NBER WORKING PAPER SERIES
GRILICHESIAN BREAKTHROUGHS:
INVENTIONS OF METHODS OF INVENTING AND
FIRM ENTRY IN NANOTECHNOLOGY
Michael R. Darby
Lynne G. Zucker
Working Paper 9825
http://www.nber.org/papers/w9825
NATIONAL BUREAU OF ECONOMIC RESEARCH
1050 Massachusetts Avenue
Cambridge, MA 02138
July 2003
This paper was initially prepared for the International Pre-Conference on “R&D, Education and Productivity”
in honor of Zvi Griliches (1930-1999) held at the NBER, Cambridge, MA, July 31, 2002. Zvi organized a
NBER workshop for Zucker and Darby to present their first joint paper on biotechnology and was a constant
source of ideas for improvement as their work progressed. We all stand on the shoulders of giants, and Zvi’s
shoulders are very high. This research has been supported by grants from the University of California's Industry-
University Cooperative Research Program, the University of California Systemwide Biotechnology Research and
Education Program, the Japan Foundation, the National Science Foundation, and the Alfred P. Sloan Foundation
through the NBER Research Program on Industrial Technology and Productivity. James R. Heath, Evelyn Hu,
and Fraser Stoddart have provided continuing guidance as we try to understand nanoscale science and
technology well enough to measure and model its growth and commercialization. Evelyn Hu’s detailed
comments particularly improved the paper, and members of the UCLA Innovation Workshop also provided
useful comments. We are indebted to Luis Arias, Chuling Chen, Rui Wu, and Josh Mason for their work on the
nanotechnology data and analysis. This paper is a part of the NBER's research program in Productivity. Any
opinions expressed are those of the authors and not those of their employers or the National Bureau of Economic
Research.
©2003 by Michael R. Darby and Lynne G. Zucker. 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.
Grilichesian Breakthroughs: Inventions of Methods of Inventing and Firm Entry in Nanotechnology
Michael R. Darby and Lynne G. Zucker
NBER Working Paper No. 9825
July 2003
JEL No. O31, L63, L65, M13, O16, R12
ABSTRACT
Metamorphic progress (productivity growth much faster than average) is often driven by Grilichesian
inventions of methods of inventing. For hybrid seed corn, the enabling invention was double-cross
hybridization yielding highly productive seed corn that was not self-propagating. Biotechnology
stemmed from recombinant DNA. Scanning probe microscopy is a key enabling discovery for
nanotechnology. Nanotech publishing and patenting has grown phenomenally. Over half of nanotech
authors are in the U.S. and 58 percent of those are in ten metropolitan areas. Like biotechnology, we
find that firms enter nanotechnology where and when scientists are publishing breakthrough academic
articles. A high average education level is also important, but the past level of venture-capital activity
in a region is not. Breakthroughs in nanoscale science and engineering appear frequently to be
transferred to industrial application with the active participation of discovering academic scientists. The
need for top scientists' involvement provided important appropriability for biotechnology inventions,
and a similar process appears to have started in nanotechnology.
Michael R. Darby Lynne G. Zucker
Cordner Professor of Money & Financial Markets Professor of Sociology & Policy Studies
Departments of Management, Economics, Director, Center for International Science,
& Policy Studies Technology, and Cultural Policy, SPPSR
Anderson Graduate School of Management Department of Sociology
University of California, Los Angeles University of California, Los Angeles
Los Angeles, CA 90095-1481 Los Angeles, CA 90095-1551
and NBER and NBER
[email protected] zucker@ucla.edu
1
Grilichesian Breakthroughs: Inventions of Methods of Inventing
and Firm Entry in Nanotechnology
by Michael R. Darby and Lynne G. Zucker
Zvi Griliches (1957a, 1957b) was the first economist to study the class of breakthrough
discoveries which he named an “invention of a method of inventing.” His case was hybrid corn, “a
method of breeding superior corn for specific localities. It was not a single invention immediately
adaptable everywhere. Griliches observed that such breakthroughs thus involve a double-
diffusion process: the timing of application of the inventing method to specific potential products
(“availability”) and the speed with which sales of each specific product reach a mature level
(“acceptance”). Griliches related the speed of both processes to their profitability.
This paper demonstrates that Grilichesian breakthroughs have a much wider applicability
and importance than generally believed. In Darby and Zucker (2003), we argue that much if not
most technological change occurs at any given time in the relatively few firms in the relatively
few industries undergoing metamorphic progress. Modern biotechnology provides an excellent
example of metamorphic progress based on an invention of a method of inventing. The invention
in question was recombinant DNA (also known as genetic engineering or gene splicing) which has
been applied since its 1973 invention to a widening range of scientific problems and products and
industries.
1
The newer applications from the same breakthrough method are generally less
profitable than the older applications as firms continue to pick fruit hanging ever higher on the tree
of applications. However, the initial breakthrough often leads to virtuous circles of further scientific
and technological productivity, making follow-on breakthroughs more likely (Zucker and Darby
1995).
2
Nanoscale science and nanotechnology (hereafter nano S&T) represent another such
breakthrough invention of a method of inventing with the potential to drive technological progress.
Nano S&T involves understanding and control of matter and processes at the atomic and molecular
level – building atom by atom (see Section II for a fuller definition).
2
Nano S&T is 15-20 years
younger than biotechnology but so far following a similar growth trajectory. By beginning to study
this technology now we hope to not only test and generalize findings by ourselves and others on the
links between science and technology in biotechnology, semiconductors, lasers and other
technologies, but to capture new elements of the metamorphic process which because of their
ephemeral or ultimately unsuccessful existence cannot be captured in retrospective studies.
Section I of this paper lays out the key elements in Grilichesian breakthroughs that lead to
metamorphic progress in existing and/or new industries; of particular interest is the neglected key
element of the hybrid corn invention which made it profitable to apply to producing particular seed
corns. Section II provides an introduction to the scientific, technological, and commercial aspects
of nanoscale science and technology, with comparison to bioscience and biotechnology. Section III
provides our initial empirical results demonstrating the close linkage between entry of firms into
nanotechnology and the strength of the local science base, suggestive of natural excludability or
other sources of knowledge localization. We lay out our conclusions on the general features of
major Grilichesian breakthroughs and their application to nano S&T in Section IV. A data appendix
presents further detail on data sources and construction beyond that given in the text.
3
I. INVENTIONS OF METHODS OF INVENTING
Before considering the special factors characteristic of an invention of a method of
inventing, it is necessary to review and extend our common understanding of investment in research
and development (R&D), highlighting elements which are crucial to Grilichesian breakthroughs.
I.A. R&D AS PURPOSIVE INVESTMENT
Individuals and firms invest in R&D in part motivated by the expected returns to and
costs of the activity. Returns include above-normal profits for a variable time until entry drives
returns to use of a planned or serendipitous invention to their normal equilibrium level, psychic
returns to the scientists and engineers doing the R&D, consumer surplus accruing to buyers from
the inventor(s) and others enabled to produce by the invention, and producer surplus of
competitors who are able to copy, invent around, or build upon the invention once it is observed.
Inventor(s) or their employer(s) are motivated by only that portion of returns which are
appropriable to them. The costs of an R&D project are those that it takes to discover and/or
apply the invention, prove its safety and effectiveness if necessary, and to move the invention
from proof of principle or prototype to commercially producible.
3
Properly taken, the
expectations operator will allow for covariances among these returns or costs depending on the
success or failure of the project.
Note first that the value of above-normal returns depends heavily – if an invention is
successful enough that they initially exist – on how quickly the invention faces competition from
frank imitators or sophisticated invent-arounders. These same processes occur in basic science
as well as commercial applications. Most of the literature has focused on the formal intellectual
4
property rights (such as patents, copyrights, and trade secrets) available to protect individual
inventions. While patents are extraordinarily important for protecting particular biotech
products, natural excludability provided important informal protection by slowing the diffusion
of the requisite knowledge (Zucker and Darby 1996, Zucker, Darby, and Brewer 1998, Darby
and Zucker 2003). Organizational boundaries and other social structure also served to limit the
rate of diffusion (Zucker, Darby, Brewer, and Peng 1996).
Natural excludability arises from the fact that prior to adequate codification or
incorporation into a commercially available instrument, cutting edge discoveries frequently
involve extensive tacit knowledge that is embodied initially only in the discoverers and passed
on by “learning by doing with” at the bench science level. Jensen and Thursby (2001) report that
university technology transfer officers estimate that 70 percent of all university inventions cannot
be licensed unless the inventors are willing to cooperate with the licensee in developing and
transferring the technology. Mowery and Ziedonis (2001) show that market channels of
university-to-firm technology transfer – particularly, exclusively licensed patents – are much
more localized around the university than are non-market spillovers as evidenced by citations to
university patents in other patents. Because the ongoing, active participation of inventing
scientists in commercialization of most inchoate discoveries is necessary for the effort to
succeed, we observe new high-tech industries growing up around universities and on commuting
routes of leading scientists.
Ultimately codification or instrumentation may replace learning by doing with, but the
returns to codifying processes such as textbook writing are low for some time relative to the
commercial and scientific returns from applying very valuable breakthrough know-how directly
(Zucker, Darby, and Armstrong 2002). Thus, for a significant period of time, typically measured
5
in years, the top discovering scientists can realize extraordinary returns to commercializing their
discoveries in the correct expectation that no competitor will enter to drive profits toward their
competitive levels. Among the small number sharing the valuable knowledge the incentive is to
find unique areas of application so long as the best next application has higher monopoly profits
than the expected oligopoly profits from inventing a close substitute to any of the products
commercialized by one of the other knowledgeable teams.
For individual scientists and engineers, the expected returns will include their psychic
returns to doing good science and the pecuniary value that goes with enhanced professional
reputation. This is especially true when we think of one of the entrepreneurial firms including
one or more top university scientists among the founding principals. More generally, firms
which permit or encourage scientists to follow their scientific tastes may well benefit from
compensating wage differentials so that rational firm managers too may be influenced by the
scientists’ returns from doing good science (Zucker and Darby 1997). Zucker, Darby, and
Torero (2002) find that star scientists at universities are more likely to begin working with firms
after observing that other stars in the same region have increased publication and citation rates
while working with firms. Firms that encourage creativity and scientific goals among their
scientists arguably also are more likely to profit from scientists and engineers making and
following up on serendipitous inventions. Darby and Zucker (2002) show that entrant firms with
the deepest involvement of star scientists go public sooner and at a higher price.
I.B. WAVES OF INVENTION AS A RESULT OF GRILICHESIAN BREAKTHROUGHS
Schmookler (1966 and 1972) argued that technological progress in general and invention
in particular are driven by the demand side – giving theoretical and empirical backing to the old
6
chestnut, “Necessity is the mother of invention.” More recently, the literature has emphasized
technological opportunity and appropriability of returns as the two major factors varying across
industries and explaining most variation in R&D intensity (see Klevorick, Levin, Nelson, and
Winter 1995). An invention of a method of inventing creates technological opportunity,
appropriability, or both across a wide range of potential products.
Platform Technologies.
Consider a platform technology like Sun Microsystems’s
Java
TM
which Sun makes freely available to software developers. Because software applications
written in Java
TM
can run on any platform, Java
TM
creates a technological opportunity to write
one program and satisfy many markets. However, this universality also reduces the
appropriability of returns to inventing a new program since it can no longer be written for a
particular niche and be protected from those unfamiliar with the peculiarities of that program.
Java
TM
has diffused but no one except for Sun Microsystems which sells complementary products
has made exceptional profits from it.
4
Griliches’s Case of Hybrid Seed Corn. As noted earlier, Griliches (1957b) examined
the double diffusion process: the order in which the many hybrid seed corns were introduced and
both the eventual market share of each type of seed corn and how rapidly that share was reached.
Griliches looked at hybrid seeds for different growing regions as different products as the seed
that had the highest expected output in one growing area might be inferior in another area. Thus
different seeds had to be developed for each and localization of demand should not be confused
with the localization of production and especially of R&D to create the inventions for each
product. If all diabetics lived in Chicago, we would expect Genentech’s human insulin to be sold
exclusively in that region while anti-tubercular agents might be sold only in Arizona. There
would be little to remark on if a California company chose to invent and market those products
7
in Chicago and Arizona, respectively. Similarly, hybrid seed corn companies would logically try
to develop winning varieties for the biggest corn markets regardless of where the seed company
happened to be headquartered.
