Content uploaded by Massimo G. Colombo
Author content
All content in this area was uploaded by Massimo G. Colombo on May 15, 2016
Content may be subject to copyright.
The contribution of university research to the growth
of academic start-ups: an empirical analysis
Massimo G. Colombo Æ Diego D’Adda Æ Evila Piva
Published online: 19 March 2009
Ó Springer Science+Business Media, LLC 2009
Abstract The aim of this paper is to analyze empirically under which circumstances the
universities located in a geographical area contribute to the growth of a special category of
local new technology-based firms (NTBFs), those established by academic personnel
(academic start-ups, ASUs). We examine the effects of a series of characteristics of local
universities on the growth rates of ASUs and we compare them with the effects of the same
university characteristics on the growth of other (i.e., non-academic) NTBFs. In the
empirical part of the paper, we estimate an augmented Gibrat law panel data model using a
longitudinal dataset composed of 487 Italian NTBFs observed from 1994 to 2003. Out of
these NTBFs 48 are ASUs. The results of the econometric estimates suggest that univer-
sities do influence the growth rates of local ASUs, while the effects on the growth rates of
other NTBFs are negligible. In particular, the scientific quality of the research performed
by universities has a positive effect on the growth rates of ASUs; conversely the com-
mercial orientation of research has a negative effect. These results indicate that universities
producing high-quality scientific research have a beneficial impact on the growth of local
high-tech start-ups, but only if these firms are able to detect, absorb, and use this
knowledge. In this perspective, a greater commercial orientation of university research
leading to a reduction of the knowledge available for absorption by these companies, can
be detrimental.
Keywords Academic start-ups New technology-based firms University research
Growth
JEL Classification L25 M13 O30
M. G. Colombo D. D’Adda E. Piva (&)
Politecnico di Milano, Department of Management, Economics and Industrial Engineering, Piazza
Leonardo da Vinci, 32, 20133 Milan, Italy
e-mail: evila.piva@polimi.it
123
J Technol Transf (2010) 35:113–140
DOI 10.1007/s10961-009-9111-9
1 Introduction
Since the seminal works by Nelson (1959) and Arrow (1962), it is widely accepted that the
geographical proximity of universities to knowledge-intensive companies generates posi-
tive knowledge externalities for technological innovation and productivity within the
private sector. In fact, spatial proximity allows knowledge transmission from universities
to companies through the local networks of university staff and industry professionals
(local labor market of graduates, faculty consulting, university seminars, conferences,
student internships, continuing education of employees), formal business relations (uni-
versity spin-off companies, technology licensing), and externalities engendered by
university physical facilities (libraries, scientific laboratories, computer facilities) (Varga
2000).
The literature on the contribution of universities to local industry has focused on three
major issues. Many studies have examined the relation across geographical areas between
university research expenditures and private innovation outputs, measured in terms of
either patent counts (Anselin et al. 1997; Del Barrio-Castro and Garcı
´
a-Quevedo 2005;
Fischer and Varga 2003; Jaffe 1989; Jaffe et al. 1993; Piergiovanni et al. 1997) or the
number of innovations introduced into the market (Acs et al. 1992; 1994; Feldman and
Florida 1994). All these works found a strong and positive relation between local inno-
vative activity and university research and concluded that innovation in the private sector is
positively affected by knowledge spillovers from universities, i.e., intentional or unin-
tentional knowledge flows to the private sector, that tend to be geographically bounded
within the area (either the state or the region/province) where the universities are located.
More recent empirical studies focused on the impact of university research on firms’
localization (Audretsch et al. 2004; Audretsch and Lehmann 2005a; Bade and Nerlinger
2000; Rodrı
´
guez-Pose and Refolo 2003; Varga 2000). These works found a significant
positive correlation between firms’ local concentration and university location and deduced
that the existence of localized knowledge spillovers leads to greater concentration of
companies in the proximity of universities. Such effect is more evident the greater the
quantity and quality of university research output.
Finally, a few studies examined how knowledge externalities affect regional economic
growth (Goldstein and Drucker 2006; Goldstein and Renault 2004; for a survey see
Drucker and Goldstein 2007). These works showed that knowledge produced by univer-
sities is captured within the regional environment and results in enhanced economic
development measured by average annual earnings (i.e., annual income excluding divi-
dends, rent, interest, and transfer payments) per worker.
All the above mentioned studies have neglected the analysis of the effects of university
research on the growth of individual local firms (Audretsch et al. 2005). Nonetheless, we
believe that this is an important avenue for research. In particular, in the present work we
examine the relations between university research and the growth of local new technology-
based firms (NTBFs), that we define as independent high-tech firms established within the
last 25 years (Little 1977). This is an ideal unit of analysis for this research because, as we
discuss in the following section, access to external scientific knowledge is particularly
relevant for the development of new firms in high-tech sectors. Moreover, for these firms
rapid growth is an unequivocal signal of wide market acceptance of their products/services
and it is thus generally associated with business success (Barringer et al. 2005; Fischer and
Reuber 2003; Feeser and Willard 1990).
To the best of our knowledge, only Audretsch and Lehmann (2005b) explored the
connection between university knowledge production and the growth of individual
114 M. G. Colombo et al.
123
companies. They used a dataset of 281 German IPO firms to empirically analyze the
impact of knowledge spillovers from local universities on the growth rates of the number
of employees in the sample firms 1 year after the IPO. They found that, with all else equal,
the closer the next university and the higher the number of academic papers published by
its researchers, the higher the growth rates of sample firms. Here we diverge from this
study in two directions.
First, one of the limits of the work by Audretsch and Lehmann (2005b) is that, even
though they are interested in the effects of knowledge spillovers to the private sector they
do not measure the flows of knowledge. It is fair to acknowledge that finding an appro-
priate proxy for knowledge spillovers has proved to be a challenging exercise and the lack
of a proper operationalization of this concept is a weakness common to many other studies.
Hence we do not set the ambitious aim of studying the impact of knowledge flows from
universities on firms’ growth, conversely we limit our analysis to testing the effects of a
series of organization-specific characteristics of universities.
Second, we adhere to the argument set forth by prior studies that the ability of firms to
get access to and assimilate the knowledge produced by local universities depends on
firms’ absorptive capacity (Fischer and Varga 2003) and the presence of academic
researchers in the board (Audretsch and Lehmann 2006). In accordance with this view, we
argue that the impact of the characteristics of local universities on individual firm growth is
affected by the specificities of focal NTBFs. Accordingly we distinguish new high-tech
ventures established by academics and/or researchers
1
that were previously employed by
public research organizations (academic start-ups, ASUs), from other (i.e., non-academic)
NTBFs. We contend that ASUs are in an ideal position to take advantage of university
research as a consequence of both their science-orientation and the social contacts and
collaborative linkages of their founders with the research personnel of public research
organizations.
To sum up, in order to extend our understanding of the contribution of universities to the
growth of individual firms, we address the following research questions: which charac-
teristics of local universities enhance or hinder the growth rates of ASUs? Do the alleged
effects of local universities on firm growth differ between ASUs and other NTBFs?
In the theoretical part of the paper we rely on prior studies on the contribution of
academic research to firms’ performances and on the peculiarities of ASUs and we for-
mulate a series of hypotheses on both the effects of selected characteristics of local
universities on the growth of ASUs and the different impact of such characteristics on
ASUs and other NTBFs. In the empirical part, we compare the determinants of the growth
rates of ASUs with those of other NTBFs through the estimation of an augmented Gibrat
law panel data model. We devote specific attention to the impact on the growth rates of the
two types of firms of the amount, quality, and commercial orientation of the scientific
knowledge produced by the universities located in the same provinces as focal companies.
For this purpose we take advantage of a unique longitudinal dataset including 487 Italian
young firms that operate in high-tech industries in both manufacturing and services and are
observed over the period 1994–2003. Out of these firms 48 are ASUs.
The remainder of the paper is structured as follows. In Sect. 2, we first describe how the
characteristics of academic institutions influence the knowledge externalities arising from
the proximity of universities to NTBFs and, thus, how they affect the growth potential of
1
Researchers are defined as individuals who perform research activities in public research organizations,
regardless of the contractual link with the parent institution. They include the research staff of parent
organizations (both full time and part time) and also Ph.D. students.
The contribution of university research to the growth of academic start-ups 115
123
these companies. Then we examine to what extent this growth potential is realized
depending on the specificities of local firms. More specifically, we formulate a series of
hypotheses concerning the effects of the characteristics of local universities on ASUs’
growth rates and the differences with respect to the effects on the growth of non-academic
NTBFs. In Sect. 4, we describe the dataset. Then we illustrate the specification of the
econometric model and the explanatory variables. In Sect. 5, we report the results of the
econometric estimates. A discussion of the main findings and limitations of the study in
Sect. 6 concludes the paper.
2 From the extant literature to the theoretical hypotheses
2.1 The conceptual framework
Our hypotheses are based on the resource- and competence-based theories of the firm. Such
theories postulate that the growth of individual companies depends on their ability to
develop internally and/or acquire rare and difficult to reproduce resources and compe-
tencies. In the high-tech industries where NTBFs operate new scientific knowledge is
particularly valuable (Fischer and Varga 2003). As universities are key sources of new
scientific knowledge, we expect the NTBFs that succeed in having access to and absorbing
the knowledge produced by universities to achieve greater growth performances.
Any firm finds it easier to have access to the knowledge produced by a university when
it is located nearby. In fact, spatial proximity allows the dissemination of academic
knowledge through channels such as the local personal networks of university researchers
or fresh graduates. Hence, ceteris paribus, being located close to universities should
facilitate also knowledge absorption and thus it should positively affect the growth
potential of NTBFs.
However, as prior studies have shown, knowledge externalities from university research
are influenced by several characteristics of the universities located in the same area of focal
firms. As we will show in Sect. 2.2, such characteristics of local universities also influence
firms’ growth potential.
As we will better show in Sect. 2.3, to what extent this growth potential is realized will
depend on the characteristics of focal NTBFs. Recent studies on the role of firms’ char-
acteristics in university-to-industry knowledge transfer (for a review see Agrawal 2001)
suggest that the mere availability of scientific knowledge as a consequence of the presence
of a university in a given geographical area is not enough to enable the firms located in the
same area to assimilate such knowledge. Conversely, one can identify a number of firm-
specific characteristics that facilitate the absorption of university knowledge. In this paper
we focus on a special group of NTBFs: the high-tech start-ups established by academic
personnel (ASUs). In particular, in Sect. 2.3 we contend that these firms have some
‘‘genetic’’ characteristics that facilitate detection, absorption, and exploitation of academic
knowledge (Colombo and Piva 2008a). Then we discuss how the characteristics of local
universities affect the growth rates of these firms. We also compare the effects on the
growth rates of ASUs to those on the growth of non-academic NTBFs, i.e., firms that do
not exhibit the genetic characteristics of ASUs. In Fig. 1 we graphically illustrate the
effects of both the characteristics of local universities and the status of ASU on firms’
growth rates.
