3
Project Descriptions: Incentives, Governance, and Innovation
A major theme in the SciSIP-funded research presented at the conference was the nexus of incentives, governance, and innovation. Researchers explored the effects of national and institutional policies, culture, and other influences on rates of innovation in both research and manufacturing.
MANUFACTURING LOCATION DECISIONS AND INNOVATION
Erica Fuchs, Carnegie Mellon University
Erica Fuchs began by noting the significance of SciSIP having funded her work, observing that “it’s pretty crazy for an engineer to look at policy problems.” The application of engineering models, she added, can help illuminate the relationship between manufacturing and innovation, and she described the way she investigated this relationship. Manufacturing located in the United States accounts for only 21 percent of manufacturing value added in the world as of 2009.20 That figure has been declining, while the percentages generated in southern and eastern Asian countries are increasing. Increasing numbers of manufacturing firms are “born global,” rather than starting out near the source of the innovations on which they rely. In this changing climate, U.S. firms whose manufacturing takes place offshore vary in size and age, in the technologies they use, and in their reasons for choosing an offshore site.
According to Fuchs, data show that for 90 percent of firms the primary reason for choosing offshore manufacturing is to reduce cost, while only 15 percent cite the goal of reaching new markets. There has been relatively little research on the connections between manufacturing locations and product decisions or innovation. Thus, she set out to learn whether manufacturing in an offshore location affects the relative competitiveness of technologies and thereby the technology trajectory of an individual firm or an industry—the types of technologies the firm or industry uses and its willingness to adopt new approaches. She investigated innovation in the automotive, telecommunications, and computing industries to pursue this question.
The existing literature shows vast differences in manufacturing processes around the world, Fuchs observed. There are differences in, for example, yields, downtimes, the quality of materials, and in how production line workers are organized, but these
________________
20In the context of economics, a sector’s value added is its contribution to the economy, which is calculated as its outputs minus the inputs required to produce them; see http://data.worldbank.org/indicator/NV.IND.MANF.ZS [January 2014].
differences have not been linked to decision making. Because these are precisely the sorts of differences that one would expect to influence manufacturing design decisions, she explored emerging manufacturing technologies in two areas to trace possible connections. Fiber-reinforced polymer composite unibody construction, she explained, offers the automotive industry a potentially perfect substitute for existing technologies, with potential benefits that include lighter vehicle weight, energy savings, and lower fuel costs. Similarly, in the telecommunications and computing industry, the capacity for monolithic integration of multiple functions into a single computer chip, originally developed by small telecom firms, offers the possibility of significantly increasing processing speed and other benefits.
Fuchs examined data from manufacturing shops in the United States and in developing countries in East Asia, and found that in both industries, the site of manufacturing affected the calculation of the relative economic benefits of the prevailing design versus the innovative design. Firms that located their manufacturing in the United States found the emerging design to be more cost-competitive, while those located in East Asian countries found the prevailing design to be more cost-competitive. Through interviews, she found that decision makers at these firms accordingly behave “like rational economic actors,” moving operations overseas and using the old technology.
This makes sense in the short term, she noted, but if market demand for lighter weight vehicles or greater processing speed increases, these firms may not be well prepared to meet it.
The new computer chip technology, which could be applied in, for example, biosensors that could be placed inside the body or the development of smaller photonics, was affected by the bursting of the “telecom bubble” in 2000, Fuchs explained. By 2003 this technology was producing only 20 percent of the revenue that had been forecast. As a result, the pressure to reduce costs was great even in the top ten firms, which collectively account for 65 percent of the total revenue in this market. As she collected data and interviewed employees, she found the engineers were saying, “We can reduce costs—just give us a little time [to develop monolithic integration].” At the same time, financial managers were focused on reducing labor and packaging costs, and as a result, seven of the eight firms that had been located in the United States moved overseas.
Fuchs’s analysis of the outcomes for the firms she studied showed that “even if you are the best in the industry at producing the old technology” in the United States, the new technology yields far better results, even for firms that are just average at using that new technology. Yet using the old technology in an offshore facility is still less expensive. Moreover, it was not possible for firms to use the new leading-edge technology in the offshore locations, which might seem to be the optimal approach. As Fuchs explained, “The engineers were constantly down on the production line trying to figure out why products were not getting out the door.” Furthermore, the firms generally could not afford to have more than one facility, so they had to choose between using the old technology in Asia or the new technology in the United States. When firms moved fabrication operations overseas, she added, they generally stopped innovating even in facilities that remained in the United States—this was not the case when only assembly operations were moved overseas.
