„Internationalisation of EU Research Organizations: A Bibliometric Stocktaking Study”, written at the request of the European Parliament (114 pp., 2019) – here.
Full document in English is here: http://www.europarl.europa.eu/stoa/en/document/EPRS_STU(2019)634444
A Debate in the European Parliament with Professor Kwiek in February 2019 is here.
International research collaboration (IRC) is at the core of contemporary higher education and science systems, and the percentage of internationally co-authored publications globally and across Europe is on the rise. The aim of this study is to analyse, based on large-scale data on publication and citation trends over time (within the last decade), the changing nature of academic knowledge production in all European Union Member States (EU-28) and the trend towards its radically increasing internationalisation.
The study combines theoretical knowledge about IRC with the most up-to-date empirical data and their analysis. This quantitative study analyses the macro-level of countries and the meso-level of flagship institutions to assess the cross-national and cross-institutional differentiation in the pace of these changes and their depth. The report uses Scopus and SciVal data for 2007-2017, and the analysis of collaboration in research is based on bibliometric data on publications and citations.
The empirical analysis is preceded by a section on the motivations and another on the major barriers connected with the processes of research internationalisation. The study suggests policy options to improve international research collaboration at the European level.
International research collaboration (IRC) is at the core of contemporary higher education and science systems. The percentage of internationally co-authored publications globally and across Europe has been on the rise, as has been the mean distance between collaborating scientists. The present study is theoretically grounded in the global research literature concerning IRC (its motivations and drivers, advantages, costs, and major barriers) and its empirical part is used to support selected findings from previous research. In this way, the report combines theoretical knowledge about IRC with the most up-to-date empirical data and its analysis.
The aim of the study is to analyse, based on large-scale data on publication and citation trends over time (within the last decade), the changing nature of academic knowledge production in all European Union Member States (EU-28) towards its radically increasing internationalisation. This quantitative study analyses the macro-level of countries and the meso-level of institutions to assess the cross-national and cross-institutional differentiation in the pace of these changes and their depth. The study examines bibliometric data about the internationalisation of research in the theoretical context of international scientific collaboration literature and suggests policy options concerning its improvement at the European level. The empirical analysis is preceded by a section on the motivations and one on the major barriers connected with the processes of research internationalisation.
- Drivers of IRC
IRC depends to a large extent on the approach of scientists as 'calculating individuals’: scientists collaborate in research internationally because it is profitable to them in terms of academic prestige, scientific recognition, and access to research funding. Consequently, the convergence between individual-level motivations and drivers for internationalisation with departmental-, institutional- and national-level research policies is needed. The drivers of IRC also include increased visibility, new knowledge and contacts of value for the future. Apart from geographic proximity (or spatial proximity) as an important factor in IRC, cultural proximity also matters. What has been reported in the literature is the role of the 'invisible college’, the tendency of graduates to collaborate only with other graduates of their schools, with similar cultural and academic traditions, forming strong professional network ties. The academic excellence issue means that, at both individual and institutional levels, attractiveness of the potential research partner plays a crucial role in IRC. Not only is the formation of a collaboration proportional to the academic excellence of its participants, but also its impact advantage. Research shows a significant relationship between academic excellence and the probability of co-authorship: the more experienced the researcher, the higher the tendency to collaborate; the more highly ranked the academic department to which the researcher belongs, the higher his propensity to collaborate; and the higher the author’s rank, the higher his or her inclination to collaborate. Not all sciences are equally driven by the internationalisation demand: the four types of international collaboration are: data-driven collaboration (as in genetics, demography, epidemiology); resource-driven collaboration (as in seismology, zoology); equipment-driven collaboration (as in astronomy, high-energy physics); and theory-driven collaboration (as in mathematics, economics or philosophy). Wagner (2005) shows that different motivations for international collaboration affect the extent and patterns of the internationalisation of research as viewed through internationally co-authored papers. Availability of resources increases the level of IRC. Beyond that, scientists create and sustain the connections that form the global knowledge network largely because they 'become resources to others … connections are retained as long as they are of mutual (or potential) interest to participating members’ (Wagner 2018: 62). In short, networks mean (international) collaboration.
