Research and development (R&D) is creative and systematic work undertaken to increase the stock of knowledge and to devise new applications of it. The OECD's Frascati Manual (2015 edition), the international standard that statistics offices use to measure it, says an activity counts as R&D only if it is novel, creative, uncertain in outcome, systematic, and transferable or reproducible. R&D is divided into basic research, applied research and experimental development.
R&D is carried out by businesses, governments, universities and non-profit bodies, and is a key input into innovation and technological progress. Under the 2008 System of National Accounts it is treated as investment rather than a running cost, so it adds to GDP and to gross fixed capital formation; the US Bureau of Economic Analysis adopted this treatment in 2013. In company accounts the rules differ: under IFRS (IAS 38) research costs are expensed while some development costs may be capitalized, whereas US GAAP generally expenses R&D as incurred.
Research and development in plain terms
Think of a chef who spends afternoons in a back kitchen testing new dishes. Most attempts are thrown away; a few end up on the menu. The ingredients and hours spent on the failures are a real cost, and nobody can tell in advance which experiments will work.
R&D is that back kitchen for companies and countries. It is paid for today in the hope of better products and methods later, which is why it is often the first budget cut in hard times and the first thing governments try to encourage.
Basic research, applied research and development
- Basic research. Work aimed at new knowledge without a particular use in mind, such as studies of materials or mathematics. Universities and public laboratories do most of it.
- Applied research. Work aimed at a specific practical goal, such as finding a battery chemistry that charges faster, or a central bank studying the design of a central bank digital currency.
- Experimental development. Using existing knowledge to produce new or improved products and processes, including prototypes and pilot plants. In most countries this accounts for the largest share of spending.
Routine testing, quality control, market research and ordinary software maintenance are not R&D under the Frascati rules, even though firms sometimes describe them that way.
Who funds and performs R&D
National R&D effort is usually measured as gross domestic expenditure on R&D (GERD) relative to GDP, known as R&D intensity:
GERD is total spending on R&D performed within a country in a year, by all sectors, and GDP is the value of the country's output in the same year. For example, an economy that spends 30 billion dollars on R&D with a GDP of 1,000 billion dollars has an R&D intensity of . Companies use a similar ratio: R&D spending divided by revenue.
According to the OECD, R&D intensity across its member countries has stayed at about 2.7% of GDP since 2020, with Israel and South Korea reporting the highest levels, above 5%. China's spending has grown much faster than the OECD average. Many emerging markets, including India and Pakistan, spend well under 1% of GDP. In advanced economies, business enterprises perform most R&D, while governments fund a large share of basic research.
Why governments support R&D
Knowledge spills over to firms that did not pay for it, a positive externality that gives research some features of a public good. As a result, private firms tend to do less R&D than would be best for society. Governments respond with direct grants and contracts, public laboratories, intellectual property protection and tax incentives.
Tax relief is the most common tool. The United States has offered a research tax credit since 1981. A 2017 law required US firms to spread domestic research costs over five years from 2022, but the One Big Beautiful Bill Act, signed on 4 July 2025, created a new Section 174A that again lets businesses deduct domestic research costs immediately, for tax years beginning after 31 December 2024. The United Kingdom merged its two R&D relief schemes into one for accounting periods starting from April 2024. Several countries also run "patent box" regimes that tax income from patents at lower rates, a feature of tax competition. Broader industrial policy often channels subsidies to research in strategic sectors.
Limitations and criticism
R&D is an input, not a result: high spending does not guarantee useful innovation, and some valuable innovation, especially in services and business models, involves little formal R&D. Generous tax credits give firms a reason to relabel ordinary costs as research, and tax authorities, including HM Revenue and Customs in the United Kingdom, have reported significant error and fraud in claims. Listed firms may cut R&D to meet quarterly targets, a concern raised in debates about short-termism and shareholder value. Finally, because knowledge travels, a country's research may raise productivity mostly in other countries.