- Split View
-
Views
-
Cite
Cite
Koen Jonkers, Frédérique Sachwald, The dual impact of ‘excellent’ research on science and innovation: the case of Europe, Science and Public Policy, Volume 45, Issue 2, April 2018, Pages 159–174, https://doi.org/10.1093/scipol/scx071
- Share Icon Share
Abstract
This article draws on innovation studies and bibliometrics to argue that excellent research has a dual impact on both science and innovation. Academic excellence thus constitutes a major objective to ensure economic impact of research through innovation and the development of new high growth sectors. The article confronts the results of empirical studies with both public policies and the production of high impact research in Europe. In the early 2000s, while policies aimed at fighting the ‘R&D deficit’ vis-à-vis the USA and the ‘European paradox’, the EU actually suffered from deficits in both excellent research and innovation in new sectors. Policies in Europe have progressively changed in response to the internationalization of R&D, the emergence of new scientific powers and the combined influence of rankings and of empirical studies. The scientific production of EU28 also improved both quantitatively and qualitatively. The notion of a ‘European paradox’ has however remained part of the narrative on innovation and has kept influencing some policies, resulting in an overemphasis on R&D intensity and insufficient recognition of the role of excellence in science. The paper underscores the diversity of performance between European countries and draws policy conclusions.
1. Introduction1
The perception of the interactions between R&D spending and economic performance that developed in Europe during the early to mid-1990s has had a deep and enduring influence on research and innovation policies. The then increasing transatlantic gap in innovation and growth has been attributed to a combination of more business investment in R&D and a better ability of the USA to generate innovation from research results (EU 1995). Policy conclusions seemed quite straightforward: increase business R&D spending and improve the transformation of research results into innovation. These conclusions have more or less intensely influenced policies at the national and EU levels over the last two decades. In particular, a strong policy focus has been put on increasing science and technology linkages to resolve the perceived ‘European paradox’ identified by the 1995 Green paper on innovation (EU 1995, 2003).
Over the last decade, both explanations of the lower European innovation performance have been questioned and empirical studies have developed a better understanding of interactions within innovation systems. First, private R&D intensity is to a large extent determined by the national economic structure, which, as a consequence, also commands firms’ demand for innovation and interactions with academic research. Secondly, there is no opposition, but rather synergies between the scientific impact and the innovation potential of a given research result. In other words, excellent research has a dual impact on subsequent scientific production and innovation. Thirdly, a number of bibliometric analyses based on data from the 1990s to the mid-2000s have pointed to what could be summarized as a ‘transatlantic excellence gap’. Various authors have produced indicators to question the productivity and quality of European science and European universities (Dosi et al. 2006; Aghion et al. 2010; Albarran et al. 2010; OECD 2013). At the time, the EU generated a larger number of scientific publications than the USA, but these publications were on average of lesser academic impact.
Also since the beginning of the 21st century, research and innovation policies have evolved in response to the fast changing global context. One of the new trends is at odds with the perception that European science would be in a global leading position. Indeed, global competition on the basis of excellence has been mounting, driven by the influence of university rankings and country level benchmarking. Based on comparisons between the EU and US academic performance, a number of researchers have argued that high-impact basic research should become one of the principle objectives of EU policies (Pavitt 2000). In this context, a number of European Member States as well as the European Commission have launched policies and programmes to promote excellence in academic research. At the European level the most important development has been the creation of the European Research Council in 2007 after a long debate on the best way to promote basic research (König 2017).
Several European Member States launched excellence initiatives in which part of the institutional funding was allocated on the basis of ex ante evaluations of proposals, including the Excellenz Initiativ in Germany. Other countries followed the UK in setting up performance-based funding systems, in which institutional funding is allocated on the basis of ex post assessments of research outputs. In total, 15 countries have implemented such systems (Jonkers and Zacharewicz 2016). This development has been promoted by the European Commission, which argued that an increase in competitively allocated funding would raise the production of excellent research.
Yet, there has not been an overall reappraisal of the diagnosis of research and innovation systems in Europe and the notion of a European paradox has still been referred to as a rationale for setting up some of the new schemes in recent years. This article aims at conducting such a reappraisal by linking two sets of evidence: firstly, empirical studies on the respective roles and interactions of business R&D intensity and academic research to generate innovation; secondly, indicators measuring the academic research performance of European countries. The discussion aims at better connecting related strands of the literature on science and science policy, including the examination of indicators used to compare research and innovation systems. Section 2 shows that since the early 2000s, an increasing amount of empirical evidence has been at odds with major tenets of innovation policies in Europe. Section 3 examines to what extent the performance of European countries in academic research is aligned with the perception by policy makers. Section 4 pulls the analysis together and underscores the correlation between research and innovation performance among European countries. The concluding section draws policy implications.
2. Diagnosis on innovation in Europe: academics versus policy makers?
Business R&D has long been recognized as a major input to the innovation process. At the firm level, conducting R&D is causing product innovation a couple of years later (Raymond et al. 2015). Across industries, business R&D intensity is correlated to proximity to science and new knowledge creation fuelling technological opportunities (Klevorick et al. 1995). At the sector and country levels, business R&D intensity is correlated to productivity performance (Guellec and Van Pottelsberghe 2001; Haskel et al. 2014). Empirical studies have found high rates of return to private R&D as well as substantial social returns through externalities (Hall et al. 2009).
As a consequence, innovation policies have tended to support business R&D through various schemes, including direct subsidies and indirect fiscal schemes. These policies have been supplemented with an increasing array of schemes to stimulate knowledge transfer and the development of innovative start-ups. Public policies have thus adapted to the need to stimulate the development of new sectors and to the adoption of open innovation practices by companies. This section argues that in doing so some of these policies have focused on bringing research closer to the market in order to increase impact. As a result, despite the objective of better connecting academic research and firms, some innovation policies in Europe have underestimated the potential of basic and excellent research2 for informing high tech innovation.
2.1 R&D intensity and innovation performance
The empirical studies reviewed below suggest that since 2000, innovation policies in Europe have tended to both overemphasize the role of business R&D intensity as an indicator of innovation potential and insufficiently acknowledge the close interactions between academic impact and economic impact of research results. A number of policies in Europe have aimed at increasing economic impact through technology transfer schemes, while most overlooked the importance of academic excellence in the dynamics of public–private R&D cooperation and its positive influence on high tech innovation performance.
The fundamental role of R&D in the innovation process does not mean that a country’s R&D intensity, and even less so its business R&D intensity, is a direct policy variable. This is why the 3 per cent EU target decided as part of the Lisbon Strategy at the beginning of the 2000s has been questioned. Indeed, at the macro-economic level, R&D intensity may be considered as an output of the innovation process (Van Pottelsberghe 2008). Since R&D intensity depends on technological characteristics, each sector exhibits a set range of R&D intensity. In other words, in each country, R&D intensity strongly depends on the sector distribution of value added (Le Ru 2012a, 2012b). This is also the case for the EU as a whole (Moncada-Paterno-Castello and Voigt 2013). Clarifying this issue is very important because R&D intensity has been a heavily used indicator in innovation policies, both as a direct policy objective and indirectly as a one indicator in innovation scoreboards.3
Table 1 presents both business R&D intensity as it is directly observed and R&D intensity adjusted for sector distribution. For each country, the structure-adjusted indicator of R&D intensity is a weighted average of its sectoral R&D intensities using the average OECD industrial structure. The table is ordered by the adjusted business R&D intensity (column B). It shows that the high business R&D intensity of Korea and Germany strongly depends on their industrial structure: when adjusted for sector structure, it is below OECD average. Norway, France or Portugal for example are in the reverse situation. Norway is rich in natural resources and France has a much lower share of manufacturing sectors than Germany. EU countries with either a large service sector or a large agriculture or natural resource sector tend to have a higher adjusted business R&D intensity (C above 1.2), while the reverse is true for countries with a large manufacturing industry (C below 0.9). Among countries with a relatively high share of manufacturing industry, those with the highest business R&D intensity are those with the largest share of high tech industries (Korea, Japan, Finland). The share of manufacturing industry is a stronger determinant of the ratio between adjusted and unadjusted R&D intensity than the level of economic development: Greece, Norway and France have a high ratio, while Germany, the Czech Republic and Hungary have a low one.
. | Business R&D intensity* . | Rank A . | Business R&D intensity adjusted for industrial structure . | Rank B . | C = B / A . |
---|---|---|---|---|---|
Finland | 4.04 | 2 | 3.78 | 1 | 0.936 |
Japan | 3.89 | 3 | 3.62 | 2 | 0.930 |
Denmark | 3.49 | 5 | 3.44 | 3 | 0.987 |
Austria | 2.82 | 8 | 3.32 | 4 | 1.178 |
Sweden | 3.70 | 4 | 3.29 | 5 | 0.888 |
France | 2.49 | 10 | 3.26 | 6 | 1.309 |
USA | 2.81 | 9 | 2.87 | 7 | 1.021 |
Belgium | 2.38 | 11 | 2.70 | 8 | 1.137 |
OECD | 2.46 | – | 2.46 | – | 1 |
Slovenia | 3.07 | 7 | 2.35 | 9 | 0.768 |
Netherlands | 1.70 | 12 | 2.22 | 10 | 1.307 |
Korea | 4.72 | 1 | 2.20 | 11 | 0.465 |
Germany | 3.08 | 6 | 2.15 | 12 | 0.698 |
UK | 1.68 | 13 | 2.10 | 13 | 1.252 |
Norway | 1.33 | 17 | 2.06 | 14 | 1.544 |
Portugal | 1.17 | 19 | 1.75 | 15 | 1.497 |
Ireland | 1.56 | 14 | 1.63 | 16 | 1.050 |
Italy | 1.12 | 20 | 1.54 | 17 | 1.378 |
Spain | 1.05 | 21 | 1.39 | 18 | 1.320 |
Estonia | 1.27 | 18 | 1.36 | 19 | 1.070 |
Czech Republic | 1.51 | 15 | 1.25 | 20 | 0.828 |
Greece | 0.46 | 24 | 1.06 | 21 | 2.318 |
Hungary | 1.35 | 16 | 1.00 | 22 | 0.741 |
Slovakia | 0.54 | 22 | 0.63 | 23 | 1.170 |
Poland | 0.53 | 23 | 0.60 | 24 | 1.122 |
. | Business R&D intensity* . | Rank A . | Business R&D intensity adjusted for industrial structure . | Rank B . | C = B / A . |
---|---|---|---|---|---|
Finland | 4.04 | 2 | 3.78 | 1 | 0.936 |
Japan | 3.89 | 3 | 3.62 | 2 | 0.930 |
Denmark | 3.49 | 5 | 3.44 | 3 | 0.987 |
Austria | 2.82 | 8 | 3.32 | 4 | 1.178 |
Sweden | 3.70 | 4 | 3.29 | 5 | 0.888 |
France | 2.49 | 10 | 3.26 | 6 | 1.309 |
USA | 2.81 | 9 | 2.87 | 7 | 1.021 |
Belgium | 2.38 | 11 | 2.70 | 8 | 1.137 |
OECD | 2.46 | – | 2.46 | – | 1 |
Slovenia | 3.07 | 7 | 2.35 | 9 | 0.768 |
Netherlands | 1.70 | 12 | 2.22 | 10 | 1.307 |
Korea | 4.72 | 1 | 2.20 | 11 | 0.465 |
Germany | 3.08 | 6 | 2.15 | 12 | 0.698 |
UK | 1.68 | 13 | 2.10 | 13 | 1.252 |
Norway | 1.33 | 17 | 2.06 | 14 | 1.544 |
Portugal | 1.17 | 19 | 1.75 | 15 | 1.497 |
Ireland | 1.56 | 14 | 1.63 | 16 | 1.050 |
Italy | 1.12 | 20 | 1.54 | 17 | 1.378 |
Spain | 1.05 | 21 | 1.39 | 18 | 1.320 |
Estonia | 1.27 | 18 | 1.36 | 19 | 1.070 |
Czech Republic | 1.51 | 15 | 1.25 | 20 | 0.828 |
Greece | 0.46 | 24 | 1.06 | 21 | 2.318 |
Hungary | 1.35 | 16 | 1.00 | 22 | 0.741 |
Slovakia | 0.54 | 22 | 0.63 | 23 | 1.170 |
Poland | 0.53 | 23 | 0.60 | 24 | 1.122 |
R&D spending over value added in industry for all activities less Real estate; Public administration and defence; compulsory social security and education; Human health and social work activities; and Activities of households as employer. This generates a higher ratio than the R&D intensity calculated in reference to total GDP.
