Accepted Article
The decline of science in corporate R&D
Authors:
Ashish Arora
Titles and affiliations: Rex D. Adams Professor. Fuqua School of Business, Duke University, and
NBER.
Address: Duke University, Fuqua School of Business, 100 Fuqua Drive, Durham, NC 27708.
Email: ashish.arora@duke.edu
Sharon Belenzon
Titles and affiliations: Associate Professor. Fuqua School of Business, Duke University.
Address: Duke University, Fuqua School of Business, 100 Fuqua Drive, Durham, NC 27708.
Email: sharon.belenzon@duke.edu
Andrea Patacconi
Titles and affiliations: Senior Lecturer. Norwich Business School, University of East Anglia.
Address: Norwich Business School, University of East Anglia, Norwich Research Park, Norwich,
NR4 7TJ, UK.
Email: A.Patacconi@uea.ac.uk
Running head: The decline of science in corporate R&D
Keywords: innovation, scientific research, development, patents, use of science in inventions.
JEL Classification: O31, O32, O16.
This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which
may lead to differences between this version and the Version of Record. Please cite this
article as doi: 10.1002/smj.2693
This article is protected by copyright. All rights reserved.
Accepted Article
The decline of science in corporate R&D
Research summary: In this paper, we document a shift away from science by large corp orations
between 1980 and 2006. We find that publications by company scientists have declined over time
in a range of industries. We also find that the value attributable to scientific research has dropped,
whereas the value attributable to technical knowledge (as measured by patents) has remained stable.
These trends are unlikely to be driven principally by changes in publication practices. Further
science continues to be useful as an input into innovation. Our evidence points to a reduction of the
private benefits of internal research. Large firms still value the golden eggs of science (as reflected
in patents) but seem to be increasingly unwilling to invest in the golden goose itself (the internal
scientific capabilities).
Managerial summary: There is a widespread belief among commentators that large Ameri-
can corporations are withdrawing from research. Large corporations may still collaborate with
universities and acquire promising science-based start-ups, but their labs increasingly focus on de-
veloping existing knowledge and commercializing it, rather than creating new knowledge. In this
paper, we combine firm-level financial information with a large and comprehensive dataset on firm
publications, patents and acquisitions to quantify the withdrawal from science by large American
corporations between 1980 and 2006. This withdrawal is associated with a decline in the private
value of research activities, even though scientific knowledge itself remains important for corporate
invention. We discuss the managerial and policy implications of our findings.
This article is protected by copyright. All rights reserved.
Accepted Article
1 Introduction
During the 20th century and especially the post-war period, the U.S. created a scientific-industrial
complex that greatly contributed to scientific progress and resulted in many important innovations.
A key component of this scientific-industrial complex was the large corporate lab in corporations
such as AT&T, Du Pont, IBM and Xerox. Research from such labs has led to many important
discoveries such as the transistor, the laser, and the first computer with a graphical user interface,
as well as breakthroughs in medicine and pharmacology.
Since the 1980s, however, the U.S. scientific-industrial complex has undergone profound trans-
formations (Mowery, 1995, 2009; Hounshell, 1996; Pisano, 2010). A key transformation has been
the redirection, by many leading firms, of resources and attention from more exploratory scientific
research toward more commercially-oriented projects. But though articles in the popular press do
lament the demise of top-flight corporate labs (e.g., Economist, 2007), researchers have yet to doc-
ument the breadth and depth of this transformation, or whether this decline in ‘corporate science’
is related to a decline in the economic value of research, and if so, whether it is private or social
economic value.
1
In this paper, we take a step toward filling these gaps. A fuller understanding of the nature
and extent of this withdrawal is a first step towards understanding the possible reasons, such
as the growth of technology markets (Arora et al., 2001; Chesbrough, 2003; Arora et al., 2016),
globalization (Bloom et al. 2016; Autor et al. 2016), and managerial short-termism (e.g., Marginson
and McAulay, 2008). Our findings are likely to be relevant to managers and scientific entrepreneurs,
who operate within and must understand the evolution of their ecosystem (e.g., Pisano, 2010), and
to policymakers, who might wish to influence it.
Our primary contribution is to establish a set of important facts about the changing nature
of corporate R&D over a quarter of a century. To do so, we develop publication-based indicators
of scientific research at the firm level. We link scientific publications in “hard science” journals
(including engineering science) from the Web of Science to publicly traded firms in the United
States, using the affiliations of the authors. Our primary firm sample consists of 4,068 publicly listed,
R&D-performing companies with at least one patent over the period 1980–2006. Collectively, these
firms account for 452,297 “firm publications”—scientific articles where at least one of the authors
1
For simplicity, we use the terms “science” and “research” (or “scientific research”) interchangeably. The key
distinction we make is between “research” (as measured by publications in scientific journals) and “development” (as
measured by patents). We largely avoid finer distinctions such as that between basic and applied research because
these distinctions are often difficult to draw in practice.
