The increasing linkage between U.S. technology and public science

ELSEVIER

research policy Research Policy 26 (1997) 317-330

The increasing linkage between U.S. technology and public science Francis Narin *, Kimberly S. Hamilton, Dominic Olivastro CHI Research Inc., 10 White Horse Pike, Haddon Heights. NJ 08035, USA

Abstract

A detailed and systematic examination of the contribution of public science to industrial technology would be useful evidence in arguing the case for governmental support of science. This paper provides such an examination, by tracing the rapidly growing citation linkage between U.S. patents and scientific research papers. Seventy-three percent of the papers cited by U.S. industry patents are public science, authored at academic, governmental, and other public institutions; only 27% are authored by industrial scientists. A strong national component of this citation linkage was found, with each country's inventors preferentially citing papers authored in their own country, by a factor of between two and four. Particularly rapid growth was found for the dependence of patented technology on U.S. papers. References from U.S. patents to U.S.-authored research papers have tripled over a six-year period, from 17,000 during 1987-1988 to 50,000 during 1993-1994, a period in which the U.S. patent system grew by only 30%. The cited U.S. papers are from the mainstream of modern science; quite basic, in influential journals, authored at top-flight research universities and laboratories, relatively recent, and heavily supported by NIH, NSF, and other public agencies. © 1997 Elsevier Science B.V.

l. Introduction

Among both scientists and economists it is widely accepted that public science - - scientific research that is performed in academic and governmental research institutions and supported by governmental and charitable agencies - - is a driving force behind high technology and economic growth. This report provides quantitative evidence of both the magnitude and the direction of that force, based on tracing tens of thousands of references from recent United States patents to the scientific research papers they cite. The premise that academic research makes an important contribution to economic growth is well

* Corresponding author. Tel.: + 1-609-546-0600;fax: + 1-609546-9633; e-mail: [email protected].

accepted across the political spectrum. For example, in his statement on Technology for America's Economic Growth (Clinton and Gore, 1993), President Clinton stated that "Scientific advances are the well-spring of technical innovations. The benefits are seen in economic growth, improved health care, and many other areas". This acceptance is quite international, as evidenced by the current Japanese expansion of basic science, which is very specifically linked to economic needs (Normile, 1997). The economic impact of science has, of course, long been a motivation for the government's support of academic research. During an earlier period of tight budgets some 30 years ago, Weinberg (1963) eloquently discussed the criteria for scientific choice, mentioning 'technological merit' as one of the important factors in determining the support of a scien-

0048-7333/97/$17.00 © 1997 Elsevier Science B.V. All rights reserved. PII S0048-7333(97)00013-9

318

F. Narin et al. / Research Policy 26 (1997) 317-330

tific field. The importance of this support was recognized even earlier in the fundamental assessments of science by Bush, 1945 (reprinted 1960), at a time when science was just emerging as a major area of governmental concern, and has recently been reviewed by Martin and Salter, 1996 in a report to the U.K. Government. Furthermore, Rosenberg and Birdzell (1990) have attributed much of the 'Western Miracle' to the ability of the free western economies to absorb and utilize scientific knowledge; they specifically assert that "close links between the growth of scientific knowledge and the rise of technology have permitted the market economies of the Western nations to achieve unprecedented prosperity". Mansfield (1991) has added a quantitative underpinning to this; from his survey data he estimates that 10% of industrial innovation would not have occurred, or would have occurred with great delay, without the contribution of academic research. The notion that technology springs from a scientific base was originally embedded in the 'linear model' of innovation: from basic research through applied research continuing into technology and resultant economic benefit. This simple linear model has been supplanted by much more complex views of the process, with many feedback loops and major contributions of technology to science, but the origins of research knowledge in basic research still lie at the core of the process (Turney, 1991). This paper will provide large-scale evidence detailing a massive, contemporary linkage between industrial technology and public science, with a tripling of the knowledge links from U.S. technology to U.S. science in just six years. This analysis of more than 100,000 patent-to-science references provides evidence that public science plays a crucial role in patented industrial technology. For example, the data show that only 17% of the science references on the front pages of drugs and medicine patents are to U.S. industry papers, while 50% are to U.S. public science, and 33% are to foreign science, most of which is also public. The data also show that: (1) the patent-to-science linkage has a strong national component, with U.S.-authored scientific papers particularly heavily cited by U.S.-invented patents; (2) the linkage is appropriately subject specific, with chemical patents citing chemistry papers, drugs and

medicine patents citing biomedical papers, and so forth; and (3) that the linkage is quite contemporaneous and quite basic, with the references on U.S. patents citing recent papers situated at the basic end of the research spectrum. This confirms the preliminary findings in our previous papers on linkage between technology and science (Narin et al., 1995). The present paper shows that these linkages are increasing at a rapid rate, and adds two major new dimensions: data on cited U.S. research institutions and data on research support, and places this information, for the first time, in a context linking U.S. industry and public science (Narin et al., 1996).