116 M. G. Colombo et al.
123
2.2 The influence of the characteristics of local universities on the growth potential
of NTBFs
In this section we consider (some of) the characteristics of universities that are likely to
influence the growth potential of local NTBFs.
2
First, the size of the university research staff in a particular domain determines the amount
of knowledge that local firms may have access to. Ceteris paribus, the greater the number of
scientists within a university, the greater the skills and knowledge available for transfer to the
private sector (O’Shea et al. 2005), and the higher the growth potential of the NTBFs located
in the same area. Second, the growth potential of NTBFs is likely to be affected by the quality
of the research performed in local universities. In fact, the more advanced the knowledge
produced within universities, the more relevant its alleged contribution to the generation of
competitive advantages for local NTBFs and thus to firm growth potential. The development
of cutting-edge scientific knowledge requires universities to hire skilled and talented indi-
viduals (Powers and McDougall 2005). Hence, the higher the quality of university research
staff, the greater the growth potential of local NTBFs (this argument is consistent with the
findings of Anselin et al. 1997; Rodrı
´
guez-Pose and Refolo 2003).
Finally, the contribution of universities to the growth potential of local NTBFs may
depend on the commercial orientation of university research, i.e., the degree to which
researchers focus on industrial needs and problems. This is reflected in the source of
funding of research: the research activity of universities that receive a greater share of their
research budget from industry is likely to be more commercially oriented (Rosenberg and
Nelson 1994). The effect of commercial orientation on firms’ growth potential is not
straightforward.
Fig. 1 The conceptual model
2
It is fair to acknowledge that, following the extant literature, we may expect a number of traits of
universities additional to those analyzed in this paper to influence the growth potential of local NTBFs. In
particular, prior studies have shown that university policies regarding intellectual property, licensing
strategies and characteristics of technology transfer offices may affect university-to-industry knowledge
transfer (for a review of these studies see again Agrawal 2001). Hence, we may expect these characteristics
to influence firms’ growth potential too; however we decided not to consider them in this work. In Italy
policies and support infrastructures for technology transfer have started spreading after 2000 and have
proliferated only in the last couple of years. Therefore, very few universities already exhibited such
properties over the period 1994–2003 that we consider in this work. As a consequence, we could not capture
the effect of these characteristics through the estimation of econometric models over this time horizon. Thus
we excluded the study of these characteristics from both the empirical analysis and the development of the
theoretical hypotheses.
The contribution of university research to the growth of academic start-ups 117
123
On the one hand, greater commercial orientation may facilitate the absorption of the
knowledge produced by universities on the part of local firms. The tendency of a university
to conduct commercially oriented research should increase the likelihood of discovering
technologies and producing knowledge that have commercial value (Di Gregorio and
Shane 2003). We expect that firms find it easier to absorb this knowledge rather than
abstract scientific knowledge. In addition, institutions performing research that is closer to
industrial needs exhibit more entrepreneurial activity, such as faculty consulting within
industry, faculty involvement in new firms and faculty and university equity participation
in start-ups (Cohen et al. 1998; Roberts and Malone 1996). These activities enhance the
contacts between academic researchers and practitioners and encourage the knowledge
flows from academia to industry.
On the other hand, greater commercial orientation of university research may inhibit
knowledge transfer to local NTBFs. As Argyres and Liebeskind (1998) suggest, univer-
sities interested in attracting private research sponsorship might offer sponsoring firms
privileged access to the results of academic research. Universities could even assign
intellectual property (IP) rights to the firms that funded the research that produced such IP.
As firms are more motivated than universities to protect IP from public disclosure and
secure exploitation rights of research results (Dasgupta and David 1994; Stern 2004), we
expect firms to be anxious to ensure that knowledge leakages do not take place until a
patent application is filled or the research results are exploited commercially. To achieve
this, firms may try to inhibit premature publication or dissemination of research results by
requiring the right to delay or, at least, to approve any proposed scientific publication or
further university initiative to disseminate the research results in advance. This clearly
hinders the generation of positive externalities to other firms from university research and,
consequently, has negative effects on the growth potential of local NTBFs.
2.3 The role of ASU’s status
As we have pointed out in Sect. 2.2, the characteristics of universities affect the growth
potential of local firms. Whether this growth potential is translated into realized growth
depends on the characteristics of focal firms. In this section we focus on the status of ASU
as an enabling factor of the exploitation of academic knowledge. Hence, we expect that
there are considerable differences between ASUs and other NTBFs as to the effect on
firms’ growth of the above mentioned characteristics of local universities.
In order to capture and deploy knowledge from external sources firms must be endowed
with adequate ‘‘absorptive capacity’’ (Cohen and Levinthal 1989; 1990). Absorptive
capacity is ‘‘the ability of a firm to recognize the value of new, external information,
assimilate it, and apply it to commercial ends’’ (Cohen and Levinthal 1990, 128). It largely
depends on the level of prior related knowledge within the firm. This in turn is a function of
firms’ internal investments in R&D. Hence, the higher firms’ R&D intensity in the fields
where local universities are active, the easier the exploitation of academic knowledge and
thus the greater its effects on firm growth.
Following this argument, we claim that ASUs should enjoy significant advantages in
exploiting university knowledge as they are endowed with greater absorptive capacity than
other NTBFs. In fact, ASUs’ founding teams exhibit greater scientific education and prior
research experience than those of other NTBFs (Colombo and Piva 2008b). Furthermore,
ASUs are likely to invest greater resources in R&D activities than other NTBFs. On the
one hand, due to the genetic characteristics of their founding teams, the marginal returns of
the investments of ASUs in these activities allegedly are greater than those of other
118 M. G. Colombo et al.
123
NTBFs. On the other hand, due to their science orientation, ASUs encounter smaller
obstacles in hiring qualified technical personnel. As a consequence of the larger initial
scientific knowledge base and the higher R&D intensity, ASUs should find it easier than
other NTBFs to evaluate the research of local universities, identify interesting results and
utilize them in their own activities.
Second, more recent studies concerning the firm-specific characteristics that influence
the ability of companies to use scientific knowledge transferred from universities have
shown that investments in R&D are not the only mechanism to develop absorptive capacity
(Cockburn and Henderson 1998; Lim 2000; Zucker et al. 2000). In particular, firm
‘‘connectedness’’ in the scientific community plays a key role in facilitating the evaluation
and utilization of academic knowledge. In fact, when researchers employed in a company
co-author papers with academics, a firm participates in research consortia with universities
and cultivates university relationships by sponsoring university research or collaborating
with faculty members, acquiring knowledge from academia is easier. As to ASUs, the
network of social contacts of academic entrepreneurs in the public research sector is wider
than that of the founders of other NTBFs, making ASUs more embedded within the
scientific community (Murray 2004). Furthermore, academic founders often keep their
position in the parent research organization even after the foundation of the new venture
(Roberts 1991), so it is easier for them to maintain and enlarge this social network. These
links in the research environment give ASUs privileged access to the results of university
research and make it easier to detect potentially valuable knowledge and skilled academic
researchers that could actively participate in firms’ activities.
Nonetheless it has been argued that under specific circumstances, connectedness in the
scientific community is not enough to enable the utilization of external knowledge pro-
duced by universities (Audretsch and Lehmann 2006; Audretsch and Stephan 1996;
Mowery and Ziedonis 2001; Zucker et al. 1998; 1999). In particular, in the fields where
knowledge is embodied in human capital it is crucial to involve university scientists in the
company as principals, consultants, employees, or members of scientific advisory boards.
In this respect, ASUs should clearly enjoy advantages in comparison with other NTBFs as,
by definition, their founding teams always include academic researchers.
To sum up, the above arguments suggest that ASUs are better equipped than other
NTBFs to detect and absorb useful knowledge produced by local universities. Hence, they
can more easily attract ideas from academia and thus benefit from a positive impact of
these ideas on firm growth. By combining these arguments with those reported in the
previous section we conclude that the number and quality of scientists of local universities
will have a positive impact on the growth of ASUs, as these firms are in an ideal situation
to absorb and exploit scientific knowledge. Such effect should be greater for ASUs than for
other NTBFs. This leads to the following hypotheses.
Hypothesis H1a The number of researchers in local universities has a positive effect on
the growth rates of ASUs.
Hypothesis H1b The number of researchers in local universities has a more positive
effect on the growth rates of ASUs than on those of other NTBFs.
Hypothesis H2a The quality of the scientific knowledge produced by local universities
has a positive effect on the growth rates of ASUs.
Hypothesis H2b The quality of the scientific knowledge produced by local universities
has a more positive effect on the growth rates of ASUs than on those of other NTBFs.
The contribution of university research to the growth of academic start-ups 119
123
Let us now consider the effects of the commercial orientation of academic research.
First of all, commercial orientation might act as a substitute for the above mentioned
characteristics of firms that facilitate the absorption of academic knowledge. Commer-
cially-oriented universities are more likely to produce knowledge that has immediate
commercial value. This knowledge can be easily exploited also by non-academic NTBFs
that exhibit low absorptive capacity and are otherwise unable to absorb more abstract
scientific knowledge. Similarly, the greater entrepreneurial attitude of researchers in
commercially-oriented universities should compensate for the weaker ties of non-academic
NTBFs within the public research sector thus having a clear positive effect on the ability of
these NTBFs to absorb academic knowledge. Conversely, commercial orientation should
not increase the already well-developed ability of ASUs to take advantage of scientific
knowledge (see Fig. 2, Graph 1). This suggests that the more commercially-oriented the
research performed by universities, the less relevant the advantages that local ASUs enjoy
with respect to other NTBFs in detecting, absorbing, and utilizing academic knowledge.
Conversely, the limited amount of knowledge freely flowing from commercially-oriented
universities should have negative consequences for both ASUs and other NTBFs (Figure 2,
Graph 2). The result of these overlapping forces is that commercial orientation should have
negative consequences for ASUs (Fig. 2, Graph 3), while the effect on other NTBFs is
questionable (Fig. 2, Graph 4).
We synthesize such arguments in the following hypotheses.
Hypothesis H3a The commercial orientation of local universities has a negative effect on
the growth rates of ASUs.
Fig. 2 The effect of commercial orientation of university research on ASUs and other NTBFs
120 M. G. Colombo et al.
123
Hypothesis H3b The commercial orientation of local universities has a more negative
effect on the growth rates of ASUs than on those of other NTBFs.
