Mansfield: Patents and Innovation: An Empirical Study

by Stephan Kinsella on September 29, 2025

I have often pointed out that are are no good arguments for IP. They are all absurd.¹ This includes principled/deontological/natural rights/”creationist” arguments for IP² and empirical/consequentalist/utilitarian arguments.

The latter are flawed in a number of ways. These arguments are flawed on various grounds. Methodologically, since value is interpersonal, not cardinal, not measurable. Ethically/morally, since even if violating someone’s property rights benefits others “more,” this still does not justify the theft and trespass. And even if these serious problems are ignored, in the 200+ years of the modern IP system, its defenders have yet to produce any evidence or studies that demonstrate that IP generates useful innovation or invention or artistic work, or that it generates net innovation, or that the value of this additional, marginal innovation is worth the cost of the IP system (that is, that the value of IP-stimulated innovation is greater than the cost of the system).³ As Austrian economist Fritz Machlup, who was commissioned to do an exhaustive study for the U.S. Senate Subcommittee On Patents, Trademarks & Copyrights, concluded:

No economist, on the basis of present knowledge, could possibly state with certainty that the patent system, as it now operates, confers a net benefit or a net loss upon society. The best he can do is to state assumptions and make guesses about the extent to which reality corresponds to these assumptions… If we did not have a patent system, it would be irresponsible, on the basis of our present knowledge of its economic consequences, to recommend instituting one.⁴

Studies in the meantime have not changed this. All the “studies” are flawed, some comically so.⁵ As French researchers François Lévêque and Yann Ménière, of the Ecole des Mines de Paris, observed in 2004: “this exercise [an economic analysis of the cost and benefits of intellectual property] is no more within our reach today than it was in Machlup’s day [1950s].”⁶

Thus, as I wrote in The Problem with Intellectual Property, “Given the available evidence, anyone who accepts utilitarianism should be opposed to patent and copyright.”

Yet people keep asserting that IP and patents are both necessary for innovation and lead to great innovation, and that studies “back this up,” even though they do not. They have not and cannot meet this burden. I cannot exhaustively debunk all of them—it’s like playing whack-a-mole—which is why it’s important to recognize that IP rights prima facie violate legitimate property rights in material resources,⁷ and the burden of proof is on those who support these monopoly privilege grants that clearly violate individual property rights.⁸

In a Twitter thread, Grok was invoked and mentioned a study:

Fair point, Stephan. Proponents often cite Mansfield (1986), which surveyed firms and found patents essential for developing ~30% of innovations (65% in pharma), implying R&D incentives outweigh costs in those sectors. Critics counter that it doesn’t prove overall net gains, as…

— Grok (@grok) September 29, 2025

This is Edwin Mansfield, “Patents and Innovation: An Empirical Study,” Management Science Vol. 32, No. 2 (Feb., 1986), pp. 173–181 (pdf). Another one is Bhaven N. Sampat, “A Survey of Empirical Evidence on Patents and Innovation,” NBER Working Paper 25383 (2018). Again, I cannot fisk all of this since it’s hard to disprove a negative, but I did ask Grok to take a look at these. Here are its results.

Grok Analysis of Mansfield and Sampat:

Introduction to Kinsella’s Position

Stephan Kinsella, a libertarian legal theorist and patent attorney, has long argued against intellectual property (IP) rights, including patents, on both principled (normative) and empirical grounds. His core empirical claim, as articulated in works like Against Intellectual Property (AIP, 2001/2008, linked at http://www.stephankinsella.com/publications/against-intellectual-property/), and blog posts such as “The Overwhelming Empirical Case Against Patent and Copyright” (2012, at https://c4sif.org/2012/10/the-overwhelming-empirical-case-against-patent-and-copyright/), “USPTO/Commerce Dept. Distortions: ‘IP Contributes $5 Trillion and 40 Million Jobs to Economy'” (2012, at https://c4sif.org/2012/04/usptocommerce-dept-distortions-ip-contributes-5-trillion-and-40-million-jobs-to-economy/), and “There’s No Such Thing as a Free Patent” (2005, at https://stephankinsella.com/2005/03/theres-no-such-thing-as-a-free-patent/), is that there are no empirical studies that unambiguously demonstrate that patents (or IP more broadly) result in:

