An evolutionary integrated view of regional systems of innovation.

Concepts, measures and historical perspectives

Simona Iammarino*

ABSTRACT

The literature on geographical systems of innovation has traditionally shown a ‘national-bias’ that
has strongly affected the identification of actors, relationships and attributes operating at the sub-
national scale. Indeed, the historical evolution of the regional dimension has rarely been
considered (implying that history really matters only at the national level). Modes of governance
have also mostly been examined from a country perspective, which neglects the complexity,
heterogeneity and path dependency of multi-level governance in current innovation systems. This
paper reviews the main literature on the concept of Regional Systems of Innovation (RSI),
adopting an integrated view that brings together both top-down and bottom-up characteristics and
evolutionary mechanisms for the purpose of identifying RSIs. After discussing conceptual
problems, and the relevance and applicability of an evolutionary integrated view of RSI, the case
of Italy is employed to support the argument that the historical perspective on regional cultures is
often unavoidable in order to assess future opportunities for regional development.

JEL Classification: O3, R1, O1.

Keywords: Technological Change, Regional Systems of Innovation, Regional Development.

*
SPRU, University of Sussex (UK), and University of Rome “La Sapienza” (Italy). Address for correspondence:
SPRU, The Freeman Centre, University of Sussex, Brighton, BN1 9QE, United Kingdom. Tel: +44 (0)1273 877565.
Fax: +44 (0)1273 685865. E-mail: s.iammarino@sussex.ac.uk.
The author is grateful for helpful and valuable comments on earlier versions of this paper from Suma Athreye, Ron
Boschma, Nick von Tunzelmann, and one anonymous referee. I would like also to thank the participants in the
conference on “Regionalisation of Innovation Policy”, DIW, Berlin, 5-6 June 2004, for the lively and stimulating
debate. The usual disclaimer applies.

1
 An evolutionary integrated view of regional systems of innovation.
Concepts, measures and historical perspectives
Simona Iammarino

1. Introduction

Since the appearance of the first literature on National Systems of Innovation (NSIs) the idea of
applying tout court a similar conceptual perspective at a smaller geographical level – regional, or
even local – has been very tempting (Breschi, 1995; Cooke et al., 1997; Howells, 1999). However,
whilst a NSI cannot by any means be considered as being the simple sum of regional systems,
attempts to find some kind of standardisation of sub-national systems of innovation have so far
encountered two major problems.

The first one, theoretical in nature, arises from the fact that geographical innovation systems have
been conceptualised and applied by considering components, relationships and attributes, which
operate and are governed mainly at the national level. Indeed, the systems of innovation (SI)
approach assumed, at least in its original formulation, that key decision-making processes
regarding the aggregate of micro-founded innovation activities are taken at a macro (national)
level and, equally, that much micro-level activity is linked through a macro (national) web of
interconnections. Hence, in the SI literature the modes of governance – i.e. markets, corporate
hierarchies, political hierarchies and networks – have been regarded mostly from a country
perspective, which ignores the complexity, heterogeneity and path dependency of multi-level
governance in current SI settings (von Tunzelmann, 2004). This ‘national-bias’ has strongly
affected identification of the relevant actors, relationships and attributes operating at the meso
(sub-national) scale and very rarely has the regional dimension effectively been considered in
terms of its historical evolution, implying that history really matters only at the national level. This
bias has also contributed to creating some confusion in the policy debate, arguably leading to a
virtual distinction between ‘stylised’ and ‘actual’ Regional Systems of Innovations (RSIs) and to
an excessive criticism of the conceptual framework, thereby undermining the search for better
methods to test its validity.

The second and related problem is the well recognised and broad ‘regional measurement issue’.
The availability of data and indicators on the characteristics and behaviours not only of firms, but
also of the multitude of actors and sets of institutions whose interactions determine the innovative

2
performance of the region, is much better at national than at sub-national level. This has further
constrained the possibility of exploring the nature and evolution of RSI based on statistically
robust evidence (see among others, Braczyk et al., 1998; Evangelista et al., 2002).

This paper takes the view that a top-down (macro-to-micro) conceptual perspective, i.e. the ‘shift’
of NSI features down to the regional scale, although it provides the necessary conditions to
distinguish RSIs is not by itself sufficient. It is necessary to integrate the top-down view with a
bottom-up (micro-to-meso) perspective – tackling also the internal dynamics of regionally
embedded social, economic and institutional structures, whose analysis in terms of historical
origins and evolution of regional cultures, may further help to refocus the nationally-biased
‘Listian’ view of innovation systems (Cooke, 2001; Dopfer et al., 2004). Moreover, whilst clear
thinking on RSIs is obviously a prerequisite for making sensible policy implications, adequate data
and methodologies are crucial to determine whether the conceptual approach is suited to
identifying structural regularities of regional systems and providing sound normative justifications
for public intervention.

The paper is organised in five sections. The next section reviews the main literature on the concept
of RSIs, adopting an integrated view that brings together top-down and bottom-up characteristics,
and evolutionary mechanisms for the purpose of identification of RSIs. Section 3 discusses the
relevance and applicability of the evolutionary integrated concept of RSIs by addressing such
questions as: to what extent is the system portrayed in such a framework applicable and replicable?
Do currently available data and indicators reproduce both structure and performance of RSIs?
Does the integration of concepts and measures actually provide a normative base for regional
policy? Section 4 attempts a first application of the evolutionary dynamic view of RSIs to the case
of Italy, which offers strong support for the argument that use of the historical perspective as a
filter is often unavoidable in assessing future opportunities for regional development. Section 5
presents some concluding remarks, and draws out the main implications and directions for future
research.

2. Top-down, bottom-up and integrated views of RSIs

The extensive literature on the advantages of geographical agglomeration with particular reference
to innovation and technology activities, has generally encompasses two distinct perspectives on
the relationship between innovation and space. The first approach, and antecedent to the second,
follows the Marshallian tradition in trying to identify these advantages and their implications for

3
overall economic growth (see, for instance, Perroux, 1950; Pred, 1967, 1977; Malecki, 1977,
1983). The pioneering works and intuitions of Marshall suggested that the accumulation of skills
and know-how takes place within spatially bounded contexts, which create a kind of favourable
‘industrial atmosphere’ capable of enhancing economic growth and spurring the generation and
diffusion of new ideas. The empirical literature that flourished over the last decades confirmed that
spatially-specific factors strongly influence firms’ innovative performance and regional patterns of
technological specialisation. However, from such a perspective, the geographical dimension
represents a factor characterising economic development, in relation to which the localised
innovative potential is assumed to be an exogenously determined explanatory variable. The second
and more recent line of research has addressed the localised structural and institutional factors that
shape the innovation capacity of specific geographical contexts. This has given rise to
heterogeneous sub-national typologies, all pointing to a broadly defined model of spatial
organisation, that is the ‘innovative cluster’. Amongst the most popular units of analysis we find
milieux innovateurs (Aydalot, 1986), new industrial districts (Becattini, 1987), technological
districts (Markusen, 1985, 1996; Storper, 1992), learning regions (Asheim, 1995; Morgan, 1997)
and regional systems of innovation (Cooke, 1992; Cooke et al., 1997; Howells, 1996, 1999;
Braczyk et al., 1998).1

Focusing on this latter sub-national model, the importance of contextual elements and the presence
of systemic interactions in the process of generation and diffusion of innovation are recognised as
a key determinant of regional technological and economic performance. This is in line with the
prolific literature on NSI – introduced by evolutionary theorists in the late 1980s (Freeman, 1987;
Lundvall, 1992; Nelson, 1993; Nelson and Rosenberg, 1993; Edquist, 1997) – which argues that
the performance of national economies cannot be explained only in terms of the strategies and
performance of firms. There are other factors and actors that play vital roles in favouring the
generation and diffusion of knowledge, including: inter-organisation networks, financial and legal
institutions, technical agencies and research infrastructures, education and training systems,
governance structures, innovation policies, etc.