- Barriers to IRC
Barriers to IRC may include macro-level factors (geopolitics, history, language, cultural traditions, country size, country wealth, geographical distance); institutional factors (reputation; resources); and individual factors (predilections, attractiveness). They also include lack of funding, finding collaborators, communications (different languages, managing personal/family commitments, managing work commitments and time commitments to initiate/conduct the collaboration. The costs of collaboration can take a variety of forms. First, travel and subsistence costs are substantial. Costs of international physical mobility have been on the rise across all European science systems for all staff categories, including scientists and management personnel. Another cost is time as an academic resource. Additional requirements can reduce the available time and energy for actual research activities. Finally, collaboration increases administrative costs of research: with more people and more institutions involved, greater effort is required to manage the research.
- Data sources and methodology
The data analysed in this report have been retrieved from Scopus, the largest abstract and citation database of peer-reviewed literature covering almost 40 000 journals, book series and conference proceedings by some 6 000 publishers (owned by Elsevier) and SciVal, an Elsevier’s research intelligence tool offering access to research performance of 230 nations, as well as 12 600 institutions and their associated researchers worldwide. SciVal uses Scopus data from 1996 to the current date, which covers 48 million records. SciVal receives a weekly update of new data from Scopus. The choice of Scopus rather than the Web of Science (WoS) global indexing data set in this report was motivated by higher coverage of academic journals, especially in EU-13 countries. The report uses the 2007-2017 data, assuming the timeframe to be long enough to analyse basic trends in research performance and changing collaboration types over time. The analysis of collaboration in research is limited to a single output data type: bibliometric data on publications. The overall approach to IRC was unambiguous: IRC was analysed in the context of the three other collaboration types: institutional RC (multi-authored research outputs, where all authors are affiliated with the same institution in a European country), national RC (multi-authored research outputs, where all authors are affiliated with more than one institution within the same European country), and single authorship (or no collaboration, single-authored research outputs where the sole author is affiliated with an institution in a European country).
Empirical analyses show that both the number of internationally co-authored articles and their percentage share in the national output have been on the rise in the last decade across all EU-28 countries. The number of articles written in international collaboration in the period studied (2007-2017) was 2,193,504 in the EU-28 and 1,437,621 in the United States of America (USA), compared with merely 588 087 in China; however, the highest growth in the number of these publications per year in the same period was for China (by 309.02 %). Within countries and between them, there is substantial cross-disciplinary differentiation, with different increases between different fields of science. In the EU-28, the largest number of articles published in international collaboration in 2017 was by far for natural sciences (175,150; and 109,624 in the USA), followed by medical sciences (84,325; and 64,029 in the USA) – and the lowest for the humanities (5,480; and 2,880 in the USA). In 2017, the share of internationally co-authored papers was 44.4 % for EU-28 countries (47.1 % for EU-15 countries and 39.2 % for EU-13 countries, 40 % for the United States and merely 22.2 % for China). IRC in Europe is thus at similar levels than in the USA and 150 % more popular than in China.
The share of national collaboration was the highest for China (30.2 %), followed by the United States (23.7 %), and EU-28 countries (18.9 %, with a significant difference between the EU-15 and EU-13 groups: 19.2% and 15%.3, respectively). The share of institutional collaboration is in the range of 45.4 % (China) and 24.1 % (EU-15, considerably bigger for EU-13 countries, 33.5 %). Finally, the share of single-authored publications is the smallest in China (2.4 %) and in the rest of the groups of countries it remains at a level of only 9.5-12.1 %. The same trends (2007-2017) and the same patterns (2017) are clear for all of the EU-28 countries studied. There is not a single EU-28 country in which IRC has not been on the rise in the period studied and in all countries it was a dominating collaboration type in academic science in 2017. The vast differences in the total number of internationally co-authored publications among the European countries studied needs to be kept in mind in all percentage-based IRC trends.
EU-28 countries also differ significantly in terms of their IRC in terms of two other parameters: their collaborating partner countries and the Field-Weighted Citation Impact (FWCI, or the ratio of citations received relative to the expected world average for the subject field, publication type and publication year) of their internationally co-authored publications. The biggest number of internationally co-authored papers is observed between China and the USA, followed by the United Kingdom (UK) and the USA, Germany and the USA, as well as France and the USA. The dominant feature of IRC in Europe is its powerful collaboration with the USA: the UK, Germany, and France collaborate more intensively with the USA than with any other European country. In 2013-2018, 172,887 papers were written jointly by UK and US scientists, 141,195 papers written jointly by German and US scientists, and 93 308 papers written jointly by French and US scientists. In contrast, the highest number of papers written by two intra-European collaborative partners is only 90,202 (papers co-authored by German and UK scientists in the period studied). While China is the most powerful global partner of US science, only one country in Europe, the UK, is collaborating widely with China (with 63,625 papers written jointly in the period studied).
