Regional Studies

ISSN: 0034-3404 (Print) 1360-0591 (Online) Journal homepage: http://www.tandfonline.com/loi/cres20

Collective Learning and Relational Capital in Local

Innovation Processes

Roberta Capello & Alessandra Faggian

To cite this article: Roberta Capello & Alessandra Faggian (2005) Collective Learning
and Relational Capital in Local Innovation Processes, Regional Studies, 39:1, 75-87, DOI:
10.1080/0034340052000320851

To link to this article: https://doi.org/10.1080/0034340052000320851

Published online: 18 Aug 2010.

Submit your article to this journal

Article views: 1189

View related articles

Citing articles: 198 View citing articles

Full Terms & Conditions of access and use can be found at

http://www.tandfonline.com/action/journalInformation?journalCode=cres20
Regional Studies, Vol. 39.1, pp. 75–87, February 2005

Collective Learning and Relational Capital in

Local Innovation Processes
ROBERTA CAPELLO* and ALESSANDRA FAGGIAN†
*Department of Management, Economics & Industrial Engineering, Politecnico di Milano, Piazza Leonardo da Vinci 32,
I-20133 Milan, Italy. Email: Roberta.Cappello@polimi.it
†Department of Economics, School of Business, University of Reading, Urs Building, Whiteknights, PO Box 219, Reading
RG6 6AW, UK. Email: A.Faggian@rdg.ac.uk

(Received April 2003: in revised form February 2004)

C R. and F A. (2005) Collective learning and relational capital in local innovation processes, Regional Studies
39, 75–87. Innovative capacity of firms has traditionally been explained through intra-firm characteristics. In the more recent
literature, much emphasis has been put on determinants that are external to the firm. These external factors, called knowledge
spillovers, refer to the positive externalities that firms receive in terms of knowledge from the environment in which they
operate. Geographers and industrial economists underline the importance of knowledge spillovers. As the paper underlines, an
important difference exists between the approach of knowledge spillover and that of socialized processes of local knowledge
creation developed by regional economists; while in the former, the mere probability of contacts explains local knowledge
transfer, in the latter, the channels through which knowledge spills over a local area are clearly identified in the relational capital
of the area. Relational capital is defined as all relationships – market relationships, power relationships and cooperation –
established between firms, institutions and people, which stem from a strong sense of belonging and a highly developed capacity
of cooperation typical of culturally similar people and institutions. The main aims of the paper are twofold. The first is to
underline the major conceptual differences between industrial and regional economists. The second is to provide a quantitative
empirical approach, using econometric techniques, to verify the existence and importance of relational capital on the innovation
activity of firms. Proxies are found to represent the channels through which local knowledge develops at the local level and
therefore indirectly of relational capital. The different regional, sectoral and firm characteristics are also analysed to understand
whether they influence the role relational capital has on innovation. It is indeed reasonable to expect that relational capital will
play a different role in different regional, sectoral and firm’s contexts.

Collective learning Local knowledge Endogenous local economic development Relational capital

C R. et F A. (2005) L’apprentissage collectif et le capital relationnel dans les processus d’innovation locale,
Regional Studies 39, 75–87. Traditionnellement, la capacité à l’innovation des entreprises s’explique par des caractéristiques intra-
entreprises. Dans la documentation plus récente, on met plutôt l’accent sur les facteurs externes. Ces facteurs externes, appellés
retombées de connaissance, traduisent les effetes externes positifs dont profitent les entreprises en termes de la connaissance qui
provient du milieu où elles opèrent. Les géographes et les économistes industriels soulignent l’importance des retombées de
connaissance. Comme le souligne l’article, il y a une importante différence entre l’approche de retombées de connaissance et
l’approche qui met l’accent sur les processus socialisés de la création de la connaissance locale, développée par des économistes
régionaux. Alors que dans celle-là, c’est la probabilité de se faire des contacts qui explique le transfert de la connaissance locale,
pour ce qui est de cette dernière, ce sont les canaux par lesquels la connaissance s’étend à une zone locale qui se voient
clairement identifier dans le capital relationnel de la zone. Le capital relationnel veut dire tous les rapports – rapports de marché,
rapports de pouvoir, partenariat – établis entre les entreprises, les institutions et les individus, qui proviennent d’un fort sentiment
d’appartenance et d’une capacité à la coopération très développée, qui caractérise des individus et des institutions qui se
ressemblent du point de vue culturel. L’article cherche primo à souligner les principales différences conceptuelles entre les
économistes régionaux et industriels, et secundo à fournir une approche empirique quantitative, employant des techniques
économétriques afin de vérifier la présence et l’importance du capital relationnel pour l’innovation des entreprises. On emploie
des variables muettes afin de représenter les canaux par lesquels la connaissance locale se développe sur le plan local et, par la
suite, indépendamment du capital relationnel. On analyse aussi les diverses caractéristiques des régions, des secteurs et des
entreprises pour comprendre si, oui ou non, elles influent sur le rôle du capital relationnel quant à l’innovation. Il est fort
probable que le rôle du capital relationnel varie suivant le contexte de la région, du secteur ou de l’entreprise.

Apprentissage collectif Connaissance locale Développement économique local autochtone Capital relationnel

0034-3404 print/1360-0591 online/05/010075-13 ©2005 Regional Studies Association DOI: 10.1080/0034340052000320851