The notion of RSI has emerged as a territorially-focused perspective of analysis derived from the
broader concept of NSI: a RSI may thus be defined as ‘the localised network of actors and
institutions in the public and private sectors whose activities and interactions generate, import,
modify and diffuse new technologies within and outside the region’ (Howells, 1999; Evangelista
et al., 2002). Indeed, the highly uneven pattern and spread of innovation in space suggests that it

1
For the ‘fuzziness’ of the cluster concept itself, see Gordon and McCann (2000). For a critical review of territorial

4
could be better depicted by assuming sub-national units of analysis, which can avoid the
distortions and the loss of information of hypothesising national systems as homogeneous entities
(Morgan, 2004). As Carlsson and Stankiewicz aptly remarked “high technological density and
diversity are properties of regions rather than countries” (Carlsson and Stankiewicz, 1991: 115).

It is therefore implicitly maintained that the elements characterising a national system can be
transposed to a smaller territorial scale to help define the RSI. Following Dopfer et al. (2004),
micro-structures refer to individual components (or actors, i.e. individuals, firms and other
organisations) that through their relationships (market and non-market) and their attributes
(competencies and functions) build up and maintain systems of rules (see also Carlsson, 2003).
Macro-structures, on the other hand, are comprised of the population of meso-systems and their
interdependence. The top-down (macro-to-micro) view of a RSI thus implies identification of the
following characteristics (Howells, 1999):

internal organisations of firms, the latter being the principal agents of innovation;
inter-organisation relationships and, more specifically, the type and intensity of interactions
between the business sector and the rest of the economic system;
role of public sector and innovation policy (assuming that, at the local level, formal
policies interact to a much greater extent with informal rules and conventions);
institutional framework (administrative, political, legal, fiscal, educational, etc.);
institutional set-up of the financial sector (i.e. whether based on a developed capital
market, on regulated and strictly controlled credit, or on relatively ‘free’ access to funds,
etc.);
industrial structure (i.e. average firm size, degree of competition and collaboration among
firms, sectoral specialisation models, demand patterns, etc.) and intensity and organisation
of R&D activities (both private and public);
spatial structure (e.g. relative geographical position, degree of urbanisation, extent of
regional network externalities) and intra-regional scale and scope of geographical
agglomerations (i.e. sub-regional clusters with specific advantages in local labour markets,
specialised suppliers and knowledge spillovers);
degree of openness, capacity to attract/absorb external resources, integration in global
innovation networks;
core/periphery hierarchical forces driven by the historical evolution of regional societies.

innovation models, see Moulaert and Sekia (2003).

5
The above list is clearly not exhaustive and underlies a broad definition of territorial systems of
innovation, encompassing the specific institutions that deliberately promote innovation and
knowledge, as well as the wider socio-economic system in which political and cultural influences
and specific modes of governance determine innovation structures and performances (Lundvall,
1992; Cooke et al., 2000; Freeman, 2002). However, as discussed in section 3 below, the actual
identification of components, relationships and attributes, as well as that of the ‘purposive nature’
of geography, required to be a relevant context for a system of innovation, is more problematic in
regional systems than in other cases (Carlsson, 2003). Nonetheless, the top-down approach is
helpful for a comparative view of RSIs, both within and across national boundaries, in assessing
the nature of the innovation system in terms of space and networks of forces, which escape the
broad and generic national dimension insofar as they “can never be made precise by their outline
or by their container” (Perroux, 1950: 102).

Yet, within the systemic approach to the geographical dimension of innovation, the distinction
between aspects related to the agglomeration of innovative activity (i.e. Marshallian-type
agglomeration forces) and the contextual mechanisms for the generation and diffusion of
innovation (i.e. the impact of space on innovation through informal and formal social networks
Pred, 1967; Hagerstrand, 1967; Lundvall, 1988, 1996) has become increasingly blurred. The first
implication of the merging of these two aspects is that the top-down perspective may account for
the necessary but not the sufficient conditions needed to identify RSIs. Thus, integration of the
top-down view with a bottom-up (micro-to-meso) perspective, which also takes account of the
internal and dynamic regularities of territorially embedded socio-economic structures, has gained
momentum (Asheim, 1995; Asheim and Gertler, 2003; Dopfer et. al., 2004).

As highlighted by the evolutionary theories of technological change, the dynamism of an economic

system, which necessarily builds on access to and efficient use of knowledge, rests upon three
main functional dimensions:

1. absorption of new knowledge, technology and innovation for adaptation to local needs;

2. diffusion of innovations throughout all the constituent parts of the regional social fabric to
strengthen the existing knowledge base;

3. generation of new knowledge, technology and innovation.

All three functional dimensions involve micro and macro changes for adapting and maintaining the
system of rules for many distinct environments, so contributing to processes at the meso level,
which in turn account for local and regional variety (Dopfer et al., 2004). The dynamism of a

6
regional system may therefore be sustained through several channels related to these different
dimensions and based on collective learning processes through which knowledge and technology
are used, diffused and created. Such learning processes are central to growth and competitiveness,
and have convincingly provided a strong argument in favour of meso perspectives of analysis
(Cooke and Memedovic, 2003; Morgan, 2004). Learning dynamics and the employment of
informal channels for the exchange of tacit and ‘sticky’ codified knowledge are in fact embedded
in distinct environments of interactions among various actors and organisations, the sharing of
common attitudes, habits and conventions, and institutional settings facilitating idiosyncratic (and
often context-specific) types of learning (see among others, Hägerstrand, 1967; Lundvall, 1988,
1992; Audretsch and Feldman, 1996; Cooke, 2001; Asheim and Isaksen, 1997, 2002).

According to technological gap theories (Abramovitz, 1986; Fagerberg, 1987, 1994; Fagerberg et
al., 1994), ‘social capability’ and ‘technological congruence’ concepts are particularly relevant
when considering the meso level insofar as both appear to be highly variable across space, even
within the same national economy. The first concept refers to the overall ability of the region to
engage in innovative and organisational processes and to make institutional change; the second
relates to the distance of the region from the technological frontier, or, in other words, its capacity
to implement the technical properties embedded in new technologies (Fagerberg et al., 1994).
Following this approach, therefore, regional systems with stronger social capabilities and a
stronger knowledge base will tend to also be better equipped to exploit new technological
opportunities, to adapt existing activities to emerging business environments, and to learn faster
about how to build new regional advantages. Even the most specialised forms of knowledge are
nowadays becoming a perishable resource due to the accelerating pace of technological change;
competencies and capabilities have to be accumulated quickly; continuous learning and adaptation
determine the innovative performance of individuals, firms and systems; economic growth
involves complex and simultaneous change in both micro and macro structures, whose relations
become apparent particularly when the meso domain is taken into account (Steinmueller, 2000;
van der Meer et al., 2003; Dopfer et al., 2004).2

Yet, on the one hand, being able to build new competencies quickly involves the ability to
establish links at all levels, from the ‘global’ to the ‘local’: the extent to which a region attracts
innovative resources from outside – i.e. spurring its external integration – depends first and

2
The remarkable regional divergence in growth rates observed within Europe in the last decades is largely attributable
to the presence or absence of social capabilities for institutional change, especially those that stimulate high rates of
technological change (see among others, Rodriguez-Pose, 1994, 1998; Fagerberg and Verspagen, 1996; Freeman,
2002).

7
foremost upon its extant absorptive capacity (Asheim and Isaksen, 2002; Cantwell and Iammarino,
2003; Simmie, 2003). As learning curve advantages are mainly people- and institution-embodied,
human capital and skill upgrading raises the ability of a region to absorb, diffuse and generate new
knowledge. On the other hand, technology diffusion is proportional to the ability of a region to
absorb innovation: as large differences in terms of absorptive capacity spur geographical
agglomeration, knowledge will flow more easily and socio-economic development in general will
be more evenly spread if high absorptive capacity is even across space. Absorptive capacity
depends significantly on diversity: innovation occurs where there is a diverse (technological,
social, economic) culture, and the most dynamic capabilities lie in the combination of both
exploration and exploitation of new and existing assets (Rantisi, 2002).

