Convergence effect of the Belt and Road Initiative on income disparity: evidence from China

The study addresses whether and how China’s Belt and Road Initiative (BRI) influences domestic regional income disparities, focusing on inter-city disparities within provinces. Against the backdrop of China’s rapid growth alongside widening inequality (e.g., Gini rising to 0.467 by 2017), the paper situates the problem within concerns over social stability, growth, and innovation. Existing explanations for regional inequality emphasize factor endowments, market integration, government roles, and knowledge spillovers, but few examine BRI’s domestic economic-geography effects. The authors pose four questions: Does the BRI affect regional disparities? Does it promote income convergence across cities? Through what mechanisms (trade openness, industrial structural transformation)? And does the effect vary across East, Central, and West China? Contributions include embedding BRI’s impact within a public policy analysis framework in economic geography, providing new evidence on city-level income convergence, and documenting regional heterogeneity.

The literature highlights multiple determinants of regional inequality: factor endowments (physical and human capital), market integration (reducing segmentation and improving transport), government interventions (public spending and reducing administrative monopoly), and knowledge spillovers (innovation diffusion and collaborative networks). Research on the BRI largely examines international effects—opening border ports, outward FDI, trade upgrading, and partner-country growth—while domestic spatial effects have been underexplored. The paper also reviews China’s macro trends (2002–2017 growth in GDP and GDP per capita), persistent inter-provincial disparities (East vs. Central/West), rising trade openness post-WTO accession, and major policy regimes (growth poles, coordinated regional development, new opening policies including pilot free trade zones and the BRI). The theoretical framing in public policy geography posits that national policies shape spatial economic outcomes, implying BRI can influence inter-city income convergence via enhanced connectivity, formation of new growth poles, trade openness, and industrial restructuring.

Data: Balanced panel of 269 prefecture-level cities across 26 Chinese provinces, 2002–2017 (4304 observations). Exclusions: Four centrally administered municipalities (Beijing, Shanghai, Tianjin, Chongqing); Tibet and Qinghai (insufficient city-level data); other cities with missing data. Variables: Dependent variable is provincial-level inter-city income disparity, primarily measured by coefficient of variation (CV) of city GDP per capita within each province-year. For robustness, Gini coefficient across cities within a province-year is also computed. Independent variable: BRI policy exposure using a difference-in-differences (DID) setup. Treat equals 1 for cities in provinces designated as affected by the BRI (per Wang and Lu, 2019), 0 otherwise. Policy equals 0 for 2002–2014 and 1 for 2015–2017. The key DID term is treat × policy. Mediators: Trade openness (traopen) proxied by FDI/GDP (due to limited availability of city import/export totals); Industrial structural transformation (transition) proxied by the ratio of tertiary industry output value to secondary industry output value. Controls: Government expenditure on science and technology (govtec: sci-tech spend/fiscal expenditure), urbanization (urban: urban population share), human capital (human: higher-education students/population), marketization (market: private-sector employees/total employees), cultural capital (culture: public books per capita), and informatization (ln net: log of international internet users). Descriptive statistics reported for all variables. Models: - Baseline two-way fixed-effects DID: citydis_it = β0 + β1 (treat×policy)_it + β2 X_it + γ_t + η_i + ε_it, where citydis is CV-based disparity; city and year fixed effects included. - Dynamic DID: replace treat×policy with treat×year2016 and treat×year2017 to trace post-policy dynamics. - Endogeneity/selection: Two-stage least squares (2SLS) with an instrumental variable (IV) based on provinces along the ancient Silk Road. First stage uses IV×policy to instrument treat×policy, motivated by strong historical correlation with BRI exposure and presumed exogeneity to modern inter-city income disparities. - Robustness: (i) Replace dependent variable with Gini; (ii) Triple-difference specification with tridid indicating treated post-period observations; (iii) Propensity score matching (KNN, radius, kernel, Mahalanobis) combined with DID. - Mechanism tests: Augment models with triple interactions treat×policy×traopen and treat×policy×transition to test mediation by trade openness and industrial restructuring. - Heterogeneity: Subsample regressions for East China vs. Central and West China. Standard errors and significance levels reported; fixed effects included in all main specifications.