http://www.regional-studies-assoc.ac.uk
76 Roberta Capello and Alessandra Faggian
C R. und F A. (2005) Kollektives Lernen und relationales Kapital in örtlich begrenzten Innovationsprozessen,
Regional Studies 39, 75–87. Innovationsfähigkeit von Firmen ist tradionell mit innerbetrieblichen Eigenschaften erklärt worden.
In der neueren Literatur dagegen hat man von außen beeinflussenden Faktoren große Bedeutung für Firmen beigelegt. Diese
außerbetrieblichen Faktoren, die auch unter dem Begriff Wissensverbreitung laufen, beziehen sich auf die positiven Außenfak-
toren, die Firmen in Form von Fachkenntnissen von der Umgebung empfangen, in der sie wirken. Wie auch dieser Aufsatz
unterstreicht, besteht ein wichtiger Unterschied zwischen dem Ansatz von Wissensverbreitung und dem in der Gesellschaft
allgemein stattfindenden Vorgang der Wissensgenerierung am Orte, wie er von Regionalwirtschaftswissenschaftlern angestrebt
wird. Während im ersteren Fall die bloße Wahrscheinlichkeit von Kontakten die Übertragung von Wissen am Ort erklärt,
werden im letzteren die Kanäle, durch welche Fachkenntnisse in örtlich begrenzte Gebiete gelangen, klar im relationalen Kapital
der Gegend herausgestellt. Relationales Kapital wird als alle Beziehungen definiert – Marktbeziehungen, Machtbeziehungen,
Zusammenarbeit – die zwischen Firmen, Institutionen und Menschen betehen, und auf einem starken Gefühl der Zusammenge-
hörigkeit sowie einer für kulturell verwandte Menschen und Institutionen hoch entwickelten Fähigkeit zur Zusammenarbeit
beruhen. Der vorliegende Aufsatz verfolgt zwei Hauptziele: das erste ist, die begrifflichen Hauptunterschiede zwischen
Betriebswissenschaftlern und Regionalwirtschaftlern zu unterstreichen, das zweite, einen quantitativen, empirischen Ansatz mit
Hilfe ökonmetrischer Techniken vorzulegen, um das Vorhandensein und Bedeutung relationalen Kapitals für die innovative
Tätigkeit von Firmen zu bestätigen. Es wird festgestellt, daß Vertreter Kanäle darstellen, durch die sich Ortskenntnisse auf
örtlicher Ebene entwickeln und daher indirekt auf relationales Kapital auswirken. Es werden auch die verschiedenen regionalen,
sektoralen und firmeneigenen Charakterzüge analysiert, um zu erfahren, ob sie die Rolle des relationalen Kapital bei Innovation
beeinflussen. Es ist, in der Tat, nur zu erwarten, daß relationales Kapital in unterschiedlichen Regional-, Sektoren- und
Firmenzusammenhängen auch eine andersartige Rolle spielen wird.
Kollektives Lernen Ortskenntnisse Endogene örtliche Wirtschaftsentwicklung Relationales Kapital
C R. y F A. (2005) Aprendizaje colectivo y capital relacional en los procesos locales de innovación, Regional
Studies 39, 75–87. La capacidad innovadora de las empresas se ha explicado tradicionalmente por medio de caracterı́sticas intra-
empresariales. En la literatura más reciente, se ha puesto mucho énfasis en los determinantes que son externos a las empresas.
Dichos factores externos, denominados desbordamientos de conocimiento (knowledge spillovers), se refieren a las externalidades
positivas que reciben las empresas en materia de conocimiento del ambiente en el que operan dichas empresas. Los geógrafos y
los economistas industriales destacan la importancia de los desbordamientos de conocimiento. Tal y como el artı́culo subraya,
existe una diferencia importante entre el enfoque de desbordamientos de conocimiento y el de los procesos socializables de
creación de conocimiento local por parte de los economistas regionales; mientras que en el primero, la mera probabilidad de
contactos explica la transferencia de conocimiento a nivel local, en el segundo, los canales por los cuales se desborda el
conocimiento sobre una área local están claramente identificados en el capital relacional de la zona. El capital relacional se
define como todas las relaciones – relaciones de mercado, relaciones de poder, cooperación – que se establecen entre las
empresas, instituciones y personas, las cuales derivan de un fuerte sentido de pertenencia y una capacidad altamente desarrollada
de cooperación que es tı́pica de personas e instituciones culturalmente similares. Este artı́culo tiene dos objetivos fundamentales:
el principal objetivo es destacar las principales diferencias conceptuales que existen entre los economistas industriales y regionales.
El segundo es ofrecer un planteamiento empı́rico cuantitativo, utilizando técnicas de tipo econométrico, con el fin de verificar
la existencia e importancia del capital relacional en la actividad innovadora de las empresas. Se encuentran una serie de proxies
para representar los canales a través de los cuales se desarrolla el conocimiento a nivel local y por tanto indirectamente respecto
del capital relacional. También se analizan las diferentes caracterı́sticas de las empresas, sectoriales y regionales, con el fin de
entender si influyen de alguna forma en el rol que el capital relacional tiene en la innovación. En efecto, es razonable esperar
que el capital relacional juegue un rol diferente en contextos sectoriales, regionales y de empresas diferentes.
Aprendizaje colectivo Conocimiento local Desarrollo económico local endógeno Capital relacional