3 The dataset
This section has purely illustrative purposes. Here we describe the dataset we use to test
our hypotheses. This dataset includes both survey data on a sample of 487 Italian NTBFs
and public information on the characteristics of the universities located in the same
provinces as the sample firms.
3.1 The sample of NTBFs
In this paper we use a unique dataset of Italian NTBFs that operate in high-tech manu-
facturing and services sectors. These firms are extracted from the RITA (Research on
Entrepreneurship in Advanced Technologies) database developed by the Department of
Management, Economics and Industrial Engineering of Politecnico di Milano. The RITA
dataset was created in 2000 and it was updated and extended in 2002 and 2004. The
development of the 2004 release of the database went through a series of steps. First, Italian
firms that were established after 1980, remained independent (i.e., not controlled by other
organizations) up to the end of 2003 and operated in high-tech sectors, both in manufac-
turing and services, were identified. For the construction of the target population a number
of sources were used.
3
These included lists provided by national industry associations, on-
line and off-line commercial firm directories, and lists of participants in industry trades and
expositions. Information provided by the national financial press, specialized magazines,
other sectoral studies, and regional Chambers of Commerce was also considered. Alto-
gether, 1,974 firms were selected for inclusion in the database. Among these firms there
were 123 ASUs. For each firm, a contact person (i.e., one of the owner-managers) was also
identified. Second, a questionnaire was sent to the contact person of target firms either by
fax or by e-mail in the first semester of 2004. The questionnaire provided detailed infor-
mation on the characteristics of the firms including the human capital characteristics of their
founders, their financing strategies and their growth performances.
Lastly, answers to the questionnaire were checked for internal coherence by educated
personnel and were compared with published data (basically data provided by firms’
annual reports and websites). In many cases, phone or face-to-face follow-up interviews
were made with firms’ owner-managers. This final step was crucial in order to obtain
missing data and ensure that data were reliable.
4
The sample used in the present work consists of all RITA firms for which we were able
to create a complete data set including data on size, measured by the logarithm of the
number of employees (including owner-managers), for at least three consecutive years (on
3
Unfortunately, data provided by official national statistics do not allow to obtain a reliable description of
the universe of Italian NTBFs. The main problem is that in Italy most individuals who are defined as ‘‘self-
employed’’ by official statistics actually are salaried workers with atypical employment contracts. Unfor-
tunately, on the basis of official data such individuals cannot be distinguished from entrepreneurs who
created a new firm.
4
Note that for only three firms the set of owner-managers at survey date did not include at least one of the
founders of the firm. For these firms information relating to the human capital characteristics of the founders
was checked through interviews with firms’ personnel so as to be sure that it did not relate to current owner-
managers.
The contribution of university research to the growth of academic start-ups 121
123
this issue see footnote 12). It is composed of 487 NTBFs. 48 of these firms are ASUs. v
2
tests show that there are no statistically significant differences between the distributions of
sample firms across industries and regions and the corresponding distribution of the 1,974
RITA NTBFs from which the sample was drawn (v
2
(4) = 4.01 and v
2
(3) = 6.10,
respectively). Sample firms are observed during the period 1994–2003.
Of course, there is no presumption here to have a random sample. First, in this domain
representativeness is a slippery notion as new ventures may be defined in different ways
(see for instance Aldrich et al. 1989; Birley 1984; Gimeno et al. 1997). Second, absent
reliable official statistics, it is very difficult to identify unambiguously the universe of
Italian NTBFs and that of Italian ASUs. Therefore, one cannot check ex-post whether the
sample used in this work is representative of the population. Third, the analysis presented
in this paper is based on survey data; so it might suffer from a sample selection bias. First
and foremost, only firms having survived up to the survey date could be considered: this
generates a survivorship bias. As in most survey-based studies, it is impossible to properly
control for the survivorship bias. What we can do is to check its extent.
For this purpose, we focused attention on the RITA 2000 sample. This sample is com-
posed of 401 NTBFs identified when the RITA database was created; it includes 25 ASUs
and 376 other NTBFs. Out of the 25 ASUs only four ceased operations or were acquired by/
merged with other firms in the period 2000–2003. Therefore, we could not test whether there
was any difference between the ASUs that exited the sample and those that did not. The
likelihood of exit was greater among non-academic NTBFs; in fact, 83 of these firms exited
the sample, corresponding to a 22.1% share. As the likelihood of exit was very low among
ASUs, the survivorship bias, if there were any, would be driven by the exit of non-academic
NTBFs. We then run a probit model of the likelihood of these latter firms having survived in
the 2000–2003 period, conditional on survival up to 2000. In addition to firm- and industry-
specific controls, the explanatory variables included all the measures of the characteristics
of local universities that are likely to affect firm growth and hence will be considered in the
analysis reported in Sect. 5. The econometric estimates show that these variables have no
effects on the likelihood of survival of non-academic NTBFs. Hence, even though this
check is very partial, we have no indication suggesting that the results of the estimates that
will be illustrated in Sect. 5 are driven by a survivorship bias.
The distribution of sample firms across industries, geographic areas and periods of
foundation is highlighted in Table 1.
Differences across industries in the number of ASUs and other NTBFs are fairly limited.
Both ASUs and other NTBFs mainly operate in Software (29.2% vs. 29.8% of the sample,
respectively) and Internet and telecommunication services (35.4% vs. 28.7%). ASUs are
less numerous than NTBFs in the ICT manufacturing sector (18.8% vs. 22.6%).
As to the geographical distribution, sample firms are mainly located in the North–West
(47.6% of the total sample), conversely they are seldom founded in southern regions
(13.6%). As to the distribution across provinces, the 487 NTBFs are located in 78 out of the
103 Italian provinces that existed in the period under scrutiny with greater concentration in
the provinces of Milan, Turin, and Rome. It is interesting to compare the geographical
distributions of ASUs and other NTBFs as the pattern of localization of ASUs differs quite
remarkably from that of other NTBFs. The percentages of ASUs located both in the regions
of the North-West and in the less developed regions of the South are lower than those of
other NTBFs (41.7% vs. 48.3% and 6.3% vs. 14.4%, respectively), conversely the per-
centage of ASUs is higher in the remaining areas (29.2% vs. 22.6% in the North–East and
22.9% vs. 14.8% in the Centre). As to the distribution across provinces, we encounter
ASUs in 24 out of the 78 provinces where sample NTBFs are located.
122 M. G. Colombo et al.
123
As to the year of foundation, sample NTBFs have been mainly founded since the mid
Nineties (50.3% of sample firms have been founded between 1995 and 2003). This is
particularly evident for ASUs. A total of 72.9% of sample ASUs were founded after 1994
and 22.9% after 1999. Conversely, the number of other NTBFs established in the same
periods is 47.8% and 9.8%, respectively.
3.2 Italian universities
Table 2 shows the geographical distribution of Italian universities. At the end of 2003 in
Italy there were 77 universities located in 49 out of the 103 Italian provinces.
5
The number
Table 1 Distribution of sample ASUs and other NTBFs by industry, geographic area and year of
foundation
ASUs Other NTBFs Total
N% N % N %
Industry
ICT manufacturing 9 18.8 99 22.6 108 22.2
Automation & Robotics 4 8.3 41 9.3 45 9.2
Biotechnologies, pharmaceutics and advanced materials 2 4.2 19 4.3 21 4.3
Software 14 29.2 131 29.8 145 29.8
Internet & TLC services 17 35.4 126 28.7 143 29.4
Multimedia content 2 4.2 23 5.2 25 5.1
Total 48 100.0 439 100.0 487 100.0
Geographical area
North–West 20 41.7 212 48.3 232 47.6
North–East 14 29.2 99 22.6 113 23.2
Centre 11 22.9 65 14.8 76 15.6
South 3 6.3 63 14.4 66 13.6
Total 48 100.0 439 100.0 487 100.0
Year of foundation
1980–1984 6 12.5 50 11.4 56 11.5
1985–1989 3 6.3 81 18.5 84 17.2
1990–1994 4 8.3 98 22.3 102 20.9
1995–1999 24 50.0 167 38.0 191 39.2
2000–2003 11 22.9 43 9.8 54 11.1
Total 48 100.0 439 100.0 487 100.0
5
Following prior studies on the Italian case (Piergiovanni et al. 1997; Rodrı
´
guez-Pose and Refolo 2003)we
use the province as the geographic unit of analysis. Although some of the largest Italian universities have
campuses in different provinces (for instance, Politecnico di Milano has six campuses over the Lombardy
region in five provinces and one campus in Emilia Romagna), in Table 2 for each university we consider the
location of the main campus only. This allows us be consistent with the data reported in the following. In
fact, all the data on the characteristics of Italian universities are available at university level only (i.e., they
are not disaggregated by provinces). Hence, in order to measure the number of researchers, the quality of the
knowledge produced and the commercial orientation of local universities for Italian provinces, for each
university we have attributed all the researchers and, coherently, all the research activities to the main
campus. As faculty members are mainly located in the main campus, this approximation can be considered
reasonably accurate.
The contribution of university research to the growth of academic start-ups 123
123
of universities per province was low: only 11 out of the 49 provinces had more than one
university. Exceptions were the provinces of Naples, Milan, and Rome with respectively 5,
7, and 8 universities. Italian universities were mainly located in the South (32.5%) and in
the Centre (28.7%).
In Table 3 we report some figures on Italian universities focusing on the three char-
acteristics that we consider in the present article: number of researchers, quality of the
scientific knowledge produced (called research quality in the following) and commercial
orientation. The main aim of this table is to show whether and to what extent Italian
universities (and accordingly Italian provinces) differ along these three dimensions.
Let us first consider university size. At the end of 2003 the mean of the total number of
researchers across the 77 Italian universities was 730. If we focus only on economic and
technical faculties, the ones that are more likely to produce knowledge that may be
commercially exploited by NTBFs, the mean number of researchers at the end of 2003
drops to 534. In both cases the standard deviations are high (843 and 644, respectively).
This suggests that there are considerable size differences among Italian universities.