Source: calculation based on data from OECD (2015).
. | Business R&D intensity* . | Rank A . | Business R&D intensity adjusted for industrial structure . | Rank B . | C = B / A . |
---|---|---|---|---|---|
Finland | 4.04 | 2 | 3.78 | 1 | 0.936 |
Japan | 3.89 | 3 | 3.62 | 2 | 0.930 |
Denmark | 3.49 | 5 | 3.44 | 3 | 0.987 |
Austria | 2.82 | 8 | 3.32 | 4 | 1.178 |
Sweden | 3.70 | 4 | 3.29 | 5 | 0.888 |
France | 2.49 | 10 | 3.26 | 6 | 1.309 |
USA | 2.81 | 9 | 2.87 | 7 | 1.021 |
Belgium | 2.38 | 11 | 2.70 | 8 | 1.137 |
OECD | 2.46 | – | 2.46 | – | 1 |
Slovenia | 3.07 | 7 | 2.35 | 9 | 0.768 |
Netherlands | 1.70 | 12 | 2.22 | 10 | 1.307 |
Korea | 4.72 | 1 | 2.20 | 11 | 0.465 |
Germany | 3.08 | 6 | 2.15 | 12 | 0.698 |
UK | 1.68 | 13 | 2.10 | 13 | 1.252 |
Norway | 1.33 | 17 | 2.06 | 14 | 1.544 |
Portugal | 1.17 | 19 | 1.75 | 15 | 1.497 |
Ireland | 1.56 | 14 | 1.63 | 16 | 1.050 |
Italy | 1.12 | 20 | 1.54 | 17 | 1.378 |
Spain | 1.05 | 21 | 1.39 | 18 | 1.320 |
Estonia | 1.27 | 18 | 1.36 | 19 | 1.070 |
Czech Republic | 1.51 | 15 | 1.25 | 20 | 0.828 |
Greece | 0.46 | 24 | 1.06 | 21 | 2.318 |
Hungary | 1.35 | 16 | 1.00 | 22 | 0.741 |
Slovakia | 0.54 | 22 | 0.63 | 23 | 1.170 |
Poland | 0.53 | 23 | 0.60 | 24 | 1.122 |
. | Business R&D intensity* . | Rank A . | Business R&D intensity adjusted for industrial structure . | Rank B . | C = B / A . |
---|---|---|---|---|---|
Finland | 4.04 | 2 | 3.78 | 1 | 0.936 |
Japan | 3.89 | 3 | 3.62 | 2 | 0.930 |
Denmark | 3.49 | 5 | 3.44 | 3 | 0.987 |
Austria | 2.82 | 8 | 3.32 | 4 | 1.178 |
Sweden | 3.70 | 4 | 3.29 | 5 | 0.888 |
France | 2.49 | 10 | 3.26 | 6 | 1.309 |
USA | 2.81 | 9 | 2.87 | 7 | 1.021 |
Belgium | 2.38 | 11 | 2.70 | 8 | 1.137 |
OECD | 2.46 | – | 2.46 | – | 1 |
Slovenia | 3.07 | 7 | 2.35 | 9 | 0.768 |
Netherlands | 1.70 | 12 | 2.22 | 10 | 1.307 |
Korea | 4.72 | 1 | 2.20 | 11 | 0.465 |
Germany | 3.08 | 6 | 2.15 | 12 | 0.698 |
UK | 1.68 | 13 | 2.10 | 13 | 1.252 |
Norway | 1.33 | 17 | 2.06 | 14 | 1.544 |
Portugal | 1.17 | 19 | 1.75 | 15 | 1.497 |
Ireland | 1.56 | 14 | 1.63 | 16 | 1.050 |
Italy | 1.12 | 20 | 1.54 | 17 | 1.378 |
Spain | 1.05 | 21 | 1.39 | 18 | 1.320 |
Estonia | 1.27 | 18 | 1.36 | 19 | 1.070 |
Czech Republic | 1.51 | 15 | 1.25 | 20 | 0.828 |
Greece | 0.46 | 24 | 1.06 | 21 | 2.318 |
Hungary | 1.35 | 16 | 1.00 | 22 | 0.741 |
Slovakia | 0.54 | 22 | 0.63 | 23 | 1.170 |
Poland | 0.53 | 23 | 0.60 | 24 | 1.122 |
R&D spending over value added in industry for all activities less Real estate; Public administration and defence; compulsory social security and education; Human health and social work activities; and Activities of households as employer. This generates a higher ratio than the R&D intensity calculated in reference to total GDP.
Source: calculation based on data from OECD (2015).
Sector composition has thus been playing a fundamental role in the EU ‘R&D deficit’ as compared to the USA. High-tech sectors represent a much higher share of business R&D in the USA. During the 2000s, the share of high tech has increased on both sides of the Atlantic, but the overall gap has remained. The EU remains specialized in mid-high-tech sectors, while the USA has further increased its specialization in high tech and knowledge intensive sectors (EU 2015). More generally, an analysis at the firm-level confirms that across-sector differences dominate over within-sector differences (Stančík and Biagi 2012). American high-tech sectors and knowledge intensive services are however particularly R&D intensive: the US adjusted R&D intensity is slightly higher than its observed intensity, resulting in a better rank for that indicator (Table 1).
Since 2000, a number of European countries have engaged in policies to increase their R&D intensity by stimulating business research spending. These policies have had some positive impacts on R&D intensity. In France, for example, business R&D intensity has reversed its erosion since 2008 despite continuing de- industrialization and the economic crisis. This means that R&D intensity has strongly increased in a number of sectors, including for example in the car industry (MENESR 2014). Business R&D intensity has also increased in a number of South and Eastern European countries (Rodriguez-Pose 2014; OECD 2014; JRC RIO, 2016). These evolutions can contribute to innovation and structural upgrading in existing firms and sectors. At the EU level, these increases in R&D intensity are nevertheless relatively modest and will not be sufficient to meet the 3 per cent objective or to generate innovation-based growth if economic structures do not evolve more strongly in a number of countries. The issue of structural evolution can be brought up in a similar way with respect to other indicators that are, despite their well-known limitations, widely used to measure innovation, such as patents or mid-high tech and high-tech exports. This point will be discussed below in relation to intra-EU diversity and the European Innovation Scoreboard (Section 4).
The importance of the industrial structure to explain R&D intensity had become clearer by the end of the 2000s, which led analysts to consider the dynamics of structural change as a major issue for the development of the knowledge based economy and innovation-based growth. Analyses dealing with these issues have underscored the role played by young high tech or knowledge intensive firms. In recent economic history, the USA has experienced a more dynamic evolution of its industrial structure than a number of European countries. The USA has in particular generated new product and service sectors based on new technologies. In Europe, some large companies manage to keep productivity and innovation up in their sector of origin, as illustrated by German companies in the automobile and mechanical industries. However, rapid growth in new sectors typically depends on new ideas and innovation brought to the market by young high growth companies. Such companies can also enter established sectors through disruptive innovation, which may or may not be technologically based.
These dynamics are confirmed by the age structure of R&D leaders of the Industrial R&D Scoreboard (EU 2011). The share of young firms4 in R&D spending of the world innovation leaders tends to be larger in sectors that are still recent and rapidly growing. Young firms have a higher impact on the innovation and growth of such sectors, including biotechnology, computer hardware and software or internet. At the end of the 2000s, all the world innovation leaders in the internet sector had been born after 1990.5 These more recent activities correspond to sectors of technological specialization for the USA as measured by relative technological advantage (Veugelers and Cincera 2010; OECD 2013). Many more world leading firms in these sectors originate from the USA than from Europe. In 2014, 22 US companies were among the global R&D leaders, for only 2 from Europe (EU 2015). As a result, the structural disadvantage of Europe in developing the knowledge economy may not be diminishing.
The rapid emergence of new knowledge based activities has been interacting with the development of open innovation practices since the late 1990s. The need to connect to science has always meant a certain degree of openness of firms’ research activities to the broader scientific community, but since the late 1990s, openness has increased and has been systematically organized by an increasing number of companies (Chesbrough 2003). This trend results from a set of converging determinants, including more competitive pressures, more focused companies, increasing R&D capabilities around the world and ever more efficient communication technologies (Sachwald 2009). In this global context, American and European companies have reduced their investment in internal scientific capability, which can be proxied by the proportion of basic research in R&D activities or by the publication of scientific articles by company researchers (Arora et al. 2015). As a result, in order to develop new inventions, firms depend more upon their innovation ecosystem, which prominently includes academic institutions, start-ups and R&D focused spin-outs. The latter can be comparable to external projects, acting as transfer channels for research results in a number of sectors where large companies often buy out the most promising start-ups. Such interactions have been repeatedly observed over the latest decades when new sectors or activities have emerged from biotechnologies to the digitalization of both manufacturing and service sectors.
Overall, these richer innovation ecosystems do not rely less on science, but allow more specialized firms and institutions to collaborate to generate new products and services. As a consequence, the development of young companies in new sectors has progressively been identified as a major objective of national and European innovation policies (Guellec and Sachwald 2008; Veugelers 2011). Over the last couple of years, many new policy instruments have been implemented to promote the creation of innovative start-ups and try to stimulate their growth and their impact on the economy.
2.2 The dual impact of research on science and innovation
Since the late 1990s, public policies in Europe have developed numerous schemes to promote the transfer of research results to firms, including by promoting public–private cooperation in R&D and the creation of innovative start-ups. In a context of fiscal consolidation, policies also try to increase the impact of public funding of public and private R&D expenditure on innovation and economic outcomes (OECD 2014). Related policies include more funding for applied research and experimental development as opposed to fundamental research, as well as various schemes to develop and speed up technology transfer of results from academic research.
Policies in favour of technology transfer have often been justified by the perception that European countries are good at creating knowledge but not at converting it into innovation. The notion of the ‘European paradox’ has for example often been referred to in presentations about the mission of the European Institute of Technology (EIT).6 The same idea has also been mentioned in 2016 as a motivation for the project to create a European Innovation Council.7
These policies tend to embrace a linear perspective of innovation in which the economic impact of research results depends on specific transfer activities and a number of steps to the market. They have focused on transfer schemes and institutions, while they overlooked the characteristics of the supply side (academics) and of the demand side (firms). Available evidence suggests though that interactions are all the more fruitful to generate innovation when firms have adequate absorptive capacity and when researchers have produced excellent research results. Besides, training of students and the inter-sectoral mobility of human resources are major pathways for research and frontier knowledge to contribute to innovation. Not only are star scientists a major origin of radically innovative ideas which in the long term can lead to disruptive technological change. Their graduate and PhD students tend to play an important role in further developing these ideas, by creating start-ups or after having been recruited by firms more generally. Throughout their research work in high tech industries they remain in contact with academia (Bonnaccorsi 2009).
Open innovation processes are designed to operate effective interactions between internal RDI capabilities and external resources. Thus, in the open innovation context, firms strongly depend on their research environment, which is composed of a mix of public research institutions, technology transfer operators and other firms. The latter include various types of innovation partners, including start-ups. Successful open innovation thus depends on a sophisticated technological and service environment and on access to relevant research results. This is also the case for non-technological innovation, which often makes use of sophisticated information and communication services.
Firms tend to interact preferably with academic teams from the same country.8 However, global innovation networks are developing and large firms as well as those relying most on science are able to identify relevant partners around the world (Sachwald 2013). Attraction to excellent science has become a relatively more important determinant for the location of R&D activities over the last decades (Arundel and Geuna 2004; Abramovsky et al. 2007; Hedge and Hicks 2008). The general trend of scientific globalization and global innovation networks makes collaboration at a distance easier (Tijssen et al. 2011).
Empirical research synthesized below indicates that productive and academically well assessed researchers tend to engage more with technology transformation and to cooperate more with firms. Conversely, firms look for excellent researchers, with whom cooperation is more fruitful. Finally, in high-tech sectors, firms also tend to develop open science strategies. In particular, those firms that draw on scientific results and researchers’ know-how develop networks with academic communities, which often implies devoting resources to scientific publication. Overall, these empirical results suggest that high academic impact also generates positive impact on innovation and firms’ competitive edge through various channels, including the building up of firm’s absorptive capacities through public–private cooperation.