For years the authors have taught the classic Griliches (1958b) article, understanding that
the founding invention was a method which produced superior hybrid corn compared to what
could be produced by standard selection and propagation of high yield plants. As far as we can
tell from footnote 3 in Griliches (1957b) and from his other writings on hybrid corn (1957a,
1958, 1960a, 1960b, 1980) that is what he understood to be the facts.
Curiously, Griliches’s case appears to be a rare one in which the invention of a method of
inventing was valuable because the method produced less valuable products than were possible
by previous methods but products with a deficiency that increased appropriability sufficiently
that private firms were willing to invest in the large, multiyear R&D projects required to develop
each product.
The precise invention is made clear in the original treatise by Edward East and Donald
Jones (1919, pp. 221-224):
5
You start with four different inbred varieties of corn A, B, C, and
D. These strains themselves are not suitable for sale as seed because the inbreeding process
reduces their vigor while ensuring certain characteristics valuable under particular growing
conditions. On average, the first generation of each of the possible crosses is better than either
parent and, more importantly, equal or better in yield but still inferior in resistance to the open-
pollination varieties then commonly used by farmers. The trick comes in crossing two of these
first generation hybrids, for example AxB with CxD. Depending on the particular inbred lines
represented by A, B, C, and D, the resulting first generation double-cross hybrid can predictably
deliver hardier plants with substantially higher yield than any of the inbred or open-pollination
8
varieties in the public domain. Obviously, with dozens of inbred lines available and the
opportunity to selectively breed more, the multi-year experimentation involved to find the best
choices of A, B, C, and D for a particular growing region is a costly R&D investment. Not
surprisingly, by 1994 nearly 94 percent of researchers with doctorates breeding new corn lines
were in industry and only 6 percent in academe or elsewhere in the public sector: “there are
more Ph.D. corn breeders in any of the large corn breeding companies than all the public corn
breeders combined in the U.S. and Canada.” (Kannenberg 1999)
Lewontin (1993, pp. 52-57) argues that what was important was that the double-cross
hybrid method produced a superior seed corn which itself did not breed true. If a farmer were
to save some of his crop as seed for next year, he would have a mixture of inferior and high yield
strains yielding far less per acre than the purchased seed. He/she would be willing to pay up to
the value of the difference in productivity to obtain new pure hybrid seed corn from the growers.
Thus the essential value of East and Jones’s discoveries 1918-1924 is this: Commercial
R&D to develop hybrid seed corn now could be profitable since the seed would not be in direct
competition with crops of prior years’ purchasers. Lewontin (1992, p. 55) quotes the inventors
to show their self-awareness of this property:
[Double-cross hybridization is] something that might easily be taken up by the
seedsmen; in fact, it is the first time in agricultural history that a seedsman is
enabled to gain the full benefit from a desirable origination of his own or something
that he has purchased. The man who originates devices to open our boxes of shoe
polish or to autograph our camera negatives, is able to patent his product and gain
the full reward for his inventiveness. The man who originates a new plant which
may be of incalculable benefit to the whole country gets nothing – not even fame –
9
for his pains, as the plants can be propagated by anyone. There is correspondingly
less incentive for the production of improved types. The utilization of the first
generation hybrids enables the originator to keep the parental types and give out
only the crossed seeds, which are less valuable for continued propagation.
- East and Jones (1919, p. 224, italicized portion omitted by Lewontin)
The inventors of double-cross hybrids seem to understand the positive social value of
appropriability better than Lewontin.
The scientist-entrepreneur (later politician) Henry A. Wallace perceived the opportunity
and established the first hybrid-seed-corn company Hi-Bred Corn Company (later Pioneer Hi-
Bred) in 1926.
6
Experimentally trying each possible combination of two pairs from 30, 40, or
more inbred lines and then testing each combination in each growing region would be an
extraordinarily expensive proposition. The best corn breeders had a sufficient implicit model of
what would work to order the pairs tried well enough that the cost of experimentation was low
enough for seed companies to prosper. Purposive entry of competitors is possible, but only if the
price for a particular seed corn is high enough to encourage duplicative invention knowing that
the duopoly price will be lower than the monopoly price (Gilbert and Newbery 1982).
Biotechnology. The founding discovery for biotechnology occurred in 1973 when
Stanley Cohen (of Stanford) and Herbert Boyer (of UCSF) realized that by combining their
methods they could cut a desirable gene from one organism and insert into another organism to
create a new organism which could reproduce itself with the new desired trait.
7
Herbert Boyer –
with the support of venture capitalist co-founder Robert Swanson – led Genentech to the
successful production in 1982 of human insulin from genetically engineered bacteria. Genetic
engineering created not only a vast opening of technological opportunities across a range of
10
products and industries, but also came with natural excludability. Scientists in the field agree
that – like playing the violin – you learned genetic engineering by working with someone who
knew how to do it. Even today new authors on an article reporting a new genetic sequence
discovery are normally coauthors of other scientists who have already appeared on such an
article (Zucker, Darby, and Torero 2002). In the early days, that meant that the essential input
for the technology were one or more of the people who could do it. When the U.S. Supreme
Court in 1980 approved patenting of life forms created by genetic engineering, appropriability
was further enhanced.
I.C. ESSENTIAL ELEMENTS OF SUCCESFUL GRILICHESIAN BREAKTHROUGHS
We believe it is possible to identify a successful invention of a method of inventing before
it is proven over time by examining whether if creates appropriable technological opportunity across
a broad range of products. The double crossing of pure inbred corn lines created the private
incentive to develop new seed lines because the corn grown from the seed was markedly inferior in
output if used as a seed for a second generation of corn. The major biotech firms nearly all had as
founders or initial key scientists top “star” discovering scientists who had access to the scarce tacit
knowledge needed to take advantage of the new technological opportunities.
11
II. NANOTECHNOLOGY: A VERY BIG, VERY SMALL THING
The U.S. government has identified nano S&T as a scientific and technological
opportunity of immense potential, formally launching a National Nanotechnology Initiative
(NNI) in January 2000. It is extremely difficult to define simply the full range of nano S&T, but
the NNI’s steering committee settled on the following definition of nanotechnology:
Research and technology development at the atomic, molecular or macromolecular
levels, in the length scale of approximately 1 - 100 nanometer range, to provide a
fundamental understanding of phenomena and materials at the nanoscale and to create
and use structures, devices and systems that have novel properties and functions because
of their small and/or intermediate size. The novel and differentiating properties and
functions are developed at a critical length scale of matter typically under 100 nm.
Nanotechnology research and development includes manipulation under control of the
nanoscale structures and their integration into larger material components, systems and
architectures. Within these larger scale assemblies, the control and construction of their
structures and components remains at the nanometer scale. In some particular cases, the
critical length scale for novel properties and phenomena may be under 1 nm (e.g.,
manipulation of atoms at ~0.1 nm) or be larger than 100 nm (e.g., nanoparticle reinforced
polymers have the unique feature at ~ 200-300 nm as a function of the local bridges or
bonds between the nano particles and the polymer).
8
Roco, Williams, and Alivisatos (1999), Roco (2001), and Roco and Brainbridge (2001) provide a
thorough review of the present state of nano S&T, the implementation of the NNI, and an
introduction to thinking about the implications of nano S&T for our economy and society.
12
Nano S&T has been a burgeoning area of science and engineering since at least 1990 as
illustrated in Figure 1. This figure reports data which we obtained by searching on the topic
“nano*” by year in the Science Citation Index Expanded (SCI-EXPANDED)--1975-May 30,
2003. The values for 1981 through 1989 (averaging one third article per thousand) reflect some
of the substantial early scientific work as well as the background error in the search strategy.
These values show no trend and none are significantly different from their mean. Since 1990 the
growth in nano S&T articles has been remarkable, and now exceeds 2 percent of all science and
engineering articles.
9
The patent data suggest a beginning date for nano S&T some five years earlier than 1990.
Figure 2 presents data on patents granted by the end of 2000 containing the string “nano” in their
title or description. The data is presented for two different dating conventions: by year the patent
was granted and by the year the patent was applied for. The latter is more precise in terms of
when the invention was actually made (typically about 3 months before the date of application),
but suffers for the last five or six years from right truncation: Some patents applied for before
1996 were still pending at the end of 2000; as were many of those applied for in 1997-2000 and
nearly all applied for in 2000. If we allow for a lag between application and grant dates (25
months was the difference in means in this sample), both series tell the same story. Patents
applied for 1976-1985 and granted 1976-1986 had mean values of 37.4 and 33.3 patents per year
and no observation within these periods differed from the respective means by as much as two
standard deviations. Patent growth takes off in 1986: No number of patents applied for after
1985 (excluding the severely truncated 2000 observation) nor granted after 1986 were within two
standard deviations of the cited means.
13
II.A. THE ENABLING INVENTIONS
At this early date it is speculative to identify the key invention or inventions enabling
rapid burgeoning of nanotechnology. Indeed, it may be that a confluence of breakthroughs will
ultimately be seen as playing this role.
10
At this point, we can say that one key enabling
invention creating technological opportunity in nano S&T is the development in the 1980s (with
continuing improvements through the present) of scanning or proximal probe microscopy: “Ask
a dozen surface scientists to identify key developments in instrumentation that are responsible
for catapulting nanotechnology to the front lines of physical science research. Nearly all of them
will point to the advent of scanning probe microscopy.” (Jacoby 2001)
The scanning tunneling microscope (STM) was the first instrument to enable scientists to
obtain atomic-scale images and ultimately to manipulate individual atoms on the surfaces of
materials. It was invented in 1981 at IBM's Zurich Research Laboratory and reported by the
inventors Gerd Karl Binnig and Heinrich Rohrer (1982 and 1983); they received the Nobel Prize
in Physics in 1986 for their STM work. The STM works by moving a very fine pointer back and
forth over a surface with each scan line displaced slightly from the next. A sensitive feedback
mechanism maintains a constant distance relative to the surface so that a three dimensional
representation is obtained. The procedure is called raster scanning in reference to the parallel
lines which make up a television picture. The STM could be used only on conductive materials
(metals) due to the electron tunneling method used to maintain the constant distance between
pointer and surface.
The atomic force microscope (AFM) which was invented five years later by Binnig,
Calvin Quate, and Christoph Gerber (1986) greatly broadened the range of materials which could
be viewed at the atomic scale and enhanced the ability to manipulate individual atoms and
14
molecules. Haberle, Horber, and Binnig (1991) report a modified AFM for use on living cells
with which they observed the effects of antibody attachment and changes in salinity on living red
blood cells. The invention of the AFM triggered an explosion in the microscopy of surfaces that
continues through the present.
11
We note that the inventions of methods or inventing by East and Jones, Cohen and Boyer,
and Binnig, Rohrer, Quate, and Gerber are literally methodologies or instruments, not theoretical
revolutions or paradigm shifts in response to nagging anomalies as argued by Kuhn (1962). The
theoretical changes follow as the range of experimentation is broadened. Table 2 reports citation
counts as of mid-July 2002 to the most-highly cited (250 or more citations) of the inventors’
STM and AFM papers.
STMs and AFMs were initially only available to the few scientists with the resources and
ability to construct one. Digital Instruments was the first of a number of firms to construct a
commercially successful STM, shipping the first units in 1987. They introduced the first
commercial AFM in 1989. The commercial availability of proximal probe devices facilitated if
not enabled the rapid diffusion of nanoscale research. The usefulness of proximal probe
microscopy was also greatly advanced in the late 1980s and 1990s as the rapid fall in the price of
computing power has permitted multiple scans to be integrated into a three dimensional picture
of a material that can be rotated and manipulated.
In conclusion, the history of breakthroughs in scanning probe microscopy as key enabling
inventions for nano S&T is consistent with the 1985-1990 initiation period which we detected by
significant increases in the rate of patenting and publication.