2. Methodology This paper is based on an analysis of the 430,226 non-patent references (NPR's) which were listed as 'other references cited' on the front pages of the 397,660 U.S. patents issued in 1987-1988, and 1993-1994. For regular U.S. patents issued in 1993 and 1994, on average there were 8.6 references/patent to U.S. patents, 1.9 references/patent to foreign patents, and 1.5 non-patent references/patent. The 'references cited' on U.S. patents are a fundamental requirement of patent law. When a U.S. patent is issued it has to satisfy three general criteria: it must be useful, it must be novel, and it must not be obvious. The novelty requirement is the primary factor leading to the references which appear on the front page of the U.S. patent, since it is the responsibility of the patent applicant and his attorney, and of the patent examiner, to identify, through various references cited, all of the important prior art upon which the issued patent improves. These references are chosen a n d / o r screened by the patent examiner, who is "not called upon to cite all references that are available, but only the best" (Patent and Trademark Office, 1995). Non-patent references (NPR's), as they appear on the front pages of U.S. patents, are a mixed set of references to scientific journal papers, meetings, books, and many non-scientific sources such as industrial standards, technical disclosures, engineering manuals, and every other conceivable kind of published material.

F. Narin et al. / Research Policy 26 (1997) 317- 330

All of the 430,226 NPR's underwent a standardization process to put them into various categories, such as science references, abstracts, and books. Of the 430,226 NPR's in those four years, approximately 242,000 were judged to be science references that is, citations to scientific journal papers, scientific meetings and other scientific publications. A further subset of approximately 175,000 of the science references are citations to papers published in the 4000 or so journals covered by the Science Citation Index (SCI). The 175,000 SCI citations were completely unified into a standardized journal, volume, page and 3,ear format, and matched to the SCI-based Science Literature Indicators Database (SLID), maintained at CHI Research for the National Science Foundation. The resulting linkages between the citing patents and the cited scientific research papers were analyzed for two specific time-periods: 1987-1988 patents citing papers published during the l 1-years 1975-1985 t40,000 citations), and 1993-1994 patents citing papers published over the 1l-years between 1981 - 1991 t,104,000 citations). These 1 l-year citation windows contain 80% of the cited papers. Of the 144,000 citations, 109,000 or 76% were successfully matched to the SLID. Most of the unmatched references were either incomplete or erroneous; for example, a reference might omit the author's name, or misspell it, or give the wrong volume or page numbers. The next step was to identify the author addresses on the papers. Since the match was to the SLID, the address of each author institution was available, and the individual institutional identifications standardized for more than 3500 major U.S. publishing institutions. The final step in our acquisition of data was to identify the agencies supporting the U.S.-authored papers. A library search was undertaken for the 45,000 matched papers with at least one U.S. author, to locate a copy of the paper in one of the Philadelphia area libraries. For the located papers, approximately 95%, a tabulation was made of all the U.S. and foreign research support agencies that were acknowledged in the papers. Of all the cited papers found, approximately 63% acknowledged some source of external support. External support was, of course, most frequently acknowledged in the academic papers. Papers in which -

-

319

the authors come from the private sector, or only from government agencies such as the National Institute of Health (NIH) or the Agricultural Research Service (ARS), tended not to contain an explicit support acknowledgment; the research reported in those papers is often funded directly from the internal budgets of the organizations themselves. Three final methodological notes are important. First, the data are extremely complex, in that there are many different ways counts can be made. A given paper may be cited in several patents and a given patent may cite a number of different papers. In addition, a paper often has authors from more than one institution, and may be supported by more than one agency. This combination of many patents citing many papers with many different institutional addresses and support sources gives rise to various complexities in counting and presenting data. The second methodological note concerns the limitation of this analysis to the science references on the front pages of the U.S. patents, omitting any analysis of references in the body of the patents. The reasons for this are both theoretical and practical. Theoretically, the front page references should be the most important ones on a patent, since they are the ones relied upon, as mentioned previously, by the examiner in establishing the patent's novelty. Furthermore, from a practical viewpoint it is far more difficult to extract the non-patent references scattered through the text of a patent. In one brief study conducted to evaluate the representativeness of the front page science references, we found that approximately half of all the science references are on the front page, and that there was great similarity between front page and text science references (Narin and Olivastro, 1988). As a result it is quite reasonable to assume that the science references analyzed here are representative of all of the science references in the patent and probably underestimate the dependence of technology upon science by a significant factor. The increasing dependence of modern technology on science revealed in this paper is, almost certainly, a very conservative estimate of that important linkage. Finally, the data presented herein measure linkages in codified knowledge, as reflected in patents and papers, and understates the important contributions that public science makes to industry through

F. Narin et al. / Research Policy 26 (1997) 317-330

320

research training, which predominates in technology-based industries such as automobiles and aircraft.

