Do ownership reform pilots accelerate academic technology transfer? Evidence from China's job-related invention scheme

Technical Summary: Ownership Reform Pilots and Academic Technology Transfer in China

Problem and Context
The central policy question addressed is whether academic inventions should be owned by the employing institution or the individual scientist. While the global consensus, largely derived from the U.S. Bayh-Dole model, favors institutional ownership, recent policy debates have suggested that granting ownership to individual scientists (the "professor's privilege") might better incentivize commercialization. Empirical evidence on this trade-off is asymmetric: studies on Norway's 2003 abolition of the professor's privilege show that removing individual rights halves academic entrepreneurship. However, the reverse operation—extending ownership rights to individuals within a system historically built for institutional control—remains largely untested at scale.

China offers a unique natural experiment for this question. Between 2016 and 2020, the country implemented a staggered ownership reform. Initially piloted in Sichuan province and later expanded nationally in October 2020, the scheme designated 40 universities and research institutes to assign ownership shares or long-term use rights (minimum 10 years) to scientists for job-related inventions. The reform aimed to address the "patent-heavy, commercialization-light" paradox in Chinese universities by aligning individual incentives with market outcomes.

Methodology and Data
The study employs a difference-in-differences (DiD) design to evaluate the "intention-to-treat" effects of the designation.

• Data Source: The authors utilize the full population of Chinese patent applications from the China National Intellectual Property Administration (CNIPA) covering 1985 through November 2025.
• Sample: The treatment group for the main specification consists of 16 "Double First-Class" universities designated in the October 2020 national list. These 16 are drawn from the 36 nationally designated institutions remaining after excluding 4 Sichuan early adopters (to ensure clean timing) and 9 other designated universities that do not belong to the "Double First-Class" tier. The control group comprises 88 comparable non-designated "Double First-Class" universities (selected from an initial pool of 90, with 2 dropped due to potential contamination from provincial schemes).
• Outcome Measures: The study constructs four outcome families at the institution-year level:
  1. Patent Output: Total application counts.
  2. Composition: Share of invention patents (the most examination-intensive category).
  3. Institutional Collaboration: University-industry co-application rates.
  4. Individual Behavior (Novel Measure): Firm-side patenting by pre-reform academic inventors. This involves tracking a roster of 586,317 inventors from sample institutions (active 2010–2019) against 9.5 million firm-filed patent applications to detect when academics appear on company patents.
• Identification Strategy: The authors use institution and application-year fixed effects. They address potential biases through event-study designs, propensity-score matching, and permutation placebo tests. Crucially, because the reform confers eligibility rather than automatic transfer (requiring disclosure and agreement), the estimates reflect the policy's eligibility effect rather than the effect of specific rights transfers.

Key Results
The analysis reveals a sharp divergence between institutional and individual responses to the reform:

1. No Institutional Response: The reform did not accelerate university-industry technology transfer through the institutional channel. Co-application rates between designated universities and firms did not increase relative to the control group. The data excludes any increase in co-patenting above two percent of the control mean. Similarly, the composition of filings (invention vs. utility models) remained unchanged.
2. Marginal Individual Response: The reform did shift individual behavior. In the 16 designated research-intensive universities, the rate of firm-side patenting by pre-designation academic inventors rose by 0.32 percentage points (approximately 6% relative to the baseline of 5.7%).
   • This effect is concentrated entirely in the "Double First-Class" universities.
   • The result is robust to stricter name-matching criteria (restricting to uncommon names) and alternative estimation windows.
   • The effect weakens when using propensity-score-matched controls, suggesting the magnitude is sensitive to the control group selection, though the sign remains positive.
3. Asymmetry with Norway: The findings contrast with the Norwegian experience. While withdrawing rights in Norway significantly reduced entrepreneurship, granting rights in China produced only a marginal shift in individual patenting without altering the institutional machinery of technology transfer.

Key Contributions

1. Causal Evidence on Ownership Extension: This paper provides the first large-sample causal evidence on extending ownership toward individual scientists within an institution-owned system. It documents a theoretical asymmetry: removing rights has large negative effects, while granting them in a system lacking complementary infrastructure yields only marginal individual gains and no institutional gains.
2. Novel Measurement of Firm-Side Patenting: The authors develop a scalable method to track academic inventors across the national patent register, linking half a million scientists to millions of firm applications. This approach detects policy responses that aggregate institutional data miss entirely.
3. Resolution of Policy Debate: The study settles a decade-long empirical debate in Chinese policy circles by demonstrating that the co-patenting channel, expected to be the primary driver of reform success, did not move. It quantifies the upper bound of any potential institutional gain, showing it to be negligible.

Significance and Interpretation
The paper argues that ownership rules are not a portable lever that functions identically in both directions. The effect of ownership allocation depends critically on the transfer infrastructure surrounding it.

• Infrastructure Dependence: In Norway, individual rights operated atop functioning channels connecting professors to investors. In China, the new rights were introduced into an environment where valuation services, liability safe harbors, and early-stage capital are still developing, and where institutional veto power remains intact.
• Policy Implication: Reassigning title alone is insufficient to drive institutional technology transfer if the binding constraints lie downstream (e.g., transfer office capacity, evaluation systems rewarding publication over commercialization). However, ownership reform can shift individual behavior at the margin, particularly for scientists in well-equipped institutions who wish to commercialize personally.
• Conclusion: The reform moved individual behavior without changing the institutional machinery. Effective transfer reform requires a "bundle" of ownership rights and complementary infrastructure, rather than a simple reallocation of title. The paper concludes that while ownership reform reached individuals, it stopped there, failing to replicate the systemic effects seen in systems where the complementary machinery was already established.