pith. sign in

arxiv: 2606.26470 · v2 · pith:QJDFGZAPnew · submitted 2026-06-25 · 💰 econ.GN · q-fin.EC

The Growing Self-Reliance of Chinese Innovation

ZIyu Chen , Christopher Esposito This is my paper

Pith reviewed 2026-06-30 10:27 UTC · model grok-4.3

classification 💰 econ.GN q-fin.EC
keywords Chinese patentsscientific citationsinnovation dependenceUS-China technologypatent analysisdomestic scienceresearch attribution
0
0 comments X

The pith

Chinese patents now draw a larger share of their underlying science from China than from the United States.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that the fraction of China-produced scientific papers cited by Chinese patents grew from 1% in 2000 to 26% in 2025. This rise caused the Chinese share to surpass the U.S. share in 2021. The result challenges the premise behind U.S. policies that restrict Chinese access to American research. A sympathetic reader would care because it indicates that China's innovation system is becoming less dependent on external science inputs.

Core claim

Linking the full corpus of Chinese invention patents to the global scientific literature shows that the share of China-produced science behind Chinese patents rose from 1% in 2000 to 26% in 2025, overtaking the U.S. share in 2021. As China's reliance on U.S.-produced science fades, policies restricting access fall out of alignment with the U.S.' actual strategic position.

What carries the argument

The linkage of Chinese invention patents to the scientific papers they cite, with attribution of each paper to the country that produced it.

If this is right

  • U.S. policies that limit Chinese access to American science are now misaligned with actual dependence levels.
  • China's technological capabilities are increasingly supported by its own domestic research output.
  • The strategic leverage once provided by controlling access to U.S. science has diminished for influencing Chinese patenting.
  • Other countries may observe similar shifts away from reliance on any single foreign science base.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Continued growth at the observed rate would make Chinese patents almost entirely independent of U.S. science within another decade.
  • The same linkage method could be applied to patents from other nations to map changing global science-dependence networks.
  • Domestic science investment may generate self-reinforcing effects on patent quality and volume that compound over time.

Load-bearing premise

The method of linking patents to scientific literature and attributing papers to countries accurately captures true dependence without systematic biases from citation practices, multi-author papers, or classification rules.

What would settle it

Re-running the patent-to-paper matching and country attribution with an alternative algorithm or dataset that produces a flat or rising U.S. share after 2021 would falsify the central trend.

read the original abstract

U.S. policy increasingly seeks to slow China's technological rise by restricting its access to American science, on the assumption that Chinese innovation depends on U.S. science. Linking the full corpus of Chinese invention patents to the global scientific literature, we show that this dependence has fallen in recent years: the share of the China-produced science behind Chinese patents rose from 1% in 2000 to 26% in 2025, overtaking the U.S. share in 2021. As China's reliance on U.S.-produced science fades, policies restricting access fall out of alignment with the U.S.' actual strategic position.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 0 minor

Summary. The paper links the full corpus of Chinese invention patents to the global scientific literature and reports that the share of China-produced science cited by these patents rose from 1% in 2000 to 26% in 2025, overtaking the U.S. share in 2021. It concludes that Chinese innovation has become less dependent on U.S. science, implying that U.S. policies restricting access to American science are misaligned with current realities.

Significance. If the measurement is robust, the result would carry substantial policy significance by providing quantitative evidence against the premise that Chinese technological progress remains heavily reliant on U.S. scientific output. It offers a direct empirical test of self-reliance trends in innovation systems.

major comments (2)
  1. [Abstract] Abstract: the headline result (China share rising from 1% to 26%, overtaking the U.S. in 2021) is computed from the fraction of cited papers classified as China-produced, yet the abstract supplies no information on the patent-to-paper matching algorithm, the corpus used, or the country-attribution rule (whole vs. fractional counting for international collaborations). This information is load-bearing for validating the claimed trend.
  2. [Abstract] Abstract: without details on handling of non-English papers, recent preprints, or journal coverage in the linkage step, it is impossible to assess whether the measured China share is inflated relative to the U.S. share due to systematic differences in citation practices or classification.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful review and for identifying areas where the abstract requires greater methodological transparency. We agree that these details are important for readers to evaluate the headline result and will revise the abstract accordingly. We respond to each comment below.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the headline result (China share rising from 1% to 26%, overtaking the U.S. in 2021) is computed from the fraction of cited papers classified as China-produced, yet the abstract supplies no information on the patent-to-paper matching algorithm, the corpus used, or the country-attribution rule (whole vs. fractional counting for international collaborations). This information is load-bearing for validating the claimed trend.

