“Instrumental Burden”: The Rising Cost of Scientific Instruments and Consequences for Translational Research

Why has the translation of science into productivity growth slowed in recent years? I suggest that the rising capital costs of modern science may have hollowed out applied research at universities, hampering the commercialization of science. Semiconductor science, for instance, requires access to the latest equipment to fabricate substrates and metrology equipment to observe test results. Work in structural biology depends heavily on high fixed-cost equipment such as Cryo-Electron Microscopes, while drug candidate selection is done with the aid of high throughput screening machines. The fixed cost of these investments, which often total in the multi-millions, act as an entry barrier for smaller research teams, potentially affecting the direction of university science. Despite their importance, no comprehensive dataset of scientific equipment and their prices exist. I therefore scrape data from 13,453 federal grants for research instrument purchases and supplement it with an industry buyer’s guide to construct a comprehensive dataset of major research tools.  I then show that select subfields of science (chemicals, materials, and the life sciences) have experienced over threefold increases in capital intensity as measured through equipment mentions in peer-reviewed papers. However, public funding for instrument purchases have failed to keep up with this growing pace. To compensate for restricted access to the latest equipment, academics in capital-intensive subfields have focused on scientifically novel research at the expense of commercial relevance (as measured through patent citations). Scientific hardware has also been increasingly concentrated in a handful of elite institutions, potentially affecting the diversity of research paths.

(Presented at the 2025 Industry Studies Association Annual Meeting, 2025 Roundtable for Engineering Entrepreneurship Research (REER) Conference)

The Rise of Absorptive Research in Corporate America: 1945-1980

With Ashish Arora, Sharon Belenzon, and Hansen Zhang

We study the post–World War II “Golden Age” of American corporate research from 1945 to 1980, using multiple indicators of corporate research activity. We use an ensemble learning approach to classify firms as either Science Leaders, Absorbers or Followers. Our analysis reveals that only a small fraction of firms, whom we call Leaders, invest in internal research that is on the scientific frontier, with the objective to generate breakthrough inventions. Absorbers invest in research principally to absorb external scientific discoveries to fuel their inventive activity. Followers typically generate incremental innovations, using older scientific knowledge. Consistent with this, we find Leaders were more likely to be at the technological frontier, enjoy greater market power, and benefit from government procurement contracts. As universities and startups began to commercialize academic discoveries, the need for “absorptive corporate labs” declined. The shift ultimately transformed the American innovation landscape, deepening the division of innovative labor between universities, startups, and incumbent corporations, with only a select group of Leader firms continuing to invest in basic science.

(Presented at the 2025 Wharton Corporate Strategy and Innovation Conference)

The Reorganization of the American Innovation Ecosystem and the Challenge of Translating Science

With Ashish Arora, Sharon Belenzon, and Andrea Patacconi, Industrial & Corporate Change (2025)

In this paper, we focus on lack of translational research as a potential explanation for the recent slowdown in productivity growth, as opposed to a slowdown in science or the decline in the novelty of science. We provide evidence that the translation of science (as measured through citations by patents) has dropped for novel science. This drop is correlated with the withdrawal of large firms from science (especially in the physical sciences), as measured through their publications. Fields where VC-backed startups have entered have experienced an increase in the translation of science, but their entry has largely occurred in the life sciences and ICT. Sectors such as chemicals and material science have instead been neglected, both by large firms and startups. Our results point to the role of large firms and their labs in facilitating the translation and commercialization of science. A growing division of innovative labor may have delivered gains from specialization in the production and use of science in some sectors, but appears to have left a void in others.

Swinging for the Fences: Startup Novelty as a Response to Entry Costs

Revise & Resubmit at Organization Science

Why are startups more likely than incumbents to commercialize novel products when faced with a scientific breakthrough? The question has become a pressing one in light of a deepening division of innovative labor in the American innovation ecosystem that increasingly tasks startups with the job of commercializing science into products. The literature has explained the higher innovative propensity of some startups by noting their superior fit with new technology and markets. However, it is also possible that the apparent innovativeness of startups may be a result of firm choice rather than inherent capability gaps. Startups may choose to develop novel products that are risky but offer high payoffs because they pay a higher entry cost than incumbents in the form of investments in new factories, sales, and distribution channels. This selection into novelty due to higher entry cost has received little attention so far because it is difficult to rule out upstream capability differences between potential entrants. To address this challenge, I link the influx of new laser science from the former Soviet Union after the end of the Cold War to patents (pre-entry) and products (post-entry) by American laser firms. Startups are approximately 37% more likely than incumbents to introduce new products using Soviet science, but not more likely to rely on Soviet science during pre-entry patenting. Moreover, the probabilities of both entry and exit are lower for startups than for incumbents, which is consistent with startups choosing more novel innovations due to higher market entry costs. The results suggest that startups have a comparative advantage in innovating with novel technologies and markets because of their absolute disadvantage in downstream commercialization.