Automotive firms do not share the constraint on operating in multiple locations, Fuchs explained in answer to participants’ questions. In this case, having plants around
the world can help firms make use of opportunities in different places to increase product diversity and innovation. Thus, for her, the question of whether governments should provide support to manufacturers to encourage them to keep their plants in the United States is complicated and the answer may vary across types of manufacturing.21
THE EFFECTS OF FUNDING POLICIES ON HUMAN STEM CELL SCIENCE
Jason Owen-Smith, University of Michigan
Jason Owen-Smith looked at the effects of funding policies on innovation in the field of human embryonic stem cell research, including pluripotent stem cell technologies.22 This is an ideal example for exploring the connection between policy and research decisions, he explained, because stem cell research is cutting edge and offers potentially great benefits, but the use of stem cells has been very controversial and highly politicized. Stem cell research has led to path-breaking discoveries, Owen-Smith noted, including the possibility of growing differentiated human tissue, and even new organs. However, the cells are obtained by culturing cells harvested from a human embryo, which results in the death of the embryo. There are differing conceptions of life and when it begins, so some people view this use of embryonic tissue as immoral.
Owen-Smith noted how federal policy on using stem cells from human embryos has swung back and forth over the past couple of decades. A 1995 amendment to an appropriations bill prohibited the use of federal funds for research resulting in the destruction of an embryo, even those left over from fertility treatments. In 1998 privately funded research discovered embryonic stem cells, and in 1999 the Clinton administration approved federally funded research with existing stem cell lines. President George W. Bush implemented a policy in August 2001 that limited the use of federal funds just to existing stem cell lines and not also to new lines that might be developed; his policy also supported research with adult and animal stem cells. Congressional attempts to expand research with embryonic stem cell lines met with presidential vetoes. Meanwhile, some states passed their own laws supporting embryonic stem cell research. President Obama issued an executive order in 2009 that expanded the scope of federal research with new lines of embryonic stem cells.
Owen-Smith and his colleagues explored the effects of these changes in policy on research in the field. They examined recent research using a census of papers published between 1998 and 2010 reporting research that used either human embryonic stem (hES) cells or induced pluripotent stem (iPS) cells, which researchers make by introducing embryonic genes into somatic (adult) cells. They also conducted random and targeted interviews with scientists who have done this type of research and examined research posters presented at conferences in 2010 that addressed this type of research.
This work yielded three primary findings. First, Owen-Smith explained, scientists do pay attention to regulation and the ethical implications of their work. They are eager to have certainty about what is acceptable, and they want easy access to diverse cell lines
________________
21For more details on this research, see Fuchs et al. (2011); Fuchs and Kirchain (2010); Yang, Nugent, and Fuchs (2013); and Fuchs (2013).
22Stem cells are undifferentiated or “blank” cells found in the human body with the potential to develop into different cell types that carry out different functions.
that are developed in a morally defensible way. Second, iPS cells are not a “magic bullet” for ethical research because the vast majority of researchers who use iPS cells also use hES cells or have been trained by hES researchers. Technical and ethical questions remain. Last, the way in which a policy is implemented matters more than the goals it was intended to achieve. Uncertainty and confusion about what will be acceptable make investigators more conservative in their choice of materials. As a result, few new hES cell lines are in wide use, and there is an increased concentration in the use of particular types of cells. These circumstances also have meant, according to Owen-Smith, that “new people aren’t entering the field very much.”
Owen-Smith observed that some of the most prestigious universities have been able to pursue alternate sources of funding through collaborations with colleagues in other countries, and a few states have attempted to fill in some of the gaps in federal funding. These initiatives have not been enough to bring stability to the situation, however. He questioned whether it would be desirable for the system to depend so heavily on state and private support.
Owen-Smith closed with a summary of his suggestions for the field:
• Clear, stable, and uniform rules regarding both hES and iPS cell science are necessary for this field to progress.
• Legislation is needed to ensure continuity in funding for hES cell research.
• Support for hES cell research is necessary for the continued development of promising iPS cell science.23
ECONOMIC SPILLOVERS FROM SCIENCE
Bruce Weinberg, Ohio State University Subhra Saha, Cleveland State University
Governments are important supporters of scientific research, observed Bruce Weinberg, and “government activity is increasingly motivated by the economic benefits it is supposed to generate.” However, the economic benefits of science are frequently disputed even among scientists, and there are no standard methods for evaluating them. He, Subhra Saha, and Lura Crispin sought to establish a set of methods that could be used to evaluate economic spillovers, and also provide estimates of their value, though he noted that the work is at a preliminary stage.
Science may have a direct economic impact, Weinberg explained, if it pushes the frontiers of knowledge in a way that can be applied in the development of new products, technologies, or treatments. These benefits are important and have received attention in the literature, so Saha and Weinberg focused on indirect benefits beyond those that directly affect workers and firms. These are benefits that spread to local economies. For example, scientific activity may attract a better educated workforce, foster the development of a “hub” for innovation, or lead to ideas that solve industrial problems.
________________
23For more details on this research, see: Owen-Smith and McCormick (2006); McCormick, Owen-Smith, and Scott (2009); Scott, McCormick, and Owen-Smith (2009); Christopher, McCormick, and Owen-Smith (2010); Scott et al. (2011); and Owen-Smith, Scott, and McCormick (2012).