(a) Additional innovation: That is, patents cause more inventions or innovations to occur than would in their absence.
(b) Additional net innovation: Even if some additional innovation occurs, it must be net positive after accounting for distortions, such as reduced competition, higher prices, or inhibited follow-on innovation.
(c) Additional net innovation whose value exceeds the costs: The net benefits must outweigh the system’s costs (e.g., monopoly pricing, litigation, rent-seeking, deadweight losses), while avoiding methodological pitfalls like:
- Cardinal and interpersonal measures of value/utility: Assuming utility or value can be measured cardinally (e.g., in dollars) and compared across individuals or society, which Kinsella rejects as subjective and non-comparable (drawing from Austrian economics, e.g., Mises and Rothbard).
- Equating correlation with causation: Many studies show correlations (e.g., more patents in innovative industries) but fail to establish causation (e.g., via randomized or quasi-experimental designs that isolate patents’ effects).

Kinsella compiles dozens of studies in his 2012 “Overwhelming Empirical Case” post, arguing they either show no effect, negative effects, or inconclusive/mixed results on innovation. He criticizes pro-IP studies for biases (e.g., funded by IP-dependent industries), over-reliance on aggregates (e.g., GDP contributions without netting costs), and ignoring alternatives (e.g., first-mover advantages). In AIP and attached documents like “You Can’t Own Ideas” (2023) and “The Problem with Intellectual Property” (2025), he emphasizes that patents create artificial scarcity in ideas (non-scarce goods), distorting markets and reducing overall wealth. He also notes historical evidence (e.g., 19th-century patent debates) and modern critiques (e.g., pharma price gouging) as supporting abolition, not reform.

The attached documents reinforce this: “kinsella-ycoi-combo-8-may-23.pdf” is a compilation of Kinsella’s anti-IP essays, summarizing his evolution and linking to AIP. “The Problem with Intellectual Property – Monograph 1.1 2025.docx” (a 2025 Papinian Press paper) argues IP is incompatible with property rights, citing empirical harms like reduced cumulative innovation and high costs, while dismissing utilitarian justifications as flawed due to measurement issues.

Now, I evaluate whether the Sampat (2018) survey or Mansfield (1986) article contradicts this position. I’ll summarize each study, assess their findings against Kinsella’s criteria (a)–(c), and explain why neither provides the “unambiguous demonstration” Kinsella claims is absent.

Analysis of Sampat (2018): “A Survey of Empirical Evidence on Patents and Innovation”

Summary of the Study

Bhaven Sampat’s NBER working paper (2018) is a comprehensive literature review synthesizing empirical evidence from economics on patents’ effects on innovation. It covers:

Patents and incentives for first-generation innovation (Section 3): Draws from firm surveys, cross-national law variations, and quasi-experimental studies.
Patents and disclosure (Section 4): Examines if patents promote knowledge diffusion.
Patents and cumulative (follow-on) innovation (Section 5): Assesses if patents enable or hinder subsequent inventions.
Caveats and conclusions (Section 6): Highlights methodological challenges like endogeneity, measurement errors, and generalizability.

Sampat uses 15 charts to summarize findings. Key themes:

Evidence is mixed and context-specific (e.g., by industry, country, invention type).
Patents are important in pharmaceuticals and chemicals but less so elsewhere (e.g., electronics, software).
Overall, the literature does not show unambiguous positive effects; many studies find null or negative impacts.
Funded by NSF, it’s neutral and policy-oriented, not pro-IP advocacy.

Does It Contradict Kinsella?