This article is protected by copyright. All rights reserved.
Accepted Article
is a company employee.
We find that, over the period 1980–2006 participation in scientific research by publicly traded
American companies diminished. A significant fraction of the decline can be attributed to entry by
firms that do not publish or publish very little. However firms with established research programs
also markedly decreased research. In terms of the nature of research, the decline is most evident in
high-quality publications. The implied value of scientific capability, as measured by stock market
valuations or by the acquisition price in M&A deals, also declined. By contrast, patenting by large
American firms increased and the implied value of patents, including the premium paid for patents
in M&A, did not decrease. We find no evidence that invention became less science-intensive, or
that the science used in inventions grew older. These patterns are present across a broad range of
industries, except perhaps biotechnology.
As mentioned above, we are not the first to note the decline of many large corporate labs, or docu-
ment that corporate scientists are publishing less (e.g., Coombs and Georghiou, 2002; Bhaskarbhatla
and Hegde, 2013). Tijssen (2004), for instance, shows that the total number of papers published by
corporate researchers in academic journals substantially declined over the period 1996-2001. Data
from the National Science Foundation (NSF) also indicates that the share of basic and applied
research in corporate R&D in the United States declined from 28 percent in 1985 to 21 percent in
2009.
2
However, a limitation of both Tijssen’s paper and the NSF statistics is that they present
aggregate patterns. Thus, it is difficult to assess whether the trends they document reflect changes
in the behavior of existing firms or other factors, such as a change in the industrial mix of reporting
firms or the entry of innovative firms that do not engage in research.
3
The first contribution of this paper is to distinguish between these effects, and quantify them. A
second key contribution is to jointly analyze changes in both the value and the quantum of scientific
outputs. By matching firm publications with stock market data as well as M&A data, we are able
to estimate the implied “value” of scientific capability. Previous work has argued that capabilities
in research are plausible sources of competitive advantage (e.g., Gambardella, 1992, 1995; Durand
et al., 2008). Griliches (1986) analyzed the drivers of productivity and profits for a sample of the
1,000 largest manufacturing firms in the U.S. For the period 1967–1977, he found that the share
of basic research in the firm’s R&D expenditure was p ositively related to measures of productivity
growth. Henderson and Cockburn (1994) showed that bundles of organizational practices capturing
2
See NSF Science & Engineering Indicators 2016, Appendix Tables 4.3, 4.4, 4.5.
3
A second issue with NSF data is that distinguishing between research and development requires subjective judg-
ments. Such judgments may not be fully reliable or consistent over time.
This article is protected by copyright. All rights reserved.
Accepted Article
pharmaceutical firms’ propensity to connect with the external scientific community were strongly
associated with success in drug discovery. Our results qualify these important early findings, as we
show that for large firms in broad range of industries, the value attached to engaging in scientific
research has declined over the last quarter of a century.
By estimating the implied value of scientific capability, we also probe the proximate causes of
the decline of science in corporate R&D. Large firms may be producing fewer scientific publications
because (i) the private value of investments in internal (in-house) scientific research has declined,
or (ii) its cost has increased. A third possibility is that firms have not reduced their investments in
science: what has changed is their propensity to publish their research results in academic journals.
We refer to this possibility as (iii) a change in publication practices. Though we cannot definitively
rule out any of these possibilities, we argue that our findings, as well as aggregate NSF data, can
most easily be reconciled with a reduction in the net future benefits from internal scientific research.
To summarize, our contribution is to go beyond case studies and aggregate data to document
the extent to which corporate engagement in research has changed over time, within and across
firms. Our findings suggest that, in reducing their engagement in science (as measured by firm
publications), firms have been following market signals. Large firms are publishing less in scientific
journals because these activities are now less privately valuable than they once were, as indicated
by the collective judgement of investors and managers. This does not mean that research is not
socially b eneficial. Our evidence that invention is not becoming less science-intensive, or that the
vintage of science used in inventions is not growing older, further suggests that the social value of
science is not declining. We discuss, but do not empirically analyze, a variety of factors that might
account for why the private value of science, for established firms, has diverged from the social value
of science. In concluding the paper, we also discuss the implications of this reduction in corporate
science for managers, as well as public policy.
2 Conceptual background
There is an extensive literature in strategy on corporate engagement in scientific research. Although
many innovations arise through serendipity or through knowledge generated outside formal R&D,
corporations have also invested in science to accelerate the introduction of new products and pro-
cesses. Innovations sometimes build directly on scientific advance (e.g., new drugs), and sometimes
arise as indirect outputs of scientific research (e.g., laser). In other cases, scientific research en-
hances the productivity of technical search by guiding it toward more fruitful pastures (Evenson
This article is protected by copyright. All rights reserved.
