Does the BRI contribute to poverty reduction in countries along the Belt and Road? A DID-based empirical test

The study addresses whether the Belt and Road Initiative (BRI) contributes causally to poverty reduction in participating countries and through which channels this effect operates. Against the backdrop of stalled global poverty reduction due to COVID-19, climate change, and conflicts, and in light of China’s domestic poverty alleviation success (e.g., Targeted Poverty Alleviation and meeting SDG poverty goals ahead of schedule), the paper examines how BRI, as a major platform of international cooperation, can align with and advance SDG1 (no poverty). It frames three questions: (1) Does BRI participation reduce poverty in member countries? (2) Does the effect vary by income group, geography, and type of BRI pathway (land vs maritime)? (3) Through what mechanisms (the “Five Cooperation Priorities”: policy coordination, facilities connectivity, unimpeded trade, financial integration, and people-to-people bonds) does BRI reduce poverty? The study argues this inquiry is important for improving BRI’s design and for global poverty governance.

The literature on BRI spans international relations, political science, and international business, debating motives (soft power vs market expansion) and documenting economic effects for China (enhanced export quality, reduced financing constraints, productivity improvements) and for BRI countries (economic growth, regional integration, lower trade costs from infrastructure, urban development, employment gains), alongside concerns about competition and environmental impacts. Global welfare analyses suggest broader benefits, including for some non-BRI countries, via connectivity and trade expansion. While links between BRI and SDGs are often asserted, causal evidence directly connecting BRI to poverty reduction and unpacking mechanisms is limited. This study extends the literature by testing the causal relationship between BRI and poverty reduction and by examining mechanisms aligned with the Five Cooperation Priorities, integrating insights from development economics (capital accumulation, trade expansion, infrastructure-led growth, institutional quality) and China’s domestic poverty reduction experience.

Design: A difference-in-differences (DID) framework evaluates the average treatment effect of BRI on poverty. Treatment is BRI participation; the policy cutoff is 2013 (initiative launch). BRI countries form the treatment group; non-BRI countries form the control group. Country and year fixed effects are included. Sample and period: Balanced/pooled panel of 151 countries from 2005–2019 (53 BRI, 98 non-BRI), covering 97.6% of global GDP and 98.3% of global population (2019 basis), subject to data availability. Outcome (poverty): Moderate poverty headcount ratio at the $3.20/day (2011 PPP) threshold from PovcalNet. The moderate line is chosen over $1.90/day given the predominance of middle-income countries among BRI participants and better variability for panel analysis. Key policy variable and DID interaction: BRI_i (1 if BRI country), Time_t (1 for years ≥2013). DID = BRI_i × Time_t; its coefficient identifies the post-2013 average effect for BRI countries relative to controls, conditional on parallel trends. Controls: Economic (log GDP per capita; GDP growth; gross fixed capital formation % GDP; natural resource rents % GDP; agriculture value added % GDP); Social/demographic (urbanization rate; labor force participation rate; population density; age dependency ratio); Governance (institutional quality index averaging voice and accountability, rule of law, regulatory quality, political stability, control of corruption, government effectiveness; -2.5 to 2.5). Data are primarily from World Bank WDI/WITS, PovcalNet, UN, China BRI Portal, and CGIT. Heterogeneity analyses: (a) Geography: neighboring vs non-neighboring countries; continuous distance to China’s capital/major cities interacted with DID. (b) Income groups: World Bank classifications (low/lower-middle combined vs upper-middle; and middle-income aggregate). (c) BRI type: Land Silk Road Economic Belt vs 21st-Century Maritime Silk Road. Mechanism (mediation) tests: Stepwise regressions and Sobel tests for Five Cooperation Priorities proxies: - Unimpeded trade: multilateral trade volume, bilateral trade with China, and number of trade partners (WITS). - Financial integration: number of commercial bank branches (bank_n, WDI), FDI inflows (WDI), and China’s greenfield investment projects (CGIT). - Facilities connectivity: mobile subscriptions and internet use (WDI) as digital infrastructure proxies. - People-to-people bonds: number of Confucius Institutes/schools in host countries. - Policy coordination: number of bilateral official visits (China Diplomatic Yearbooks) and technical cooperation exchanges (World Bank). Estimation details: Country and year fixed effects in all core and mediation regressions; standard errors summarized by t-statistics. The paper treats 2013 as a common treatment start; robustness includes heterogeneity and mediation analyses; event-time or staggered adoption methods are noted as future work.