basis, and in Table 1 on a country-by-technology basis. Fig. 1 shows that science linkage is increasing fastest for U.S. and U.K.-invented patents, but also increasing steadily in U.S. patents with French, German, and Japanese inventors, as it is for patents from almost all other inventor countries. The highly science-linked position of the U.S. and the U.K., compared to the other three major countries, is the result of a combination of two factors: a tendency within all technologies for U.S. and U.K. patents to be more science-linked, and the comparatively high level of U.S. and U.K. patenting activity in drugs and medicine, biotechnology, and medical instrumentation, all of which are highly science-linked. Another important characteristic of science linkage is that it is very subject specific. Fig. 2 shows that drugs and medicine patents invented in all countries cite almost exclusively to papers in the scientific fields of clinical medicine and biomedical research. Similarly, chemical patents cite chemistry

3. Overall linkage characteristics The U.S. patent system is quite representative of the world's technology. Approximately half of the inventors of U.S. patents are foreign, and each country's inventors patent in the U.S. in rough proportion to their country's Gross Domestic Product (GDP) (Narin, 1991). In addition, the patent system covers the whole range of technology, from old but still active classes representing such basic mechanical technologies as railroads, to the most modern human genetics technology. Across all of these countries and technologies there has been a steady increase in science linkage for at least two decades. The last decade of this increase is shown in Fig. 1 on a country-by-country 1.6

US 1.4

1.2 UK

:5

0.8.

,~

0.6.

g 0.4.

0.2

0 1985

I 1986

~

÷ 1987

1988

; 1989

[ 1990

~

4 1991

1992

....

4

.... 4--

1993

1994

I 1995

Fig. 1. Science linkage: citations from patents to papers are increasing fastest in U.S. and U.K. invented U.S. patents.

321

F. Narin et al. / Research Policy 26 (1997) 317-330

Table 1 Science referencing: the average number of science references per U.S. patent are consistently higher for U.S. and U.K. invented U.S. patents Inventor country

Chemical patents

U.S L .K. Japan France Germany All Countries

Drug and medicine patents

Electrical component patents

Prof. and sci. instruments patents

1985

1990

1995

1985

1990

1995

1985

1990

1995

1985

1990

1995

0,94 0.68 0.44 0,32 0.44 0.74

1.85 1.05 0.67 0.63 0.63 1.30

4.63 2,50 1.28 1.05 1.34 3.18

3.05 1.33 1.06 1.24 0.97 2.17

5.48 2,66 1.57 1.27 1.67 3.78

11.61 5.26 3.26 2.49 3.54 8.66

0.53 0,44 0.31 0.54 0.44 0.45

0.93 0,70 0.46 0,64 0.64 0.73

1.28 1.20 0.69 0.79 0.98 1.00

0.58 0,39 0.13 0.31 0,24 0.41

0.84 0,76 l/. 17 0.43 0.33 0.58

1.72 1.35 0.42 1.02 0.55 1.27

papers heavily, and electronics patents cite to physics and e n g i n e e r i n g papers, and there is throughout the linkage p h e n o m e n o n a subject-specific couple between the t e c h n o l o g y of the patent and the science upon w h i c h it is building. T h e linkage is also quite recent; the peak cited year for papers cited in drugs and m e d i c i n e patents is 4 - 6 years prior to patent grant, only a year or two s l o w e r than citing from b i o m e d i c a l papers to b i o m e d i c a l papers.

Another important characteristic of this linkage is the strong national c o m p o n e n t . Each c o u n t r y ' s inventors in the U.S. patent system cite their o w n c o u n t r y ' s papers two to four times as often as expected, when adjusted for the size of a c o u n t r y ' s scientific publication rate. For e x a m p l e , a p p r o x i m a t e l y 7% o f all the papers in the SCI are authored by scientists at German institutions. C o m p a r e d to that, a p p r o x i m a t e l y 17% of all of the science papers cited on the front

e,i

e,in

Mad & Biomed Res

;tW

:hnology Field of

Cited Paper

Germany

Fig. 2. Within discipline citing: drug and medicine patents heavily cite to clinical medicine and biomedical research journals.

322

F. Narin et al./Research Policy 26 (1997) 317-330

pages of German-invented U.S. patents are Germanauthored papers. The ratio of those two percentages represents a national citation ratio of approximately 2.4, in German-to-German patent-to-paper citation. This is a completely general phenomena, and is apparent for all the five major countries in Fig. 3. If each country's inventors were equally likely to cite any of the world's science, then the height of each column in Fig. 2 would be 1. The fact that the diagonal columns are significantly greater than unity, ranging from one to four, indicates the strong domestic component that exists in science linkage, showing that each country's inventors are preferentially building upon their own domestic science. It should be noted that the same national preference is apparent when looking at cross-national citations from papers-to-papers and patents-to-patents. In both of those cases there is just as strong a national component in the linkage; all of these citing phenomena show that there are still very strong national

ties between scientists within a country and inventors within a country, and between national inventors and scientists, implying that a strong scientific base is necessary for a strong national technology, in science dependent areas of technology such as biotechnology, agricultural chemicals and plastics. The final general point about this linkage is that the papers cited in patents are published in prestigious, mainstream, basic scientific research journals. Table 2 shows the 25 journals most heavily cited by 1993-1994 patents for the four most heavily cited scientific categories: biomedicine, combining clinical medicine and biomedical research, physics, chemistry, and engineering and technology. The journals shown on Table 2 are clearly prestigious and influential, and for the biomedical and chemistry papers, quite basic. The physics journals, however, are not as basic, with much of the physics cited in patents published in applied physics journals, rather than the more

