Executive Summary
UK spending on research and development, by business, universities, government and charities together, was £79.4 billion in 2024, 2.71% of GDP. Businesses carried out 70% of it, universities 23% and government bodies 6%.1 On the OECD's figures, that puts the UK about level with the OECD average, ahead of France and the European Union as a whole, and behind Germany, the United States, Japan and South Korea.2
The government's own R&D budget is rising to £22.6 billion a year by 2029-30, under the 2025 Spending Review.3 Most of it flows through the Department for Science, Innovation and Technology, which plans to spend £58.5 billion on R&D between 2026-27 and 2029-30, including £38.6 billion through UK Research and Innovation (UKRI), £1.2 billion for the Advanced Research and Invention Agency (ARIA) and £8.7 billion on membership of EU research programmes such as Horizon Europe.4
Two problems sit underneath the headline. First, UKRI funds most university research grants at 80% of their full economic cost, and universities recover less than that, a gap UKRI says is filled by "cross-subsidy from international student income".5 International recruitment is now falling, and the Office for Students modelled in November 2025 that around 45% of English higher education providers could report a deficit in 2025-26.6 Second, R&D tax relief, worth £8.2 billion in 2024-25 and comparable in size to UKRI's annual budget, is a major public subsidy for business research, yet HMRC's statistics on it include no assessment of error, fraud or how much additional research it produces.7
Key Proposals
Pay the full cost of the research the government commissions, phased in and published. Raise the share of full economic cost paid on public research grants in stages, publish the remaining gap each year, and stop relying on international students' fees to fund public research.
Publish a regular independent evaluation of R&D tax relief. Every three years, publish how much additional research each pound of R&D tax relief produces, and the estimated level of error and fraud, so that the £8 billion a year it costs is judged on the same evidence as direct grants.See how Budget 2026 relates to this
Set a published, protected floor for curiosity-driven research. Turn the government's stated first priority, "protecting curiosity-driven, foundational science", into a ring-fenced minimum within UK Research and Innovation's budget, published for each year of the spending period.
1. Where Research Money Comes From and Goes
The UK spent £79.4 billion on research and development in 2024, £3.6 billion more than in 2023. As a share of GDP that was 2.71%, slightly below 2.72% the year before because the economy grew faster than research spending. Businesses performed £55.6 billion of it, universities £17.9 billion, government bodies £4.7 billion and charities and other non-profit organisations £1.2 billion.1
On the OECD's comparable series, the UK spent 2.75% of GDP on R&D in 2023. The OECD average in 2024 was 2.72%. Germany spent 3.13%, the United States 3.44%, Japan 3.62% and South Korea 5.13%; France spent 2.18% and the European Union as a whole 2.13%.2 The UK is not a laggard, but nor is it among the countries that spend most heavily on research.
Government's own R&D budget is due to rise to £22.6 billion a year by 2029-30.3 The Department for Science, Innovation and Technology's plans allocate £38.6 billion to UKRI between 2026-27 and 2029-30, rising to almost £10 billion a year by the end of the period; £1.2 billion to ARIA, the agency set up to fund high-risk, high-reward research; and £8.7 billion to UK contributions to EU programmes, including Horizon Europe, which the UK rejoined as an associated country, and its successor, whose budget is not yet settled.4 The Treasury's narrower classification of public spending on "science and technology" was £9.7 billion in 2024-25.8
Not all public research money is rising. The Foreign Office's research funded from the aid budget falls by 44% over 2026-27 to 2028-29, as part of the cut in aid to 0.3% of national income.9
2. Research Funded Below Its Cost
Most government research funding for universities comes in two forms: competitive grants for specific projects, largely through UKRI's research councils, and a block grant based on research quality. UKRI awards most of its project grants at 80% of what it calls the full economic cost of the research, expecting universities to meet the rest. UKRI's own analysis, published in 2023, found that the share of cost universities actually recover through the Transparent Approach to Costing (TRAC), the sector's standard costing method, is "consistently lower than this and declining", and that funding for postgraduate research students has been below 50% of cost for many years.5
The gap has to be filled from somewhere. UKRI's analysis says directly that "research relies on cross-subsidy from international student income", alongside a deficit on publicly funded teaching that also needs cross-subsidy.5 In other words, part of the cost of the UK's publicly funded research has been paid by the fees of students from other countries, especially China.
That arrangement is now under pressure. The Office for Students' November 2025 update found that larger research-intensive universities had seen a 7% fall in the number of study confirmations issued to international postgraduate students, with an 11.6% fall in those issued to students from China. Its modelling suggested that, without further action, 124 providers, around 45% of the sector in England, could report a deficit in 2025-26, improving to 41% in 2026-27 once inflation-linked rises in home tuition fees take effect.6
UKRI funds most grants at 80% of full economic cost, cost recovery is lower and declining, and UKRI states that research relies on cross-subsidy from international student income.5 The Office for Students found international postgraduate recruitment at research-intensive universities down 7% and modelled around 45% of English providers in deficit in 2025-26.6
This connecting argument is this project's own. When public research is funded below cost and the gap is filled by international fees, the UK's research capacity depends on immigration policy, exchange rates and the decisions of students in other countries, none of which are research policy. A fall in international recruitment then forces universities to cut research or teaching that the government is paying for only in part.
One option is to accept the current model and rely on international recruitment recovering. A second is to fund fewer projects at full cost within the same budget. A third, which this pillar proposes, is to raise the share of full cost paid on public grants in stages, using part of the planned rise in the R&D budget, and to publish the remaining gap each year so the trade-off between the number of projects and their funding rate is visible.