JEL classifications: R00, R11

INTRODUCTION These contrasting results have led industrial econom-

Innovative capacity of firms has traditionally been
explained through intra-firm characteristics, firms’ size
being the most important. A wave of empirical studies
identifies small firms as the engines of technological
change and innovative activity, at least in certain indus-
tries (A and A, 1993; A and
V, 1996; P et al., 1987; R,
1989). This statement and these empirical findings
contrast the well-known observation that since research
and development (R&D) expenditure is concentrated
in large firms and that innovative output strongly
depends on R&D inputs (S, 1982), large firms
are expected to drive the technological process.
ists to look for other explanatory variables. In the
recent literature, much emphasis has been put to deter-
minants that are external to the firm; these external
factors are called knowledge spillovers and refer to the
positive externalities that firms receive in terms of
knowledge from the environment in which they
operate (A et al., 1997; 2000; A and
A, 1990; A and F,
1996, F, 1994; F and A,
1999; A et al., 2000; D G et al., 2001).
The importance of knowledge spillovers has been
emphasized by geographers, as well as by industrial
and regional economists. As this paper underlines,
 Collective Learning and Relational Capital in Local Innovation Processes
important differences exist in the way the subject is
treated by different schools of thoughts, at both the
theoretical and empirical levels. Geographers, sup-
ported by industrial economists, tend to stress the
importance of physical proximity as a condition for the
diffusion of local knowledge; firms located in areas
characterized by a high number of scientific activities
register increasing returns in innovation processes
thanks to knowledge they acquire from the external
environment. The channel through which knowledge
is spread around is a simple epidemic process. At the
same time, another stream of thought, led by regional
economists, has put much emphasis on the concept of
relational capital as a fundamental concept to under-
stand the way in which knowledge spills over a local
area.
Relation capital is defined as the set of all relation-
ships – market relationships, power relationships and
cooperation – established between firms, institutions
and people that stem from a strong sense of belonging
and a highly developed capacity of cooperation typical
of culturally similar people and institutions. The exist-
ence of high relational capital in an area generates stable
cooperation between firms and their local suppliers and
customers, an efficient local labour market with a high
internal mobility of employees and spin-offs from local
firms, which are considered as the main channels
through which knowledge spreads over a local area
(A, 1986; A and K, 1988;
C, 1991, 1999; M et al., 1993; R
et al., 1997; RERU, 1999; C and
C, 2000). New insights in the way knowledge
develops over space emerge in this way.
The concept of relational capital has a striking resem-
blance with the concept of ‘social capital’ developed
by P (1993); social capital is intended as all
social networks, collective and institutional rules that
through the development of trust and sense of
belonging to a local community guarantee to overcome
more easily market failures in the coordination of
decision processes and gives rise to a community gover-
nance.1 The present authors prefer to refer to ‘relational’
capital rather than to ‘social’ capital for a specific reason.
Social capital exists wherever a local society exists,
while relational capital refers to the (rare) capability of
exchanging different skills, interacting among different
actors, trusting with each other and cooperating even
at a distance with other complementary organizations.
Especially the capacity of interaction and cooperation
among local actors will be taken into consideration
herein.
The main aims of the paper are twofold. The first is
to go into more detail about the way knowledge
spillovers are treated and interpreted in the literature
by underlining the theoretical and empirical differences
in the approaches used. The second is to provide
a quantitative empirical approach using econometric
techniques to verify the existence and importance of
77
relational capital on the innovation activity of firms.
Proxies are found to represent the channels of collective
knowledge and therefore indirectly of relational capital.
The different regional, sectoral and firms’ characteristics
will also be analysed in order to understand whether
they influence the role relational capital has on firms’
innovation; being relational capital meant to be the
capacity of local firms to cooperate, it can be influenced
by regional, sectoral and firm’s contexts.
The paper is organized as follows. The second
section summarizes the main points of industrial and
regional economists on the determinants of innovation.
The third section describes the methodology used. The
fourth section presents the results of the empirical
analysis. Finally, the fifth section suggests some conclu-
sions and policies.
SPACE AS AN INPUT FOR
INNOVATION ACTIVITY
Determinants of innovation: different theoretical approaches
In the last two decades, several authors focused their
attention on the innovation phenomenon and its deter-
minants. So many contributions were produced that
the term ‘Innovation Economics’ was coined to gather
them under a common label. On the one hand, this
large interest created the preconditions to take big steps
forward on the subject. On the other hand, the different
theoretical approaches developed did not always agree
with each other.
As far as the determinants of innovation are con-
cerned, the so-called ‘intra-firm’ determinants of
innovation became the main explanation of different
innovation performances. Among these determinants,
a crucial role was played by firm size and R&D
expenditure, both internal and external to the firm.
Empirical analyses led in the past to some contrasting
results, underlining the need for introducing other
variables vital for fostering the innovation process. Both
large and small firms can prove to be very innovative
due to certain factors, which are completely exogenous
to the firm itself, such as their sector and location.2 As
G (1995) underlines, the proximity to other
firms can be essential in increasing the innovation
capacity of a firm, independently of internal firm
characteristics.
There is an agreement in the literature on the fact
that ‘physical proximity’ among firms plays a crucial
role in improving the innovative capacity of firms.
Space matters because of the existence of ‘knowledge
spillovers’.
However, the way in which space is conceptualized
into innovation processes through the concept of
‘knowledge spillovers’ is very different among different
schools of thought.
One school, represented by geographers and indus-
trial economists, adds to the traditional ‘intra-firm’
78 Roberta Capello and Alessandra Faggian
variables the space dimension, as simple physical prox-
imity among innovative actors. Very solid econometric
results prove the importance of physical proximity
and shed light on some of the previously mentioned
contrasting results. A second school, represented by
regional economists, deals with a more complex con-
cept of ‘space’. As will be shown in more detail
below, physical space is coupled in this approach with
‘relational’ space, which is made of all the different
relationships built among local actors.
The well-known concept of ‘milieu innovateur’ has
been developed by this second school of thought and
is based on what regional scientists call ‘territory’,
rather than ‘space’ (C, 2004). The ‘milieu’ is a
place capable of maintaining a long-term competi-
tiveness thanks to its adaptability to external changes.
It is a dynamic concept that partly overlaps with the
static industrial district view. The firms belonging to
the milieu are close not only in geographical terms,
but also, and above all, in terms of culture. As M
and L (1992, p. 2) underline ‘though the milieu
has a spatial dimension, it does not correspond to a
well defined geographical area; the milieu is an organic
framework including market and non-market links’.
The concept of ‘innovative milieu’ is interesting and
it found further refinements during the 1990s. In
particular, two main refinements can be identified. On
the one hand, some authors tried to draw a parallel
between the milieu innovateur and some concepts
borrowed from the biological sciences. In this respect,
the contribution of G (1998), for instance,
underlines that the correct theoretical approach to the
milieu should be ‘systematic’. The reasons for the
success of a milieu are not simply additive. There are
feedbacks and reciprocal causality relationships between
the factors of success of a milieu. It seems, therefore,
that the best approach to study a milieu is a non-
linear one, a ‘study of complexity’ (P and
S, 1983), that stresses that the milieu is a
system in continuous evolution, not in equilibrium,
and is strongly dependent on its past. This current of
thought makes some very good points on the concept
of milieu, but it does not answer the criticism of
abstractness, which has been moved to the concept.
On the other hand, some authors tried to show the
similarities between the concept of milieu and some
concepts belonging to mainstream economics in order
to give a more formalized nature to the milieu concept
and also to attempt an empirical measurement of its
effects. The reason for doing so was to answer some of
the first criticisms on the milieu innovateur concept.
Some authors, indeed, noticed how the milieu inno-
vateur was a sort of ‘abstract entity’, an ideal archetype,
not applicable to reality. The contribution of C
(2002) shows how the milieu innovateur concept can
be integrated into a neo-classical endogenous growth
model à la R (1986) and L (1988), without
distorting its initial hypotheses or conclusions. Also,
some empirical tests of the milieu innovateur concept
have been published recently (C, 1999a, b).
The present paper can be included in this second
approach, since its main purpose is to formalize the
milieu concept, allowing one to test the importance of
relational space on the innovativeness and economic
performance of firms and to analyse the best conditions
under which relational capital maximizes its incentive
to innovation.
The next section will focus on the definitions of
relational capital, collective learning and the channels
that make collective learning possible. It is essential to
define these concepts clearly before proceeding,
because they are the starting points of the following
empirical analysis.
Geographical and relational knowledge spillovers: similarities
and differences
As already noted above, the concept of knowledge
spillovers has long been recognized as essential in
studying the innovation process. Many authors address
the problem. A and V (1994),
for instance, try to measure the effect of knowledge
spillovers on innovation – measured in terms of new
patents (using data on Italian firms) – and they find a
significant effect of these spillovers on small- and
medium-sized firms. The definition they use of spill-
overs, though, is not very wide, including only the
physical proximity (physical distance) to universities or
research centres.3
A-B (1999) extends the definition
of spillovers also to include the proximity of a high
number of firms belonging to the same sector. Again,
as in A and V (1994), he finds a
significant positive relationship between knowledge
spillovers – measured in terms of R&D expenditure
and researchers of firms in the local area – and the
innovative performance of firms.
Despite recognizing that proximity to universities,
research centres and other firms – belonging both to
the same or different sectors – in the local area is
important, the phenomenon of knowledge spillovers is
much more complex. A high concentration of firms
belonging to the same sector in an area is not enough
to explain the high innovation of the area itself. It
is necessary to define which channels convey these
knowledge spillovers and allow them to spread over
the territory.
The concept of relational space, first introduced in
the milieu theoretical framework, becomes crucial in
this respect.4 Relational space is defined as the set of all
relationships – market relationships, power relationships
and cooperation – established between firms, institu-
tions and people that stem from a strong sense of
belonging and a highly developed capacity of
cooperation typical of culturally similar people and
institutions. The concept of relational capital helps to
 Collective Learning and Relational Capital in Local Innovation Processes
Fig. 1. Physical versus relational space
underline the difference between the approach of two
schools of thought mentioned above.
Fig. 1 underlines the comparison between the two
approaches. On the one hand, if one starts from a
concept of pure physical space, the precondition for
knowledge spillovers is the physical proximity to firms
of the same sector (to exploit specialization economies),
to firms of different sectors (to exploit economies
coming from diversification), and to universities and
research centres, typical places where knowledge is
produced. Physical proximity increases the probability
of contacts between the economic actors, therefore
allowing knowledge to spread more easily and produce
useful spillovers.
On the other hand, if one takes into account the
concept of relational space, the precondition for the
creation of knowledge spillovers becomes the cultural
proximity of actors, i.e. their sense of belonging to the
area, their capability of interacting and the sharing of
common values. This cultural proximity is the basis for
the existence of relational capital, which in turn is
formed by the following:
79
Ω Explicit cooperation among actors.
Ω Implicit cooperation among actors.
Ω Public and private partnership.