15
II.B. CONCENTRATION OF NANOSCALE SCIENCE & TECHNOLOGY
As is frequently observed in other instances of metamorphic progress, nano S&T has
been highly concentrated in a few countries and a few regions in those countries. Figure 3 and
Panel A of Table 1 illustrate that for the ISI’s High Impact nano articles, 54 percent of the
authors’ addresses have been in the United States with another 29 percent divided among
Australia, France, Germany, Japan, Switzerland, and the United Kingdom.
12
Within the United States, nano S&T publishing is similarly concentrated as shown by
Figure 4 and Panel B of Table 1. The region including Los Angeles and Santa Barbara has the
most authors’ addresses on articles with at least one top-112 research university author. Silicon
Valley and Boston follow close behind. When we look at things in terms of articles per million
population, outstanding universities in relatively less urbanized regions make the areas around
the University of Illinois at Champagne-Urbana and the North Carolina Research Triangle look
outstanding not only within the U.S. but also for the 1991-1998 high-impact nano articles which
are comparable to the last column of Panel A.
We expect that as the nanotechnology industry grows, we will also find that growth
within the industry is concentrated in relatively few, very rapidly growing firms as is
characteristic of metamorphic progress (Darby and Zucker 2003).
II.C. COMPARISON OF NANO S&T WITH BIOTECHNOLOGY
Given the remarkable transformations due to biotechnology in treatment and diagnosis of
disease and even the flowers we give, it might seem foolhardy to suggest that nano S&T is
following a similar trajectory. However, just as biotechnology covers all living things, nano
S&T covers the atomic and molecular level for all matter, organic or inorganic. The intersection
of biotech and nano S&T is certainly a large and important area, but even more is going on with
16
respect to inorganic science and technology. The distribution of nano publishing is illustrated in
Figure 5 which shows the percentage distribution for each semiannual volume of the Virtual
Journal of Nanotechnology.
13
Figure 6 illustrates the remarkable increase in publishing and patenting that occurred
during the first twenty years of the biotechnology revolution and that is occurring now in nano
S&T.
For articles, nano S&T is maintaining a growing lead over biotechnology articles. Recall
that we so far identify nano articles as simply those that include the string “nano” in the ISI topic
search (in the case of our aggregate data) or the article title (for our ISI flat files discussed
below). We are working actively with our colleagues at the California NanoSystems Institute
(CNSI) to improve the search methodology for identifying both articles and patents, aiming for
defined subareas via new keywords. Defining biotech articles as any that report a genetic
sequence (i.e., appear in GenBank) is also conceptually overly narrow, but it has been proven in
practice a very useful measure in our work on biotech.
Nano S&T patents were ahead of biotech patents early in the process because practically
none were issued in biotech until the courts gave the go ahead in 1980. Thirteen years into the
biotech revolution (1986), biotech patenting took off as gene sequences were patented with little
proof of their use and many variations on drug candidates were patented in an attempt to prevent
quick competition from me-too drugs if one particular candidate were proved safe and effective.
Taken as a whole the scientific and patenting growth of nanotechnology is of the same
order of magnitude as biotechnology at a similar state of development.
17
III. EMPIRICAL RESULTS ON ENTRY OF FIRMS INTO NANOTECHNOLOGY
Zucker, Darby, and Brewer (1998), Darby and Zucker (2001), and Torero, Darby, and
Zucker (2001) have shown that entry into high-tech industries undergoing metamorphic progress
(U.S. and Japan biotech and U.S. semiconductors) can be explained well in terms of a poisson
process: The probability of entry per unit time in a particular region depends upon the size of the
academic science base – especially of active, star researchers – and the local economic climate as
represented by the past level of venture capital funding, employment, and average wage per job
(which serves as an indicator of the quality of the labor force). We attempt here to follow that
strategy in explaining where and when firms enter nanotechnology.
In our previous work, we have stressed the importance of natural excludability as
reinforcing formal intellectual property protection and ensuring that scientists with the requisite
know-how were the scarce resources around which firms were formed or transformed. With the
enabling technology now commercially accessible, it could be that natural excludability is less
operative in nano S&T. However, these instruments are being put to such novel uses and with
sufficient tacit knowledge that the inventions are not likely to be licensed without the inventors’
hands on involvement in making them work commercially. Where commercial opportunity is
built on fast-advancing academic science it is generally more economical to establish
commercial laboratories and even manufacturing facilities near the universities than to try to
move the scientists and their network to an existing firm location.
This section is organized in four subsections. The first describes the data being used.
The second presents the empirical results for entry into nanotechnology by all firms regardless of
industry of application or science base. The third subsection presents a novel analysis relating
18
entry of firms to the region’s specific science base relevant to the applications pursued by those
firms. The final subsection summarizes the results.
III.A. DATA
Following the approach of Zucker, Darby, and Brewer (1998), our variables are all
defined for the years 1981-1999 for each of 172 distinct regions comprising every county of the
United States. These regions correspond to the U.S. Bureau of Economic Analysis’s Functional
Economic Areas based on commuting and shopping patterns and described in Johnson (1995).
Since the earliest evidence of increased nano S&T activity occurred in 1985 and 1999 is the last
year for which we have data for all the variables, our regressions are estimated for 1985-1999
with 1981-1984 values used to get good starting values for the various knowledge stock
measures.
14
These regions are the reporting units for the BEA’s Regional Accounts Data. We
downloaded employment and average wages for each region from
http://www.bea.doc.gov/bea/regional/reis/ on July 15, 2002, but this data is available on the
BEA’s Regional Economic Information System and numerous other sources. Table 3
summarizes the definitions of all the variables used in the study. The sample statistics for the
variables are given in Table 4.
Our major source of information for the regional nanoscale specific science base is the
Institute of Scientific Information (2000a) database of all publications with at least one author
affiliated to one of the top-112 research universities in the U.S. The ISI list of the top-112
research universities is included as Table Al. For this paper, we identified nano S&T articles by
searching for the string “nano” in their title.
15
We used computer-based and hand matching to
divide this base into those articles with at least one author affiliated with a firm and the
19
remaining all-academic articles. We followed the same procedure with the Institute of Scientific
Information (2000b) database of High-Impact (highly cited) papers.
We used the top-112 universities and high-impact academic papers separately to define
measures of the science base by region and year. First we coded the specific region for each of
the addresses given for authors on the paper. We then credited a fractional part of the paper to
each region represented in the addresses equally.
16
Thus for each region and year we have
separate counts for academic top-112 articles and academic high-impact articles. These two
counts are accumulated year by year with a 20% depreciation rate being applied to the
knowledge stock of the previous year. We take the starting value in 1980 (before the invention
of the STM) to be zero for all regions. We hypothesize that high-impact papers have a higher
average quality level than the average top-112 paper and hence expect them to have a larger
impact on firm birth.
We plan in future years to develop a database with extensive data on nanotech firms
comparable to the one which we have developed for biotechnology. This task of integrating
directory, archival, web-search, and survey data will take years and, meanwhile, there are no
good alternative databases existing. Fortunately, for science-driven technologies essentially all
significant firms encourage their scientists and engineers to publish their significant research
results in academic journals – delaying publication only for the few months required to file for
patent protection. As a result, we can use the rest of the ISI data – the top-112 and high-impact
articles with firm authors – to identify firms active in nanotechnology and proxy for their date of
entry.
Specifically we code each of these articles as to where the firm authors are located and
count the firm as entering nanotechnology in a given region in the year of their first publication
20
in a that region in the firm-article dataset. Thus, one firm may have multiple entry dates if it has
employees authoring articles in different regions. This methodology identifies 202 firm entries
in all, with 187 of them occurring 1991-1999. Over 68 percent of the 1991-1999 firm entries
occurred in the 10 regions for which entries by year are illustrated in Figure 7. Note the
unevenness across regions of the upward trend in firm entry.
In subsection III.C below we use a modified definition of firm entry in which a firm is
identified as entering a specific area of nanotechnology in a given region in the year of their first
publication in a that region in a firm-article sub-dataset restricted to articles in a relevant science-
area grouping. Thus – on this special definition – one firm may have multiple entry dates if it
has employees authoring articles in different regions and/or has employees authoring articles in
different major science-technology areas in the same region, but there will be at most one entry
date per firm per region for any one of the major science-technology areas. The five major
science-technology areas used are listed in Table A2 as defined in the Darby and Zucker (1999)
monograph comparing California’s science base with that in other hightech states. The
distribution of firm entry across these fields is reported in Figure 8 for the same 10 regions as in
Figure 7. Note that the proportional distribution of total entries by field is not even across
regions, consistent with the hypothesis that differences in relative strengths of the science bases
will cause uneven distribution of firm entry.
We have two non-article based measures of regional science base. The first is based on
the NSF data on federal funding received by the top-100 universities in 2000 and tracking them
back to 1981 (see Data Appendix for details). This funding is added up by region and year but
can not be broken down further into the five major science-technology areas. The second
measure is based on the 1993 National Research Council (NRC) study of U.S. research-doctorate
21
programs (Goldberger, Maher, and Flattau 1995). We count the number of doctoral programs in
science and engineering which are ranked in the top 10 (including all universities tied for 10
th
place).
17
We counted these highly ranked programs by region and used the same regional value
for each year 1985-1999. We believe that given the current state of nanotechnology data it is
acceptable to take the 1993 values as exogenous regional constants. For subsection III.C, we
made these counts separately based only on rankings for doctoral programs in the same five
broad science-technology categories listed in Table A2.
III.B. EMPIRICAL RESULTS FOR ALL SCIENCE-TECHNOLOGY AREAS COMBINED
Poisson regressions are appropriate as here when dealing with count variables with
numerous zero values (Hausman, Hall, and Griliches 1984). The intuition of the model is that there
are thousands of individuals and who might start a nanotech firm or subunit but the probability that
any one of them will do so in a short period of time is very small. The time and region varying
probability λ is what is estimated as a function of the independent variables. All the regressions
reported here were estimated using LIMDEP 7.0, with the Wooldridge (1991) regression-based
correction for the variance-covariance matrix estimates.
18
Table 5 reports the poisson regressions for firm entry. Column (a) and (b), respectively,
report separate regressions for the science-and-engineering-base and regional economic variables
with the venture capital variables excluded since it is problematic whether they should be viewed
as reflecting the science base or the regional economic base. Column (c) reports the regression
with all the variables in (a) and (b).
In column (a) we see that the high-impact articles dominate the (insignificant) top-112
university articles measure. This implies that – as with biotechnology – it is primarily the top
science and scientists who spawn firms locally. Federal research funding is also significant and
22
can be considered as either an input measure or, given the peer review involved, a supplementary
quality measure. The NRC survey results are positive but not significant.
19
Column (b) shows
that average wage as a measure of work-force skill levels is positive and significant while
employment as a measure of size is positive but not significant. In the combined regression (c)
Federal research funding becomes insignificant and only the region’s high-impact articles and
skill level appear significant.
Measures of the frequency and size of recent venture capital deals in the region are added
to these regressions in columns (d), (e), and (f). The venture capital variables generally are
insignificant – although they enter inconsistently in regression (e). Comparing columns (d)-(f)
with (a)-(c), we see that neither do the venture capital variables have much effect on the other
significant variables in these regressions: The only exception being that regional size becomes
significantly positive in regression (e) with the science-and-engineering base excluded.
The overall message of Table 5 then is that firms enter nanotechnology near where top
scientists are making breakthrough discoveries and where skill levels in the work force are high.
This does not exclude the possibility that federal research funding, highly-rated university
doctoral publishing, and top-112-universities articles play a role – indeed each of these enter
significantly positively in Poisson regressions (not reported here) which exclude the other 3
science-base measures. It does suggest that these variables play a role to the extent that they
promote breakthrough discoveries as indicated by the high-impact article measure.
As we have emphasized above and in Darby and Zucker (2003), metamorphic progress is
highly concentrated. In order to ensure that we are not capturing simply the concentration of
entry in a relatively few regions, we report fixed-effects Poisson regressions in Table 6
analogous to those in Table 5, except that we estimate a different constant for each region, so
23
that the remaining coefficients provide estimates based on deviations from average for each
region by year.
