Skip all navigation and go to page content.

Chapter 5. Academic Research and Development

Outputs of S&E Research: Articles and Patents

Chapter 2 of this volume discusses the human capital outputs of higher education in S&E. This section of the current chapter continues that theme by examining the intellectual output of S&E research. The section presents indicators derived from both published research articles and U.S. patents.

Researchers have traditionally published the results of their work in the world’s peer-reviewed S&E journals. These bibliometric data (see sidebar, “Bibliometric Data and Terminology”) are indicators of national and global scientific activity. For example, a count of the coauthorships on U.S. articles is an indicator of the partnerships involved in the U.S. scientific effort. Likewise, measures involving citations and patents can be indicators of international patterns of influence and of invention based on scientific research. Bibliometric indicators are calculated for different countries and—within the United States alone—for different sectors.

Overall, the indicators provide insight into five broad areas. The first section, “S&E Article Output,” examines the quantity and national origin of S&E publications. The second section, “Coauthorship and Collaboration in S&E Literature,” examines the national partnerships in these publications. The third section, “Trends in Citation of S&E Articles,” examines various patterns of national scientific sharing and influence. The fourth section, “Citation of S&E Articles by USPTO Patents,” examines the utilization of S&E literature by inventors. And, finally, the fifth section, “Academic Patenting,” examines patenting and related activities in academia.

Discussions of regional and country indicators will examine patterns and trends in developed and developing countries, as classified by the World Bank. Countries classified by the World Bank as high income are considered developed; those classified as upper- and lower-middle income and as low income are considered developing.[43]

S&E Article Output

This section begins by describing and comparing the S&E article output of the United States to other regions, countries, and economies in the world. The article output of China and other developing countries has increased much more rapidly than that of the United States and other developed countries over the last 15 years. Although the United States remains a major producer of S&E articles, its global share of article production has declined. This section then examines U.S. article output in academia, the largest producer of U.S. articles, and other institutional sectors.

Article Output by Country

A growing number of countries produce S&E articles. Over the period from 1988 to 2012, a total of 199 countries were authors on at least one S&E article (appendix table 5-24).[44]

The four major producers of the world’s S&E articles in 2011 were the European Union (EU; see “Glossary” for member countries) (31%), the United States (26%), China (11%), and Japan (6%).[45] Together, they accounted for 73% of the world’s S&E publications in 2011 (figure 5-19; appendix table 5-26). The EU, the United States, and Japan have been major producers for several decades. China emerged as a major producer in the mid-2000s. Overall, 47 countries—less than a quarter of those that produced S&E articles in 2011 (see appendix table 5-24)—accounted for 98% of global output (table 5-20).

Between 2001 and 2011, the total world S&E article output grew at an average annual rate of 2.8% (table 5-20). The total for developing countries grew more than three times faster (9.9% average annual) than the world total. China propelled growth of developing countries (15.6%), resulting in its global share climbing from 3% to 11% (figure 5-19). The fifth-largest S&E article producer in 2001, China surpassed Japan in 2007 to become the third-largest S&E article producer, behind the EU and the United States (appendix table 5-26). China’s growth in S&E publication is concurrent with its enormous growth in GDP over the last decade, which is consistent with findings by many researchers that there is a high correlation between these two measures (Price 1969; Narin, Stevens, and Whitlow 1991).

Among other larger emerging economies, over the decade Brazil grew at a 6.4% average annual rate and India grew at a 7.6% average annual rate, resulting in their global shares increasing 1 percentage point to reach 2% and 3%, respectively (table 5-20). Rapid growth of S&E articles in Brazil, India, and China coincided with increased R&D expenditures and growth in S&E degrees awarded at the bachelor’s-degree and doctoral-degree levels (see chapter 2, “Higher Education in Science and Engineering”).

Smaller developing countries with rapid S&E article growth (11%–23% annual average) included Iran, Malaysia, Pakistan, Thailand, and Tunisia.

Developed economies’ S&E article production grew more slowly (1.5%) than that of developing economies (9.9%) over the decade. U.S. growth in S&E article production was even slower (1.1%) than the average for all developed economies. The U.S. global share fell from 30% to 26%, mostly as a result of developing economies’ more rapid growth.

The EU, the world’s largest producer, grew slightly more slowly (1.4%) than all developed countries. Among EU member countries, growth rates were slower for the three largest—France, Germany, and the United Kingdom—and generally much faster in Ireland, Portugal, and other smaller member countries. Although EU article production grew slightly faster than that of the United States, the EU’s global share fell from 35% to 31% because of far more rapid growth of developing countries.

S&E article production of Japan, the fourth-largest producer, contracted (-1.7% annual average) over the decade. As a result, Japan’s global share dropped from 9% to 6%, a far greater decline (35%) compared to the declines of the shares of the United States and the EU (15% and 12%). The weakening of Japan’s position may reflect its lengthy economic stagnation despite recent increases in R&D expenditures and reform of its research universities.[46] Also among major developed nations, Russia saw its S&E article output decline (-1.0% annual average) over the decade.

Publication output by developed economies outside of the EU, the United States, and Japan grew much faster, primarily due to rapid growth (6%–9% annual average) in three Asian locations—South Korea, Taiwan, and Singapore.

The distribution of S&E article output by field provides an indication of the priority and emphasis of scientific research in different locations.[47] The S&E article portfolios of the four major producers—the EU, the United States, China, and Japan—have distinct differences (table 5-21; appendix tables 5-275-39). The United States is focused primarily on biological sciences and medical sciences, more so than the world at large; together, these fields account for 52% of U.S. 2011 articles. The United States also produces a higher proportion of S&E articles than the rest of the world in other life sciences, psychology, and social sciences, although this may be due in part to how Thomson Reuters selects journals to include in its database.[48]

Like the United States, the EU is also focused primarily on biological sciences and medical sciences. However, the EU has placed a greater emphasis than the United States on physics, chemistry, and engineering.