- Main effect: BRI participation significantly reduces moderate poverty. DID coefficient with controls is -3.42 percentage points, indicating BRI countries’ poverty headcount ratio fell by 3.42 pp more than non-BRI countries post-2013. - Time dynamics: Effects exhibit a lag and strengthen over the longer term (as stated in the abstract and discussion). - Controls (direction and magnitudes reported): - Higher economic scale (log GDP per capita) and GDP growth associate with lower poverty; a 1 pp rise in economic scale corresponds to -1.79 pp poverty; GDP growth coefficient is modestly negative (-0.05 pp per 1 pp growth). - Physical capital formation reduces poverty; agricultural share increases poverty (+0.35 pp per 1 pp increase), suggesting need for upgrading agriculture and structural transformation. - Urbanization reduces poverty (-0.65 pp per 1 pp increase); dependency ratio raises poverty (+0.45 pp per 1 pp). Labor force participation is positively associated with poverty (consistent with prevalence of low-quality/informal jobs). Better institutional quality substantially reduces poverty (-5.12 pp per 1-unit improvement). - Heterogeneity: - Geography: Stronger effects in countries neighboring China (e.g., DID around -9.50 in neighbors vs -0.87 in non-neighbors); effect attenuates with greater distance (positive interactions with distance measures). - Income: Stronger poverty reduction in lower-middle-income countries than in upper-middle and especially high-income countries; interaction terms confirm larger gains for lower-income participants. - BRI type: Both routes reduce poverty; estimated reductions: Maritime Silk Road about -5.99 pp vs Land Silk Road about -1.85 pp; maritime effect larger, likely due to trade via sea transport. - Mechanisms (mediation evidence): - Unimpeded trade: BRI increases multilateral and bilateral trade; higher trade and more trade partners are associated with lower poverty, indicating trade is a key channel. - Financial integration: BRI increases the number of commercial bank branches; more branches reduce poverty. China’s greenfield investments rise and are linked to poverty reduction; aggregate FDI inflows are not a significant channel. - Facilities connectivity: BRI increases mobile subscriptions and internet use; mobile subscriptions are associated with poverty reduction (internet’s direct effect on poverty not significant in this setup). - People-to-people bonds: Confucius Institutes increase but show no significant short-run impact on poverty (education/knowledge channels likely longer-term). - Policy coordination: More bilateral visits and technical cooperation; technical cooperation is associated with lower poverty, supporting knowledge/technology transfer as a pathway.

The findings directly address the core questions by establishing a negative causal effect of BRI participation on moderate poverty and identifying operative channels consistent with the Five Cooperation Priorities. The stronger impacts in neighboring and lower-middle-income countries suggest that proximity and development stage shape how effectively BRI platforms translate into growth and inclusion. Mechanism tests show that trade expansion, improved financial access and greenfield projects, digital connectivity (notably mobile penetration), and policy/technical cooperation are pivotal in transmitting BRI’s benefits to poor populations. The limited immediate effects of people-to-people educational exchanges on poverty underscore the long horizon needed for human-capital channels. Overall, the results indicate BRI can function as a complementary international cooperation mechanism for SDG1 when paired with inclusive policies and strong institutions in partner countries.

The paper contributes by (1) providing causal evidence that BRI participation reduces moderate poverty in member countries, (2) documenting heterogeneity—stronger effects for neighbors and lower-middle-income economies and larger impacts along the Maritime route, and (3) unpacking mechanisms aligned with the Five Cooperation Priorities, notably unimpeded trade, financial integration (including greenfield investment), facilities connectivity (digital), and policy coordination via technical cooperation. The study suggests that to maximize poverty reduction, BRI should continue expanding partnerships with low- and middle-income countries, deepen trade networks, bolster credit and financial infrastructure, promote greenfield investment, prioritize digital infrastructure, and enhance knowledge/technology exchange via policy coordination and people-to-people channels. Future research should employ staggered adoption (time-varying DID) and extend the analysis beyond 2019 to assess impacts through COVID-19 and geopolitical shocks, including effects on poverty vulnerability and resilience.

- Identification timing: A single 2013 cutoff is used despite staggered entry of countries; a time-varying (staggered) DID would yield more precise estimates. - Period coverage: Data end in 2019; impacts through COVID-19 and recent conflicts are not captured. - Mechanism measures: Proxies (e.g., Confucius Institutes, bank branches, mobile/internet) may not fully capture complex channels; some effects (education/knowledge) are long-term and may be underestimated in the sample window. - External validity: Estimates pertain to moderate poverty ($3.20/day) in a largely middle-income country sample; extreme poverty impacts may differ.