Production of University Technological Knowledge in European Regions: Evidence from Patent Data

Abstract

Acosta M., Coronado D., León M. D. and Martínez M. Á. Production of university technological knowledge in European regions: evidence from patent data, Regional Studies. This paper explores the European regional distribution of the production of new technological knowledge generated by universities, as measured by patent counts. The empirical basis for this study is a unique panel data set of 4580 European university patents from 1998 to 2004. The main findings were a strong regional and sectoral concentration of patents, and no average relation between university technological specialization and industrial specialization. Furthermore, the results suggest that variations in regional research and development funding do affect patenting activities in regions, with elasticities showing constant returns to scale, but no evidence was found regarding the industrial potential of the region encouraging the production of new university technological knowledge.

Full text

For Peer Review Only
The production of university technological knowledge in
European regions: evidence from patent data
Journal:
Regional Studies
Manuscript ID:
CRES-2007-0229.R1
Manuscript Type:
Main Section
JEL codes:
O32 - Management of Technological Innovation and R&D < O3 -
Technological Change|Research and Development < O - Economic
Development, Technological Change, and Growth, O33 -
Technological Change: Choices and Consequences|Diffusion
Processes < O3 - Technological Change|Research and Development
< O - Economic Development, Technological Change, and Growth
Keywords:
university patents, european regions, knowledge production,
technological specialization
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
peer-00526546, version 1 - 15 Oct 2010
Author manuscript, published in "Regional Studies 43, 09 (2009) 1167-1181"
DOI : 10.1080/00343400802154573

For Peer Review Only
1
The production of university technological knowledge in European
regions: evidence from patent data
Manuel Acosta (*)
Daniel Coronado
M. Dolores León
M. Ángeles Martínez
All: Universidad de Cadiz - Facultad de CC Economicas y Empresariales
Duque de Najera, 8 Cadiz Cadiz 11002
Spain
(*) Correspondence to: manuel.acosta@uca.es
daniel.coronado@uca.es
marilo.leon@uca.es
mariangeles.martinez@uca.es
First received: August 2007
Accepted: January 2008
Abstract: This paper explores the European regional distribution of the production of new
technological knowledge generated by universities, as measured by patent counts. The empirical
basis for this study is a unique panel data set of 4,580 European university patents from 1998 to
2004. Our main findings were a strong regional and sectoral concentration of patents, and no
average relation between university technological specialization and industrial specialization.
Furthermore, our results suggest that variations in regional R&D funding do affect patenting
activities in regions, with elasticities showing constant returns to scale, but no evidence was
found regarding the industrial potential of the region encouraging the production of new university
technological knowledge.
Page 1 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
2
Key words: European regions, knowledge production, university patents, technological
specialization.
JEL codes: O33, O32
La producción de conocimiento tecnológico universitario en las regiones europeas:
evidencia a partir de las patentes
En este artículo exploramos la distribución regional en Europa de la producción de
conocimiento tecnológico generado en las universidades, medido a través de las
patentes. La base empírica para este estudio consiste en un conjunto de 4.580 patentes
universitarias europeas con fecha de solicitud desde 1998 hasta 2004. Nuestros
principales resultados se resumen en una fuerte concentración regional y sectorial de
patentes, con una relación media inexistente entre especialización tecnológica
universitaria y especialización industrial. Además los modelos sugieren que variaciones
en los fondos regionales de I+D influyen en la producción de patentes universitarias,
con elasticidades que muestran rendimientos constantes a escala; sin embargo, no se
obtuvieron evidencias de la capacidad del potencial industrial de la región para
estimular la producción de conocimiento tecnológico universitario.
Palabras clave: Regiones europeas, producción de conocimiento, patentes universitarias,
especialización tecnológica.
-------------------------------------------------------
Page 2 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
3
La production de la connaissance technologique universitaire dans
les régions européennes: des preuves provenant des données sur les brevets.
Acosta et al.
Cet article cherche à examiner la distribution régionale européenne de la production de
la nouvelle connaissance technologique universitaire en fonction du nombre de brevets.
La base empirique de cette étude est un unique ensemble de données provenant d’une
enquête par panel pour 4.580 brevets déposés par des universités européennes entre
1998 et 2004. Il en a résulté principalement une forte concentration régionale et
sectorielle des brevets et aucun rapport moyen entre la spécialisation technologique
universitaire et la spécialisation industrielle. Qui plus est, les résultats laissent supposer
que la variation du financement régional pour la R et D influent sur les activités
régionales brevetables, dont les elasticités montrent des rendements d’échelle constants,
mais il n’y avait pas de preuves quant à une corrélation étroite entre le potentiel
industriel de la région et la production de la nouvelle connaissance technologique
universitaire.
Régions européennes / Production de la connaissance / Brevets universitaires /
Spécialisation technologique
Classement JEL: O33; O32
Die Produktion von universitärem technischem Wissen in europäischen
Regionen: Belege von Patentdaten
Manuel Acosta, Daniel Coronado, M. Dolores León and M. Ángeles Martínez
Abstract:
In diesem Beitrag wird die regionale Verteilung der Produktion von neuem
technischem Wissen, das von Universitäten in Europa erzeugt wird, anhand der
Anzahl der Patente untersucht. Die empirische Grundlage für diese Studie
bildet ein eindeutiger Paneldatensatz mit 4580 Patenten europäischer
Universitäten aus der Zeit von 1998 bis 2004. Unsere wichtigsten Ergebnisse
waren eine ausgeprägte regionale und sektorale Konzentration der Patente
sowie keine durchschnittliche Beziehung zwischen der technischen
Spezialisierung der Universitäten und der industriellen Spezialisierung. Darüber
hinaus weisen unsere Ergebnisse darauf hin, dass sich Abweichungen in der
Finanzierung der regionalen Forschung und Entwicklung auf die
Patentaktivitäten in den Regionen auswirken, wobei die Elastizitäten einen
konstanten Skalenertrag aufweisen. Hingegen wurden keinerlei Belege
hinsichtlich des industriellen Potenzials der Region zur Förderung der
Produktion von neuem technischem Wissen der Universitäten gefunden.
Key words:
Europäische Regionen
Page 3 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
4
Wissensproduktion
Universitäre Patente
Technische Spezialisierung
JEL codes: O33, O32
Page 4 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
5
1. Introduction
In a modern conception, universities are increasingly viewed as active contributors to
technological development and regional economic growth. The importance of universities in
encouraging regional technological change has been reported in some well-known studies of
economically successful regions (MARKUSEN et al., 1986; SAXENIAN, 1994; STERNBERG and
TAMÁSY, 1999; WEVER and STAM, 1999) and in studies that have used the national/regional
systems of innovation as a framework (LUNDVALL et al., 2002; GOLDFARB and HENREKSON, 2003;
GITTELMAN, 2006). Regional innovation systems are a suitable framework for understanding the
general innovation process at a regional scale, and they also provide some relevant clues to
assist in understanding the role of universities and the knowledge production of universities in a
regional context. Universities are assumed to accomplish a number of different functions in a
regional innovation system. FRITSCH and SLAVTCHEV (2007) emphasized some of these functions.
By conducting R&D activities, universities generate and accumulate knowledge (scientific and
technological). One of the most important channels for the transfer of academic knowledge into
the private sector is the teaching and training of students. Academic knowledge can also
disseminate through R&D cooperation with private sector firms or by providing innovation-related
services. Moreover, universities may serve as ‘‘incubators’’ for knowledge-intensive spin-offs.