Job creation is one potential benefit that gets a lot of attention, Weinberg noted, but he suggested that tracing the links between science and jobs is very difficult. Instead, he and Saha focused on ways in which science may raise the productivity of firms in a particular area. The production of ideas and knowledge is likely to cause firms to expand and hire and to build larger plants and factories. Such activity should, in turn, increase the demand for both labor and land and yield higher wages and real estate prices. However, Weinberg added, such activity might attract a wide range of workers in terms of skill levels, and average wages may not rise even though economic activity is stimulated.
Weinberg also cautioned that the numerous differences among cities make comparisons complicated. The cities in which scientific activity and innovation are prevalent may be different from other cities in certain ways—having higher than average educational attainment levels, for example. Other factors besides the presence of science researchers may make a city an attractive place to live, so a credible estimate of the impact of science must control for those effects. Thus, Weinberg and Saha’s project began with an analysis of longitudinal data on metropolitan areas to establish basic differences among cities, and census data to establish and control for worker characteristics. The researchers used data on science funding collected by the National Science Foundation, information on patent applications from the U.S. Patent and Trademark Office, and other data to assess scientific activity. They focused on cities that are home to institutions that receive large amounts of funding for research.
The researchers used an economic cost function productivity model to estimate the benefits of scientific activity. Their results suggest that the presence of scientific activity is associated with higher wages in cities; the data are not sufficiently clear to support conclusions about the effect on real estate prices. The researchers intend to explore the specific influence of the patenting of scientific discoveries, which may have the largest economic effects, as well as the preliminary finding that local spillovers from science are increasing over time.
Weinberg noted that the magnitude of the benefits of scientific activity is large, though he acknowledged that the estimates are “a little bit fuzzy.” He suggested that increasing science spending by, for example, $1 billion over a year or two across the nation could increase wages and real estate prices by slightly more than one-quarter of 1 percent. “That does not sound very large,” he noted, “but if you use conventional estimates of the share of labor and real estate in the economy, you wind up with an increase in productivity of just under one-fifth of a percent.” The return would be greater the longer the science spending were sustained, he noted, but “certainly this would seem like a nice return.”
THE IMPACT OF OPEN-ACCESS INSTITUTIONS AND POLICY ON LIFE SCIENCES RESEARCH
Scott Stern, Massachusetts Institute of Technology
Scott Stern described a systematic research program aimed at establishing not just correlations between policies or institutions and positive outcomes for science, but actual causal linkages between open access and particular sorts of scientific progress. He pointed to Isaac Newton’s observation that each generation stands on the shoulders of
giants, and explained that he and his colleagues hoped to shed light on how knowledge is actually transferred. “Simply producing knowledge does not at all guarantee its accessibility,” he noted. “Knowledge transfer by its very nature is very costly,” he added, and “knowledge … can be maintained as a secret … in a way that makes it difficult to facilitate follow-on research across research generations.”
Specifically, Stern and his colleagues wondered how open-access institutions (those that require their researchers to make their published results freely available to others) and policies that support a “scientific commons” approach to science contribute to the accumulation of knowledge and research productivity. They hoped to learn what conditions give researchers and research funders incentives to contribute to open access, and what roles institutions and public policy play in fostering those conditions.
Stern first described his and his colleagues’ findings with respect to biological resource centers, though the project has covered other areas as well, including mouse genetics and the Human Genome Project. Biological resource centers are places that store large stocks of specialized cell lines that are made available to scientists. These include stem cells as well as many other types of cell lines. Biological resource centers have many advantages: They provide independent access to the material they hold so researchers need not compete for it, and they can preserve material for a long time—potentially longer than any individual researcher would be able to. These centers also authenticate the material so that those who use it can be confident in its source, an issue that has presented problems in the past.
To establish the outcomes of making particular biological material accessible, Stern and his colleagues traced the pattern of publications on related topics before and after the material was made available, and also compared those rates to control rates for other areas where there had not been a comparable sharing of research material. Whenever material is deposited in a resource center for whatever reason, it is linked to a particular citation, which will then be included in any future work with that material. Particularly beneficial for the before-and-after comparison were cases where the material was deposited in a resource center many years after the discovery had first been made and published.
Stern and his colleagues found “a relatively big effect” of opening access to these cell lines, he explained: a doubling in the rate of citation of the articles linked to open-access cell lines, compared with the controls. Moreover, the citation rates increase over time. “The biggest effect is associated with the number of institutions that cite an article, the range of journals that cite the article, and the geographic reach of the citations,” Stern noted. They found similar results when they examined mice bred for scientific research, some of which have been patented. When particular mice are made available through open-access programs, he explained, there is a large increase not only in the number of authors and institutions involved in related research, but also in the diversity of research topics being pursued using the mice.