Paying full cost means funding fewer projects for the same money, unless the budget rises. This pillar judges that a smaller amount of securely funded research is better than a larger amount that depends on a fragile cross-subsidy. A reader who values the number of research projects more highly is weighing the same evidence differently.
Raise the share of full economic cost paid on public research grants in stages over the spending period, and publish the remaining gap annually. The test below sets out what would show this is not working.
3. Eight Billion Pounds of Tax Relief
Alongside direct funding, the government supports company research through tax relief. In 2024-25 companies claimed £8.2 billion of R&D tax relief, up 5% on the year before, supporting £51.0 billion of R&D spending. The number of claims fell by 17% to 40,325, after reforms that merged the main schemes and tightened eligibility. Large companies received £4.6 billion and smaller ones £3.6 billion.7
That makes R&D tax relief comparable in scale to UKRI's entire annual budget. Yet HMRC's statistics state that they "do not include any assessment of the levels of error and fraud", and they do not measure how much research the relief causes that would not have happened anyway, its additionality.7 HMRC has published separate estimates of error and fraud and has commissioned evaluations in the past, but there is no regular, published assessment of value for money that sits alongside the annual cost.
4. Discovery Research and Missions
The government's research plans set three priorities, the first of which is "protecting curiosity-driven, foundational science", alongside research that supports its industrial strategy and its policy missions. The plans set out budgets by organisation, but they do not set a separate, protected amount for curiosity-driven research.4 At the same time, the 2025 Spending Review created a new £500 million R&D Missions Accelerator Programme to direct research towards the government's missions.3
Both kinds of research matter. Mission-led funding concentrates effort on problems the government has chosen; curiosity-driven research produces the discoveries nobody knew to ask for. The risk, when a budget is under pressure and governments want visible results, is that the second is squeezed to fund the first, and without a published floor there is no way to see whether that is happening.
5. The Steel Man
The strongest case against this pillar is that it misdiagnoses the problem. The UK's weakness is not how research is funded but how little of it is turned into products, companies and jobs at home: the UK is good at discovery and poor at commercialisation, and money should go to the latter. On this view, paying universities more for each grant rewards institutions that have over-expanded on international recruitment, and the right response to falling international fees is for universities to become more efficient, not for taxpayers to fill the gap.
R&D tax relief, the argument continues, is already heavily scrutinised and has been reformed twice in recent years. Adding a statutory evaluation risks creating uncertainty that deters the very investment the relief is meant to encourage. And ring-fencing curiosity-driven research would reduce the flexibility of research funders to move money to where it is most needed, which is exactly what ARIA and the missions programme are designed to do.
This case has force on commercialisation, which this pillar does not address and which the Economy pillar covers in part. But it does not answer the evidence that public research is currently funded below cost and dependent on a falling source of income. Universities can and should become more efficient; that does not change the fact that the government is paying only part of the cost of the research it commissions. And regular evaluation is how a large scheme earns continued support, not a threat to it.
6. Counter-Arguments
'Full-cost funding just bails out universities that over-recruited international students'
The cross-subsidy exists because public research is funded below cost, which is a government funding decision, not a university one.5 The proposal phases in higher rates and publishes the gap, so that the cost and the trade-off are visible rather than hidden in international fee income.
'R&D tax relief has already been reformed, so evaluation is unnecessary'
The reforms tightened eligibility and merged schemes, and claims fell by 17% in 2024-25 while the cost rose by 5%.7 Whether the reformed scheme produces more research per pound is exactly what a published evaluation would show.
'Ring-fencing discovery research makes funding inflexible'
The government has already said curiosity-driven science is its first priority.4 A published floor only makes that commitment visible and checkable; everything above the floor remains flexible.
Cross-Pillar Dependencies
| Pillar | Connection |
|---|---|
| The Economy | Research is one of the main routes to higher productivity, and the Economy pillar addresses how investment and innovation reach the wider economy. |
| Education | University finances link research and teaching: teaching of home students also runs at a deficit, and both depend on international fees. |
| Immigration | International students' fees cross-subsidise research, so visa and immigration policy directly affects research funding. |
| Overseas Aid | Aid-funded research falls by 44% under the cut to 0.3% of national income. |
| Artificial Intelligence | Public investment in AI research and computing is covered in the AI pillar. |
8. Proposals for Change
These are the evidence-based proposals of this pillar, put forward for public discussion and challenge.
-
P1 Raise the share of full economic cost paid on public research grants in stages over the current spending period, using part of the planned rise in the R&D budget, and publish each year the remaining gap and the number of projects funded, so that the trade-off is visible.
-
P2 Publish every three years an independent evaluation of R&D tax relief, covering the additional research it produces per pound and the estimated level of error and fraud, alongside the annual cost.
-
P3 Set a published, ring-fenced minimum for curiosity-driven research within UKRI's budget for each year of the spending period, so that the government's stated first priority is protected in practice.
The UK's problem in science is not the size of its ambition. It is that the research it pays for is underfunded, propped up by a source of income it does not control, and that one of its largest research subsidies is not regularly checked. The Generational Reset does not argue for a different total. It argues that the money should pay for what it is meant to, and be seen to work.
The Generational Reset is a non-partisan, public-interest project. It is not affiliated with any political party, does not accept corporate funding, and publishes all its work under open licence for public discussion and adaptation.
For public discussion. Not affiliated with any political party. | generationalreset.org
The Generational Reset | S1_15: Science and Research | For public discussion. Not affiliated with any political party. | generationalreset.org