0 .m

iv C 0

U

pan

f Cited Paper 81-93 Cour

1993-94

US

Fig. 3. National citation ratios: the linkage between technologyand science has a strong national component.

x5 76 53 78 66

I I? 112 I I2 IO0 Y9

J. Heteru. Chem. J. Catal. Polym. Bull. Heterocyclc Organometal.

28X

266 283 256 251

409

400 380 36.5 362

FEBS Lett. Blood J. Cell Biol. N. Eng. J. Med.

IO1 90 98 IO9 IO1

I68 166 164 I51 I26

J. Appl. Poly. Polymer J. Phys. Chem. Synthesis-S B. Chem. S. J.

308 335 310 380 300

4YO 482 476 472 422

Analyt. Biochrm. Mol. Cell B J. Exp. Med. Infec. Immun. Biochem. Biophys. A J. Clin. Inv.

140 145,

212 204

449 381

593 496

J. Virology J. Bact.

179 I I4 149

241 224 218

406 485 420

737 679 679

Gene Cancer Res EMBO J.

216

277

J. Cryst. Growth J. Pol. Sci. Angew. Chetn. Macromol. Chem. Inorg. Chem. Chem. Lett.

543

821

632 412 324 346 270 I94 I88

945 590 472 460 379 314 307

II29 I243 646 644 669 686 524

2344 1874 II36 II21 1072 979 922

Science J. Biol. Chem. J. Med. Chem. Cell Nucl. Acid Res. J. Immunol. Biochem. Biophys. Res. Biochem.

673

811

# Papers

991

117.5

2398

Nature

1394

J. Am. Chem. SW. Tetrahedron Len. J. Org. Chem. J. Chem. Sot. Analyt. Chem. J. Elchem. Sot. J. Chromatogr. Macromolec. Tetrahedron

2134

3835

P. Nas. US.

# Cites

Journal

Chemistry journals

for lYY3-‘YJ

# Papers

# Cites

Journal

Biomedical journals

Table 2 Numbers of citations and ctted papers for top 25 cited journals

535 569 435 372 I68 242 I95 204

773 771 670 472 349 347 299 298

Jpn. J. Appl. Phys.

Molec. Cryst. Physica C Appl. Phyx Opt. Commun. Solid State C’ommun

J. Opt. Sot.

114 102 Y2 80 80

70 82 61 67 56

94

I08 103 I08 77 98

172 172 I54 I52 I36 134

I20 122

I56 182 I46

I88 I81

228 222 203

Rev. Sci. Ins. Nucl. Instrum. Phys. Rev. Opt. Eng. Solid State Technology J. Magn. Res. IEEE Acoust. IEEE Photon. Synth. Metal J. Phys.

286

Thin Solid Films

I74

I356

2210

Appl. Phys. Lett.

J. Appl. Phys. J. Vat. Sci. Technol. Appl. Optics Phys. Rev. Lett. Optics Lett. IEEE J. Q. El. J. Lightw. T.

# Papers

# Cites

Physics journals Journal

AT&T Tech. J J. Alloy Corn. Ind. Eng. Res. J. Elec. Mat.

J. Metal.

Comm. Acm. Tappi J. IEEE Patt. A J Mater Res Metallurg. T.

J. Non-Cryst. Computer

IEEE Cons. E IBM J. Res. IEEE Nucl. S

J. Mater. Sci.

67 62 60 56

70

87 86 73 70 70

92 89

122 122 93

I52

345 233 I81 166 I64 I63 162

721

1812

# Cites

and technology

IEEE Elec. D IEEE Commun J. Am. Ceram. P. IEEE IEEE Comput. IEEE J. Sel. Sensor Actuat.

Electr. Lett.

IEEE Ttrana.

Engineering

43 47 37 41

28

59 59 50 44 44

70 65

104 I IO

220 I43 144 91 103

521

1206

# Papers

journals

3

7 2

,‘ : u k

2

2 p \

ze ;.