Finally, scientific outputs and informal relationships can be important ways of transferring
academic knowledge to the private sector. Some of the ideas in the innovation system concept
are included in the “triple helix” model (ETZKOWITZ and LEYDESDORFF, 1997, 2000; LEYDESDORFF,
2000). The thesis of the triple helix is that the university can play an essential role in the process
of innovation, and thus strengthen knowledge-based societies. In these types of models, different
possibilities are proposed concerning the relationship between the institutional spheres—
university, industry, and government—that may help to generate alternative strategies for
Page 5 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
6
“economic growth and social transformation” (ETZKOWITZ and LEYDESDORFF, 2000). From a
regional perspective, in the triple helix, academia plays a role as a source of innovation for
regional development. ETZKOWITZ and KLOFSTEN (2005) summarized the basic elements of the
triple helix model and its importance from a regional perspective: “First, it presumes a more
prominent role for the university in innovation, on a par with industry and government in a
knowledge–based society. Second, there is a movement toward collaborative relationships
among the three major institutional spheres in which innovation policy is increasingly an outcome
of interaction rather than a prescription from government. Thirdly, in addition to fulfilling their
traditional functions, each institutional sphere also ‘takes the role of the other’ operating on a y-
axis of their new role as well as an x-axis of their traditional function. An entrepreneurial
university, taking some of the traditional roles of industry and government, is the core institution of
an Innovating Region”.
The positive effects of universities in regions may occur through a variety of university outputs
that potentially have important impacts on regional economic development, as noted previously.
In this paper we focus on one of these outputs: the generation of university patents. University
patenting may have an influence on innovation in regions in two different but complementary
ways. On the one hand, there is a potential direct contribution when a university produces useful
new patented technological knowledge with potential applications to the industrial processes. This
new patented knowledge may be transferred to private firms through licensing, by increasing
private innovation, and by the inducement of regional economic growth. Furthermore, patents are
valuable as their licensing helps to generate employment, especially among graduates, when a
spin-off firm is created to exploit the patent. On the other hand, the production of a patent may
have an indirect contribution to regional innovation due to the flow of technological knowledge
between universities and firms. This flow of knowledge can take place through a variety of
Page 6 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
7
channels of interaction between academics and firms (when reading the patent, or via direct
conversation or informal meetings with the inventors, etc.)1 The flow of knowledge has important
potential benefits for regions because of spillovers from university to industry (see for instance,
AGRAWAL and COCKBURN, 2003; ANSELIN et al., 1997, 2000; CALDERINI and SCELLATO, 2005;
FISCHER and VARGA, 2003; JAFFE, 1989; JAFFE et al., 1993; MAURSETH and VERSPAGEN, 2002;
VERSPAGEN and SCHOENMAKERS, 2000). The proliferation of a consistent literature illustrating the
importance of physical proximity for knowledge flows and for the promotion and development of
innovation, allied with the high degree of self-government enjoyed by many European regions,
makes it clear that the study of knowledge from universities is relevant not only in national or
supranational contexts but also at the regional level. However, although most of the studies that
stress the importance of universities at a regional scale have analysed the effects of scientific and
technological knowledge on the innovation of firms, little is known about the causes. For instance,
are there any particular (economic) characteristics of regions that encourage the production of
technological knowledge in their universities? As we will point out below, the bulk of the empirical
literature on the characteristics of a university’s technological knowledge and the causes of
technology generation in universities (measured by patents) is focused on countries, universities,
laboratories, or individual inventors from universities. Very few papers have aimed to analyse this
issue at a regional scale. Nevertheless, universities in the same country are not homogeneous
institutions; the financial resources, the level of regional development, the industrial structure of
regions, legal regulations, and other particular characteristics of the regions can affect the
motivation and capacities of universities to produce new technological knowledge. As BOUCHER et
al., 2003, reported in their case studies review of the role of universities in regional development,
the interactions of institutional and social factors can foster or hinder the contributions of
universities to their region’s development. This is the main argument for carrying out this research
in the case of the European regions. The purpose of this paper is to address these significant
Page 7 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
8
issues. We will consider, from a regional point of view, the magnitude, technological
characteristics, regional peculiarities, and explanatory causes of the direct contribution of
universities in European regions to the development of industrial technology. More specifically,
we will seek to answer the following research questions.
What is the distribution of the production of university technological knowledge across European
regions?
What relationship exists between the regional technological specialization of the universities and
the economic specialization?
Finally, we investigate which factors determine the regional development of university
technological knowledge, in particular R&D funding. The answer to this question is especially
relevant because the identification of these factors may help to implement regional policies that
encourage the production of industrial knowledge by universities.
The empirical basis for this study is a unique panel data set of 4,580 family patents (patents that
cover the same inventions) from the Derwent Innovation Index. We have considered the family
patent if it includes at least one European university patent, covering the years from 1998 to
2004. This is an original data set that was organized and regionalized using Nomenclature of
Territorial Units for Statistics (NUTS) II. Our empirical methodology involves two parts. First, we
undertook a descriptive analysis of regionalized data to map the technological knowledge
production generated from European universities, and to answer some of the research questions.
In the second step, we carried out an econometric analysis (a knowledge production function) to
test the importance of the classical inputs (regional university R&D) and the demand factors (the
industrial potential of the regions) in the process of generating new university technological
knowledge across European regions.
Page 8 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
9
This paper contributes to the existing literature in several ways. First, it presents an overview of
how university patenting is spread across European regions. Second, although some studies
focus on the determining factors in the production of knowledge (measured by patent counts) by
countries, universities, etc., there are none that discuss the European regions as a whole. Third,
most of the European regions have enough autonomy to organize their own systems of
innovation. Therefore, a better understanding of the characteristics and causes of the
technological generation of new knowledge in universities may help to define policies that support
regional technological development.2
The paper is organized as follows. In the first section, we review some empirical papers focusing
on university patents and their determining factors. Next, using patent counts, we explore some of
the characteristics of the production of new knowledge by universities in the European regions. In
the following sections, using our econometric specification, we test how regional spatial inputs
and demand factors are related to the production of university patents. The main conclusions and
some policy implications are drawn at the end of the paper.
2. Literature review
Previous research on the generation of university patents and their explanatory causes has
mainly related to US universities. Recently, there has been growing concern about this issue in
Europe, but the evidence remains scarce. Little is known about the effects of regional peculiarities
on the production of patents by universities. In the following paragraphs, we highlight some of the
conclusions of the recent empirical literature.
Page 9 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
10
One of the more relevant studies on the characteristics and explanatory causes of university
patenting is by HENDERSON et al., 1998, who compared the US universities’ patents for the period
1965–1988. Regarding explanatory causes of the evolution of university patents, they
emphasized three essential aspects: the legal framework, or changes in the federal laws that
facilitate patent applications by universities (MOWERY et al., 2001, put this finding in doubt);
increases in industrial funds destined to support university research; and the increase in the
numbers of interface centres and institutions. COUPÉ, 2003, estimated a production function for
university patents by means of empirical count models, in which the principal explanatory factors
were academic expenditures on R&D, and the institutional factors considered previously by
HENDERSON et al., 1998. The results of this study confirmed the evidence on the institutional
effects and the significant influence of R&D expenditure on the output of university patents. In
addition, COUPÉ, 2003, found some indications of constant returns to scale at the institutional
level. However, once fixed effects are controlled for, he found much smaller coefficients,
indicating decreasing returns to scale. PAYNE and SIOW, 2003, discussed the effect of federal
funding on university patents. After analysing the data under different specifications, they
suggested that when universities receive more funding, they will produce more patents at the
margin. As possible explanatory causes of the mechanisms by which universities develop
patents, these authors suggested know-how (the accumulation of previous patents or experience
in the particular field), the personnel dedicated to technology transfer, and contractual links with
companies that patent. In two similar papers, FOLTZ et al., 2000, 2003, examined the production
of patents in ag-biotech sectors for US universities. In their first study, they used a static count
data model to stress the role of the technology transfer offices and star scientists. In their second
paper, they estimated a panel count data model of university ag-biotech patent production with a
sample of 127 universities, 65 if which had received at least one ag-biotech patent between 1994
and 1999. They found strong evidence of a correlated dynamic effect in which patenting
Page 10 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
11
experience helps to produce more patents and that patent production is enhanced by the overall
university propensity to patent (patenting culture effects), a strong land grant effect, and a
biological science research funding effect. They did not find convincing evidence that university
reliance on industry financing increases patent production (there were no industry effects
present).
