Science magazine

R E S E A R C H A N D D E V E LO P M E NT Where Is the New Science
most new science in developed rather thandeveloping economies for reasons that may not in Corporate R&D?
always characterize the U.S. situation.
Jerry Thursby and Marie Thursby*
The idea that the United States domi- or currently planned R&D facility both views of corporate R&D. For example, outside and inside the home country.
Respondents identified 145 facilities in devel- reported that 4% of U.S. industry expenditure U.S. share of patents and the growth of corpo- oped economies (primarily the United States on R&D was for basic research, 19% was for rate spending on research and development applied research, and 77% was for develop- (R&D) in emerging countries like China and economies (primarily China and India). They ment (1). By contrast, in our taxonomy, 38.8% India (1–3). Because scientific discovery is were asked to characterize the technological of R&D at identified sites involves new sci- critical to economic growth, these trends have and market focus of R&D at the site. The tech- ence, while 61.2% is familiar science.
sparked concerns as to what is driving compa- nological focus was defined as either (i) a The focus here is on the percent of effort nies to conduct R&D in these countries and novel application of science as an output of the devoted to new science, regardless of whether the implications for future competitiveness, R&D (it could be patentable or not) or (ii) it is for new or familiar markets. The his- particularly given problems with the U.S.
of intellectual property (IP) in emerging economies (4–9). Similar concerns pervade European innovation policy initiatives (10).
The popular press has fueled these concerns with reports of R&D moving to emerging countries in search of low costs (11).
intensive companies headquartered primarily revealed that respondents expect their R&D to grow in emerging economies and to decline in (12, 13). Lower R&D cost in emerging eco- new markets and (iv) as familiar markets.
tograms in the chart (p. 1548) give responses factors, collaboration with university scien- for the percent of effort devoted to new sci- tists, and quality of R&D personnel were all at R&D: new science to create new markets, new ence in developed versus emerging economy least as important as cost (12, 14).
science to improve familiar markets, familiar sites. The percent of effort devoted to new sci- Here we focus on the type of R&D con- science to create new markets, and familiar ducted in different countries and argue that science to improve familiar markets.
evenly distributed than it is for sites in emerg- appropriate policies in the face of globaliza- To clarify, when Pfizer developed Viagra, ing economies. In the latter, almost 71% of the tion should focus not only on the factors af- it was a new molecular structure with applica- sites conduct 25% or less new science. On fecting location but also on the type of R&D tion in a market not served by Pfizer. It was average, 49.6% of R&D effort in developed conducted. We categorize R&D according to a new science for a new market. Cialis, based economy sites is for new science; in emerging taxonomy suggested by R&D executives as on the same molecular structure, was later economy sites, it is only 22%. The contrast is one they use in tracking internal R&D. This more striking when responses are weighted by allows us to focus on the extent to which com- for Lilly. It was familiar science for a new the number of technical employees at each panies use cutting-edge science and show that market. Once-a-week versions developed by facility: The weighted averages for new sci- the type of industrial R&D differs substan- either company would be familiar science for ence are 56% in developed economy sites and tially in developed versus emerging country sites. An econometric model is used to relate the type of R&D at various sites to country effort at the site devoted to each of the four science at a site, a logistic regression ap- characteristics. In the survey, respondents categories (see figure, above). The R&D exec- proach for grouped data was used to relate the were asked to identify a recently established utives we interviewed claimed this classifica- ratio of new to familiar science in the identi- tion is more relevant to their R&D than fied facilities to respondent views of a variety the more “linear and sequential” taxonomy of other country-specific characteristics (16).
J. Thursby is at Emory University, Atlanta, GA 30322 and M.
Thursby is at Georgia Institute of Technology, Atlanta, GA of basic or curiosity-driven research, applied The model controls for industry, the firm’s research designed for specific end use, or de- total worldwide technical employment, and velopment to improve products or processes whether the country of the facility is developed *Author for correspondence. E-mail: marie.thursby@mgt.
gatech.edu (15). The two taxonomies provide different or emerging. Data for the other country char- corporate sentiment. U.S. universities have terms, enough so as to instigate policy discus- university research agreements (18). Recent research on university industry collaboration in the European economies that have adopted U.S. policies regarding university research shows similar concerns (19). This dynamic will only be accentuated as the quality of uni- versities in emerging economies improves.
the country in which thefacility is located. They 1. National Science Foundation, Science and Engineering Indicators (NSF, Washington, DC, 2006).