Japan’s articles are fairly evenly divided among biological sciences, medical sciences, chemistry, and physics.

China’s S&E portfolio is dominated by chemistry, physics, and engineering, with a far higher concentration in these fields than the three other major producers and most other countries. These fields largely fueled China’s rapid growth in article output. Compared to the rest of the world, China and Japan put very little emphasis on publication in other life sciences, psychology, and social sciences.

Article Output by U.S. Sector

Six U.S. institutional sectors produce S&E articles: the federal government, industry, academia, FFRDCs, private nonprofit organizations, and state and local governments.[49] This section describes patterns and trends in the sector distributions of U.S. article output.

The U.S. academic sector is the largest producer of S&E articles, accounting for three-fourths of U.S. S&E article output. This sector was largely responsible for the slight growth of U.S. S&E article output over the last 15 years. The number of academic S&E articles rose from 138,000 to 163,000 between 1997 and 2012. As a result, academia’s share of all U.S. articles rose from 73% to 76% (figure 5-20).

S&E publications in the non-academic sectors decreased slightly from 52,000 to 51,000 during this period. These sectors had divergent trends:

  • Articles in the private nonprofit sector grew from 15,000 to 18,000 and at an even greater pace than the academic sector between 1997 and 2012 (appendix table 5-40). However, this sector’s much smaller size resulted in a lesser impact on total U.S. growth.
  • Articles in FFRDCs fluctuated between 5,000 and 6,000.[50]
  • Industry and the federal government exhibited similar trends, starting the period at 14,000 articles and then declining, especially over the past 10 years. However, industry articles dropped further than federal government articles to end the period at 12,000, compared with 13,000 for the federal government.

Except for the FFRDCs, the research portfolios of the U.S. sectors are dominated by life sciences (biological sciences and medical sciences), with nearly half or more of all articles in these fields (table 5-22). The dominance of life sciences is especially pronounced in the nonprofit sector, where 79% of the articles are in the biological sciences and medical sciences. With a much larger number of articles, academia has 49% of its S&E literature in life sciences. The research portfolio of FFRDCs is dominated by physics (36%), chemistry (19%), and engineering (16%), with far less concentration in life sciences (11%). This reflects the FFRDCs’ more specialized and mission-oriented research programs in these and other physical sciences.

Coauthorship and Collaboration in S&E Literature

Collaborative S&E research facilitates knowledge transfer and sharing among individuals, institutions, and nations. It can be an indicator of interconnections among researchers in different institutional settings and the growing capacity of researchers to address complex problems by drawing on diverse skills and perspectives. Collaboration on S&E research publications over the last 15 years has been increasing, with higher shares of scientific articles with more than one named author and a higher proportion of articles with institutional and international coauthorships (figure 5-21). The largest increase was in international collaboration; the percentage of articles with authors from different countries rose from 16% to 25% between 1997 and 2012.

The following two sections explore the growth of collaborative publication.[51] The first section looks at international collaboration. The second section examines collaboration across institutional sectors—including academia, the federal government, and industry—within the United States. (Data on sectors for other countries are not available.)

International Collaboration

International scientific collaborations reflect wider patterns of relationships among countries. Linguistic and historical factors (Narin, Stevens, and Whitlow 1991), geography, and cultural relations (Glänzel and Schubert 2005) play a role in these relationships. In recent years, coauthorships in Europe have risen in response to EU policies actively encouraging intra-European, cross-border collaboration. Strong ties among science establishments in Asia, though without the formal framework that characterizes Europe, have led to similar collaboration.

Rates of international collaboration by field. Inter-national collaboration on scientific articles, as measured by the shares of articles coauthored by institutional authors in different countries, has increased markedly over the last 15 years. S&E articles with coauthors from more than one country have grown to nearly one-fourth of the world’s S&E articles, rising from 16% in 1997 to 25% in 2012. This is a slightly larger increase than the increase in purely domestic coauthorships during the same period (from 36% to 44%) (figure 5-21).

Researchers in different fields have different tendencies to collaborate internationally. Astronomy is the most international field, with over half of its articles internationally coauthored (56%) (figure 5-22). Geosciences, computer sciences, mathematics, physics, and biological sciences have relatively high rates of international collaboration, with shares in the range of 27%–34%. Fields with low rates of collaboration (17%–21%) include psychology, chemistry, social sciences, and other life sciences. Possible factors influencing variations among fields include the existence of formal international collaborative programs, expensive infrastructure (e.g., atomic colliders and telescopes) that results in cost sharing and collaboration among countries, the geographic scope (local versus international) of research fields, and path dependencies from earlier, relatively local ways of doing research.

International collaboration has risen across all scientific fields over the last 15 years. The two fields with the highest rates of international collaboration—astronomy and geosciences—had increases of 17 and 14 percentage points, respectively, in their shares between 1997 and 2012. Physics and chemistry had far lower gains of just 5 and 7 percentage points, respectively. Psychology and other life sciences had strong gains yet remain among the four fields with the least amount of international collaboration.

Rates of international collaboration by country/region. Countries vary widely in the proportion of their S&E articles that are internationally coauthored, ranging from 25% (Iran) to as much as 80% (Saudi Arabia) for articles in 2012 (appendix table 5-41; see also appendix tables 5-425-54 for individual fields). The shares of larger countries are generally lower (from 25% to 60%) than smaller countries (from 50% to 80%). The difference is likely because the bigger and more diversified scientific establishments in larger countries allow opportunities for collaborative scientific teams within their borders, whereas smaller countries do not have the research infrastructure or personnel to support such collaboration.