As we emphasized above, the analysis of the European case began more recently than the
analysis of the US situation, and the empirical evidence for Europe is not extensive. The recent
review by GEUNA and NESTA, 2006, of European academic patenting concluded that the broadly
defined research area of biotechnology and pharmaceuticals tends to be an area of extremely
high university patenting activity across (European) countries. Second, historical developments in
Italy and Germany seem to support the view that university patenting is not a new phenomenon.
The authors also suggested that the rapid rise of academic patenting was driven more by the
growing technological opportunities in the biomedical sciences than by policy changes affecting
the universities’ rights to own patents arising from publicly funded research. In the next
paragraphs, we review what the empirical European literature says about university patenting and
its determining factors.
SARAGOSSI and VAN POTTELSBERGHE, 2003, carried out a descriptive analysis of patenting activity
in six Belgian universities. They found an increase in patents, which they attributed to two major
changes: the new technological opportunities resulting from research activities related to the
biotechnology sector, and an increased propensity to patent technologies developed by Belgian
universities (also related to more effective technology transfer offices). BALDINI et al., 2006,
analysed a set of 637 patent applications filed at the Italian Patent and Trademark Office, the
European Patent Office, and the US Patent and Trademark Office between 1965 and 2002, with
Page 11 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
12
at least one applicant belonging to the official list of higher education institutions. Their empirical
results showed that, in the last 10 years, the number of Italian university patent applications, in
Italy and/or abroad, rose substantially; patenting activities almost tripled in universities with an
internal Intellectual Property Right regulation, after controlling for several universities’
characteristics, previous patenting activity, and time trends. Each time a university creates its own
patent regulation, there is a 9% increase in the likelihood that universities without any internal
patent regulation will adopt one. Furthermore, consistent with previous studies on academic
patenting, BALDINI et al., 2006, found greater patenting activity in the north of Italy, where there is
a higher level of industrial development. In a broad paper, in which the authors considered a
patent as an input and an output, AZAGRA-CARO et al., 2003, estimated a patent production
function using data on patent applications from 43 departments of the Polytechnic University of
Valencia (Spain) from 1991 to 2000. Their results showed that the aggregate R&D expenditure
has a positive effect on patents, but they found decreasing returns to scale. When they used R&D
split by source of funding, they found that government and industry funding has a significant and
positive effect, and that the public funding is more important for patenting than private funding.
The internal characteristics of departments are relevant in patent generation.
We have briefly reviewed some of the key papers on the determinants of university patents, but
none of these studies focused on the regional level. To our knowledge, only AZAGRA-CARO et al.,
2006, discussed the analysis from a regional perspective. They built a university patent
production function to identify the factors determining the generation of patents in 17 Spanish
autonomous regions (NUTS-II). They used a sample of 1,479 patents (1,398 national and 81
international) over a time span of 14 years (from 1988 to 2001). As in previous research, they
found a significant positive relation between the number of university patents and academic R&D
in all estimations (they found that academic R&D was the only significant determinant of
Page 12 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
13
international university patents), but their elasticities were extremely high compared with previous
papers applied in other contexts. The authors explained this as being the result of using different
units of observation (regions instead of universities) and the inclusion of all the funds (not only
public funding). Other significant variables to explain the production of national patents are control
variables (trend), and an index of technological distance between university patenting and other
institutions. They did not find evidence of the effects of variables such us the number of university
researchers, university structure, or the joint research structure (specific economic variables
reflecting demands factors from the regions were not included in this paper). This was an
interesting attempt to analyse the issue at a regional scale, but as a drawback it should be noted
that although they used a panel for 14 years, the data included only Spanish regions, which
represent very heterogeneous spatial units, making it difficult to obtain econometrically consistent
results. That is, the number of universities and the number of university patents vary significantly
across Spanish regions (which probably generates spatial heterogeneity biases). As far as we are
aware, empirical studies on university patenting at a macro regional scale in Europe are
nonexistent.
3. University patenting in the European regions: an overview
This section presents a description of regional university patenting activity across the European
Union. In our empirical analysis, we use 4,580 European university patents (obtained from the
Derwent Innovation Index) related to 202 European regions and 378 universities for the period
from 1998 to 2004.3 As regions for our analysis, we chose the territorial units from Eurostat in
each country at the NUTS II level of aggregation. As is well known, a NUTS II is a territorial unit
with some degree of administrative and policy authority.
Page 13 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
14
In order to regionalize the university patents, we followed two steps. First, we identified all the
universities for each region. In the second step, we gathered all the patents from each university
situated in the same region. A patent was assigned to a university (and in consequence to a
region) when the name of the university appeared in the list of applicants of the patent, so the
name of the university was the criterion for detecting a university patent.4 When more than one
university appeared on the list of applicants, we assigned the patent to the first applicant.
Therefore, what we are considering in this paper is a university-owned patents (those that have at
least a university as one of the applicants), and not the production of a patent by an inventor
(researcher/professor) from a university. The review by VERSPAGEN (2006) includes some figures
for six European countries where the importance of each category of patents differs substantially
among countries. As GEUNA and NESTA, 2006, reported in a recent review of the literature on
university patenting in Europe, the number of university-invented patents (patents with at least
one inventor from a university but where the owner of the patent is a firm) is higher than the
number of patents owned by universities. However, the data in our sample are dramatically
different to those reported in the paper by GEUNA and NESTA, 2006, for two reasons. First, they
summarized the available evidence for five European countries only (they pointed out that there
were no other data available), whereas we include all European regions. Second, the papers
reviewed by GEUNA and NESTA, 2006, considered a longer time period than our paper does. In
the next paragraphs, we describe the data and attempt to answer some of the research
questions.
3.1 Spatial distribution of the production of technological knowledge across European regions
The regional distribution of university patent counts from 1998 to 2004 is shown in Fig. 1. Of the
202 regions, 54 have no university-owned patents; 42 regions have between one and five such
Page 14 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
15
patents; 45 have between six and 15; 35 have 16 to 50; 16 have 51 to 100; and 10 regions have
more than 100 university-owned patents. On average, there are 3.2 patents per region for the
whole period. University patents appear to be concentrated in the regions of the UK, with other
important clusters occurring in north Europe (Belgium and Holland) and in two regions of France
(Rhône-Alpes and Île de France). It is worth remarking that the 54 regions where there is no
university patenting activity are scattered across the European Union.5
Figure 1. Regional distribution of university patents
A summary of the main statistics from the whole sample is reported in Table 1. From this Table, it
is evident that the number of university patents increased from 390 to 936 during the period under
examination. The coefficient of variation and the Herfindahl index reveal no substantial changes
in the spatial concentration of university patenting activities from 1998 to 2004.
Table 1. European university-owned patents (1998–2004)
A closer look at the top 10 regions with the highest number of university patents confirms that the
production of university technological knowledge is highly concentrated in a few regions. As is
shown in Table 2, more than 25% of all the university patents are concentrated in five regions,
and 10 regions account for 40% of all university patents (but only 10.5% of the population). Two
regions in the UK are the most active in university patent production, namely Inner London; and
Berkshire, Buckinghamshire, and Oxfordshire. The last column of Table 3 presents the number of
university patents normalized by the size of the geographical unit, expressed per number of
inhabitants.
Page 15 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
16
Table 2. Regions with high numbers of European university-owned patents (1998–2004)
3.2 Technological specialization vs. economic specialization in regions
University-owned patents can be analysed by looking at the distribution across industrial sectors.
In order to assign a patent to an industrial sector, we applied an economic concordance table,
which was recently developed by SCHMOCH et al., 2003, between the four-digit of the International
Patent Classification (IPC) codes and the 44 industrial sectors. Using this equivalence table, we
converted the original IPC data (at the four-digit level) to the Classification of Economic Activities
in the European Community (NACE) at the two-digit level based on the industrial sector where
the patent originated. To avoid duplication, when a patent included more than one IPC code, we
considered only the primary IPC category.