20
The results in Table 6 are remarkably similar to those in Table 5, with the
main notable difference being that – with the means taken care of – we do see that fast growing
regions (as measured by employment) do have significantly more entry than slow growing
regions.
III.C. EMPIRICAL RESULTS FOR SPECIFIC SCIENCE-TECHNOLOGY AREAS
We would have liked to estimate separate poisson regressions for entry by firms into each
of the five science-and-technology areas listed in Table A2. Unfortunately, as indicated by a
close inspection of either Figure 8 or Table 4, most of these areas have so far had too little entry
for these regressions to be estimable – while there are still 2580 observations, nearly all the left-
hand variables are zeroes. Put another way, so far about 62 percent of entries (counted by region
and science-and-technology areas) have occurred in the area of semiconductors, integrated
circuits, and superconductors, and another 24 percent of these entries are accounted for by the
biology, medicine, and chemistry area.
Tables 7 and 8 shows that entry of firms working in the broad area of semiconductors,
integrated circuits, and superconductors, are explained by essentially the same processes as
estimated in Tables 5 and 6 for all entries combined. This similarity is hardly surprising given
the dominance of this science-and-technology area in the initial applications of nanotechnology.
However, it is reassuring that the science base measured specifically in terms of this particular
subset of articles and doctoral programs (but not research funding) gives the same answers as
when the combined variables for all areas of science and engineering are used.
Tables 9 and 10 report the same set of regressions where the independent and dependent
variables are defined in terms of the area of biology, medicine, and chemistry. The results are
24
clearly fragile with relatively few entries through 1999 in bio-nanotechnologies. Federal
research funding and average wage seem significant in Table 9 when the science base or regional
economic variables are estimated separately, but only the average wages is significant in the full
Poisson regression (f). Even that significance disappears in the fixed-effects Poisson regressions.
There were no reportable results for the other three areas. We conclude that there has simply not
been enough time as of yet to estimate separate equations by science-and-technology areas.
III.D. SUMMARY OF THE EMPIRICAL EVIDENCE
Based on the robust results for entry for all science-and-technology areas combined, we
conclude that nanotechnology is following the same pattern as biotechnology: Firms are
entering where and when academic scientists are making major scientific breakthroughs as well
as where the quality of the labor force is high and, perhaps, growing rapidly. Other measures of
the science base so far appear to be better viewed as increasing the probability of a breakthrough
discovery rather than having a detectible independent result over and above that outcome.
Venture capital and capitalists appear to flow to the technological opportunity rather than
exerting an independent effect on where and when growth occurs, given that all the regions
considered here are tied into the American financial system.
25
IV. SUMMARY AND CONCLUSIONS
We have argued that Grilichesian breakthroughs or inventions of methods of inventing
occur when a scientific or engineering breakthrough creates appropriable technological
opportunity across a broad range of distinct products. Most often these revolutionary inventions
are in the less glamorous areas of methodology and instrumentation rather than high-theory
paradigm shifts. We have seen that in Griliches’s own case of hybrid seed corn the breakthrough
was primarily on the appropriability side of the equation since other means of producing higher
yield corn were available but could not yield a return to investing in their creation because of
competition from saving part of the crop as seed.
Biotechnology combined formal protection on individual products (from 1980) with strong
informal protection due to the high degree of natural excludability inherent in genetic engineering.
This ensured an extended period of above-normal profits to those scientists who became involved in
commercializing their discoveries. It still does so today to those at the frontiers of that rapidly
expanding science, and provides incentives to the top scientists to move part of their labor to firms,
a necessary part of the technology transfer process.
Nanotechnology opens a new range of possibilities to operate, control, and build at the
atomic scale where unique processes and properties are available. The key enabling inventions
classed as scanning probe microscopes became commercially available in a matter of five years or
less after their invention; so there is relatively little natural excludability inherent in the instruments.
However, the uses that these instruments are being put involves areas which are very imperfectly
understood and other new methods of inventing may be at least as important as scanning probe
microscopy. Thus the involvement of the inventing scientists may be very valuable in
26
commercialization, and we note that many of the best nano scientists and engineers maintain their
academic positions and research programs while co-founding and guiding new entrants or
continuing relationships collaborating at the bench-science level with scientists from incumbent
firms.
At this point in our empirical work, our conclusions have been supported through a variety
of empirical measures and methods for biotechnology. In this paper we have presented evidence of
a close coincidence in time and place of firms engaging in nanotechnology and the location of
academic scientists and engineers making major breakthroughs. As our research on the
commercialization of nanoscale science and technology continues we hope to provide further
evidence on the major features of the two-way flows between academe and commerce and new
insights based on more nearly real time data collection than is the norm for research on science,
technology, and growth.
27
Data Appendix
This appendix supplements the descriptions of the data set in the text and Tables 3 and 4.
All current dollar amounts are converted to 1996 dollars by deflating by the chain-type price index
for personal consumption expenditures (PCECPI, U.S. Bureau of Economic Analysis 2002)
Articles as Measure of Science Base
Journal publishing is at the heart of the scientific enterprise and has been successfully used
to measure the local science base in many past studies. We use two partially overlapping databases
licensed from the Institute of Scientific Information (ISI, 2000a and 2000b). As described in the
text and Table 3. Appendix Table A1 lists the 112 universities identified by ISI as top research
institutions on the basis of receipt of federal research funding. ISI (2000a) lists all articles which
have at least 1 author listing an address at one of these 112 institutions. The ISI (2000b) CD lists all
articles, regardless of where authored, that meet the ISI threshold for most highly cited. We have
divided these two data bases into those in which any firm is listed as an author’s address and all
others which we term purely academic. We assign these articles to regions using equal fractional
weights for each listed address so that the total weights for each article equal 1. The articles are
dated by the year of their publication. We calculate knowledge stocks by region and year (see
below) for all purely academic articles for both databases.
21
We also disaggregate these articles into five broad science and engineering areas for which
we can make meaningful comparison across a variety of data sources. Table A2 lists these areas and
corresponding ISI journal fields. Table A.3 reports on the correspondence of these broad areas to
the standard program names used in the NRC study of U.S. doctoral programs (Goldberger, Maher,
and Flattau 1995, National Research Council 1995). This correspondence permits us to calculate
area specific breakdowns of top-10 departments.
28
Knowledge Stocks
Research and development is a multiyear process and inputs impact output measures over a
number of years although the impact decreases over time. Therefore various inputs are measured as
a knowledge stock:
X
it
= A
it
+ (1 – δ) X
i,t-1
where A
it
represents the current input for a given region i and year t, and δ is the depreciation rate of
knowledge. We assume a conventional value for δ of 0.2; see Griliches (1990).
Research Funding
Our data for university receipts of federal research funding is due to the U.S. National
Science Foundation (NSF). We limit the series to the 100 institutions reported for 1993-2000 in
NSF (2000, Table B-6). Our sources for funding for these institutions before 1993 are 1991-
1992 NSF (1997, Table B-6), 1988-1990 NSF (1990, pages 18-19), and 1980-1987 NSF (1987,
pages 15-16) with the following exceptions:
1981-1983 in NSF, (1981, 1982, 1983, Table C-16 for each year) for Mississippi State
University, University of South Florida, Medical University of South Carolina, Thomas
Jefferson University, Medical College of Wisconsin.
1984-1989 in NSF (1989, Table B-6) for Thomas Jefferson University.
1989 in NSF (1989, Table C-16) and 1990 in NSF (1997, Table B-6) for Medical
University of South Carolina.
We coded which region each university was in, aggregated by region and year, and deflated by the
PCECPI to obtain research funding in millions of 1996 dollars by region and year. The funding
data could not be disaggregated into the five major science-and-technology areas.
29
Venture Capital
Venture capital data was obtained by licensing the Venture Economics database, geocoding
recipient firms by region and then aggregating number of deals and funds received by region and
year. Since the venture capital funds are not specific to nanotechnology, we use the same methods
as used to calculate knowledge stocks to calculate venture-capital deals and funding stocks as a
general measure of past local availability of venture capital. These data indicate the general level of
venture-capital activity in a region and we did not attempt to disaggregate these measures into the
five major science-and-technology areas.
30
REFERENCES
Binnig, Gerd Karl, Calvin F. Quate, and Christoph Gerber, “Atomic Force Microscope,”
Physical Review Letters, March 3, 1986, 56(9): 930-933.
Binnig, Gerd Karl, and Heinrich Rohrer, “Scanning Tunneling Microscopy,” Helvetica Physica
Acta, 1982, 55(6): 726-735.
Binnig, Gerd Karl, and Heinrich Rohrer, “Scanning Tunneling Microscopy,” Surface Science,
1983, 126(1-3): 236-244.
Binnig, Gerd Karl, and Heinrich Rohrer, “In touch with atoms,” Reviews of Modern Physics,
March 1999 Special Issue, 71(2): S324-S330.
Cohen, Stanley N., and Herbert W. Boyer, “Process for producing biologically functional
molecular chimeras,” United States Patent 4,237,224, December 2, 1980.
Cohen, Stanley, A. Chang, Herbert Boyer, and R. Helling, “Construction of Biologically
Functional Bacterial Plasmids in vitro,” Proceedings of the National Academy of
Sciences, November 1973, 70
(11): 3240-3244.
Darby, Michael R., and Lynne G. Zucker, California's Science Base: Size, Quality and
Productivity, A Report to the California Council on Science and Technology, February 3,
1999, Sacramento, CA: California Council on Science and Technology, 1999.
Darby, Michael R., and Lynne G. Zucker, "Change or Die: The Adoption of Biotechnology in
the Japanese and U.S. Pharmaceutical Industries," Comparative Studies of Technological
Evolution, 2001, 7: 85-125.
Darby, Michael R., and Lynne G. Zucker, “Going Public When You Can in Biotechnology,”
National Bureau of Economic Research Working Paper No. 8954, May 2002.
31
Darby, Michael R., and Lynne G. Zucker, “Growing by Leaps and Inches: Creative Destruction,
Real Cost Reduction, and Inching Up,” Economic Inquiry, 2003, 41
: 1-19.
East, Edward M., and Donald F. Jones, Inbreeding and Outbreeding: Their Genetic and
Sociological Significance, Philadelphia, PA: J. B. Lippincott Company, 1919.
Gilbert, Richard J., and David M.G. Newbery, “Preemptive Patenting and the Persistence of
Monopoly,” American Economic Review, June 1982, 72(3): 514-526.
Goldberger, Marvin L., Brendan A. Maher, and Pamela Ebert Flattau, eds., Research-Doctorate
Programs in the United States: Continuity and Change, Washington, DC: National
Academy Press, 1995.
Griliches, Zvi, “Hybrid Corn: An Exploration in the Economics of Technological Change,”
Ph.D. Dissertation, Department of Economics, University of Chicago, 1957. (1957a)
Griliches, Zvi, “Hybrid Corn: An Exploration in the Economics of Technological Change,”
Econometrica, October 1957, 25(4): 501-522. (1957b)
Griliches, Zvi, “Research Costs and Social Returns: Hybrid Corn and Related Innovations,”
Journal of Political Economy, Oct. 1958, 66(5): 419-431.
Griliches, Zvi, “Are Farmers Irrational?”, Journal of Political Economy, Feb. 1960, 68(1): 68-71.
(1960a)
Griliches, Zvi, “Hybrid Corn and the Economics of Innovation,” Science, Jul. 29, 1960, 132(No.
3422): 275-280. (1960b)
Griliches, Zvi, “Hybrid Corn Revisited: A Reply,” Econometrica, Sept. 1980, 48(6): 1463-1465.
Griliches, Zvi, “Patent Statistics as Economic Indicators: A Survey,” Journal of Economic
Literature, December 1990, 28(4):1661-1707.
32
Haberle, W., J. K. H. Horber, and Gerd Karl Binnig, “Force Microscopy on Living Cells,”
Journal of Vacuum Science & Technology B, March-April 1991, Part 2, 9
(2): 1210-1213.
Hausman, Jerry, Bronwyn H. Hall, and Zvi Griliches. "Econometric Models for Count Data with an
Application to the Patents-R&D Relationship." Econometrica, July 1984, 909-938.