Relational capital is therefore the ‘substratum’ of collec-
tive learning exactly like physical space is the necessary
condition for the ‘traditional’ knowledge spillovers.
It can be seen in Fig. 1 that the parallel between the
two approaches is almost complete, but an important
difference must be emphasized. In the industrial
approach, there is no clear definition of the channels
through which physical proximity materializes into
geographical knowledge spillovers. All that is known is
that the proximity to other firms or research centres
positively influences the performance and the innova-
tiveness of a firm, but it is not clear how this happens.
Everything is due to pure probabilistic mechanisms.
Conversely, in the regional approach, the channels
through which the relational capital becomes collective
learning are clearly defined:
Ω High mobility of local labour force.
Ω Stable and fruitful relationships with local customers
and suppliers.
Ω Spin-offs.
C (1995, p. 203) defines collective learning as
the ‘dynamic and cumulative process of production of
knowledge, which is due to interaction mechanisms
typical of an area characterised by a strong sense of
belonging and relational synergies’. The internal cohe-
sion promotes the introduction of new products or
production techniques and reduces the uncertainty
linked to innovations. The space is not just ‘physical’,
it is something more. It is a space created by men,
both the result of and the precondition for collective
learning, an active input rather than a passive surface
(C and B, 1996).
The definition of relational capital influences the
indices that will be used in the empirical analysis below.
Cultural as well as social proximity is the basis for the
existence of relational capital, which in turn gives
rise to collective learning processes; the existence of
collective processes through strong local relationships
indirectly assumes that social and cultural proximity is
present locally.
Towards an empirical analysis: testable propositions
Before presenting the database and empirical results,
the steps to be followed, and the theoretical hypotheses
behind them, will be stated more clearly.
The main idea behind the paper is that ‘relational
capital matters’ in promoting innovation and some
indicators can be found to prove this point. Very few
empirical works (C, 1999a) attempt to verify
empirically the role of relational capital on innovation
and not without a reason. The lack of contributions is
directly proportional to the size of the challenge. Most
80 Roberta Capello and Alessandra Faggian
of the concepts being dealt with are very abstract. So
abstract, that many authors dispute the possibility of
measuring them empirically.
Relational capital is an intangible and complex asset
that cannot be directly measured. The real challenge is
to find a way to make this concept empirically verifi-
able. If relational capital does exist, there must be a
way to evaluate its effects on firms.
The first challenge is to try to find an empirical
counterpart to the concept of relational capital to show
its importance on the innovative performance of firms.
Is it true that, irrespective of their location, sector and
intra-firm characteristics, all firms benefit from the
presence of relational capital?
The second step is then to understand whether the
effect of relational capital on innovation varies accord-
ing to the characteristics of the different areas consid-
ered. Is it reasonable to assume that different areas with
different productive structures (i.e. the present case
studies) can all benefit from relational capital? And if
so, which components of relational capital are more
important in each territorial context?
Lastly, once it is identified which component of
relational capital works at best in each territorial con-
text, do firms’ size and sector influence the capability
of exploiting these components?
DATA AND THE SAMPLE
Database and the variables
The empirical analysis is based on micro-data taken
directly from interviews to firms’ managers via a
questionnaire jointly prepared by three research groups
belonging to three different Italian universities. The
database includes 217 firms located in three different
areas: Milan, Piacenza and the Belluno (Cadore) area
in the Veneto region.
The questionnaire used can be divided into different
sections. The first includes general questions about
firms such as their age, sector, dimensions (both in
terms of employees and turnover), exports and com-
petitive position in the local and global market. The
second focuses on the innovative capacity of the firm,
which is the core of the analysis. As already noted, one
of the basic hypotheses of the paper is that knowledge
spillovers turn into a higher innovative capacity. It is
essential, therefore, to have detailed information on
this variable both in terms of its dynamics over time
and of its distribution in space. The rest of the question-
naire is devoted to specific features that can influence
the learning process of firms and consequently its
innovative behaviour. These features include local
labour market characteristics, relationships with local
suppliers and customers, and spin-offs and cooperation
with other firms both locally and non locally.
There are several advantages to using the common
questionnaire as described above. First, the joint effort
Table 1. Variables used
Variables Method used to build the
variables
Degree of innovation Percentage of turnover due to
sales of innovative products
(average in the last 5 years)
Dimension Turnover (millions lire)
Research and development Percentage of turnover invested
expenditure in research and development
expenditure
Relational capital (measured via
collective learning channels):
Local labour market Percentage of new employees
(i.e. employees hired in the last
5 years) from local firms
Specialized local labour Percentage of new employees
market (i.e. employees hired in the last
5 years) from local firms
belonging to the same sector
Diversified local labour Percentage of new employees
market (i.e. employees hired in the last
5 years) from local firms
belonging to different sectors
Innovative cooperation with Importance of local suppliers and
local suppliers and customer customers in fostering product
innovation (average of scores)
High-technology sectors Dummy variable (1óhigh-
technology firm)
of three different universities facilitated the inter-
viewing of a larger number of firms, making it possible
to obtain more robust econometric estimations.
Second, the larger sample allowed the innovation pro-
cess to be studied from different angles. The sample
indeed includes firms that differ in size, sector and
location. All previous studies based on micro-data
focused on only one of these dimensions due to the
limits imposed by the sample size. Widening the sample
meant one could analyse the subject in its entirety.
Table 1 summarizes the variables used in the empiri-
cal analysis. As input of innovation, the percentage of
turnover spent in R&D was chosen. As a proxy for
the innovation, the percentage of turnover due to sales
of innovative products was chosen. This last variable
seemed to have a very high degree of reliability as
reported by the interviewers. It was more difficult to
build a variable representing relational capital.
It was decided to use some variables representing
so-called ‘collective learning channels’ (C,
1999a, b). The literature describes three channels: the
local labour market, cooperation with local suppliers
and customers, and spin-offs. The present research,
however, focused on the first two, since spin-offs are
less common and would have restricted the sample
considerably. A good proxy for the importance of local
labour market was the percentage of new employees
from firms5 belonging to the local area. The importance
of cooperation with local suppliers and customers was
 Collective Learning and Relational Capital in Local Innovation Processes 81
Table 2. Location quotients by sector in the three analysed areas important role in the Emilia-Romagna region, espe-
cially the food industry. The location quotient in this
Cadore
Milan Piacenza (Belluno) sector for Piacenza is 1.32 with respect to Italy.
The same holds for the mechanical sector and the
Sector production of non-metal products. A high location
Food 0.57 1.32 0.57
Textile 0.53 0.40 0.67 quotient is exhibited in the production of energy,
Timber 0.39 0.78 1.86 gas and water (1.85). The chemical and electronic
Press 1.79 0.79 0.45 industries, which are so strong in the Milanese area, are
Chemical 2.68 0.22 0.24 incredibly low in Piacenza (0.22 and 0.61, respectively).
Plastic 1.22 0.92 0.62 Belluno, the province where the Cadore area is
Non-metalliferous products 0.40 1.71 1.21
Metal 0.91 1.28 1.04 located, has a dichotomized productive structure. On
Mechanical 1.17 1.81 1.53 the one hand, there are some non-specialized sectors
Electronic and optical 2.01 0.61 6.25 with location quotients\0.50 such as press, textile
Transportation 0.48 0.83 0.03 industry and transportation. On the other hand, there
Electricity, water and gas 0.82 1.85 0.80 is one sector with a surprisingly high location quotient,
Hotels and restaurant 0.75 1.06 1.80
Monetary and financial brokerage 1.49 0.94 0.51 i.e. the electrical and optic sector (6.25). This is due to
Real estate, information technology the Cadore area being an industrial district specializing
and research activity 1.66 0.80 0.44 in the production of glass, which belong to the category
‘electrical and optical sector’ according to the National
Source: The authors’ elaboration on Istat data, 1996.
Statistical Office (Istat). Other sectors with high loca-
tion quotients, although much lower than the electrical
represented by the average of the scores attributed by and optical sector, are the timber industry (1.86), and
managers to the contribution of local customers and hotels and recreation (1.80).
suppliers to innovative activity. A dummy variable was This brief economic description of Milan, Piacenza
introduced for high-technology sectors6 to test if firms and Cadore emphasizes how different is the productive
belonging to this sector had considerably different structure of these three areas. Milan is a highly diversi-
behaviour. Table 1 summarizes the variables and fied metropolitan area specialized in more advanced
method used to build them. sectors, Piacenza is a middle-sized town more special-
ized in traditional industries and Cadore is a typical
industrial district.
Areas of analysis
The sample includes firms belonging to three different
Description of the sample
geographical areas: Milan, Piacenza and Belluno
(Cadore). The three areas were chosen for both theo- As already noted, the database includes 217 observa-
retical and practical reasons. From a theoretical point tions:7 62 firms located in Milan, 65 in Piacenza and
of view, there are two reasons for the choice. On the 90 in the Cadore district in Belluno province.
one hand, the three areas are located in three successful The firms differ for internal characteristics (dimen-
Italian regions, which have had good economic perfor- sions, R&D expenditure) and for the sectors in which
mance in the last decade. This ensures that the three they belong. Table 3 summarizes some of the main
areas have a comparable macroeconomic background. characteristics of the samples in the three different areas.
On the other hand, and more crucially, the three areas The firms in Piacenza are on average larger than
have very different productive structures, which allows those in Milan and Cadore, but they seem less innova-
one to test if the hypotheses hold under different tive. The percentage on turnover due to sales of innova-
territorial and productive conditions. From a practical tive products is, indeed, only 20% in Piacenza, 25% in
point of view, the choice was favoured by the location Cadore and 32% in Milan. R&D expenditure on
of the three research groups, which could exploit their average is higher in Piacenza in absolute value (1212
specific knowledge of the territory to select better the million lire), but lower than Milan in terms of percent-
sample. age on turnover (6% in Milan compared with 2% in
Table 2 shows the location quotients of the different Piacenza). In all the considered areas, the local labour
sectors in the three analysed areas. Each area has its market plays a crucial role. In Milan and Piacenza,
own specialization. Milan reports a very high location around one-third of new employees come from local
quotient in high-technology sectors (2.01) and in the firms; in Cadore, the percentage is even higher, 48%.8
chemical industry (2.68). High location quotients are Milan, though, differs from the other two areas for the
also exhibited in some advanced tertiary sectors, such origin sector of new employees. Indeed, while in
as monetary and financial brokerage, real estate, Piacenza and Cadore a high percentage of new
information technology services and research activities. employees come from local firms belonging to the
The fields of specialization in Piacenza are com- same sector as the new firm (21% in Piacenza, 37% in
pletely different. More traditional sectors play an Cadore), in Milan, 20% of new employees come from
82 Roberta Capello and Alessandra Faggian
Table 3. Characteristics of the sample by geographical area (averages)
Milan Piacenza Cadore (Belluno)
Degree of innovation (per cent of positive answers) 32 (60)* 20 (65) 25 (27)
Turnover (millions lire) 9785 (62) 60 226 (65) 14 371 (68)
Research and development expenditure (millions lire) 639 (60) 1212 (65) 79 (65)
Importance of the local labour market (per cent of positive answers) 38 (62) 33 (65) 48 (90)
Importance of the diversification of the local labour market (per cent of positive answers) 20 (62) 12 (65) 11 (90)
Importance of the specialization of the local labour market (per cent of positive answers) 18 (62) 21 (65) 37 (90)
Importance of suppliers on innovation (scores from 1 to 10) 3.9 (62) 1.8 (65) 4.9 (54)