2.2.1 Academics economic engagement increases with their scientific impact
A number of empirical studies of the determinants of researchers’ engagement with firms have used individual and institutional level data. They cover countries with different types of innovation systems (USA, UK, Sweden, Germany, Spain and Italy) and reach a number of converging conclusions.9 They found that researchers with high scientific productivity and/or high scientific impact tend to engage more in technology transfer activities. Productivity indicators are based on the number of articles produced by researchers. Studies use different proxies to build scientific impact indicators: volume of government grants based on peer review processes and citations to scientific publications in particular. Indicators of scientific impact generally have a positive relationship with indicators of engagement in transfer activities, both at the individual and institutional levels and both through contract research and through commercialization (licensing and start-up creation).
The determinants of technology transfer activities also depend quite strongly on the discipline of the researcher and, at the institutional level, the presence of certain disciplines. Applied sciences and in particular engineering have unsurprisingly the most consistently positive impact on both engagement and commercialization. The effect of being a life scientist appears stronger for commercialization than for engagement. This could be related to the more science-based character of biotechnology and health technologies.10 Patents in these fields are those citing non-patent literature the most, which suggests a more direct impact of academic research on technology.11 Differences among disciplines have also been observed through the relative importance of different university activities for firms at the regional level. A study on Italian data has jointly analysed the effects of education and research activities of universities on local firms’ technological performance (Leten et al. 2014). The latter is measured by the number of patents in an industry and province, weighted by the number of forward citations received over a five-year window. Results indicate that for chemical and mechanical industries, the positive impact of universities on local technological performance is transmitted through the production of skilled labour in science and engineering. In electrical and pharmaceutical industries, scientific publications by researchers from local universities generate an additional positive effect on technological performance. Empirical results thus suggest that research with a high academic impact is positively correlated to technology transfer activities and does not prevent it. It may be related to the fact that firms pick researchers with an excellent track record. A study of Italian universities found that industrial partners take the scientific output of universities into account as a selection criterion for contract research (Van Looy et al. 2011). Besides, industry experience is associated with positive effects on productivity and the training of students and junior faculty (Dietz and Bozeman 2005; Lin and Bozeman 2006). From a policy perspective there is thus no need to compromise scientific impact in order to promote technology transfer.
2.2.2 The contribution of excellent research to innovation
Empirical studies have analysed various technology transfer pathways, including R&D cooperation with firms, spin offs by researchers and licensing. For all pathways, studies conclude that scientific impact, proxied by citation impact, has a positive effect on transfer.
Studies on cooperation have shown that agreements with academic researchers are about explorative R&D and novelty. Cooperation with academic researchers generates more patents or significant technology than cooperation with suppliers or customers (Cassiman et al. 2007). A study of collaboration between firms and universities in Japan showed that resulting patents had a higher ‘quality’ than those flowing from firms’ internal R&D (Motohashi and Muramatsu 2012). Firms entering in such public–private cooperative research tend to have a solid absorptive capacity and conduct exploratory research to serve an ambitious innovation strategy (Miotti and Sachwald 2003; Bercovitz and Feldman 2007). Cooperation with academic research leads to higher levels of novelty while it does not represent risks in terms of value appropriation, which overall generate positive performance effects (Belderbos et al. 2004, 2014; Faems et al 2005).
Some studies have been able to take into account the profile of the researchers or academic institutions involved. They tend to conclude that firms’ proximity with academic researchers is conducive to more innovation when those researchers achieve high academic impact. Otherwise, firms extend their search for academic partners further away as they tend to prefer the right partner to one nearby. In turn, this can be related to the fact that companies opt for collaborating with academia on research activities of a rather basic nature (Belderbos et al. 2004).
The case of star scientists in the development of emerging technologies offers an extreme example of the strong interactions between the commercialization of a technology and the underlying science. Star scientists have played a key role in the development of ICT in the USA from the 1950s on (Bonaccorsi 2011). The presence of star scientists has been an important factor in the development of biotechnology companies during the 1990s when they played a central role in both the development of the science and its successful commercialization (Zucker and Darby 1996). Zucker and Darby (2003) have also shown that the location of scientists publishing breakthrough articles across US metropolitan areas is correlated with firm entry in nanotechnology. These empirical results on the importance of proximity to outstanding researchers can be related to the more general observation that a technology which is transferred to a start-up is typically at an early stage and needs to be further developed in connection with the researcher herself (Mowery and Ziedonis 2014). In other words, technology transfer requires an interaction between codified knowledge (e.g. publications and academic patents) and complementary tacit knowledge embodied in the researcher. Collaboration with researchers with a very strong reputation is more likely to yield positive results, which may be due to their superior knowledge, skill set or strategic acumen. In addition, collaboration with star scientists may facilitate young firms to gain access to additional resources.
In relation with the evaluation of the impact of cluster policies, a number of results suggest that scientific impact may be at least as important as proximity to generate fruitful R&D cooperation with firms. As part of the evaluation of a Japanese cluster policy conducted during the early 2000s, a study has estimated the effect of the University-Industry Partnership (UIP) scheme on innovation (Nishimura and Okamuro 2010). The scheme aimed at stimulating public–private research partnerships between SMEs and ‘national’ universities, which are the highest ranking universities in the Japanese system. The analysis shows that collaborations with academic research within one region or with a neighbouring region tend to reduce the number and quality of patents12 derived from R&D spending. However, local collaboration supported by the public scheme involving a ‘national’ university has a positive effect on the number of patents. In other words, local collaborations only stimulated innovation when they involved an excellent academic partner.
This result in the case of Japan is consistent with empirical studies in European countries on the effect of public private cooperation on innovation. A study on Italian data found no significant relationship between contract research and regional R&D intensity and concluded that this fits an interpretation in which the search for academic partners extends beyond regional boundaries (Van Looy et al. 2011). More generally, empirical studies emphasize the importance of the choice of a relevant innovation partner, even if the latter is located far away. National or international rather than regional relationships may be justified by relevance, complementarity and the search for excellent partners (Miotti and Sachwald 2003; D’Este and Iammarino 2010, Fitjar and Rodríguez-Pose 2011). The positive effect of excellent science on an innovation systems absorptive capacity can extend to the facilitation of (international) partner search.
Recent contributions have further explored the impact of proximity to an excellent university on the probability of new local economic activities, either by the creation of new firms or through the attraction of new establishments. An American study shows that university patents have a positive impact on the probability of firm creation in the neighbourhood (Hausman 2012). This impact is measured by establishing a correspondence between the technologies in which universities patent and the sectors in which new companies are created. The study measured a positive impact of higher levels of federal funding (from NIH and DOD) on the creation of new companies around the beneficiary institutions. Overall, university patents have a positive impact on local employment (75 miles around the university) and a stronger impact close to the university. Competences of the university in specific technologies attract companies, with a mix of start-ups and new establishments of existing companies. This is the case in particular in the pharmaceutical industry, where R&D centres have been attracted to excellent academic institutions in the USA and in Europe.13
A recent study using data on new knowledge intensive firms (KIFs) in Italy investigates more directly the role of university knowledge with respect to start-up creation (Bonaccorsi et al. 2013). It shows that knowledge codified in academic patents from one Italian province positively affects new KIFs creation in other provinces, having a spatial range of 200 km. Knowledge codified in publications and embedded in university graduates is more localized: their effect on new KIFs creation is confined within the boundaries of the province in which universities are located. Besides, the spatial range of university knowledge is shaped by the academic quality of the universities producing this knowledge. Here again, the scientific impact of university knowledge appears to increase its economic impact. Another study on Italian data found that universities with a stronger scientific productivity seem to find themselves in an advantageous position for developing entrepreneurial activities, including both contract research and start-up creation (Van Looy et al. 2011). A positive relationship between scientific productivity and start-up creation was found both directly and indirectly through contract research. Comparing the observed relations between contract research and spin-offs the authors suggested that the former is driven more by the distributed efforts of all faculty, while the latter benefits from a more dedicated expert-oriented approach.
Few studies have been able to analyse precisely the differences between countries in technology transfer performance,14 but the few available results suggest that research performance can explain at least part of the differences.
A comparison of the propensity of universities to license in the USA and in European countries confirms the positive impact of research excellence on technology transfer according to Conti and Gaule (2011). They analyse the determinants of the transatlantic ‘licensing gap’, both in terms of numbers of licenses and in terms of licensing revenues. Their analysis shows that the number of licenses of a university depends positively on the presence of an engineering department and on the volume of scientific publications.15 but negatively on the so called ‘professor privilege’, which was still enforced in Sweden, Finland and Norway in the mid-2000s.16 The ‘licensing gap’ disappears when, on top of these determinants, the explanatory model includes the size and age of the technology transfer office (TTO). Income from licensing is also positively related to publications, and to the presence of star researchers in biomedical disciplines.17 These results are consistent with the difference between engineering disciplines, which generate numerous patents, and biomedical disciplines, which may lead to blockbusters in the pharmaceutical industry. Income from licensing is also positively impacted by the local GDP per capital,18 which the authors interpret as an indicator of demand for technology transfer. In terms of income, a transatlantic ‘licensing gap’ persists when these factors are all taken into account. The qualitative analysis of the authors relates this remaining gap to the fact that US universities tend to employ more TTO staff with industrial experience than their European counterparts.
2.2.3 Companies’ open innovation and open science strategies
Firms in high-tech sectors as well as large firms tend to rely more on R&D and access to academic research than firms from other sectors or small firms. This is related to their stronger absorptive capacity, which results from previous investments and human resource choices. Large firms are able to devote more resources to the reading of scientific publications, participation in conferences and building connections with academic researchers.
A firm’s exposure to and engagement in research activities are important predictors of its ability to exploit scientific knowledge. The opportunities for successful exploitation of scientific research are concentrated in certain sectors and activities. In knowledge-intensive sectors, firms’ open innovation activities can include participation in open science activities. Such a strategy results in close and productive links with academic research provided that the firms conform to academic rules about disclosure and publication. In particular, that these firms’ research personnel conduct basic research and can publish the results in academic journals. Simeth and Lhuillery (2015) have found that firms publishing in scientific journals are more likely to hire PhD holders in their R&D teams.
Information from patents and data on publications by firms has been used to generate indicators of knowledge flows between science and technology. The different types of indicators point to similar sets of technologies as being the closest to science: biotechnology and biomaterials, pharmaceuticals, nanotechnology, digital and basic communication, computer technology and some fields of chemistry (OECD 2013; Simeth and Raffo 2013).
The first set of indicators is based on backward citations which are used to assess an invention’s patentability and to define the legitimacy of its claims. They are also used to uncover the extent to which the patented inventions rely on science contained in non patent literature (NPL): scientific publications, conference proceedings, databases and other relevant literature. The share of NPL citations of USPTO patents strongly increased during the 1990s, when collaborative R&D grew more systematic and new sectors such as biotechnology developed (Narin et al. 1997; Mc Millan et al. 2000; EU 2003). Patents continue to cite science at the same rate and the age of cited publications is constant, indicating that new scientific results are similarly relevant for innovation (Arora et al. 2015).
The second set of indicators is based on scientific publications by firms, which represent an investment in academic work. The authors can be from the private sector only, or include both authors from a company and academics. In the biotech sector, firms that disclose valuable research results in scientific publications exhibit higher levels of innovative output (Jong and Slovova 2014). The positive relationship is measured both for corporate publications and for co-publications with academics. Besides, a more substantive publication track record is more beneficial for radical innovations as measured by New Chemical Entities (Jong and Slovova 2014).
Simeth and Cincera (2015) measure a positive impact of science-related indicators on a sample of US high-tech companies’ market valuation, beyond the effects of R&D and patent indicators. Their empirical analysis found that the active involvement in open science, as reflected by disclosure in scientific journals results in high stock-market values. The authors suggest that this positive impact stems from scientific signalling to upstream stakeholders, which allows firms to become members of the scientific community and benefit from a better access to frontier research.