(Presented at the 2021 NBER Productivity Seminar, 2021 CCC Doctoral Conference, HEC Lausanne PhD Workshop on Innovation & Entrepreneurship)

The Rise of Scientific Research in Corporate America

With Ashish Arora, Sharon Belenzon, Konstantin Kosenko and Yishay Yafeh, Organization Science (2024)

It is widely believed that university and corporate research are complementary: Companies invest in research in part to develop the capacity to absorb the knowledge emerging from universities. However, as we show in this paper, corporate research in the United States emerged when American universities were behind the world frontier in scientific research. Why, then, did for-profit businesses choose to invest in creating new knowledge, much of which could spill over to rivals, and whose conduct presented many managerial challenges? We argue that corporate research in America arose in the 1920s to compensate for weak university research, not to complement it. Using newly assembled firm-level data from the 1920s and 1930s, we find that companies invested in research because inventions increasingly relied on science, but American universities were unable to meet their needs. Large firms, close to the technological frontier, and operating in concentrated industries were likely to invest in research, especially in scientific disciplines where American universities lagged behind the scientific frontier. Corporate science seems to have paid off, resulting in novel patents and high market valuations for firms engaged in research.

(Presented at 2022 AOM Annual Meeting, XIX World Economic History Congress, 2021 Strategy Science Conference, 2021 Wharton Technology & Innovation Conference, 2021 American Economic Association, Boston University IP Day 2020: “Intellectual Property and Technology Markets”, Boston University Technology and Declining Economic Dynamism Conference)

Science and the Market for Technology

With Ashish Arora and Sharon Belenzon, Management Science (2022)

Well-functioning Markets for Technology (MFT) allow inventors to sell their inventions to others that may derive more value from them. We argue that the growing use of science in inventions enhances MFT. Science-based inventions have higher gains from trade and lower transaction costs. This relationship is amplified in equilibrium because science-based inventions are also likely to feature smaller inventors with a greater propensity to trade. Using large-scale data, we show that patents citing science are more likely to be traded, especially for novel patents and for smaller inventors. We conclude that the growing use of science in invention is beneficial by encouraging the expansion of MFT and supporting a division of innovative labor.

(Mentioned in New Things Under the Sun)

The Changing Structure of American Innovation: Cautionary Remarks for Economic Growth

With Ashish Arora, Sharon Belenzon and Andrea Patacconi, NBER Innovation Policy and the Economy (2020)

A defining feature of modern economic growth is the systematic application of science to advance technology. However, despite sustained progress in scientific knowledge, recent productivity growth in the United States has been disappointing. We review major changes in the American innovation ecosystem over the past century. The past three decades have been marked by a growing division of labor between universities focusing on research and large corporations focusing on development. Knowledge produced by universities is not often in a form that can be readily digested and turned into new goods and services. Small firms and university technology transfer offices cannot fully substitute for corporate research, which had previously integrated multiple disciplines at the scale required to solve significant technical problems. Therefore, whereas the division of innovative labor may have raised the volume of science by universities, it has also slowed, at least for a period of time, the transformation of that knowledge into novel products and processes.

(Mentioned in Financial Times Alphaville, The Economist, AEI, The Atlantic, Quartz)

Why the U.S. Innovation Ecosystem is Slowing Down

With Ashish Arora, Sharon Belenzon and Andrea Patacconi, Harvard Business Review Online (2019)  

The data suggests that American innovation is sputtering: Productivity growth in the United States, which is powered by innovation, has been decelerating. Total factor productivity grew substantially in the middle of the 20th century, but started slowing in 1970. This slow growth continues today, with productivity lower than it was more than 100 years ago, despite increased investment in scientific research. What explains this? Research suggests that the U.S. innovation ecosystem has splintered since the 1970s, with corporate and academic science pulling apart and making application of basic scientific discoveries more difficult. If we want to see greater productivity growth, we need to explore alternative ways to translate science into invention