The U.S. international collaboration rate was 35% in 2012, significantly lower than France, Germany, and the United Kingdom (figure 5-23). However, because the United States has a higher share of articles with domestic coauthors, its overall proportion of coauthored articles is similar to that of the three EU countries.

The higher international collaboration rates of large EU member countries relative to the United States are likely due to their smaller science establishments, which increase the need for collaboration teams with international participation. In addition, the EU’s Framework Programmes for Research and Technological Development and other programs designed to increase collaboration among EU member countries and with other countries likely boost their international collaboration.

Japan and China have even lower international collaboration shares than the United States (figure 5-23). One factor that may explain their low shares is that Asia does not have a formal framework like the EU to facilitate international collaboration. Another possible factor is that some Chinese and Japanese scientists may not speak English or publish their research in that language, which could limit their visibility in the international scientific community, where English is commonly used.

Rates of international collaboration have generally risen over the last decade, though to varying degrees (figure 5-23). The U.S. rate rose 10 percentage points to reach 35% between 2002 and 2012. Canada had a similar increase (from 40% to 50%) over the same period.

The increase has been even more dramatic for EU members and other European countries. The shares of France, Germany, and the United Kingdom increased by 12–16 percentage points to reach over 50%. The EU’s Framework Programmes for Research and Technological Development, now in their seventh year, have likely been a major factor in these countries’ increases.

China is an exception to the general trend of increasing international collaboration. China’s rate of international collaboration (27%) remained stable over the last decade during China’s period of very rapid article growth. In contrast, Chinese domestic collaboration increased in this period: the proportion of its articles that had multiple domestic institutional authors rose by 11 percentage points, reaching 44% (appendix table 5-41).

Preferred collaboration partners. Different countries have different preferred partners for international scientific collaboration. The remainder of this section describes global partnership patterns, with particular emphasis on patterns of U.S. involvement in international collaboration.

The nation that most often coauthors with the United States is China, a collaborator on 16% of U.S. internationally coauthored articles (table 5-23). [52] As shown in figure 5-24, other countries that are important partners for the United States are the United Kingdom (14%), Germany (13%), Canada (11%), France (9%), Italy (7%), and Japan (7%). Canada and China are notable among these countries for having unusually high rates of U.S. participation in their own internationally coauthored articles (49% and 48%, respectively). For the other five countries, the comparable rates range from 29% to 37%.

For most countries, the percentage of U.S. internationally coauthored papers on which they are coauthors has stayed stable over the decade. China and Japan are exceptions. China’s share of U.S. internationally authored articles tripled from 5% in 2002 to 16% in 2012, coinciding with its rapid expansion of article production. China swiftly moved up from the sixth-largest collaborating country in 2005 to the second-largest collaborating country in 2010 before becoming the largest in 2011. Japan’s share of U.S. coauthored articles dropped from 10% to 7%, coinciding with its decline in article production.

Several countries that collaborate on relatively few U.S. internationally coauthored articles have very high U.S. participation in their own internationally coauthored articles. Three economies—Israel, South Korea, and Taiwan—have more than 50% of their international articles coauthored with the United States. Other countries with relatively large U.S. shares of their internationally coauthored articles include Mexico, Chile, Brazil, and Turkey.

An index of international collaboration is useful for highlighting rates of international scientific collaboration that differ substantially from chance (see sidebar, “Normalizing Coauthorship and Citation Data”). When collaborative authorship between two countries is exactly proportional to their overall rates of international collaborative authorship, the index value is 1; a higher index value means that a country pair has a stronger-than-expected tendency to collaborate, and a lower index value means the opposite.

U.S. collaboration with countries as measured by the index of international collaboration shows variable trends (table 5-24; appendix tables 5-55 and 5-56). In North America, the Canada-U.S. index shows a rate of collaboration that is slightly greater than would be expected, and the index has not changed much over the past 15 years. The U.S.-Mexico index is just about as would be expected and has been stable.

In scientific collaboration with EU member countries, the United States has a weaker-than-expected tendency to collaborate with the United Kingdom, Germany, and France despite a comparatively high volume of internationally coauthored articles. U.S. collaboration with these countries became slightly stronger between 1997 and 2012.

In contrast to EU member countries, U.S. collaboration with Asia has generally been stronger than expected. U.S. collaboration is relatively strong with China, South Korea, and Taiwan. However, U.S. collaboration with Japan is slightly weaker than expected despite a high volume of coauthored papers. Between 1997 and 2012, U.S.-Japan collaboration has shifted from as expected to weaker than expected.

Collaborations between Latin American countries are notably stronger than expected. The collaboration index of Mexico-Argentina is 3.88, far above expected levels. The collaboration index of Argentina-Brazil is even higher, at 5.81, one of the highest in the world, and was high, at 4.94, even 15 years ago.

Among European countries, collaboration patterns are mixed, but most have increased between 1997 and 2012. Among the large publishing countries (Germany, the United Kingdom, and France), collaboration was less than expected in 1997 but grew to just about what would be expected in 2012. A particularly strong collaboration network has developed between scientists in Poland and the Czech Republic, with the index for their countries standing at 5.97 in 2012.

The Scandinavian countries increased their collaboration indexes with many countries elsewhere in Europe over the last 15 years (appendix table 5-55).[53] Within Scandinavia, the indexes are among the highest in the world (table 5-24).