These different results suggest that public–private co-publications at the national level will depend on a set of interdependent characteristics: scientific and technological specialization, high impact research and active connections between the public and private research communities. Generally, countries where the degree of public–private co-publications is more intense also tend to have two other characteristics: specialization in high-tech and high impact research output. An examination of public private co-publications by disciplines tends to confirm the set of interrelated determinants. Between 2003 and 2013, scientific publications co-authored by researchers from universities and industry in computer science were most intense in the USA (8 per cent of total publications), Finland and the Netherlands (7 per cent). In pharmacology, these co-publications were most prevalent in Denmark (15 per cent), Sweden (13 per cent), Belgium (10 per cent) and the UK (8 per cent) (OECD 2016). Public–private co-publications are more frequent in some EU countries including Denmark, the Netherlands, Sweden, Austria and Belgium than in the USA (OECD 2016). The degree of public–private co-publications is much less intense both in China and in EU countries where various other dimensions of the innovation system are less developed. The rate of public private co-publications may also be influenced by specific policy instruments.
3. How excellent is European research?
Since excellent research is one determinant of the intensity of public–private R&D cooperation, technology transfer and innovation, it is important to measure to what extent the EU scientific base generates such high impact research. Various bibliometric indicators suggest that the EU scientific performance has substantially improved since 2000. It is important to study the drivers of this favourable trend and in particular whether it could be explained by internal dynamics and public policies.
3.1 Volume and proportion of european excellent research
3.1.1 Indicators and data sources
This article looks at both size-dependent and size independent indicators of scientific impact in order to compare the EU with other regions and European countries among themselves. Two types of size independent indicators have been widely used to assess scientific impact at the level of national research systems.19 The first one is the field weighted average of relative citations indicator, a measure of the average impact of a research system's publications relative to the world average of 1. The second focuses on the top percentiles of the distribution of a country’s publications (Tijssen et al. 2002). High impact publications are defined as the proportion of a country’s scientific output which falls within the top 10 or top 1 per cent most highly cited publications worldwide – PPtop-10% or PPtop-1%. The percentile values can be compared because they are normalized among scientific fields, document types and citation windows.20 By giving a measure of the density of highly cited papers in the national production of publications, they constitute an indicator of its ‘excellence’.
This article also considers the size dependent countries’ share of the world production of highly cited publications (King 2004; Leydesdorff et al. 2014). Indeed, the volume of highly cited publications can also be an indicator of the innovative potential of a country, region or organization, in particular for incremental innovation. The size- independent indicator (PPtop10%) decreases with an increased production of research with low scientific impact. A larger proportion of low impact science in a country can increase search costs and (importantly) constitutes a draw on research funding. It is a question of debate to which extent it influences innovative potential of a system beyond that. For example, publications in trade/applied journals, which receive less citations, do not contribute to frontier science but may play a role in knowledge diffusion.
The article builds on the results of two sources of bibliometric data: the ‘Analysis and Regular Update of Bibliometric Indicators’ carried out by ScienceMetrix for the DG RTD of the European Commission (Campbell et al. 2013) and the National Science Board’s Science and Engineering Indicators (National Science Board 2016).21 Both sources are based on the Scopus database and use fractional counting, i.e. publications are assigned to countries according to the ratio with which they appear in the publication’s author addresses. Fractionalization is done at the article level: i.e. if two authors are based in France, one in the USA and one in Canada, the article is assigned for 50 per cent to France.22 However, even if produced by the same company based on the same database, the two sources do not use exactly the same methodological approach to elaborate the various indicators, which results in minor differences.23 The differences between the two sources have only limited impact on the indicators used here.
3.1.2 Trends in scientific publications by the USA, EU28 and China
The rapid growth of publications by authors located in China has radically altered the world map of scientific production. Since 2004, China has overtaken Japan as the third producer of scientific publications24 and in 2013 it published 3 times as many 10 per cent most highly cited papers than Japan25 A detailed comparison between the weight of high impact scientific production by the EU, the USA and China has to take into account different dimensions. It is necessary in particular to distinguish the dynamics of the number of publications from that of their scientific impact. For example, the world share of a country’s top-10 per cent most cited publications is the product of its share of total publications and of its PPtop-10%.
Table 2 compares the evolution of volume and impact indicators for the USA, the EU and China between 2002 and 2012. It shows that the three regions experience distinctive dynamics. In 2012, the USA remains the first producer of top-1 per cent most cited publications, but from the top-5 per cent category on, the EU has become the first producer. The EU28 also produces almost as many top-1 per cent papers as the USA does. This stands in marked contrast to the situation in 2002, when the USA produced a third more of the most highly cited papers. However, the USA share of highly cited publications in its national output remains higher than that of the EU28 from PPtop-50% upwards (Table 2). In other words, the EU produces more highly cited papers than the USA, but even more low impact publications.
Top most cited . | USA . | EU . | China . | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Share of national publications (%) . | Share of USA, EU and Chinese publications (%) . | Share of national publications (%) . | Share of USA, EU and Chinese publications (%) . | Share of national publications (%) . | Share of USA, EU and Chinese publications (%) . | |||||||
. | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . |
1% | 1.8 | 1.9 | 58.1 | 42.9 | 1.0 | 1.3 | 38.4 | 41.1 | 0.5 | 0.8 | 3.5 | 16.1 |
5% | 7.8 | 8.3 | 53.6 | 39.5 | 5.1 | 6.3 | 42.3 | 43.0 | 2.9 | 4.1 | 4.1 | 17.5 |
10% | 14.7 | 15.4 | 51.7 | 37.9 | 10.3 | 12.3 | 44.0 | 43.7 | 5.8 | 8.3 | 4.3 | 18.4 |
25% | 33.1 | 33.7 | 48.7 | 35.2 | 25.9 | 29.4 | 46.4 | 44.2 | 15.9 | 21.9 | 4.9 | 20.6 |
50% | 59.1 | 59.3 | 45.7 | 32.9 | 51.6 | 55.0 | 48.4 | 43.9 | 36.6 | 46.6 | 5.9 | 23.2 |
Top most cited . | USA . | EU . | China . | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Share of national publications (%) . | Share of USA, EU and Chinese publications (%) . | Share of national publications (%) . | Share of USA, EU and Chinese publications (%) . | Share of national publications (%) . | Share of USA, EU and Chinese publications (%) . | |||||||
. | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . |
1% | 1.8 | 1.9 | 58.1 | 42.9 | 1.0 | 1.3 | 38.4 | 41.1 | 0.5 | 0.8 | 3.5 | 16.1 |
5% | 7.8 | 8.3 | 53.6 | 39.5 | 5.1 | 6.3 | 42.3 | 43.0 | 2.9 | 4.1 | 4.1 | 17.5 |
10% | 14.7 | 15.4 | 51.7 | 37.9 | 10.3 | 12.3 | 44.0 | 43.7 | 5.8 | 8.3 | 4.3 | 18.4 |
25% | 33.1 | 33.7 | 48.7 | 35.2 | 25.9 | 29.4 | 46.4 | 44.2 | 15.9 | 21.9 | 4.9 | 20.6 |
50% | 59.1 | 59.3 | 45.7 | 32.9 | 51.6 | 55.0 | 48.4 | 43.9 | 36.6 | 46.6 | 5.9 | 23.2 |
The total number of publications for the three regions has increased by a factor 2 between 2002 and 2012.
Source: calculations from National Science Board (2016).
Top most cited . | USA . | EU . | China . | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Share of national publications (%) . | Share of USA, EU and Chinese publications (%) . | Share of national publications (%) . | Share of USA, EU and Chinese publications (%) . | Share of national publications (%) . | Share of USA, EU and Chinese publications (%) . | |||||||
. | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . |
1% | 1.8 | 1.9 | 58.1 | 42.9 | 1.0 | 1.3 | 38.4 | 41.1 | 0.5 | 0.8 | 3.5 | 16.1 |
5% | 7.8 | 8.3 | 53.6 | 39.5 | 5.1 | 6.3 | 42.3 | 43.0 | 2.9 | 4.1 | 4.1 | 17.5 |
10% | 14.7 | 15.4 | 51.7 | 37.9 | 10.3 | 12.3 | 44.0 | 43.7 | 5.8 | 8.3 | 4.3 | 18.4 |
25% | 33.1 | 33.7 | 48.7 | 35.2 | 25.9 | 29.4 | 46.4 | 44.2 | 15.9 | 21.9 | 4.9 | 20.6 |
50% | 59.1 | 59.3 | 45.7 | 32.9 | 51.6 | 55.0 | 48.4 | 43.9 | 36.6 | 46.6 | 5.9 | 23.2 |
Top most cited . | USA . | EU . | China . | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|
Share of national publications (%) . | Share of USA, EU and Chinese publications (%) . | Share of national publications (%) . | Share of USA, EU and Chinese publications (%) . | Share of national publications (%) . | Share of USA, EU and Chinese publications (%) . | |||||||
. | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . | 2002 . | 2012 . |
1% | 1.8 | 1.9 | 58.1 | 42.9 | 1.0 | 1.3 | 38.4 | 41.1 | 0.5 | 0.8 | 3.5 | 16.1 |
5% | 7.8 | 8.3 | 53.6 | 39.5 | 5.1 | 6.3 | 42.3 | 43.0 | 2.9 | 4.1 | 4.1 | 17.5 |
10% | 14.7 | 15.4 | 51.7 | 37.9 | 10.3 | 12.3 | 44.0 | 43.7 | 5.8 | 8.3 | 4.3 | 18.4 |
25% | 33.1 | 33.7 | 48.7 | 35.2 | 25.9 | 29.4 | 46.4 | 44.2 | 15.9 | 21.9 | 4.9 | 20.6 |
50% | 59.1 | 59.3 | 45.7 | 32.9 | 51.6 | 55.0 | 48.4 | 43.9 | 36.6 | 46.6 | 5.9 | 23.2 |
The total number of publications for the three regions has increased by a factor 2 between 2002 and 2012.
Source: calculations from National Science Board (2016).
China experiences the largest increase for both the size-dependent and the size-independent indicators in comparison to its low starting position. Its PPtop-1% increases by 60 per cent and PPtop-10% by 27 per cent, while its share of the three region’s top-1 per cent publications multiplied by 4.6.
Table 3 presents the size dependent and size independent indicators for the top-10 per cent most cited publications together. This allows for an easier discussion of the way trends in the number of publications and in their citation impact combine to determine the world share of highly cited publications. The table shows similar trends as Table 2 with slight differences in figures due to methods of counting.
. | Share of publications, % . | PPtop-10% . | World Share (%) of top- 10% most cited publications . | |||
---|---|---|---|---|---|---|
. | 2000 . | 2010 . | 2000 . | 2010 . | 2000 . | 2010 . |
EU28 | 34 | 28 | 9.9 | 11.4 | 33.2 | 32.1 |
USA | 28 | 21 | 14.6 | 14.6 | 41.0 | 31.0 |
China | 6 | 18 | 4.3 | 6.7 | 2.6 | 11. 9 |
World | 100 | 100 | 10 | 10 | 100 | 100 |
. | Share of publications, % . | PPtop-10% . | World Share (%) of top- 10% most cited publications . | |||
---|---|---|---|---|---|---|
. | 2000 . | 2010 . | 2000 . | 2010 . | 2000 . | 2010 . |
EU28 | 34 | 28 | 9.9 | 11.4 | 33.2 | 32.1 |
USA | 28 | 21 | 14.6 | 14.6 | 41.0 | 31.0 |
China | 6 | 18 | 4.3 | 6.7 | 2.6 | 11. 9 |
World | 100 | 100 | 10 | 10 | 100 | 100 |
Source: Author’s calculation on the basis of Scopus based data provided by ScienceMetrix to the European Commission DG Research and Innovation <http://www.science-metrix.com/en/publications/reports#/en/publications/reports/bibliometrics-and-patent-indicators- for-the-science-and-engineering-indicators>.
. | Share of publications, % . | PPtop-10% . | World Share (%) of top- 10% most cited publications . | |||
---|---|---|---|---|---|---|
. | 2000 . | 2010 . | 2000 . | 2010 . | 2000 . | 2010 . |
EU28 | 34 | 28 | 9.9 | 11.4 | 33.2 | 32.1 |
USA | 28 | 21 | 14.6 | 14.6 | 41.0 | 31.0 |
China | 6 | 18 | 4.3 | 6.7 | 2.6 | 11. 9 |
World | 100 | 100 | 10 | 10 | 100 | 100 |
. | Share of publications, % . | PPtop-10% . | World Share (%) of top- 10% most cited publications . | |||
---|---|---|---|---|---|---|
. | 2000 . | 2010 . | 2000 . | 2010 . | 2000 . | 2010 . |
EU28 | 34 | 28 | 9.9 | 11.4 | 33.2 | 32.1 |
USA | 28 | 21 | 14.6 | 14.6 | 41.0 | 31.0 |
China | 6 | 18 | 4.3 | 6.7 | 2.6 | 11. 9 |
World | 100 | 100 | 10 | 10 | 100 | 100 |
Source: Author’s calculation on the basis of Scopus based data provided by ScienceMetrix to the European Commission DG Research and Innovation <http://www.science-metrix.com/en/publications/reports#/en/publications/reports/bibliometrics-and-patent-indicators- for-the-science-and-engineering-indicators>.
