What Drives Innovation?
A DISCUSSION OFWhat Does Innovation Today Tell Us About the US Economy Tomorrow?
Read Responses From
In “What Does Innovation Today Tell Us about the US Economy Tomorrow?” (Issues, Fall 2017), Jeffrey Funk starts with an assertion that puzzles me, but after that he develops and provides evidence for a point of view that is quite consistent with my knowledge. He asserts early on that most scholars of innovation see new scientific knowledge as generally providing the focus and capability for successful efforts to develop new products and processes—what is often called the linear model. However, though many years ago the linear model was widely believed in the innovation research community, it has been almost completely abandoned over the past quarter century, a bit too sweepingly in my view since it is not a bad first approximation to much of what is going on in a few fields, particularly biomedicine.
But after that assertion, Funk makes it clear that he does not hold to that theory, and he spends most of his article providing evidence that knocks it down. His report on the referencing to science in patents taken out by a collection of successful start-ups is a useful contribution to a wide range of empirical evidence we now have that in most industries and technologies, invention generally does not rely on new science. And Funk’s data, as that in other studies, show biotech as something of an exception.
Funk argues that in most fields of technology what he calls the Silicon Valley model fits what is going on in innovation much better than the linear model, and I think most scholars of innovation would agree. In these fields, most particular innovations are incremental, but the effort is cumulative and progress can be very rapid. Funk makes a point that progress at any stage often is driven by recent advances in component technologies and materials, again an argument that is consistent with a number of other empirical studies. He gives us some very nice examples.
Toward the close of his article Funk comes out agreeing with Robert Gordon and Tyler Cohen that in recent years rapid technological advance has been occurring in only a few sectors, those drawing on biotech and those involved in information processing being the most important ones. He takes the position that an important route toward broadening the innovation that is going on is to take steps to make research done at universities more focused on explicitly creating the bases for technological advance in areas where the need is great. For reasons that he surely knows but does not go into, this is a very controversial argument. But he will have many people who strongly agree with him.
Richard R. Nelson
Jeffrey Funk’s essay contains informative vignettes about the contributions of science to technological innovation in selected industries and about technological innovations in other industries that are formed instead by what the economist W. Brian Arthur has called “fresh combinations of what already exists” and are essentially independent of scientific advances. Other than perhaps eliding the case of technological advances serving as essential building blocks to scientific advance, these vignettes add to but largely restate well-known propositions in studies of causal linkages between scientific discoveries and technological innovation.
The essay’s efforts to relate its vignettes to national science and technology policy issues are hampered, however, by an overdone reliance on a stylized dichotomy between a linear, science-based, model and technology-built (Silicon Valley) models of innovation. Its presentation and interpretation of data on the percentage of patents citing scientific and engineering publications to “identify the parts of the US innovation system that are working well and those that are not” also are unconvincing.
The crux of the argument is that the frequency of citations of publications in patents is a measure of the importance of the science-based model. Thus, as Funk reported in Table 1, low percentages to the Billion Dollar Startup Club are presented as indicating little contribution of science to economic growth. Note needs to be made here of the difference between this finding and that of Francis Narin and colleagues, who in 1997 used patent citations to publications in highly regarded journals to demonstrate 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.” Relationships clearly may have changed over the past 20 years. But the more likely explanation for the difference is Funk’s limited and uncritical use of data. The data in Table 1 are largely consistent with a priori expectations about which industries would rely on patents for protection of intellectual property and which would not (biotechnology and e-commerce, respectively), and relatedly on knowledge embedded in scientific papers. In Funk’s interpretation, no allowance is made for differences in an industry’s reliance on patents that are, say, relative to trade secrets to establish intellectual property rights or to the relative importance of different patents within a firm’s patent portfolio.
The issue of the relative mix of mission-oriented and science-oriented investments addressed by the essay is of longstanding importance. In 1987, likely the nadir of US international competitiveness in traditional manufacturing sectors and the peak of concern about declining leadership in scientific and technological endeavors, the economist Henry Ergas introduced the distinction between mission-oriented and diffusion-oriented national research strategies. He described the former as “big science deployed to meet big problems” and the latter as “policies that seek to provide a broadly based capacity for adjusting to technological change throughout the industrial structure.”
Much of US science and technology policy from the early 1980s on, encompassing the Bayh-Dole Act, the Stevenson-Wydler Act, the National Cooperative Research and Production Act, and the Small Business Innovation Development Act, as well as agency-specific initiatives such as the National Science Foundation’s funding of Engineering Research Centers, may be seen as experiments designed to foster a more mission-oriented cast to federal investments in research and development (R&D). Some of these policy initiatives have worked well, others less so.