The analysis at the macro-level of countries is accompanied in this report by the analysis at the meso-level of (selected, flagship) institutions. In most general terms, collaboration trends over time and collaboration patterns for 2017 (according to the four collaboration types: institutional, national, international and single-authorship) are similar for EU-28 countries and for their flagship institutions; however, the internationalisation trends are more intense for flagship institutions than for countries.
The percentage share of international collaboration is on average lower for flagship universities located in EU-13 countries than for those located in EU-15 countries. While no flagship universities located in EU-13 countries exceeded the level of 60 % of international collaboration for the period 2007-2017 and only three exceeded that of 50 %, five flagship universities in EU-15 countries exceeded the level of 60 % of international collaboration (University of Luxembourg, University of Vienna, Karolinska Institutet, KU Leuven and University of Oxford). Only in four EU-28 flagship universities was the share of internationally co-authored publications in a single year of 2017 smaller than 50 % (all of them located in central and eastern Europe). For all the universities studied, the percentage share of internationally co-authored papers increased substantially between 2007 and 2017.
The patterns indicate that the biggest increase in citation impact per international collaboration is observable for institutions located in EU-13 countries: the top five includes institutions from the Czech Republic, Slovakia, Croatia, Poland and Romania. The increase is as follows: Charles University (Prague) by 336.9 %, Comenius University (Bratislava) by 290 %, University of Zagreb by to participating members’ (Wagner 2018: 62). In short, networks mean (international) collaboration.
Policy option 1: IRC should be at the centre of national research policies.
National higher education systems focused on increasing the international visibility of their academic knowledge production need to install the internationalisation of research at the centre of their national research policies (Norway being a prime positive example, see Gornitzka and Langfeldt 2008). European countries have been transforming governance and funding modes in their higher education systems and internationalising their research policies to increase their global competitiveness (Horta and Yudkevich 2016; Shin et al. 2014; Kwiek 2013; Kwiek 2015b).
At the same time, global and intra-European competition in research is reflected on several planes:
- human resources, or the competition for talent (including scientific prize winners and Highly Cited Researchers)
- funding, or the competition for EU research funds (including highly competitive individual research funding from ERC; see Bloch and Schneider 2016)
- research performance, or the competition for highly-cited publications and publications in highly-cited journals (for instance, publications in the top 1 % or 10 % citation percentile and publications in the top 1 % or 10 % journal percentiles; see Bornmann et al. 2013; Bornmann et al. 2014; and Didegah and Thewall 2013)
- international academic rankings (and especially those fully research-based like Leiden Ranking based on WoS data).
If IRC should move to the centre of national research policies, English should also be acknowledged as the language of global science today because, increasingly, 'non-native English speakers face challenges when trying to publish’ (Powell 2012). Academic and scientific English holds the key to success on the international scale.
Installing the internationalisation of research at the centre of national research policies refers to all levels of operation of HE systems, from national to institutional to departmental to individual. In most general terms, internationalisation-supportive research policies should promote top international publications in academic employment, rather than merely top national publications and should promote international, rather than merely national, collaboration in research. They should promote international publication channels both in direct block funding to their institutions and in indirect, individual-level competitive research funding in their national research councils (or their equivalents). They should also promote the internationalisation of research in their award and reward systems in science at the level of individual scientists.
Consequently, national models of successful universities, departments, research teams and individual scientists need to be clear: no academic success is possible and no large funding is awarded at any level to those units and individuals that are not internationalised in research. No professorships are available (or renewable) to scientists whose research performance profile is predominantly national – rather than international. In some national systems, detailed guidance is needed (numbers or percentages, percentiles of publications or journals, or national journal ranking lists); in others, general guidance suffices for the research internationalisation agenda to be implemented.