By applying the concordance table to the 4,580 patents, we obtained an initial picture of the
sectoral distribution of university patenting (see Table 3, which shows the five most dynamic
sectors for 1998 to 2004). As shown in this table, there is a high degree of concentration in a few
activities, as five industrial sectors account for 69.5% of all patents. The most active sector was
patents related to pharmaceuticals, which accounted for 39.1% of all university patents. This
sectoral distribution is similar to that reported by GEUNA and NESTA, 2006, in their review of
papers on university patenting in Europe. The last column of Table 3 presents the most dynamic
regions in each sector; note that Inner London is at the top in three of the five sectors.
Table 3. Sectors with the highest number of university-owned patents (1998-2004)
Page 16 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
17
In addition to this previous general view, Table 4 shows the specialization in the five most
dynamic sectors of the 20 regions with the highest number of university patents. As a measure of
sectoral specialization, we use the index of Revealed Technological Advantage (RTA), which
provides information on the specialization of each region compared to the whole set of regions.
The index of technological specialization is calculated as RTA = (Pij/ ΣPj)/ (ΣPi/ ΣP), where Pij is
the number of university patents in sector i of region j; Pj is the number of university patents in
region j; Pi is the number of patents in sector i across all regions; and P is the total number of
university patents for all the regions. The index is greater than one when the region has a
comparative advantage in that sector and is less than one when it has a disadvantage. For
instance, Comunidad Valenciana (ES) and Région Wallonne (BE) are the two regions with the
highest RTA in sector 10 Basic chemical; Utrecht (NL) and Tübingen (DE) have the highest RTA
in Pharmaceuticals, etc.
Table 4. Technological specialization of the 20 regions with the highest numbers of university patents
In order to identify to what extent the technological specialization of the European regions is
associated with their productive specialization pattern, we used an index based on sectoral
employment, which is similar to productive specialization, for each region. Once both indexes
were calculated (the industrial specialization and the above technological specialization index for
all the sectors in which there was at least one patent), the correlation between them was obtained
(Table 4, last column). This result shows that a high intraregional correlation is found in some
regions (intraregional correlation)—such as Berkshire, Buckinghamshire, and Oxfordshire
(0.812), and Derbyshire and Nottinghamshire (0.617) in the UK, or Tübingen from DE (0.910).
However, there are also some regions with no relation or with a negative correlation. However, if
we calculate the interregional correlations in the main sectors, they confirm that, on average, the
Page 17 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
18
relation between technological specialization and industrial specialization is non-existent: the
coefficients are 0.018 in the Basic chemical and Pharmaceuticals sectors (calculated with 130
observations); –0.085 in the Office machinery and computers sector (with 64 observations); and
0.025 in the Medical equipment and Measuring instruments sectors (with 112 observations).
We put forward two hypotheses to explain this lack of relation between industrial specialization in
regions and technological university knowledge specialization within and across some regions.
The first hypothesis is that a situation may exist in which the universities are producing
technological knowledge that does not cover all of the industrial specializations in the regions
where they are situated, but instead they are attending to specific demands for high technology
from their region or another region. The second hypothesis involves a worse situation:
universities are producing technological knowledge that has nothing to do with the industrial
structure of the regions because they are focusing upon academic objectives only (such as
increasing academic publications). This is an important unresolved issue that requires more
research.
4. Empirical framework
This section aims to establish an econometric framework to test the importance of the inputs and
the regional demand factors in the process of generating new university technological knowledge,
measured by patent counts. In particular, we attempt to test how university R&D funding (the
main input) influences the university patent output, and at the same time to determine the
potential role of private demand in regions in encouraging the creation of new technological
knowledge in universities.
Page 18 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
19
4.1 The model
In order to study the relationship between outcomes measured by patent counts and R&D
university funding, and other determining characteristics of the regions, we regress patents
measured on funding and regional characteristics using regional level data as follows:
KPi = f(Ri,CHi, ui),
where KP is the technological new knowledge produced in universities in a region i, and R is the
amount of university regional funding. R is produced as a lagged input because, as is well known,
it takes time before R&D expenditure yields an output in the form of a patent. In addition, we
consider a set of variables that reflect the industrial characteristics of the environment and that
can affect the production of technological knowledge by a university. The acceptance of an
interactive model of innovation assumes that the knowledge flows are bidirectional between the
academic and industrial sectors. The existence of informal contacts between academic scientists
and company researchers can generate mutual benefits for both that could have a positive impact
on the universities’ portfolio of patents. These factors are captured by CH, a vector of variables
that picks up the potential demand for university technology in the region i. The term u represents
unobservable regional differences. Note that this standard departure model is a “knowledge
production function” (GRILICHES, 1979). Essentially, the model involves a neoclassical production
function where knowledge is measured by means of a proxy variable (e.g., patents) and the
inputs incorporate university R&D expenses, together with other spatial variables. As we pointed
out above, a similar theoretical base was used by AZAGRA-CARO et al., 2006; COUPÉ, 2003; FOLTZ
et al., 2000; FOLTZ et al., 2003; PAYNE and SIOW, 2003. However, our framework is substantially
different because the models in these papers included only the supply-side factors (funding and
Page 19 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
20
university characteristics), whereas we also introduce potential regional demand-side effects as a
factor that may encourage or hinder the production of university patents.
4.2 Variables and econometric specification
Our dependent variable is the count of university patents, used as a measure of the production of
new university technological knowledge in regions.
Our explanatory variables are as follows.
- University funds. We include R&D university regional funding measured as millions of
Purchasing Power Standards (PPS) in 1995 prices (R variable). We lag the funding measure in
separate specifications by three years (following PAYNE and SIOW, 2003) and two years (FOLTZ et
al., 2003) in order to prevent endogeneity. Furthermore, this allows us to compare the results.
- Potential demand for university technology. As reviewed above, some empirical papers without
a specific focus on regions have stressed the importance of the demand side in encouraging
university patenting activities. In particular, SARAGOSSI and VAN POTTELSBERGHE (2003)
highlighted the new technological opportunities resulting from research activities related to the
biotechnology sector, and BALDINI et al. (2006) found greater patenting activity in the north of
Italy, where there is a higher level of industrial development. From a regional viewpoint, the
recent paper by KOO (2007) tackled a related topic: the endogenous relationship between
spillovers and the role of regional and industrial attributes (the demand side). This paper provided
substantial evidence that agglomeration, industry structure, small establishment, and local
competition play important roles in the localization of technology spillovers. On the other hand, it
Page 20 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
21
should be borne in mind that there are serious doubts as to whether the potential demand for
university knowledge remains local for several sectors. In particular, this is the case for some
centres or universities of “excellence”. For instance, COOKE’S (2004, 2005) comprehensive
research on bioscience megacentres reported some instances where academic and public
laboratory scientists evolved “dynamic capabilities” that attracted entrepreneurs from different
locations to engage in tacit knowledge exchange. However, once firms are located near
universities, those companies will get early access to local inventions (for example, university
patents) generating what COOKE (2005) referred to as “precipitatory knowledge”.
Our main interest including this variable is to question whether regional demand is relevant in
promoting university patenting. In order to capture the regional demand-side factors that can
determine the production of university patents in universities, we include the variable gross
industrial domestic product (GIP) in the models as a proxy of the industrial capacity of the region,
and, consequently, of the regional potential to demand university technology. This variable also
accounts for the (industrial) size of regions, preventing spatial heterogeneity bias.
- Spatial fixed effects. We consider national dummies in order to capture the different political and
institutional contexts that can encourage university patenting activity. Particularly relevant in this
respect is the legal situation regarding university patents in Europe. This issue is important for
two reasons. On the one hand, the legal context is an institutional mechanism that enables (or
hinders) the development of university knowledge production. It is important to note in this regard
that each European country has its own legal system of patents (see OECD, 2003, for details).