We measure the research quality of each university by the ratio of the number of
citations obtained till year 2001 by articles published in the international journals moni-
tored by the Institute for Scientific Information (ISI) in the 1995–1999 period by
researchers employed by the university to the number of these researchers. Data on
Table 2 Geographical distribu-
tion of Italian universities at the
end of 2003
Geographical area Universities
N%
North–West 17 22.1
North–East 13 16.9
Centre 22 28.6
South 25 32.5
Total 77 100.0
Number of universities per province Italian provinces
None 54 52.4
1 37 35.9
2 7 6.8
3 2 1.9
[3 3 2.9
Table 3 Number of researchers, quality of the scientific knowledge produced and share of privately funded
research of Italian universities
Number
of universities
Mean Standard
deviation
Number of researchers 77 730 843
Number of researchers in economic or technical faculties 77 534 644
Quality of research in medical field 69 23.5 48.0
Quality of research in engineering field 69 5.0 18.5
Quality of research in other scientific fields 69 17.7 24.2
Share of privately funded research 77 0.25 0.19
124 M. G. Colombo et al.
123
publications (and thus citations) per researcher are not comparable across scientific areas,
scientific disciplines may differ in terms of technological opportunity and commercial
potential (Wright et al. 2004). Therefore, in measuring the research quality we distin-
guished medical, engineering, and other scientific fields. The mean ranges from five
citations per researcher in the engineering field to more than 23 in the medical field, and the
standard deviations again are high in every field. Hence, we conclude that Italian uni-
versities exhibit large variety also as to research quality.
Finally, we measured the commercial orientation of universities as the share of uni-
versity research funded by private companies out of the total research budget of the
university in 2003. The mean share is 25%, with a minimum of 0% and a maximum of
86%. Again the differences among Italian universities are remarkable.
4 The methodology of the econometric analysis
4.1 The specification of the econometric model
In order to capture the differences between ASUs and other NTBFs, we estimate the
following augmented Gibrat law dynamic growth model:
LSize
i;t
¼b
0
þ b
1
LSize
i;t1
þ b
2
LAge
i;t1
þ b
3
LSize
i;t1
2
þb
4
LAge
i;t1
2
þ b
5
ðLSize
i;t1
LAge
i;t1
Þþb
6
U
i;t1
þ b
7
CV
i;t1
þ b
8
DASU
i
þ b
9
LSize
i;t1
DASU
i
þ b
10
LAge
i;t1
DASU
i
þ b
11
LSize
i;t1
2
DASU
i
þ b
12
LAge
i;t1
2
DASU
i
þ b
13
LSize
i;t1
LAge
i;t1
DASU
i
þ b
14
U
i;t1
DASU
i
þ b
15
CV
i;t1
DASU
i
þ e
i;t
ð1Þ
In Eq. 1 LSize
i,t
indicates the size of sample firms measured by the logarithm of the
number of employees (including owners-managers) at time t, and LAge
i,t-1
is the logarithm
of firms’ age at t-1. Here we adopt the approach of Evans (1987a, b), thus we include in
the model specification also the squared values of LSize
i,t-1
and LAge
i,t-1
, and the inter-
active term between LSize
i,t-1
and LAge
i,t-1
. U
i,t-1
is a group of variables describing the
characteristics of the universities located in the same province as sample NTBFs. CV
i,t-1
is
a group of control variables including measures of (i) the human capital of firms’ founders,
(ii) firms’ obtainment of venture capital (VC), (iii) the level of economic development of
the provinces where firms are located, and (iv) firms’ sector of activity (for an in depth
description of the U
i,t
and CV
i,t
variables see Sect. 4.2). Finally, DASU
i
is a dummy
variable equal to one if the firm is an academic start-up and e
i,t
are i.i.d. disturbance terms.
The inclusion in Eq. 1 of the lagged dependent variable as one of the covariates and the
possible endogenous nature of the relationship between VC financing and firm size require
the use of appropriate estimation techniques. In fact, as long as regressors are correlated
with disturbance terms, both pooled ordinary least squares and random effects estimators
are likely to produce biased estimates. Therefore, following the recent literature on
dynamic panel data models (see Arellano and Bond 1991; Blundell and Bond 1998; Bond
2002), we resort to the generalized method of moments (GMM) procedure and estimate
model (1) by the GMM-System estimator. This approach, originally proposed by Blundell
and Bond (1998), extends the GMM-DIF estimator (first differenced), and additional
moment conditions are used in order to obtain more efficient estimates. In particular, in
The contribution of university research to the growth of academic start-ups 125
123
addition to using lagged levels of the series as instruments for first differences (as in
GMM-DIF), additional information is extracted using first differences as instruments for
variables in levels.
6
This augmented GMM estimator requires the assumption of mean
stationary of the series and is particularly appropriate where series are highly persistent
(see Bond 2002).
Finally, since we perform the analysis on micro units using several aggregate variables
at the province level as covariates, we need to control for the potential downward bias in
the estimated errors (see Moulton 1990). We do so by clustering the standard errors at firm
level.
4.2 The explanatory variables of the econometric model
The explanatory variables used in the estimation of model (1) are presented in Table 4.
They include measures of the characteristics of the universities located in the same
provinces as sample NTBFs and a series of control variables measuring founder-, firm-,
and industry-specific characteristics.
Let us first consider the variables measuring the characteristics of local universities.
Faculty size is measured by ResearcherDensity
t-1
that we calculated as the sum of the
researchers employed in either economic or technical faculties
7
in the universities located
in the same provinces as sample start-ups at time t-1, t-2, and t-3 divided by the
population (in thousands of residents) of the province at t-1 to control for the size of the
area.
We measured the research quality of universities located in a given province by the ratio
of the number of citations obtained till year 2001 by articles published in the international
journals monitored by the ISI in the 1995–1999 period by researchers employed by these
universities to the number of these researchers (Breno et al. 2002). Consistently with the
descriptive statistics on the Italian university system provided in Sect. 3.2, we distin-
guished the citation indexes in engineering (EngQuality), medical (MedQuality), and other
scientific fields (ScienceQuality). For each scientific field the effects of the research quality
of universities on the growth of local NTBFs may differ according to the sector of activity
of recipient companies; hence, in the model we include the interactive terms between the
indexes of citations and the industry dummies DBio and DNoBio. DBio is equal to 1 for
companies operating in biotechnology, pharmaceutics, and advanced materials, conversely
DNoBio equals 1 for companies in other industries. We do not include in the model
DBio 9 EngQuality as university research in engineering fields is likely to have negligible
effects on biotech companies. For similar reasons we exclude also DNoBio 9 MedQuality.
The variables measuring the research quality are time-invariant. This clearly is a lim-
itation of our study. However, as the citation ratio is unlikely to have varied substantially in
the period under scrutiny, differences across provinces are likely to have remained quite
constant over time. Therefore we are confident that considering the research quality as a
constant between 1994 and 2003 is unlikely to lead to biased results.
Finally, the commercial orientation of local universities (%PrivateBudget) has been
calculated through a two-step procedure. First for each province we computed the share of
6
In particular, considering LSize
i,t-1
as endogenous implies the use of instruments dated t-3 for the
equation in first differences and instruments dated t-2 for the equation in level.
7
The economic faculties include economics, management, and political sciences, while the technical
faculties are engineering, chemistry, physics, geology, mathematics, biology, medicine, pharmaceutics, and
computer science.
126 M. G. Colombo et al.
123
Table 4 Definition of the explanatory variables
Variable Description
DASU One for academic start-ups
Gibrat variables
LSize
t-1
Logarithm of the size of the firm at t-1 measured by the number of employees
LAge
t-1
Logarithm of the number of years since firm’s foundation at t-1
LSize^2
t-1
Squared logarithm of the size of the firm at t-1 measured by the number of
employees
LAge^2
t-1
Squared logarithm of the number of years since firm’s foundation at t-1
LAge
t-1
9LSize
t-1
Product of LAge and LSize
Characteristics of local universities
ResearcherDensity
t-1
Average number of researchers (i.e., professors, lecturers, Ph.D. students)
employed in local universities in either economic or technical faculties at
t-1, t-2 and t-3, divided by the population of the province (in thousands
of residents) in t-1
DBio 9 MedQuality Ratio of the number of citations obtained till 2001 by articles published in the
1995–1999 period in medical fields to the number of researchers in the
province multiplied by a dummy equal to one for firms operating in
biotechnology, pharmaceutics, and advanced materials
DNoBio 9 EngQuality Ratio of the number of citations obtained till 2001 by articles published in the
1995–1999 period in engineering fields to the number of researchers in the
province multiplied by a dummy equal to one for firms not operating in
biotechnology, pharmaceutics, and advanced materials
DBio 9 ScienceQuality Ratio of the number of citations obtained till 2001 by articles published in the
1995–1999 period in other scientific fields to the number of researchers in the
province multiplied by a dummy equal to one for firms operating in
biotechnology, pharmaceutics, and advanced materials
DNoBio 9 ScienceQuality Ratio of the number of citations obtained till 2001 by articles published in the
1995–1999 period in other scientific fields to the number of researchers in the
province multiplied by a dummy equal to one for firms not operating in
biotechnology, pharmaceutics, and advanced materials
%PrivateBudget Average share over the fiscal years 2002 and 2003 of privately funded research
out of the total research budget of local universities
Founder-specific variables
WorkExp Average number of years of industrial work experience gained by founders
before firm’s foundation
Education Average number of years of founders’ education
DManager One for companies with one or more founders with a prior management
position in a large or medium company (i.e., number of employees greater
than 100)
Firm-specific variables
DVC
t-1
One for companies that obtained VC financing at t-1
DVC
t-2
One for companies that obtained VC financing at t-2
DVC
t-3
One for companies that obtained VC financing at t-3
Geographical and industry controls
Infrastructure Value of the index measuring regional infrastructures in 1989 (mean value
among Italian regions = 100; source: Centro Studi Confindustria, 1991)
DICTManufacturing One for companies that operate in aerospace and ICT manufacturing
DBiotech One for companies that operate in biotechnology, pharmaceutics, and advanced
materials
The contribution of university research to the growth of academic start-ups 127
123
the research performed by local universities and funded by private companies out of the
total research budget of these universities both in 2002 and in 2003.
8
Then we assigned
%PrivateBudget the average value over this 2 year period for the province where the focal
NTBF is located.
9
In this paper we are interested in (i) examining the effects of the characteristics of local
universities on the growth rates of ASUs and (ii) assessing whether such characteristics
differently affect the growth rates of ASUs and those of other NTBFs. Hence, the
explanatory variables in Eq. 1 also include the interactive terms between the dummy
DASU which equals 1 for academic start-ups, and the above mentioned measures of the
characteristics of local universities. According to hypotheses H1a,b and H2a,b, we expect
the coefficients of ResearcherDensity
t-1
? DASU 9 ResearcherDensity
t-1
, DASU 9
ResearcherDensity
t-1
, DNoBio 9 EngQuality ? DASU 9 DNoBio 9 EngQuality, DBio
9 MedQuality ? DASU 9 DBio 9 MedQuality, DNoBio 9 ScienceQuality ? DASU 9
DNoBio 9 ScienceQuality, DBio 9 ScienceQuality ? DASU 9 DBio 9 ScienceQuality,
DASU 9 DNoBio 9 EngQuality, DASU 9 DBio 9 MedQuality, DASU 9 DNoBio 9
ScienceQuality, and DASU 9 DBio 9 ScienceQuality to be positive. Conversely, fol-
lowing hypotheses H3a,b, we predict that the coefficients of %PrivateBudget ? DASU 9
%PrivateBudget and DASU 9 %PrivateBudget are negative.