“Scientists by and large want to do the right thing,” Stern concluded, but it is important to understand what motivates them as they make decisions, and what characteristics of the research environment “make it easy for them to do the right thing.” Researchers who might pursue more speculative kinds of work that could build on what has already been done, he explained, are encouraged and supported when there are formal
institutions and policies that “provide independent, low-cost access to tools, databases, materials, and even pieces of intellectual property.”
COMMUNICATION, COLLABORATION, AND COMPETITION
Jerry Thursby, Georgia Institute of Technology
The ways in which scientists decide to share their work with colleagues and the factors that influence these decisions was the focus of work by Jerry Thursby and his colleagues. In a prior study, they had explored what they call “specific sharing,” in which a scientist “makes a specific request to another scientist for materials, for an algorithm, or data.” They found that the tradeoffs and incentives in these situations are very different from those that influence decisions about public presentations of intermediate results, which they call “general sharing.” They decided to follow up with a closer look at general sharing.
The researchers began with three models of how sharing might be viewed, which are based on game theory (the study of strategic decision making in conflict situations). The competition/collaboration model describes the calculation that disclosure could lead either to competition from other researchers that is undesirable (because a competitor might beat the original researcher to a solution or breakthrough), or to productive collaboration. The mathematician model describes a situation in which competition is desirable because no one can receive credit for a discovery until the work is completed. For example, a mathematician might have a theorem that has largely been proven, but be unable to complete the last portion of the proof. The mathematician might share it in order to create interest, in hopes that another mathematician will solve the last portion. The research leader model describes a situation in which the reputation of a researcher is enhanced if his or her work inspires others to work in the same area, thus demonstrating the powerful influence of the original work.
In considering the competition/collaboration model, Thursby went on, it is important to distinguish among types of researchers who may behave differently. The focal researcher—the decision maker—has made a discovery that is of partial value in solving a problem. That person will have to decide whether to share results with colleagues in the field, who are either close colleagues who can be trusted not to compete, or general colleagues who might decide to compete. The main potential benefits to sharing are gaining credit in the field for the initial discovery and possibly finding a collaborator. The risk is that a rival with better resources or skills will engage in the research and outpace the focal researcher.
Thursby and his colleagues developed simulation models to explore the results for each of the three models for how sharing might be weighed. They assigned values to each of the factors that would influence the outcome of sharing or not sharing with the two types of colleagues. The factors include the relative ability of the researchers involved, time saved if two able researchers collaborate, the opportunity cost of not working on some other research, and the like. The results varied across the three sharing models. In the competition/collaboration model, researchers are much more likely to share incremental work than breakthrough discoveries, while the opposite is true in the
mathematician model. In the research leader model, the outcomes depend on the relative capability of the researchers involved.
The next step will be to survey approximately 60,000 U.S. researchers and 20,000 German researchers, and to interview a subset of this group, to collect data on their decisions about what they disclose, the point in their research at which they disclose, and the reasons for their decisions.24
GENERIC DRUGS AND INCENTIVES FOR RESEARCH AND DEVELOPMENT
Matthew Higgins, Georgia Institute of Technology
There has been an exponential increase in spending on research and development (R&D) in the pharmaceutical field during the past few decades, noted Matthew Higgins, but the approval of new products has remained constant or lagged behind these expenditures. Anecdotal evidence, he added, suggests that pharmaceutical firms have moved away from developing certain kinds of drugs partly in response to competition from generics. While it is likely that a combination of factors are affecting this industry, Higgins and his colleagues explored the possibility that regulation policies that speeded approvals for generic drugs may partially account for the discrepancy.
Current regulations for new drugs are laid out in the Drug Price Competition and Patent Term Restoration Act of 1984, P.L. 98-417, also known as the Hatch-Waxman Act. The law provides patent and other protections for new drugs, and when these protections expire, other manufacturers can file a request to sell the product as a generic drug. In some situations, manufacturers can use legal challenges to get earlier access to new drugs. The law covers only chemical-based or small-molecule drugs, Higgins noted, not biological drugs. Current law governing biologics allows them 12 years of market exclusivity and makes no provision for allowing manufacturers of generics into the market.
Higgins noted that the current regulations for chemical-based drugs have created a situation in which drug developers have a very short time to reap a return on their investment, because once generics enter the market, their revenue declines very rapidly. This is not the case for biological drugs because there are no “biosimilars,” drugs that can mimic exactly the properties of the original ones. The competition for these drugs is newly developed biologics, rather than generic versions of the same ones.
Higgins and his colleagues investigated three “seemingly simple questions.” The first was whether allowing generic drugs to enter markets earlier than they had been previously was welfare-enhancing (beneficial overall in terms of the ratio between gains and losses). The second question was whether early market entry for generic drugs has reduced incentives for innovation in pharmaceutical research in particular markets. The answers to these two questions provide the backdrop for exploration of the somewhat broader question of whether the result of current drug approval policies is that the United States is sacrificing future innovation for the sake of access to inexpensive drugs today.
