Does the Belt and Road Initiative promote international innovation cooperation?

The study investigates whether the Belt and Road Initiative (BRI) promotes international innovation cooperation between China and partner countries. In a globalized context where talent, knowledge, and technology flow rapidly across borders, international collaboration can enhance innovation quality and enable technological catch-up. The BRI has created platforms and policies to encourage scientific and technological collaboration (e.g., policy frameworks since 2016 and the 2019 Silk Road of Innovation). Despite extensive research on the BRI’s economic effects (e.g., FDI, trade, energy efficiency), little work directly examines its impact on innovation cooperation and the mechanisms involved. This paper focuses on the top 80 innovation-capable countries and tests whether signing BRI cooperation documents serves as a quasi-natural experiment that improves innovation cooperation with China. The study posits that the BRI can: (1) promote innovation cooperation between China and BRI partners; (2) reduce institutional distance; (3) enable cultural distance to positively affect cooperation; and (4) enhance talent exchanges, thereby promoting cooperation.

Two main literatures are reviewed. (1) Socio-economic effects of the BRI: Prior studies find that the BRI increases China’s outward FDI and exports, boosts corporate innovation via competition and capacity utilization, creates jobs, improves city-level economic growth quality, and enhances trade integration and energy efficiency for partner countries. Some negative views highlight potential increases in carbon emissions, outward transfer of excess capacity and high-pollution industries, income inequality, and investment hostility. (2) Determinants of innovation cooperation: Cooperation is more likely when R&D costs and difficulty are high and in supportive external environments (e.g., technology parks). Partner types matter for likelihood and outcomes of collaboration. Institutional distance (differences in laws and governance) and cultural distance (differences in values and norms) significantly affect cooperation and knowledge transfer, often hampering spillovers when distances are large. The gap in existing work is the limited evidence on how a major international initiative like the BRI shapes cross-border innovation cooperation and through which mechanisms. Research hypotheses: H1: The BRI promotes innovation cooperation between China and partner countries. H2: The BRI shortens institutional distance, promoting cooperation. H3: The BRI enables cultural distance to have a positive effect on cooperation. H4: The BRI enhances talent exchange, promoting cooperation.

Research design: The study uses the top 80 countries in global innovation capability (WIPO) as the sample because they account for over 87% of world GDP and over 97% of multinational cooperative invention patents. The period is 2008–2018 (five years before and after the 2013 BRI proposal). China and Hong Kong are excluded; Montenegro, Serbia, and Bulgaria are removed due to missing data, yielding 825 country-year observations. BRI participation is defined by official signing of BRI cooperation documents with China.

Empirical strategy: A staggered difference-in-differences (DID) model estimates the impact of BRI participation on innovation cooperation, exploiting different signing years across countries. The core interaction is Treat × Post, where Treat=1 if a country signed BRI documents; Post=1 in and after the signing year. Country and year fixed effects are included; Treat and Post main effects are collinear with fixed effects and omitted. Standard errors are clustered at the country level; additional tests cluster at the country-year level for event-study checks.

Outcome measures: Innovation cooperation is measured using PCT international patent data (WIPO) on coinvented patents between China and each country (2008–2018). Two relative indicators are constructed: HOMIC = 100 × (co-patents with China / total China patents) and HOSIC = 100 × (co-patents with China / total patents of the partner country). Relative measures help control for country-level patenting scale and capabilities.

Controls: National-level controls include GDP per capita (PGDP), manufacturing value added share (ADV), working-age population share (LABOR), urbanization rate (URBAN), and FDI inflows (OPEN), sourced from the World Bank.

Event-study (parallel trends): Interact Treat with year dummies and take the pre-period as baseline to assess pre-trends. Results show no significant pre-trends for HOSIC and non-positive significant pre-trends for HOMIC, while post-BRI effects for HOMIC are significantly positive, supporting the DID identification.

Additional analyses: Lagged Treat × Post terms account for policy and patenting delays. A triple-differences (DDD) model introduces Income (1 if per-capita GNI ≥ US$3896) to test heterogeneity by economic foundation. Robustness checks include: alternative treatment definition (countries along BRI routes, shock year 2013), dropping 2013 and 2014, excluding countries with zero co-patents throughout, excluding CEE ‘16+1’ countries to remove policy interference, placebo tests with nonparticipants as false treated, and PSM-DID to mitigate selection bias.

Mechanism tests: Institutional distance (SD) is computed from World Governance Indicators as a normalized squared difference across six dimensions; regress SD on Treat × Post. Cultural distance (CD) is measured using the Kogut–Singh index (Hofstede dimensions), interacted with time to create panel variation; regress HOMIC on CD in full sample and separately for BRI partners vs nonpartners. Talent exchange is proxied by the number of Chinese students abroad (Talex1) and foreign students in China (Talex2), regressed on Treat × Post.

Further analysis on innovation capabilities: For BRI partner countries, regress total patents, scientific and technical journal papers, and R&D expenditure-to-GDP ratio on Treat × Post and its lag to test whether BRI improves innovation inputs and outputs.