China experienced a 3-fold increase in its share of world publications and an even higher increase in its share of top-10 per cent most cited publications (4.6-fold). This higher increase is due to the fact that its PPtop-10% has increased from 4.3 per cent to 6.7 per cent. Despite a reduction of its world share of publications by 21%, the EU nearly maintained its share of top-10% most cited publications (−3 per cent). This is due to the fact that its PPtop-10% has increased from 9.9 per cent to 11.4 per cent. For the USA, the decrease of both its share in total publications and of top-10 per cent most cited publications has been stronger (32–33 per cent). This is due to fact that its PPtop-10% has remained flat, while that of the EU has increased.
As a result, the EU28 now produces a larger share of the world’s top-10 per cent most cited publications than the USA does.26 For scientific results published in 2010, the share of the top-10 per cent most cited publications from both the EU28 and USA was still three times higher than that of China, which is due to its still low PPtop-10%.
The EU28 has overtaken the USA in 2008 for the number of top 10 per cent most cited publications and in 2013 for the top 1 per cent. The specific years, or even whether the USA or the EU produces most of these papers is relevant only for symbolic purposes, if any. Leydesdorff et al. (2014), who base themselves on Web of Science rather than Scopus data finds that the EU28 overtook the USA in 2010 on the top-10 per cent and is still below the USA for the share of the top-1 per cent.
National Science Board S&E indicators data show that in 2002 the EU produced much less highly cited publications than the USA in most scientific fields. However, in 2012, the EU had overtaken the USA in all fields, except in health sciences, life sciences, biology and psychology. China's position is especially strong in engineering, in which its production of top 1 per cent most cited publications is on par with that of the USA, though still somewhat behind the EU28. In chemistry, China was the leading producer of highly cited publications in 2012. In the biological, life, medical, psychology and social science fields, China is still far behind both the EU and the USA. Also on this indicator it is still more specialized in traditional and applied research fields, whereas the EU and especially the USA have invested heavily in the life sciences.27
Overall, bibliometric evidence gives an indication that while the proportion of highly cited publications in the US scientific output remains considerably higher, it is close to losing its position of first producer of the most highly cited publications28 to the EU. One does still observe a transatlantic gap with unweighted indicators, due to the strong specialization of the USA in health related sciences, where both publications and citations are relatively numerous. At the same time, China has rapidly emerged as a large producer of high impact science in some disciplines.
3.2 Drivers of the evolution of EU and US scientific performance
Different factors can contribute to explain the contrasting profiles of the USA and the EU in terms of scientific outputs since 2000. Some of these factors are internal, like policies in Europe aimed at upgrading the science base. Others are related to the global dynamics of science and increasing international cooperation. In particular, any analysis at the world level has to pay attention to the increasing weight and influence of China in world research and global scientific networks. In China as in a number of European countries, the issue of the attractiveness of the national system for global talents is high on the policy agenda (Jonkers 2010).
The higher share of world publications of the EU, including for highly cited publications, can first be explained by the larger size of the European academic research system. Actually, the data suggest a nuanced answer, depending on whether one compares the number of researchers or the amount of funding. European academic research systems employ more than twice as many researchers (870,000 versus 400,000)29 than the US system. Expenditure on public R&D represents on the contrary similar amounts (Eurostat 2016). Total expenditure is not corrected for discipline composition and differences in standards of living, but it roughly indicates that on average, European researchers benefit from lower levels of resources. As a result, it appears that the EU is more productive in volume of publications per euro spent, whereas the USA is considerably more productive in terms of high impact publications per researcher.
Many of the European comparatively low funded researchers do publish, including in non-English journals with low scientific impact as well as in trade journals. The low scientific impact of these publications may have a number of causes: they may be focused on application, involve less topical work, target specific niche areas or topics on which few scientists draw for subsequent work or do not represent a clear enough contribution. Some analysts argue that the academic culture in European research systems is less conducive to the production of high impact science than it is in the USA. This they argue may be due to weak reviewing practices, inadequate preparation of graduate students for academic publishing and lower command of research methods (Lyytinen et al. 2007). The examination of this issue is beyond the scope of this article and would require detailed analyses of different disciplines as the situation Lyytinen et al. 2007 describe above may be specific to Information Systems research.
A related issue is that of resource concentration in universities and their performance in terms of excellent research. Indeed, US universities exhibit the highest size independent scientific impact indicators (PP top-10% and PP top-1%).30 Among European universities, only a handful of British and Swiss organizations make it to the top league, followed by a somewhat larger contingent of British and continental European ones. The relative under-performance of continental European universities may be due in part to the existence of large Public Research Centres (OECD 2011) competing for research funding resources with universities. Partially, in response to the perceived ‘excellence gap’, various policies have been promoting excellence in science in Europe over the past decade. The most prominent example at the EU level has been the establishment of the European Research Council. Among the Member States, various measures have been implemented in order to promote excellence; including research performance based funding regimes. Such systems for allocating institutional funding aim to concentrate resources in the best performing organizations while providing incentives for scientific output and excellence.31 Their direct and indirect, as mediated through reputational competition, effects may offer a partial explanation for the increase in the production of highly cited articles in the EU (Jonkers and Zacharewicz 2016).
Another type of explanation can be related to the emergence of China as a major producer of scientific publications. China has increased the world total number of publications, especially at the low impact end of the distribution. In turn, this tends to push up the proportion of high impact publications of all countries whose impact indicators are higher than that of China. For example a country with a PPtop-10% of 9 per cent could experience an increase, say to 10 per cent, because China’s is lower than 9 per cent (see Table 2). This higher PPtop-10% has to be multiplied by the total number of publications of the country to get its number of top 10 per cent most cited publications. As a result, this number will grow in relation to the growth of the country’s total number of publications. It will thus grow faster for the EU than for the USA. China itself has combined a rapid growth of its total number of publications with an increase in impact, starting from a very low level.
4. Correspondance between excellence and innovation performance in EU countries
Excellent researchers move the frontier of knowledge forward and develop various interactions with their environment, in particular through teaching and cooperation with firms. As a result, they have a particularly high contribution to the overall capacity of their innovation system not only to generate knowledge but also to absorb knowledge. The creativity and tacit knowledge embedded in excellent researchers and the graduates they train can form the basis of both radical scientific breakthroughs and innovation with high socio-economic impact.
This section use a number of indicators to examine to what extend the dual impact of excellent research that is measured at the individual level (researchers, firms, universities) holds at the national or macroeconomic level.32 A full diagnosis of the contribution of the EU scientific base to knowledge transformation and innovation includes a combined analysis of specialization and excellence of European research. This section first examines the diversity of scientific impact and innovation performance among the EU28 and then focuses on the issue of the scientific specialization of European countries.
4.1 Diversity of scientific impact and innovation performance among the EU28
Figure 1 presents the evolution of the PPtop-10% indicator in EU countries, the USA and China between 2000 and 2010. EU Countries are ordered into four groups based on 2010 data: the PPtop-10% of the first group is above 12.5 per cent; that of the second group is between 12.5 per cent and the world average; that of the third group lies between 10 per cent and 7.5 per cent and that of the fourth group is lower.
The figure shows a general increase in the propensity of the Member States to produce publications with high scientific impact. In 2000, 18 Member States did not reach the world average and by 2010, 5 had crossed this threshold. Slovenia was at the same level as China in 2000, but has increased its proportion of highly cited papers even more rapidly over the decade. In the fourth group however, all countries have a lower proportion of highly cited papers than China. In the first group, the Netherlands and Denmark have experienced a substantial increase in their propensity to publish highly cited papers and reached a higher performance than the USA in 2010.
Figure 2 shows that the four groups represent very different shares of the world's highly cited publications. The first two groups account for more than 90 per cent of EU total.
The first two groups exhibit a similar evolution as the USA, i.e. a high or increasing PPtop-10% but a declining world share of highly cited publications. The grouping marks some important inter-country variation. For example the drop in world share in the first group is almost exclusively due to the drop in the share of the UK (−20 per cent) and Sweden (−5 per cent). Denmark and the Netherlands maintain their share. In the second group, one observes a large increase for Spain (+10 per cent) and Italy (+5 per cent) which is of set by a similar loss for France (−7 per cent) and Germany (−9 per cent). The third group is approaching world average PPtop-10% scores in 2010 which is reflected in their increasing world share of publications. The fourth group has a profile more comparable to that of China with both an increasing share of total publications and an increase in PPtop-10% though from a much lower base. The generally increasing trend in the EUs share of high (citation) impact publications, thus masks considerable variations between countries in both the evolution of their share as well as in the concentration of this type of papers in their total research output.
The EU is not alone in displaying a high level of heterogeneity between its constituent parts. Also in the USA, the production of high impact science is geographically concentrated. However, its proportion of highly cited papers is still higher than in the EU, and this proportion is even higher in the states where the best research universities are concentrated (IAU 2017). The USA counts a greater number of cities, like Boston, San Francisco, New York, Los Angeles, Philadelphia, Baltimore, Seattle and Chicago that concentrate a large number of high impact publications. In Europe, the cities reaching similar performance produce a smaller number of highly cited scientific publications. It is the case of Oxford, Copenhagen, Paris or Amsterdam. London and Cambridge combine relatively high scientific impact with a number of publications similar to major US academic cities, attaining similar levels of high impact science production as cities like New York and Boston. Microeconomic studies reviewed in Section 2 suggest that the proximity of universities producing excellent research output has a positive impact on innovation. Such university hot spots tend to be larger and to reach higher academic performance in the USA. A mapping of these hot spots based on the local level of the PPtop-10% indicator would be useful to make a more precise comparison between the USA and the EU.
The first group of EU countries reaches a proportion of excellent papers comparable to that of the USA (Figure 1). Among this group, in the early 2010s, Denmark and Sweden were also ‘innovation leaders’ according to the EU Innovation scoreboard, which indicates they have a similar ‘innovation performance’ as that of the USA (EU 2015). The UK, Luxembourg and Belgium were among the ‘innovation followers’. On the other hand, Finland and Germany were among the ‘innovation leaders’, while they are only in the second group in terms of scientific excellence. This can be largely explained by their very high performance with respect to business R&D and related indicators included in the synthetic indicator of the scoreboard (EU 2015). But, as for R&D intensity, a number of those other indicators are influenced by the industrial structure of the country. It is the case in particular of patent-based indicators, export based indicators and a number of the indicators on the innovation performance of SMEs. It is also the case for the scientific co-publication indicator that depends on the proportion of firms conducting R&D and among them of those conducting basic and applied research (see Section 2). A simulation of the impact of the adjustment of R&D intensity for industrial composition shows that it has an impact on the synthetic indicator (see Annex 2). With the adjustment of this single sub-indicator, Italy and Portugal gain one place in the overall ranking. This suggests that countries like Germany could have a lower rank in the scoreboard if all these indicators were to be adjusted like R&D intensity in Table 1. On the contrary, countries like the Netherlands or the UK would improve their overall score. In other words, adjusting a number of innovation indicators for the industrial composition is likely to reinforce the correlation between scientific excellence and innovation performance at the national level.
4.2 Scientific impact and technological specialization in new sectors
As emphasized above, the evolution of sector distribution is a major issue for innovation and growth in Europe. It is more specifically crucial for those Member States which are closest to the technology frontier and need to be able to generate a constant stream of innovations. Policies thus try to promote the development of high growth sectors, which tend to be more recent and relatively more connected to science. In this context, policy makers have been keen to stimulate connections between research and technology and to improve the exploitation of research results. Information & Communication technologies (ICT) and health are major research areas both in the EU and in the USA. As FP7 priorities, the evolution of the EU scientific specialization and impact in these areas is of major interest. The data on which the next paragraphs are based makes use of journal based categories for all FP7 fields, including ICT and Health of Scopus data constructed by ScienceMetrix (Annex 1).
