NBER WORKING PAPER SERIES
CLIMBING ATOP THE SHOULDERS OF GIANTS: THE
IMPACT OF INSTITUTIONS ON CUMULATIVE RESEARCH
Jeffrey L. Furman
Scott Stern
Working Paper 12523
http://www.nber.org/papers/w12523
NATIONAL BUREAU OF ECONOMIC RESEARCH
1050 Massachusetts Avenue
Cambridge, MA 02138
September 2006
We thank each of the scientists who graciously offered their insights, the personnel of the American Type
Culture Collection, and especially Dr. Raymond Cypess and Robert Hay. We thank Andrew Bernard,
Bronwyn Hall, Shane Greenstein, Ben Jones, Michael Kremer, Robert Litan, Megan MacGarvie, Paul Romer,
Mathieu Trepanier and participants in numerous seminars for comments and suggestions. Jason Corradini,
Mercedes Delgado, Lorraine DeLeon, Chijoke Emineke, Jeremy Gagne, Anna Harrington, Martha Kam, Julia
Lo, Kinga Piekos, Courtney Mason, Tracy Myers, Michael Vitulli, and Naomi Wohl provided excellent
research assistance. All errors are our own. Financial support for this research was provided by the
Brookings Institution and by the Boston University School of Management Junior Faculty Research Fund.
Author contact information: Scott Stern, Kellogg Graduate School of Management, 2001 Sheridan Road,
Evanston, IL 60208 s-stern2@northwestern.edu, and Jeffrey L. Furman, Boston University School of
Management, 595 Commonwealth Ave -- #653a, Boston, MA 02215, furman@bu.edu. The views expressed
herein are those of the author(s) and do not necessarily reflect the views of the National Bureau of Economic
Research.
©2006 by Jeffrey L. Furman and Scott Stern. All rights reserved. Short sections of text, not to exceed two
paragraphs, may be quoted without explicit permission provided that full credit, including © notice, is given
to the source.
Climbing Atop the Shoulders of Giants: The Impact of Institutions on Cumulative Research
Jeffrey L. Furman and Scott Stern
NBER Working Paper No. 12523
September 2006
JEL No. H4, L3, O3, O33
ABSTRACT
While the cumulative nature of knowledge is recognized as central to economic growth, the
microeconomic foundations of cumulativeness are less understood. This paper investigates the
impact of a research-enhancing institution on cumulativeness, highlighting two effects. First, a
selection effect may result in a high correlation between “high-quality” institutions and knowledge
of high intrinsic quality. Second, an institution may have a marginal impact – an incremental
influence on cumulativeness, conditional on the type and quality of knowledge considered. This
paper distinguishes these effects in the context of a specific institution, biological resource centers
(BRCs). BRCs are “living libraries” that authenticate, preserve, and offer independent access to
biological materials, such as cells, cultures, and specimens. BRCs may enhance the cumulativeness
of knowledge by reducing the marginal cost to researchers of drawing on prior research efforts. We
exploit three key aspects of the environment in which BRCs operate to evaluate how they affect the
cumulativeness of knowledge: (a) the impact of scientific knowledge is reflected in future scientific
citations, (b) deposit into BRCs often occurs with a substantial lag after initial research is completed
and published, and (c) “lagged” deposits often result from shocks unrelated to the characteristics of
the materials themselves. Employing a difference-in-differences estimator linking specific materials
deposits to journal articles, we find evidence for both selection effects and the marginal impact of
BRCs on the cumulativeness of knowledge associated with deposited materials. Moreover, the
marginal impact increases with time and varies with the economic and institutional conditions in
which deposit occurs.
Jeffrey L. Furman
Boston University – SMG
595 Commonwealth Avenue – #653a
Boston, MA 02215
and NBER
furman@bu.edu
Scott Stern
Management and Strategy Department
Kellogg School of Management
2001 Sheridan Road
Evanston, IL 60208
and NBER
s-stern@northwestern.edu
1
“If I have been able to see further, it was only because I stood on the shoulder of giants.”
Isaac Newton, 1676
I. INTRODUCTION
At least since the development of scientific societies and related research institutions in
the 17
th
century, the centrality of cumulativeness in scientific and technical advance has been
recognized.
1
However, from the perspective of economic theory, cumulativeness has only been
incorporated recently, in models of endogenous economic growth [Romer, 1990; Grossman and
Helpman, 1991; Jones, 1995] and step-by-step technical progress within industries [Scotchmer,
1991; Gallini and Scotchmer, 2003]. In order to serve as a foundation for long-term growth,
scientific research and technological progress must exert a positive intertemporal spillover; as
Jones [1995] emphasizes, to avoid diminishing returns to research investments, research itself
must “stand on the shoulders” of prior knowledge.