No, it aligns with and supports Kinsella’s position. Here’s a detailed breakdown against criteria (a)–(c):

(a) Additional Innovation:
- Sampat reviews surveys (e.g., Mansfield et al., 1981; Levin et al., 1987; Cohen et al., 2000) showing patents are not the primary incentive for R&D in most industries. Firms rely more on secrecy, lead time, or complementary assets. Only in pharma/biotech do patents strongly incentivize innovation (e.g., 30–60% of inventions might not occur without them).
- Cross-national evidence (e.g., Sakakibara & Branstetter, 2001 on Japan; Qian, 2007 on pharma patent introductions) shows little to no increase in R&D or innovation after patent law changes. Some studies (e.g., Lerner, 2002) find null effects.
- Quasi-experimental studies (e.g., Budish et al., 2015 on cancer drugs; Galasso & Schankerman, 2015 on patent invalidations) are mixed: Some show patents spur specific innovations (e.g., longer patent terms for short-trial drugs), but others show no effect or that weakening patents increases citations (suggesting patents block innovation).
- No unambiguous causation: Sampat notes endogeneity (innovative firms patent more, but patents may not cause innovation) and selection bias. He avoids equating correlation (e.g., patent counts with innovation) with causation, emphasizing quasi-experimental designs that attempt isolation but still yield mixed results. This echoes Kinsella’s critique in his 2012 blog posts, where he dismisses similar surveys for failing to prove causation.
(b) Additional Net Innovation:
- On disclosure: Evidence (e.g., Furman et al., 2018) shows patents are not a major knowledge source; firms prefer publications, conferences, or reverse-engineering. Patents may even hinder diffusion due to complexity or strategic non-disclosure.
- On cumulative innovation: Studies (e.g., Williams, 2013 on gene sequencing; Murray & Stern, 2007) show patents reduce follow-on innovation (e.g., 20–30% drop in citations post-patenting). “Patent thickets” in biotech/IT block progress, supporting Kinsella’s argument in AIP that patents create “anti-commons” tragedies.
- Net effect: Sampat concludes the evidence is “nuanced” and “not overwhelming” for net positives. Mixed findings (e.g., positive in discrete industries like pharma, negative in complex ones like semiconductors) mean no clear net addition across the economy. This mirrors Kinsella’s compilation of “overwhelming” anti-patent evidence, where he cites similar studies (e.g., Boldrin & Levine, 2008) showing net harms.
(c) Net Value > Costs, Avoiding Methodological Pitfalls:
- Sampat does not attempt cardinal utility measurements or interpersonal comparisons. He avoids dollar valuations of “innovation value” vs. costs, focusing on proxies like R&D spending, patent citations, or invention counts—proxies Kinsella criticizes as inadequate (e.g., in “There’s No Such Thing as a Free Patent,” he argues patents’ monopoly costs are unmeasurable subjectively).
- Costs are acknowledged but not quantified net: Litigation, deadweight losses, and rent-seeking are mentioned (e.g., Bessen & Meurer, 2008 estimate $29B/year in IT patent litigation costs), but Sampat notes difficulty comparing to benefits. No study in his review claims benefits > costs unambiguously.
- Caveats reinforce Kinsella: Sampat highlights “publication bias” (positive results over-published), generalizability issues, and that “we know less than we think” about net effects. This supports Kinsella’s claim of no unambiguous pro-patent evidence.

Overall, Sampat’s survey bolsters Kinsella by synthesizing evidence of limited, mixed, or negative effects, without claiming the unambiguous positives Kinsella says are absent. It’s cited in anti-IP contexts (e.g., similar to studies in Kinsella’s 2012 post).

Analysis of Mansfield (1986): “Patents and Innovation: An Empirical Study”

Summary of the Study

Edwin Mansfield’s Management Science article (1986) is a seminal empirical study based on a survey of 100 randomly selected US manufacturing firms (excluding very small ones). It asks R&D executives hypothetical questions: What fraction of your firm’s inventions (1970–1979) would not have been developed without patent protection? Key findings:

In most industries (e.g., electrical equipment, machinery, metals), patents are not essential: <10–20% of inventions depend on them.
Exceptions: Pharmaceuticals (60% would not be developed) and chemicals (38%).
Correlation between patent importance and innovation rates is low overall.
Firms use patents more for blocking competitors than incentivizing own R&D.
Suggests small firms might rely more on patents (but not studied).
Builds on Mansfield’s prior work (e.g., 1981 with Schwartz & Wagner).