F. Narin et al. /Research Policy 26 (1997) 317-330

324

basic theoretical and high energy physics journals. The Physical Review, for example, the largest and one of the most prestigious physics journals, is only thirteenth on the physics list, led by more applied journals. As would be expected in engineering and technology, many of the journals are electrical engineering journals, since there is such a very large amount of patenting in the electronics field, and many of these patents cite to papers in IEEE and other electronics-related journals.

proximately 58% of the SCI-covered papers cited in patents in the four years had at least one U.S. author, a total of 44,765 papers. An attempt was made to locate each of these papers in the library, in order to verify the institutional addresses of the authors, and to tabulate the research support acknowledgments. Fig. 4 shows the most dramatic finding: that there has been a remarkable increase in linkage between U.S. patents and U.S.-autbored scientific papers; in just six years the number of U.S.-authored papers cited in patents has more than doubled, the number of citations to these papers almost tripled, and the number of research support acknowledgments on the papers more than tripled. Since over this period of time the number of patents in the U.S. system has only grown by 30%, an increase in science linkage of 200% is truly remarkable, and indicative of a rapidly increasing dependence of patented technology upon contemporary science.

4. Cited U.S. science This section will describe the U.S.-authored scientific papers cited in 1987-1988 and 1993-1994 U.S. patents. Because of the tendency for U.S.-invented patents to cite U.S. papers, the high activity of the U.S. in science-linked areas of technology, and the large size of the U.S. scientific establishment, ap-

[ ] 1987-88 citing 1975-85

70000-

m 1993-94 citing 1981-91

60000-

50000-

30000-

20000-

10000-

Number of Papers Cited

Number of Citations

Number of Support Sources

Fig. 4. Increasein science linkage: citations to U.S. authored papers have tripled in six years.

F. Narin et al. / Research Policy 26 (1997) 317 330

A second important point is that these cited research papers are authored at prestigious universities and laboratories, just as they were published in prestigious journals. This is shown in Table 3, which lists the top 25 author institutions for the four categories. Note that in Table 3 we have combined the universities and their medical schools. Most of the biomedical papers listed there as Harvard University are papers which are published at Harvard Medical School. Note also that those are whole counts of citations. In a whole count, if a paper has authors at two institutions, say Harvard and Stanford, it is counted as one citation to each institution. In a fractional count, used in some of the data later on, a cited paper with authors at Harvard and Stanford would be counted as one-half a citation to each. For the biomedical papers it was not surprising that the National Cancer Institute (NCI) appears very high on the list, since NCI has a large Intramural

325

research program. The high ranking of the Veterans Administration (VA) was somewhat surprising. However, almost all of these VA papers are coauthored with scientists at the medical schools associated with the VA hospitals, indicating a close and productive linkage between the VA and the academic community. A number of the major private companies are high on the list for chemistry papers, with DuPont listed as fourth, A T & T Bell Labs. sixth, and IBM seventh. For both physics papers, and engineering and technology papers, the top two cited organizations are A T & T Bell Labs. and IBM, both of which have large numbers of patents, and many Bell Labs. and IBM patents cite their own company papers. However, there is also much citation IYom other patents to the A T & T and IBM papers, and this is by no means a reflection of excessive company self-citation. For example, only 19% of the patent citations

'Fable 3 Number of citations for top 25 U.S. author institutions for 1993-'94 Biomedical papers Harvard Univ. Nat'l. Cancer Inst. Veteran's Administration Univ. Cal. San Francisco Stanford Univ. Univ. Washington MIT Scripps Clin. and Res. Fdn. Univ. Cal. Los Angeles Mass Gen. Hosp. Johns Hopkins Univ. Washington Univ. Univ. Cal. San Diego Univ. Pennsylvania Merk & Co. Inc. Yale Univ. Nat'l. Inst. Allergy Infectious Dis. Univ. Wisconsin Univ. Michigan Cornell Univ. Genentech Inc. Columbia Univ. Duke Univ. Univ. Colorado Rockefeller Univ.

2506 1279 1033 930 920 845 756 690 642 625 610 588 534 517 484 463 456

Chemistry papers

Physics papers

MIT 171 Univ. Texas. Austin 171 Harvard Univ. 160 Dupont Co. 142 Univ. Cal. Berkeley 139 AT&T Bell Labs. 130 IBM Corp. 122 Merk & Co. Inc. 102 Cornell Univ. 96 Texas A & M Univ. 95 Penn. State Univ. 89 Univ. Wisconsin 87 Purdue Univ. 83 Univ. Illinois 83 Univ. Cal. Los Angeles 79 Univ. Massachusetts 72 Va. Polytech. Inst. 71

AT&T Bell Labs. IBM Corp. Stanford Univ. Bellcore USN, Res. Lab. Lincoln Labs. MIT Univ. Illinois Univ. Cal. Santa Barbara Comell Univ. Univ. Cal. Berkeley Xerox Corp. Univ. Pennsylvania N. Carolina State Univ. Caltech Sandia Natl. Labs. Lawr. Livermore Nat'l. Labs. General Electric Cox Lawr. Berkeley Lab. Gen, Motors Corp. Jet Prop. Lab., Caltech Los Alamos Nat'l. Labs. Natl. Inst. Standards and Technology Hewlett Packard Co. Univ. Cal. San Diego Univ. Houston

448 447 432 432 428 413

Rice Univ. Univ. N. Carolina Dow Chemical Co. Stanford Univ. Univ. Florida Univ. Nebraska

70 70 67 67 66 66

409 405

Univ. Cal. San Diego Iowa State Univ.