Since the beginning of the 21st century, the EU has greatly increased its scientific production and its level of specialization in ICT. Its specialization index (SI) grew from 0.88 to 0.91. The USA lost their specialization, its SI decreasing from 1.21 to 0.65 between 2000 and 2010. China on the contrary reinforced its specialization in ICT. Trends in terms of scientific impact are quite contrasted with the trends in the production of scientific documents. The USA achieves one of its top scores for PPtop-10% in ICT and its score is well above that of the EU28. The Chinese score hardly increases; it is much lower and also lower than for the total Chinese scientific production. These trends result in large changes in world shares of highly cited publications. Whereas in 2000 the USA had a 15 percentage points higher share of world highly cited ICT publications than the EU, by 2010 the USA share was 8 percentage points below that of the EU. These data indicate that the EU science base in ICT has improved considerably, even if its PPtop-10% remains below that of the USA. The Chinese share of high impact publications increases by more than 3-fold despite the constant proportion of these publications in the national production.
The USA is more specialized in health research than the EU. The amount of funding allocated to health-related research in the USA is almost three times as high as in the EU28.33,34 The USA has a much higher PPtop-10% in the medical and life science fields and the transatlantic gap has been increasing over the last decade. As is shown by the National Science Board (2016), the USA also consistently outperforms the EU28 in the production of the 1 per cent most highly cited articles in the medical, biological and other life science fields. That being said, due to a stronger increase in the number of scientific publications in this area, the EU28 has been narrowing the transatlantic gap for the world share of top 10 per cent most cited publications, from 14 points in 2002 to 6.5 per cent percentage points in 2012. The same can be said for the biotechnology field, though here the EU28 and the USA are now equal in terms of production of highly cited articles compared to a 20 per cent point USA lead in 2000.35 Overall, in Health research, the USA remains firmly in the lead. The increase in the world share of China’s highly cited publications is fairly modest in comparison to other fields, reflecting its decreasing scientific specialization in health research.
A recent EU report pointed to a ‘strong mismatch between scientific and technological specializations in the EU’ (EU, 2014) In ‘the areas of health and ICT, there is a relatively strong scientific specialization (coupled with citation rates which are slightly above average) but a weak technological specialization. This situation compares unfavourably to the US and China where health and ICT, respectively, are areas of strong S&T co-specialization’ (EU 2014, p. 14). A number of methodological caveats36 such as the strong interrelation between patent-based indicators and industrial structure and the potentially long lags before the development of positive feedback loops between a strong science base and innovative activities materialize, suggests one needs to be careful in interpreting this analysis. If one nonetheless follows this correspondence one finds that the match is actually quite good for those Member States with high impact life science research. Denmark is, for example, strongly specialized in the life sciences, health technologies and biotechnology. Belgium and the UK also exhibit fairly good matches between health-related science and technologies. So, the idea that there would be ‘European paradoxes’ in some high technologies areas, as suggested by the ‘mismatch’ notion is not supported by the data produced by the report. Actually, despite the increase in the specialization of EU countries in ICT, there remains a transatlantic gap in the production of excellent research in both ICT and Health, in relative though in the case of ICT not in absolute terms.
5. Conclusions
From the 1990 onwards, innovation policies have developed a portfolio of instruments to foster innovation. In Europe, in particular, policies have been actively promoting technology transfer, public–private cooperation and the development of high-tech start-ups to complement the more traditional support to R&D activities. This powerful trend has generated numerous new schemes and an increasing pressure to demonstrate the economic value of research. A related tendency is the idea that research should be more applied and more directly influenced by the needs of firms and of the society. This has sometimes been summarized by the idea that innovation policies should be less ‘techno-push’ and more ‘demand-pull’. As a result, proposals for research projects may have to anticipate their potential economic or societal impact.
These policy trends are not specific to Europe. In Europe, however, they have been reinforced by the notion of a ‘European paradox’, which has been a very influential narrative in policy circles for nearly two decades. As early as 2000, Keith Pavitt had proposed an alternative narrative of the transatlantic innovation gap by pointing at the European deficit in high scientific impact publications and universities rather than at the R&D deficit. Despite the development of a number of schemes to promote excellent universities and high impact research, the ‘European paradox’ narrative has remained influential, in particular to support technology transfer and innovation policies. This may result in unbalanced and inefficient policies.
This article has developed an alternative narrative by drawing on two strands of the literature: empirical studies of the interactions between excellent research and innovation-based microeconomic data; bibliometric analyses based on more aggregated data. Our review of the empirical literature on the interactions between academic researchers and firms suggests that excellent research, even basic research, is attractive to companies and tends to have a strong scientific impact as well as a potential economic impact through high-tech innovation performance. This emerges from numerous empirical studies of technology transfer and public–private co- operations and is of particular interest for the promotion of innovation and the development of high growth innovative start-ups. Indeed, it means that research with high scientific impact can also generate higher economic returns. The opposition here is not between short- and long-term projects, but between low and high scientific impact research. Policies fostering excellent research in Europe can be beneficial to raise the overall level of absorptive capacity in European innovation systems while generating a greater number of hot spots where a high level of high impact science production is concentrated. Some of the resources to achieve this might be found in reducing the resources devoted to the production of low impact science, unless contrary findings suggests this type of research has clear added value on innovation, teaching or alternative societal ends beyond high tech innovation.
This article aims at contributing to evidence based policy making. Its policy implications are 3-fold. First, the dual (scientific and economic) impact of excellent research suggests a careful review of the policies aimed at increasing the economic and social impact of academic research, so that new policies or strategic orientations in favour of applied and mission oriented research do not endanger the funding of excellent research. Secondly, given the influence of rankings and scoreboards on policy makers, it is necessary to ensure a better measurement of innovation and its determinants through composite indexes, in particular to take into account the influence of economic structures. Such a review also implies a careful check of the correlation of individual indicators on synthetic indicators. In the case of the EU policies, such an exercise will allow a better analysis of intra EU diversity. Thirdly, whenever systems target high levels of high-tech innovation, scientific excellence should remain a major evaluation indicator. More qualitative impact studies can bring additional qualitative insight in how the produced scientific knowledge has a societal or economic impact. Designed well such societal impact assessments do not have to contradict incentives towards scientific excellence.
Footnotes
The data and analysis presented in this article do not necessarily represent the views of the European Commission or HCERES. These organizations nor anyone acting on its behalf can be held responsible for any use made thereof.
In this article, ‘excellence’ in research is used interchangeably with ‘high academic impact’ research and highly cited publications.
This second point is discussed and related to other indicators in Section 4.
Veugelers and Cincera (2010) have coined the term ‘yollies’ for young innovative firms aged less than 25. Yollies are not young start-ups: on average these firms have 10,000 employees.
EU-1000 and non EU-1000 largest R&D spenders as monitored by the European Industrial R&D Scoreboard (Cincera and Veugelers 2014).
See for example, <https://www.era-learn.eu/manuals-tools/smart-coordination/positioning-of-the-era-net-scheme/other-methods-of-coordination/eit-kics> accessed 20 Feb 2016; <https://www.ffg.at/sites/default/files/03_eit_ict_labs_2014_udo_bub.pdf> accessed 20 Feb 2016; <http://ec.europa.eu/education/eit/index_en.html> accessed 20 Feb 2016.
It was for example mentioned by the EU research Commissioner Carlos Moedas during the 2016 Science Business Horizon 2020 conference (Kelly 2016).
Not necessarily from the same region, at least for strategic or important R&D partnerships (Dhont-Peltrault 2005). See also below on the interaction between the proximity of the academic partner and its academic profile.
Proximity to science has been analysed by sectors studies and examination of prior art in patents (Laursen and Salter 2004; Callaert et al. 2006).
More than 60 per cent of biotechnology, biomaterials and pharmaceutical patents cite patents with non-patent literature (NPL) references, while 50 per cent of patents in digital communication and 40% of patents in organic and food chemistry, nanotechnology and ICT related technologies do so (OECD 2013). On the contrary, engineering, mechanical and transport technologies hardly ever cite patents with NPL.
As measured by the number of claims or the number of forward citations.
One example is the investment of 0.5 bn USD by AstraZeneca in the Cambridge Biomedical Campus announced on July 2014 (see also: Garnsey and Heffernan 2005).
Jacobsson et al. (2013) emphasize the lack of comparable indicators in this area for example
Number of articles in science and engineering published in 2004–2006 as reported in the ISI Web of Science.
Corresponding to the data used in the article.
A star is defined as appearing on the ISI Web of Science list of highly cited researchers.
Country level for Europe and state level for the USA.
The use of citation indicators to assess the scientific impact of science has well-known limitations and bibliometric data should be interpreted with caution (e.g. Tijssen et al. 2002; Wilsdon 2016). At the aggregate level of research systems most of these are less pronounced than in the case of individual or organisational level analysis. The use of quantitative indicators is considered a challenge especially in the social sciences and humanities for several reasons, including the fact that a lot of research is published in non-English language journals.
Each publication’s citation count is divided by the average citation count of all publications of the corresponding document type that were published in the same year in the same field.
www.scival.com. These data were downloaded in February 2016. Scival © 2016 Elsevier B.V. All rights reserved. SciVal® is a registered trademark of Reed Elsevier Properties S.A., used under license.
There appear to be some unexplained differences between the number of publications taken into account in the ScienceMetrix studies carried out for the European Commission and the National Science Board.
Glänzel et al. (2008) considered that the triad had then become a tetrad.
OECD (2016) adapted from <http://dx.doi.org/10.1787/888933433519>
The indicator is field normalized. Considering the greater specialization of the USA in the Health Sciences field the balance would be different if no field normalisation would have been applied.
These data also illustrate that while preferable for some purposes PPtop-1% can lead to conclusions that are not necessarily an accurate reflection of reality. For example the National Science Board S&E indicator report remarks on the strong performance of China on this indicator in the social sciences. However, the absolute share of highly cited publications made in this field by authors based in China remains quite small.
Using the field weighted total number of highly cited publications as an indicator.
Comparable data on non-business researchers for the EU28 and USA were not readily available. The number of non-business researchers was calculated based the OECD STI Scoreboard 2015 (OECD 2015) for the USA and estimated from Eurostat data for the EU.
See for example, the CWTS Leiden University ranking <http://www.leidenranking.com/> accessed January 2017; the Scimago institutions ranking <http://www.scimagoir.com/> accessed January 2017; or for a complementary perspective OECD (2016).
The USA, Switzerland and the Netherlands have alternative ways of concentrating institutional funding in top level research universities: they have binary university systems.
In the previous section, high impact researchers were identified either through bibliometric indicators or through the propensity to get research grants. This section relies on bibliometric indicators which are more widely available and allow more comparisons between countries and disciplines.
Eurostat GBAORD by NABS 2007. The definitions used in the USA and EU28 differ, which may influence reported data. OECD GBAORD data by NABS also indicates that the USA allocates around 50 per cent of its civil R&D budget to health related fields.
The comparison is not available for the number of researchers working in this research area.
Author's calculation on the basis of Scopus-based data provided by ScienceMetrix (Canada) to the European Commission DG Research and Innovation
Matching patent and publication classified as falling within FP7 thematic domains intuitively appears a useful way to explore the potential mismatch between scientific and technological specialization. There is some doubt to which extent it really is. Patenting in the field of transport for example may not rely too much on scientific publications classified as belonging to this field. A stronger case may be made for health research. In ICT, Bonaccorsi (2011) argues that it is especially physics papers which are heavily cited in the ICT patent literature. Whether publications in more basic scientific fields are also classified to this FP7 classification would require some further scrutiny of the classification.