Institute of Scientific Information (ISI), U.S. University Science Indicators, machine-readable
database on CD-ROM. Institute of Scientific Information, Philadelphia, PA, 2000.
(2000a)
Institute of Scientific Information (ISI), High Impact Papers, 81-98, in 22 Broad Fields,
machine-readable database on CD-ROM. Institute of Scientific Information,
Philadelphia, PA, 2000. (2000b)
Jacoby, Mitch, “New tools for tiny jobs,” Chemical and Engineering News, Oct. 16, 2000,
78(42): 33-35.
Jensen, Richard A., and Marie C. Thursby, “Proofs and Prototypes for Sale: The Tale of
University Licensing,” American Economic Review, March 2001, 91(1): 240-259.
Johnson, Kenneth P., “Redefinition of the BEA Economic Areas,” Survey of Current Business,
February 1995, 75(2): 75-81.
Kannenberg, L. W., “Corn Hybrids: Corn Breeding in the 20th Century,” Ontario Corn Producer
on-line edition, December 1999, http://www.ontariocorn.org/dec99feat.html.
Klevorick, Alvin K., Richard C. Levin, Richard R. Nelson, and Sidney G. Winter, “On the
Sources and Significance of Interindustry Differences in Technological Opportunities,”
Research Policy, March 1995, 24
(2): 185-205.
Kuhn, Thomas S., The Structure of Scientific Revolutions, Chicago: University of Chicago Press,
1962.
33
Lewontin. Richard C., Biology As Ideology: The Doctrine of DNA, New York, NY: Harper
Perennial, 1993.
Mowery, David C., and Arvids A. Ziedonis, “The Geographic Reach of Market and Non-Market
Channels of Technology Transfer: Comparing Citations and Licenses of University
Patents,” National Bureau of Economic Research Working Paper No. 8568, October
2001.
National Research Council, Research-Doctorate Programs in the United States: Data Set,
machine-readable database, Washington, DC: National Academy Press, 1995.
Roco, Mihail C., “From Vision to Implementation of the U.S. National Nanotechnology Initiative,”
Journal of Nanoparticle Research, Feb. 2001, 3(1): 5-11.
http://kapis.wkap.nl/sample.pdf?324433
Roco, Mihail C., and William Sims Bainbridge, eds., Societal Implications of Nanoscience and
Nanotechnology, Boston, MA: Kluwer Academic Publishers, 2001.
http://www.wtec.org/loyola/nano/societalimpact/ anosi.pdf
Roco, Mihail C., R. Stanley Williams, and Paul Alivisatos, eds, Nanotechnology Research
Directions: IWGN Workshop Report –Vision for Nanotechnology in the Next Decade,
Boston, MA: Kluwer Academic Publishers, 1999.
http://www.wtec.org/loyola/nano/IWGN.Research.Directions/
Schmookler, Jacob, Invention and Economic Growth, Cambridge, MA: Harvard University
Press, 1966.
Schmookler, Jacob, Patents, Invention, and Economic Change: Data and Selected Essays, Zvi
Griliches and Leonid Hurwicz, eds., Cambridge, MA: Harvard University Press, 1972.
34
Torero, Máximo, Michael R. Darby, and Lynne G. Zucker, “The Importance of Intellectual
Human Capital in the Birth of the Semiconductor Industry,” UCLA Working Paper,
January 2001.
U.S. Bureau of the Census, Statistical Abstract of the United States: 1991, 111
th
ed.,
Washington: U.S. Government Printing Office, 1991.
U.S. Bureau of Economic Analysis, “Table 7.4. Chain-Type Quantity and Price Indexes for
Personal Consumption Expenditures by Major Type of Product,” National Income and
Product Accounts Tables, http://www.bea.doc.gov/bea/dn/nipaweb/, rev. June 27, 2002
U.S. National Science Foundation, Federal support to universities, colleges, and selected
nonprofit institutions. Fiscal Year 1981, Washington, DC: U.S. Government Printing
Office, 1981.
U.S. National Science Foundation, Federal support to universities, colleges, and selected
nonprofit institutions. Fiscal Year 1982, Washington, DC: U.S. Government Printing
Office, 1982.
U.S. National Science Foundation, Federal support to universities, colleges, and selected
nonprofit institutions. Fiscal Year 1983, Washington, DC: U.S. Government Printing
Office, 1983.
U.S. National Science Foundation, Federal support to universities, colleges, and selected
nonprofit institutions. Fiscal Year 1987, Washington, DC: U.S. Government Printing
Office, 1987.
U.S. National Science Foundation, Federal support to universities, colleges, and selected
nonprofit institutions. Fiscal Year 1989, Washington, DC: U.S. Government Printing
Office, 1989.
35
U.S. National Science Foundation, Federal support to universities, colleges, and selected
nonprofit institutions. Fiscal Year 1990, Washington, DC: U.S. Government Printing
Office, 1990.
U.S. National Science Foundation, Federal science and engineering support to universities,
colleges, and nonprofit institutions. Fiscal Year 1997, Arlington, VA : Division of
Science Resources Studies, Directorate for Social, Behavioral and Economic Sciences,
National Science Foundation, 1997; http://www.nsf.gov/sbe/srs/nsf99331/tables.htm.
U.S. National Science Foundation, Federal science and engineering support to universities,
colleges, and nonprofit institutions. Fiscal Year 2000, Arlington, VA: Division of
Science Resources Studies, Directorate for Social, Behavioral and Economic Sciences,
National Science Foundation, 2000; http://www.nsf.gov/sbe/srs/nsf02319/sectb.htm.
Vettiger P, M. Despont, U. Drechsler, U. Durig, W. Haberle, M. I. Lutwyche, H. E. Rothuizen,
R. Stutz, R. Widmer, and G. K. Binnig, “The ‘Millipede’ – More than one thousand tips
for future AFM data storage,” IBM Journal of Research and Development, May 2000,
44
(3): 323-340.
Wooldridge, Jeffrey M., "On the Application of Robust, Regression-Based Diagnostics to Models
of Conditional Means and Conditional Variances," Journal of Econometrics, January 1991,
47: 5-46.
Zucker, Lynne G., and Michael R. Darby, "Virtuous Circles of Productivity: Star Bioscientists
and the Institutional Transformation of Industry," National Bureau of Economic Research
Working Paper No. 5342, November 1995.
36
Zucker, Lynne G., and Michael R. Darby, “Star Scientists and Institutional Transformation:
Patterns of Invention and Innovation in the Formation of the Biotechnology Industry,”
Proceedings of the National Academy of Sciences, November 12, 1996, 93(23): 12,709-
12,716.
Zucker, Lynne G., and Michael R. Darby, “Present at the Biotechnological Revolution:
Transformation of Technical Identity for a Large Incumbent Pharmaceutical Firm,”
Research Policy, December 1997, 26(4&5): 429-446.
Zucker, Lynne G., Michael R. Darby, and Jeff Armstrong, "Commercializing Knowledge:
University Science, Knowledge Capture, and Firm Performance in Biotechnology,"
Management Science, January 2002, 48(1): 138-153.
Zucker, Lynne G., Michael R. Darby, and Marilynn B. Brewer, “Intellectual Human Capital and
the Birth of U.S. Biotechnology Enterprises,” American Economic Review, March 1998,
88
(1): 290-306.
Zucker, Lynne G., Michael R. Darby, Marilynn B. Brewer, and Yusheng Peng, “Collaboration
Structure and Information Dilemmas in Biotechnology: Organizational Boundaries as
Trust Production,” in Roderick M. Kramer and Tom R. Tyler, eds., Trust in
Organizations, Thousand Oaks, CA: Sage, 1996.
Zucker, Lynne G., Michael R. Darby, and Máximo Torero, “Labor Mobility from Academe to
Commerce,” Journal of Labor Economics, July 2002, 20
(3): 629-660.
37
FOOTNOTES
1
Biotechnology covers a large field, and other basic discoveries like polymer chain reaction
(PCR) and monoclonal antibodies (MABS) both were instrumental in the success of genetic
engineering and led to independent trajectories of invention and innovation.
2
Nano- is a prefix meaning one billionth (i.e., 1/1,000,000,000) so nanoscale refers to working at
sizes measured in the range of one to hundred billionths of a meter. A human hair is on the order
of 50,000 nanometers in diameter.
3
Many important inventions arise in universities and national labs in the course of research
supported, in whole or part, by the federal and/or state governments. For those inventions
discovery is not at issue and only the licensing fees and other costs of transferring the knowledge
to the firm apply.
4
Sun Microsystems charges computer manufacturers only a nominal fee to install the Java
TM
Virtual Machine software on their machine and any significant profits result from Sun’s selling
complementary hardware with the aura of controlling the past and future development of Java
TM
.
5
Interestingly, East and Jones did not attempt to patent their invention as UCSF and Stanford
later did Cohen and Boyer’s discovery, but instead confronted Aiken’s theorem: "Don't worry
about people stealing your ideas. If your ideas are any good, you'll have to ram them down
people's throats.” (Howard H. Aiken invented the first programmable mechanical computer –
the Mark I completed in 1943 – and shares parental credit for the electronic computer, itself a
pretty good invention of a method of inventing.) George H. Shull working at the Carnegie
Institution’s Cold Spring Harbor Laboratory had also observed the increased yield with double-
cross hybrids in 1910-1911 but did not publish his results nor appreciate their importance.
6
The case of hybrid seed corn deserves a thorough study by another Griliches to sort out why the
38
proof of concept for appropriable double-cross hybrids appears to have been completed by 1920,
the first firm was founded in 1926, and as late as 1934 a trivial percentage – considerably under
5 percent on all accounts – of corn acreage was planted with hybrid seed corn. Ten years later in
1944, 59 percent of all U.S. acreage, including 90 percent of the Corn Belt, was planted using
hybrid seed (Kannenberg 1999). One possible clue is that Henry A. Wallace became Secretary
of Agriculture in 1933 and Vice President in 1940.
7
The invention is reported in Cohen, Chang, Boyer, and Helling (1973) and patented in Cohen
and Boyer (1980). See footnote 1 above on other inventions enabling biotechnology.
8
Subcommittee on Nanoscale Science, Engineering and Technology (NSET), Committee on
Technology, National Science and Technology Council, February 2000, as posted at
http://nano.gov/omb_nifty50.htm.
9
Beginning in 1990 the counts of nano articles per 1000 S&E articles was significantly greater
than the 1981-1989 mean and increasing every year.
10
Besides scanning or proximal probe microscopy, candidates for enabling breakthroughs
include quantum dots, carbon nanotubes, and the phenomenal drop in the cost of computing.
11
Remarkably, Vettiger, …, and Binnig (2000) report on the “millipede” which uses an array of
over 1024 tiny AFMs with read/write capability to achieve high speed data storage and retrieval
at a density of 100-200 gigabytes per square inch (Gb/in
2
). “The very large scale integration
(VLSI) of micro/nanomechanical devices (cantilevers/tips) on a single chip leads to the largest
and densest 2D array of 32 x 32 (1024) AFM cantilevers with integrated write/read storage
functionality ever built. Time-multiplexed electronics control the write/read storage cycles for
parallel operation of the Millipede array chip. Initial areal densities of 100-200 Gb/in
2
have been
achieved with the 32 x 32 array chip, which has potential for further improvements. In addition
39
to data storage in polymers or other media, and not excluding magnetics, we envision areas in
nanoscale science and technology such as lithography, high-speed/large-scale imaging,
molecular and atomic manipulation, and many others in which Millipede may open up new
perspectives and opportunities.”
12
On a per capita basis (Table 1, Panel A), the United States loses its lead to Switzerland. Since
Switzerland is a scientific powerhouse and the origin of the enabling technology (by employees
of a firm headquartered in America), it is natural that they are world leaders.
13
The Virtual Journal of Nanoscale Science & Technology is an internet publication of the
American Institute of Physics and the American Physical Society. It provides links to nano S&T
articles published in regular refereed journals. Since it started in January 2000, it has served as a
central resource for the nano S&T community, the editorial board has identified and classified
7,466 articles.