However, as this report strongly emphasises, IRC depends to a large extent on individual approaches of scientists as 'calculating individuals’: scientists collaborate internationally in research, including top-level international publishing, because it is profitable to them in terms of academic prestige, scientific recognition and access to academic rewards and research funding. Consequently, the convergence between individual-level drivers for internationalisation and departmental-, institutional-, and national-level research policies is needed.
For the research internationalisation agenda to be successful, highly internationalised institutions, departments, research teams and scientists need to be better off than local ones; the international needs to be promoted over the local in research in the different variants of national research assessment exercises, usually leading to different intra-national rankings of institutions or their organisational units (Ponomariov and Boardman 2010). IRC should matter more for funding and academic prestige and it needs to be consistently promoted at all levels of academic organisation. Usually, major opponents to national assessment exercises and rankings of institutional units or institutions come from the humanities and their major supporters come from the natural sciences; consequently, national and institutional systems need to guarantee cross-disciplinary flexibility so that the whole idea of systematic promotion of research internationalisation is not in jeopardy; in each system, there is a limited number of local academic disciplines, usually linked to national languages, literatures and history.
Policy option 2: Large-scale funding should be provided for IRC.
Increasingly, top scientists globally opt for collaborative, networked science that is locally rooted through training and institutions and nationally funded. European countries should consider supporting their academic faculty to become more internationalised in research and providing large-scale funding for IRC to avoid creeping isolation at a global level.
Internationalisation costs are increasing across all national systems in Europe: suffice to compare institutional and national budgets for research, including budgets for new ministerial programmes or programmes of national research councils directed at IRC. Internationalisation costs include both such traditional items as travel and subsistence costs for hundreds of thousands of travelling scientists and such new items as subscriptions to global indexing data sets and global academic journals. Doctoral students, postdocs, junior and senior scientists travel for academic business increasingly frequently, and use access to global knowledge bases (publications and data provided by Clarivate Analytics, Elsevier and other commercial providers) to an unprecedented degree. Journal and book subscription and ICT infrastructural costs are critical to the success of IRC and they are also increasing, both globally and in EU-28 countries. As international academic travel, global academic journals and books and ICT infrastructure are at the core of internationalisation, the rise of internationalisation-related costs needs to be noted and reflected in both budget size and its internal distribution. IRC costs – and it costs a lot.
Consequently, national systems seeking to increase the international visibility of their knowledge production need not only to install international research at the centre of national research policies but also to consider substantial public investments in research internationalisation. One option is to increase public investments, and another is to choose spending priorities differently, with internationalisation in research in focus. In different systems, different options are possible; however, disregarding both options may lead to gradual international isolation of national science systems across Europe, and especially in EU-13 countries, traditionally heavily underfunded in research in the last three decades in almost all cases and almost all academic disciplines.
Policy option 3: Individual scientists should be at the centre of national internationalisation agendas.
National systems determine conditions in which academic institutions operate, thriving or fighting for survival; however, in IRC the critical node is the individual scientist who will (or will not) collaborate internationally in research, will (or will not) publish in international collaboration and will (or will not) publish in top academic journals.
The national aggregate of individual-level research performance determines national research performance, and the aggregate of individual-level collaboration patterns in research determines dominating national collaboration patterns, as different as they have been discussed in the two sections on empirical findings in this report. In IRC, the abstract levels of 'countries’ (Section 5) and 'institutions’ (Section 6) are ultimately aggregates of individual scientists collaborating and publishing, more (or less) internationally. Understanding this individual-level determination of successes or failures of IRC is critical in understanding the future of IRC. 'It’s the individual scientist, stupid!’, to paraphrase Bill Clinton (in the multi-layered context of IRC which includes institutional and national award and reward structures in science, systems of academic promotion, levels of research funding and modes of its distribution etc.)
The individual scientist matters so much for IRC today because the modalities of IRC depend almost entirely on scientists themselves. They decide whether and with whom to collaborate, institutionally, nationally and internationally, and the decision to internationalise in research depends on individual choices based on reputation, resources, research interests, and the attractiveness of the potential research partner (Wagner 2018; Da Fonseca Pachi et al. 2012). In the empirical section of this report, different national (28 countries) and different institutional (22 flagship universities) collaboration patterns have been shown in detail, with different levels of IRC between systems and within systems. However, the data used are merely aggregates of individual-level data derived from publications. And publications are only (co-)published by (more or less heavily) internationally collaborating individuals.