The findings are complex. For example, as Higgins noted, the impact of a new generic drug depends on the nature of the drug and the condition for which it might be used. With some conditions, such as epilepsy, insurance companies will not require
________________
24For more details on this research, see Haeussler et al. (2013).
patients to switch to an available generic, and doctors may prefer to stay with a drug that is effective. In other cases, the availability of generics reduces prices and provides patients and doctors with more choices. The analysis Higgins and his colleagues conducted took numerous factors into account and showed that, overall, the gains to consumers have been greater than the losses to producers in the current market place. However, they also found that the entry of a generic drug leads to “about an 8 percent decrease in innovation,” with the effect apparently being greater on later stage innovation than on early-stage innovation. Early-stage innovation is also boosted when there are “technological opportunities,” Higgins added, which are promising possibilities identified through basic science research.
There is some indication that current policies have pushed companies to focus more resources on biologic drugs than chemical ones, Higgins observed, and he closed with the observation that both push and pull mechanisms influence the drug market. He suggested that it would be beneficial to have more preventive and curative drugs and that possibly shifts in the regulatory structure would improve the incentives for drug companies to develop the drugs that are most needed. Funding for basic science research might also be used as a mechanism to “push” the industry to develop drugs that are particularly needed.25
PATENT RIGHTS IN THE SOVIET UNION
Lisa Cook, Michigan State University
Lisa Cook set out to learn whether patent rights are necessary to spur robust innovation. It is difficult to study this question in a market economy, she noted, but socialist countries provide an “interesting laboratory” because they traditionally have not extended patent rights to their citizens.26 She noted that there was a slowdown in technological advancement in the Soviet Union during the 1970s that caused policy makers significant concern. In response, she explained, planners developed incentives to encourage individual inventors. Until now it has not been possible to test whether these incentives worked, but newly available archival data make it possible to assess the contributions of Soviet inventors during this period.
In general, Cook explained, the Soviet Union provided recognition to inventors in the form of inventor’s certificates, but the certificates did not designate anyone as the original inventor of anything or give the recipient any control over the invention. Technically, inventors could apply for patents, but most were granted to foreigners—just 0.01 percent of patents granted between 1973 and 1991 were issued to Soviet residents.
The market-like incentives that were introduced, Cook noted, included increased compensation and tax advantages for individual inventors, as well as such non-pecuniary benefits as housing privileges, promotions, and prizes. Authorities also instituted a requirement that inventors at research institutes produce innovations of national or international significance, although the significance of this was difficult to measure. An entity called Licensintorg was established in 1962 to facilitate the licensing and
________________
25For more details on this research, see Branstetter, Chatterjee, and Higgins (2013a and 2013b).
26Cook noted that her work builds on previous work by Moser (2004, 2005); Brunt, Lerner, and Nicholas (2012); and Moser and Rhode (2011).
marketing of Soviet inventions abroad, Cook noted. By 1962, 62 licensing agreements had been negotiated with the United States and other countries.
Further changes included a 1979 reform of research and development designed to encourage individual investors, and patent reform carried out as part of perestroika. This reform provided limited rights for inventors and promoted research and development in civilian areas, rather than defense. Cook noted that these measures reflected conflicting objectives in the Soviet Union: despite reasons for maintaining secrecy during the Cold War years, the leadership was eager to demonstrate the country’s technological prowess.
To assess the outcomes of these efforts, Cook used data from several sources. Files of the United States Patent and Trademark Office and the National Bureau of Economic Research allowed her to trace patents granted to Soviet inventors between 1959 and 1991 (at least one of the inventors was a resident of the Soviet Union or Russia). The patent offices of Russia and Germany were a source for inventor’s certificates and patents issued by the Soviet Union.
Cook used a statistical model to examine the effects of the policies on the share of patents assigned to individuals. She noted that if the market incentives were effective, one would expect to see greater patent activity in response to a rise in Gross National Product, and that was in fact evident. She highlighted several additional findings. “Soviet inventors were inventing like crazy outside the Soviet Union,” she observed. They obtained many patents from other countries, and their output was comparable to that of medium-sized industrial countries such as Austria, Australia, and Belgium. In general, she concluded, the incentives did have the effect of stimulating inventions. This finding is important, she added in response to a question, because developing countries and others that lack robust intellectual property rights are interested in technological and economic growth and in ways to provide incentives for innovation. She noted that there are follow-up questions to be explored, such as what other factors influenced individual inventors, what other factors affected technological spillovers among countries, and how inventors responded to the fall of the Soviet Union. In another paper, Cook and Ivanya (2012) explored the extent and quality of technological spillovers from these Soviet inventors. In particular, they examined the effect of the boycott of the Moscow Summer Olympics on these Soviet inventors, which led to a decline in their patent activity in the U.S., and found that there was a significant subsequent effect on Soviet inventors’ patent activity in East Germany.
