Grilichesian Breakthroughs: Inventions of Methods of Inventing and

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.

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

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.

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).

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).

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.

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.

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

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

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

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

which Sun makes freely available to software developers. Because software applications

written in Java

can run on any platform, Java

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

has diffused but no one except for Sun Microsystems which sells complementary products

has made exceptional profits from it.

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

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):

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

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 –

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.

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.

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

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.

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).

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.

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.

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.

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.

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

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.

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.

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.

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

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.

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.

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

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.

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.

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

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.

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

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

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

place).

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.

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

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.

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

that the remaining coefficients provide estimates based on deviations from average for each

region by year.

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

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.

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

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.

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.

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.

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:

= A

+ (1 – δ) X

i,t-1

where A

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.

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.

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.

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.

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.

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.

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

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