14
Thus, N, the number of observations equals 15 years (1985-1999) times 172 regions which is
2,580. In our work on biotechnology we used the BEA definition of 183 functional economic
areas operative at the time the industry was forming. For nanotechnology, we have switched to
the current definition of 172 economic areas since it more nearly corresponds to regions during
the formation of this industry.
15
We are developing what we hope will prove to be an improved search methodology with the
goal of increasing the completeness of coverage without an unreasonable rate of false positives.
The resulting list of articles will be available for future work by other researchers and ourselves.
16
ISI does not associate each author with a specific address, so we allocated credit equally
among the addresses listed. If there were 3 addresses on a paper and 2 were in region 10 and 1 in
region 20, then region 10 is credited with 2/3 of an article and region 20 with 1/3 of an article.
40
Papers with foreign authors will add up to less than 1 when summed across the 173 U.S. regions.
17
We also experimented with including a count of all other doctoral programs in science and
engineering which are ranked in the top 30 (including all universities tied for 30
th
place). This
was uniformly insignificant when included with other measures of the science base so we
dropped it to simplify the discussion.
18
The significance of key variables in these regressions is generally not sensitive to the
Wooldridge correction, but to achieve an estimate of the variance-covariance matrix that is not
restricted by first-moment parameter estimates, we apply the Woodridge correction. An
alternative would be to implement a binomial specification, but as explained in Wooldridge
(1991), this procedure may bias both first and second moment estimates, whereas the Poisson
process potentially biases only the second moment parameters
19
The insignificance of top-112 university articles and top-10 doctoral programs occurs when all
four, correlated measures of science base are in the same equation. Each variable is very
significantly positive in a one-variable regression.
20
We must drop the top-10 doctoral program from the estimates because the values are constant
for a given region. Conventionally, the 172 individual constants for each region are unreported.
We cannot compute Wooldridge corrections for the standard errors but individual regional
constants should obviate any problem arising from excess zeroes.
21
We also experimented with knowledge stocks based on purely academic articles with at least one
top-112 university author which were published in the top 5 percent and separately the second 5
percent of ISI journals in terms of average citations per article within each of the 5 broad S&E fields
in Table A2. These measures were dropped since they were always insignificant when the top-112
university articles or high-impact articles measures were in the regression.
41
Table 1
Concentration of Nanoscale Science and Engineering Articles
by Nations and in the U.S.
Panel A -- National Data
Nations Nano S&T Articles
1990-98 1990-98 1991-98
High Impact
a
High Impact
a
High Impact
a
% of World Total per million
c
per million
c
Australia 4.1% 0.77 0.71
France 4.0% 0.22 0.22
Germany 6.9% 0.27 0.26
Japan 5.5% 0.14 0.14
Switzerland 4.5% 2.08 2.08
United Kingdom 3.8% 0.21 0.21
United States 53.9% 0.68 0.63
Subtotal 82.7% 0.44 0.42
Rest of World 17.3% 0.01 0.01
World Total 100.0% 0.06 0.06
Panel B -- United States Data
Regions Nano S&T Articles
1990-99 1990-99 1991-98
Top-112 Univs.
a
Top-112 Univs.
b
High Impact
a
% of U.S. Total per million
c
per million
c
Atlanta 3.5% 25.79 2.32
Boston 8.7% 39.24 2.30
Champagne-Urbana 4.5% 312.76 10.20
Chicago 4.5% 14.23 1.10
Hartford-New Haven 3.4% 27.25 0.30
Los Angeles/Santa Barbara 10.0% 18.15 0.74
New York 5.3% 7.94 0.70
Philadelphia 4.6% 16.61 1.10
Raleigh-Durham 3.8% 94.67 8.19
San Francisco Bay 9.6% 38.63 2.68
Subtotal 57.8% 22.24 1.36
Rest of the United States 42.2% 5.51 0.34
United States Total 100.0% 10.29 0.63
Notes: a. High Impact Papers, 81-98, Institute of Scientific Information (2000b).
b. U.S. University Science Indicators, Institute of Scientific Information (2000a).
c. Population estimates for 1990 taken from U.S. Bureau of the Census (1991, no. 1434)
42
Table 2
Highly Cited Articles (over 250 Citations) by the Inventors of
The Scanning Tunneling Microscope (STM) and the Atomic Force Microscope (AFM)
Cites Publication
_________________________________________________________________________________________________________________________________________________________________
546 Binnig, Gerd Karl, Heinrich Rohrer, Christoph Gerber, and E. Weibel, “Tunneling
through a Controllable Vacuum Gap,” Applied Physics Letters, 1982, 40(2): 178-180.
739 Binnig, Gerd Karl, and Heinrich Rohrer, “Scanning Tunneling Microscopy,” Helvetica
Physica Acta, 1982, 55(6): 726-735.
308 Binnig, Gerd Karl, and Heinrich Rohrer, “Scanning Tunneling Microscopy,” Surface
Science, 1983, 126(1-3): 236-244.
947 Binnig, Gerd Karl, Heinrich Rohrer, Christoph Gerber, and E. Weibel, “7X7
Reconstruction on Si(111) Resolved in Real Space,” Physical Review Letters, 1983,
50(2): 120-123.
3,436 Binnig, Gerd Karl, Calvin F. Quate, and Christoph Gerber, “Atomic Force Microscope,”
Physical Review Letters, March 3, 1986, 56(9): 930-933.
409 Binnig, Gerd Karl, and Heinrich Rohrer, “Scanning Tunneling Microscopy,” IBM
Journal of Research and Development, July 1986, 30(4): 355-369.
258 Binnig, Gerd Karl, Christoph Gerber, E. Stoll, T. R. Albrecht, and Calvin F. Quate,
“Atomic Resolution with Atomic Force Microscope,” Europhysics Letters, June 15,
1987, 3(12): 1281-1286.
296
a
Binnig, Gerd Karl, and Heinrich Rohrer, “Scanning Tunneling Microscopy – From Birth
to Adolescence,” Reviews of Modern Physics, July 1987, 59(3): 615-625.
_________________________________________________________________________________________________________________________________________________________________
Source: ISI Science Citation Index Expanded (SCI-EXPANDED)--1975-present (Updated July 15, 2002)
Note: a. 296 is the total for three versions of this article published nearly simultaneously.
43
Table 3
Definitions of the Variables
For those variables marked with an asterisk (*), there are alternative series calculated depending
on whether the variable refers to all science-and-technology areas combined or to one of the five
specific science-and-technology fields as defined in Tables A2 and A3.
Article stocks* Knowledge stock in a given year and region as measured by counts of
nano articles of a specified type excluding articles with firm authors
(see immediately below) cumulated from 1981 according to X
it
= A
it
+
(1 – δ) X
i,t-1
where X
i,1980
= 0, A
it
is the number of articles of the
specified type in region i in year t and δ is the depreciation rate of
knowledge. We assume a conventional value for δ of 0.2; see Griliches
(1990).
Articles-high-impact* Knowledge stock where the input article series A
it
is a count of articles
with no firm authors in the ISI High-Impact database.
Articles-top-112* Knowledge stock where the input article series A
it
is drawn from articles
with no firm authors and in the ISI High-Impact database.
Average wage BEA series “Average Wage per Job” 1996 $1000s
a
by year and region.
Doctoral programs This is a count of science and engineering programs ranked among the
in top 10* top 10 in the U.S. in the 1993 National Research Council (NRC) study
Employment BEA series “Total Full-time and Part-time Employment” in millions of
jobs by year and region
Entry* Number of firms entering nano S&T in a given year and region. For this
paper, we define a firm’s entry as the first year that an author affiliated
with the firm publishes a nano article in the given region.
Entry to date* Cumulative entry of firms in a given region from 1981 up through given
year.
Region One of the 172 functional economic areas described in Johnson (1995).
Research funding Research funding in millions of 1996 dollars
a
is the sum for the region
and year of federal funding going to the top-100 universities as reported
by the National Science Foundation.
Venture capital Stock of venture capital funding reported in the Venture Economics
funding database in billions of 1996 dollars
a
according to the knowledge stock
formula (20% depreciation)
Venture capital deals Stock of venture capital deals reported in the Venture Economics
database (20% depreciation)
___________________________________________
Note: a. All current dollar amounts are converted to 1996 dollars by deflating by the chain-type
price index for personal consumption expenditures (U.S. Bureau of Economic
Analysis 2002)
44
Table 4
Sample Statistics of the Variables, 1985-1999
Variables Units Mean Std. Dev. Minimum Maximum
Variables invariant with science & technology fields:
Research funding millions of 1996 dollars 0.05 0.14 0 1.12
Research funding stock millions of 1996 dollars 0.22 0.58 0 5.00
Venture capital funding billions of 1996 dollars 0.09 0.49 0 16.87
Venture capital funding stock billions of 1996 dollars 0.28 1.11 0 28.69
Venture capital deals count of deals 14.91 64.19 0 1547.00
Venture capital deals stock count of deals 56.10 223.21 0 4363.35
Employment millions of jobs 0.83 1.48 0.03 14.16
Average wage thousands of 1996 dollars 23.73 3.27 16.60 38.39
Variables for all science & technology fields combined:
Entry count of firms 0.08 0.45 0 7
Entry to date count of firms 0.37 2.09 0 41
Articles-top-112 flow count of articles 1.32 4.55 0 53.33
Articles-top-112 stock count of articles 3.66 12.81 0 156.40
Articles-high-impact flow count of articles 0.06 0.31 0 4.38
Aritcles-high-impact stock count of articles 0.20 0.86 0 9.36
Doctoral Programs in top 10 count of programs 1.38 5.22 0 46
Variables for Biology/Medicine/Chemistry:
Entry count of firms 0.02 0.17 0 3
Entry to date count of firms 0.11 0.59 0 10
Articles-top-112 flow count of articles 0.33 1.31 0 21.82
Articles-top-112 stock count of articles 0.95 3.30 0 46.60
Articles-high-impact flow count of articles 0.01 0.07 0 1.00
Aritcles-high-impact stock count of articles 0.02 0.14 0 2.13
Doctoral Programs in top 10 count of programs 0.60 2.38 0 21
Variables for Integrated Circuit/Semiconductors/Superconductors:
Entry count of firms 0.06 0.35 0 6
Entry to date count of firms 0.26 1.66 0 33
Articles-top-112 flow count of articles 0.79 2.82 0 37.82
Articles-top-112 stock count of articles 2.18 7.90 0 108.66
Articles-high-impact flow count of articles 0.04 0.21 0 3.00
Aritcles-high-impact stock count of articles 0.12 0.56 0 6.66
Doctoral Programs in top 10 count of programs 0.24 1.06 0 8
(concluded on following page)
45
Table 4 (concluded)
Variables Units Mean Std. Dev. Minimum Maximum
Variables for Computer/Information processing/Multimedia:
Entry count of firms 0.00 0.03 0 1
Entry to date count of firms 0.00 0.06 0 2
Articles-top-112 flow count of articles 0.00 0.06 0 1.33
Articles-top-112 stock count of articles 0.01 0.10 0 1.84
Articles-high-impact flow count of articles 0 0 0 0
Aritcles-high-impact stock count of articles 0 0 0 0
Doctoral Programs in top 10 count of programs 0.12 0.57 0 5
Variables for Other Engineering:
Entry count of firms 0.00 0.07 0 2
Entry to date count of firms 0.01 0.11 0 2
Articles-top-112 flow count of articles 0.03 0.22 0 4.50
Articles-top-112 stock count of articles 0.06 0.43 0 9.06
Articles-high-impact flow count of articles 0 0 0 0
Aritcles-high-impact stock count of articles 0 0 0 0
Doctoral Programs in top 10 count of programs 0.18 0.64 0 5
Variables for Other Sciences:
Entry count of firms 0.01 0.09 0 2
Entry to date count of firms 0.04 0.27 0 5
Articles-top-112 flow count of articles 0.07 0.37 0 6.33
Articles-top-112 stock count of articles 0.22 0.92 0 14.70
Articles-high-impact flow count of articles 0.02 0.15 0 3.00
Aritcles-high-impact stock count of articles 0.06 0.36 0 6.11
Doctoral Programs in top 10 count of programs 0.23 0.97 0 8
Note: The number of observations (N) is 15 years x 172 regions = 2,580.