CULTURE AND NATIONAL INNOVATION RATES
Mark Zachary Taylor, Georgia Institute of Technology
Policies and institutions have a significant influence on a nation’s relative success in science and technology, noted Mark Taylor, but “they only explain anywhere from half to 75 percent of the story.” There are well-designed policies and institutions that have limited effects in some countries, as well as countries that have success in science and technology despite poor policies. Taylor attempted to understand the unexplained differences by trying to determine whether national culture affects rates of innovation. He noted that anthropologists and sociologists do not agree on the definition of culture, and that culture can be viewed as a spectrum that encompasses a very wide range of values,
even within a country. Nevertheless, a notion of culture as “a country’s ‘central tendencies’ in terms of values, beliefs, and preferences” has value, he explained. Despite significant variation within cultures, nations do tend to have what he described as “national cultural values,” such as individualism, collectivism, or social or intellectual autonomy, to different degrees.
Discussions of culture may reflect biases, Taylor explained, but social scientists have attempted to quantify the cultural values that tend to be associated with national cultures. Researchers27 have used survey instruments in numerous countries to establish cultural values by asking large numbers of people about their attitudes and preferences regarding personal traits and values. Looking across this body of work, he added, one can “triangulate”: “If there is some objective thing out there that multiple scientists are trying to measure independently using different methodologies, then the noise should cancel out and the signal should come through.” There are also multiple, independent measures of innovation rates, so, collectively, these data support conclusions about the relationships between the two, in Taylor’s view.
Taylor described several findings. First, individualism as a cultural value seems to correlate strongly with national innovation rates, regardless of how either is measured. This finding held when other possible factors, such as level of development, trade openness, military and education spending, and research and development spending, are controlled. An interesting difference between two different types of collectivism was also evident, Taylor added. In-group collectivism—strong identification with friends, family. or tribe—has a negative correlation with national innovation rates, whereas institutional collectivism—such as patriotism or loyalty to institutions—has a positive correlation with innovation rates. Taylor cautioned, however, that qualitative research is needed to illuminate the causal mechanisms and confirm or disconfirm the picture evident in the aggregate statistics. Participants followed up on this point, noting that historical factors such as immigration might also play key roles but are not captured in the data Taylor described.
The findings do suggest, Taylor noted, that it is important to be skeptical of stereotypes, such as the idea that cultures that value collectivism must therefore be unfavorable for innovation. The findings also suggest that particular policies may work better in one culture than another because of “cultural fit.” He suggested some tentative points related to cultural fit. For example, free markets, democratic systems, and political decentralization may all be more important to innovation in individualistic societies than in collectivist ones. The primary conclusion he reached was that there is strong evidence—stronger than the case studies on which scholars have tended to rely—that culture influences innovation in significant ways. This is important, he noted in closing, because policy makers and businesses would do well to consider culture when designing policies and institutions.28
________________
27Taylor mentioned Geert Hofstede, Shalom Schwartz, Robert House, Fons Trompenaars, Charles Hampden-Turner, and Ronald Inglehart as researchers who have done this work.
28For more details on this research, see Taylor and Wilson (2012).
William Ribarsky, University of North Carolina at Charlotte
William Ribarsky, Wenwen Dou, and colleagues demonstrated digital tools for examining the relationship between science funding and the evolution of research fields. These tools use what they describe as a visual analytics approach. The tools make it possible to trace the progression of activity in a particular area. With these tools, Ribarsky explained, it is possible to analyze empirically and then represent visually the prevalence of particular ideas in proposals, papers, and the broader media. The visual representations can reveal trends, the impact of events or relationships, and the possible cause and effect relationships. The tools (ParallelTopics and LeadLine were primary examples) can be used for general exploration and to identify general trends or to conduct more detailed analyses. They can be used to forecast future impacts or to support decision making, he observed.
The visual representation tools make it possible to trace and depict the prevalence or impact of particular ideas, beginning with their presence in funding proposals (such as to the National Science Foundation) through the publication of papers and the granting of patents. Ribarsky and his colleagues have amassed data on 5 million patent applications, for example, and also have folded in textual analysis of papers and such other sorts of data as online news sources, technical and business blogs, and the like to identify research trends and their impact. Using these tools, Ribarsky noted, it is possible to answer such questions as whether funding is lagging behind research in a particular area, how relationships between funding and research evolve over time, and which proposals and papers are shaping funding decisions. Such answers can be useful in decisions about investments in research and programs or the structure of review panels, Ribarsky concluded.
An example of the information that visual representation tools can encapsulate and summarize is provided in Figure 3-1.
FIGURE 3-1 ParallelTopics visualization: Overview of awarded proposals in NSF Information and Intelligent Systems Division from 2000 to 2010.
SOURCE: Presentation to SciSIP Principal Investigators’ Conference by William Ribarsky, 2012.