Therefore, the legal situation is an element of the national innovation system that has to be taken
into account as a cause of differences in the production of university knowledge among regions in
different countries. BALDINI et al. (2006) reported some studies showing that patent policies are
Page 21 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
22
among the determinants of intercountry and interorganizational differences. On the other hand, in
our empirical analysis, we consider a database of “university own patents”. This is one of the
categories of patents that involve university knowledge, as we mentioned earlier, with the other
type including patents applied by inventors employed by universities. The importance of one or
the other type of patent differs between countries depending on the patent law of each (see
VERSPAGEN, 2006).
We also consider regional spatial effects, but regional dummies are impossible to include
because of a lack of degrees of freedom, so we consider only a single regional variable that takes
a value of one if the region i is an Objective 1 region (producing less than 75% of the average
GDP per capita of the European Union) and 0 otherwise. Objective 1 regions are those that are
lagging in terms of their development, and they normally have a weak regional system of
innovation and little capacity to demand technology from a university.
- Temporal effects. Finally, we include dummy variables for years to control for fixed temporal
effects.
Because we are dealing with a count variable as the dependent variable, the nature of the data
suggests the formulation and estimation of a counting (Poisson or negative binomial) model to
detect the intensity of the technological knowledge creation. As in most previous studies, our
baseline specification assumed that the dependent variable followed a Poisson distribution,
where the set of regressors is given by X = (R, GIP, OBJ1, YEAR DUMMIES, SPATIAL
DUMMIES).
Page 22 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
23
The standard procedure for computing the estimators is the Newton–Raphson iterative method.
Convergence is ensured because the logarithmic likelihood function is globally concave.
However, one restriction of the Poisson model is that it assumes that the mean and variance of
the dependent variable are equal, so this framework breaks down when the data are
overdispersed; that is, when the variance of the dependent variable is greater than the mean, a
requirement that cannot always be met in practice. If the data show overdispersion, the standard
errors of the Poisson model will be biased toward the low end, giving spurious high values for the
t-statistics (CAMERON and TRIVEDI, 1986). Because equality of the mean and variance does not
hold in the data, we consider a number of different models that do permit overdispersion. The
most common formulation for taking overdispersion into account is the negative binomial model
(NB2, in the terminology of CAMERON and TRIVEDI, 1986, which assumes that the variance is a
quadratic function of the mean (the approach for the density function, logarithmic likelihood
function, first-order conditions, etc., is similar to the above, and is discussed in detail in CAMERON
and TRIVEDI, 1998).
4.3 Data
Our sample contains 4,580 European university patents from the European Patent Office
(obtained from the Derwent Citation Index) related to 202 European regions and 378 universities
in the period from 1998 to 2004 (this sample is the same as described above -in Section 3-,
where we explained how the data were obtained). Unfortunately, however, we have no data for
the explanatory variables for all regions (with the exception of the dummy “Obj1”). University R&D
funds and regional industrial domestic product (obtained from EUROSTAT) are only available for
some regions and years. Another problem is that the panel is unbalanced (the regions in one year
Page 23 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
24
are not exactly the same as those reported in the following years). Table 5 contains the data used
in our models.
Table 5. Number of regions with data available for the explanatory variables
4.4 Results
Our baseline specification assumed that the dependent variable followed a Poisson distribution,
but the presence of overdispersion led us to consider negative binomial models as well. As is
shown in Table 6, four models have been estimated with all the variables described above,
including those to control for spatial and temporal fixed effects. For the first group (Models I and
II), we have lagged the inputs by three years, and for the second group (Models III and IV), we
have lagged the inputs by two years. Models I and III include Poisson specifications with robust
standard errors to avoid spurious high values for the t-statistics. Models II and IV include negative
binomial specifications.
The results in all the models are quite similar. Given that the university funds are expressed in
logarithmic form, the coefficients for the R variable provide estimates of elasticities. As shown in
Table 6, these elasticities for university funds present coefficients of around one (0.93–1.04),
indicating constant returns to scale. This result is in line with the first part of the paper by COUPÉ,
2003, who found constant returns to scale in his analysis of university patents (or decreasing
returns after controlling for fixed effects), although he used universities as a unit of observation in
his papers. PAYNE and SIOW, 2003, also obtained decreasing returns, but they considered only a
part of university funding. Finally, our results contrast strongly with those obtained by AZAGRA-
CARO et al., 2006, whose paper focused on Spanish regions and found extremely high
elasticities, which were attributed to a high level of aggregation for these spatial units.
Page 24 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
25
We did not find evidence of the possible effects that the industrial potential may have on the
production of university technological knowledge. The variable GIP is not relevant in all the
specifications. This is an important issue because it indicates that the production of university
technological knowledge does not appear to be supported by potential industrial demand in
regions, a surprising result that requires further investigation. On the one hand, it is possible that
universities are responding to global demands coming from outside the regional borders and, as
a consequence, the role of the economic regional demand is limited in encouraging the
production of university patenting. However, on the other hand, GIP may not be the best variable
to represent regional technological demand, and future research may need to consider better
proxies.
The spatial fixed effects captured by the country dummies are all significant and with wide
differences between them (note in Table 6 that the base category is the UK). These results
emphasize the importance of the political and institutional context in producing differences in the
production of university technological knowledge in regions. Finally, the dummy variable Obj1 is
not significant, but, as we comment below, this is because its effect overlaps with that of the
country dummies.
Although the results of the models are similar, it is necessary to determine which one is better. In
order to compare the Poisson models with the NB models, two overdispersion tests were applied:
a t-test on the overdispersion parameter alpha, and an LR test comparing the log likelihood of the
Poisson models against the log likelihood of the NB models. Both tests select the NB models over
the Poisson models (see the statistics in Table 6). On the other hand, there are no substantial
differences between the models with three or two years of lags, probably because in some
Page 25 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
26
regions two years are needed to translate resources into patents, but in other regions an extra
year is required. Note, however, that the models with two-year lags present a slightly better fit.
Table 6. Effects of regional university R&D funds and industrial potential of regions on university patents
The final issue is to determine how robust our results are. We have run other regressions to
address this issue, as follows.
First, we carried out a sensitivity analysis with some alternative specifications, dropping some
variables from the base models and estimating new models with some variables omitted. The
results can be summarized as follows. Elasticities of around one (0.99–1.10) are obtained in all
the models if we consider only R as the explanatory variable (always with two or three lags).
Taking out fixed temporal and spatial effects from the models does not greatly change the original
elasticities; however, more variability appears, as indicated by coefficients with values between
0.87 and 1.11). The coefficient for GIP is not significant in most of the specifications. The variable
OBJ1 becomes significant when we remove the country dummies from the models, suggesting an
overlap effect, but the fit is considerably worse.
Second, we obtained intragroup estimations (obtained not by pooling the data for all the available
years, but by taking a sample of regions for every year) with two and three lags for the
explanatory variables, and ran both Poisson and negative binomial models. The results for these
models were not substantially different from the previous models. Again, elasticities were around
one for R, coefficients for GIP were not significant, and there were considerable spatial
differences among countries. Table 7 summarizes all the results.
Table 7. Summary of results
Page 26 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
27
Conclusions
This paper was a first attempt to understand the distribution of technological knowledge
generated in universities, measured by patent counts, at a regional level in Europe. Our research
design involved two parts. First, a descriptive analysis was carried out to analyse the spatial
distribution of university patents. Our main findings from this part of the analysis were a strong
regional and sectoral concentration of patents, and no average relation between technological
specialization and industrial specialization. In the second part of the research, an econometric
analysis was undertaken to identify which factors determine the production of university patents.