    Authors: We agree that the abstract should briefly convey these elements. The manuscript (Section 3) details a title-abstract-author fuzzy matching procedure applied to the full corpus of Chinese invention patents linked against Web of Science, Scopus, and Crossref records. Country attribution uses fractional counting for internationally co-authored papers. In the revised abstract we will add one sentence summarizing the linkage approach and confirming fractional counting. revision: yes

  2. Referee: [Abstract] Abstract: without details on handling of non-English papers, recent preprints, or journal coverage in the linkage step, it is impossible to assess whether the measured China share is inflated relative to the U.S. share due to systematic differences in citation practices or classification.

    Authors: The referee correctly notes that these choices could affect relative shares. Our linkage incorporates non-English records via multilingual indexing in the source databases and includes cited preprints from major repositories. Journal coverage follows the union of the three databases. To permit direct assessment, the revised abstract will note these inclusions, and we will add a short robustness subsection comparing results with and without non-English and preprint citations. revision: yes

Circularity Check

0 steps flagged

No circularity: result is direct empirical measurement

full rationale

The paper computes its headline share (China-produced science cited by Chinese patents rising from 1% to 26%) by linking patents to the literature corpus and applying country attribution rules. This is a data-processing pipeline with no equations, fitted parameters renamed as predictions, or self-citation chains that reduce the output to its own inputs. The derivation chain is therefore self-contained against external benchmarks and receives the default non-finding.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Based solely on the abstract; no free parameters, axioms, or invented entities are stated or implied.

pith-pipeline@v0.9.1-grok · 5621 in / 1091 out tokens · 49531 ms · 2026-06-30T10:27:50.877110+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

28 extracted references · 1 canonical work pages

  1. [1]

    Chatterji, F

    A. Chatterji, F. Murray, How Geopolitics Is Changing the Economics of Innovation, inEn- trepreneurship and Innovation Policy and the Economy(University of Chicago Press), vol. 5 ofNBER Book Series(2025)

    2025

  2. [2]

    Kwon, Dual-Use Research under Scrutiny.Science392(6802), 1032–1035 (2026), doi:10

    S. Kwon, Dual-Use Research under Scrutiny.Science392(6802), 1032–1035 (2026), doi:10. 1126/science.aee2479,https://www.science.org/doi/10.1126/science.aee2479

    work page doi:10.1126/science.aee2479 2026

  3. [3]

    Congress, Securing Innovation and Research from Adversaries Act, S

    U.S. Congress, Securing Innovation and Research from Adversaries Act, S. 4525, 119th Congress (2025–2026) (2026),https://www.congress.gov/bill/119th-congress/ senate-bill/4525, introduced by Sen. J. Banks and Rep. J. Moolenaar, May 2026

    2025

  4. [4]

    J. Mervis, Lawmakers Propose Banning All U.S.–Chinese Research Collabora- tions, Science (News) (2026),https://www.science.org/content/article/ lawmakers-propose-banning-all-u-s-chinese-research-collaborations, accessed: 2026-05-28

    2026

  5. [5]

    Y. Hu,et al., Layered copper-containing oxide material and preparation method and use thereof (2017),https://patents.google.com/patent/CN104795551B/en, priority date 2014-07-17; granted 2017-07-14

    2017

  6. [6]

    J. Wang, X. Li, Y. Wang, Chinese firms are increasingly leveraging science for innovation, and geopolitical tensions have accelerated the process.Working Paper(2025)

    2025

  7. [7]

    M. Marx, A. Fuegi, Reliance on Science: Worldwide Front-Page Patent Citations to Scientific Articles.202041(9), 1572–1594 (2020)

    2020

  8. [8]

    S. Arts, N. Melluso, R. Veugelers, Beyond Citations: Measuring Novel Scientific Ideas and their Impact in Publication Text.Review of Economics and Statistics(Forthcoming), 1–33 (2025)

    2025

  9. [9]

    S. Arts, N. Melluso, Hidden Reliance on Scientific Ideas and Firm Innovation.Working Paper (2026). 12

    2026

  10. [10]

    Congress, CHIPS and Science Act of 2022, Public Law No

    U.S. Congress, CHIPS and Science Act of 2022, Public Law No. 117–167, 136 Stat. 1366 (2022),https://www.congress.gov/bill/117th-congress/house-bill/ 4346, h.R. 4346, 117th Congress

    2022

  11. [11]

    R. Wu, C. Esposito, J. Evans, Shifting power asymmetries in scientific teams reveal China’s rising leadership in global science.Proceedings of the National Academy of Science122(44) (2025)

    2025

  12. [12]

    AlShebli, S

    B. AlShebli, S. Ali Memon, J. Evans, T. Rahwan, China and the U.S. Produce more Impactful AI Research when Collaborating Together.arXiv(2023),https://arxiv.org/abs/2304. 11123

    2023

  13. [13]

    Esposito, Global Science Sustains U.S

    C. Esposito, Global Science Sustains U.S. Innovation.Working Paper(2026)

    2026

  14. [14]

    C. S. Wagner, China Surpasses the US in Research Spend- ing: The Consequences Extend Far Beyond Scientific Ranking and Clout, The Conversation (2026),https://theconversation.com/ china-surpasses-us-in-research-spending-the-consequences-extend-far-beyond-scientific-ranking-and-clout-280543, reports OECD data showing China’s R&D spending reached parity with...