Regional systemic advantages (or disadvantages) are therefore also supposed to depend on
attributes such as ‘untraded interdependencies’, informal flows of knowledge, interactive learning,
degree of embeddedness,3 which generate the bulk of territorial externalities at the meso level
(Saxenian, 1994; Storper, 1998). Moreover, dense social networks may prove to be critical
channels for knowledge diffusion and learning, recombining old and new pieces of knowledge. It
has been recently argued that geographical proximity would not produce knowledge spillovers per
se, but social proximity in the region would (Breschi and Lissoni, 2001; Faggian and McCann,
2004).

The above discussion may help to clarify what is meant by top-down versus bottom-up views. The
sum of micro units does not give a macro aggregate, in the same way that the decomposition of the
macro does not translate into many micro, due to the presence of meso units. As already argued,
socio-economic growth implies complex change in both micro and macro structures: however,
evolutionary mechanisms and structural regularities are mainly captured in terms of the
relationship between such structures, that is at the meso level (Dopfer et al., 2004). This is not to
deny the importance attributed by evolutionary economic thinking of the meso domain, but rather
to contribute an improved analytical framework – that of geographical systems of innovation –
which seems to be lagging with respect to the strong theoretical foundations and empirical testing
of other meso perspectives (such as those, for example, of sectoral and technological SIs). As
pointed out by Dopfer et al. (2004: 6): “The essential point to grasp here is that macro is not a
behavioural aggregation of micro, but, rather, it offers a systems perspective on meso viewed as a

3
Following Cooke (2001), embeddedness refers to a set of characteristics appropriate for systemic innovation and
reflecting the extent to which a social community operates in terms of shared norms of cooperation, trustful interaction
and untraded interdependencies (see also Dosi, 1988; Fritsch, 2001).

8
whole. Similarly, micro is not the reduced essence of an economic system; it is a ‘bottom up’
systems perspective on meso when viewed in terms of its component parts.”

Adapting from Howells (1996, 1999), among the idiosyncratic characteristics of RSIs which
emerge from a bottom-up view are:

• localised communication patterns relating to innovation processes, both at individual

and corporate levels;
• localised invention and learning patterns (individual, organisational, institutional and
societal);
• localised knowledge sharing (inter-individual, intra- and inter-organisation);
• localised search and scanning procedures relating to innovation and technology;
• localised network integration (within and between networks, intra- and extra-region)
and consequent degree of alignment of governance modes;
• historical path dependency of localised innovation processes.

Again, this list may not be complete, but it helps to illustrate the degree of regional embeddedness
of the knowledge and technology generation and diffusion processes, as well as the type of
learning attitudes and the potential for capability building of a particular RSI. These are crucial
elements insofar as regions with similar responses to the top-down criteria may show different
abilities to accommodate innovation and adjust in ways that enhance innovative capabilities. Also,
if the top-down view has the merit that it emphasises the role of networks as forms of governance
between (and beyond) the market and the hierarchy (corporate or political), the bottom-up
approach brings to the fore the complexity and idiosyncratic nature of linking the two within and
between localised networks developed for different purposes, thereby determining the overall
direction of regional governance (von Tunzelmann, 2004).4

An integrated micro–meso–macro view can be a better lens for identifying differences among
RSIs. The more science-based, codified and hierarchical the mechanisms by which interdependent
pieces of knowledge are integrated and recombined, the better the overall process governing
technological change and the character of the regional knowledge base can be grasped by a top-
down view. Conversely, the more important the idiosyncratic knowledge and/or the more informal
the mechanisms for knowledge absorption, integration and diffusion, the more suitable is a
bottom-up outlook for picking up the process managing technological change and the nature of the
regional knowledge base (Patrucco, 2004).

4
Following von Tunzelmann (2003), we adopt here the broad definition of governance as ‘organising collective
action’, first coined by Prakash and Hart (1999).

9
The RSI integrated approach emphasises evolutionary mechanisms such as routines, technological
trajectories, selection environments, heterogeneity and path dependency. As argued by Boschma
and Lambooy (1999) and Lambooy and Boschma (2001), variety in characteristics, behaviours and
performance is a key assumption of evolutionary perspectives, formed by its surrounding selection
environment, which includes both market and non-market factors. Regional environments are
heterogeneous (due to chance) and path dependent (as a result of historical contingency):5 they act
as selection mechanisms that may, or may not, provide conditions favourable to meeting the new
requirements of technical change (i.e. social capabilities for institutional change). Hence, new
growth opportunities are shaped and constrained by meso path dependency, or, in other words, by
the inheritance of local structural regularities from past knowledge accumulation and learning.6 In
such a dynamic perspective, the interdependence between structures and actors should be regarded
as a ‘feedback’ mechanism: not only do the characteristics of the selection environment and their
changes influence the actors, but the latter also change the environment (Lambooy and Boschma,
2001).

When integrating and applying the two lists of identifying features of RSIs, however, related
questions emerge such as: to what extent is the RSI portrayed by the evolutionary integrated view
applicable? Are its actual manifestations replicable? Do available data and indicators reproduce
both RSI structure and performance? Does the combination of concepts and measures provide a
sound normative base for regional policy? The next section discusses the relevance and
applicability of the evolutionary integrated view of RSI by addressing some of these issues.

3. Applicability and replicability of the evolutionary integrated view of RSIs

3.1 ‘Stylised’ and ‘actual’ RSIs

The framework derived from the integration of top-down and bottom-up approaches undoubtedly
helps in the definition of a ‘stylised’ RSI. The identifying features, in practice, combine in
different ways, have different intensity, quantity and quality and depend on the historical paths of
regional cultures made up by socially accepted values, norms, routines and general customary
practices. It is intuitive that the more diverse is the cultural base of a national system, the more

5
Historical contingency refers to the actual existence of selection mechanisms in socio-economic processes: change is
neither solely based on structural elements subject to general rules, nor purely driven by random effects. At each point
in time in a system’s evolution, a number of paths is theoretically possible, but only a few will be chosen by the actors
because each path must conform to the logic of socio-economic dynamics (Schwerin and Werker, 2003).
6
Indeed, as emphasised by Dopfer et al. (2004), path dependency is mainly a micro-meso concept.

10
meaningful the meso analysis appears to be and the less the assumptions of a Listian view of
innovation systems hold.

Indeed, the presence of strong geographical asymmetries in socio-economic and innovative

variables raises the question of to what extent the top-down and bottom-up criteria can be applied
(even allowing for strong and weak RSIs) and the extent to which actual (and in particular
successful) RSIs are replicable. The answer points to some limitations: first of all because any
‘system’ to be distinguished as such, is required to have internal coherence, a collective identity
and conform to the ‘rules of the game’ which result in a whole or a model, trickier to grasp at the
meso level of observation (Cooke et al., 1997). However, paraphrasising Cooke (2001), the
hypothetical world is only a special case. The integration of the two views7 is arguably sufficiently
flexible and consistent to address specific and diverse empirical cases. Its somewhat restricted
applicability and replicability do not seem to challenge its positive scope, that is, to provide
guidelines for empirical analysis, which in principle are oriented to testing the goodness of the
theory in describing the world as it is, and not as it might hypothetically be. The frequent ‘non-
existence’ of RSIs in the real world is not grounds for rejecting the concept as such (this being a
misconception of the null hypothesis); it rather gives an indication of the fact that not all regions
work as (innovation) systems and that replicating RSIs is a difficult and hazardous venture.
Although in principle this could also apply to NSIs, it is apparent that the regional scale poses
additional difficulties insofar as: “there are a few fully functioning RSIs and even fewer where the
economic performance of such regions is outstanding, at least in Europe” (Cooke, 2001: 958).