Note: Number of valid observations is reported in parentheses.

local firms belonging to different sectors of origin and
only 18% from firms belonging to the same sector.
As far as the importance of local suppliers on innova-
tion is concerned, the average scores are low. The
minimum is 1.8 in Piacenza, followed by Milan (3.9)
and Cadore (4.9). The problem is in understanding if
these low values reflect the actual unimportance of
suppliers or rather the misperception of their impor-
tance by the interviewed managers.
EMPIRICAL RESULTS
Role of relational capital in fostering innovation
As mentioned above,9 the productive structure in the
areas selected as case studies is very different. To test
whether relational capital should play a crucial role in
fostering innovative activities irrespective of location,
the internal characteristics of a firm and sector, the
starting point was the estimation of the same econo-
metric model for all firms included in the database
(pooling) without any distinction of location, size or
sector.
The first step was the estimation of the relationship
between ‘intra-firm’ characteristics and degree of inno-
vation. The second step was then to introduce, in the
same model, some relational capital variables to test
their significance. Table 4 shows the results obtained.
Both relational capital variables, i.e. the interfirm
mobility of local labour market10 and the cooperation
with local suppliers, have a positive and significant
effect. Moreover, R2 rises from 0.19 to 0.28. An F-test
has also been carried out to test the overall significance
of the models.
The result supports the initial hypothesis, i.e. the
importance of relational capital variables in fostering
the innovative performance of a firm. Regional eco-
nomists are therefore correct in underlining that not
only are intra-firm characteristics crucial for innova-
tion, but also (and maybe most of all) that the location
of firms in an area where the local labour market and
the tight links with suppliers foster the exchange of
local knowledge are vital for innovation. This seems to
hold no matter where firms are located, although, as
will be shown, with different features from one case to
the other.
The last two columns in Table 4 show two further
models in which a dummy variable for high-technology
firms was introduced to test if there was a substantial
difference between high-technology and traditional
firms. The dummy is statistically insignificant when the
model is tested on the total sample, but becomes
significant if the Cadore case is dropped and Milan and
Piacenza are only considered. This may be partly due
to the lack of variance in the sectorial composition of
the firms in the Cadore area, almost all belonging to
the same sector.