2. J. Landefield, R. Mataloni, Working Paper 2004-06 (U.S.
New science in developed and emerging economy sites.
Bureau of Economic Analysis, Department of Commerce, Washington, DC, 2004);www.bea.gov/bea/working_papers.htm.
facility. Responses on agreement and impor- two IP factors were similar to those for quality 3. Economist Intelligence Unit, “Scattering the seeds of tance were combined to create a measure of of personnel, in that the IP factors were statis- invention: The globalization of research and develop- the extent to which a factor drove the location tically important in location decisions, but ment” (White paper, Economist, London, 2004).
decision. We then tested the hypotheses that were not significantly related to the ratio of 4. R. Nelson, The Sources of Economic Growth (Harvard some of these factors are also central to the new to familiar science. Thus IP protection 5. Committee on Prospering in the Global Economy, Rising type of R&D conducted (see table below). Above the Gathering Storm: Energizing and Employing Because one would expect the availability new and familiar science. In terms of the America for a Brighter Economic Future (NationalAcademies Press, Washington, DC, 2006).
of high-quality personnel to be important for Viagra/Cialis example, it would not be sur- 6. R. Freeman, NBER Working Paper 11457 (National any kind of scientific research, it is not clear prising that Pfizer and Lilly consider IP pro- Bureau of Economic Research, Cambridge, MA, July how the ratio of new to familiar science would tection equally important for both products, 7. A. Jaffe, J. Lerner, Innovation and Its Discontents: Our vary (if at all) with the quality of personnel.
even though the former represents new sci- Broken Patent System Is Endangering Innovation and Our regression analysis showed that, although Progress (Princeton Univ. , Princeton, NJ, 2004).
quality of R&D personnel affects location The most striking result is that the factors 8. N. Zamiska, WSJ (3 June 2006), p. A3.
decisions, it is not significantly related to the related to universities (presence of university 9. L. Branstetter, Q. J. Econ. 121, 321 (2006).
10. Aho Group report, Creating an Innovative Europe type of science. Cost was significantly related faculty with special expertise and ease of (European Commission, Brussels, 2006); http://ec.
to the type of science with an increase in cost europa.eu/invest-in-research/action/ research06_en.htm.
decreasing the ratio of new to familiar science.
strongest impact on the type of science con- 11. Taken from 38 of 61 articles on R&D moving off-shore in the WSJ and New York Times 2002–2006.
Growth potential and supporting sales were ducted. Each is statistically significant in the 12. J. Thursby, M. Thursby, Here or There? A Survey on the expected to be more important for familiar Factors in Multinational R&D Location (National than new science, because R&D in those leads to a substantial increase in new relative cases is likely to be product localization. An 13. The companies included are R&D-intensive firms large enough feasibly to have multiple R&D facilities. Because increase in market potential or a facility that of confidentiality agreements that made this study possi- supports sales is associated with a decreased summarized in the table, left (17). With regard ble, firm-specific data are not available.
ratio of new to familiar science. Results for the to government and university policy, these 14. For R&D facilities in developed economies, the same factors are important, except cost is not important and IP protection is important. Weak IP protection is a deterrent 15. J. Marburger, Science 308, 1087 (2005).
16. Details are in the supporting online material.
17. The most important factor has rank 1. The rank is based on the absolute size of the elasticity showing the impact of the factor on the type of science. The factors ranked as “not important” are not statistically significantly related 18. www7.nationalacademies.org/guirr/Meetings.html 19. F. Valentin, R. Jensen, Effects on Academia-Industry cantly related to university factors.
Collaboration of Extending University Property Rightswww.cbs.dk/forskning_viden/fakulteter_institutter_centre/ institutter/oekonomi/biotech_business/menu/ 20. Supported by the E. M. Kauffman Foundation and Government University Industry Research Roundtable ofthe National Academies.
*Costs of R&D are exclusive of tax breaks and government assistance; growth refers to market growth potential in that country, www.sciencemag.org/cgi/content/full/314/5805/1547/DC1 Ease of ownership is the ease of negotiation for ownership of IP from research relationships, and IP protection refers to its strength.

Source: http://scheller.gatech.edu/directory/faculty/thursby_m/pubs/where%20is%20the%20new%20science.pdf