However, determining the causal links between possible explanatory causes and university
patenting is a difficult task, because regional heterogeneity, endogeneity, correlations between
inputs, and spatial/time fixed effects have to be taken into account. Controlling for all these
difficulties in our models, we found that variations in regional R&D funding do affect patenting
activities in regions, with elasticities showing constant returns to scale. Furthermore, all the
specifications showed significant country dummies, revealing the importance of the institutional
context in producing university patents in regions.
Our empirical analysis raises some questions that are relevant to the debate regarding regional
innovation policies at different levels. According to our data, universities in regions play an
important role in generating technological knowledge (patents) with potential application to the
market. Promoting this role should be a priority in order to foster entrepreneurial universities that
strengthen regional innovation systems. European institutions and national governments should
have a strong interest in promoting such a role for three reasons. First, there is a large skewed
spatial concentration of patents (10 regions have more than 100 university patents, but more than
Page 27 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
28
50 regions have no university patents); amending this imbalance is recommended in order to
avoid wider regional disparities in the future. Second, money is important; there is a direct relation
between the amount of university R&D resources and university performance in terms of numbers
of university patents (with constant returns to scale). Therefore, financial grants are essential to
promote entrepreneurial universities. Another implication of this finding is that investing extra
money in university R&D in large regions is as effective in stimulating the production of university
patenting as is investing extra money in small regions. Third, the institutional context is relevant to
encouraging the production of university patenting. In particular, each country has its own legal
framework that generates differences in university patent productions. It may be necessary to
consider a more (supranational) homogeneous legal system for industrial property in universities.
Finally, we found an intraregional correlation between technological specialization in universities
and regional economic specialization only in a few regions. This is probably a result of the fact
that universities in many regions are involved in projects unrelated to the technological
necessities of their environment. Therefore, there are some potential steps for the regional
governments with competencies in R&D to consider, such as reinforcing (or establishing) the
mechanisms to encourage interactions with local firms, and promoting the kind of university
technological knowledge that can support industries in the same regions where the universities
are located.
Page 28 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
29
Acknowledgments
The authors would like to thank two anonymous referees for their useful comments that improved
the final version of this paper. The authors are grateful for the financial assistance provided by the
Ministerio de Educación y Ciencia (SEJ 2005-08972/ECON) and the Junta de Andalucía,
Consejería de Innovación, Ciencia y Empresa (P06-SEJ-02087).
Page 29 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
30
NOTES
1 Note that other important channels of interaction such us employment of graduates by firms or
training of firm members have nothing to do with a patent.
2 For instance, in the US, the state governments and regions have promulgated specific policies
and programmes to exploit research universities as drivers of innovation and high technology-
based economic growth. Universities and their research spending are essential assets and the
foundation for high-tech regional growth initiatives (SONKA and CHICOINE, 2004).
3 It should be noted that the number of European university patents is considerably lower than the
number of national ones. The reason for gathering only European patents in our sample is to
avoid the national distortions arising from the different patent application requirements in different
countries. Furthermore, the higher costs for a national patent compared to a European patent
suggest that we are dealing with valuable technological knowledge with real potential
applications.
4 We have considered that a patent belongs to a university when the university’s name appears in
the list of applicants. To avoid problems with the university name, we considered that institutions
included in the European Indicators, Cyberspace, and the Science–Technology-Economy –
System (EICSTES) Project and in the Worldwide Web of Universities (www.univ.cc, web site with
links to 7,884 universities in 190 countries).
5 As we pointed out in the description of the data, note that we are considering university-owned
patents in this paper. Therefore, this spatial distribution does not necessarily coincide with other
criteria (for example, it excludes patents that are invented by academic researchers but where the
university does not appear on the patent’s applicant list).
Page 30 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
31
References
AGRAWAL, A. and COCKBURN, I. (2003) The anchor tenant hypothesis: exploring the role of large,
local, R&D-intensive firms in regional innovation systems, International Journal of Industrial
Organization 21, 1227-53.
ANSELIN, L., VARGA, A. and ACS, Z. J. (1997) Local geographic spillovers between university
research and high technology innovations, Journal of Urban Economics 42, 422-48.
ANSELIN, L., VARGA, A. and ACS, Z. J. (2000) Geographic and sectoral characteristics of academic
knowledge externalities, Papers of Regional Science 79, 435-43.
AZAGRA-CARO, J. M., FERNANDEZ-DE-LUCIO, I. and GUTIERREZ-GRACIA, A. (2003) University
patents: output and input indicators... of what?, Research Evaluation 12, 5-16.
AZAGRA-CARO, J. M., YEGROS-YEGROS, A. and ARCHONTAKIS, F. (2006) What do university patent
routes indicate at regional level?, Scientometrics 66, 219-30.
BALDINI, N., GRIMALDI, R. and SOBRERO, M. (2006) Institutional changes and the commercialization
of academic knowledge: A study of Italian universities' patenting activities between 1965 and
2002, Research Policy 35, 518-32.
BOUCHER, G., CONWAY, C. and VAN DER MEER, E. (2003) Tiers of Engagement by Universities in
their Region's Development, Regional Studies 37, 887-97.
CALDERINI, M. and SCELLATO, G. (2005) Academic research, technological specialization and the
innovation performance in European regions: an empirical analysis in the wireless sector,
Industrial and Corporate Change 14, 279.
CAMERON, A. and TRIVEDI, P. (1986) Econometrics models based on count data: comparisons and
applications of some estimators and tests Journal of Applied Econometrics 1, 29.
CAMERON, A. and TRIVEDI, P. (1998) Regression Analysis of Count Data. Cambridge University
Press, Cambridge.
Page 31 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
32
COOKE, P. (2004) Regional Knowledge Capabilities, Embeddedness of Firms and Industry
Organisation: Bioscience Megacentres and Economic Geography, European Planning Studies
12, 625 - 641.
COOKE, P. (2005) Rational Drug Design, the Knowledge Value Chain and Bioscience
Megacentres, Cambridge Journal of Economics 29, 325-341.
COUPÉ, T. (2003) Science Is Golden: Academic R&D and University Patents, Journal of
Technology Transfer 28, 31-46.
ETZKOWITZ, H. and KLOFSTEN, M. (2005) The Innovating Region: Toward a Theory of Knowledge-
Based Regional Development, R&D Management 35, 243-255.
ETZKOWITZ, H. and LEYDESDORFF, L. (1997) University and the Global Knowledge Economy. A
Triple Helix of University-Industry-Government Relations.Pinter Publishers, London.
ETZKOWITZ, H. and LEYDESDORFF, L. (2000) The Dynamics of Innovation: From National System
and ‘Mode 2’ to a Triple Helix of University-Industry-Government Relations, Research Policy 29,
109-123.
FISCHER, M. and VARGA, A. (2003) Spatial knowledge spillovers and university research: Evidence
from Austria, Annal of Regional Science 37, 303-22.
FOLTZ, J., BARHAM, B. and KIM, K. (2000) Universities and agricultural biotechnology patent
production, Agribusiness 16, 82.
FOLTZ, J. D., KIM, K. and BARHAM, B. (2003) A Dynamic Analysis of University Agricultural
Biotechnology Patent Production, American Journal of Agricultural Economics 85, 187-97.
FRITSCH, M. and SLAVTCHEV, V. (2007) Universities and Innovation in Space, Industry and
Innovation 14, 201-218.
GEUNA, A. and NESTA, L. J. J. (2006) University patenting and its effects on academic research:
The emerging European evidence, Research Policy 35, 790-807.
Page 32 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
33
GITTELMAN, M. (2006) National Institutions, Public–Private Knowledge Flows, and Innovation
Performance: A Comparative Study of the Biotechnology Industry in the Us and France,
Research Policy 35 1052-1068.
GOLDFARB, B. and HENREKSON, M. (2003) Bottom-up Versus Top-Down Policies Towards the
Commercialization of University Intellectual Property, 32.
GRILICHES, Z. (1979) Issues in assessing the contribution of research and development to
productivity growth, Bell Journal of Economics 10, 92-116.