Table 4. Role of relational capital on innovative activity (pooling)

Model 5
(only Milan
Variable Model 1 Model 2 Model 3 Model 4 and Piacenza)
Constant 1.97 (2.54)* 2.02 (2.70) 1.49 (1.5) 0.91 (1.02) 0.93 (1.0)
Firm size (turnover in ln) ñ0.32 (ñ4.32) ñ0.30 (ñ4.13) ñ0.23 (ñ3.08) ñ0.21 (ñ2.59) ñ0.21 (ñ2.36)
Research and development expenditure as a
percentage of turnover (ln) 0.24 (4.08) 0.23 (3.99) 0.22 (3.84) 0.21 (3.66) 0.18 (2.53)
Local labour market (ln) 0.14 (3.27) 0.14 (3.4) 0.14 (3.31) 0.20 (4.5)
Cooperation with local suppliers and clients (ln) 0.21 (2.9) 0.20 (2.68) 0.22 (3.02)
High-technology sector 0.21 (0.87) 0.46 (1.79)
R2 0.19 0.24 0.28 0.29 0.40
Number of valid cases 148 148 148 146 122

Notes: Dependent variable: degree of innovation (per cent of turnover from sales of new products – last 5 years).
*Student’s t-test is in parentheses.
F-test is 9.47 when passing from model 1 to 2, and 7.95 when passing from model 2 to 3. Both values are significant at the 0.01 level.
 Collective Learning and Relational Capital in Local Innovation Processes
Relational capital and innovative activity in the three
different areas
Once it is shown that relational capital does play a
crucial role in innovation, it becomes interesting to
understand better the differences among the areas
analysed. Is it possible to identify specific territorial
characteristics that allow the relational capital to be
more effective? To answer this question, one can start
by estimating separate models for the different geo-
graphical areas. Unfortunately, in trying to do so, we
had to deal with a high number of missing values for
the Cadore area in both the dependent and explanatory
variables. This reduced considerably the degrees of
freedom and made the results for this area unreliable.
Therefore, it was decided to present the detailed analysis
only for the areas of Milan and Piacenza. The Cadore
area still contributes to the results obtained when
pooling all the data. Tables 5 and 6 show the results of
the estimations.
In the Milan area, the intra-firm characteristics, i.e.
small size and high R&D expenditure, are crucial in
fostering innovation. As far as the relational capital
variables are concerned, the most surprising result is
that cooperation with local suppliers no longer plays
an important role in Milan. The local labour market is
the most important collective learning channel with a
83
coefficient of 0.14 in both models 1 and 2 and a
Student’s t-test of 2.37 and 2.38, respectively (Table 5).
Furthermore, a more detailed analysis of the compo-
nents of local labour market shows something very
interesting. In Milan, innovation performance is posi-
tively linked to the diversification of the local labour
market rather than to its specialization. As models 3
and 4 in Table 5 show, the variable ‘diversified local
labour market’ has a coefficient of 0.12 and is highly
significant (2.36 in model 3, 2.33 in model 4), while
the variable ‘specialized labour market’ is insignificant
in model 3 and has been removed in model 4 to test
the robustness of the other coefficients. The economic
meaning of this seems clear. In Milan, there are urban-
ization economies that foster innovation. Also, the
dummy variable for the high-technology sector is statis-
tically significant in the Milan area. Because of the
high specialization encountered in these sectors in
Milan, one can easily draw the conclusion that these
sectors are organized around filières.
Piacenza has some similarities to Milan, but also
some peculiar characteristics. In the Piacenza area, as
in Milan, the local labour market is the most effective
collective learning channel with a coefficient of 0.25
(even higher than that of Milan), which is highly
significant (4.04). Unlike Milan, though, cooperation

Table 5. Role of relational capital on innovative activity: Milan

Variable Model 1 Model 2 Model 3 Model 4 Model 5
Constant 3.4 (2.07)* 3.26 (2.15) 3.67 (2.30) 3.51 (2.27) 3.93 (2.53)
Firm size (turnover in ln) ñ0.40 (ñ2.44) ñ0.39 (ñ2.50) ñ0.42 (ñ2.61) ñ0.41 (ñ2.60) ñ0.51 (ñ3.17)
Research and development expenditure as a
percentage of turnover (ln) 0.30 (3.04) 0.30 (3.09) 0.29 (2.89) 0.29 (2.94) 0.23 (2.279)
Local labour market (ln) 0.14 (2.37) 0.14 (2.38) – – –
Cooperation with local suppliers and clients (ln) ñ0.04 (–0.22) – – – –
Diversified local labour (ln) – – 0.12 (2.36) 0.12 (2.33) 0.12 (2.38)
Specialized local labour (ln) – – 0.02 (0.48) – –
High-technology sector – – – – 0.57 (2.06)
R2 0.24 0.24 0.24 0.24 0.29
Number of valid cases 59 59 59 59 57

Notes: Dependent variable: degree of innovation (per cent of turnover from sales of new products – last 5 years).
*Student’s t-test is in parentheses.

Table 6. Role of relational capital on innovative activity: Piacenza

Variable Model 1 Model 2 Model 3
Constant 0.51 (0.35) 1.64 (1.13) 1.65 (1.13)
Firm size (turnover in ln) ñ0.16 (ñ1.22) ñ0.16 (ñ1.24) ñ0.16 (ñ1.23)
Research and development expenditure as a percentage of turnover (ln) 0.15 (1.51) 0.21 (2.12) 0.21 (2.10)
Local labour market (ln) 0.25 (4.04) – –
Cooperation with local suppliers and clients (ln) 0.28 (3.05) 0.25 (2.40) 0.25 (2.37)
Diversified local labour (ln) – 0.18 (2.8) 0.18 (2.85)
Specialized local labour (ln) – 0.18 (3.02) 0.18 (3.00)
High-technology sector – – ñ0.08 (–0.11)
R2 0.37 0.42 0.42
Number of valid cases 65 63 63

Notes: Dependent variable: degree of innovation (per cent of turnover from sales of new products – last 5 years).
*Student’s t-test is in parentheses.
84 Roberta Capello and Alessandra Faggian
with local suppliers is significant with a coefficient of
0.28 and a Student’s t-test of 3.05 (Table 6, model 1).
Moreover, intra-firm characteristics do not play an
important role in Piacenza. Models 2 and 3 show
the results, once the local labour market has been
decomposed into ‘diversified’ and ‘specialized’. Surpris-
ingly enough, in Piacenza the specialized component
is significant. Both components, indeed, have the same
coefficient of 0.18, but the diversified one is statistically
more significant (around 3.0 in both models). It seems,
therefore, that in the Piacenza area, specialization11
rather than urbanization economies are at work.
Relational capital and innovation activity: a comparison
between firm size and sectoral specialization
The last aspect to be considered is how the size of a
firm and the degree of specialization of a sector can
affect the effectiveness of relational capital in fostering
innovation. The following model has been estimated
separately for Milan and Piacenza to analyse the
problem:
IóaòbDIMòcCR*DIMòdCR*QLòe
where I is the degree of innovation of a firm, DIM is
the size of a firm (turnover), CR is relational capital
(local labour market) and QL is the sectoral location
quotient.
As proxy for the relational capital (CR), the most
significant variable in the regressions presented above
is chosen, i.e. the diversified local labour market
in Milan and the specialized local labour market in
Piacenza. The results of the estimation of equation (1)
are presented in Table 7.
Differentiating equation (1) with respect to size
(DIM) and the location quotient (QL), it is possible
to evaluate the effect of relational capital on innovative
capacity of a firm for different dimensional levels of a
firm and different sectoral specialization. The calcula-
tion results in the following two equations for Piacenza
and Milan, respectively, are as follows:
(LI/LCR)Piacenzaó0.02DIM ò 0.12QL (2a)
(LI/LCR)Milanó0.01DIM – 0.27QL (2b)
The equations are shown in Fig. 1 (with a logarithmic
scale) for different levels of location quotients.
In the Piacenza area, the importance of the special-
ized local labour market on innovation becomes
increasingly more important as the location quotients
increase (Fig. 1a). It seems reasonable that firms
belonging to more specialized sectors can exploit the
specialized local labour market more efficiently to
increase their innovation. Moreover, as shown in Fig.
1a, the larger the firm, the better it can exploit the
local labour market.
In Milan, conversely, the capacity to exploit the
diversified local labour market decreases as the location
quotients increase. Firms belonging to less specialized
sectors can exploit better the ‘variety’ of the local
labour market. They localize in the Milan area to
(1) exploit a wider pool of potential employees with
differentiated characteristics and abilities.12 Moreover,
as in the previous case, the larger the firm, the better
it can exploit the relational capital. As shown in Fig. 1b,
location quotients being equal, larger firms exploit
better the diversified local labour market.
Similar results are presented in Fig. 2 (with a logarith-
mic scale), where equations (2a) and (2b) were calcu-
lated for different location quotients. The importance
of relational capital on innovation increases as size
increases in both Milan and Piacenza. The difference
is that, given a certain firm’s dimensions, the impor-
tance of relational capital increases with location
quotients in Piacenza (specialization economies) and
decreases in Milan (urbanization economies).

