Last month, China’s principal science funder selected 44 young researchers (all under 35) and committed to funding them for up to 15 years with no short-term performance reviews. India should adopt a version of this, taking the sensible bits and leaving out the rest.
China’s process does have some checks. The funding comes in five-year chunks, and only a proportion of the researchers carry on at the end of the five- and ten-year marks. The amounts increase with each funding block; so, most of the funding comes in at the end.
The best data on long-term grants comes from the US where a private foundation, Howard Hughes Medical Institute (HHMI), funds scientists for many years. When you compare the output of HHMI researchers with that of their peers in similar fields, you find that they produce almost double the number of papers in the top 1% by citations. They also have more “failures”, as you would expect when tackling harder problems. HHMI is by no means a hands-off funder. Every few years, an external panel of experts examines each researcher’s five best papers. Approximately one in six lose their funding at the first review and over one in four at the second. The experts look at the work as a whole, not as a tick-box list.
What happens over 15 years with no such reviews at all ought to be studied. China is currently running that experiment.
{{/usCountry}}What happens over 15 years with no such reviews at all ought to be studied. China is currently running that experiment.
{{/usCountry}}Long grants can go wrong in two ways, other than the factor of the scientist(s) involved. The first is what happens when the funding stops. A 2026 study of Japan’s World Premier International Research Centres found that they outperformed their home universities while they were funded. After the funding stopped, the ones dependent on the university went right back down to the level of the average university. The rest, which had other sources of funding, continued to perform. So, a fellowship has to have a plan for exit.
The second thing that matters is the level of interruption. When US labs are delayed in their renewal, they cut spending by about half. In 2022-23, India’s new public R&D funding system — which sent unspent funds back to the government on March 31 — meant that many labs couldn’t pay their staff.
A US survey found that longer grants only make scientists bolder if they feel secure about their jobs.
One has to ask two questions to work out which projects are suitable for a long grant. First, how quickly it can be known whether the project is succeeding. And second, what and how much is lost in the case of failure. For the vast majority of projects, a quick answer and little sunk cost is the right model for a standard three-year grant. A quick answer with a lot of sunk cost is the model for a programme with regular milestones (such as Advanced Research Projects Agency-Energy, or ARPA-E, in the US, whose directors have a quarterly review of progress). A slow answer with little sunk cost (maths, long field studies, etc.) is the domain of the long grant. A slow answer with a lot of sunk cost (big instruments etc.) is the type of scenario where you need an institution funded in stages, with a definite end date.
India’s fellowship could last for seven years, be renewable once (after outside experts have read a fellow’s five best pieces of work), and then taper off over a further two years. Seven years is long enough for the first cohort of PhD students to complete their research. India’s longest individual award at the moment is five years.
The funds should be sufficient to run a decent group. For a lab scientist, around ₹9 crore over the first seven years, and around ₹3.5 crore for a theorist, would be adequate. That’s in the ballpark of China’s first chunk of funding (though it’s worth noting that extra money buys progressively less research). The current early-career grant covers about a quarter of a lab’s running costs. Fifty fellows a year would cost about ₹600 crore per year. Grants with a review point out-perform those without, and doubling the funds every five years gives you about half the research per rupee as a flat grant. The release of funds is also important, adding around 15% to every design.
That last point is the most important. The Anusandhan National Research Foundation (ANRF) Act enables the board to establish special funds with their own financial rules. The ₹1,000-crore RDI Fund is an example of this, with its own rules that supersede the general rules. A fellowship fund could follow the same route: One sanction for the entire period, funds advanced to the host annually, unspent funds carried forward, and audits by sampling rather than a certificate being issued prior to every instalment.
The 2025 purchase-rules relaxation allows institute-heads to buy outside the government portal and to sanction large import tenders, but only for equipment and consumables. Computing and cloud services should be included in that.
Another worthwhile thing to do could be to have the selection process combine a panel and a lottery. A German foundation found that lottery winners were published as often as jury-picks, though its grants were small. Shortlisted candidates who are not drawn could form a comparison group.
India must also note that the US’s DARPA is the wrong model to emulate. Their programme managers have very short terms, set the problem, and then cancel anything that doesn’t hit milestones, which is fine for a fast-answer scenario. A fellowship requires a standing board of working scientists with fixed, staggered terms and no serving government secretary; the board must have published rules, too. Other countries have actually enshrined such independence in law. The UK ARIA Act makes it a duty to allow for failure and Germany’s SPRIND may have a proportion of its funds set aside for flexible use. India already has the ANRF Act in place which can be used for this. Powering research needs careful strategising in this age of knowledge-led disruptions, and the country must ensure it follows the right roadmap.
Arindam Goswami is staff research analyst, High-Tech Geopolitics Programme, Takshashila Institution. The views expressed are personal