At this basic, individual level of particular collaborating scientists affiliated to particular institutions, there is always a trade-off between the time and energy spent on IRC and research and publishing outcomes of this collaboration. If a given collaboration in research is beneficial individually, it will occur; but if it is not, it will not occur.
Therefore the crucial point is to create sufficiently attractive internationalisation-supportive research policies at various levels, from institutional to national (and international), to make sure that scientists are increasingly involved in IRC. A bottom-up approach, with maximum flexibility as to how, with whom and on which topic to collaborate internationally in research, unreservedly combined with the hard line of research excellence as defined through top publications alone, should always work better than any other set of recommendations for IRC programmes.
If the global network of science emerges because scientists 'connect with each other on a peer-to-peer basis, and a process of preferential attachment selects specific individuals into an increasingly elite circle’ (Wagner 2018: x), then scientists not collaborating internationally in every country (with the possible exception of the USA) are gradually being excluded from the ongoing global scientific conversation.
Across Europe, internationalists compete directly with locals, or scientists collaborating internationally in research compete directly with scientists not collaborating internationally (in sharp contrast to the United States, see Goodwin and Nacht 1991; Finkelstein and Sethi 2014), and locals increasingly stand to lose out. As the rules governing academic prestige, incentives, and awards become increasingly homogeneous across the continent, individual evaluations based on prestigious international publications become ever more important for individual academic careers. Across Europe, academic institutions (competing for public funding and high international rankings) tend to use the same research-based metrics because their aggregated institutional success hinges on the disaggregated individual research successes of the academics they employ.
The international visibility of national research output hinges on prevailing patterns of collaboration (international, national) and of publication (international channels, national channels). These can be changed over time by means of careful policy measures that promote advantageous patterns while discouraging others.
What is important in increasing the international visibility of individual scientists, institutions, or countries is not only IRC; it is also the changes in the publication behaviour of scientists and the increasing role of the academic journal stratification in which all journals have their clear positions in global science systems, with all disciplines having their own top-tier journals (van Raan 1998). As part of their IRC policies, faculties, institutions, and nations should no longer be focused on merely the international publications of their scientists; they should be increasingly focused on highly-cited publications in highly-ranked academic journals. Only these publications can increase their position in global rankings and guarantee stable public funding. This is particularly true in the context of widespread national 'research excellence’ initiatives additionally supporting financially only selected parts of higher education systems. Generally, understanding that IRC rests on individual scientists and their individual decisions to internationalise their own research should be installed in the centre of national internationalisation agendas. European international collaboration trends in research are merely aggregates of individual research decisions taken by millions of scientists involved in the global academic enterprise, day by day, year by year.
About the Author:
Marek Kwiek. Professor (full) and Director of the Center for Public Policy Studies (since 2002), Chairholder, UNESCO Chair in Institutional Research and Higher Education Policy, University of Poznan, Poland (www.cpp.amu.edu.pl). ORCID: 0000-0001-7953-1063. Contact: email@example.com
His research area is quantitative studies of science & sociology of science. His focus is on international research collaboration, academic productivity, stratification in science and global academic elites, and he uses global bibliometric datasets and large-scale international surveys.
His recent monograph is Changing European Academics. A Comparative Study of Social Stratification, Work Patterns and Research Productivity (London Routledge 2019). He has been advising extensively to national governments and international organizations on university funding & governance reforms and science policy in 13 countries (OECD, World Bank, USAID, Council of Europe, UNDP, E&Y and PWC). His most recent research report is „The International Collaboration of EU Research Organizations” for the European Parliament (114 pp., July 2019). Since 2000, he has been a Principal Investigator or country Team Leader in 25 international higher education research projects (global and European) funded by the European Commission (6th and 7th Framework Programs); the European Science Foundation (ESF); and the Fulbright, Ford, and Rockefeller foundations. He was a partner in large-scale EU-funded comparative projects: EDUWEL: Education and Welfare (2009–2013), WORKABLE: Making Capabilities Work (2009–2012), EUROAC: The Academic Profession in Europe (2009–2012), EUEREK: European Universities for Entrepreneurship (2004–2007), and GOODUEP: Good Practices in University-Enterprises Partnerships (2007–2009). He has published approximately 180 papers and 8 monographs and publishes mostly in leading international journals.
Translations of the report are here:
Chinese (traditional): here.