The Shifting Landscape of Global Science: A Challenge for United States Policy

Caroline S. Wagner, Gregoire Cote, Eric Archambault

Abstract

Rapid changes in national shares of publication numbers and citations reflect an underlying shift in the global landscape of science. Many more countries of the world are producing world-class science and scientists than was the case even 15 years ago. For a number of reasons (some related to data collection and calculation practices) the United States’ quality indicators are not increasing compared to other nations. The growth in output of many other countries and regions is nudging the United States from its dominant position in scientific rankings[1]. The shifting positions include the rise to equality of the European Union with the United States in quantity and quality of output, and the very rapid rise of Asian entrants into global science. The shifting landscape can work to its benefit if the United States can take advantage of the distributed knowledge base emerging in science and technology. This could involve developing a policy of global knowledge sourcing. This article discusses the changing landscape, the concept of global sourcing, and the implications of these factors for policy.

Shifts in Global Science

Since the middle of the 20^th century, the United States has led the world rankings in scientific research output in gross numbers and quality indicators[2]. In numbers, U.S. output accounted for over 20 percent of the world’s papers in 2009. U.S. research institutions have topped most lists of quality research institutions since 1950. The United States vastly out-produces most other countries or regions in patent numbers. This privileged status was partly due to the historical anomaly whereby at the end of World War II the United States had a newly developed, full-fledge and expanding scientific system whereas the rest of the world (save a few countries such as Canada and Australia) had to rebuild their war-torn science systems; it also is partly due to the significant, sustained commitment made by the U.S. government to research and development spending.

Many governments, responding to the perceived significance of science to economic growth, have increased investment in R&D[3]. In 1990, six countries of the world were responsible for 90 percent of R&D spending; by 2008, this number has grown to include 13 countries (not including spending by the European Commission which is about 3 percent more, over and above the individual countries)[4]. According to UNESCO, since the beginning of the 21^st century, global spending on R&D has nearly doubled to close to 2 percent of all the world’s global domestic product[5], or USD 1.1 trillion[6]. Developing countries have more than doubled their spending on R&D during the same period[7].

The number of scientific papers cataloged has grown as information retrieval processes have improved, and as competition has arisen in delivering scientific publications to readers (from different services and from the Internet).  Figure 1 plots the increase in the Science Citation Index from 1990 through 2011; the number of journals accepted for inclusion in these services increases at about 4 percent net growth per year[8]. In 2009 and 2010, the SCI was expanded to reflect greater regional coverage, in part to balance the dominance of the United States output in the data. Thus, these two years show a large jump in the numbers of journals in SCI, causing a discontinuity in trend line and influencing the position of the United States.

[insert figure 1 about here]

The two years of discontinuity does not negate the overall trend line, which shows significant changes for the United States. Over 30 years from 1980 until 2010, the growth in output of scientific articles (as recorded in the cataloging services) has contributed to a drop in the relative position of the United States. The convergence of three factors is influencing the shift: 1) increased capacity to conduct R&D in many countries, 2) more output of scientific articles by all countries, although some more than others, and 3) expanded cataloging of journals in the databases. Additionally, UNESCO notes the trend in many nations to train more technical people: The number of researchers has increased significantly from 5.7 million in 2002 to 7.1 million in 2007. The distribution of talent is spread more widely, and the quality of contributions from new entrants has increased[9]. All these factors are cutting into the historical position of the U.S., although to what extent each one is a factor is unknown[10].

The European Union (27) countries plus Switzerland have shown the most notable improvement in quality relative to the United States[11]. The EU-15 surpassed the U.S. in SCI impact factors in 1994[12].  As the EU has expanded, comparisons with the United States use all 27 member countries.  Using citation counts as a proxy for quality, the EU-27 plus Switzerland have shown the most significant quality gains on the United States. Switzerland surpassed the United States in citation quality measures in 1985, albeit based on a small number of publications compared to the U.S.[13] More recently, some individual European Union member countries also have passed the United States in citation quality measures: Denmark and the Netherlands in 2003, Belgium in 2007, the United Kingdom and Germany in 2008, and Sweden and Austria in 2009.