Against the backdrop of projected near-term austerity in real levels of federal R&D funding, it is not clear what Funk has in mind when he contends that “US policy makers should be moving more of the nation’s R&D investment toward a mission-based approach, and they should be experimenting with different approaches to implementation.” What type of investment/performer/societal objective (other than economic growth) funding mechanism will bear the opportunity costs of reduced funding? Without providing specific answers to these questions, depending on one’s choice of metaphors, moving in the direction of his recommendation is either opening a black box or a Pandora’s box.
Jeffrey Funk draws attention to two perspectives on innovation and offers a potential remedy for improving the current innovation approach. The first perspective distinguishes between the optimists’ and pessimists’ views of the economic impact of the current state of innovation in the United States. The second distinguishes between science-based (science-focused) and Silicon Valley-based (mission-focused) innovation. Funk postulates that Silicon Valley innovation is not as dependent on basic or applied science, and that it commercializes faster but penetrates smaller sections of the economy. He proposes speeding science-based innovation and focusing on critical areas for its application—merging academic research with mission-based goals for society. Before addressing this solution, I want to consider the issues with Funk’s assessment of the situation.
Funk’s observations and examples reveal the difficulty of defining, measuring, and tracking innovation, major problems in evaluating innovation’s effectiveness, economic penetration, and upstream and downstream effects. Researchers such as Funk lack time series data (data points indexed in time order) or other relevant data, relying instead on a series of individual surveys, personal anecdotes, and inconsistent methodologies. The assumptions made about data quality and compatibility result in measurement and forecasts that fail to provide policy-makers with the appropriate information to make critical decisions.
Better insights can be achieved with indicators that more comprehensively measure the multiple facets of innovation. Changing the measure of the nation’s gross domestic product to include R&D, computer software and databases, entertainment, and literary and artistic originals as investments rather than expenses, as proposed by the economist Marissa J. Crawford in 2014, would be a step in the right direction. Since companies expect their current spending in these investments to generate future returns and investors consider them in assessing the firm’s market value (versus book value), the investments are indicative of the value of the economy. Broadening the definition of innovation to include investments in design, branding, new financial products, organizational capital, and firm-provided training and other intangibles would provide a more challenging but improved measurement.
Broadening the definition of innovation beyond commercialization would include many additional activities and outputs. For example, currently unmeasured is what the economist Eric von Hippel has called “free innovation,” where innovators have a specific and often personal need to create new products or processes and to make them available to all. Free innovation takes many forms, from medical devices to sports equipment to open source software. A comprehensive measure of innovation would include these free innovations.
These new measures can take advantage of new ways to collect data, such as opportunity data from crowdsourcing and the internet. These new sources of data would supplement the current survey and national accounts measures and provide new insights into current measures.
Funk’s solution to improving the economic impact of innovation is also problematic. If as he proposes innovation must be made more mission-focused, who will decide the missions, the critical areas for supporting research, and the mission-based goals? How will the decisions be made? And what happens to funding for pure basic research? Although tools can be created without basic science research—even cave dwellers and crows have done it—basic science is the feedstock for improving those tools. How does society prevent this mission approach from removing the feedstock for future improvement?
As Funk suggests, there is a need for an improved link between innovation and its economic impact. Better measurement tools and approaches are needed to assess the total economic impact of innovation before a more mission-focused strategy can arguably improve economic impact.
Professor of Statistics
I largely agree with Jeffrey Funk’s analysis and his prescriptions for improving the yield of academic research projects. He makes a strong separation between science-based innovation and the Silicon Valley process of technology change. And he is correctly critical of the linear model. But I think he would do well to celebrate the positive role that technology-driven innovation plays in providing new challenges that advance science. For example, the invention of the transistor came from technology-driven needs to replace vacuum tubes, later leading to the discovery of the transistor effect, which earned the inventors a Nobel Prize in Physics. There are many such examples, but linear model advocates often rewrite history to favor their misguided model.
I like Funk’s analyses of the MIT Technology Review predictions and his claims that large economic impacts are more likely to come from technology-based innovations. I agree with his recommendations that tying scientific research more closely to national priorities and mission-driven projects would be helpful and that a slightly more centralized approach would be beneficial, as is emerging with the Engineering Research Centers sponsored by the National Science Foundation and the Manufacturing.gov partnerships. Of course, there should always be room for blue-sky explorations and theory-driven science.
One concern is that Funk appears to believe that change can come only through top-down government policy shifts, but bottom-up changes can happen from individual researchers, laboratory directors, and campus leaders who recognize the paths to high-impact research by working more closely with business, government, and nongovernmental organization partners. There is evidence that both paths to change are happening, so more articles such as this are helpful to accelerate these changes that will lead to better research that has higher societal impact.
Professor of Computer Science
University of Maryland