46
Table 5
Entry into Nanotechnology in the U.S. – All Science-Technology Areas Combined
Poisson Regressions for 1985-1999
Independent Variables Coefficients
(a) (b) (c) (d) (e) (f)
Constant -3.7032*** -12.1967*** -10.2842*** -3.7168*** -11.5434*** -10.9341***
(0.1354) (0.9844) (1.006) (0.1368) (1.043) (0.9691)
Articles-high-impact stock 0.2117*** 0.2269*** 0.2749*** 0.1856*
(0.0539) (0.0623) (0.0741) (0.0754)
Articles-top-112 universities stock -0.0024 -0.0029 -0.0072 0.0034
(0.0037) (0.0048) (0.0054) (0.0064)
Research funding stock --top-100 1.0262*** 0.2779 1.0320*** 0.1642
Universities (0.0965) (0.2299) (0.1307) (0.2412)
Doctoral programs in top 10 0.0085 -0.0121 0.0142 0.0081
(0.0099) (0.0143) (0.0220) (0.0336)
Employment 0.0573 0.0614 0.0821* 0.0387
(0.0393) (0.0654) (0.0374) (0.0720)
Average wage 0.3516*** 0.2674*** 0.3230*** 0.2938***
(0.0372) (0.0400) (0.0404) (0.0378)
Venture capital deals stock -0.0005 0.0007* -0.0001
(0.0005) (0.0003) (0.0008)
Venture capital funding stock 0.0941 -0.0992* -0.0578
(0.0729) (0.0507) (0.1034)
Log-likelihood
-454.110 -433.087 -410.844 -452.6914 -426.699 -408.3245
Restricted log-likelihood
-798.228 -798.228 -798.228 -798.228 -798.228 -798.228
Sample size
2580 2580 2580 2580 2580 2580
Note: All models were estimated as a Poisson process with standard errors (in parentheses) corrected following Wooldridge (1991).
Significance levels: * <
0.05, ** < 0.01, *** < 0.001.
47
Table 6
Entry into Nanotechnology in the U.S. – All Science-Technology Areas Combined
Poisson Regressions with Fixed Effects for 1985-1999
Independent Variables Coefficients
(a) (b) (c) (d) (e) (f)
Constants not reported – fixed effect for
Each region
Articles-high-impact stock
0.2595*** 0.2633*** 0.3116*** 0.2849***
0.0647 0.0691 0.0764 0.0833
Articles-top-112 universities stock
-0.0053 -0.0135** -0.0017 -0.0024
0.0041 0.0058 0.0056 0.0073
Research funding stock --top-100
1.3642*** 0.9116** 1.2044*** 0.0241
Universities
0.3597 0.4238 0.3635 0.4943
Doctoral programs in top 10
Employment
0.6383* 0.9230** 1.2342*** 1.0307**
0.2720 0.4538 0.3307 0.4758
Average wage
0.3125*** 0.1839** 0.4183*** 0.3454***
0.0570 0.0786 0.0690 0.0979
Venture capital deals stock
-0.0013 -0.0003 -0.0019**
0.0007 0.0006 0.0008
Venture capital funding stock
0.0850 -0.0826 0.0619
0.0789 0.0681 0.0823
Log-likelihood
-266.720 -270.626 -261.463 -264.870 -263.917 -255.435
Restricted log-likelihood
-798.228 -798.228 -798.228 -798.228 -798.228 -798.228
Sample size
2580 2580 2580 2580 2580 2580
Note: All models were estimated as a Poisson process with standard errors (in parentheses). The Wooldridge (1991) corrections could not be
computed for these fixed-effect regressions with separate constants for each of the 172 regions.
Significance levels: * <
0.05, ** < 0.01, *** < 0.001.
48
Table 7
Entry into Nanotechnology in the U.S. – Integrated circuits-Semiconductors-Superconductors Area
Poisson Regressions for 1985-1999
Independent Variables Coefficients
(a) (b) (c) (d) (e) (f)
Constant -4.1771*** -12.7174*** -10.6840*** -4.1679*** -12.0283*** -11.5235***
(0.1668) (1.1110) (1.1660) (0.1687) (1.1160) (1.227)
Articles-high-impact stock 0.3482*** 0.2577*** 0.3472*** 0.1825*
(0.0666) (0.0818) (0.0718) (0.0858)
Articles-top-112 universities stock 0.0001 0.0018 0.0013 0.0144
(0.0039) (0.0066) (0.0049) (0.0077)
Research funding stock --top-100 0.9981*** 0.2748 0.9946*** 0.1039
Universities (0.0738) (0.2337) (0.0784) (0.2312)
Doctoral programs in top 10 0.1251*** 0.0676 0.0487 0.0095
(0.0379) (0.0541) (0.0513) (0.0963)
Employment 0.0627 0.0077 0.1015*** 0.0574
(0.0422) (0.0633) (0.0396) (0.0912)
Average wage 0.3564*** 0.2685*** 0.3247*** 0.3006***
(0.0412) (0.0470) (0.0422) (0.0476)
Venture capital deals stock 0.0005 0.0010*** 0.0009
(0.0003) (0.0003) (0.0007)
Venture capital funding stock -0.0486 -0.1565*** -0.2082
(0.0545) (0.0496) (0.1135)
Log-likelihood
-345.198 -331.618 -317.263 -343.452 -322.790 -311.824
Restricted log-likelihood
-607.585
-607.585 607.585 -607.585 607.585 607.585
Sample size
2580 2580 2580 2580 2580 2580
Note: All models were estimated as a Poisson process with standard errors (in parentheses) corrected following Wooldridge (1991).
Significance levels: * <
0.05, ** < 0.01, *** < 0.001.
49
Table 8
Entry into Nanotechnology in the U.S. – Integrated circuits-Semiconductors-Superconductors Area
Poisson Regressions with Fixed Effects for 1985-1999
Independent Variables Coefficients
(a) (b) (c) (d) (e) (f)
Constants not reported – fixed effect for
Each region
Articles-high-impact stock
0.8560*** 0.7796*** 0.8547*** 0.7504***
0.1981 0.2031 0.2149 0.2243
Articles-top-112 universities stock
-0.0011 -0.0066 0.0073 0.0105
0.0066 0.0086 0.0087 0.0097
Research funding stock --top-100
0.8235 0.3027 0.7187 -0.6580
Universities
0.4303 0.4927 0.4321 0.5848
Doctoral programs in top 10
Employment
0.3048 0.3096 1.1215** 0.7562
0.3157 0.4781 0.4057 0.5221
Average wage
0.3491*** 0.2137* 0.4854*** 0.4335***
0.0672 0.0891 0.0841 0.1202
Venture capital deals stock
-0.0002 -0.0001 -0.0010
0.0008 0.0007 0.0008
Venture capital funding stock
-0.0444 -0.1499 -0.0615
0.0901 0.0907 0.0934
Log-likelihood
-193.487 -199.576 -190.297 -192.393 -191.665 -184.303
Restricted log-likelihood
-607.585
-607.585 607.585 -607.585 607.585 607.585
Sample size
2580 2580 2580 2580 2580 2580
Note: All models were estimated as a Poisson process with standard errors (in parentheses). The Wooldridge (1991) corrections could not be
computed for these fixed-effect regressions with separate constants for each of the 172 regions.
Significance levels: * <
0.05, ** < 0.01, *** < 0.001.
50
Table 9
Entry into Nanotechnology in the U.S. – Biology-Medicine-Chemistry Area
Poisson Regressions for 1985-1999
Independent Variables Coefficients
(a) (b) (c) (d) (e) (f)
Constant
-4.7784*** -13.8448*** -12.4097*** -4.7752*** -12.9833***
-12.6577***
(0.2624) (1.5960) (1.7410) (0.2672) (1.6150)
(1.6860)
Articles-high-impact stock
0.4661 0.4567 0.6464
0.6947
(0.4043) (0.7675) (0.7105)
(0.8169)
Articles-top-112 universities stock
0.0165 0.0074 -0.0116
0.0106
(0.0216) (0.0285) (0.0317)
(0.0400)
Research funding stock --top-100
0.8557*** 0.2833 0.9732***
0.2239
Universities
(0.2015) (0.3461) (0.2700)
(0.4088)
Doctoral programs in top 10
0.0405 -0.0154 0.0051
0.0664
(0.0502) (0.0739) (0.1180)
(0.1594)
Employment
0.0185 -0.0256 0.0251
-0.0787
(0.0634) (0.1410) (0.0663)
(0.1654)
Average wage
0.3699*** 0.3107*** 0.3358***
0.3219***
(0.0579) (0.0660) (0.0612)
(0.0637)
Venture capital deals stock
-0.0002 0.0002
-0.0011
(0.0015) (0.0008)
(0.0018)
Venture capital funding stock
0.1013 0.0017
0.1056
(0.1809) (0.1234)
(0.2008)
Log-likelihood
-194.334 -181.956 -178.493 -192.765 -180.865 -177.808
Restricted log-likelihood
-276.847 -276.847 -276.847 -276.847 -276.847 -276.847
Sample size
2580 2580 2580 2580 2580 2580
Note: All models were estimated as a Poisson process with standard errors (in parentheses) corrected following Wooldridge (1991).
Significance levels: * <
0.05, ** < 0.01, *** < 0.001.
51
Table 10
Entry into Nanotechnology in the U.S. – Biology-Medicine-Chemistry Area
Poisson Regressions with Fixed Effects for 1985-1999
Independent Variables Coefficients
(a) (b) (c) (d) (e) (f)
Constants not reported – fixed effect for
Each region
Articles-high-impact stock
0.4462 0.2979 0.6668 0.5033
0.4552 0.4913 0.5150 0.5483
Articles-top-112 universities stock
0.0122 -0.0080 0.0032 -0.0015
0.0246 0.0378 0.0300 0.0427
Research funding stock --top-100
1.1933* 0.4516 1.2715* 0.2926
Universities
0.6104 0.7753 0.6284 0.8964
Doctoral programs in top 10
Employment
0.8226 0.7471 1.0466 0.8404
0.5886 1.0126 0.6677 1.0465
Average wage
0.2761** 0.2087 0.3192* 0.2564
0.1122 0.1527 0.1291 0.1873
Venture capital deals stock
-0.0010 -0.0008 -0.0013
0.0012 0.0011 0.0012
Venture capital funding stock
0.1245 0.0474 0.0994
0.1192 0.1071 0.1219
Log-likelihood
-107.581 -106.548 -106.249 -107.014 -106.175 -105.666
Restricted log-likelihood
-276.847 -276.847 -276.847 -276.847 -276.847 -276.847
Sample size
2580 2580 2580 2580 2580 2580
Note: All models were estimated as a Poisson process with standard errors (in parentheses). The Wooldridge (1991) corrections could not be
computed for these fixed-effect regressions with separate constants for each of the 172 regions.
Significance levels: * <
0.05, ** < 0.01, *** < 0.001.