On the other hand, if a normative interpretation is to be given to the integrated evolutionary

framework, then the question becomes one of what an RSI ought to be, and drawing implications
(particularly those for public policy) from stylised constructs should be done with caution because
the risk of providing inadequate answers to complex and diverse problems is considerable. Again
following Howells (1999), the lack of agglomeration and localisation advantages for innovative
activities in some regions may be attributable to a combination of top-down and bottom-up
failures, such as for example:

o low density of interactions, both at individual and organisation level;

o peripherality with respect to political and economic centres of power and decision sittings;
o presence of social and cultural marginalisation;
o weak social networks and capital;

11
 o lack of dynamic innovative firms and institutions;
o weak access to and weak attraction for external knowledge and information flows;
o inflexibility of organisational and institutional structures, which hampers the capacity to
adequately monitor, evaluate, absorb and diffuse innovation produced elsewhere.

The above circumstances are indicators of either weakness or lack of social capabilities, that is, as
previously mentioned, the capacity of the region as a ‘social system’ to make institutional change
for growth. The integration of top-down and bottom-up characteristics defining RSIs renders the
inclusion of weaker regions in a systemic vision highly questionable (for a good example see
Vilanova and Leydesdorff, 2001). It is not only the lack of a critical mass to generate, import,
modify and diffuse new technologies, but also the dearth of internal coherence within the societal
structure that makes up the regional governance that prevent the attribution of the distinctive status
of (innovation) ‘system’ to the region.

Following von Tunzelmann (2003, 2004), each of the conventional modes of governance may be
associated with a certain type of failure. Along with the typical institutional (market, corporate,
state) failures, pervasive network (social) failures may emerge when integration – both within and
between localised networks – falls short. Hence, whilst network governance ‘alignment’ arises
when top-down and bottom-up elements are pulling in similar directions even when their aims are
different, network governance ‘misalignment’ occurs when these same elements are pulling in
contrasting (and sometimes contradictory) directions, threatening the unity and coherence of the
regional whole. As a consequence, regional modes of governance, that is the combined and
interrelated roles of markets and hierarchies as intercoupled through various levels of networking,
account largely for systemic success and growth. The ‘purposive nature’ of the geographical
context that is required for a regional system of innovation to develop does not necessarily have to
rely on a complete and adamantine mix of characters (such as those in the top-down and bottom-up
lists above). It does however have implications for the functional dimensions based on collective
learning, that is absorption, diffusion and creation of new knowledge, technology and innovation.
The null hypothesis here is the Marshallian cluster as arising out of ‘something in the air’: for an
RSI, the air has to be blown about by some deliberate and integrated decision-making process.

Weak or vulnerable regions often show innovation barriers such as organisational ‘thinness’,
institutional lock-in, social fragmentation and overall anti-developmental network governance
(von Tunzelmann, 2004; Tödtling and Trippl, 2004). Moreover, they are bound to be highly

7
A different and complementary perspective is to integrate the ‘regionalisation approach’ with the ‘regionalism
approach’, the first related to the region’s competence capacity, the second connected to the region’s cultural base

12
dependent on actual innovation systems, particularly those belonging to the same NSI; their need
for new technologies is satisfied mainly by mere adaptation of imported innovation, and they have
limited or null capacity to recombine and integrate old and new pieces of knowledge; their high
dependency on external providers is usually coupled with a low degree of openness and
attractiveness towards external resources; etc.. Even intermediate regional cases are often not
eligible for full RSI status: this applies, for instance, to those regions whose industrial base is open
to innovation, but which have only a weak scientific system; or regions where, in spite of a
relatively strong local scientific and research infrastructure, this is not sufficient to ensure spillover
effects (Vilanova and Leydesdorff, 2001; Evangelista et al., 2002).

Furthermore, it is essential to underline the shifts towards new techno-economic paradigms, as

they are likely to have a significant impact upon the evolutionary concept of a RSI, as well as on
actual regions. The spread of Information and Communication Technology (ICT) is far from being
uniform across space; also it is affecting the boundaries between codified and tacit knowledge,
with corresponding implications for the rationale of geographical agglomeration (see among
others, Mansell and Steinmueller, 2000; Steinmueller, 2001; Ernst et al., 2002); it is likely to
change the nature and the relative balance of the top-down and bottom-up features of RSIs; it
might broaden (or help reduce?) the gap between full RSIs and regional peripheries. To give an
example of these on-going transformations, the rising techno-economic paradigm is having major
consequences for local labour markets, leading to more unstable environments, transformation of
work regulations, individualisation of labour processes, changing institutional protection and
unemployment structure. These changes, which will alter the links between individuals and
societies, not only may create opportunities for rising RSIs, but also increase the risk of lock-in for
‘low road’ regions (Iammarino et al., 2004).

The advent of new paradigms also casts doubts on the stability of RSIs, once identified as such. In
an evolutionary perspective, the capacity of regions to withstand the (macro) processes of
technological change and globalisation is determined by the comparative advantage upon which
they can rely; and, as seen above, the features of the region’s internal organisation and structure
are subject to (micro-meso) change over time (as learning processes are, by definition, not stable).
This may lead to the strengthening or the breaking down of systemic coherence, even where a
system has been identified as existing. In other words, network governance alignment does not
only imply alignment of objectives (market/non-market, private/public) and levels of decision

from which the degree of systemic potential stems (Cooke et al., 1997).

13
making (micro, meso, macro); system integration increasingly also requires the alignment of old
and new technologies (von Tunzelmann, 2003).

To sum up, the conceptual framework as depicted here – that is the evolutionary integrated view of
RSIs – is believed to provide a suitable, though perhaps not perfect, interpretation tool for
analysing diverse regional configurations despite the fact that the replicability of RSIs on the basis
of stylised constructs has proved to be rather risky, because the idiosyncratic nature of
evolutionary mechanisms operating at the meso level rarely allows for problem-solving using
standardised procedures. As is argued below, to successfully accomplish the task of improving the
interpretative and normative framework to respond to the growing complexity and variety of the
actual world, the scope for testing its validity further should be considerably enhanced.

3.2 ‘Structures’ and ‘performance’ of RSIs

Measuring RSIs is one of the most discussed issues in this field of research. The conceptualisation
of geographical innovation systems has never overcome the serious drawbacks to data collection
and measurement with regard to smaller (than the national level) territorial units. It may be argued
that, somewhat surprisingly, in spite of the longstanding acknowledgement of the interactive
nature of innovative processes – which led to the substitution of the linear model by the chain-
linked model based on feedbacks, interactions and networks (Kline and Rosenberg, 1986) – the
empirical analysis of RSIs is still stuck in the consequential and oversimplified logic of the old
theoretical models of technology-push and demand-pull (Evangelista et al., 2002).

Two issues are worth highlighting here. The first is that generally available data and indicators are
appropriate for measuring RSI ‘performance’ – usually expressed in terms of economic
accomplishments, but they are of little help for examining regional ‘structures’, which reduces the
scope for testing the conceptual framework sketched above. Firm performance – as measured by
value added, exports, employment, innovative output, etc. –does not say much about the system, if
regional structural regularities have not been clearly identified: for instance, whose performance
are we measuring? The scant possibility to appraise, for example, interdependence at the inter-
regional level is due not only to the lack of sub-national data, but also to the shortage of indicators
appropriate to give account of the degree of attractiveness, dependence and openness of a region
or, in other words, of interregional innovation flows and networks. Turning to bottom-up
characteristics, the drawbacks become even more serious: the current state of the art on measures
and estimations of intra-regional innovative features and flows is, to say the least, not very

14
advanced in spite of the emphasis that regional evolutionary scholars have put on systemic
innovation, interactive learning processes and network modes of governance.