Though extremely insightful in deriving the implications of cumulativeness for related
economic variables (such as the equilibrium growth rate or the incentives for innovation), these
models do not articulate the conditions that result in a cumulative research environment. For
example, as Mokyr [2002] elegantly and persuasively argues, the mere production of knowledge
does not guarantee that others will be able to exploit it. Effective diffusion of knowledge across
researchers and over time requires that individuals are aware of extant knowledge and pay the
costs of accessing that knowledge. Further, since any individual researcher only captures a small
share of the benefit from the process of certifying knowledge and making it accessible, there may
be a significant gap between the private and social returns associated with investments that
contribute to the diffusion of scientific knowledge. Overall, then, the ability of a society to stand
on the shoulders of giants depends not only on the amount of knowledge it generates, but on the
1
Though certainly not the first example, Newton’s recognition of cumulativeness is (famously) recognized in his
classic 1676 letter to scientific rival Robert Hooke in the context of a dispute over the nature of light: “What
Des-Cartes did was a good step. You have added much several ways, & especially in taking ye colours of thin
plates unto philosophical consideration. If I have seen further it is by standing on ye sholders of Giants.”
Economic historians and economists of technical change, most notably Nathan Rosenberg (1963), highlighted the
centrality of cumulativeness in economic growth long before this idea was incorporated into formal models. As
well, the role of institutions in promoting the explosion of scientific research in 17
th
century England is the
cornerstone of Merton’s seminal contributions to the sociology of science (Merton, 1957; 1973), which itself has
served as a foundation for the “new” economics of science (Dasgupta and David, 1994).
2
quality of mechanisms for storing knowledge, the fidelity of knowledge, and the costs of
accessing knowledge.
Institutions and public policy are often suggested as central to the cumulative process.
Social scientists face a considerable challenge, however, in assessing the extent to which any
particular institution influences the way in which the “knowledge stock” is created, maintained,
and extended. It is empirically difficult to isolate the impact of a particular piece of knowledge
from the impact of the particular institution in which it is embedded, although the two are
conceptually distinct. Two forces may be at work: First, a selection effect may result from a
high correlation between “high-quality” institutions and knowledge of high intrinsic quality.
However, for policy analysis, we are often more interested in the marginal impact of an
institution – the incremental influence of an institution on cumulativeness, conditional on the
nature and quality of knowledge embodied in it. Without an identification strategy capable of
isolating selection and marginal effects, research examining the role of institutions on knowledge
diffusion will confound these two effects.
The main contribution of this paper is to provide direct statistical evidence of the impact
of a specific institution – biological resource centers (BRCs) – on the cumulativeness of
knowledge, distinguishing the marginal impact of the institution from the effect of selection into
that institution. Biological resource centers play a central (but invisible) role in life sciences
research. BRCs collect, certify and distribute biological organisms for use in biological research
and in the development of commercial products in the pharmaceutical, agricultural and
biotechnology industries. BRCs maintain large and varied collections of biological materials,
including cell lines, microorganisms, and DNA material, and distribute tools that allow
researchers to access and exploit these materials. The ability to exploit prior research in the life
sciences depends on access to the cells, cultures, and specimens used in that research. BRCs are
a key institutional arrangements through which scientists can obtain materials for research
purposes. A key objective of our empirical analysis is to evaluate whether the ability to access
research materials through a BRC is associated with enhancing the impact of the scientific
research article that initially described those research materials.
At a broad level, our analysis contributes to understanding the microfoundations of
knowledge accumulation, which plays an important role in leading to ideas-driven growth. More
specifically, our empirical approach extends recent studies using citation analysis to investigate
3
the impact of institutions and technological communities on the cumulativeness of discovery and
innovation [Jaffe, et al, 1993; Griliches, 1998]. We exploit three aspects of our empirical setting
to develop and implement a differences-in-differences estimate of the impact of BRCs on
knowledge spillovers. First, in most cases, each material deposited in a BRC is associated with a
journal article that describes its initial characterization and application. Second, for specific
types of BRC deposits, there is a significant lag between that initial article and its deposit into a
BRC, and, in certain cases, the transfer of the materials was motivated by factors unrelated to the
extent of their use. Specifically, we examine the deposit of materials into BRCs that are
associated with “special collections” that are transferred in toto from smaller collections from
which they had previously circulated into a major BRC, as a result of either the departure of a
principal investigator or in response to institution-wide funding difficulties unrelated to the
culture collection itself. Third, we take advantage of the fact that each of the special collections
we analyze was a unified collection prior to deposit and was transferred as a group. Thus, the
institutional shock is consistent for every material within a collection, and any systematic
differences in the impact of that shock on future knowledge diffusion can be linked to the
characteristics of the materials themselves.
In our empirical analysis, we focus on these “special collections” and evaluate whether
articles associated with BRC deposited materials receive a boost in citations after deposit has
occurred (after accounting for an article-specific fixed effect, and controlling for age and year
fixed effects). In so doing, we are able to separately identify the role of selection (the likelihood
that materials deposited into BRCs are associated with intrinsically important research) from the
marginal impact of BRCs (the impact of BRCs in enhancing diffusion, controlling for the
intrinsic importance of that knowledge). Our approach builds on the considerable advances that
have been made in recent years in evaluating differences-in-differences estimation [Bertrand,
Duflo, Mullainathan, 2004]. Beyond our main specifications, we are able to provide a series of
checks of our identification assumptions that reinforce our overall approach.
Our results provide strong empirical support for both the selection effect and the marginal
impact of BRCs. Even in the period before their materials are accessible through a BRC, those
research articles that are ultimately linked to BRCs experience nearly double the citation rate
compared to a set of control articles drawn from the same journal and published in the same year.
Even if the marginal impact of BRCs were to be zero, this result is important, as it suggest that