Collaboration indexes within Asia and across the South Pacific between the large article producers are generally higher than expected, but some have declined between 1997 and 2012. The collaboration index of China-Japan declined from 1.61 to 1.23; the South Korea–Japan index fell from 2.20 to 1.93. The Australia–New Zealand collaboration index, although much higher than expected, fell from 4.33 to 3.65. Other partnerships strengthened during this period. The Australia-China collaboration shifted from slightly weaker to slightly stronger than expected. India’s collaborations with both South Korea and Japan grew stronger between 1997 and 2012.

Collaboration among U.S. Sectors

U.S. coauthorship data at the sector level—academic, nonprofit, industry, FFRDCs, federal and state government—are indicators of collaboration among U.S. sectors and between U.S. sectors and foreign institutions. The academic sector, the largest article producer among U.S. sectors, is the center of U.S. sector and foreign collaboration. In 2012, the academic sector published 119,371 articles coauthored with other U.S. sectors and foreign institutions, three and a half times more than the 33,973 such articles published by the nonprofit sector, the second largest (table 5-25).

Although the largest producer of articles coauthored with other U.S. sectors and foreign institutions, academia has the lowest coauthored share of total articles, compared to other U.S. sectors.

Figure 5-25 shows the share of U.S. articles coauthored with foreign institutions, U.S. academic institutions, and other U.S. sectors (outside of self and academia). FFRDCs are notable for their very high level of foreign collaboration (46%) compared to a 31%–34% range for most other U.S. sectors. With a high concentration of FFRDCs being focused on physics research (36% of FFRDC articles, table 5-22), which often requires the use of globally shared instruments, a high degree of international collaboration can be expected. State and local governments have the lowest foreign collaboration shares but the highest share of collaboration with other U.S. sectors. Industry has the lowest collaboration share (57%) with academia, compared to 63% or higher for other U.S. sectors.

Over the last decade, collaboration with other U.S. sectors and with foreign institutions increased strongly in almost all sectors (table 5-25). In the academic sector, the number of articles coauthored with other U.S. sectors and foreign institutions increased by more than half, from 76,622 to 119,371. The largest increase was for articles coauthored with foreign institutions, which increased by 83% (from 41,978 to 76,907). As a result, articles with foreign coauthors increased their share of all U.S. academic articles, from 24% to 34%. U.S. academic articles coauthored with other U.S. sectors increased by 41% (from 43,587 to 61,329 articles).

The nonprofit sector had the largest increase in the number of coauthored articles with other U.S. sectors and foreign institutions (from 20,703 to 33,973, a 64% increase). Nonprofit articles coauthored with foreign institutions led the increase, more than doubling (from 6,337 to 13,740). The percentage of articles coauthored with foreign institutions increased their share from 22% to 34%.

Articles with at least one author from industry grew the least over the time period, less than 8%, and in turn had the smallest increase in articles coauthored with other U.S. sectors and foreign institutions (25%).

Much of the growth of industry-coauthored articles was with foreign institutions; foreign coauthorships increased by 57%. Articles coauthored with the academic sector rose by only 29%, the smallest increase among sectors coauthoring with academia.

Trends in Citation of S&E Articles

Citations indicate influence, and they are increasingly international in scope.[54] When scientists and engineers cite the published papers resulting from prior S&E research, they are formally crediting the influence of that research on their own work.

Citations are generally increasing with the volume of S&E articles. (For the analysis of citations from articles to articles, citation counts are limited to a fixed 3-year citation window that begins 4 years and ends 2 years prior to the year of the citing article.[55]) As cited by 1992 articles, an earlier S&E article received, on average, 1.85 citations. In contrast, an S&E article cited by 2012 articles received, on average, 2.47 citations (figure 5-26). Articles with U.S. authors tended to receive more citations than others, but that gap has narrowed slightly in the most recent 4 years.

The next sections examine two aspects of article citations in a global context: the overall rate of citation of a country’s scientific publications, and the share of the world’s most highly cited literature authored by different countries. The discussion of article citations will conclude with an examination of citations to articles authored by researchers at U.S. academic institutions and in other U.S. sectors.

International Citation Patterns

Like the indicators of international coauthorship discussed earlier, cross-national citations are evidence that S&E research is increasingly international in scope. Citations to a country’s articles that come from articles authored outside that country are referred to as international citations. Between 1992 and 2012, the international share of citations increased in all but one of the world’s major S&E article–producing countries.

China is the exception. In 1992, 69% of citations to Chinese S&E articles came from outside China; by 2012, the proportion had dropped to 49% (figure 5-27). This suggests that China’s expanding S&E article output is being used mostly within China. However, changes in the composition of the Thomson Reuters database probably also play a role in accounting for this trend.[56] The trend toward domestic citations is also related to the unusually large role of domestic articles in Chinese output growth; the lack of international coauthors may explain, in part, the relatively low rate of international citations.

The relative citation index normalizes cross-national citation data for variations in publication output, much like the collaboration index (see sidebar, “Normalizing Coauthorship and Citation Data”). The expected value is 1.0, but unlike the collaboration index, citation indexes are not symmetric. When country A cites an article by country B, this does not mean that country B is also citing an article by country A. Table 5-26 shows the relative citation index for the year 2012 for major publishing locations in four regions: North America, the EU, Asia, and South America. These data show the following:

  • U.S. articles are most highly cited by articles from Canada (1.29) and the United Kingdom (1.15).
  • U.S. authors cite Chinese articles much less than expected (0.32).
  • Mexico is heavily cited by South American countries, ranging from 22% to 44% more than expected (index values from 1.22 to 1.44); likewise, Mexican authors cite South American articles more than they cite articles from other areas of the world.
  • Inter-European influence is strong, with most country pairs exhibiting index values greater than 1.0. Asian authors show similar interconnectedness, with the exception of Japan.