This ParallelTopics visualization traces trends in proposals awarded by the NSF Information and Intelligent Systems Division in the Computer and Information Science and Engineering Directorate from 2000 to 2010. The data shown include about 4,000 awarded proposals. Each of the color-coded streamlets in Figure 3-1 represents a topic, which is identified by an automatically generated string of keywords. For example, consider the topic “robotics” indicated in light blue and the topic “using interfaces to help people with impairment” shown in green. The two topics selected are labeled with their leading keywords, but all the topics have a set of keywords (indicated by the other colors in the diagram). Selection of a topic at a specified time range gives the proposals’ abstracts for that topic awarded during that time range; a proposal can have more than one topic. The visual representations in the diagram show both the main themes of proposals over time and the programs from which they come. The width of a streamlet scales as the number of award abstracts for that topic at that time. The results in Figure 3-1 show a large jump in the number of awards that is visible in 2003–2004 and again in
2006-2007. Robotics maintained its numbers, while interfaces for people with impairments grew. These and other trends depicted were confirmed through evaluations by a former program manager for the directorate.29
Maryann Feldman, University of North Carolina at Chapel Hill
Private foundations play a key role in funding academic research, and Maryann Feldman’s research explored how attributes of philanthropic organizations affect the conduct of university research, their relationship to other funding sources and to commercial outcomes of scientific research, and possibilities for a new model of strategic foundation funding of research. She and her colleagues reviewed data on 19,000 projects that received foundation funding between 2000 and 2012. These projects were conducted by 6,000 principal investigators, who received more than $3.2 trillion in funding. Grant proposals provided information about agreements, contractual terms, and principal investigator characteristics, Feldman explained, and other sources provided data about corporate and foundation giving and about technology transfer. Overall, philanthropic funding for research increased during this period.
Approximately 60 percent of academic research and development funding comes from the federal government, Feldman noted, and 8 percent comes from philanthropy. However, internal university research grants, which account for another 20 percent, are also often the product of foundation gifts. Many foundations have adopted innovative funding models because they hope to improve the impact of their gifts, Feldman explained. One such innovation is venture philanthropy, a model in which funders adapt the approaches of venture capitalists in identifying ideas that have commercial potential and using their investments to nurture them and guide their development. For example, a foundation may manage the pipeline of a new drug from the research stage to production for the open market. Typically, a foundation will accomplish this by building a team focused on a particular goal, identifying progress milestones in advance, and assembling a portfolio of work.
There are several advantages to this approach, Feldman observed. Projects funded in this way are often more efficient than others, and funding can directly target areas for which there have been funding gaps, such as riskier research or that being conducted by younger researchers. Faculty members seem to favor this approach, Feldman noted, but the approach is not financially advantageous for universities, which may lose both licensing revenue and overhead payments. Feldman also noted that, in this approach, a small group of funders may dictate research priorities that may not align with the ideas of the research community or the needs of the general public.30
________________
29For more details on this research, see Dou et al. (2011); and Dou et al. (2012).
30For more details on this research, see Feldman and Graddy-Reed (2013, forthcoming).
EVIDENCE ON PATENT POOLS UNDER THE NEW DEAL
Ryan Lampe, DePaul University
As described by Ryan Lampe, patent pools are agreements among patent holders to combine related patents and possibly license them to third parties. For example, the holders of patents on airplane wings and propellers, respectively, might agree to join forces so that airplanes can be developed more efficiently and profitably. Such agreements reduce the risk of litigation among patent holders and may also reduce licensing fees. On the other hand, a pool member may share the benefits of another’s invention without making a contribution to the effort. The need to share profits may also reduce pool members’ motivation, Lampe added.
Lampe and co-author Petra Moser had noted that a nineteenth-century pool agreement concerning sewing machine technology had the effect of reducing patent grants, and also innovation, as measured by sewing machine speed. To gain a broader picture of the influence of pooling on innovation, they studied patent pools in 20 industries that were affected by the New Deal legislation of the 1930s. Under the New Deal, Lampe explained, there was a window of time in which regulations affecting patents were relaxed, and he and Moser wanted to know whether permissive policies regarding patent pooling had encouraged or discouraged innovation. Twenty pools formed between 1930 and 1938 concerning technology used for diverse applications, including railroad springs, furniture slip covers, and aircraft instruments. Lampe and Moser compared patents granted for pooled technologies with those granted for non-pooled developments in related technologies. They used court records and license agreements held in archives to research a total of 75,396 patents issued from 1921 to 1948. They identified subclasses of patent pools by pool size, technology type, and other factors to explore the reasons for pooling and the outcomes.
Lampe shared several findings. In general, patent rates declined after a pool was created. For example, the existence of a pool delayed the switch from black-and-white film to color film. The presence of a pool arrangement reduced competition among researchers in the pool to develop improved substitutes for existing technologies. In the absence of the antitrust regulations that were relaxed under the New Deal, pools also discouraged innovation by weakening competition in general, he concluded.