Asia lags the U.S. and Europe in citedness of scientific output. Asian journals are not well represented in the catalog services, compared to the United States and Europe. It is noteworthy that in the SCI, Singapore’s impact has been growing fastest among this group of countries. (A contributing factor may be that the government of Singapore has offered incentives to researchers to publish in high impact journals.) In fact, should current trends continue, Singapore will rank fifth in quality measures in 2015, breaking a barrier for Asian scientists in scientific output and impact in Scopus and SCI. The table below shows the number of journals in each region of the world compared to the number of researchers working in that region.

[insert table here]

The trend of flat U.S. quality measures has been discussed elsewhere in the context of U.S. R&D spending[14], output shares[15], and other factors[16],[17],[18]. Several explanations have been offered including that the rate of international collaboration is increasing rapidly, and thus the U.S. has been increasingly sharing citation counts with other countries[19]. A second explanation put forth is that the United States is producing output at a maximum level of efficiency, and even with more input and higher efficiency, could not increase the level or quality of output[20]. A third is that other countries and regions have made a concerted effort to enhance the quality of their R&D, and they have seen good results. All of these explanations may be factors. Additional research is needed to understand if these factors is clearly influencing the U.S. position.

As other parts of the world have enhanced their science bases, the U.S. percentage shares of all aspects of the knowledge system are giving way to a broader representation of countries. Over the past two decades, scientifically-advanced countries[21] have continued to strengthen their positions relative to the U.S. New entrants – particularly China and South Korea – two countries that are vastly increasing their investment as well as the quantity and quality of their output[22] are rapidly taking leadership positions in scientific output[23]. Between 1996 and 2008, the United States dropped 20 percent in relative terms in its share of global publications as other nations have increasingly placed quality scientific publications in journals cataloged by Thomson-Reuters and/or Elsevier[24].

The sustained rate of growth of China has caught the attention of many who track global science. Its rise may be due to the increasing availability of human capital at Chinese universities and research institutions. In addition, the Chinese Academy of Sciences is providing incentives for researchers to publish in cataloged journals[25]. Chinese scientists who have been living abroad have been encouraged to return to China[26], or to collaborate and coauthor with co-nationals[27]. These changes have increased the number of Chinese scientists who seek to publish in the cataloged journals, contributing to the growth in overall numbers in the Science Citation Index, and the drop in percentage share of other leaders. Other notable new entrants among rapidly increasing producers are South Korea, Turkey, and Iran[28]. At the same time that Asian countries have supported exponential growth in scientific publications, the U.S. and other scientifically-advanced countries have maintained slow growth.

To create a measure that would normalize the quality measures across-the-board and allow analysis over 30 years, one of us calculated the average of relative citations (ARC) by paper, by address of each author. The ARC is obtained by counting the number of citations received by each paper during the year in which the paper is published and for the two subsequent years. To account for different citation patterns across fields and subfields of science (e.g., there are more citations in biomedical research than in mathematics), each paper’s citation count is divided by the average number of citations in that field – in other words, the calculation obtains the average citation rate of papers in that same field during the same time period. The National Science Foundation establishes the taxonomy of scientific fields. This calculation creates a relative citation count (RC). The ARC of a given entity is the average of the relative counts of the papers authored by its researchers; an ARC value above one (1) means that a country’s publications are cited more than the world average, and below one, less than average. Counts are aggregated from the paper level by field up to the country level.

In order to validate the results obtained by studying citations by country, we added a calculation of the position of U.S. output based on quality of the journals in which papers were published, as approximated by journal impact factors. The specific measure used here is the average of relative impact factors (ARIF) that, similar to the ARC used above, takes into account interfield variations in the propensity to cite. The impact factor used here is similar to that calculated by Thomson-Reuters except that only articles, notes and reviews are used to calculate both citeable and cited materials. (This is in contrast to Thomson Reuters’ Impact Factor is asymmetrical[29].)  Despite its inherent limitations[30], for very large numbers of papers, the Impact Factor is a useful indicator of quality based on the accepted view that the most cited journals are, by and large, able to select the best papers.