These data indicate the strong influence that geographic, cultural, and language ties have on citation patterns.

U.S. articles are more influential than those produced by the world’s other major publishing regions or countries. They receive 31% more citations than expected. U.S. index values for physics and chemistry are especially high, at 1.49 and 1.43, respectively, but in every field, U.S. articles are disproportionately cited (see figure 5-28).[57]

Trends in Highly Cited S&E Literature by Country

Another indicator of the performance of a national or regional S&E system is the share of its articles that are highly cited. High citation rates generally indicate that an article has a relatively great impact on subsequent research.

World citations to U.S. research articles show that, in all broad fields of S&E, U.S. articles continue to have the highest citation rates. In both 2002 and 2012, as displayed in appendix table 5-58, the U.S. share of articles in the 99th citation percentile was higher than its share in the 95th percentile, and these were higher than its share in the 90th percentile.[58] In 2012, although the United States authored 27% of the world’s total of 2.4 million articles in the cited period shown, the United States authored 46% of the articles in the 99th citation percentile.

U.S. publications uniquely display the preferred citation pattern: the higher the citation percentile, the higher the share of U.S. articles in the citation percentile. In contrast, EU articles are found disproportionately in the middle citation percentiles, while Chinese and Japanese articles are found disproportionately in the lower citation percentiles (see appendix table 5-58). Nevertheless, as the U.S. share of all articles produced declined between 2002 and 2012, its share of articles in the 99th percentile (i.e., the top 1%) of cited articles also declined, particularly in some fields. Shares in the top percentile increased for the EU and China but dropped slightly for Japan.

Between 2002 and 2012, 1.6%–1.8% of U.S.-authored S&E articles have appeared in the world’s top 1% of cited articles, compared with 0.7%–0.9% of articles from the EU (figure 5-29). The share of China’s articles in the top 1% remained behind the United States and the EU but increased from 0.1% to 0.6% over the period.

The high citation of U.S. articles has changed little over the past 10 years, remaining much higher than expected when compared to the overall U.S. share of world articles (figure 5-30; appendix table 5-57). Between 2002 and 2012, the EU index of highly cited articles for all fields combined rose slightly, to almost 1.0. The Japanese index remained the same and well below the expected value. China’s index rose substantially from 0.1 in 2002 to 0.6 in 2012, the same as Japan’s index.

U.S. articles are highly cited across all broad scientific fields, with indexes ranging from 1.3 to 2.2. The U.S. indexes across all these fields showed little change between 2002 and 2012. The greatest gain in the index of highly cited articles was in engineering, which grew from 1.7 to 2.0. The indexes for two fields—chemistry and social sciences—declined slightly (appendix table 5-57).

The EU’s articles are more highly cited than expected in two fields, agriculture (1.2) and physics (1.2) for 2012. The EU’s index values are what would be expected in two fields—astronomy and chemistry.

China is less highly cited than expected in all science fields except computer sciences, chemistry, and geosciences. Impressively, China’s index in computer sciences leaped from 0.2 in 2002 to 1.3 in 2012. Chinese geosciences articles experienced a similar rise from 0.2 to 1.1, while the index for chemistry has now just reached the expected value of 1.0.

Japan’s production of highly cited articles is lower than expected across all fields, although its index increased substantially in astronomy.

U.S. Cross-Sector Citation Trends

The relative citation index (described in the section on “International Citation Patterns”) can also be used to examine the influence that each U.S. sector has on U.S. S&E literature. Figure 5-31 shows the relative citation index values for each of the six sectors of U.S. institutions and how they have changed over the past 20 years. U.S. academic articles are at the citation level that would be expected and have maintained this level over the entire time period. State and local governments, industry, and FFRDCs historically have produced the U.S. articles with the lowest citation rates. Index values for industry articles have gradually declined over time. In contrast, articles authored at FFRDCs have shown a marked improvement since 2008, rising above the expected value of 1.0 by 2011 and finally ending the period as the second most highly cited U.S. sector.

Articles authored at federal government institutions always have been cited within the United States more than expected. Although the index value declined almost to 1.0 in the 1990s, it has since risen to 1.09. The U.S. articles with the relative greatest impact are those by nonprofit organizations. Counter to the federal government trend, index values rose over the 1990s to 1.29 but have been in decline in the past 10 years, dropping to 1.14 by 2012.

Citation of S&E Articles by USPTO Patents

Citations to the S&E literature on the cover pages of issued patents are one indicator of the contribution of research to the development of inventions.[59] To measure trends consistently, the analysis limits the cited article years to a specific moving window, just as is done for references from articles to articles. Unlike article-to-article citations, however, patents reference much older research, largely due to the length of time that passes from patent application to patent grant (i.e., pendency). Therefore, indicators in this section are based on an 11-year citation window after a 5-year lag. For example, citations from 2012 are references from patents issued in 2012 to articles published from 1997–2007.

According to this indicator, research links to invention increased sharply in the late 1980s and early 1990s (Narin, Hamilton, and Olivastro 1997). At the same time, patenting activity by academic institutions was increasing rapidly, as were patent citations to S&E literature produced across all sectors (NSB 2008:5-49–5-54).

After a slowdown in the late 1990s and early 2000s, referencing from patents to scientific literature is once again increasing. Of utility patents awarded to both U.S. and foreign assignees, 12% cited S&E articles in 2003, and this figure grew to 15% in 2012 (appendix table 5-59). In addition, the share of patent citations to foreign S&E articles has increased, coinciding with a growth in the percentage of U.S. utility patents awarded to foreign assignees and the share of world articles authored outside the United States. Starting in 2009, U.S. patents cited more foreign articles than U.S. articles.