EFFECTS OF CHANGES IN FEDERAL FUNDING ON THE BIOMEDICAL SCIENCES
Meg Blume-Kohout, New Mexico Consortium Krishna Kumar and Neeraj Sood, RAND Corporation
Meg Blume-Kohout and her colleagues used econometric analyses to explore the effects of federal funding for research and development (R&D) on non-federal investment in life sciences R&D at U.S. universities. They took advantage of a change in budget policies at NIH, which resulted in a significant tightening of research funding after 2006, to compare outcomes under different funding regimes. They used multiple longitudinal data sources, including the NSF Survey of Research and Development Expenditures at Universities and Colleges, NIH administrative records, Congressional
budget appropriations by NIH Institute and Center, and Congressional subcommittee membership.
Blume-Kohout and her colleagues found that before the reduction in funding, “each federal dollar U.S. universities received spurred an additional $0.25 in research funding from non-federal sources,” including industry, state and local governments, philanthropic donors and nonprofit organizations, and other institutional sources. However, the more competitive funding environment from 2006 onwards had significant impact on less research-intensive universities, Blume-Kohout explained. Like the larger, more research-intensive PhD-granting institutions, these less research-intensive institutions benefitted from dramatic increases in federal funding from FY1998 through FY2004, but from FY2005 onwards, less research-intensive institutions experienced a much steeper decline in federal funds. Nonetheless, when these latter institutions did succeed in attracting federal funds post-2006, their complementary non-federal R&D investment increased by over $0.63 the following year. In contrast, more research-intensive PhD-granting universities appear to have substituted non-federal R&D funding at least 1:1 when federal funding availability declined (see Figure 3-2).
FIGURE 3-2 Non-federal life sciences R&D funding substitutes for federal dollars at Carnegie doctoral high /very high research universities.
SOURCE: Presentation to SciSIP Principal Investigators’ Conference by Meg Blume-Kohout, 2012.
The nonlinearity in non-federal R&D investment responses to the changing fiscal environment could reflect several possible influences. For example, the strong and continuing positive impact of federal R&D funds at historically less research-intensive institutions may be due to signaling effects, or due to a unique role in federal R&D funding in building those universities’ productive capacity, particularly via investments in facilities, equipment, or human capital. On the other hand, more research-intensive institutions may make strategic decisions to pursue non-federal funding when federal award success rates and total funding decline.
EXTRACTING AND ASSESSING THE PUBLIC VALUES OF SCIENCE AND INNOVATION POLICIES
Daniel Sarewitz, Arizona State University
Science and innovation policy can have powerful impacts on individuals and society, and Daniel Sarewitz explored ways to understand and measure this impact. He noted that it is important to ask, with respect to any project, what public values it might serve, whether the reasoning about how it might serve those goals is sound, whether the necessary human and institutional resources are in place, and what strategies the project leaders have for linking the institutions and people involved.
Sarewitz explained that by public values, he means desired social outcomes that would justify public investment in particular research. Examples, he noted, include increasing the length and quality of people’s lives and eliminating health disparities, ensuring a safe and affordable food supply, and fostering a reliable energy system that is environmentally and economically sustainable. Scientists’ perspectives on issues are often different from those of the general public, however, as he illustrated in Figure 3-3, which compares scientists’ perceptions of various risks with those of the general public.
FIGURE 3-3 Public versus scientists’ values on perceived risks.
SOURCE: Presentation to SciSIP Principal Investigators’ Conference by Daniel Sarewitz, 2012.
Sarewitz noted that the traditional model by which the logic of science policy is mapped does not capture public values well. Figure 3-4 compares the traditional model to the model that he proposes. His model includes a non-economic analysis of public values and the potential impact of a policy on them. Sarewitz noted that there are a variety of ways that science policy may fail the general public. A policy might focus on effects likely within a short time frame in a case where understanding of longer-term outcomes would likely point to a different course of action. Research and development associated with energy is an example: immediate needs for affordable sources of energy may obscure appreciation for the long-term downsides of particular approaches. Figure 3-5 shows Sarewitz’s listing of possible policy failures.
FIGURE 3-4 Traditional versus enhanced science policy logic models.
SOURCE: Presentation to SciSIP Principal Investigators’ Conference by Daniel Sarewitz, 2012.
Sarewitz used climate change research to illustrate this point, noting that criteria 1 and 3 in his map apply to failures in this area. Policy in this area has lacked a mechanism for articulating and aggregating public values (criterion 1), he stated. The criteria for assigning priorities to different values are too broad or vague, and the connections between priorities and values or outcomes have not been established. An additional problem is that the stakeholders in climate change science policy are not clearly defined. There is also a scarcity of providers of expertise (criterion 3), he added. Despite recognized needs that lie outside of natural science, priorities have tended to be driven by research in the natural sciences, rather than social science research that might more readily reveal competing values.
Sarewitz concluded by emphasizing that science and innovation policy can have significant impact on people’s lives, and that the Public Value Mapping tool can assist researchers, funders, and policy makers in making optimal use of investments in science and innovation.31
________________
31For more details on this research, see Bozeman and Sarewitz (2011, 2005); and Sarewitz (2007).
