Somewhere in a national laboratory, a result that could become a company is being written up for a journal and then forgotten. This is not a failure of science; it is the default outcome of a system that rewards publication and provides no path to market. Governments across India, the Gulf and Southeast Asia have spent the past decade building that path — research foundations, proof-of-concept funds, technology-transfer offices, scouting agencies — with results that range from transformative to invisible. This article distils what the evidence and four decades of practice say about the difference, and lays out a program design that research councils and universities can commission.
The idea in brief. Research commercialisation is a funnel with a well-documented shape: for every hundred invention disclosures, a handful become companies, and most value arrives through licences and partnerships rather than spin-outs. The system-level lever is not more patents; it is the institutional machinery — technology-transfer capability, inventor incentives, proof-of-concept funding across the “valley of death”, and surrogate entrepreneurs who can carry a technology the scientist cannot. India’s new national research foundation, the Gulf’s university-anchored innovation hubs and Thailand’s science-agency model are all bets on that machinery. A commercialisation program that works scouts broadly, triages ruthlessly, validates with the market before it patents, and treats team formation as its central product.
The gap, measured
The scale of the opportunity is easiest to see in the inputs. India’s gross expenditure on research and development has hovered around 0.6 to 0.7 per cent of GDP for years, according to the Department of Science and Technology — low by the standards of innovation-intensive economies, but on a base large enough to produce one of the world’s biggest scientific workforces and a rapidly rising patent output. The Indian patent office’s annual grants crossed the one-hundred-thousand mark in the 2023–24 fiscal year, a multiple of the figure a few years earlier, following procedural reforms and a sharp rise in domestic filing. The Anusandhan National Research Foundation, established by an Act of Parliament in 2023 with a stated budget of ₹50,000 crore over five years and an explicit mandate to bring industry into research funding, is the institutional response to the gap between that output and the economy.
The chart makes a point that is often lost in national debates: research intensity and commercialisation capability are different things. Israel and South Korea sit at the top of the intensity table because their systems convert research into firms and exports with unusual efficiency, and that conversion capacity is what justifies the spending politically. A country that raises intensity without building conversion capacity buys more journal articles. A country that builds conversion capacity first creates the demand that justifies raising intensity. The Gulf states, whose research bases are young but whose universities — KAUST, Khalifa University, Qatar’s Education City institutions — were built with commercialisation in their charters, are testing the second path at speed.
What Bayh–Dole did, and what it did not do
Every commercialisation policy conversation eventually reaches the United States’ Bayh–Dole Act of 1980, which allowed universities to own and license inventions arising from federally funded research. Its reputation as the origin of American academic entrepreneurship is partly deserved and partly myth, and the distinction matters for anyone copying it.
David Mowery, Richard Nelson, Bhaven Sampat and Arvids Ziedonis examined the evidence in a landmark 2001 paper in Research Policy. Their conclusion was that university patenting and licensing were already rising before the Act, driven by the growth of biomedical research and the emergence of biotechnology, and that Bayh–Dole accelerated and formalised a trend rather than creating it. The Act’s real contribution was institutional: it gave universities a reason to build technology-transfer offices, standardise licensing, and share royalties with inventors — machinery that the science alone had not produced.
Why some universities then generated far more start-ups than others was the question Dante Di Gregorio and Scott Shane asked in a 2003 Research Policy study. Two of their findings translate directly into policy. Institutional eminence and the willingness of a university to take equity in start-ups were associated with more company formation. Counter-intuitively, giving inventors a larger share of licensing royalties was associated with fewer start-ups — because a generous royalty makes licensing to an existing firm more attractive to the scientist than founding a company. The design of incentives shapes the pathway, not merely the volume.
Donald Siegel, David Waldman and Albert Link’s 2003 analysis in Research Policy examined the transfer offices themselves and found large differences in productivity explained by organisational practices — staffing, incentives, and the business skills of the people in the office. Markus Perkmann and colleagues’ 2013 review in Research Policy added a broader frame: academic engagement with industry (collaborative research, consulting, informal exchange) is far more widespread than formal commercialisation, is driven by different motives, and produces much of the economic value that patents are credited with. A program that counts only patents and spin-outs misses most of the traffic between labs and firms.
The four pathways
Commercialisation programs go wrong when they treat the spin-out as the only respectable destination. There are four, and the right one depends on the technology, the market and the people.
| Pathway | Best when | Value captured by | Typical timescale | Common error |
|---|---|---|---|---|
| Licensing to an existing firm | Technology improves an existing product line; incumbent has distribution | Royalties, milestones; inventor share | 1–3 years to first revenue | Over-valuing early-stage IP; exclusive licences that go unused |
| Spin-out company | Platform technology; no natural acquirer; team available | Equity; jobs; follow-on research | 5–10 years to exit | Scientist as CEO by default |
| Contract research and consulting | Industry needs capability, not a product | Revenue; relationships; future licences | Immediate | Undervalued and unreported |
| Open dissemination / standards | Public-good technology; adoption matters more than rent | Impact; reputation; ecosystem growth | Variable | Treated as failure to commercialise |
The licensing pathway carries the most volume in mature systems. The spin-out pathway carries the most attention and, per venture, the most risk and the most upside. A research council’s portfolio should contain all four, and its metrics should count all four; the temptation to report spin-outs alone because they photograph well is the same input-counting error described in article 32.
The funnel per hundred disclosures
The single most useful expectation-setting device for a minister or vice-chancellor is the shape of the commercialisation funnel. The figures below are approximate ratios drawn from the orders of magnitude reported over many years by the Association of University Technology Managers’ licensing survey of North American institutions; they are illustrative, and individual institutions vary widely.
Several things follow from the shape. A program that hopes for ten companies must expect to process on the order of a few hundred disclosures, which means it must scout far beyond the researchers who volunteer. Most patents will never be licensed, so patent counts are a cost metric, not a success metric — filing should follow market validation, not precede it. Licences outnumber start-ups by a wide margin, so a transfer office that cannot license is failing at its main job, whatever i