Citations to U.S. articles in 2012 USPTO patents were dominated by articles in biological sciences (48%) and medical sciences (23%), along with chemistry (11%), engineering (7%), and physics (7%). These five fields account for 96% of the total (figure 5-32; appendix table 5-60). The patents citing U.S. articles are concentrated in three technology areas—pharmaceuticals, chemicals, and biotechnology—that together make up 63% of the total (figure 5-32).

The proportion of U.S. articles cited in U.S. patents that were authored by industry and federal government dropped between 2003 and 2012, largely because citations to academic articles increased (appendix table 5-59). Citations to academia grew from 59% to 65% of total citations to U.S. articles in that time period. This trend was stronger in some fields than in others. It was especially pronounced in engineering (from 50% to 68%), mathematics (from 71% to 89%), physics (from 51% to 68%), and psychology (from 67% to 83%). Despite the increasing proportion of citations to academic articles overall, citations to academic agricultural science articles actually decreased (from 67% to 63%) (appendix table 5-60).

Articles from other sectors receive far fewer citations in patents, but this varied by field (figure 5-33). After academia, industry articles capture the next-largest share of citations in every major field except medical sciences, ranging from 12% (medical sciences) to 22% (engineering). In medical sciences, nonprofit articles garner 16% of patent citations.

Energy and Environment–Related Patent Citations

Clean energy and energy conservation and related technologies—including biofuels, solar, wind, nuclear, energy efficiency, pollution prevention, smart grid, and carbon sequestration—are closely linked to scientific R&D and have become a policy focus in the United States and other countries. NSF developed a method for identifying patents with potential application in these technologies. (See sidebar “Identifying Clean Energy and Pollution Control Patents” for details on the filters.)

Chapter 6 of this volume presents extensive data on the patents in four technology areas related to clean energy—alternative energy, pollution mitigation, smart grid, and energy storage—including the nationality of their inventors. (See chapter 6, “Industry, Technology, and the Global Marketplace,” section “Patenting of Clean Energy and Pollution Control Technologies.”) This section reports on the citations in those patents to the S&E literature, using those citations to indicate the linkages between S&E R&D and the potential for practical use of the results of those R&D projects in new inventions and technologies.[60] The citation data are based on patents issued between 2003 and 2012.

U.S. patents in these four areas of clean energy technology cite more foreign literature than U.S. literature (appendix table 5-61). In contrast, patents in all technology areas have consistently cited more U.S. literature than foreign literature (appendix table 5-59).

Within citations to U.S. literature, articles authored by the academic sector accounted for the most citations (70%) among U.S. sectors in 2012. Industry and FRRDCs were the next largest, accounting for 12% and 10% of citations, respectively. Between 2003 and 2012, academia’s share of citations to U.S. literature increased from 59% to 70%. Industry’s share fell from 22% to 12%.

Four broad S&E fields dominate the citations to S&E literature in these four patent areas: chemistry, physics, engineering, and biological sciences. The range of S&E fields cited indicates that these developing technologies rely on a wide base of S&E knowledge.

The S&E fields cited by these patents are shown in table 5-27. These four categories of energy and environment–related patents show somewhat different patterns of reliance on S&E literature. In both energy storage and smart grid, referencing is concentrated in a single field. For energy storage patents, over half of all citations are to chemistry articles; for smart grid patents, engineering is similarly dominant. Alternative energy and pollution mitigation citations are more evenly distributed across the four fields; for both of these technologies, however, chemistry is the most heavily cited field, receiving roughly one-third of all citations.

Using patent citations as an indicator, the data show that chemistry research contributes heavily to invention in all areas of green technology with the exception of smart grid, where engineering dominates. Geoscience articles, which in this taxonomy include environmental sciences, are prominent as well, but only in pollution mitigation.

Academic Patenting

Academic institutions whose research leads to intellectual property attempt to protect and benefit from the fruits of their labor through patents and associated activities. The majority of U.S. universities did not become actively involved in managing their own intellectual property until late in the 20th century, when the Bayh-Dole Act of 1980 gave colleges and universities a common legal framework for claiming ownership of income streams from patented discoveries that resulted from their federally funded research. Other countries implemented policies similar to the Bayh-Dole Act by the early 2000s, giving their academic institutions (rather than inventors or the government) ownership of patents resulting from government-funded research (Geuna and Rossi 2011). To facilitate the conversion of new knowledge produced in their laboratories to patent-protected public knowledge that potentially can be licensed by others or form the basis for a startup firm, many U.S. research institutions established technology management/transfer offices (AUTM 2009).

The following sections discuss overall trends in university patenting and related indicators through 2011 and 2012.

Trends and Patterns in Academic Patenting

USPTO granted 8,700 patents to U.S. and foreign universities and colleges in 2012, 3.4% of USPTO patents granted to all U.S. and foreign inventors (figure 5-34). U.S. universities and colleges were granted 5,100 USPTO patents, with foreign universities receiving 3,600.