The journal-level analysis finds the output of thirty years of ARIF data showing that U.S. scientists continue to publish in high-impact journals. The persistence of the U.S. position in journals is due to persistent quality of U.S. output. It also may be due to an Anglo-Saxon bias for English-language journals documented by Luukonen[31] and Decker et al.[32]. (For example, journals are often edited by U.S. academics: this may contribute to a bias towards publications from English-speaking authors.) The continued leadership of the U.S. is being matched by other countries with trends similar to those seen in the average of relative citations at the paper level: Swiss and Dutch scientists already publish in journals that, on average, are more frequently cited. Singapore is on a trajectory to overtake the U.S. Because several of these countries are increasing the quality of their output faster than the U.S., projecting these trends forward suggests that by 2015, the U.S. would occupy ninth place. 

The Challenge to U.S. Policy

The shifts in the global S&T landscape have evoked several responses within the United States. One has been alarm at the challenge to U.S. leadership[33] and concern about the implications of the emerging global order for U.S. economic competitiveness[34]. Concern about the implications of the decline of the U.S. position has raised calls from President Barack Obama to increase R&D funding, improve S&T education, and renew infrastructure. The National Academy of Sciences, in its 2010 update to the “Rising above the Gathering Storm” report, said “The unanimous view of the committee members…is that our nation’s outlook has worsened[35]” since the first report was issued in 2005, due in part to the rising positions of many other countries in science.

Another view has been to see these changes as the addition of new resources in the knowledge system. As new researchers and new knowledge centers arise, those in a position to access and share information can benefit[36],[37]. Unlike some economic resources (such as factories or commodities), knowledge resources have a feature that economists call “non-rivalous goods”-- meaning the consumption or use of the good by one individual does not reduce the availability or usefulness of the good for consumption by others. In fact, economists note that knowledge (particularly science and technology) has the feature of non-excludability, meaning no one can be effectively excluded from using it. Quite the opposite, in fact: scientific knowledge increases in value as it is used, as can be seen by the use of citations to measure quality. There is an additional economic theorem that applies here: shared technology can have network effects. This is where the value of knowledge--a product, standard, or a service--increases in value as more people use it.

As centers of excellence emerge in new places, it may be that the U.S. and other scientically-advanced countries can take advantage of these features of S&T knowledge. Global scanning for emerging knowledge is a service that other governments (such as Japan, Finland, and Canada) conduct effectively.  The concept behind global scanning is a simple one: at the global level, the scientific system does not need to create or maintain redundant capabilities. This is the common practice in megascience—those areas of science where the cost to any one nation of investing in S&T is too great (e.g., astrophysics, space science, climate change). This practice of global scanning, coupled with collaboration and networking, could improve efficiency of U.S. S&T investment at a time of budgetary pressure.

Rather than creating capacity across the entire frontier of science—a suggestion made by the National Academy of Sciences in a 1997 study, global knowledge sourcing would require making careful choices. The United States can draw upon a worldwide knowledge system. A more aggressive policy and strategy of collaboration and networking can identify redundant capabilities where the U.S. would not be weakened by depending upon an internationally-shared resource. This could free up national investments to focus on more critical, cutting-edge capacity building needed locally—an approach called global knowledge sourcing[38]. In business parlance, global knowledge sourcing means the integrating and coordinating of common materials, processes, designs, technologies, and suppliers across worldwide operating locations. Applying a similar vision to national-level investments could result in significant efficiencies.

Although the U.S. S&T system remains the world’s largest and among the best, it is clear that a new era is rapidly emerging, one that, with preparation and strategic policymaking, can work to U.S. advantage. Intelligent knowledge sourcing can speed innovation. The vast size of the U.S. system can actually be a hindrance to moving towards a more efficient, networked knowledge system. The size of the U.S. S&T system means it can use vast networks to identify and draw in knowledge. On the other hand, the U.S. capabilities means that any one knowledge unit (researchers, institution, disciplines) does not need to look far to find the knowledge it needs. One is just as likely to find a collaborator within the U.S. as in another country; this is not true in smaller countries. At the international level, as a percentage of all output, the United States is one of the least internationalized countries in the world. (Only Japan is less internationally engaged in percentage terms[39].) U.S. public strategy would need to be reworked to source knowledge globally and reintegrate locally[40].