References
ICT . | Specialization Index . | PPtop-10% . | World share of 10% most cited publications (%) . | |||
---|---|---|---|---|---|---|
2000 . | 2010 . | 2000 . | 2010 . | 2000 . | 2010 . | |
Austria | 0.83 | 1.36 | 12.50 | 12.52 | 0.6 | 1.0 |
Belgium | 0.84 | 0.79 | 11.19 | 15.12 | 0.9 | 0.9 |
Denmark | 0.65 | 0.64 | 9.30 | 13.70 | 0.4 | 0.5 |
Finland | 1.45 | 1.36 | 12.34 | 10.93 | 1.2 | 0.8 |
France | 0.86 | 0.97 | 9.3 | 11.86 | 3.7 | 4.1 |
Germany | 0.81 | 0.90 | 9.54 | 11.34 | 4.9 | 4.9 |
Greece | 2.29 | 1.38 | 7.67 | 12.88 | 0.8 | 1.1 |
Italy | 1.11 | 0.89 | 10.98 | 14.99 | 4.0 | 3.9 |
Netherlands | 0.83 | 0.78 | 14.16 | 13.89 | 2.0 | 1.6 |
Poland | 0.5 | 0.73 | 2.41 | 8.52 | 0.1 | 0.7 |
Portugal | 1.48 | 1.28 | 6.15 | 7.92 | 0.3 | 0.5 |
Spain | 0.87 | 0.97 | 7.80 | 14.15 | 1.6 | 3.5 |
Sweden | 0.79 | 0.77 | 9.79 | 11.78 | 1.1 | 0.8 |
UK | 0.8 | 0.73 | 10.27 | 14.22 | 5.9 | 5.1 |
EU28 | 0.88 | 0.91 | 9.53 | 12.18 | 28.6 | 31.6 |
USA | 1.12 | 0.65 | 14.79 | 15.09 | 45.5 | 24.0 |
China | 1.48 | 1.79 | 5.64 | 5.66 | 5.0 | 18.0 |
World | 1 | 1 | 10 | 10 | 100 | 100 |
ICT . | Specialization Index . | PPtop-10% . | World share of 10% most cited publications (%) . | |||
---|---|---|---|---|---|---|
2000 . | 2010 . | 2000 . | 2010 . | 2000 . | 2010 . | |
Austria | 0.83 | 1.36 | 12.50 | 12.52 | 0.6 | 1.0 |
Belgium | 0.84 | 0.79 | 11.19 | 15.12 | 0.9 | 0.9 |
Denmark | 0.65 | 0.64 | 9.30 | 13.70 | 0.4 | 0.5 |
Finland | 1.45 | 1.36 | 12.34 | 10.93 | 1.2 | 0.8 |
France | 0.86 | 0.97 | 9.3 | 11.86 | 3.7 | 4.1 |
Germany | 0.81 | 0.90 | 9.54 | 11.34 | 4.9 | 4.9 |
Greece | 2.29 | 1.38 | 7.67 | 12.88 | 0.8 | 1.1 |
Italy | 1.11 | 0.89 | 10.98 | 14.99 | 4.0 | 3.9 |
Netherlands | 0.83 | 0.78 | 14.16 | 13.89 | 2.0 | 1.6 |
Poland | 0.5 | 0.73 | 2.41 | 8.52 | 0.1 | 0.7 |
Portugal | 1.48 | 1.28 | 6.15 | 7.92 | 0.3 | 0.5 |
Spain | 0.87 | 0.97 | 7.80 | 14.15 | 1.6 | 3.5 |
Sweden | 0.79 | 0.77 | 9.79 | 11.78 | 1.1 | 0.8 |
UK | 0.8 | 0.73 | 10.27 | 14.22 | 5.9 | 5.1 |
EU28 | 0.88 | 0.91 | 9.53 | 12.18 | 28.6 | 31.6 |
USA | 1.12 | 0.65 | 14.79 | 15.09 | 45.5 | 24.0 |
China | 1.48 | 1.79 | 5.64 | 5.66 | 5.0 | 18.0 |
World | 1 | 1 | 10 | 10 | 100 | 100 |
In the EU, countries with more than 1100 publications in 2013.
Information and Communication Technologies corresponding to the FP7 priority, see Campbell et al. (2013).
ICT . | Specialization Index . | PPtop-10% . | World share of 10% most cited publications (%) . | |||
---|---|---|---|---|---|---|
2000 . | 2010 . | 2000 . | 2010 . | 2000 . | 2010 . | |
Austria | 0.83 | 1.36 | 12.50 | 12.52 | 0.6 | 1.0 |
Belgium | 0.84 | 0.79 | 11.19 | 15.12 | 0.9 | 0.9 |
Denmark | 0.65 | 0.64 | 9.30 | 13.70 | 0.4 | 0.5 |
Finland | 1.45 | 1.36 | 12.34 | 10.93 | 1.2 | 0.8 |
France | 0.86 | 0.97 | 9.3 | 11.86 | 3.7 | 4.1 |
Germany | 0.81 | 0.90 | 9.54 | 11.34 | 4.9 | 4.9 |
Greece | 2.29 | 1.38 | 7.67 | 12.88 | 0.8 | 1.1 |
Italy | 1.11 | 0.89 | 10.98 | 14.99 | 4.0 | 3.9 |
Netherlands | 0.83 | 0.78 | 14.16 | 13.89 | 2.0 | 1.6 |
Poland | 0.5 | 0.73 | 2.41 | 8.52 | 0.1 | 0.7 |
Portugal | 1.48 | 1.28 | 6.15 | 7.92 | 0.3 | 0.5 |
Spain | 0.87 | 0.97 | 7.80 | 14.15 | 1.6 | 3.5 |
Sweden | 0.79 | 0.77 | 9.79 | 11.78 | 1.1 | 0.8 |
UK | 0.8 | 0.73 | 10.27 | 14.22 | 5.9 | 5.1 |
EU28 | 0.88 | 0.91 | 9.53 | 12.18 | 28.6 | 31.6 |
USA | 1.12 | 0.65 | 14.79 | 15.09 | 45.5 | 24.0 |
China | 1.48 | 1.79 | 5.64 | 5.66 | 5.0 | 18.0 |
World | 1 | 1 | 10 | 10 | 100 | 100 |
ICT . | Specialization Index . | PPtop-10% . | World share of 10% most cited publications (%) . | |||
---|---|---|---|---|---|---|
2000 . | 2010 . | 2000 . | 2010 . | 2000 . | 2010 . | |
Austria | 0.83 | 1.36 | 12.50 | 12.52 | 0.6 | 1.0 |
Belgium | 0.84 | 0.79 | 11.19 | 15.12 | 0.9 | 0.9 |
Denmark | 0.65 | 0.64 | 9.30 | 13.70 | 0.4 | 0.5 |
Finland | 1.45 | 1.36 | 12.34 | 10.93 | 1.2 | 0.8 |
France | 0.86 | 0.97 | 9.3 | 11.86 | 3.7 | 4.1 |
Germany | 0.81 | 0.90 | 9.54 | 11.34 | 4.9 | 4.9 |
Greece | 2.29 | 1.38 | 7.67 | 12.88 | 0.8 | 1.1 |
Italy | 1.11 | 0.89 | 10.98 | 14.99 | 4.0 | 3.9 |
Netherlands | 0.83 | 0.78 | 14.16 | 13.89 | 2.0 | 1.6 |
Poland | 0.5 | 0.73 | 2.41 | 8.52 | 0.1 | 0.7 |
Portugal | 1.48 | 1.28 | 6.15 | 7.92 | 0.3 | 0.5 |
Spain | 0.87 | 0.97 | 7.80 | 14.15 | 1.6 | 3.5 |
Sweden | 0.79 | 0.77 | 9.79 | 11.78 | 1.1 | 0.8 |
UK | 0.8 | 0.73 | 10.27 | 14.22 | 5.9 | 5.1 |
EU28 | 0.88 | 0.91 | 9.53 | 12.18 | 28.6 | 31.6 |
USA | 1.12 | 0.65 | 14.79 | 15.09 | 45.5 | 24.0 |
China | 1.48 | 1.79 | 5.64 | 5.66 | 5.0 | 18.0 |
World | 1 | 1 | 10 | 10 | 100 | 100 |
In the EU, countries with more than 1100 publications in 2013.
Information and Communication Technologies corresponding to the FP7 priority, see Campbell et al. (2013).
Health . | Specialization index . | PPtop-10% . | World share of 10% most cited publications (%) . | |||
---|---|---|---|---|---|---|
2000 . | 2010 . | 2000 . | 2010 . | 2000 . | 2010 . | |
Austria | 1.21 | 1.1 | 9.30 | 11.35 | 0.7 | 0.7 |
Belgium | 1.13 | 1.19 | 9.60 | 13.38 | 1.0 | 1.2 |
Denmark | 1.17 | 1.42 | 11.34 | 14.65 | 0.9 | 1.1 |
Finland | 1.11 | 1.00 | 13.63 | 13.16 | 1.0 | 0.6 |
France | 1.06 | 1.08 | 7.63 | 8.62 | 3.7 | 3.2 |
Germany | 1.06 | 1.18 | 9.10 | 11.3 | 6.0 | 6.1 |
Greece | 0.88 | 1.16 | 4.29 | 7.02 | 0.2 | 0.5 |
Italy | 1.16 | 1.24 | 6.95 | 10.00 | 2.6 | 3.5 |
Netherlands | 1.21 | 1.47 | 12.53 | 16.76 | 2.6 | 3.5 |
Poland | 0.71 | 1.01 | 1.64 | 2.23 | 0.1 | 0.3 |
Portugal | 0.64 | 0.75 | 3.72 | 6.46 | 0.1 | 0.2 |
Spain | 1.1 | 1.12 | 4.95 | 7.26 | 1.3 | 2.0 |
Sweden | 1.21 | 1.31 | 11.47 | 12.87 | 1.9 | 1.5 |
UK | 1.14 | 1.25 | 11.85 | 13.25 | 9.6 | 7.9 |
EU28 | 1.08 | 1.15 | 8.92 | 10.53 | 32.2 | 33.4 |
USA | 1.13 | 1.28 | 14.28 | 17.31 | 46.4 | 39.8 |
China | 0.52 | 0.45 | 2.91 | 5.65 | 1 | 4 |
World | 1 | 1 | 10 | 10 | 100 | 100 |
Health . | Specialization index . | PPtop-10% . | World share of 10% most cited publications (%) . | |||
---|---|---|---|---|---|---|
2000 . | 2010 . | 2000 . | 2010 . | 2000 . | 2010 . | |
Austria | 1.21 | 1.1 | 9.30 | 11.35 | 0.7 | 0.7 |
Belgium | 1.13 | 1.19 | 9.60 | 13.38 | 1.0 | 1.2 |
Denmark | 1.17 | 1.42 | 11.34 | 14.65 | 0.9 | 1.1 |
Finland | 1.11 | 1.00 | 13.63 | 13.16 | 1.0 | 0.6 |
France | 1.06 | 1.08 | 7.63 | 8.62 | 3.7 | 3.2 |
Germany | 1.06 | 1.18 | 9.10 | 11.3 | 6.0 | 6.1 |
Greece | 0.88 | 1.16 | 4.29 | 7.02 | 0.2 | 0.5 |
Italy | 1.16 | 1.24 | 6.95 | 10.00 | 2.6 | 3.5 |
Netherlands | 1.21 | 1.47 | 12.53 | 16.76 | 2.6 | 3.5 |
Poland | 0.71 | 1.01 | 1.64 | 2.23 | 0.1 | 0.3 |
Portugal | 0.64 | 0.75 | 3.72 | 6.46 | 0.1 | 0.2 |
Spain | 1.1 | 1.12 | 4.95 | 7.26 | 1.3 | 2.0 |
Sweden | 1.21 | 1.31 | 11.47 | 12.87 | 1.9 | 1.5 |
UK | 1.14 | 1.25 | 11.85 | 13.25 | 9.6 | 7.9 |
EU28 | 1.08 | 1.15 | 8.92 | 10.53 | 32.2 | 33.4 |
USA | 1.13 | 1.28 | 14.28 | 17.31 | 46.4 | 39.8 |
China | 0.52 | 0.45 | 2.91 | 5.65 | 1 | 4 |
World | 1 | 1 | 10 | 10 | 100 | 100 |
In the EU, countries with more than 3600 publications in 2010.
Health corresponding to the FP7 priority, see Campbell et al. (2013).
Source: Author’s calculation on the basis of Scopus based data provided by ScienceMetrix (Canada) to the European Commission DG Research and Innovation.
Annex 1a and 1b only give details for Member states with more than 1100 and 3600 scientific publications in 2010 in the respective fields. These countries represent the very large majority of total EU publications (89% and 93%, respectively) and thus when European averages are given in terms of specialization or impact, it largely reflects the position of the countries included in the tables.