Table 4 continued
Variable Description
DSoftware One for companies that operate in software
DInternet One for companies that operate in internet and TLC services
DMultimediaContent One for companies that operate in the multimedia content sector
8
The first available data on the amount of financing of Italian universities refer to the fiscal year 2000. In
fact, it was only in 2001 that the National Committee for the Evaluation of the Academic System (Comitato
Nazionale per la Valutazione del Sistema Universitario, CNVSU) started collecting the financial accounts of
Italian universities and making available aggregated data. The reports of the CNVSU on the fiscal years
2000 and 2001 classified the revenues of Italian universities in three categories: financing from MIUR (the
Italian Ministry of University and Research), university internal funds and financing from external sources.
Since the publication of the report on the fiscal year 2002, three subcategories have been distinguished
within the ‘‘external sources’’ category: financing from the European Union, from other public research
organizations and from other organizations. As we are interested in the share of research funded by the
private sector, we have to focus on the last subcategory. Hence, in calculating %PrivateBudget, we could not
use data on the fiscal years 2000 and 2001. Note that the category ‘‘financing from other organizations’’
includes both financing from firms and financing from other sources (e.g., foundations). However, the role of
foundations in financing academic research in Italy was fairly limited in the observation period. Hence, even
though %PrivateBudget is likely to overestimate financing from firms, absent more fine grained data, it can
be considered as a reasonably good proxy.
9
%PrivateBudget was calculated as the average value over two subsequent years in order to reduce random
fluctuations. We are aware that this variable, being time-invariant, might generate biases that may distort the
estimates. In particular, as financing to universities from the Italian Central Government has decreased
during the last decade, universities have started looking for additional sources of income thus probably
raising the share of funds from private sources out of the total budget. This increase might have been more
relevant for northern provinces as the industrial system of this area is more developed and local companies
are likely to be more prone to finance academic research. Therefore, our measure of %PrivateBudget might
overestimate the commercial orientation of universities located in northern provinces in the first years of the
period under scrutiny. As a consequence, if in this period the growth rates of NTBFs located in northern
provinces were higher (lower) than those of the companies located elsewhere because of unobserved effects,
the estimates of the coefficient of %PrivateBudget might reveal a upward (downward) bias.
128 M. G. Colombo et al.
123
As to the control variables, we first introduce into the model measures of founder-
specific characteristics of sample firms. These variables allow to control for the alleged
positive impact on firm growth of the human capital of the founding team (for a survey see
Colombo and Grilli 2005; Storey 1994). We consider the level of education measured by
the mean number of years of university education of founders (Education) and the mean
number of years of prior industrial work experience (WorkExp). We also add a dummy
(DManager) equal to 1 if prior to the establishment of the new venture, one (or more)
founder(s) had a managerial position in a medium or large company (i.e., number of
employees greater than 100).
10
As to firm-specific controls, we first control for the obtainment of VC financing which
according to the financial literature has a positive effect on firm growth (for a survey see
Bertoni et al. 2007). In order to do so we include in our estimates three dummy variables,
DVC
t-1
, DVC
t-2
, and DVC
t-3
, which are equal to 1 if, respectively, at time t-1, t-2, and
t-3 sample NTBFs got access to VC financing. We also control for firm location: being
located in a developed area is likely to positively influence growth as it enables NTBFs to
benefit from positive externalities that may arise from external assets with public good
nature (e.g., transport system, telecommunication infrastructure, efficient market for sup-
port services). Hence, in the model we include the variable Infrastructure, that reflects the
level of economic development in 1989 of the province where firms are located (source:
Centro Studi Confindustria 1991). It is calculated as the average of the following indexes:
per capita value added, share of manufacturing out of total value added, employment
index, per capita bank deposits, automobile-population ratio, and consumption of electric
power per head.
11
We also allowed firm- and founder-specific control variables to differ
according to the status (ASU or non-ASU) of firms. For this purpose we inserted in the
model specification a series of interactive terms between the control variables and the
dummy DASU.
Finally, the industry-specific controls are five sectoral dummies: DInternet, DSoftware,
DMultimediaContent, DICTManufacturing, and DBiotech. These variables equal 1 for
firms in the internet and telecommunication services, software, multimedia content, ICT
manufacturing, and biotechnology and pharmaceutics industry, respectively. The baseline
of the estimates is the robotics and automation industry. As to these latter variables we
have no specific predictions.
5 Results of the econometric analysis
The results of the econometric analysis are illustrated in Tables 5 and 6.
12
In Table 5 we
present the estimates of the equation including the control variables only (Model 1) and the
10
In small family-owned Italian companies decision authority is often centralized in the owner-managers’
hands (see Colombo and Delmastro 1999), while salaried managers are assigned execution tasks. So,
entrepreneurial learning associated with such managerial positions generally is fairly limited.
11
We also controlled for other characteristics of the provinces where sample NTBFs are located, namely
provincial deflated GDP per capita, provincial deflated GDP per capita rescaled on a national basis and
provincial population or provincial density of population. The results are almost unchanged. They are
available from the authors upon request.
12
As we mentioned in footnote 6, GMM-SYS estimate requires the use of instruments for LSize
i,t-1
,
(LSize
i,t-1
)
2
and LSize
i,t-1
9LAge
i,t-1
dated t-3 for the equation in first differences. Hence, the sample used
for the estimates presented in Tables 5 and 6 excludes all RITA NTBFs for which data on size were
available for less than three consecutive years.
The contribution of university research to the growth of academic start-ups 129
123
control variables plus the independent variables (Model 2), but the model specification
does not make any distinction between ASUs and other NTBFs. In Table 6 we present the
estimates of the equation including also the interactive terms between the independent and
control variables and DASU.
As to the control variables, the estimates of Model 1 in Table 5 show that the coefficient
of firm size (LSize
t-1
) is significantly smaller than 1 (and higher than 0) while the coef-
ficient of LSize^2
t-1
is positive and significant. Hence, we may conclude that the relation
Table 5 The determinants of the growth of NTBFs: a GMM-system model
LSize
t
Model 1 Model 2
a
0
Constant 0.553(0.269)* -0.422(0.652)
a
1
ResearcherDensity
t-1
– -0.033(0.020)
a
2
DBio 9 ScienceQuality – 0.041(0.044)
a
3
DBio 9 MedQuality – 0.005(0.011)
a
4
DNoBio 9 ScienceQuality – 0.003(0.005)
a
5
DNoBio 9 EngQuality – 0.000(0.008)
a
6
%PrivateBudget – -0.102(0.209)
a
7
LSize
t-1
0.696(0.086)** 0.639(0.089)**
a
8
LSize^2
t-1
0.045(0.022)* 0.059(0.023)**
a
9
LAge
t-1
-0.011(0.142) 0.061(0.142)
a
10
LAge^2
t-1
-0.004(0.043) -0.018(0.043)
a
11
LAge
t-1
9 LSize
t-1
0.000(0.056) -0.008(0.061)
a
12
DManager -0.112(0.155) -0.097(0.163)
a
13
WorkExp -0.005(0.007) -0.005(0.007)
a
14
Education -0.012(0.010) 0.003(0.013)
a
15
DVC
t-1
0.330(0.140)* 0.366(0.147)*
a
16
DVC
t-2
0.299(0.069)** 0.293(0.069)**
a
17
DVC
t-3
0.154(0.071)* 0.157(0.071)*
a
18
Infrastructure 0.002(0.002) 0.003(0.002)
Industry controls Yes Yes
Wald v
2
test
a
1
= a
2
= a
3
= a
4
= a
5
= a
6
= 0 – 6.13(6)
a
7
= a
8
= a
9
= a
10
= a
11
= 0 3814.63(5)** 3835.92(5)**
a
12
= a
13
= a
14
= 0 2.30(3) 1.22(3)
a
15
= a
16
= a
17
= 0 25.00(3)** 22.20(3)**
Number of observations 3201 3201
Number of groups 487 487
AR(1) -9.56 ** -9.28**
AR(2) -1.02 -1.02
Sargan 150.25(139) 144.65(133)
Hansen 144.30(139) 137.38(133)
Legend: * Significance level greater than 5%; ** Significance level greater than 1%
AR(1) and AR(2) are tests of the null hypothesis of respectively no first- or second-order serial correlation.
Sargan and Hansen are tests of the validity of the overidentifying restrictions based on the efficient two-step
GMM estimator. Standard deviations in round brackets. Standard errors clustered at firm level
