The waiting has a price, and economists are beginning to put a number on it. Benjamin Jones and Lawrence Summers, in a 2021 NBER working paper, calculated the social return of all US public R&D investment (every NIH grant, every DARPA programme, every public university laboratory) and arrived at a conservative estimate of $4 spent, with a benefit-to-social gains for every $1 spent, with a benefit-to-cost ratio of 5. When health benefits, inflation bias, and international spillovers are properly accounted for, the social return climbs to over $20 per dollar spent and an internal rate of return of approximately 100 per cent. These are not numbers from a utopian model. They are backward-looking estimates derived from the observed record. India has been spending 0.64 per cent of its GDP on R&D. In 2020-21, that translated to Rs 1,27,381 crore, a figure that sounds large until you note that China spends 2.65 per cent of its GDP on R&D, South Korea 5.32 per cent, and Israel 6.33 per cent. In absolute terms, India’s estimated $15.7 billion of R&D expenditure in 2024 ranks seventh globally; measured as a share of output, India still looks like a country that treats research as a discretionary luxury rather than a compounding asset.
The theoretical case for why this is a mistake is well established. Zvi Griliches argued in 1979 that the social return to research was systematically and substantially larger than the private return (roughly twice as large) because knowledge has a spillover characteristic: it leaks, diffuses, and recombines into ideas that the original investor never intended and cannot capture. Paul Romer formalised the point in his endogenous growth theory. Knowledge is non-rivalrous, meaning it can be used simultaneously by any number of people without being depleted. A protein-folding breakthrough in pharmaceutical research becomes relevant to a materials scientist working on battery chemistry. The stock of knowledge compounds.
The case for public subsidy is not a preference but a mathematical consequence of the structure of the good. The empirical case for developing countries specifically has been made most forcefully by the Brazilian Agricultural Research Corporation, Embrapa. In 1973, Brazil was a food-aid recipient. Its yields for maize, wheat, and rice had been flat for two decades because the Green Revolution varieties (designed for the temperate soils, rainfall patterns, and pest environments of Mexico and the Philippines) were simply ill-suited to the Cerrado savanna. Embrapa was established that year with an explicit mandate to develop seed varieties from scratch, decentralised across every major biome so that researchers worked inside the environments they were studying. By 2010, Brazil had become the world’s third-largest agricultural exporter (and the second-largest by 2035). A working paper by Akerman, Moscrona, Pellegrina, and Sastry puts a number on this. Embrapa accounts for 110 per cent of Brazil’s agricultural TFP growth between 1970 and 2010, with a benefit-to-cost ratio of 17 and an internal rate of return of 25 per cent.
India’s own agricultural research record tells a cognate story, half completed. The Indian Council of Agricultural Research (established in 1929 as the Imperial Council of Agricultural Research, the imperial survival in the name longer than it should in most conversations about Indian governance) has 113 institutes and 71 agricultural universities across the country. A 2024 study found that every rupee invested in ICAR’s agricultural research yields Rs 13.85 in returns. An international assessment put the figure higher still. Every dollar of public research adds approximately $18 to India’s agricultural output in present-value terms. Yet in 2020-21, India spent only 0.54 per cent of its agricultural GDP on research and 0.11 per cent on extension, well below the global average. The benefit-to-cost ratio is 13.85; the funding intensity is 0.54 per cent. This is the arithmetic of a country leaving money on the table.
ISRO makes the same case from a different direction. Between 2014 and 2024, India’s space sector generated an estimated USD 60 billion for the national economy, created 4.7 million jobs, and contributed USD 24 billion in tax revenues. ISRO’s own estimate of its return on investment stands at 2.54 times expenditure, modest by the Jones-Summers standard, but a number from a programme chronically underfunded relative to its peers. The Chandrayaan-3 mission cost Rs 615 crore. (Christopher Nolan’s Interstellar, a film about space, had a production budget of approximately $165 million, roughly Rs 990 crore at 2014 exchange rates.) Between 2015 and 2024, India launched 391 foreign commercial satellites, earning approximately $439 million in foreign exchange. The space economy, valued at around Rs 6,700 crore today, is projected to grow to 10 per cent of the global total by 2034. These are not the returns of a programme too expensive to afford; they are the returns of a programme too cheap not to expand.
And yet there is a persistent leak in the system. Approximately one-third of IIT graduates leave India each year. India loses an estimated $160 billion annually from brain drain, income tax foregone, pension contributions unrealised, and the uncaptured return on public investments in education and training. The problem is not merely financial. It is the problem Griliches identified in 1979. Knowledge spills over, but in India’s case the spillovers are flowing outward. The non-rivalry of knowledge, which should compound India’s growth, is instead compounding Silicon Valley’s. The Washington Consensus (that developing countries should reduce trade barriers, stabilise their economies, attract foreign investment, and let technology flow in) had no theory of publicly funded research as an instrument of development. The assumption was that invention was for the rich and adoption was for the rest. The assumption is now collapsing under its own contradictions. South Korea had a per-capita GDP lower than Ghana’s in 1960. It chose institutions (KIST, KAIST, ETRI) over adoption. By the mid-2010s, it commanded over 60 per cent of the global memory chip market. It now spends 5.32 per cent of its GDP on R&D.
India’s Economic Survey of 2020-21 observed, with characteristic understatement, that India’s GERD as a percentage of GDP had been ‘stagnant’ for two decades. The Annual Science Report of 2024 noted that India ranked 39th in the Global Innovation Index, up from 81st in 2015. Progress is occurring, but not yet compounding. The gap between what India spends and what the evidence says it should spend is not a policy question with two legitimate sides. It has a number, approximately the difference between 0.64 per cent and 2 per cent of GDP, the level at which Romer’s framework suggests a country begins to sustain growth rather than merely borrow it. ANRF and 1 lakh crore RDIF is definitely a good start. Mahalanobis’s 1959 note ended with a sentence that has aged poorly in its optimism: ‘India has the talent, it requires only the institutional architecture.’ The institutional architecture, sixty-five years later, is still under construction. The talent, meanwhile, is boarding flights.
Aditya Sinha writes on macroeconomics and geopolitics. Shaurya Pandey is a Technology Strategy Consultant.