- Average DID effect: treat×policy coefficient around −0.034 to −0.035 (t ≈ −5.4 to −5.7), significant at 1%, indicating that BRI exposure reduces inter-city income disparity within provinces. - Dynamic effects: treat×year2016 ≈ −0.039 (1%); treat×year2017 ≈ −0.020 (5%), suggesting a persistent but tapering convergence effect over time. - IV-2SLS: First-stage IV×policy strongly predicts treatment (F-stat > 10). Second-stage treat×policy remains significantly negative at 1%, with larger magnitude than OLS/FE, supporting a robust causal effect. - Robustness: Using Gini as dependent variable yields consistent negative and significant coefficients. Triple-difference estimates remain significantly negative and similar in magnitude. PSM-DID average treatment effects range from −0.030 to −0.044 across matching methods, all significant. - Mechanisms: Triple-interaction results show treat×policy×traopen significantly negative (≈ 5% level), and treat×policy×transition significantly negative (≈ 1% level), indicating BRI’s disparity-reducing effect operates via increased trade openness and industrial structural upgrading. - Regional heterogeneity: East China subsample shows a small, statistically insignificant effect of BRI on inter-city disparity; Central and West China subsample shows a larger, significant reduction (e.g., treat×policy ≈ −0.048, significant), indicating stronger convergence effects outside the East. - Overall, evidence supports Hypotheses 1–3: BRI reduces regional income disparities, with dynamic persistence, operating through trade openness and industrial restructuring, and with marked regional heterogeneity.

The findings demonstrate that the BRI, as a national public policy, exerts meaningful economic-geography effects by narrowing inter-city income disparities within provinces. The DID and dynamic-DID results directly address the primary research question, showing that treated provinces experience greater convergence after 2015. The IV and multiple robustness checks bolster causal interpretation. Mechanism analyses confirm that BRI’s impact on disparity operates through enhanced trade openness (e.g., facilitation, connectivity, FDI inflows) and industrial structural transformation (greater tertiary activity and upgrading), aligning with theoretical expectations from Stolper-Samuelson, spatial economics, and structural change literature. The stronger effects in Central and West China reflect both policy targeting and higher marginal returns to new connectivity and openness in less developed regions, while East China’s saturation with prior policies yields attenuated marginal effects. The results contribute to the public policy geography framework by evidencing how a globalization-oriented policy reshapes intra-provincial economic distributions, and reconcile part of the prior mixed literature by focusing on within-country, inter-city disparities rather than international or aggregate national outcomes.

This paper provides new empirical evidence that China’s BRI promotes convergence of inter-city income disparities within provinces. Using a DID framework on 269 cities (2002–2017), supported by IV, triple-difference, and PSM checks, the study shows significant and dynamic disparity-reducing effects. Mechanism tests indicate that increased trade openness and industrial structural transformation are key channels. The effects are heterogeneous, being significant and larger in Central and West China but insignificant in the East. Contributions include: (1) integrating BRI into a public policy geography lens to explain domestic spatial income dynamics; (2) documenting causal, dynamic convergence effects; (3) identifying mechanisms; and (4) detailing regional heterogeneity. Future research should extend to cross-country analyses of BRI’s distributional effects and examine post-2017 developments, including the impacts of COVID-19 and evolving BRI modalities. Policy recommendations include aligning BRI with coordinated regional development strategies, enhancing interprovincial connectivity to diffuse benefits to non-treated areas, promoting balanced trade openness across cities, and leveraging international cooperation for industrial upgrading, with special attention to improving inclusion in Eastern provinces.

- External validity: The analysis focuses on Chinese prefecture-level cities; results may not generalize to other BRI countries or to different administrative scales. - Time horizon: Data end in 2017; longer-run effects and post-2017 policy adjustments (including COVID-19 impacts on BRI) are not captured. - Measurement constraints: Trade openness proxied by FDI/GDP due to limited city-level trade data; alternative openness measures could refine mechanism estimates. - Policy selection: Although IV and robustness checks mitigate endogeneity, residual selection concerns may remain regarding policy targeting and concurrent policies.