The second and related issue is that there are, in any case, inherent difficulties in drawing
inferences about performance from the data used in regional analysis. The concept of regional
competition is indeed well established as far as some aspects are concerned – i.e. conventional
economic strengths and weaknesses, but much less so in relation to others, such as innovation
processes, which are usually detected at the national level (and, at best, simply ‘regionalised’).
Moreover, in most of the existing empirical analyses the innovativeness of a specific region
corresponds simply to the sum of selected micro behaviours – e.g. the innovative activities of
resident firms – and not to that of the meso system as a whole. Hence, as argued by Dopfer et al.
(2004), we may well have good micro-macro arithmetic, but still we cannot capture the behaviour
of our unit of observation. What is depicted mainly is not the overall innovative performance of
the region, but rather the local reaction to often exogenous decisions and strategies. Therefore,
what performance are we measuring?

Data on performance in general do not reveal whether regional behaviour actually affects firm
behaviour, or whether it is simply an aggregation of it. There is evidence of regions identified as
RSIs displaying poor values in relation to traditional economic indicators, as well as of relatively
weak or even non-systems scoring rather well using the same standard measures (Cooke et al.,
1997; Evangelista et al., 2002).8 Yet, as stressed by Feldman and Martin (2004), firms’ success
and regional economic growth are mutually dependent and their interdependence may set up
virtuous (or vicious) historical cycles. Firms’ success depends on their external environment: there
is no unilateral causality nexus but, rather, a process of coevolution.

These shortcomings have so far prevented a complete integration between concepts and measures
of RSIs, thereby hampering the possibility of, on the one hand, enhancing the theoretical model
and, on the other, attaining a better normative justification for public policy. We very much agree
with Schwerin and Werker (2003) in that policy could learn from historical experience how to
identify localised patterns of socio-economic change. Although regional heterogeneity and the role
of chance will always provide margins for failure, a deeper understanding of regional path
dependency and historical contingency could strengthen the knowledge base upon which policy is

8
A further element that should not be disregarded given its increasing importance in innovative processes and
technological change, particularly in the ICT paradigm, is the role of the demand for innovation. This implies that
regions might be ranked according to their capacity to generate demand (and thus to create the conditions to provide a
supply to meet this demand). The demand for innovation comes from a variety of actors (public sector, firms, social
organisations) and their interactions, and it is usually difficult to detect using available information (Evangelista et al.,
2002).

15
built and policy learning occurs. A prerequisite for any sound normative base is to dig up far more
data and devise more adequate indicators than those currently available. As Schwerin and Werker
maintain “policy is intelligent in the sense that learning from past experience takes place”
(Schwerin and Werker, 2003: 402). In this light, the effectiveness of policy will be constrained as
long as available methods and measures do not allow testing of the concepts, or the hypotheses, to
correctly identify regional structural regularities and coevolution processes.9

4. Historical perspectives of RSIs: the Italian example

According to the evolutionary perspective, economic growth and change are the outcomes of
cumulative processes that operate through non-linear and self-reinforcing feedbacks between
technological and structural change. An array of constraints and conditions that are basically
idiosyncratic to structures and the historical paths of regions affect learning and knowledge
accumulation (Patrucco, 2004). Such conditions and constraints cannot be overlooked when
considering how knowledge and technology are absorbed, diffused and generated across space.
Nonetheless, as argued above, little progress has been made in testing the definition of RSI on the
basis of the importance attributed to localised knowledge generation and diffusion and
idiosyncratic learning capabilities. Even less work has been devoted to considering this definition
from the historical perspective likely to be enlightening about critical questions such as: which
regions are RSIs? Why? What does this status imply? Where do the modes of regional governance
come from in such systems and how do they evolve?

As was emphasised by Freeman (1995, 2002), a long-term historical approach is essential in order
to establish a true link between socio-economic growth, and innovation systems. The very nature
of technical and institutional change implies that it could take centuries for gaps between different
countries, or regions within a country, to show up, and a similarly long time may be required to
close them. As Carlsson pointed out, only a small percentage of innovation systems studies “can
be considered ‘dynamic’ in the sense that they focus on a historical process or development over
time rather than on a snapshot of a system in a particular time period” (Carlsson, 2003: 11).10 In

9
It is worth highlighting the meagreness of socio-economic analyses at the regional level in a dynamic perspective,
which should in principle be the basis of any effective policy. We are well aware of the huge costs associated with
data collection. However, in spite of the magnificent efforts of the EU National Institutes of Statistics and Eurostat to
provide territorial variables in many domains and, more rarely, to reconstruct historical time series of, at least, regional
GDP, the fundamental problem of sub-national data on innovation in the Union seems, rather surprisingly, to be rather
underestimated. For instance, it is still not clear whether and to what extent the regional dimension was one of the
sampling requirements in the most recent version of the Community Innovation Survey (CIS3).
10
There have recently been a few interesting attempts to investigate regional and local systems of innovation from a
historical perspective: see Rantisi (2002), Schwerin (2004), Boschma and Wenting (2005).

16
the main, the prevailing approach still adheres to a static view of the world (von Tunzelmann,
2003). The persistence of the empirical and historical ‘national bias’ is all the more serious in
countries characterised by a high degree of spatial imbalances, such as Italy, in which, therefore,
the scope for either the existence or the non-existence of RSIs is rather large.

Italy is a frequently cited example of a non-homogeneous NSI.11 It illustrates some crucial points
raised in the previous conceptual discussion, such as: sharp territorial differences in terms of both
heterogeneity and historical contingency; coexistence of actual RSIs and lack of a system among
its 20 administrative regions; highly problematic measurement issues (i.e. inadequacy of GDP-
based indicators to proxy socio-economic development due to persistent and significant inter- and
intra-regional differentiation; relatively low technological-intensity of specialisation patterns and
prevailing processes of ‘innovation without R&D’, which call into question the effectiveness of
standard indicators to grasp innovative phenomena); copious evidence of policy failure,
particularly with respect to innovation and technology policies.

The Italian ‘regional problem’ is arguably at least a thousand years old. It is interesting to contrast
some historical synopses with short descriptions of innovation features in the current Italian
regional panorama, outlining the long-term diversity of Italian regional trajectories in order to
support the case of history as the filter necessary to assess the existence of RSIs and, more
generally, opportunities for regional development. For the sake of brevity, a short overview by
current macro-regions (i.e. by aggregating the NUTS2 20 administrative regions in conventional
larger territorial units) is provided below.12

North-west:13 the Italian RSIs

Economic wealth and industrial orientation in the Italian north-west were historically rooted in the
substantial stability (since the 11th century), both political and administrative, of the Savoia family
in Piemonte, and the power of the Visconti and Sforza families (1300-1500) and the Asburgo
regime (up to Italian unification in 1860) in Lombardia. Agricultural development, incorporating
sophisticated irrigation techniques, was already advanced at the end of the Middle Ages; this
allowed intensive cropping, plant breeding and production of new cultivars. The abundance of
water in the area favoured the diffusion of water mills and looms, which played a crucial role in
the establishment of the first industrial settings for the production of wool, cotton, linen, silk and
manufactured textiles. Indeed, such activities stimulated diverse and related manufacturing

11
A number of studies have addressed the structure and performance of the Italian NSI as a whole. See, for example,
Antonelli, 1988; Malerba, 1993; Archibugi et al., 1991; Evangelista et al., 1997.
12
Based mainly on Cipolla (1974) and Zamagni (1990). For a more detailed discussion see Iammarino (2005).
13
Piemonte, Valle d’Aosta, Liguria and Lombardia.