CONCLUSIONS AND POTENTIAL

Table 7. Interaction among relational capital, firm size and
FURTHER DEVELOPMENTS
location quotients
The aim of this paper was to present a theoretical and
Variable Piacenza Milan
empirical analysis of the role of relational capital on
Constant ñ1.38 1.19 innovation activity. Relational capital – in the sense of
(ñ0.84) (0.88) the capacity of interaction among local actors due to a
DIM (turnover in ln) ñ0.03 ñ0.16
(ñ0.20) (ñ1.3) strong sense of belonging and a close cultural prox-
CR (division of local labour ñ 0.01 imity – is, indeed, the basis for local collective learning.
market)*SIZE (2.22) From a theoretical point of view, the paper focused
CR (division local labour market)*LQ ñ ñ0.27 on the differences between the industrial economists
(1.8) and regional economists in defining the local externali-
CR (specialization of local labour 0.02 ñ
market)*SIZE (3.39) ties generated by innovation activity referred to as
CR (specialization of local labour 0.12 ñ ‘knowledge spillovers’. Regional economists have the
market)*LQ (1.5) benefit of having defined the channels through which
R2 0.22 0.14 knowledge spills over space and of having introduced
Number of valid cases 63 59 the concept of relational space – rather than the purely
Note: Dependent variable: degree of innovation. physical one – as the key to understand collective
LQ, location quotient. learning.
 Collective Learning and Relational Capital in Local Innovation Processes
Fig. 2. Role of sectoral specialization: (a) Piacenza and
(b) Milan
From an empirical point of view, the paper shows
the importance of relational capital on the innovative
capacity of firms. The presence of relational capital is
measured indirectly via the collective learning channels,
which are believed here to represent the contents of
the knowledge spillovers ‘black box’.
Identifying the mechanisms through which know-
ledge spills over the space is not only important from
a theoretical point of view, but also it is fundamental
when formulating useful and effective policies. As
A (1996, p. 139) underlines,
although knowledge spillovers between firms play a
critical role in the evolution technology little is known
about such spillovers. . . . By understanding the mecha-
nisms and determinants of knowledge flows, company
managers and public policy makers can influence know-
ledge diffusion more effectively.
Underlining the crucial role of the local labour
market as a channel for spreading knowledge, the
present paper shows how important it is for the govern-
ment to implement policies that both improve the
quality of local labour market and foster the mobility
of highly trained employees.
Lastly, policies to encourage cooperation with local
suppliers and customers should be implemented, since
both are important to the innovative capacity of firms.
Although the paper represents a first attempt in
85
Fig. 3. Role of firm size: (a) Piacenza and (b) Milan
opening the spillovers ‘black box’ and testing empiri-
cally some very abstract concepts, further developments
can be added in the future. First, additional indicators
looking at different aspects of relational capital can be
devised and used to test the consistency of the results.
Further, a second round of interviews with local cor-
porate decision-makers and industry-watchers could
provide a deeper understanding of some more general
local socio-cultural factors that are strictly linked with
the concept of relational capital.
NOTES
1. On the concept of ‘social capital’, see P (1993),
G and V B (2001) and S
(2002).
2. In this respect, we refer to the contrasting empirical
results traditionally obtained by industrial economists on
the role of the physical dimension of firms on innovative
activity (e.g. A and A, 1993; L
and S, 1991; S-L, 1986;
P et al., 1987; R, 1989).
3. When dealing with knowledge spillovers, we refer to
J (1989) as a seminal work, followed by, among
others: A et al. (1994), dealing with the capacity of
large versus small firms to exploit knowledge spillovers;
A and F (1996) and F and
A (1999), dealing with the importance of
diversified versus specialized knowledge spillovers; and
A et al. (2000), dealing with the definition of
86 Roberta Capello and Alessandra Faggian
the physical distance over which knowledge spillovers
disappear. For a recent review on the role of knowledge
spillovers on regional development, see D G et al.
(2001).
4. On the relationship between relational capital and
innovation, see C (1991), K and W-
 (1999, 2000), L and L (1999),
and C and C (2002). For the concept
of organizational and cultural proximity, see also the
French school on ‘la proximité’ (e.g. R, 1993).
5. Both belonging to the same or to a different sector.
6. High-technology sectors include: electronics, informa-
tion technology, telecommunications, and electric and
optical machinery
7. The trade-off between the collection of primary data
and sample size is well known to every applied researcher,
but it was fundamental for our research purpose not to
rely on simple secondary data, but to collect data directly.
Given this choice, the number of observations in our
sample seems reasonable. Although we are aware that a
much larger sample would obviously be preferable, we
still believe that the results of our empirical analysis are
sound and shed light on the main relationships of interest.
8. This is not surprising since Cadore is an industrial
district.
9. See the end of the section entitled ‘Areas of analysis’.
10. This result is in line with A and K (1999),
who found that interfirm mobility of engineers in the
semiconductor industry in Silicon Valley indeed influ-
enced the local transfer of knowledge among patent
holders.
11. Although the terms ‘specialization economies’ and
‘urbanization economies’ are well known in urban eco-
nomics, note that for specialization economies, we mean
economies external to the firm but internal to the sector,
while urbanization economies are external to both the
firm and the sector. The latter arise from the fact that
they are located in a big city.
12. This is a typical advantage offered by large cities and
is part of what regional economists call ‘urbanization
economies’. For an empirical analysis on urban milieu
effects, see C (2001).

