1. Research excellence and specialisation

Quantity and quality of scientific production, 2005 and 2015
Number of documents and percentage among the world’s 10% most cited publications, fractional counts

Source: OECD calculations based on Scopus Custom Data, Elsevier, Version 4.2017; and 2015 Scimago Journal Rank from the Scopus journal title list (accessed June 2017), July 2017. StatLink contains more data. See chapter notes.


Did you know?

Scientific production is 20-30% higher in China than the United States in fields such as computer or materials science, but is 70% lower in neuroscience.

The indicator of top-cited publications provides a “quality adjusted” measure of research output. In 2015, the United States led the production of scientific publications with nearly half a million. China accounts for the second largest number of top-cited documents, with nearly as many as Germany and the United Kingdom combined. Switzerland has the largest share of domestic scientific documents with a high citation impact, closely followed by the Netherlands.

Countries exhibit specialisation in different scientific domains. A specialisation index provides evidence of the fields in which a given country accounts for a relatively high share of scientific production, compared to the global distribution of scientific output across fields. The relationship between specialisation and citation impact is analysed in four selected domains: Biochemistry, Computer science, Materials science and Neuroscience. While higher specialisation is associated with greater citation impact in the case of biochemistry, this is not the case for other domains. Most countries appear to be similarly specialised in biochemistry, while there is wider heterogeneity for neurosciences. India exhibits high levels of specialisation and output in the area of computer science. The Russian Federation is highly specialised in Materials science. For countries with high levels of scientific output, normalised impact scores are similar across fields. For smaller countries, pockets of excellence can be found in specific areas.


Estimates of scientific publication output are based on counts of citable documents (articles, reviews and conference proceedings), indexed within Elsevier’s Scopus database, by authors with affiliations in each country. Documents are assigned on a fractional basis, according to the number of authors and their respective affiliations in that particular country.

The specialization indicator is calculated by dividing a field’s share of documents within a given country by the global share of that particular field. Economies that have field distributions very similar to that of the entire world exhibit specialisation values very close to 1.

The indicator of scientific excellence (top-cited publications) shows the percentage of a country’s scientific output that is included in the group of the 10% most-cited publications in their respective scientific fields. The 2015 Scimago Journal Rank indicator is used as a complement to sort publications with identical numbers of citations within each class.

Specialisation and citation impact in science, selected fields, 2015
Percentage of documents in the top 10% ranked documents and relative specialisation, by field, fractional counts

Source: OECD calculations based on Scopus Custom Data, Elsevier, Version 4.2017; and 2015 Scimago Journal Rank from the Scopus journal title list (accessed June 2017), July 2017. StatLink contains more. See chapter notes.



Scientific publications provide a measure of scientific production activity based on the numbers of documents published in peer-reviewed journals and indexed by data providers. Publication norms vary by field and sector (OECD and CSIC, 2016), depending on the use of review and dissemination mechanisms as well as organisational disclosure practices. Indexing may also exhibit language-related biases. Scientific excellence is approximated by measures of the distribution of citation “impact” normalised by year of publication, type of document and field(s). High citation rates may incorporate self-citation, refer to retracted papers, and fail to capture relevance to non-publishing communities.

In the case of Scopus, Elsevier uses its All Science and Journal Classification (ASJC) to classify each journal under one or more field subject. Field assignment on a journal basis is approximate, as a given journal’s classification may not provide an accurate representation of each document’s thematic content. To minimise this problem, documents published in generic multidisciplinary journals have been allocated on a fractional basis to the ASJC codes found in both citing and cited papers.