130 M. G. Colombo et al.
123
Table 6 The determinants of the growth of ASUs and other NTBFs: a GMM-system model
LSize
t
a
0
Constant -0.433(0.928) a
19
DASU 0.412(0.996)
a
1
ResearcherDensity
t-1
-0.037(0.043) a
20
DASU 9 ResearcherDensity
t-1
0.091(0.063)
a
2
DBio 9 ScienceQuality 0.028(0.054) a
21
DASU 9 DBio 9 ScienceQuality 0.135(0.100)
a
3
DBio 9 MedQuality 0.012(0.016) a
22
DASU 9 DBio 9 MedQuality dropped
a
4
DNoBio 9 ScienceQuality -0.023(0.014)* a
23
DASU 9 DNoBio 9 ScienceQuality 0.018(0.021)
a
5
DNoBio 9 EngQuality -0.009(0.013) a
24
DASU 9 DNoBio 9 EngQuality 0.107(0.052)**
a
6
%PrivateBudget 0.749(0.531) a
25
DASU 9 %PrivateBudget -2.066(0.834)**
a
7
LSize
t-1
0.600(0.144)*** a
26
DASU 9 LSize
t-1
0.531(0.385)
a
8
LSize^2
t-1
0.073(0.033)** a
27
DASU 9 LSize^2
t-1
-0.106(0.084)
a
9
LAge
t-1
0.174(0.175) a
28
DASU 9 LAge
t-1
-0.993(0.414)**
a
10
LAge^2
t-1
-0.039(0.064) a
29
DASU 9 LAge^2
t-1
0.294(0.126)**
a
11
LAge
t-1
9 LSize
t-1
-0.025(0.098) a
30
DASU 9 (LAge
t-1
9 LSize
t-1
) -0.044(0.174)
a
12
DManager -0.367(0.359) a
31
DASU 9 DManager 0.502(0.468)
a
13
WorkExp 0.013(0.015) a
32
DASU 9 WorkExp -0.015(0.025)
a
14
Education 0.029(0.047) a
33
DASU 9 Education -0.024(0.054)
a
15
DVC
t-1
0.470(0.197)** a
34
DASU 9 DVC
t-1
-0.102(0.296)
a
16
DVC
t-2
0.269(0.090)*** a
35
DASU 9 DVC
t-2
-0.054(0.181)
a
17
DVC
t-3
0.143(0.087) a
36
DASU 9 DVC
t-3
0.010(0.227)
a
18
Infrastructure 0.006(0.003)*
Industry controls YES
Wald v
2
test on the effects of the characteristics of local universities
a
1
= a
2
= a
3
= a
4
= a
5
= a
6
= 0 6.22(6)
a
20
= a
21
= a
23
= a
24
= a
25
= 0 15.25(5)***
a
1
? a
20
= a
2
? a
21
= a
4
? a
23
= a
5
? a
24
= a
6
? a
25
= 0 16.08(5)***
The contribution of university research to the growth of academic start-ups 131
123
Table 6 continued
LSize
t
Wald v
2
test - hypotheses testing
H1a: a
1
? a
20
= 0 1.48(1)
H2a: a
2
? a
21
= 0 2.15(1)
H2a: a
4
? a
23
= 0 0.10(1)
H2a: a
5
? a
24
= 0 3.86(1)**
H3a: a
6
? a
25
= 0 4.56(1)**
Wald v
2
test on the control variables
a
7
= a
8
= a
9
= a
10
= a
11
= 0 1267.95(5)***
a
26
= a
27
= a
28
= a
29
= a
30
= 0 11.02(5)*
a
8
= a
10
= a
11
= 0 9.06(3)**
a
27
= a
29
= a
30
= 0 8.23(3)**
a
12
= a
13
= a
14
= 0 1.33(3)
a
31
= a
32
= a
33
= 0 1.15(3)
a
15
= a
16
= a
17
= 0 11.05(3)**
a
34
= a
35
= a
36
= 0 0.36(3)
Number of groups 487
AR(1) -8.97***
AR(2) -0.97
Sargan 122.81(116)
Hansen 112.93(116)
Legend: * Significance level greater than 10%; ** Significance level greater than 5%; *** Significance level greater than 1%
AR(1) and AR(2) are tests of the null hypothesis of respectively no first- or second-order serial correlation. Sargan and Hansen are tests of the validity of the overidentifying
restrictions based on the efficient two-step GMM estimator. Standard deviations in round brackets. Standard errors clustered at firm level
132 M. G. Colombo et al.
123
between firm size and firm growth is U-shaped. That is, firm growth decreases with firm
size up to a size corresponding to about 29 employees and then it increases when firm size
grows larger. Conversely, firm’s age (LAge
t-1
and LAge^2
t-1
) and its interactive term with
firm size (LAge
t-1
9 LSize
t-1
) do not significantly affect growth.
In line with prior studies, VC financing has a positive impact on the growth of NTBFs.
The coefficients of the three lagged VC variables are all positive and significant. Con-
versely, the variables measuring the human capital of NTBFs’ founding teams, the index
measuring regional infrastructures and the five sectoral dummies are not statistically sig-
nificant at conventional confidence levels.
In Model 2 we insert the independent variables in the model specification. The impact of
the size of the research staff of local universities, and the quality and commercial orien-
tation of the knowledge produced by these universities on the growth of NTBFs is
negligible. In fact, the null hypothesis that the corresponding variables have no effect on
firm growth cannot be rejected at conventional confidence level (v
2
(6) = 6.13).
Table 6 presents the estimates of the equation including the interactive terms of all the
variables listed in the first column of Table 5 with DASU. Even though the findings are
somewhat weaker than expected, they suggest that, consistently with our predictions, the
characteristics of local universities have different effects on the growth of ASUs and other
NTBFs. In fact, as is apparent from the first Wald tests at the bottom of Table 6, while the
null hypothesis that the characteristics of local universities jointly have no effects on the
growth rates of non-academic NTBFs cannot be rejected (v
2
(6) = 6.22), the same
hypothesis indeed is rejected for ASUs (v
2
(5) = 15.25). Similarly the null hypothesis that
the coefficients of the corresponding interactive terms be null is also rejected
(v
2
(5) = 16.08).
Let us now consider in detail the theoretical hypotheses developed in Sect. 2.3.
Hypothesis H1a is not supported by our results. In fact, the coefficient of Researcher-
Density
t-1
? DASU 9 ResearcherDensity
t-1
is positive, as we hypothesized, but not
significant. The results regarding research quality are mixed. On the one hand the estimates
suggest that, consistently with hypothesis H2a, ASUs greatly benefit from the quality of the
scientific knowledge produced by local universities in engineering fields, with the coef-
ficient of DNoBio 9 EngQuality ? DASU 9 DNoBio 9 EngQuality being positive and
significant at 5%. On the other hand, the coefficient of DBio 9 ScienceQuali-
ty ? DASU 9 DBio 9 ScienceQuality is positive, but insignificant, while the one of
DNoBio 9 ScienceQuality ? DASU 9 DNoBio 9 ScienceQuality, in contrast with H2a,
is negative, but again not significant. Lastly, the results regarding commercial orientation
confirm our expectations. Consistently with H3a, the greater the commercial orientation of
university research, the lower the growth rates of local ASUs, as is apparent from the
negative coefficient of %PrivateBudget ? DASU 9 %PrivateBudget, significant at 5%.
Conversely, as to non-academic NTBFs, %PrivateBudget exhibits a positive (though not
significant) coefficient. We interpret this result as a signal that the commercial orientation
of local academic institutions prevents ASUs from having access to the results of uni-
versity research thus negatively affecting the growth of these firms. Conversely, the
estimates provide some (admittedly weak) indication that the greater the share of research
funded by private organizations out of the total research budget of local universities, the
more positive might the impact of this research be on the growth of non-academic NTBFs.
We also inserted two squared terms (Sq%PrivateBudget and DASU 9 Sq%PrivateBudget)
into the model specification so as to check whether a curvilinear specification better fits the
data. The corresponding coefficients were insignificants (the estimates are available from
the authors upon request).
The contribution of university research to the growth of academic start-ups 133
123
Let us now consider the hypotheses on the differences between ASUs and other NTBFs.
First, the argument that the size of the research staff of local universities has a more
positive effect on the growth rates of ASUs than on those of other NTBFs (H1b) is not
supported: the coefficient of DASU 9 ResearcherDensity
t-1
is positive but not significant.
Second, the coefficients of DASU 9 DNoBio 9 ScienceQuality, DASU 9 DBio 9
ScienceQuality, and DASU 9 DNoBio 9 EngQuality are positive as was predicted by
hypothesis H2b, but only the latter one is significant (at 5%). Altogether these results
provide an admittedly weak indication that ASUs’ superior ability to detect, absorb, and
effectively use the knowledge produced within local universities strengthen the contribu-
tion of university research to firm growth. Third, the coefficient of DASU 9
%PrivateBudget is negative and significant at 5%. In line with hypothesis H3b, the
commercial orientation of university research has a more negative effect on the growth
rates of ASUs than on those of other NTBFs.
13
In order to evaluate the magnitude of the effects of these characteristics of local uni-
versities on the growth of ASUs and other NTBFs we performed the following simple
exercise. First, we calculated the predicted size at the fifth year of existence of an ASU and
a non-academic NTBF located in a province where the research quality and commercial
orientation of local universities were equal to the means across our sample. For this
purpose we used the coefficients reported in Table 6 and we set all the dummy variables to
their median and all the remaining variables to their mean. We used these values as the
benchmark and compared them respectively with the estimated size at the fifth year of
existence of an ASU and a non-academic NTBF located in a province where the research
quality of local universities is high (i.e., equal to the 90th percentile; again all the
remaining continuous variables were set to their mean and the dummy variables to their
median). The difference with respect to the benchmark appears large for the ASU and
marginal for the other NTBF: the size results 171% greater than the benchmark for the
ASU, and just 6% smaller for the other NTBF. Finally we compared the benchmark with
the estimated size of an ASU and a non-academic NTBF located in a province where the
commercial orientation of local universities is low (i.e., equal to the 10th percentile). In
this latter case the size resulted 56% greater than the benchmark for the ASU, and 13%
lower for the other NTBF. This calculation suggests that the effects of both research
quality and commercial orientation on the growth of ASUs are of great economic
magnitude.
Let us now briefly consider the control variables. As to the five Gibrat’s law variables, a
Wald test on the coefficients of the five interactive terms which include the dummy DASU
rejects the null hypothesis that the dependence of the growth rates of ASUs and other
NTBFs on firms’ size and age be the same (v
2
(5) = 1267.95). As to non-academic NTBFs,
the coefficient of LSize
t-1
is significantly smaller than 1 (and higher than 0) while the
coefficient of LSize^2
t-1
is positive and significant. Hence, the relation between firm size
and firm growth is U-shaped for these firms. Conversely, the coefficient of LSize
t-1
?
DASU 9 LSize
t-1
is significant and above unity, while the one of LSize^2
t-1
?
DASU 9 LSize^2
t-1
is not significant. This finding is opposed to the stylized fact high-
lighted by the empirical literature on Gibrat’s law (see Evans 1987a, b; Hart and Oulton
13
It is fair to acknowledge that 95 out of the 487 NTBFs included in our sample are located in provinces
where there were no universities in the period under consideration. For these firms all the variables mea-
suring the characteristics of local universities are always equal to zero over the time horizon 1994–2003. As
a check for robustness, we reestimated the models using only the data on the 392 firms located in provinces
where at least one university existed. The results are almost unchanged. They are available from the authors
upon request.
134 M. G. Colombo et al.
123
1996; Sutton 1997; Caves 1998) that smaller firms tend to grow faster than larger ones.
Second, while age seems to have no effects on the growth rates of non-academic NTBFs
(neither LAge
t-1
nor LAge^2
t-1
are significant), it significantly affects those of ASUs.
More specifically, the coefficient of LAge
t-1
? DASU 9 LAge
t-1
is negative while the
one of LAge^2
t-1
? DASU 9 LAge^2
t-1
is positive and both are significant at 5%. So, the
growth rates of ASUs first decrease and then increase with firm age.
As to the other controls, the coefficients of the interactive terms relating to: (i) the
variables capturing VC financing, (ii) the variables measuring the human capital of firms’
founding teams, (iii) the measure of the level of economic development of the province
where sample firms are located, and (iv) the sectoral dummies are not statistically sig-
nificant at conventional levels as is shown by the Wald tests at the bottom of Table 6.In
other words, we were not able to detect any significant differences in the impact of these
variables on the growth rates of ASUs and other NTBFs. Therefore, we simplified the
specification of the model dropping the interactive terms between DASU and these control
variables. Of course we could not exclude the Gibrat’s law variables as is shown by the
Wald test at the bottom of Table 6. The results of these estimates do not differ from those
discussed above (they are available from the authors upon request).
