1. R&D specialisation

Business R&D intensity adjusted for industrial structure, 2015
As a percentage of value added in industry

Source: OECD calculations based on the ANBERD Database, http://oe.cd/anberd, the National Accounts (SNA) Database, the Structural Analysis (STAN) Database, http://oe.cd/stan, Main Science and Technology Indicators Database, http://oe.cd/msti, and Research and Development Statistics Database, http://oe.cd/rds, June 2017. See chapter notes.


Did you know?

Manufacturing industries account for over 85% of all business R&D carried out in China, Germany, Japan and Korea, while the OECD average is 54%.

When comparing countries’ business R&D intensity (business R&D expenditure relative to value added), their relative specialisation in different industries with varying R&D intensities can impact the analysis. Comparison is facilitated by calculating the notional value of a country’s business R&D intensity assuming it had an “OECD average industrial structure”. If this were the case – and assuming countries retained their own R&D intensities within industries – adjusted business R&D intensity for Germany and Korea would be below the OECD average of 2.5%, because these economies are relatively specialised in R&D intensive industries. For Belgium and France, business R&D intensity would shift above the OECD average, because they are more specialised in less R&D-intensive industries, while Canada, Norway and the United Kingdom would converge towards the OECD average. For most Southern and Eastern European countries, the change would not greatly alter overall R&D intensity. The United States’ industrial structure already closely resembles the OECD average.

R&D-intensive industries typically account for the largest share of business enterprise expenditure on R&D (BERD) in manufacturing. In most OECD countries, more than one-third of BERD is performed within service industries. This share has increased over the last decade, boosted in part by the role of information and communication service industries. Cross-country comparisons of the sectoral distribution of BERD should be made with care, however, as varying business structures and statistical practices can affect the extent to which R&D performers (including some involved in manufacturing-related R&D) are recorded in the specific R&D services industry – which may account for over half of all R&D in services industries.


R&D intensity adjusted for industry structure is a weighted average of the R&D intensities of a country’s industrial sectors, using the OECD industrial structure’s sector value added shares as weights instead of the actual shares used in the unadjusted measure of R&D intensity. Calculations are based on the ISIC Rev.4 classification.

The R&D intensity groups are defined in Galindo-Rueda and Verger (2016). High and medium-high R&D intensive manufacturing includes “chemicals and pharmaceutical products” (ISIC Rev.4 20 and 21) and “computer, electronic and optical products, electrical equipment, machinery, motor vehicles and other transport equipment” (ISIC Rev.4 26 to 30). R&D services correspond to ISIC Rev.4 Division 72.

Business R&D in manufacturing, by R&D intensity group, 2015
As a percentage of manufacturing R&D

Source: OECD, ANBERD Database, http://oe.cd/anberd; Research and Development Statistics Database, http://oe.cd/rds, June 2017. StatLink contains more data. See chapter notes.


R&D in services industries, 2015
As a percentage of business expenditure on R&D

Source: OECD, ANBERD Database, http://oe.cd/anberd; Research and Development Statistics Database, http://oe.cd/rds, June 2017. StatLink contains more data. See chapter notes.



Distributing R&D to industries presents several challenges. In most countries, R&D expenditure is assigned to a firm’s principal industrial activity based on value-added (“main activity” approach). However, in some, allocation is based on the main R&D activity of the firm or the content of the R&D itself (“industry-orientation” approach). The Frascati Manual 2015 recommends classifying statistical units by their main activity and suggests that their R&D activities be classified both on that basis and by industry orientation. In practice, some countries use a combination of the two approaches. The methodology used to adjust R&D intensity for industrial structure is also sensitive to the industrial classification used. Other impacting factors are the level of aggregation at which the sectoral weights are calculated and the countries included in the benchmark. Countries where the disaggregation level for R&D is too low and where there are no comparable SNA data on Value Added by industry have been excluded.