Health . | Specialization index . | PPtop-10% . | World share of 10% most cited publications (%) . | |||
---|---|---|---|---|---|---|
2000 . | 2010 . | 2000 . | 2010 . | 2000 . | 2010 . | |
Austria | 1.21 | 1.1 | 9.30 | 11.35 | 0.7 | 0.7 |
Belgium | 1.13 | 1.19 | 9.60 | 13.38 | 1.0 | 1.2 |
Denmark | 1.17 | 1.42 | 11.34 | 14.65 | 0.9 | 1.1 |
Finland | 1.11 | 1.00 | 13.63 | 13.16 | 1.0 | 0.6 |
France | 1.06 | 1.08 | 7.63 | 8.62 | 3.7 | 3.2 |
Germany | 1.06 | 1.18 | 9.10 | 11.3 | 6.0 | 6.1 |
Greece | 0.88 | 1.16 | 4.29 | 7.02 | 0.2 | 0.5 |
Italy | 1.16 | 1.24 | 6.95 | 10.00 | 2.6 | 3.5 |
Netherlands | 1.21 | 1.47 | 12.53 | 16.76 | 2.6 | 3.5 |
Poland | 0.71 | 1.01 | 1.64 | 2.23 | 0.1 | 0.3 |
Portugal | 0.64 | 0.75 | 3.72 | 6.46 | 0.1 | 0.2 |
Spain | 1.1 | 1.12 | 4.95 | 7.26 | 1.3 | 2.0 |
Sweden | 1.21 | 1.31 | 11.47 | 12.87 | 1.9 | 1.5 |
UK | 1.14 | 1.25 | 11.85 | 13.25 | 9.6 | 7.9 |
EU28 | 1.08 | 1.15 | 8.92 | 10.53 | 32.2 | 33.4 |
USA | 1.13 | 1.28 | 14.28 | 17.31 | 46.4 | 39.8 |
China | 0.52 | 0.45 | 2.91 | 5.65 | 1 | 4 |
World | 1 | 1 | 10 | 10 | 100 | 100 |
Health . | Specialization index . | PPtop-10% . | World share of 10% most cited publications (%) . | |||
---|---|---|---|---|---|---|
2000 . | 2010 . | 2000 . | 2010 . | 2000 . | 2010 . | |
Austria | 1.21 | 1.1 | 9.30 | 11.35 | 0.7 | 0.7 |
Belgium | 1.13 | 1.19 | 9.60 | 13.38 | 1.0 | 1.2 |
Denmark | 1.17 | 1.42 | 11.34 | 14.65 | 0.9 | 1.1 |
Finland | 1.11 | 1.00 | 13.63 | 13.16 | 1.0 | 0.6 |
France | 1.06 | 1.08 | 7.63 | 8.62 | 3.7 | 3.2 |
Germany | 1.06 | 1.18 | 9.10 | 11.3 | 6.0 | 6.1 |
Greece | 0.88 | 1.16 | 4.29 | 7.02 | 0.2 | 0.5 |
Italy | 1.16 | 1.24 | 6.95 | 10.00 | 2.6 | 3.5 |
Netherlands | 1.21 | 1.47 | 12.53 | 16.76 | 2.6 | 3.5 |
Poland | 0.71 | 1.01 | 1.64 | 2.23 | 0.1 | 0.3 |
Portugal | 0.64 | 0.75 | 3.72 | 6.46 | 0.1 | 0.2 |
Spain | 1.1 | 1.12 | 4.95 | 7.26 | 1.3 | 2.0 |
Sweden | 1.21 | 1.31 | 11.47 | 12.87 | 1.9 | 1.5 |
UK | 1.14 | 1.25 | 11.85 | 13.25 | 9.6 | 7.9 |
EU28 | 1.08 | 1.15 | 8.92 | 10.53 | 32.2 | 33.4 |
USA | 1.13 | 1.28 | 14.28 | 17.31 | 46.4 | 39.8 |
China | 0.52 | 0.45 | 2.91 | 5.65 | 1 | 4 |
World | 1 | 1 | 10 | 10 | 100 | 100 |
In the EU, countries with more than 3600 publications in 2010.
Health corresponding to the FP7 priority, see Campbell et al. (2013).
Source: Author’s calculation on the basis of Scopus based data provided by ScienceMetrix (Canada) to the European Commission DG Research and Innovation.
Annex 1a and 1b only give details for Member states with more than 1100 and 3600 scientific publications in 2010 in the respective fields. These countries represent the very large majority of total EU publications (89% and 93%, respectively) and thus when European averages are given in terms of specialization or impact, it largely reflects the position of the countries included in the tables.
1. SII with market sector R&D . | 2. SII with market sector R&D adjusted (Table 1) . | Change in SII with R&D adjustment (1–2) in % . | ||||
---|---|---|---|---|---|---|
1 | Sweden | 0.732 | Sweden | 0.730 | 1 | −0.33 |
2 | Denmark | 0.699 | Denmark | 0.701 | 2 | 0.27 |
3 | Germany | 0.645 | Germany | 0.635 | 3 | −1.52 |
4 | Finland | 0.629 | Finland | 0.629 | 4 | 0.00 |
5 | Netherlands | 0.613 | Netherlands | 0.619 | 5 | 1.07 |
6 | Austria | 0.565 | Austria | 0.572 | 6 | 1.39 |
7 | Ireland | 0.562 | Ireland | 0.563 | 7 | 0.14 |
8 | Belgium | 0.544 | Belgium | 0.549 | 8 | 0.92 |
9 | United Kingdom | 0.533 | UK | 0.538 | 9 | 0.99 |
10 | France | 0.508 | France | 0.519 | 10 | 2.12 |
11 | Estonia | 0.448 | Estonia | 0.448 | 11 | 0.11 |
12 | Slovenia | 0.407 | Norway | 0.403 | 12 | −1.72 |
13 | Norway | 0.394 | Slovenia | 0.400 | 13 | 2.28 |
14 | Czech Republic | 0.346 | Italy | 0.350 | 14 | −1.16 |
15 | Italy | 0.346 | Portugal | 0.342 | 15 | 1.16 |
16 | Portugal | 0.336 | Czech Republic | 0.342 | 16 | 1.19 |
17 | Spain | 0.314 | Spain | 0.318 | 17 | 1.27 |
18 | Greece | 0.301 | Greece | 0.307 | 18 | 1.99 |
19 | Slovakia | 0.263 | Slovakia | 0.263 | 19 | 0.00 |
20 | Hungary | 0.251 | Hungary | 0.246 | 20 | −1.99 |
21 | Poland | 0.219 | Poland | 0.218 | 21 | −0.46 |
1. SII with market sector R&D . | 2. SII with market sector R&D adjusted (Table 1) . | Change in SII with R&D adjustment (1–2) in % . | ||||
---|---|---|---|---|---|---|
1 | Sweden | 0.732 | Sweden | 0.730 | 1 | −0.33 |
2 | Denmark | 0.699 | Denmark | 0.701 | 2 | 0.27 |
3 | Germany | 0.645 | Germany | 0.635 | 3 | −1.52 |
4 | Finland | 0.629 | Finland | 0.629 | 4 | 0.00 |
5 | Netherlands | 0.613 | Netherlands | 0.619 | 5 | 1.07 |
6 | Austria | 0.565 | Austria | 0.572 | 6 | 1.39 |
7 | Ireland | 0.562 | Ireland | 0.563 | 7 | 0.14 |
8 | Belgium | 0.544 | Belgium | 0.549 | 8 | 0.92 |
9 | United Kingdom | 0.533 | UK | 0.538 | 9 | 0.99 |
10 | France | 0.508 | France | 0.519 | 10 | 2.12 |
11 | Estonia | 0.448 | Estonia | 0.448 | 11 | 0.11 |
12 | Slovenia | 0.407 | Norway | 0.403 | 12 | −1.72 |
13 | Norway | 0.394 | Slovenia | 0.400 | 13 | 2.28 |
14 | Czech Republic | 0.346 | Italy | 0.350 | 14 | −1.16 |
15 | Italy | 0.346 | Portugal | 0.342 | 15 | 1.16 |
16 | Portugal | 0.336 | Czech Republic | 0.342 | 16 | 1.19 |
17 | Spain | 0.314 | Spain | 0.318 | 17 | 1.27 |
18 | Greece | 0.301 | Greece | 0.307 | 18 | 1.99 |
19 | Slovakia | 0.263 | Slovakia | 0.263 | 19 | 0.00 |
20 | Hungary | 0.251 | Hungary | 0.246 | 20 | −1.99 |
21 | Poland | 0.219 | Poland | 0.218 | 21 | −0.46 |
1. SII with market sector R&D . | 2. SII with market sector R&D adjusted (Table 1) . | Change in SII with R&D adjustment (1–2) in % . | ||||
---|---|---|---|---|---|---|
1 | Sweden | 0.732 | Sweden | 0.730 | 1 | −0.33 |
2 | Denmark | 0.699 | Denmark | 0.701 | 2 | 0.27 |
3 | Germany | 0.645 | Germany | 0.635 | 3 | −1.52 |
4 | Finland | 0.629 | Finland | 0.629 | 4 | 0.00 |
5 | Netherlands | 0.613 | Netherlands | 0.619 | 5 | 1.07 |
6 | Austria | 0.565 | Austria | 0.572 | 6 | 1.39 |
7 | Ireland | 0.562 | Ireland | 0.563 | 7 | 0.14 |
8 | Belgium | 0.544 | Belgium | 0.549 | 8 | 0.92 |
9 | United Kingdom | 0.533 | UK | 0.538 | 9 | 0.99 |
10 | France | 0.508 | France | 0.519 | 10 | 2.12 |
11 | Estonia | 0.448 | Estonia | 0.448 | 11 | 0.11 |
12 | Slovenia | 0.407 | Norway | 0.403 | 12 | −1.72 |
13 | Norway | 0.394 | Slovenia | 0.400 | 13 | 2.28 |
14 | Czech Republic | 0.346 | Italy | 0.350 | 14 | −1.16 |
15 | Italy | 0.346 | Portugal | 0.342 | 15 | 1.16 |
16 | Portugal | 0.336 | Czech Republic | 0.342 | 16 | 1.19 |
17 | Spain | 0.314 | Spain | 0.318 | 17 | 1.27 |
18 | Greece | 0.301 | Greece | 0.307 | 18 | 1.99 |
19 | Slovakia | 0.263 | Slovakia | 0.263 | 19 | 0.00 |
20 | Hungary | 0.251 | Hungary | 0.246 | 20 | −1.99 |
21 | Poland | 0.219 | Poland | 0.218 | 21 | −0.46 |
1. SII with market sector R&D . | 2. SII with market sector R&D adjusted (Table 1) . | Change in SII with R&D adjustment (1–2) in % . | ||||
---|---|---|---|---|---|---|
1 | Sweden | 0.732 | Sweden | 0.730 | 1 | −0.33 |
2 | Denmark | 0.699 | Denmark | 0.701 | 2 | 0.27 |
3 | Germany | 0.645 | Germany | 0.635 | 3 | −1.52 |
4 | Finland | 0.629 | Finland | 0.629 | 4 | 0.00 |
5 | Netherlands | 0.613 | Netherlands | 0.619 | 5 | 1.07 |
6 | Austria | 0.565 | Austria | 0.572 | 6 | 1.39 |
7 | Ireland | 0.562 | Ireland | 0.563 | 7 | 0.14 |
8 | Belgium | 0.544 | Belgium | 0.549 | 8 | 0.92 |
9 | United Kingdom | 0.533 | UK | 0.538 | 9 | 0.99 |
10 | France | 0.508 | France | 0.519 | 10 | 2.12 |
11 | Estonia | 0.448 | Estonia | 0.448 | 11 | 0.11 |
12 | Slovenia | 0.407 | Norway | 0.403 | 12 | −1.72 |
13 | Norway | 0.394 | Slovenia | 0.400 | 13 | 2.28 |
14 | Czech Republic | 0.346 | Italy | 0.350 | 14 | −1.16 |
15 | Italy | 0.346 | Portugal | 0.342 | 15 | 1.16 |
16 | Portugal | 0.336 | Czech Republic | 0.342 | 16 | 1.19 |
17 | Spain | 0.314 | Spain | 0.318 | 17 | 1.27 |
18 | Greece | 0.301 | Greece | 0.307 | 18 | 1.99 |
19 | Slovakia | 0.263 | Slovakia | 0.263 | 19 | 0.00 |
20 | Hungary | 0.251 | Hungary | 0.246 | 20 | −1.99 |
21 | Poland | 0.219 | Poland | 0.218 | 21 | −0.46 |
Author notes
Koen Jonkers and Frédérique Sachwald are joint first authors.
This article belongs to the Special Issue ‘Towards evidence-based industrial research and innovation policy’.