65 63

Engineering and tech papers 854 566 300 174 167 150 133 120 110 106 100 95 93 90 87 83 80

AT&T Bell Labs. IBM Corp. Univ. Cal. Berkeley MIT Stanford Univ. General Electric Co. Texas Instruments Inc. USN Res. Lab. N. Carolina State Univ. Bellcore Xerox Corp. Univ. Illinois Penn. State Univ. Univ. Cal. Los Angeles Lincoln Labs. Sandia Natl. Labs. Carnegie Mellon Univ.

471 428 189 174 162 IIl ~6 88 84 78 69 64 60 59 57 54 53

74 74 65 64 63 62

Univ. Texas, Austin Hewlett Packard Co. Motorola Inc. Cornell Univ. Univ. Michigan Univ. Southern Cal.

53 5I 51 48 48 48

60 56 56

Univ. Washington Oak Ridge Natl. Lab,

44 43

Advanced Research Projects Agency Sloan Foundation Am. Cancer Soc. Environmental Prot. Agency Nat'l. Inst. Arthr. and Muscloskl., Skin Dis. Nat'l. Aero and Space Admin. Research Corporation United Nations/Sup. Nat'l. Org. Nat'l. Inst. ofNeuro. Disord. and Stroke Nat'l. Inst. ofEnvmntl. Hlth. Sci. Dept. of Agriculture

591 591 510 435 425 364 336 335 304 293 292

Arthritis Foundation Army Nat'l. Inst. of Dental Res. Muscular Dystrophy Assn.

Nat'l. Fdm. March of Dimes Nat'l. Inst, of Mental Health

25 22

46 46 48 26

85 83 78 56 52

281 267 197 197 172 171 129

Nat'l. Heart, Lung and Blood Inst. 102 Nat'l. Inst. Allergy and Infectious Dis. 94

Petroleum Research Fdtn. Nat'l. Inst. of Health (gen.) Air Force Army Welch Foundation Nat'l. Ctr. for Research Resources Am. Chemical Soc.

2131 2104 1651 1347 1198 1142 1092

1566 694 512 402 356

755 718

Nat'l. Science Foundation Nat'l. Inst. ofGen'l. Med. Sciences Dept. of Energy Navy Nat'l. Cancer Inst.

7970 4493 4334 4057 2294

Nat'l. Cancer Inst. Nat'l. Heart, Lung and Blood Inst. Nat'l. Inst. of Gen'l. Med. Sciences Nat'l. Inst. Allergy and Infectious Dis. Nat'l. Inst. Arthr. and Muscloskl., Skin Dis. Am. Cancer Soc. Nat'l. Sci. Foundation Nat'l. Inst. of Health (gen.) Am. Heart Assn. (St. and Loc.) Nat'l. Inst. of Neuro. Disord. and Stroke Nat'l. Ctr. for Research Resources Nat'l. Inst. of Child Health and Hum. Dvlpmt. Veteran's Administration Natl. Inst. Diab. and Digest an Kidney Dis. Dept. of Energy Nat'l. Eye Inst. Dept. of Agriculture Nat'l. Inst, of Aging United Nations/Sup, Nat'l. Org.

Chemistry papers

Biomedical papers

Table 4 Number of citations for top U,S, support agencies 1993-'94 citing 1981-'91

Sloan Foundation Nat'l. Inst, of Health (gen.)

United Nations/Sup. Nat'l. Org. Welch Foundation

22 14

23 23

Nat'l. Science Foundation 1072 Navy 708 Dept. of Energy 637 Air Force 549 Advanced Research 352 Projects Agency Army 299 Nat'l. Aero and Space Admin. 255 Dept. of Defense 192 Nat'l. Ctr. for Research Resources 34 Nat'l. Inst. of Gen'l. Med. Sciences32 Nat'l. Heart, Lung and Blood Inst. 30 Nat'l. Cancer Inst. 23

Physics papers

Nat'l. Science Foundation Navy Dept. of Energy Air Force Advanced Research Projects Agency Army Nat'l. Aero and Space Admin. Dept. of Defense Dept. of Agriculture

Engineering and tech. papers

152 120 90 11

606 408 301 282 224

t~

k,a

6O ~x

327

F. Narin et a l . / Research Policy 26 (1997) 317-330

to A T & T Bell Labs. papers come from A T & T itself, while only 21% of the cites to IBM papers come from IBM patents. Note also that in physics and engineering some of the major national laboratories also appear quite high on the list, as do the private companies and major universities; this list clearly shows a high degree of interdependence between the academic, industrial and governmental R & D communities which has been shown to be an important component of technological progress (Klevorick et al., 1995). A third important point is that many of the cited research papers are supported by U.S. governmental and other research support agencies. Table 4 lists the top support agencies. As would be expected, in biomedicine the various components of the National Institutes of Health are the first five support sources, with the American Cancer Society sixth, and the

