The government would get a better return on its investment in research if it axed tax breaks for companies doing research and development and ploughed the money into the research councils instead, an analysis suggests.
The study by business school academics adds that the £3.5 billion spent on publicly funded research annually generates a return to UK companies of £45 billion a year.
The findings will boost research organisations that are striving to make the case for investment ahead of the Comprehensive Spending Review that is expected to follow the forthcoming general election.
The research was presented at a seminar organised by the British Academy and the Economic and Social Research Council last week.
"Our data suggest the strong impact of research council funding on business, and so any cuts would be costly if they were to hit it," said Jonathan Haskel, professor of economics at Imperial College Business School and co-author of the paper "Public Support for Innovation, Intangible Investment and Productivity Growth in the UK Market Sector".
The study analyses the relationship between public investment in research and economic impact by looking back over the past 20 years at any correlation between the investment and increases in private sector productivity when other factors are controlled.
The results show that while there is no correlation with government spending on civil and defence R&D or funding councils' spending on research via the university block grant, private productivity is highly correlated with research council spending.
The study also suggests that this spending delivers greater "bang for the buck" than the R&D tax-credit scheme, which gives companies tax breaks on some areas of spending. Professor Haskel said the scheme had provided £4 billion in tax relief since it was introduced in 2000.
The paper concludes: "These findings tentatively suggest that in a world of constrained fiscal spending, government innovation policy should focus on direct spending on innovation, specifically through research councils, rather than through tax incentives."
It also finds that the results of research council-funded work are taken up very quickly by industry. Professor Haskel said the time lag between investment and impact was usually one to three years.
The findings came as the Russell Group of large research-intensive universities released a series of case studies showing the economic impact of research.
MPs on the Science and Technology Select Committee also released a report this week warning that unless investment is boosted, the government risks "devastating" British science and the economy.
"Given that the best literature on the subject concludes that reliable quantification of the economic impact of investment in science and research is deeply problematic at best, the suggestion that the Treasury is waiting for 'hard figures' on the benefits of research causes us great concern," the committee says.
Commenting on the Imperial study, Tim Bradshaw, head of science, technology and innovation at the CBI, questioned whether it was possible to review parts of the "innovation ecosystem" in isolation.