Patenting by academic institutions has increased markedly over the last two decades—from 1,800 in 1992 to 8,700 in 2012—resulting in their share of all USPTO patents doubling from 1.8% to 3.4%. Patenting by U.S. institutions outpaced overall growth of USPTO patents in the 1990s, resulting in their share of all patents increasing from 1.6% in 1992 to 2.4% in 1999. Although the number of U.S. academic patents continued to grow from 2000 to 2012, the U.S. university and college share of all USPTO patents declined slightly (appendix table 5-62). In contrast, USPTO patents granted to foreign universities and colleges grew much more rapidly than those granted to U.S. universities and colleges in the 2000–12 period. U.S. patents to foreign universities and colleges grew sixfold to reach 3,600 patents; their share of all USPTO patents rose from 0.4% in 2000 to 1.4% in 2012 (figure 5-34).[61]

Patenting by U.S. and foreign universities and colleges in another major patent office, the European Patent Office (EPO), shows a similar trend of increasing activity (figure 5-35). The academic share of all patents granted by EPO increased from 0.9% in 1992 to 2.4% in 2012. After steadily increasing in the 1990s and early 2000s, the number of EPO patents granted to U.S. universities and colleges has remained flat at approximately 500–600 patents since 2003. In contrast, patenting by foreign universities and colleges grew more rapidly in the 2000s, and they surpassed U.S. universities in 2007.

The top 200 R&D-performing institutions dominate among U.S. universities and university systems receiving patent protection, with 98% of the total patents granted to U.S. universities between 1997 and 2012 (appendix table 5-62).[62] Among these institutions, 19 accounted for more than 50% of all patents granted to the top 200 (some of these were multicampus systems, like the University of California and the University of Texas). The University of California system received 11.3% of all U.S. patents granted to U.S. universities over the period, followed by the Massachusetts Institute of Technology, with 4.2%.

Biotechnology patents accounted for the largest share (25%) of U.S. university patents in 2012 (appendix table 5-63). Biotechnology has been the largest technology area for U.S. academic patenting since 1991. Pharmaceuticals, the next-largest technology area, has had a declining number of patents over the past decade, dropping from an average of 491 a year in 1998–2002 to 369 a year in 2008–12 (figure 5-36). Medical equipment shows a similar, but much smaller, decline. The other major technology areas have been increasing. Patents for semiconductors have made the greatest increase, from around 90 patents per year in 1993–97 to around 210 in 2008–12.

Commercialization of U.S. Academic Patents

Universities commercialize their intellectual property by granting licenses to commercial firms and supporting start-up firms formed by their faculty. Data from the Association of University Technology Managers (AUTM) indicate continuing growth in a number of such patent-related activities. Invention disclosures filed with university technology management/transfer offices describe prospective inventions and are submitted before a patent application is filed. These grew from 12,600 in 2002 to 19,700 in 2011 (notwithstanding small shifts in the number of institutions responding to the AUTM survey over the same period) (figure 5-37). Likewise, new U.S. patent applications filed by AUTM university respondents also increased, nearly doubling from 6,500 in 2002 to 12,100 in 2011. However, U.S. patents awarded to AUTM respondents stayed flat over the period, rising only in the last 2 years and reflecting a similar rise in the number of patents granted to all assignees (see appendix table 5-62).[63]

Despite the economic slowdown of the past 5 years, the number of new startup companies formed continued to rise, as did the number of past startups still operating; AUTM survey respondents reported a low of 348 startup companies formed in 2003 and a maximum of 617 in 2011, with a total of extant startup companies in 2011 of 3,573 (appendix table 5-64). Licenses and options that generated revenues also increased over the period. Active licenses increased steadily from 18,800 in 2001 to 33,300 in 2011.

Most royalties from licensing agreements accrue for relatively few patents and the universities that own them, and many of the AUTM respondent offices report no income. (Thursby and colleagues [2001] report that maximizing royalty income is not the dominant objective of university technology management offices.) At the same time, large one-time payments to a university can affect the overall trend in university licensing income. In 2011, the 157 institutions that responded to the AUTM survey reported a total of $1.5 billion in net royalties from their patent holdings. This is essentially the same amount reported for the last 3 years. Perhaps as a result of the nation’s economic downturn, this number is down sharply from the high value of $2.1 billion reported in 2008 (appendix table 5-64).