6 Concluding remarks
In this paper we aimed at examining empirically the impact of university research on the
growth of individual local firms. More specifically, we analyzed the effects on academic
start-ups’ growth rates of the characteristics of local universities (i.e., universities located
in the same provinces as focal firms); we also examined the differences with other NTBFs.
We argued that the characteristics of local universities influence firms’ growth potential.
To what extent this potential is realized depends on firm-specific characteristics. ASUs
exhibit stronger ties in the research environment than other NTBFs, and are better able to
detect and absorb academic knowledge; hence, we expected the growth rates of ASUs to be
more sensitive to the characteristics of local universities than those of other NTBFs. In the
empirical part of the work, we studied the determinants of firm growth within a sample
composed of 48 ASUs and 439 non-academic NTBFs that operate in Italy in both man-
ufacturing and service high-tech sectors. Sample firms were observed from 1994 to 2003.
Their growth rates were analyzed through the estimation of an augmented Gibrat law
dynamic panel data model.
The results of the econometric estimates provide evidence of a substantial impact of
university research on the development of local ASUs. In particular, higher research
quality in the engineering field positively contributes to the growth of ASUs. This result is
in line with the view that ASUs are endowed with sufficient absorptive capacity and social
contacts in the research environment so as to be able to benefit from academic knowledge.
Instead, research quality does not influence the growth of non-academic NTBFs. Quite
interestingly, the estimates also suggest that the commercial orientation of university
research may have negative consequences on the growth of local companies as it may
reduce the knowledge available for absorption by the companies better able to exploit it. In
particular, the commercial orientation of university research has a clear negative impact on
the growth of ASUs, while the effect is negligible for non-academic NTBFs. Conversely,
faculty size has a negligible impact on the growth of both ASUs and non-academic NTBFs.
While describing the results of the study, a number of limitations are to be noted as they
open up interesting avenues for future research. The main limitations of the study come
The contribution of university research to the growth of academic start-ups 135
123
from the characteristics of the dataset. The most serious problem of this work is the failure
to control for possible unobserved heterogeneity. In particular, the positive impact of
university quality on the growth of ASUs may depend on unobserved factors. Academic
founders are assumed to learn from their experiences within their parent research orga-
nizations, and to exploit such knowledge in their own firms. It is also natural to assume that
higher quality universities produce more valuable knowledge, thus researchers from these
organizations can more effectively draw on their experiences to enhance the growth of the
ASUs they found. Hence, ASUs from higher quality universities are likely to exhibit higher
growth rates than the remaining ASUs.
14
In this work we cannot disentangle this latter
effect from the one of the quality of the research performed in local universities. In order to
properly distinguish the two effects we would need for each ASU both a measure of the
quality of the parent university at firm foundation and one of the quality of local univer-
sities in subsequent years. Unfortunately these data were not available at a suitable
aggregation level.
Second, as we mentioned above, the findings are weaker than expected. This possibly is
a consequence of the low number of ASUs included in our sample. Thus in order to
strengthen our conclusions extending the dataset seems warranted. One may also wonder
whether the results illustrated in this paper can be generalized. These results might clearly
be influenced by the specific institutional setting in which Italian NTBFs are embedded.
Similarly, the 10-year period under consideration might exhibit specific characteristics
relating to such aspects as the policy of research organizations towards knowledge dif-
fusion. Therefore, these results wait for further corroboration from replications of this
study in different countries and time periods.
A third limitation possibly regards the geographic unit of analysis. Even though the
extant studies on the effects of university research on Italian companies usually use the
province as the unit of observation (Piergiovanni et al. 1997; Rodrı
´
guez-Pose and Refolo
2003), it would be interesting to extend this analysis to understand what scale of geo-
graphic aggregation is the most appropriate to investigate the contribution of university
research on local firm growth.
Besides the limitations of this study we think that altogether these results both con-
tribute to the extant literature and have interesting policy implications. In terms of
literature development, we extend previous empirical works on the effects of university
research on local companies by showing that, besides contributing to firms’ innovation
performances and influencing firms’ localization choices, university research may also
affect the growth of individual local firms. However, the mere presence of a university in a
province is not sufficient to stimulate the growth of the companies located in the same
province. Our results clearly indicate that the effect on growth depends on the charac-
teristics of both local universities and recipient firms.
Prior research had already recognized that knowledge externalities from university
research and their effects within the private sector are influenced by several characteristics
of the universities close to focal firms. Conversely, the role of the specificities of recipient
firms had not been explored yet. In accordance with the ‘‘absorptive capacity’’ literature
(Cohen and Levinthal 1989, 1990; Zahra and George 2002), we documented that ASUs
because of their ‘‘genetic characteristics’’ (Colombo and Piva 2008a), are better positioned
than other NTBFs to detect, absorb, and utilize the scientific knowledge generated by local
universities. It would be interesting to highlight whether and how NTBFs that are not
endowed with favorable genetic characteristics can exploit this knowledge. For instance,
14
For a similar argument in a different context see Klepper (2007).
136 M. G. Colombo et al.
123
one may wonder whether the establishment of long-term collaborative relations with local
universities may have similar beneficial effect on firm growth.
Let us now focus on the policy implications of this study. First, as was said above, our
findings suggest that ASUs are better equipped than other NTBFs to exploit the results of
academic research. This makes them appealing alliance partners or acquisition targets. As a
corollary, ASUs may favor the circulation in the private sector of valuable knowledge
generated within universities, playing a crucial bridging role (for a similar view see Stuart
et al. 2007). In fact, ASUs can easily detect the source of this knowledge, translate it, and
transfer it to their customers, alliance partners and to the firms that may acquire them, thus
allowing the successful exploitation of university research results by other companies.
Through this indirect mechanism, academic knowledge would affect also the growth of
companies that are endowed with smaller absorptive capacity and thus are unable to
directly profit from the proximity of universities. Hence, in order to enable more wide-
spread exploitation of the knowledge produced within universities, university managers
should not only encourage academic personnel to create new ventures, but help these new
ventures to establish collaborations with other companies.
Second, our study suggests that although university cooperation with private companies
through contract research agreements may be an effective mechanism to commercialise
research results, it could inhibit knowledge transfer to other local companies and thus
hinder the generation of positive externalities from university research. This negative
consequence of commercial orientation needs to be duly taken in consideration by uni-
versity managers while designing and assessing the effectiveness of policy instruments
aimed at regulating the interactions between university and the private sector.
Acknowledgments Support from the REBASPINOFF project promoted by the PRIME Network of
Excellence, the PICO project (‘‘Academic entrepreneurship, from knowledge creation to knowledge dif-
fusion’’, contract n°028928) sponsored by the Sixth Framework Programme, the FIRB 2003 fund, the PRIN
2006 project 2006132439_002 and a grant from Unicredit is gratefully acknowledged. The authors would
like to thank E. Rasmussen, A. Hughes, the participants in the XX RENT conference, and two anonymous
referees for their useful comments.
References
Acs, Z. J., Audretsch, D. B., & Feldman, M. P. (1992). Real effects of academic research. The American
Economic Review, 82(1), 363–367.
Acs, Z. J., Audretsch, D. B., & Feldman, M. P. (1994). R&D spillovers and recipient firm size. The Review
of Economics and Statistics, 76(2), 336–340. doi:10.2307/2109888.
Agrawal, A. (2001). University-to-industry knowledge transfer: Literature review and unanswered ques-
tions. International Journal of Management Reviews, 3(4), 285–302. doi:10.1111/1468-2370.00069.
Aldrich, H. E., Kallenberg, A., Marsden, P., & Cassell, J. (1989). In pursuit of evidence: Sampling pro-
cedures for locating new businesses. Journal of Business Venturing, 4(6), 367–386. doi:10.1016/
0883-9026(89)90008-6.
Anselin, L., Varga, A., & Acs, Z. J. (1997). Local geographic spillovers between University research and
high technology innovations. Journal of Urban Economics, 42(3), 422–448. doi:10.1006/juec.
1997.2032.
Arellano, M., & Bond, S. (1991). Some tests of specification for panel data: Monte Carlo evidence and an
application to employment equations. The Review of Economic Studies, 58(2), 277–297. doi:
10.2307/2297968.
Argyres, N. S., & Liebeskind, J. P. (1998). Privatizing the intellectual commons: Universities and the
commercialization of biotechnology. Journal of Economic Behavior & Organization, 35(4), 427–454.
doi:10.1016/S0167-2681(98)00049-3.
The contribution of university research to the growth of academic start-ups 137
123
Arrow, K. (1962). Economic welfare and the allocation of resources for invention. In R. Nelson (Ed.), The
rate and direction of inventive activity. Princeton, NJ: Princeton University Press.
Audretsch, D. B., Keilbach, M., & Lehmann, E. E. (2005). Entrepreneurship and economic growth. London:
Oxford University Press.
Audretsch, D. B., & Lehmann, E. E. (2005a). Does the knowledge spillover theory of entrepreneurship hold
for regions? Research Policy, 34(8), 1191–1202. doi:10.1016/j.respol.2005.03.012.
Audretsch, D. B., & Lehmann, E. E. (2005b). Mansfield’s missing link: The impact of knowledge spillovers
on firm growth. The Journal of Technology Transfer, 30(1–2), 207–210. doi:10.1007/s10961-
004-4367-6.
Audretsch, D. B., & Lehmann, E. E. (2006). Entrepreneurial access and absorption of knowledge spillovers:
Strategic board and managerial composition for competitive advantage. Journal of Small Business
Management, 44(2), 155–166. doi:10.1111/j.1540-627X.2006.00161.x.
Audretsch, D. B., Lehmann, E. E., & Warning, S. (2004). University spillovers: Does the kind of knowledge
matters? Industry and Innovation, 11(3), 193–205. doi:10.1080/1366271042000265375.
Audretsch, D. B., & Stephan, P. E. (1996). Company-scientist locational links: The case of biotechnology.
The American Economic Review, 86(3), 641–652.
Bade, F., & Nerlinger, E. A. (2000). The spatial distribution of new technology-based firms. Papers in
Regional Science, 79(2), 155–176. doi:10.1007/s101100050041.
Barringer, B. R., Jones, F. F., & Neubaum, D. O. (2005). A quantitative content analysis of the charac-
teristics of rapid-growth firms and their founders. Journal of Business Venturing, 20(5), 663–687. doi:
10.1016/j.jbusvent.2004.03.004.