    2026

  15. [15]

    Fleming, O

    L. Fleming, O. Sorenson, Technology as a complex adaptive system: evidence from patent data.Research Policy30, 1019–1039 (2001). 13 Supplementary Materials for The Growing Self-Reliance of Chinese Innovation Ziyu Chen, Christopher Esposito∗ ∗Corresponding author. Email: christopher.esposito@anderson.ucla.edu This PDF file includes: Materials and Methods S...

    2001

  16. [16]

    bandwidth

    Therefore, phrases in the abstract and titles appearing for the first time after 1900 are considered to be scientific phrases and may appear in our dataset. Second, for each scientific phrase in each patent, we identified all OpenAlex papers containing the same phrase in their title or abstract and published at least one year before the patent’s applicati...

    1900

  17. [17]

    Advanced Communications—Keywords:advanced communications technology, communications technology, impacts to communications from long term changes in atmospheric condition, wireless communication, optical communication, 5G, internet of things, satellite communication, network protocol, digital signal processing, telecommunication system, mobile network, bro...

  18. [18]

    Artificial Intelligence—Keywords:zero shot learning, artificial general intelligence, large language model, transfer learning, foundation model, deep reinforcement learning, human in the loop deep reinforcement learning, generative adversarial network, neural network, generative artificial intelligence, convolutional neural network, recurrent neural netwo...

  19. [19]

    Biotechnology—Keywords:biotechnology, bioeconomy, medical technology, biomanufacturing, genomics, bioremediation, synthetic biology, genetic sequencing, biopharmaceutical, biomechanics, tissue engineering, biomaterial, regenerative medicine, biomedical imaging, medical device, drug delivery, genetic engineering, bioinformatics, molecular biology, cell the...

  20. [20]

    Data Management and Security—Keywords:data management, data governance, data quality, database system, data security, data warehousing, data integration, metadata management, information lifecycle management, data privacy, data architecture, master data management, data lake, data virtualization, NoSQL database, SQL, data cleaning, data migration, ETL, ex...

  21. [21]

    Disaster Resilience—Keywords:anthropogenic disaster prevention, climate resilience and adaptation, anthropogenic disaster mitigation, climate resiliency, natural disaster prevention, climate adaptation, natural disaster mitigation, disaster prevention, disaster mitigation, climate change, environmental degradation, industrial accident, pollution, deforest...

  22. [22]

    Energy—Keywords:energy efficiency technology, battery management control, industrial efficiency technology, battery system integration, batteries, nuclear technology, battery management system, renewable energy, solar power, wind energy, energy storage, smart grid, energy efficiency, bioenergy, photovoltaic, battery technology, nuclear energy, hydroelectr...

  23. [23]

    High Performance Computing—Keywords:high performance computing, parallel processing, high throughput computing, petascale grid computing, cluster computing, floating point operations per second, supercomputer, distributed computing, supercomputing, multi-core processing, load balancing, GPU computing, SIMD, MIMD, MPI, message passing interface, OpenMP, CU...

  24. [24]

    Materials Science—Keywords:materials science, composite 2D material, next generation material, sustainable material, carbon footprint, biodegradable material, nanomaterial, biomaterial, energy harvesting, organic electronics, lightweight material, photovoltaic, green building material, bio-based polymer, polymer, composite, metallurgy, ceramic, supercondu...

  25. [25]

    Quantum Tech—Keywords:quantum information science, quantum computer, quantum sensor, quantum technologies, quantum mechanics, quantum entanglement, quantum algorithm, quantum bit, qubit, superposition, quantum computing, quantum cryptography, quantum communication, quantum error correction, quantum key distribution, quantum teleportation, quantum gate, qu...

  26. [26]

    Robotics and Advanced Manufacturing—Keywords:manufacturing processes, lean manufacturing, additive manufacturing, industrial engineering, supply chain management, quality control, production planning, mass production, computer aided manufacturing, CAM, computer numerical control, CNC, robotics, 3D printing, process optimization, materials science, industr...

  27. [27]

    Semiconductors—Keywords:semiconductor heterostructure, silicon carbide, organic semiconductor, photodetector, MEMS, microelectromechanical system, VLSI, very large scale integration, semiconductor packaging, electronic band structure, compound semiconductor, semiconductor metrology, charge carrier transport, sputtering, thermal oxidation, electron beam li...

  28. [28]

    huawei”, “hisilicon

    Not Applicable— Use this when the IPC code does not clearly fit any CTA above. Classification rules: •Choose the single best-fitting CTA based on the IPC code title. •Be conservative: if the code is generic or does not clearly match, use ‘Not Applicable’. •Watch for false positives in Quantum Tech, especially optical communications, nanomaterials, and qua...