HENDERSON, R., JAFFE, A. B. and TRAJTENBERG, M. (1998) Universities as a source of commercial
technology: a detailed analysis of university patenting, 1965-1988, The Review of Economics and
Statistics 80, 119-28.
JAFFE, A. (1989) Real effects of academic research, American Economic Review 79, 957-70.
JAFFE, A., TRAJTENBERG, M. and HENDERSON, R. (1993) Geographic localisation of knowledge
spillovers as evidenced by patent citations, Quarterly Journal of Economics 108, 577-98.
KOO, J. (2007) Determinants of Localized Technology Spillovers: Role of Regional and Industrial
Attributes, Regional Studies 41, 995-1011.
LEYDESDORFF, L. (2000) The Triple Helix: An Evolutionary Model of Innovations, Research Policy
29, 243-256.
LUNDVALL, B.-Å., JOHNSON, B., ANDERSEN, E. S. and DALUM, B. (2002) National Systems of
Production, Innovation and Competence Building, Research Policy 31, 213-231.
MARKUSEN, A. R., HALL, P. and GLASMEIER, A. (1986) High Tech America: the what, how, where
and why of the sunrise industries. Allen and Unwin, Boston.
MAURSETH, P. B. and VERSPAGEN, B. (2002) Knowledge Spillovers in Europe: A Patent Citations
Analysis, Scandinavian Journal of Economics 104, 531-45.
Page 33 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
34
MOWERY, D. C., NELSON, R. R., SAMPAT, B. N. and ZIEDONIS, A. A. (2001) The growth of patenting
and licensing by U.S. universities: an assessment of the effects of the Bayh-Dole act of 1980,
Research Policy 30, 99-119.
OECD (2003) Turning Science into Business. Patenting and Licensing at Public Research
Institutions.OECD, Paris.
PAYNE, A. and SIOW, A. (2003) Does Federal Research Funding Increase University Research
Output?, Advances in Economic Analysis & Policy 3, 1018.
SARAGOSSI, S. and VAN POTTELSBERGHE, B. D. L. P. (2003) What Patent Data Reveal about
Universities: The Case of Belgium, Journal of Technology Transfer 28, 47-51.
SAXENIAN, A. (1994) Regional advantage: culture and competition in Silicon Valley and Route
128. Harvard University Press, Cambridge.
SCHMOCH, U., LAVILLE, F., PATEL, P. and FRIETSCH, R. (2003) Linking technology areas to
industrial sectors, Final Report to the European Commission, DG Research.
SONKA, S. T. and CHICOINE, D. L. (2004) Value and University Innovation, American Journal of
Agricultural Economics 86, 1337-44.
STERNBERG, R. and TAMÁSY, C. (1999) Munich in Germany’s N 1 high technology region:
empirical evidence, teoretical explanations and the role of small firm/large firm relationships,
Regional Studies 33, 367-77.
VERSPAGEN, B. (2006) University Research, Intellectual Property Rights and European Innovation
Systems Journal of Economic Surveys 20, 607-632.
VERSPAGEN, B. and SCHOENMAKERS, W. 2000. "The Spatial Dimension of Knowledge Spillovers in
Europe: Evidence from Firm Patenting Data." Technische Universiteit Eindhoven: The
Netherlands.
WEVER, E. and STAM, E. (1999) Cluster of High Tecnology SMEs: The Ductch Case, Regional
Studies 33, 391-400.
Page 34 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
35
Table 1. European university-owned patents 1998–2004
1998 1999 2000 2001 2002 2003 2004 1998–04
No. of Patents 390 402 606 639 710 897 936 4,580
Mean 1.9 1.9 3.0 3.1 3.5 4.4 4.6 3.2
Maximum 37 42 74 57 74 79 47 410
Minimum 0 0 0 0 0 0 0 0
Std.Dev. 4.3 4.6 7.7 6.8 8.0 9.2 8.1 7.2
Coeff. of Variation (*)
2.2 2.3 2.5 2.1 2.2 2.0 1.7 2.2
Herfindahl (**) 0.030 0.032 0.038 0.028 0.031 0.026 0.020
0.026
Obs. 202 202 202 202 202 202 202 202
(*) C.V. = Std.Dev/Mean
(**) The Herfindahl index of geographical concentration shows how concentrated the production of
university technological knowledge is across regions (the coefficient takes a value of one for maximum
concentration, that is, if all patents were assigned to one region, and 1/n for an equal distribution, where
n = 202 regions).
Page 35 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
36
Table 2. Regions with high numbers of European university-owned patents (1998–2004)
Region NUT II Patents % % Concentration
Patents/hab(*)
Inner London UKI1 410 8.95 8.95 142.99
Berkshire, Buckinghamshire, and Oxfordshire
UKJ1 257 5.61 14.56 122.41
Vlaams Gewest BE2 220 4.80 19.37 36.69
Zuid-Holland NL33 188 4.10 23.47 54.65
Île de France FR10 144 3.14 26.62 12.83
Eastern Scotland UKM2 132 2.88 29.50 67.06
South-western Scotland UKM3 131 2.86 32.36 55.98
East Anglia UKH1 127 2.77 35.13 57.96
Rhône-Alpes FR71 112 2.45 37.58 19.19
Gloucestershire, Wiltshire, and North
Somerset UKK1 110 2.40 39.98 50.67
Others 2,749 60.02 100.00 8.03
All regions 4,580 100.00 11.97
(*) Number of university patents per million inhabitants
Source: Derwent Innovation Index
Page 36 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
37
Table 3. Sectors with the highest number of university-owned patents (1998–2004)
Field no. Description Patents %
%
Concentr
ation
Regions with the highest number of patents in each
sector
13 Pharmaceuticals 1,793 39.15 39.15 - Inner London (UK) (11.2%)
- Berkshire, Buckinghamshire, and Oxfordshire
(UK) (6.9%)
- Vlaams Gewest (BE) (5.1%)
- Zuid-Holland (NL) (4.6%)
- Île de France (FR) (3.8%)
38 Measuring instruments 430 9.39 48.54 - Inner London (UK) (10.2%)
- Berkshire, Buckinghamshire, and Oxfordshire (UK)
(5.6%)
- Vlaams Gewest (BE) (5.6%)
- East Anglia (UK) (4.2%)
- South-western Scotland (UK) (4.0%)
10 Basic chemical 388 8.47 57.01 - Comunidad Valenciana (ES) (10.3%)
- Berkshire, Buckinghamshire, and Oxfordshire (UK)
(6.4%)
- Eastern Scotland (UK) (4.9%)
- Region Wallone (BE) (3.6%)
- Northumberland and Tyne and Wear (UK) (3.6%)
37 Medical equipment 374 8.17 65.17 - Inner London (UK) (13.6%)
- Rhône-Alpes (FR) (5.1%)
- Tübingen (DE) (3.7%)
- Île de France (FR) (3.7%)
- Zuid-Holland (NL) (3.7%)
28 Office machinery and computers 200 4.37 69.54 - Berkshire, Buckinghamshire, and Oxfordshire (UK)
(9.5%)
- Inner London (UK) (7.0%)
- Greater Manchester (UK) (6.5%)
- South-western Scotland (UK) (5.5%)
- Île de France (FR) (5.0%)
Others 1,395 30.46 100
All patents 4,580 100
Source: Derwent Innovation Index
Page 37 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
38
Table 4. Technological specialization of the 20 regions with the highest numbers of university
patents
Sectors (*)
Regions No. of
Patents
% Pat
High
tech 10 13 28 37 38
Intraregional
Correlation
(***)
Inner London (UK) 410 0.94 0.35 1.25 0.78 1.52 1.14 0.061
Berkshire, Buckinghamshire, and
Oxfordshire (UK) 257 0.96 1.15 1.22 1.69 0.62 0.99 0.812
Vlaams Gewest (BE) 220 0.89 0.54 1.07 0.62 0.61 1.16 0.498
Zuid-Holland (NL) 188 0.95 0.82 1.11 0.61 0.91 0.91 0.123
Île de France (FR) 144 0.94 0.25 1.21 1.59 1.19 0.74 –0.228
Eastern Scotland (UK) 132 0.95 1.70 0.87 0.35 1.11 1.29 –0.397
South-western Scotland (UK) 131 0.94 0.27 0.88 1.92 0.93 1.38 0.056
East Anglia (UK) 127 0.94 1.02 0.95 1.26 0.48 1.51 –0.472
Rhône-Alpes (FR) 112 0.96 1.37 0.50 1.64 2.08 1.14 –0.140
Gloucestershire, Wiltshire, and
North Somerset (UK) 110 0.93 0.00 1.07 1.04 0.89 1.07 –0.277
Greater Manchester (UK) 99 0.97 1.19 0.88 3.01 1.24 1.61 –0.139
Hampshire and Isle of Wight (UK) 96 0.86 0.12 0.43 1.67 0.77 0.55 –0.362
Région de Bruxelles-
Capitale/Brussels Hoofdstedelijk
Gewest (BE) 85 0.95 0.56 1.11 1.62 0.29 0.38 0.142
South Yorkshire (UK) 76 0.95 1.09 1.31 0.90 0.97 0.28 0.269
Southern and Eastern (IE) 75 0.97 0.79 1.19 1.83 0.33 0.85 0.449
Utrecht (NL) 73 0.97 0.81 1.68 0.00 1.17 0.29 0.589
Derbyshire and Nottinghamshire
(UK) 71 0.94 0.33 1.04 0.32 1.90 1.20 0.617
Tübingen (DE) 70 0.99 0.67 1.35 0.00 2.45 1.83 0.910
Comunidad Valenciana (ES) 65 0.88 7.26 0.12 0.00 0.75 0.33 –0.304
Région Wallonne (BE) 62 0.94 2.67 0.91 0.74 0.40 0.86 0.198
10 Basic chemical
13 Pharmaceuticals
28 Office machinery and computers
37 Medical equipment
38 Measuring instruments
(*) Technological specialization (index of Revealed Technological Advantage).