Bertoni, F., Colombo, M. G., & Grilli, L. (2007). Venture capital financing and the growth of new tech-
nology-based firms: A longitudinal analysis. Working paper.
Birley, S. (1984). Finding the new firm. Proceedings of the Academy of Management Meetings, 47, 64–68.
Blundell, R., & Bond, S. (1998). Initial conditions and moment conditions in dynamic panel data models.
Journal of Econometrics, 87(1), 115–143. doi:10.1016/S0304-4076(98)00009-8.
Bond, S. (2002). Dynamic panel data models: A guide to micro data methods and practice. Portuguese
Economic Journal, 1(2), 141–162. doi:10.1007/s10258-002-0009-9.
Breno, E., Fava, G. A., Guardabasso, V., & Stefanelli, M. (2002). La ricerca scientifica nelle universita
`
italiane. Una prima analisi delle citazioni della banca dati ISI (Scientific research in Italian Uni-
versities. A first analysis of citations from ISI database). CRUI: Roma.
Caves, R. (1998). Industrial organization and new findings on the turnover and mobility of firms. Journal of
Economic Literature, 36(4), 1947–1982.
Cockburn, I., & Henderson, R. (1998). Absorptive capacity, coauthoring behavior, and the organization of
research in drug discovery. The Journal of Industrial Economics, 46(2), 157–182.
Cohen, W. M., Florida, R., Randazzese, L., & Walsh, J. (1998). Industry and the academy: Uneasy partners
in the cause of technological advance. In R. G. Noll (Ed.), Challenges to research universities (ch. 7).
Washington, DC: Brookings Institute Press.
Cohen, W. M., & Levinthal, D. A. (1989). Innovation and learning: The two faces of R&D. The Economic
Journal, 99(397), 569–596. doi:10.2307/2233763.
Cohen, W. M., & Levinthal, D. A. (1990). Absorptive capacity: A new perspective on learning and inno-
vation. Administrative Science Quarterly, 35(1), 128–152. doi:10.2307/2393553.
Colombo, M. G., & Delmastro, M. (1999). Some stylized facts on organisation and its evolution. Journal of
Economic Behavior & Organization, 40(3), 252–274. doi:10.1016/S0167-2681(99)00067-0
.
Colombo, M. G., & Grilli, L. (2005). Founders’ human capital and the growth of new technology-based
firms: A competence-based view. Research Policy, 34(6), 795–816. doi:10.1016/j.respol.2005.03.010.
Colombo, M. G., & Piva, E. (2008a). Strenghts and weaknesses of academic start-ups: A conceptual model.
IEEE Transactions on Engineering Management, 55(1), 37–49. doi:10.1109/TEM.2007.912807.
Colombo, M. G., & Piva, E. (2008b). Firms’ genetic characteristics, competence enlarging strategies and
performances: A comparison between academic and non-academic high-tech start-ups. Working paper.
Dasgupta, P., & David, P. (1994). Towards a new economics of science. Research Policy, 23(5), 487–521.
doi:10.1016/0048-7333(94)01002-1.
Del Barrio-Castro, T., & Garcı
´
a-Quevedo, J. (2005). Effects of university research on the geography of
innovation. Regional Studies, 39(9), 1217–1229. doi:10.1080/00343400500389992.
Di Gregorio, D., & Shane, S. (2003). Why do some universities generate more start-ups than others?
Research Policy, 32(2), 209–227. doi:10.1016/S0048-7333(02)00097-5.
Drucker, J., & Goldstein, H. A. (2007). Assessing the regional economic development impacts of univer-
sities: A review of current approaches. International Regional Science Review, 30(1), 20–46. doi:
10.1177/0160017606296731.
138 M. G. Colombo et al.
123
Evans, D. S. (1987a). The relationship between firm growth, size, and age: Estimates for 100 manufacturing
industries. The Journal of Industrial Economics, 35(4), 567–581. doi:10.2307/2098588.
Evans, D. S. (1987b). Tests of alternative theories of firm growth. The Journal of Political Economy, 95(4),
657–674. doi:10.1086/261480.
Feeser, H. R., & Willard, G. E. (1990). Founding strategy and performance: A comparison of high and low
growth high tech firms. Strategic Management Journal, 11(2), 87–98. doi:10.1002/smj.4250110202.
Feldman, M. P., & Florida, R. (1994). The geographic sources of innovation: Technological infrastructure
and product innovation in the United States. Annals of the Association of American Geographers,
84(2), 210–229. doi:10.1111/j.1467-8306.1994.tb01735.x.
Fischer, E., & Reuber, R. (2003). Support for rapid growth firms: A comparison of the views of founders,
government policy makers and private sector resource providers. Journal of Small Business Man-
agement, 41(4), 346–365.
Fischer, M. M., & Varga, A. (2003). Spatial knowledge spillovers and university research: Evidence from
Austria. The Annals of Regional Science, 37(2), 303–322. doi:10.1007/s001680200115.
Gimeno, J., Folta, T., Cooper, A., & Woo, C. (1997). Survival of the fittest? Entrepreneurial human capital
and the persistence of underperforming firms. Administrative Science Quarterly, 42(4), 750–783. doi:
10.2307/2393656.
Goldstein, H. A., & Drucker, J. (2006). The economic development impacts of universities on regions: Do
size and distance matter? Economic Development Quarterly, 20(1), 22–43. doi:10.1177/089124
2405283387.
Goldstein, H. A., & Renault, C. S. (2004). Contributions of universities to regional economic development:
A quasi-experimental approach. Regional Studies, 38(7), 733–746. doi:10.1080/003434004200026
5232.
Hart, P. E., & Oulton, N. (1996). Growth and size of firms. The Economic Journal, 106(438), 1242–1252.
doi:10.2307/2235518.
Jaffe, A. B. (1989). Real effects of academic research. The American Economic Review, 79(5), 957–970.
Jaffe, A. B., Trajtenberg, M., & Henderson, R. (1993). Geographic localization of knowledge spillovers as
evidenced by patent citations. The Quarterly Journal of Economics, 108(3), 577–598. doi:
10.2307/2118401.
Klepper, S. (2007). Disagreements, spinoffs, and the evolution of Detroit as the capital of the U.S. auto-
mobile industry. Management Science, 53(4), 616–631. doi:10.1287/mnsc.1060.0683.
Lim, K. (2000). The many faces of absorptive capacity: Spillovers of copper interconnect technology for
semiconductor chips. Mimeo: MIT.
Little, A. D. (1977). New Technology-based firms in the United Kingdom and the federal Republic of
Germany. London: Wilton House.
Moulton, B. R. (1990). An illustration of a pitfall in estimating the effects of aggregate variables on micro
units. The Review of Economics and Statistics, 72(2), 334–338. doi:10.2307/2109724.
Mowery, D. C., & Ziedonis, A. A. (2001). The geographic reach of market and non-market channels of
technology transfer: Comparing citations and licenses of university patents. NBER Working Paper
8568.
Murray, F. (2004). The role of academic inventors in entrepreneurial firms: sharing the laboratory life.
Research Policy, 33(4), 643–659. doi:10.1016/j.respol.2004.01.013.
Nelson, R. (1959). The simple economics of basic scientific research. The Journal of Political Economy,
67(3), 297–306. doi:10.1086/258177.
O’Shea, R. P., Allen, T. J., Chevalier, A., & Roche, F. (2005). Entrepreneurial orientation, technology
transfer and spin-off performance of US universities. Research Policy, 34(7), 994–1009. doi:
10.1016/j.respol.2005.05.011.
Piergiovanni, R., Santarelli, E., & Vivarelli, M. (1997). From which source do small firms derive their
innovative inputs? Some evidence from Italian industry. Review of Industrial Organization, 12(2),
243–258. doi:10.1023/A:1007781501147.
Powers, J. B., & McDougall, P. P. (2005). University start-up formation and technology licensing with firms
that go public: A resource-based view of academic entrepreneurship. Journal of Business Venturing,
20(3), 291–311. doi:10.1016/j.jbusvent.2003.12.008.
Roberts, E. B. (1991). High tech entrepreneurs: Lessons from MIT and beyond. New York: Oxford Uni-
versity Press.
Roberts, E. B., & Malone, D. (1996). Policies and structures for spinning off new companies from research
and development organizations. R & D Management, 26(1), 17–48. doi:10.1111/j.1467-9310.1996.
tb00927.x.
Rodrı
´
guez-Pose, A., & Refolo, M. C. (2003). The link between local production systems and public and
university research in Italy. Environment & Planning A, 35(8), 1477–1492. doi:10.1068/a35297.
The contribution of university research to the growth of academic start-ups 139
123
Rosenberg, N., & Nelson, R. (1994). American universities and technical advance in industry. Research
Policy, 23(3), 323–348. doi:10.1016/0048-7333(94)90042-6.
Stern, S. (2004). Do scientists pay to be scientists? Management Science, 50(6), 835–853. doi:
10.1287/mnsc.1040.0241.
Storey, D. J. (1994). Understanding the small business sector. London: Thomson Learning.
Stuart, T. E., Ozdemir, S. Z., & Ding, W. W. (2007). Vertical alliance networks: The case of university-
biotechnology-pharmaceutical alliance chains. Research Policy, 36(4), 477–498. doi:10.1016/
j.respol.2007.02.016.
Sutton, J. (1997). Gibrat’s legacy. Journal of Economic Literature, 35(1), 40–59.
Varga, A. (2000). Local academic knowledge transfers and the concentration of economic activity. Journal
of Regional Science, 40(2), 289–309. doi:10.1111/0022-4146.00175.
Wright, M., Birley, S., & Mosey, S. (2004). Entrepreneurship and university technology transfer. The
Journal of Technology Transfer, 29(3–4), 235–246. doi:10.1023/B:JOTT.0000034121.02507.f3.
Zahra, S. A., & George, G. (2002). Absorptive capacity: A review, reconceptualization, and extension.
Academy of Management Review, 27(2), 185–203. doi:10.2307/4134351.
Zucker, L. G., Darby, M. R., & Armstrong, J. S. (1998). Intellectual capital and the firm: The technology of
geographically localized knowledge spillovers. Economic Inquiry, 36, 65–86.
Zucker, L. G., Darby, M. R., & Armstrong, J. S. (2000). University science, venture capital and the
performance of U.S. biotechnology firms. Mimeo, Los Angeles: University of California.
Zucker, L. G., Darby, M. R., & Brewer, M. (1999). Intellectual capital and the birth of U.S. biotechnology
enterprises. The American Economic Review, 88(1), 290–306.
140 M. G. Colombo et al.
123
