring transparent and efficient income distribution, and planning for a gradual transition toward reduced state support as broader PV subsidies decline. The Chinese experience suggests targeted community-scale PV, when embedded in appropriate local institutions and incentives, can be an effective poverty alleviation tool and offers transferable insights for other developing contexts.

• Time horizon: The analysis covers a short post-implementation period (2013–2016), limiting assessment of long-term impacts.
• Data gaps: Missing county-level data (e.g., rural income for some pilot counties such as in Qinghai and Tibet; solar irradiance proxied by sunshine hours) may constrain precision and external validity.
• Identification limits: DID controls for time-invariant unobservables, and PSM improves comparability, but time-varying unobservables correlated with treatment may remain.
• Mechanisms not fully disentangled: The study discusses direct and indirect channels but does not causally separate them; future work is needed to quantify pathway contributions.
• Governance and implementation variability: Potential rent-seeking, bureaucratic barriers, underutilized projects, and differences in local policy execution can bias impacts and limit generalizability.
• Policy environment dynamics: Changes in broader PV feed-in-tariffs and financing conditions may affect program sustainability and scalability beyond the study window.