(**) The Herfindahl Concentration Index for 44 industrial sectors (maximum concentration = 1; minimum concentration =
0.022)
(***) Intraregional correlation: Spearman Correlation Coefficient between technological specialization (university patents)
and economic specialization (employment). This coefficient was calculated with all the sectors and not only with those
presented in this table.
Page 38 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
39
Table 5. No. of regions with data available for the explanatory variables
Variables Variables YEAR
R&D (t-3) GIP (t-3) Both
Obs.
three lags
(1) R&D (t-2) GIP (t-2) Both
Obs. two
lags (2)
1998 (*) - 0 0 (*) 0 0 0
1999 (*) - 0 0 121 163 90 90
2000 121 163 90 90 114 163 81 81
2001 114 163 81 81 127 163 92 92
2002 127 163 92 92 150 165 117 117
2003 150 165 117 117 154 165 119 119
2004 154 165 119 119 115 164 79 79
Unbalanced panel 499 578
(*) Data on lagged university R&D or GIP for these years are not available or include too much missing data.
(1) Regions (observations) included in the models with three lags in the explanatory variables.
(2) Regions (observations) included in the models with two lags in the explanatory variables.
Page 39 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
40
Table 6. Effects of regional university R&D funds and industrial potential of regions on university
patents
THREE LAGS FOR R and GIP TWO LAGS FOR R and GIP
Variable MODEL I
Poisson Count (1) MODEL II
Negative Binomial Count MODEL III
Poisson Count (1) MODEL IV
Negative Binomial Count
Coeff. Std. Err. Coeff. Std. Err. Coeff. Std. Err. Coeff. Std. Err.
C –3.872 *
0.853 –4.380 *
0.717 –4.258 *
0.898 –4.682 *
0.698
LOG(R) 0.933 *
0.072 1.031 *
0.078 0.955 *
0.069 1.043 *
0.076
LOG(GIP) 0.122 0.121 0.125 0.100 0.142 0.124 0.142 0.095
Year04 0.528 *
0.192 0.605 *
0.161 0.978 *
0.230 0.987 *
0.183
Year03 0.598 *
0.182 0.637 *
0.158 0.735 *
0.213 0.786 *
0.184
Year02 0.206 0.199 0.306 *
0.171 0.566 *
0.209 0.595 *
0.180
Year01 0.051 0.208 0.152 0.171 0.530 *
0.219 0.559 *
0.190
Year00 0.263 0.247 0.208 0.192
Small countries (2) –0.547 *
0.170 –0.781 *
0.186 –0.786 *
0.151 –0.901 *
0.182
DE –1.092 *
0.173 –1.149 *
0.198 –1.493 *
0.163 –1.576 *
0.206
GR –4.192 *
1.050 –4.110 *
1.039 –4.301 *
1.021 –4.223 *
1.045
ES –1.143 *
0.162 –1.165 *
0.173 –1.472 *
0.155 –1.476 *
0.177
FR –1.152 *
0.164 –1.246 *
0.145 –1.509 *
0.148 –1.522 *
0.155
IT –1.860 *
0.177 –2.019 *
0.180 –2.311 *
0.169 –2.393 *
0.201
PT –2.389 *
0.367 –2.334 *
0.464 –2.883 *
0.406 –2.820 *
0.460
FI –2.748 *
0.571 –2.813 *
0.476 –2.948 *
0.487 –2.895 *
0.414
OBJ1 0.071 0.140 0.013 0.144 0.099 0.127 0.062 0.142
Alpha –0.838 *
0.144 –0.697 *
0.131
Log likelihood -1010.11 –863.69 –1147.83 –958.30
LR statistic (15 df) 3286.32 *
3579.16 3898.68 *
4277.73
*
Pseudo-R2 0.62 0.67 0.63 0.69
LR=2(lnLNB-lnLP) 292.84 *
379.06 *
Notes:
(1) Eicker–White standard errors.
(2) This variable takes a value of one for some small countries where data were available only for one or two
regions, and zero otherwise. Other European countries do not appear in the table because of a lack of R&D data.
The base category is the UK.
The symbol * denotes significant coefficients with a p value of less than 0.05.
Page 40 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
41
Table 7. Summary of results
MAIN SPECIFICATIONS VARIABLES MAIN RESULTS
Log (R) - Significant coefficients of around one for all the models (0.93–1.04), indicating
constant returns to scale.
Log (GIP) - No evidence of the possible effects that the industrial potential may have on
the production of university technological knowledge.
Year dummies - Significant coefficients for several years. Evidence of temporal effects.
Country dummies - Significant coefficients showing the importance of the political and institutional
context in countries to generate differences in patent production.
MODELS I to IV (From Table 6)
OBJ1 - Coefficient no significant for this variable, but note that this effect overlaps
with that of the country dummies.
ROBUTNESS
SENSIVITY ANALISIS I (Poisson
and NB models including only Log
(R) as explanatory with two and
three lag).
Log (R) - Significant coefficients in all the specifications. Elasticities of around one
(0.99–1.10)
Log (R) - Significant coefficients for Log (R). Elasticities with more variability (between
0.87 and 1.11).
- No significant coefficients for Log (GIP)
SENSIVITY ANALISIS II (Poisson
and NB models including Log (R)
and Log (GIP) as explanatory
variables with two and three lag). Log (GIP)
Log (R) - Elasticities were around one for R.
Log (GIP) - No significant coefficients for most of the models.
INTRAGROUP ESTIMATIONS
(Poisson and NB models for every
year with data available).
Country dummies - Significant coefficients for country dummies. Considerable spatial differences
among countries
Page 41 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010

For Peer Review Only
42
Figure 1. Regional distribution of university patents
Page 42 of 42
http://mc.manuscriptcentral.com/cres Email: regional.studies@fm.ru.nl
Regional Studies
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
peer-00526546, version 1 - 15 Oct 2010