National Science Foundation seventh, followed by other NIH Institutes and other government and private agencies. The National Science Foundation is the most widely acknowledged support agency for chemistry, physics and engineering and technology. The National Institute of General Medical Science, one of the NIH Institutes is second for chemistry, followed by a variety of government agencies including other parts of NIH, and some major private foundations. In physics and engineering the various Department of Defense agencies, the Department of Energy and NASA all show up much more strongly than they do in chemistry or in biomedicine, as would be expected. Overall, it is quite clear that the U.S. scientific papers being cited in patents are being supported by public agencies that are as mainstream and presti-

3000

2 5 0 0 - ~ D No US Inventors ~ A t Least 1 US Inventor

2000-

1500-

O 1000-

500

I

Clin Med & IBiomed Res

Biology

Chemistry

Physi~:s

Earth & Space

Fig. 5. Citation by inventor country: U.S. papers supported by NSF are preferentially cited by patents with U.S. inventors.

328

F. Narin et a l . / Research Policy 26 (1997) 317-330

gious as the institutions in which the papers are being authored, and the journals in which they are being published. A final way of looking at the data is to build more specifically upon the earlier observation of a strong national component to the patent-to-science linkage. In Fig. 5 we look at this in more detail, by tabulating whether the patents citing papers supported by the U.S. National Science Foundation are U.S. or foreign-invented. It is quite clear from the figure that papers acknowledging NSF support, in all the fields in which it funds science, are overwhelmingly cited by patents with at least one U.S. inventor. The ratio of citations from patents with at least one U.S. inventor to citations from patents with only foreign inventors is close to 5:1 in most areas, in a patent system in which half the patents are foreign-invented. This 5:1 ratio is the combined result of the two effects noted earlier: U.S.-invented patents are more highly science linked than foreign-invented U.S. patents, and the U.S. inventors have a propensity to cite to U.S. papers. The result is the highly preferential utilization of NSF supported research by U.S.-invented technology.

5. The science base of U.S. industry This last section will look at the patent-to-paper linkage data from the perspective of U.S. industry patents. The data show that even when the data are restricted to patents assigned to U.S. industry, the dependence on U.S. public science is very substantial. In this section the set of citing patents was restricted to those patents whose assignees were U.S. non-governmental, non-academic organizations, al-

most all of which are U.S. industry. Within these lists there may be a few small clinics or hospitals; however, almost all the patents are from U.S. industrial companies. The cited data consists of papers published in 1988, the peak cited year for science references in 1993-1994 patents. The reason for restricting the data to one cited year is that, while sector and institution had already been assigned to all the U.S. institutions in the SLID database, the same had not been done for papers with foreign-authored addresses, and therefore a time-consuming assignment of an organizational sector to the cited foreign papers had to be carried out specifically for this paper. Confining the data to one cited year kept the effort within reasonable bounds. The overall data are shown in Table 5, both for all industry patents, and for industry patents in three subsets: drugs and medicine (including biotechnology), chemicals (excluding drugs and medicine), and electrical components and accessories and communication equipment. The table shows the strong reliance of U.S. industry patents on public sector science; overall, only 20.4% of the cited papers are from U.S. industry, 6.3% from foreign industry, and the remaining 73.3% are from public science, in the U.S. and overseas. This implies that U.S. industry is far from self-sufficient in science. The great majority of the science base of U.S. industry comes from the public sector; public science appears to be crucial to the advance of U.S. industrial technology. As shown in Table 5, the area with the largest private science base is electrical components, for which approximately half of the cited papers are from industry. In the case of electronics, the large electronics companies, IBM, Hitachi, AT&T, and others, publish heavily, and this private sector com-

Table 5 Institutional origin of papers published in 1988 and cited in 1993-1994 U.S. industry patents All patents U.S. industry U.S. public Foreign industry Foreign public No. of matched cites

20.4% 43.9% 6.3% 29.4% 100.0% 5217

Drug and medicine patents 16.7% 50.3% 4.2% 28.8% 100.0% 1584

Chemical patents 18.3% 42.7% 6.0% 33.0% 100.0% 1784

Electrical component patents 37.5% 29.6% 13.3% 19.6% 100.0% 585

F. Narin et al. / Research Policy 26 (1997) 317-330

ponent of the scientific literature is quite heavily utilized in patents. Nevertheless, even in this case half of the science base is public. For drugs and medicine, chemicals, and almost all of the rest of the database more than 75% of the science base of U.S. industry, as expressed by these citations, has its origin in public science. The U.S. public science component is larger than the foreign, but foreign public science is still an important contributor. The sectoral dimension of this dependence is shown in more detail in Fig. 6, which shows the author sectors for the U.S. papers cited in the U.S. industry patents. The U.S. papers are 50% academic, written by scientists at U.S. universities, colleges, and medical schools, while 32% are written by industry scientists. The remaining papers are from government labs., particularly NIH in the biomedical area, from private non-profit organizations such as Scripps, and from federally funded research and development centers, largely the Department of Energy labs. A separate analysis of patents citing to non-U.S, papers is very similar; academic papers are the largest source at 60% followed by industry at 18%, with the remainder originating at non-profit institutions and government laboratories. Regardless of how the data are arranged, it is quite clear that public science plays an overwhelming role in the science base of U.S. industry.