REFERENCES
A Z. and A D. B. (1990) Innovation and Small Firms. MIT Press, Cambridge, MA.
A Z. and A D. (Eds) (1993) Small Firms and Entrepreneurship: An East–West Perspective. Cambridge University
Press, Cambridge.
A Z., A D. and F M. (1994) R&D spillovers and recipient firm size, Review of Economics and Statistics
76, 336–340.
A Z., D  M J. and P G. (2000) Regional innovation: in search of an enabling strategy, in A Z. (Ed.)
Regional Innovation, Knowledge and Global Change, pp. 37–52. Pinter, London.
A P. and K B. (1999) Localization of knowledge and the mobility of engineers in regional networks, Management
Science 45, 905–917.
A L., V A. and A Z. (1997) Local geographic spillovers between university research and high technology
innovations, Journal of Urban Economics 42, 422–448.
A L., V A. and A Z. (2000) Geographic and sectoral characteristics of academic knowledge externalities, Papers
in Regional Science 79, 435–443.
A M. M. (1996) How does knowledge flow? Interfirm patterns in the semiconductor industry, Strategic Management
Journal 17, 137–154.
A D. and F M. (1996) R&D spillovers and the geography of innovation and production, American Economic
Review 86, 630–640.
A D. and V M. (1994) Small firms and spillovers: evidence from Italy, Revue d’Economie Industrielle no.
67, 225–235.
A D. B. and V M. (1996) Firm size and R&D spillovers: evidence from Italy, Small Business Economics 8,
249–258.
A-B C. (1999) Geographic knowledge spillovers and technological proximity. Paper presented at the Inter-
national Conference ‘Knowledge Spillovers and the Geography of Innovation: A Comparison of National Systems of
Innovation’, 1–2 July, University Jean Monet Saint-Etienne, France.
A Ph. (Ed.) (1986) Milieux Innovateurs en Europe. Groupe de Recherche Européen sur les Milieux Innovateurs, Paris.
A Ph. and K D. (Eds) (1988) High Technology Industry and Innovative Environment. Routledge, London.
C R. (1991) Local milieu, uncertainty and innovation networks: towards a new dynamic theory of economic space,
in C R. (Ed.) Innovation Networks: Spatial Perspectives, pp. 121–144. Belhaven–Pinter, London.
C R. (1995) Global network and local milieu: towards a theory of economic space, in C S., M E. and
O P. (Eds) The Industrial Enterprise and Its Environment: Spatial Perspectives, pp. 195–214. Avebury, Aldershot.
C R. (1999) The city as a milieu: applying the Gremi approach to urban evolution, Révue d’Economie Régionale et
Urbaine no. 3, 591–606..
C R. and C R. (Eds) (2002), Apprendimento collettivo e competitività territoriale. FrancoAngeli, Milan.
C R. (1988) La diffusione spaziale dell’innovazione: il caso del servizio telefonico, Economia e Politica Industriale no. 58,
141–175.
C R. (1999a) Spatial transfer of knowledge in high-technology milieux: learning vs. collective learning processes,
Regional Studies 33, 353–365.
C R. (1999b) A measurement of collective learning effects in Italian high-tech milieux, Révue d’Economie Régionale et
Urbaine no. 3, 449–468..
 Collective Learning and Relational Capital in Local Innovation Processes 87
C R. (2001) Urban innovation and collective learning: theory and evidence from five metropolitan cities in Europe,
in F M. M. and F J. (Eds) Knowledge, Complexity and Innovation Systems, pp. 181–208. Springer, Berlin.
C R. (2002) Apprendimento collettivo e teorie della crescita endogena: una prospettiva territoriale, in C R.
and C R. (Eds) Apprendimento collettivo e competitività territoriale, pp. 57–84. Franco Angeli, Milan.
C R. (2004) Economia Regionale. IP Mulino, Bologna.
C W. S. and B A. (1996) Economic restructuring – a conceptual framework, in L W. and B A. (Eds)
The Spatial Impact of Economic Changes in Europe. Aldershot, Avebury.
C O. and C R. (Eds) (2000) Les milieux urbains: innovation, systèmes de production et ancrage. EDES, Neuchâtel.
D G H., N P. and A Z. (Eds) (2001) Papers of the Regional Science Association [Special Issue: Knowledge
Spillovers, Innovation and Regional Development] 80(3).
F M. (1994) The Geography of Innovation. Kluwer, Boston.
F M. and A D. (1999) Innovation in cities: science-based diversity, specialisation and localised competition,
European Economic Review 43, 409–429.
G E. (1998) The genesis of high technology milieu: a study in complexity, International Journal of Urban and Regional
Research 22, 361–377.
G P. (1995) Markets for technology: knowledge, innovation and appropriability, in S P. (Ed.) Handbook of the
Economics of Innovation and Technological Change. Blackwell, Oxford.
G Z. (1957) Hybrid corn: an exploration in the economics of technological change, Econometrica 25, 501–525.
G C. and V B T. (2001) Understanding and Measuring Social Capital: A Synthesis of Findings and
Recommendations from the Social Capital Initiative. Social Capital Initiative Working Paper No. 24. World Bank, Washington, DC.
J A. B. (1989) Real effects of academic research, American Economic Review 79, 957–970.
K D. and W F. (1999) Collective learning and knowledge development in the evolution of regional clusters
of high-technology SMEs in Europe, Regional Studies 33, 295–303.
K D. and W F. (Eds) (2000), High-Technology Clusters, Networking and Collective Learning in Europe. Ashgate,
Aldershot.
L C. and L E. (1999) Collective learning, tacit knowledge and regional innovation capacity, Regional Studies 33,
305–317.
L G. and S W. (1991) The re-emergence of small-scale production: an international perspective, Small
Business Economics no. 3, 1–38..
L R. E. (1988) On the mechanics of economic development, Journal of Monetary Economics no. 22, 3–42.
M D. and L B. (1992) New technologies and transformation of regional structures in Europe: the role of the
milieu, Entrepreneurship and Regional Development 4, 1–20.
M D., Q́ M. and S L. (1993) Réseaux d’Innovation et Milieux Innovateurs: un Pari pour le Développement
Régional. EDES, Neuchâtel.
M E. (1961) Technological change and the rate of imitation, Econometrica 29, 741–766.
P K., R M. and T J. (1987) The size distribution of innovating firms in the UK: 1945–1983, Journal
of Industrial Economics no. 55, 291–316.
P I. and S I. (1983) Order Out of Chaos. Bantam, New York.
P R. D. (1993) Making Democracy Work. Princeton University Press, Princeton.
R A. (1993) Choix de proximité et processus d’innovation technologique, Revue d’Economie Régionale et Urbaine no. 3,
365–386..
R R., B A. and G R. (Eds) (1997) The Dynamics of Innovative Regions. Ashgate, Aldershot.
RERU (1999) Le paradigme de Milieu Innovateur dans l’économie contemporaine, Revue d’Economie Régionale et Urbaine no. 3.
R P. (1986) Increasing returns and long run growth, Journal of Political Economy 94, 1002–1037.
R R. (1989) Small firms, innovation and industrial change, Small Business Economics no. 1, 51–64..
S F. M. (1982) Demand–pull and technological invention: Schmookler revisited, Journal of Industrial Economics 30, 225–237.
S S. (2002) Fiducia e capitale sociale nell’analisi economica della regolazione sociale, in C R., F R.
and M M. (Eds) Scritti in memoria di Eugenio Benedetti, pp. 39–68. CEDAM, Padua.
S-L P. (1986) Le classi sociali negli anni ottanta. Laterza, Rome.