Of public funds dedicated to R&D out of the U.S. government budget, an estimated six percent goes to international collaboration[41]. Among these funds, international engagement can be classified into two large categories: 1) top-down, planned and targeted research projects (including megascience funding) and 2) bottom-up research projects initiated by researchers. Within both categories, projects can fall along a spectrum from those that take place at a centralized location (such as CERN) to those that are distributed and virtual and conducted using electronic media (e.g., distributed and virtual). Increasingly, research projects combine some aspects of both features. United States Federal appropriations for R&D in fiscal year 2009 (prior to stimulus funding) totaled just over $151 billion[42]. Assuming that international collaboration is about 6 percent of all U.S. government spending, the amount spent on international collaboration (most of which is spent in the US) may be close to 9 billion--clearly a large contribution to global S&T[43].

The R&D funds spent by the U.S. government, for the most part, are not set aside at the start or otherwise dedicated to “international R&D.” This can be a perplexing detail for those outside the U.S. wishing to collaborate within the American science system. Generally the U.S. government funds mission-oriented research; if international collaboration furthers that research, it can become a part of U.S. activities, but it is rarely a goal of U.S. government-funded R&D appropriations or allocations. Those seeking to understand U.S. S&T policy (particularly towards international engagement) are often surprised to find that there is no such policy: the great majority of international R&D linkages result from the choices of individual scientists to work with foreign counterparts.

Still, the rapidity of the shifts in the global landscape may warrant a policy-level reassessment of this bottom-up approach to global engagement, perhaps to incorporate some global sourcing into government-wide planning. An inventory of international S&T engagement would be a first place to start on this venture.

United States policy currently lacks a strategy for encouraging and using global knowledge sourcing. Up until now, the size of the U.S. system has insulated it from having to make these choices. Meantime, smaller scientifically-advanced nations such as The Netherlands, Denmark, and Switzerland have been forced by budgetary realities as well as collaborative opportunities to shift policy much sooner than the United States. These nations have made strategic decisions to fund excellence in selected fields, and to collaborate in others. This may account in part for the rise in their quality measures. In the U.S., an explicit policy of global knowledge sourcing and collaboration would require restructuring of S&T policy to identify those areas where linking globally makes the most sense. Currently, there are a number of obstacles to global sourcing of science: 1) The United States government does not track global centers of excellence—this function is within the purview of practitioners who scan within their own networks for connections but do not have a strategic view of global science. 2) U.S. S&T budgeting and funding for scientific projects is not aligned to funding in other parts of the world. As a result, plans to collaborate are often scuttled because groups cannot get parallel funding. 3) Most agencies limit R&D funding awards to U.S.-based researchers.

One recent example of movement in the direction of global knowledge sourcing is the United States government participation with other governments in the Interdisciplinary Program on Application Software towards Exascale Computing for Global Scale Issues. Following the Group of 8 (G8) meeting of research directors in Kyoto, an agreement was reached to initiate a pilot collaboration towards multilateral research. The participating agencies are the U.S. National Science Foundation, the Canadian National Sciences and Engineering Research Council (NSERC), the French Agence Nationale de la Recherche (ANR), the German Deutsche Forschungsgemeinschaft (DFG), the Japan Society for the Promotion of Science (JSPS), the Russian Foundation for Basic Research (RFBR), and the United Kingdom Research Councils (RC-UK). These agencies will support on a competitive basis, collaborative research projects that are comprised of researchers from at least three of the partner countries—a model similar to the one used by the European Commission. Proposals will be jointly reviewed by the participating funding organizations and successful projects are required to demonstrate added value through multilateral collaboration. Support for U.S.-based researchers will be provided through awards made by the National Science Foundation. It would be useful to begin discussions about metrics of success of these types of activities.

Tapping the best and brightest minds in S&T, and gathering the most useful information anywhere in the world and bringing it back to the U.S., would greatly serve the economy and social welfare. Looking for the opportunity to collaborate with the best center in any field is a prudent perspective, since it seems unlikely that the rise of S&T at the global level can be halted while the U.S. catches up. Moreover, it may be that the U.S. is producing scientific publications at the maximum level at which it can operate and that additional spending may only produce diminishing returns[44]. Thus, seeking and integrating knowledge from elsewhere is a very rational and efficient strategy, requiring global engagement and an accompanying shift in culture.  Leadership at the policy level may be needed to speed this cultural shift.

Acknowledgements

The authors wish to thank Derek Bell and Loet Leydesdorff for helpful comments on earlier drafts.