17
productions, and were the basis for what was to become a machinery and equipment industry (still
a distinctive strength in the national specialisation model). This industrial development
overlapped, particularly in Lombardia, with the merchanting activity that had flourished since the
11th century in autonomous and prosperous communes (e.g. Milan, Como, Lodi, Cremona, Pavia),
where social institutions and networks in the modern sense had been in operation since their
establishment. Moreover, since the 16th century the North-west underwent wide institutional
reform and restructuring processes aimed at reducing the property and power of the church and the
nobility and making education accessible across social classes, thus providing significant dynamic
and innovative impulse to the knowledge base of the area. Following its accession to the Austro-
Hungarian empire (1714), Lombardia benefited from the political and custom unification of
Asburgo leadership, becoming integrated in economic networks at the continental level. At the
same time, the institutional reorganisation carried out in 1720-30 by Vittorio Amedeo II had
transformed Piemonte into the most organised and efficient bureaucratical state in the peninsula; in
the 1850s, under the political leadership of Cavour, a British-type economic policy spurred
commercial liberalisation, promotion of education and the expansion of transport networks. The
features peculiar to the Liguria region were connected to the history of Genoa – a crucial naval,
merchant and financial centre at the European level – and to its strong economic integration with
Piemonte, which were the basis for its regional development up to the present time.

Today Lombardia and Piemonte (and to a lesser extent Liguria) represent the technological heart
of Italian industry. The full range of links and interactions that form the skeleton of an innovation
system is very apparent in these two regions in the form, for example of: structured relationships
among firms and between firms and other organisations (universities, research institutes, industry
associations, etc.); a good scientific and technological infrastructure; strong R&D intensity;
diffused networks of technological services; attractiveness for external sources of technology;
effective regional innovation policies; relatively high institutional flexibility and adaptability to
change; and structured social ties and networks.

North-east and Centre-north:14 the Learning Regions

The commercial, industrial and economic ‘core’ of the North-east was certainly Venice, which, in
terms of political influence and economic power, was among the top European cities through the
period 1200-1700. Under the Venetian influence, clusters of manufacturing activities were
established in the area, some mainly specialised in traditional skill-intensive sectors (Murano’s
glasses, Vicenza’s jewels, Bassano’s ceramic tiles); some oriented towards ‘human resources

14
Veneto, Friuli Venezia Giulia and Trentino Alto Adige; Emilia Romagna and Toscana.

18
building’ (the prestigious university of Padua (1222) and the art centre of Verona); and others
geared toward trade, with docks and construction activities (Trieste). It might be argued that the
pattern of development in the Italian North-east followed a kind of ‘urban driven’ path, although
apart from the few urban agglomerations, most of the area suffered from being a ‘boundary region’
ruled by Venetian, and later on, Austrian political powers, with a strong inclination towards
primary activities and with a rather narrow diversification of manufacturing production. However,
in the late 20th century the North-east went through one of the most impressive processes of socio-
economic change to occur in Italy – the so-called ‘miracle of the Third Italy’ – with the rapid
transformation of its areas into fast-growing industrial and entrepreneurial regions (from
essentially agricultural-based production systems and societies).

The two regions of the Centre-north show some similarities in the historical character of their local
knowledge bases. Local governance, entrepreneurship and indigenous culture were firmly rooted
in Emilia and Toscana, comparable to Lombardia’s communes. In Emilia, various types of
sophisticated activity flourished under the progressive traditions of the dukedoms of Parma and
Piacenza (Visconti-Sforza in 1300-1500; Farnese family until 1730), Modena and Reggio (Estensi,
1300-1800) and Ferrara; the autonomous commune of Bologna, with its university established in
1088, exerted a powerful influence on the historical evolution of regional knowledge linkages.15
The industrial districts of Carpi, Sassuolo and Faenza are among the oldest in Italy. The
agricultural production systems, favoured by the enlightened political leaders of the big families,
the rather anti-clerical, innovative and open local culture, and the natural fertility of the soil, led to
rapid industrialisation in the form of a food production.16 The economic prosperity of Toscana
started in the 11th century based on the unbeatable cultural traditions of its communes (Lucca, Pisa,
Pistoia, Siena, Arezzo), all fierce competitors for market share, product variety, trade intensity,
cultural creativity and social dynamism, until finally Florence prevailed and absorbed the most
positive elements of Toscana’s typical pattern of development. Resources such as iron from Elba
and marble from Carrara were the basis for related manufacturing activities; and the strength of the
local communities and social ties, together with deliberate actions from local authorities, gave rise
to specialised dynamic clusters, which became the archetype for the ‘industrial district’ (e.g. the
widely-studied case of Prato).

15
Actually, between 1513 and 1859 Bologna was still formally part of the Catholic Church State, but was ruled by an
independent local authority.
16
It is interesting to note that the provinces of Emilia-Romagna that benefited least from the general wealth and
openness of the region were those in the east that had been subject to the Catholic Church State since 1500.

19
Nowadays, in all these regions knowledge flows and systemic interactions in innovation take the
form mainly of inter-firm user-producer interactions, which are particularly dense in those
industrial areas organised as districts. Such technological links, largely informal in nature and
loosely structured, are enhanced by spatial proximity and by an economic and cultural
homogeneity based on localised competencies. The innovation process is accomplished by
relatively modest research and development (R&D) activities: the public R&D system is also
rather weak. However, the innovation activities of firms rely upon a mix of codified knowledge
(engineering skills) and locally embedded competencies based on long established cumulative
learning processes. Linkages are facilitated by the high product specialisation of firms, and by
tightly integrated organisational models of production and favourable context-specific conditions
represented by the plurality of active institutional actors (e.g. specialised business services,
technology-transfer agencies, private business associations, chambers of commerce, and training
agencies).17

Centre:18 the Conservative Regions

This whole area was under the conservative political influence of the Roman Catholic Church for
varying lengths of time: Lazio since its foundation, Umbria and Marche since 1500. Whilst in the
latter two, rather small regions agriculture and ceramics industries were the major points of
strength, economic and social activities in Lazio were basically linked to the demands of a peculiar
pattern of urbanisation and of a wide bureaucratic structure (administrative, legal and
representative services, paper industry, etc.). The tertiary-oriented nature of the region continued
throughout the period of Italian unification and was even reinforced after the choice of Rome as
the capital of the kingdom.

As a result of its past and present role of capital-region, a large proportion of the national public
R&D infrastructures and expenditure is currently concentrated in Lazio. The most frequent forms
of linkages are those between a restricted number of science-based firms and public and private
research institutes. Yet, in the whole area systemic interactions do not play a very critical role;
support from local governments is neither proactive nor particularly effective; collaborative
relationships between firms, as well as other forms of knowledge and technological linkages, are
far from being intense; and regional social environments, more generally, are not particularly
cohesive, nor are their economic structures oriented towards technological change and institutional
reformism.

17
See also Boschma (2003). The literature that emphasises the diversity of industrial district evolution is not included
here for reasons of space.

20
South and Islands (Mezzogiorno):19 the Peripheral Regions

The so-called Italian Mezzogiorno had its golden age under the rule of the Normans (Frederick II),
who made the area economically and culturally advanced and promoted the magnificence of cities
like Naples and Palermo until 1266. Since then, the ‘Southern Kingdom’ was ruled by Spain: the
colony was administered on the basis of undisputed privileges given to the nobility, and the
exploitation of the local resources and the peasantry. The principal source of wealth was
agriculture, but this was often damaged by periods of famine and suffered from lack of any
innovations (under the long-lasting system of wide land estates); only in Puglia, positioned on the
eastern side, did agricultural and trading activities flourish. The Kingdom became ‘independent’
under the Bourbons (1734): the Bourbons have traditionally been blamed for the backwardness of
the area, but in fact were in power for only 126 years (roughly between 1734 and 1859, with the
inclusion of few years of French rule). Sardinia’s development was strongly affected by its long
colonisation, first by the Spanish, then by the Savoia from Piemonte who exploited the natural
resources of the island, without instituting any structural reforms or innovation processes (in 1840,
agricultural productivity in Sardinia was approximately 1/10 of that of Lombardia).