52
ARIZONA STATE UNIV NEW MEXICO STATE UNIV UNIV CALIF SAN FRANCISCO UNIV NEW MEXICO
BAYLOR COLL MED NEW YORK UNIV UNIV CALIF SANTA BARBARA UNIV OREGON
BOSTON UNIV NORTHWESTERN UNIV UNIV CALIF SANTA CRUZ UNIV PENN
BRANDEIS UNIV OHIO STATE UNIV UNIV CHICAGO UNIV PITTSBURGH
BROWN UNIV OREGON HLTH SCI UNIV UNIV CINCINNATI UNIV ROCHESTER
CALTECH OREGON STATE UNIV UNIV COLORADO UNIV SO CALIF
CARNEGIE MELLON UNIV PENN STATE UNIV UNIV CONNECTICUT UNIV TENNESSEE
CASE WESTERN RESERVE UNIV PRINCETON UNIV UNIV DELAWARE UNIV TEXAS AUSTIN
COLORADO STATE UNIV PURDUE UNIV UNIV FLORIDA UNIV TEXAS DALLAS
COLUMBIA UNIV RICE UNIV UNIV GEORGIA UNIV TEXAS HOUSTON
CORNELL UNIV ROCKEFELLER UNIV UNIV HAWAII UNIV TEXAS SAN ANTONIO HLTH SCI CTR
CUNY RUTGERS STATE UNIV UNIV ILLINOIS CHICAGO UNIV UTAH
DARTMOUTH COLL STANFORD UNIV UNIV ILLINOIS URBANA UNIV VERMONT
DUKE UNIV SUNY BUFFALO UNIV IOWA UNIV VIRGINIA
EMORY UNIV SUNY STONY BROOK UNIV KANSAS UNIV WASHINGTON
FLORIDA STATE UNIV SYRACUSE UNIV UNIV KENTUCK
Y
UNIV WISCONSIN MADISON
GEORGETOWN UNIV TEXAS A&M UNIV UNIV MARYLAND BALTIMORE UTAH STATE UNIV
GEORGIA INST TECHNOL TUFTS UNIV UNIV MARYLAND COLLEGE PARK VANDERBILT UNIV
HARVARD UNIV TULANE UNIV UNIV MASS AMHERST VIRGINIA COMMONWEALTH UNIV
INDIANA UNIV UNIV ALABAMA UNIV MASS WORCESTER VIRGINIA POLYTECH INST
IOWA STATE UNIV UNIV ALASKA UNIV MASSACHUSETTS W VIRGINIA UNIV
JOHNS HOPKINS UNIV UNIV ARIZONA UNIV MIAMI WAKE FOREST UNIV
LEHIGH UNIV UNIV CALIF BERKELE
Y
UNIV MICHIGAN WASHINGTON STATE UNIV
LOUISIANA STATE UNIV UNIV CALIF DAVIS UNIV MINNESOTA WASHINGTON UNIV
LOYOLA UNIV UNIV CALIF IRVINE UNIV MISSOURI WAYNE STATE UNIV
MICHIGAN STATE UNIV UNIV CALIF LOS ANGELES UNIV N CAROLINA CHAPEL HILL WOODS HOLE OCEANOG INST
MIT UNIV CALIF RIVERSIDE UNIV NEBRASKA YALE UNIV
N CAROLINA STATE UNIV UNIV CALIF SAN DIEGO UNIV NEW HAMPSHIRE YESHIVA UNIV
Table A1. List of Top-112 Universities as Defined by the Institute of Scientific Information
Source: Institute of Scientific Information, U.S. University Science Indicators, machine-readable database on CD-ROM, Philadelphia: Institute of Scientific Information, 2000.
2000. [Although the data base aims at the top 100 research universities, the stopping rule appears to include the 13 universities
tied for 100th place in their covered list of 112 universities.]
53
Table A2. Zucker-Darby S&E Field Categorization and ISI Category Descriptions
Zucker-Darby
Categorization
ISI Journal Category Description
Biology/
Medicine/
Chemistry
Agriculture/Agronomy, Anesthesia & Intensive Care, Animal & Plant
Sciences, Animal Sciences, Neurosciences & Behavior, Biochemistry &
Biophysics, Biology, Biotechnol & Appl Microbiol, Cardiovasc & Respirat
Syst, Cell & Developmental Biol, Oncogenesis & Cancer Res, Agricultural
Chemistry, Chemical Engineering, Chemistry & Analysis, Chemistry,
Cardiovasc & Hematology Res, Dentistry/Oral Surgery & Med,
Dermatology, Medical Res, Diag & Treatmt, Endocrinol, Nutrit & Metab,
Entomology/Pest Control, Environment/Ecology, Experimental Biology,
Food Science/Nutrition, Gastroenterol and Hepatology, General & Internal
Medicine, Hematology, Immunology, Inorganic & Nucl Chemistry, Clin
Immunol & Infect Dis, Molecular Biology & Genetics, Microbiology,
Resrch/Lab Med & Med Techn, Medical Res, General Topics, Neurology,
Endocrinol, Metab & Nutrit, Medical Res, Organs & Syst, Oncology,
Ophthalmology, Organic Chem/Polymer Sci, Orthopedics & Sports Med,
Otolaryngology, Pediatrics, Physical Chem/Chemical Phys, Pharmacology &
Toxicology, Plant Sciences, Pharmacology/Toxicology, Psychiatry,
Physiology, Clin Psychology & Psychiatry, Radiol, Nucl Med & Imaging,
Reproductive Medicine, Rheumatology, Environmt Med & Public Hlth,
Surgery, Urology, and Veterinary Med/Animal Health
Computer/
Information
Processing/
Multimedia
AI, Robotics & Auto Control, Computer Sci & Engineering, Engineering
Mathematics, Info Technol & Commun Syst, and Mathematics
Integrated
Circuit/Semi- &
Super-conductor
Appl Phys/Cond Matt/Mat Sci, Elect & Electronic Engn, Mechanical
Engineering, Metallurgy, Materials Sci and Engn, Optics & Acoustics, Physics,
and Spectrosc/Instrum/Analyt Sci
Other
Engineering
Aerospace Engineering, Civil Engineering, Environmt Engineering/Energy,
Engineering Mgmt/General, Geol/Petrol/Mining Engn, Instrumentation/
Measurement, Nuclear Engineering, and Space Science
Other Sciences Aquatic Sciences, Earth Sciences, and Multidisciplinary
Source: Darby and Zucker (1999).
54
Table A3. Zucker-Darby S&E Field Categorization and NRC Standard Doctoral Programs
Zucker-Darby Categorization and NRC standard programs
Biology/
Medicine/
Chemistry
Biochemistry & Molecular Biology
Cell & Developmental Biology
Molecular & General Genetics
Ecology, Evolution & Behavioral
Pharmacology
Chemistry
Biomedical Engineering
Chemical Engineering
Neurosciences
Physiology
Computer/
Information
Processing/
Multimedia
Computer Sciences
Mathematics
Integrated
Circuit/Semi-
& Super-
conductor
Physics
Electrical Engineering
Materials Science
Mechanical Engineering
Other
Engineering
Aerospace Engineering
Civil Engineering
Industrial Engineering
Other
Sciences
Oceanography
Astrophysics/Astronomy
Statistics/Biostatistics
Geosciences
Source: Darby and Zucker (1999).
55
Figure 1. Nano Articles per 1000 Science Articles
0
5
10
15
20
25
1
9
8
1
1
9
8
2
1
9
8
3
1
9
8
4
1
9
8
5
1
9
8
6
1
9
8
7
1
9
8
8
1
9
8
9
1
9
9
0
1
9
9
1
1
9
9
2
1
9
9
3
1
9
9
4
1
9
9
5
1
9
9
6
1
9
9
7
1
9
9
8
1
9
9
9
2
0
0
0
2
0
0
1
2
0
0
2
2
0
0
3
Source: ISI Web of Science (updated to May 30, 2003. 2003 data are (365/150)* totals for January 1-May 30, 2003.)
nano articles per 1000 science articles
Nano articles per 1000 science articles 1981-1990 average 1991-2003 average
1981-1990 average
1991-2003 average
nano articles per 1000
science articles
56
Figure 2. Nano Patents Granted 1981-2000
by year granted and year of application
0
50
100
150
200
250
300
350
400
1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000
Number of patents per year
Patents Granted Patents Applied for
Right Truncation Effect
57
1990
1991
1992
1993
1994
1995
1996
1997
1998
A
u
s
t
r
a
l
i
a
J
a
pa
n
Fr
a
n
c
e
G
e
r
m
a
n
y
U
n
i
t
e
d
K
i
n
g
d
o
m
S
w
i
t
z
e
r
l
a
nd
U
n
i
t
e
d
S
t
a
t
e
s
0
5
10
15
20
25
30
Number of Published Articles per year
Each country plotted has 1% or more of the 1990-99 world
total of citations to all ISI high-impact (highly cited) nano-
articles.
Figure 3. High-Impact Nano-Articles Published Worldwide
All Sciences and Engineering, 1990-1998 by Country
58
1
9
9
9
1
9
9
8
1
9
9
7
1
9
9
6
1
9
9
5
1
9
9
4
1
9
9
3
1
9
9
2
1
9
9
1
H
a
r
t
f
o
r
d
-
N
e
w
H
a
v
e
n
A
t
l
a
n
t
a
R
a
le
i
g
h
-
D
u
r
h
a
m
C
h
a
m
p
a
g
n
e
-
U
r
b
a
n
a
C
h
i
c
a
g
o
P
h
il
a
d
e
l
p
h
i
a
N
e
w
Y
o
r
k
B
o
s
t
o
n
S
a
n
F
r
a
n
c
is
c
o
B
a
y
L
o
s
A
n
g
e
le
s
/
S
a
n
t
a
B
a
r
b
a
r
a
0
10
20
30
40
50
60
Number of Published Articles per
Year
BEA-defined functional economic
areas
Figure 4. Nano-Articles Published by Top-112 Research Universities
All Sciences and Engineering, 1991-1999
Each FEA plotted has at least 3%
of the 1991-99 US total articles
published by top research
universities, All 10 have 58% of
this total.
59
1-6/2000
7-12/2000
1-6/2001
7-12/2001
1-6/2002
Optical Properties and Quantum
Optics*
Electronic Structure/Transport
Structural Properties
Nanomagnetism
Imaging Science/Technology
Advances in
Fabrication/Processing
MEMS/NEMS**
Carbon Nanotube Science and
Technology
Nanoscale Devices***
Quantum Coherence, Computing,
and Information Storage
Supramolecular and Biochemical
Assembly**
Organic-Inorganic Hybrid
Nanostructures**
Surface and Interface Properties
Chemical Synthesis Methods**
Miscellaneous
0%
5%
10%
15%
20%
25%
*In 1
(
1-2
)
: Nanooptics and
Quantum Optics.
**First used in 1(2 or 3).
***Dropped in 3(7).
Figure 5. Percentage Distribution of
Virtual Journal of Nanoscale Science & Technology
Articles by Classification, Data from Volumes 1-5, January 2000-June 2002
60
Figure 6. Nanotech and Biotech Publishing and Patenting Compared
1973-1994 for Biotech and 1985-2002 for Nanotech
0.00
5.00
10.00
15.00
20.00
25.00
30.00
35.00
40.00
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Years from Base Year (1973 for Biotech and 1985 for Nanotech)
Number of Articles per 1000 S&E Articles
Number of Patents Granted in 100s
Nano articles per 1000 science & engineering articles Biotech Articles (GenBank) per 1000 S&E articles
Nano Patents (in 100s) Biotech Patents (in 100s)
nano articles per
1000 S&E articles
nano patents
biotech patents
biotech articles (GenBank)
per 1000 S&E articles
61
Pittsburgh
Chicago
Minneapolis
Philadelphia
Detroit-Ann Arbor
Washington-Baltimore
Boston
Los Angeles-Santa Barbara
San Francisco Bay area
New York City
1
9
9
9
1
9
9
8
1
9
9
7
1
9
9
6
1
9
9
5
1
9
9
4
1
9
9
3
1
9
9
2
1
9
9
1
0
1
2
3
4
5
6
7
Figure 7. Firm Nanotech Entry by Region and Year, 1991-1999
62
Pittsburgh
Chicago
Minneapolis
Philadelphia
Detroit Ann Arbor
Washington-Baltimore
Boston
Los Angeles-Santa Barbara
San Francisco Bay area
New York City
C
o
m
p
u
t
e
r
/
I
n
f
o
r
m
a
t
i
o
n
p
r
o
c
e
s
s
i
n
g
/
M
u
l
t
i
m
e
d
i
a
O
t
h
e
r
E
n
g
i
n
e
e
r
i
n
g
O
th
e
r
S
c
i
e
n
c
e
B
i
o
l
o
g
y
/
M
e
d
i
c
i
n
e
/
C
h
e
m
i
s
t
r
y
I
n
t
e
g
r
a
t
e
d
C
i
r
c
u
i
t/
S
e
m
i
&
s
u
p
e
r
c
o
n
d
u
c
t
o
r
0
5
10
15
20
25
30
Figure 8. Firm Nanotech Entry by Region and Science-Technology Area