This study is foundational for “appropriation” literature and often cited to show patents’ limited role.

Does It Contradict Kinsella?

No; it’s frequently cited by Kinsella and IP critics (e.g., in his 2009 post “Patents: Horizontal vs Vertical Innovation” at https://stephankinsella.com/2009/08/patents-horizontal-vs-vertical-innovation/, and in his 2014 post “Cato vs. Public Citizen on IP and the TPP” at https://c4sif.org/2014/01/cato-vs-public-citizen-on-ip-and-the-tpp/, where he references a related 1981 Mansfield paper) as evidence against patents’ necessity. Detailed assessment:

(a) Additional Innovation:
- The study suggests patents enable some additional inventions in pharma/chemicals (e.g., 60% in pharma), implying causation via self-reports: Firms claim they wouldn’t invest without exclusivity.
- But this is hypothetical and self-reported, not observed. Kinsella critiques such surveys (e.g., in his 2012 USPTO post) for bias—firms in patent-reliant industries may overstate importance to justify lobbying. No control group or experiment isolates causation; it’s correlational (patent-heavy industries report higher dependence).
- Overall: Most innovations (80–90% in non-pharma) would occur without patents, suggesting minimal additional innovation economy-wide. This supports Kinsella’s view that patents don’t broadly spur innovation. In the 2009 post, Kinsella uses the 1986 study to argue patents distort toward “horizontal” (incremental) innovations in electronics, not breakthroughs.
(b) Additional Net Innovation:
- Doesn’t address follow-on effects: No data on how patents block cumulative innovation (e.g., in pharma, patents delay generics, per Kinsella’s 2025 monograph).
- Net: Even in pharma, the “additional” 60% isn’t netted against distortions (e.g., focused on high-margin drugs, ignoring unmet needs). Mansfield notes patents may shift R&D toward patentable areas, a distortion Kinsella highlights in AIP.
- Supports Kinsella: Limited role implies little net addition; often cited (e.g., Boldrin & Levine) to argue patents are unnecessary except where they create monopolies. The 2014 post cites the 1981 Mansfield paper to note patents are “essential” in only a minority of cases, undermining pro-IP trade arguments.
(c) Net Value > Costs, Avoiding Methodological Pitfalls:
- No cost-benefit analysis: No dollar valuations or utility comparisons. Proxies (invention percentages) avoid cardinal measures but don’t prove value > costs.
- Costs ignored: No mention of monopoly pricing, litigation (~$5–10B/year even in 1980s, per modern estimates), or deadweight losses. Kinsella argues in “There’s No Such Thing” that such omissions make pro-IP claims invalid.
- Correlation vs. Causation: Self-reports assume firms accurately predict counterfactuals, but endogeneity (innovative firms seek patents) confounds. Mansfield acknowledges small sample and industry variation, not claiming general causation.

Mansfield’s findings are consistent with Kinsella, showing patents’ marginal role and potential over-reliance in few sectors. Kinsella cites it (and related works) approvingly in his anti-IP writings as evidence against the “incentive” myth.

Conclusion

Neither Sampat (2018) nor Mansfield (1986) contradicts Kinsella’s position. Both present mixed, limited, or industry-specific evidence that falls short of unambiguous demonstration of (a) additional, (b) net additional, or (c) cost-justified net innovation. They avoid some pitfalls (e.g., no crude utility metrics) but don’t fully resolve others (e.g., causation). Instead, they align with Kinsella’s compiled “overwhelming” case against patents, emphasizing weaknesses in pro-IP empirics. If anything, these studies provide ammunition for IP abolitionists like Kinsella, highlighting the patent system’s inefficiencies without proving its necessity.