Various historical factors contributed to the weakness of the territorial systems and the scantiness
of social capabilities for institutional change in the whole of Southern Italy.20 Economic factors
include the late advent of a feudal system of agriculture (at a time when in the North the society of
communes was already established), a sharp separation between financial capitals and the
management and organisation of production, unstable and hierarchical economic relations, a rather
concentrated urban geography, and, more recently, a process of State-led industrialisation based on
resource- and scale-intensive sectors, with low degree of production interdependencies and limited
scope for either pecuniary or knowledge externalities.21

The current Mezzogiorno is a backward area, particularly in terms of innovation and technology
indicators, both in Italy and within the EU. Its vulnerability is related not only to the weak
technological performance of its firms, but also to the absence or weakness of any systemic

18
Lazio, Umbria, Marche.
19
Abruzzo, Molise, Campania, Puglia, Basilicata, Calabria, Sicilia and Sardinia.
20
It is clearly beyond the scope of this paper to go into the details of the complex ‘Mezzogiorno problem’, although its
relevance for the topic discussed here (whether and why some regions are RSIs and some are not even ‘systems’) is
crucial. For a more extended discussion see Iammarino (2005).
21
One of the causes underlying poor historical accounts is indeed the scarcity of interdisciplinary studies carried out at
the meso level. Excellent contributions, for examples, can be found in the work of Italian geographers on the historical
evolution of the country territorial polarisation. See, for example, Muscarà (1967, 1992), Bagnasco (1977), Celant
(1994).

21
dimension of innovation processes, at least at the level of administrative regions.22 Indeed, some
elements seem to suggest that firms in the main carry out innovation activities in isolation, and
have little contact with other firms, R&D institutes and the broader institutional context. The lack
of a critical mass of qualified components, relationships and attributes suggests that the status of
RSI cannot be identified in these regions. However, the emergence of “many Mezzogiorni”,
stressed by the economic literature particularly in the last decade, seems to suggest that both
catching up and falling behind processes are taking place in the area, giving rise to an increasing
intra-South differentiation (see, for instance, Guerrieri and Iammarino, 2002; Viesti, 2003).

This succinct picture serves the purpose of highlighting the relevance of history in understanding
the link between regional socio-economic and institutional trajectories and innovation systems.
Both heterogeneity, due to chance events, and path dependency, due to historical contingency,
underlie the longevity of Italian regions and their paths of development. Economic differentiation,
even more pronounced than political fragmentation, has been consolidating over the centuries,
preventing Italian economic development from achieving a single ‘natural dimension’ until quite
recently. The persistence of governance structures headed by communes, provincial dukedoms, the
Catholic Church, foreign leadership, etc., neatly indicates decision-making processes that are
overwhelmingly regionally, or even locally, based: in other words, the techno-economic aspects of
innovation have to be necessarily associated with some component of governance. There is a key
issue here about active and passive decisions concerning innovation: the extent to which the
accumulation of ‘surpluses’ (in the sense of economic development), and especially their
deployment to innovation-related or industrial-oriented activities, may be a crucial determinant of
either the presence/absence or the success/failure of geographical systems of innovation.

Hence, if on the one hand the long-term backwardness of the Mezzogiorno can be explained by
naturally poor resource endowments and unfavourable geo-climatic conditions, on the other it
definitely cannot be seen only as the outcome of accident or fate. Rather, it was coupled with the
accumulation of what small surpluses there were being in the hands of anti-developmental
governance, i.e. from southern barons in the 18th century to the more recent Mafia.23 In the Central
regions natural endowments were presumably richer, but the dominance of the Catholic Church –

22
Indeed, some elements (rules, incentives, structural and historical similarities, etc.) seem to suggest that the whole
Mezzogiorno could be considered as a ‘uniform’ macro-region but, within it, the degree of differentiation, particularly
in terms of performance, is indeed quite striking.
23
To quote an example of wealth concentration, at the end of 18th century 60% of the total income produced in the
continental southern regions was owned by 650 families, of which 90 controlled two-thirds and 20 approximately a
quarter of the total population respectively (Villani, 1973 not in refs, quoted in Zamagni, 1990).

22
universally condemned by historians as responsible for wealth concentration and anti-innovative
practices – took most of such surpluses. In northern Italy, and especially in the North-west, the
endowments were richer still and some were used ‘productively’ by both local community actions
and progressive foreign powers.

Indeed, what emerges from current attempts to apply the RSI concept to the Italian case (Silvani et
al., 1993; Iammarino et al., 1996, 1999; Camagni and Capello, 1999; Evangelista et al., 2001,
2002) is a variety of regional patterns, differing in terms not only of firms’ strategies and
performance, but also in terms of contextual and systemic characteristics, density and quality of
interactions and scope and effectiveness of functions. Thus, regional diversity depends on to
components, relationships and attributes, to social capabilities for institutional change and to the
degree of alignment in network governance.

The portrait of the contemporary Italian regional divide consistently reproduces the broad
historical trajectories. Though currently largely disregarded in academic and policy debates,
quantitative and qualitative historical research based on carefully collected data and in-depth case
studies may help reveal the structural regularities of socio-economic evolution and the degree of
integration of governance design controlling the efficiency of knowledge absorption, diffusion and
creation at both micro and meso levels.

5. Conclusions

It has been argued that by adopting an evolutionary integrated approach, the RSI framework does
provide an adequate conceptual base from which to investigate whether and why a region is an
innovation system. However, it is also argued that the severe constraints on empirical testing of the
significance of structures and performances of RSIs – due to both insufficient information and
methodological bottlenecks – have so far seriously hampered the achievement of a better
delineation of the concept, as well as of a sounder normative base for regional policy. Clear
thinking is a prerequisite for fruitful research efforts, but it is not a sufficient condition per se:
adequate data and robust indicators to identify structural regularities and patterns of socio-
economic change are urgently required. The extent of applicability and replicability of RSIs will
arguably widen only if supported by richer sets of both quantitative and qualitative variables and
by historical evidence on territorial paths of evolution.

Any attempt to analyse why a region is a particularly successful RSI must first, because systems
change over time, explain what are the factors underlying such status and their dynamics.

23
Therefore, an evolutionary approach may be the only approach of value to deal with the unsolved
questions highlighted by the recent literature on regional development in Europe: for example,
how can regional trajectories be abandoned? Is it possible to diverge from the social, economic
and institutional past? How can structural and localised change be planned and managed? Is there
a role for regional policy at all levels of governance? (Boschma, 2004).

A further issue is related to the ‘alignment of networks’ perspective, and concerns the top-down
versus bottom-up distinction. The question then becomes one of how the contrasting levels can (or
cannot) be integrated. The agents do not have to be the same in both cases, and may indeed have
different political or other objectives, but there has to be a joint and reasonably consistent set of
views about enhancing the regional knowledge base. Only advanced historical knowledge can
make visible the extent of the integration between and within networks as modes of governance in
the region. As evolutionary mechanisms like ‘technological trajectory’, ‘selection environment’,
‘heterogeneity’ and ‘path dependency’ are crucial determinants of the geography of innovation,
history in terms of inheritance of regional structures and governance often acts as a filter for
assessing new opportunities for social and techno-economic growth. As was effectively pointed
out by Lambooy and Boschma (2001), this is all the more important when policy implications are
to be drawn: regional policy – at local, national and supranational (EU) level – cannot start from
scratch, but rather it has to be based on a deep understanding of how historical trajectories affect
change. As Lambooy and Boschma have argued, in an evolutionary systemic environment, it is
unlikely that policy makers will fundamentally change the course of development of a region as:

Regional policy is likely to fail when local strategies deviate considerably from the local
context. In such circumstances, policy makers have to account for the fact that adaptation
to change is largely constrained by boundaries of the spatial system laid down in the past.
However, this also implies that the potential impact of regional policy may be quite large
when the policy objectives are strongly embedded in the surrounding environment
(Lambooy and Boschma, 2001: 113).