[43] For more information on the World Bank economic classification of countries, see
[44] Countries with indexed S&E articles can change their borders over time. Data on Hong Kong, for example, were formerly reported separately but are now included in totals for China. See appendix table 5-24 for a list of the locations represented in the data.
[45] Statements that a country “authors” a certain number of articles are somewhat imprecise, especially given the growing rates of international collaboration discussed later in this chapter. See the sidebar “Bibliometric Data and Terminology” for more information on how S&E article production and collaboration are measured.
[46] See Eades et al. (2005) for a discussion of recent reforms in Japan’s higher education system. Japan’s R&D expenditures increased by 14% to reach 17.4 trillion yen between 2000 and 2008, according to the Organisation for Economic Co-operation and Development (
[47] Publication traditions in broad S&E fields differ somewhat. For example, although all fields publish journal articles, computer scientists often publish their findings in conference proceedings, and social scientists often write books and also publish in journals. Proceedings and books are poorly covered in the data currently used in this chapter.
[48] Social science journals tend to focus on local issues, have less international author diversity, and publish in a language other than English more often than natural sciences journals—all criteria for exclusion from the Thomson Reuters databases. The lower concentration of articles in social sciences, other life sciences, and psychology in foreign countries may be partially attributed, then, to journal coverage. For further details on Thomson’s journal selection process, see
[49] The U.S. sector identification in this chapter is quite precise; to date, sector identification has not been possible for other countries.
[50] The 16 federally funded research and development centers (FFRDCs) sponsored by the Department of Energy (DOE) dominated S&E publishing by this sector. Across all fields of S&E, DOE-sponsored labs accounted for 83% of the total for the sector in 2005 (NSB 2008). Scientists and engineers at DOE-sponsored FFRDCs published 96% of the sector’s articles in chemistry, 95% in physics, and 90% in engineering (see “S&E Articles From Federally Funded Research and Development Centers” [NSB 2008:5-47]). Nine other federal agencies (including the Departments of Defense, Energy, Health and Human Services, Homeland Security, Transportation, and Treasury; the National Aeronautics and Space Administration; the Nuclear Regulatory Commission; and the National Science Foundation) also sponsor another 23 FFRDCs (NSF/SRS 2009).
[51] Coauthorship is a broad, though limited, indicator of collaboration among scientists. Previous editions of Indicators discussed possible underlying drivers for increased collaboration, including scientific advantages of knowledge sharing and instrument sharing, decreased costs of travel and communication, and national policies (NSB 2006). Katz and Martin (1997), Bordons and Gomez (2000), and Laudel (2002) analyze limitations of coauthorship as an indicator of research collaboration. Despite these limitations, other authors have continued to use coauthorship as a collaboration indicator (Adams et al. 2005; Gomez, Fernandez, and Sebastian 1999; Lundberg et al. 2006; Wuchty, Jones, and Uzzi 2007; Zitt, Bassecoulard, and Okubo 2000).
[52] Readers are reminded that the number of coauthored articles between any pair of countries is the same; each country is counted once per article in these data. However, countries other than the pairs discussed here may also appear on the article.
[53] Finland is included here as one of the Scandinavian countries; Iceland is not.
[54] “Influence” is used here broadly; even citations that criticize or correct previous research indicate the influence of that previous research on the citing article.
[55] For example, 2012 citation rates are from references in articles in the 2012 data file to articles contained in the 2008–10 data files of the Thomson Reuters Science Citation Index and Social Sciences Citation Index databases. Analysis of the citation data shows that, in general, the 2-year citing lag captures the 3 peak citation years for most fields, with the following exceptions: in astronomy and physics, the peak citation years are generally captured with a 1-year lag; in computer sciences, psychology, and the social sciences, the peak citation years are generally captured with a 3-year lag.
[56] Some part of this percentage decrease may reflect the increase in Chinese journals in the Science Citation Index and Social Sciences Citation Index databases used in this chapter. Since more Chinese authors in these journals are available to cite their Chinese coauthors, international citations to Chinese-authored articles are declining as a share of total citations. However, accounting for the “nationality” of a journal is not straightforward, and the data file used by the National Science Foundation (NSF) excludes journals that are primarily of regional interest. NSF’s estimate of “Chinese” journals shows an increase of 75% over the past decade, compared to an increase of 334% for Chinese-authored articles.
[57] Because different S&E fields have different citation behaviors, these indicators should be used with caution. For example, articles in life sciences tend to list more references than, for example, articles in engineering or mathematics. Thus, a country’s research portfolio that is heavily weighted toward life sciences (e.g., the United States) may receive proportionately more citations than a country whose portfolio is more heavily weighted toward engineering or mathematics.
[58] Percentiles are specified percentages below which the remainder of the articles fall. Thus, the 99th percentile identifies the number of citations 99% of the articles failed to receive. Across all fields of science, 99% of articles from 2008 to 2010 failed to receive at least 21 citations in 2012. Matching numbers of citations with a citation percentile is not precise because all articles with a specified number of citations must be counted the same. Therefore, the citation percentiles discussed in this section and used in appendix tables 5-57 and 5-58 have all been counted conservatively, and the identified percentile is in every case higher than specified (i.e., the 99th percentile is always greater than 99%, the 95th percentile is always greater than 95%, and so forth). Actual citations/percentiles per field vary widely because counts were cut off to remain within the identified percentile. For example, using this method of counting, the 75th percentile for engineering contained 2008 to 2010 articles with 3–4 citations from 2012 articles, whereas the 75th percentile for astronomy contained articles with 6–10 citations. A country whose research influence was high would have greater proportions of articles in the higher-citation percentiles, whereas a country whose influence was low would have greater proportions of articles in lower-citation percentiles.
[59] Patent citations to S&E research discussed in this section are limited to the citations found on the cover pages of successful patent applications. These citations are entered by the patent examiner and may or may not reflect citations given by the applicant in the body of the application. Patent cover pages also contain references to scientific and technical materials not contained in the article data used in this chapter (e.g., other patents, conference proceedings, industry standards). Analyses of the data referred to in this section found that nonjournal references on patent cover pages accounted for 19% of total references in 2008. The journals/articles in the Science Citation Index and Social Sciences Citation Index databases used in this chapter—a set of relatively high-impact journals—accounted for 83% of the journal references, or 67% of the total science references, on the patent covers.
[60] In this discussion, patent data are patents granted by the U.S. Patent and Trademark Office to all assignees, not just U.S. assignees. S&E publication data are for all publications in all U.S. sectors and for all country authors.
[61] Patent-based data must be interpreted with caution. Year-to-year changes in the data may reflect changes in U.S. Patent and Trademark Office processing times (so-called “patent pendency” rates) and attempts to reduce the backlog of patent applications that build up from time to time. Likewise, industries and companies have different tactics and strategies for pursuing patents and otherwise protecting intellectual property, and these also may change over time.
[62] The institutions listed in appendix table 5-62 are slightly different from those listed in past volumes, and data for individual institutions may be different. In appendix table 5-62, an institution is credited with a patent even if it is not the first assignee, and therefore some patents may be double counted. Several university systems are counted as one institution, and medical schools may be counted with their home institution. Universities also vary in how they assign patents (e.g., to boards of regents, individual campuses, or entities with or without affiliation with the university).
[63] Other than for general trends, the patent counts reported by Association of University Technology Managers respondents in figure 5-37 and appendix table 5-64 cannot be compared with the patent counts developed from U.S. Patent and Trademark Office data as in appendix tables 5-62 and 5-63.