• Baseline DID: Treat × Post significantly increases HOMIC (coefficient ≈ 0.0338, p<0.01), indicating a higher share of China’s patents that are coinvented with BRI partners; no significant effect on HOSIC (coefficient ≈ -0.0043, n.s.). With lagged interaction, HOMIC remains significantly positive (≈ 0.0263**, lag ≈ 0.0163*), HOSIC remains insignificant.
• Heterogeneity by economic foundation (DDD): Treat × Post × Income is significantly positive for HOMIC, showing stronger promotion where partner countries have higher per-capita income; HOSIC interaction is positive but not significant.
• Parallel trends: Event-study shows non-significant or decreasing pre-trend coefficients for HOMIC and significantly positive post-BRI coefficients; HOSIC shows no significant pre/post effects, supporting identification.
• Robustness: Findings persist under alternative treatment definition (routes, 2013 shock), dropping special years (2013/2014), excluding zero co-patent countries, excluding CEE ‘16+1’, placebo tests (false treated become non-significant; HOMIC turns significantly negative due to potential resource reallocation away from nonpartners), and PSM-DID.
• Mechanisms: • Institutional distance (SD): Treat × Post significantly reduces SD (coefficients negative at 10–5% levels), indicating BRI shortens institutional distance, facilitating cooperation. • Cultural distance (CD): In full sample, CD’s effect on HOMIC is positive but mostly insignificant; in BRI partner subsample, CD becomes significantly positive, suggesting BRI facilitates trust/exchanges so that cultural diversity contributes positively to cooperation. • Talent exchange: Treat × Post significantly increases Talex1 and Talex2 (e.g., 0.516*** and 0.155**), indicating intensified two-way student flows underpinning cooperation.
• Partner countries’ innovation capabilities: For BRI partners, Treat × Post (and/or its lag) significantly increases total patents, scientific and technical papers, and R&D/GDP (e.g., contemporaneous R&D/GDP ≈ 0.144***; lagged patents ≈ 0.261***; papers ≈ 0.129*), showing improved innovation inputs and outputs even though HOSIC share did not significantly rise.

The BRI facilitates international innovation cooperation primarily by reallocating China’s collaboration efforts toward partner countries, increasing the share of China’s patents that are coinvented with them. As the initiative’s originator, China has stronger incentives and policy support to prioritize BRI partners in investment and cooperation, which explains the significant rise in HOMIC. Many partners—being developing or underdeveloped—face constraints in innovation capacity and resources, which likely limits immediate gains in their co-patenting shares (HOSIC). BRI’s policy architecture reduces institutional frictions via coordinated frameworks and agreements, enhances trust and communication across cultural boundaries, and expands talent exchanges; together, these channels lower barriers and enable the benefits of cultural diversity for collaboration. Beyond cooperation shares, the BRI shows positive externalities by raising partner countries’ R&D intensity and scientific outputs (patents and papers), contributing to their longer-term innovation capacity building. Nonetheless, the initiative entails risks including political instability, interest misalignment across countries, economic volatility affecting long-cycle innovation projects, and heterogeneous economic bases that can limit returns and dampen private actors’ incentives. Addressing governance, coordination, and capability gaps is essential to sustain and broaden the cooperation gains.

This paper provides causal evidence that the BRI promotes international innovation cooperation, notably by increasing the share of China’s co-patents with partner countries, with stronger effects for partners with better economic foundations. It identifies three key mechanisms—reduced institutional distance, enhanced talent exchanges, and the positive utilization of cultural differences—that underlie the cooperation gains. While BRI did not significantly raise partners’ co-patent share (HOSIC), it substantially improved their innovation foundations and capabilities, increasing total patents, scientific publications, and R&D intensity. Policy implications include: (1) encouraging enterprises, universities, and research institutes to leverage BRI platforms to deepen collaboration and talent exchanges; (2) maintaining adaptive policy coordination and clear communication to ensure mutual trust and win-win outcomes; and (3) proactively managing political and economic risks and capacity disparities by building long-term cooperation frameworks and supporting infrastructure and capability development. Future research should examine heterogeneous effects across different cooperation forms (e.g., joint R&D institutions, funding programs, coauthored publications) and patent types, and expand coverage beyond the 80-country sample as data become available.

• Measurement scope: Innovation cooperation is measured via coinvented PCT patents; other cooperation forms (joint R&D institutions/funds, coauthored papers beyond what is captured, technology licensing) are not directly included, potentially understating broader cooperation effects.
• Sample coverage: The analysis focuses on the top 80 innovation-capable countries (and excludes some due to data gaps), limiting generalizability to all BRI and non-BRI countries; a global full-sample assessment awaits broader data availability.
• Timing and dynamics: The 2008–2018 window may not capture longer-run effects or later BRI phases; policy lags in patenting and institutional change could lead to understatement of impacts on partner shares.
• Mechanism proxies: Institutional and cultural distance measures and student flows proxy complex mechanisms; unobserved factors could still influence estimated channels despite controls and robustness checks.