Planning for science: China's "grand experiment" and global implications

The paper examines China's National Medium and Long-Term Plan for the Development of Science and Technology (2006–2020) as a "grand experiment" in state-led S&T planning. It seeks to assess progress toward goals of becoming an innovation-oriented nation and global scientific power, identify persistent structural and governance challenges, and discuss implications for China and the international community at the juncture between the completed 2006–2020 MLP and the forthcoming 2021–2035 MLP. The study is situated in the broader context of China's top-down S&T governance, international competition/cooperation, and evolving policy priorities such as indigenous innovation and S&T self-reliance.

This is a policy and systems analysis based on review of official plans (MLP 2006–2020; 11th–13th Five-Year Plans; outline of the 14th Five-Year Plan and 2035 long-range prospects), program documents, reforms of national S&T programs, and secondary sources. The authors synthesize quantitative indicators and benchmarks (e.g., GERD/GDP, contribution of S&T progress, dependence on foreign technology, triadic patent counts, citation impact, Global Innovation Index rankings) and draw on case evidence from mega-engineering/-science programs, funding structures, and talent policies (e.g., Thousand Talents Program). They evaluate target fulfillment, structural allocation of R&D (basic/applied/development), governance arrangements, and international dynamics to infer implications and likely directions of the new MLP.

• Fulfillment of MLP quantitative targets was mixed but largely positive by 2020: GERD/GDP reached 2.40% (target 2.5%); contribution of S&T progress (STP) hit 59.5% in 2019 (target 60%); dependence on foreign technology (DFT) declined to 31.2% in 2016 (target ≤30% before metric was discontinued); China rose to 3rd globally in triadic patents (5,323 in 2018, up from 524 in 2005) and 2nd in citations to international scientific papers (2018).
• China’s innovation standing improved in global benchmarks, e.g., GII rank improved from 29th (2007) to 14th (2020).
• Structural imbalance in R&D persists: basic research share hovered near 5% for decades, reaching 6% in 2019 (vs. U.K. 18.6% and U.S. 16.6% in 2018); applied research was low and declining (11.3% in China 2019 vs. U.S. 19.15% in 2018), with a heavy skew toward experimental development. Government share of GERD fell from about 40% (2000) to ~20% (2019), constraining research-intensive universities and public institutes.
• Incentive distortions and data integrity concerns: enterprise-reported R&D and performance indicators may be inflated to secure incentives; patenting surges may reflect subsidy-driven behavior rather than true innovation.
• Mega-Engineering Programs (MEPs) and Mega-Science Programs propelled sectoral development (e.g., large aircraft, manned space and lunar exploration, high-definition Earth observation) and overlapped with Strategic Emerging Industries, yet outcomes were uneven. Programs with clear public-sector end-users and specific goals fared better; others (e.g., ICs, core electronics/software) lagged amid continued foreign dependence and external constraints (e.g., U.S. export controls).
• Reform of national S&T programs (2014) consolidated schemes into five types, with the National Key R&D Programs integrating 973/863 and other funds to improve coordination and reduce overlap. Nonetheless, challenges remain: short project cycles bias toward applied outcomes; unclear differentiation of program functions; lingering issues of accountability and efficiency.
• Talent policies (e.g., Thousand Talents Program) boosted output and capacities in host institutions but faced limited full-time return rates, rising international scrutiny, and geopolitical headwinds that constrain cross-border talent flows, risking setbacks in technological learning.
• Governance insights: state-led planning enables resource mobilization but risks Goodhart’s law, market distortions, and inefficiencies. Macro-level coordination across ministries remains imperfect despite reforms (e.g., IMJC; MOST-NSFC/SAFEA changes). Examples such as the He Jiankui case highlight needs for stronger research ethics governance by the scientific community.
• External dynamics: escalating U.S.–China tensions create risks of technological and talent decoupling; despite increased IP receipts, China’s IP payments rose substantially (US$6.63B to US$37.78B, 2006–2020), underscoring continued reliance on foreign technologies even as DFT metrics improved.

The findings suggest that while China met or nearly met most MLP targets and climbed global innovation rankings, quantitative goal attainment does not equate to achieving a genuinely innovation-oriented system. The structural tilt toward development over basic/applied research, declining government funding share for scientific research, and incentive distortions impede long-term scientific capability. Mission-oriented mega-programs are effective when goals are specific and public-sector driven but less so in market-driven or exploratory domains, especially where foreign dependencies persist (e.g., semiconductors). The reform of national S&T programs improved coordination but has not fully realigned funding modalities to support the full pipeline from discovery to application without sacrificing basic research. Governance must balance the powerful role of the state with market mechanisms and autonomy of the research community; over-intervention can depress total factor productivity and misalign incentives, while underdeveloped ethics and advisory systems can permit problematic practices. Internationally, planning as a governance tool has utility but should avoid perverse target-driven behaviors (Goodhart’s law) and accommodate openness; geopolitical frictions necessitate strategies that maintain global collaboration while building resilient domestic capacities.

The paper contributes a system-level appraisal of China’s 2006–2020 MLP, highlighting substantial advances alongside structural and governance shortcomings, and outlines directions for the forthcoming 2021–2035 MLP. The new plan is expected to prioritize: (1) S&T self-reliance and self-improvement, especially in key and core technologies, while remaining open to global collaboration; (2) strengthening strategic S&T forces by increasing basic research funding (targeting 8% of GERD), clarifying roles of universities, research institutes, and enterprises, and establishing mission-oriented national laboratories in priority areas (e.g., quantum information, micro-nano electronics, AI, biomedicine, energy); (3) optimizing the S&TI system structure with better balance between government and market, emphasizing enterprise-centered, market-oriented innovation and deeper integration across actors; and (4) improving the innovation ecosystem, potentially by creating more specialized, relatively independent funding agencies and enhancing research ethics and governance. Future research should monitor implementation of these reforms, evaluate the effectiveness of new national laboratories and funding architectures, and assess how geopolitical dynamics shape China’s integration into global innovation networks.

• Comprehensive assessment of the MLP’s overall outcomes is inherently difficult given the plan’s scope and limited transparency; many mega-programs lack publicly available, trackable progress metrics, especially those with broad or exploratory goals (e.g., drug innovation; major disease control).
• Reliance on aggregate indicators (e.g., GERD/GDP, DFT before discontinuation, patent counts, citation ranks) may be susceptible to Goodhart’s law and may not capture innovation quality or industrial impact.
• Enterprise-reported R&D and performance data may include inaccuracies due to incentive structures, complicating precise measurement of funding shares and outcomes.
• The analysis synthesizes policy documents and secondary sources; causal attribution between specific policies and outcomes is limited, and some conclusions are inferential given data constraints.