329

123

103

±

69

0

20

40

60

80

1O0

Finally, in Fig. 7, we examine the data at one further level of disaggregation, considering the institutional origin of the scientific papers cited in the patents of IBM, a leader among companies in both publishing and patenting. For example, in 1993 IBM

so]

40

°

3o

1

I 20

104

Fed Funded R&D C~rs

Pdvate Non*Profit

Aoademlo

140

Fig. 7. IBM Science dependence: eighty percent of the papers cited in IBM's 1993-'94 patents are externally authored.

6O

~-

120

Number of References

Government

Industry

Unknown

Fig. 6. Author sector: half of the U.S. papers cited in U.S. industry patents are authored at academic institutions.

330

F. Narin et al. / Research Policy 26 (1997) 317-330

scientists published 860 papers, and IBM inventors obtained 1087 patents. IBM technical staff have been publishing and patenting at a high rate for many years. Fig. 7 shows that even with this environment of substantial publishing, I B M ' s patents are only 21% self-sufficient in science, with the remaining 79% of the science citations on the front page of I B M ' s patents in 1993-1994 to science originating outside of IBM. The largest component is U.S. public science papers, followed by IBM, by foreign companies, and by foreign universities and U.S. companies. Clearly, even for a giant company such as IBM, public science, both U.S. and foreign, plays a major role in providing a base for its patented technology.

6. Conclusion Therefore, we conclude that public science plays an essential role in supporting U.S. industry, across all the science-linked areas of industry, amongst companies large and small, and is a fundamental pillar of the advance of U.S. technology. The underlying hypothesis described in the beginning of this paper - - that public science is a driving force behind high technology - - is clearly supported by the data shown herein. Furthermore, the data shows that the science that is contributing to high technology is mainstream; it is quite basic, quite recent, published in highly influential journals, authored at major universities and laboratories, and supported by NSF, NIH, the Departments of Defense and Energy, and by other public and charitable institutions.

7. Unlinked Reference Narin and Olivastro, 1992

Acknowledgements The authors wish to acknowledge the encouragement and guidance of Ms. Jennifer S. Bond, Program

Director, Science and Engineering Indicators Unit, the National Science Foundation, throughout the course of the work, and for many years before. Work supported by NSF Grant No. SRS-9411378 and NSF Contract No. SRS-9301815.

References Bush, V., 1945 (reprinted 1960), Science: The Endless Frontier, National Science Foundation,Washington. Clinton, W.L., Gore Jr., A. 1993. Technology for America's Economic Growth, A New Direction to Build Economic Strength, Office of the President of the United States. Klevorick, A.K. et al., 1995. On the sources and significanceof inter-industrydifferences in technologicalopportunities. Res. Policy 24 (2), 185-205. Mansfield, E., 1991. Academicresearch and industrialinnovation. Res. Policy 20, 1-12. Martin, B., Salter, A., 1996. The Relationshipbetween Publicly Funded Basic Research and Economic Performance, H.M. Treasury, London. Narin, F., Hamilton, K.S., Olivastro, D., 1996. Citation Linkage Report: 1993-1994. Work supported by NSF Grant No. SRS941 1378 and NSF Contract No. SRS-9301 815. Narin, F., Olivastro, D., 1992. Status report - - linkage between technology and science. Res. Policy 21 (3), 237-249. Narin, F. et al., 1995. Linkagebetweenagency-supportedresearch and patented industrial technology. Res. Evaluation 5 (3), 183-187. Narin, F., 1991. Globalizationof research, scholarly information and patents - - ten year trends. In: Proceedings of the North American Serials Interest Group (NASIG)6th Annual Conference June 14-17, 1991, The Serials Librarian 21, pp. 2-3. Narin, F., OlivastroD., 1988. Patent citation analysis:new validation studies and linkage statistics. In: van Raan, A.F.J., Nederhoff, A.J., Moed, H.F. (Eds.), Science Indicators:Their Use in Science Policy and Their Role in Science Studies, DSWO Press, The Netherlands, pp. 14-16. Normile, D., 1997. Basic science spending to jump in 1997. Sci. 275, 21. Rosenberg, N., Birdzell, L.E. Jr., 1990. Science, technology and the Western miracle. ScientificAmerican263 (5), 42-54. Tumey, J., 1991. What drives the engines of innovation?.New Scientist 40. Patent and Trademark Office, U.S. Department of Commerce, Manual of Patent ExaminingProcedures, Section 904.02, 6th edition, 1995. Weinberg, A.M., 1963. Criteria for scientific choice, Minerva 1, 2, and 1964, criteria for scientificchoice II: the two cultures. Minerva 3, 3.