The Italian experience summarised in this paper gives some preliminary support to the fact that
regions differ not only according to the specific strategies and performance of firms, but also with
respect to the density of systemic interactions, quantity and quality of idiosyncratic factors, and
evolution of regional social capabilities and modes of governance, favourable (or unfavourable) to
innovation. This is not to deny the importance of the national system, but rather to stress the
complementarity between an NSI and its regions/RSIs. The importance of history, which is
emphasised here, goes against an a priori determination of the spatial level relevant for innovation

24
systems to exist and operate. It rather shows that there is an array of conditions and constraints
dictated by the heterogeneity and path dependency of localities, regions and countries, affecting
learning, knowledge accumulation and, ultimately, socio-economic growth.

Future research will be directed to exploring more in depth the link between regional socio-
economic growth and systems of innovation, as well as the complex relationship between
governance and technology, at both the historical and empirical levels, and devising indicators
able to reflect, at least to an extent, the region as a ‘social system’.

25
References

ABRAMOVITZ, M. (1986) Catching up, forging ahead, and falling behind, Journal of Economic History,
46, pp. 385-406.
ANTONELLI, C. (1988) New Information Technology and Industrial Change: The Italian Case.
Dordrecht: Kluwer.
ARCHIBUGI, D., CESARATTO, S. and SIRILLI G. (1991) Sources of innovative activities and industrial
organisation in Italy, Research Policy, 20, pp. 299-313.
ASHEIM, B.T. (1995) Industrial districts as learning regions: a condition for prosperity?, STEP Report No.
3, Oslo: STEP Group.
ASHEIM, B.T. and GERTLER, M.S. (2003) Regional Innovation Systems and the Geographical
Foundations of Innovation, TEARI WP no. 11, October.
ASHEIM, B.T. and ISAKSEN, A. (2002) Regional innovation systems: The integration of local ‘sticky’
and global ‘ubiquitous’ knowledge, Journal of Technology Transfer, 27, pp. 77-86.
ASHEIM, B.T. and ISAKSEN, A. (1997) Location, agglomeration and innovation: Towards regional
innovation systems in Norway?, European Planning Studies, 5, 3, pp. 299-330.
AUDRETSCH, D.B. and FELDMAN, M.P. (1996) Knowledge Spillovers and the Geography of Innovation
and Production, American Economic Review, 86, 3, pp. 630-640.
AYDALOT, P. (Ed.) (1986) Milieux Innovateurs in Europe. Paris: Gremi.
BAGNASCO, A. (1977) Tre Italie. La problematica dello sviluppo italiano. Bologna: Il Mulino.
BECATTINI, G. (Ed.) (1987) Mercato e forze locali. Il distretto industriale. Bologna: Il Mulino.
BOSCHMA, R.A. (2004) Rethinking Regional Innovation Policy. The making and breaking of regional
history, in G. FUCHS and P. SHAPIRA (Eds) Rethinking regional innovation and change: path-
dependency or regional breakthroughs? Dordrecht: Kluwer International Publishers.
BOSCHMA, R.A. (2003) Social capital and regional development: an empirical analysis of the Third Italy,
University of Utrecht, mimeo.
BOSCHMA, R.A. and WENTING, R. (2005) The spatial evolution of the British automobile industry in
Great Britain, 1895-1968, in P. NIJKAMP and A. REGGIANI (Eds) Spatial evolution and modelling.
Cheltenham: Edward Elgar.
BOSCHMA, R.A. and LAMBOOY, J.G. (1999) Evolutionary economics and economic geography,
Journal of Evolutionary Economics, 9, pp. 411-429.
BRACZYK, H.J., COOKE, P. and HEIDENREICH, M. (Eds) (1998) Regional Innovation Systems.
London: UCL Press.
BRESCHI, S. (1995) La dimensione spaziale del mutamento tecnologico: una proposta interpretativa,
Economia e Politica Industriale, 86, pp. 179-207.
BRESCHI, S. and LISSONI, F. (2001) Knowledge Spillovers and Local Innovation Systems: A Critical
Survey, Industrial and Corporate Change, 10, 4, pp. 975-1005.
CAMAGNI, R. and CAPELLO, R. (1999) Innovation and Performance of SMEs in Italy: The Relevance of
Spatial Aspects, in M.M. FISCHER, L. SUAREZ-VILLA and M. STEINER (Eds) Innovation, Networks
and Localities. Berlin: Springer-Verlag.
CANTWELL, J.A. and IAMMARINO, S. (2003) Multinational Corporations and European Regional
Systems of Innovation. London and New York: Routledge.
CARLSSON, B. (2003) Innovation Systems: A Survey of the Literature from a Schumpeterian Perspective,
Case Western Reserve University, mimeo.

26
CARLSSON, B. and STANKIEWICZ, R. (1991) On the nature, function and composition of technological
systems, Journal of Evolutionary Economics, 1, pp. 93-118.
CELANT, A. (1994) Eliminating the Gap, Public Policies and the Development of the Italian South, in
SOCIETÀ GEOGRAFICA ITALIANA, Restructuring Processes in Italy. Rome.
CIPOLLA, C.M. (1974) Storia economica dell’Europa pre-industriale. Bologna: Il Mulino.
COOKE, P. (2001) Regional innovation systems, clusters, and the knowledge economy, Industrial and
Corporate Change, 10, 4, pp. 945-974.
COOKE, P. (1992) Regional innovation systems: competitive regulation in the New Europe, Geoforum,
23, pp. 365-382.
COOKE, P. and MEMEDOVIC, O. (2003) Strategies for Regional Innovation Systems: Learning
Transfer and Applications, Policy Papers, Vienna: UNIDO.
COOKE, P., BOEKHOLT, P. and TODTLING, F. (2000) The Governance of Innovation in Europe.
Regional Perspective on Global Competitiveness. London: Pinter.
COOKE, P., GOMEZ URAGA, M. and ETXEBARRIA, G. (1997) Regional innovation systems:
institutional and organisational dimensions, Research Policy, 26, pp. 475-491.
DOPFER, K., FOSTER, J. and POTTS, J. (2004) Micro-meso-macro, Journal of Evolutionary
Economics, 14, pp. 263-279.
DOSI, G. (1988) Sources, procedures and microeconomic effects of innovation, Journal of Economic
Literature, 26, pp. 1120-1171.
EDQUIST, C. (1997) Systems of innovation: technologies, institutions and organisations. London:
Pinter.
ERNST, D., FAGERBERG, J. and HILDRUM, J. (2002) Do Global Production Networks and Digital
Information Systems make knowledge spatially fluid?, TIK WP no. 13, Oslo: University of Oslo.
EVANGELISTA, R., IAMMARINO, S., MASTROSTEFANO, V. and SILVANI, A. (2002) Looking for
regional systems of innovation. Evidence from the Italian innovation survey, Regional Studies, 36, 2, pp.
173-186.
EVANGELISTA, R., IAMMARINO, S., MASTROSTEFANO, V. and SILVANI, A. (2001) Measuring the
regional dimension of innovation. Lessons from the Italian innovation survey, Technovation, 21/11, pp.
733-745.
EVANGELISTA, R., PERANI, G., RAPITI, F. and ARCHIBUGI, D. (1997) Nature and impact of
innovation in manufacturing industry: some evidence from the Italian innovation survey, Research Policy,
26, pp. 521-536.
FAGERBERG, J. (1994) Technology and International Differences in Growth Rates, Journal of Economic
Literature, 32, pp. 1147-75.
FAGERBERG, J. (1987) A Technology Gap Approach to Why Growth Rates Differ, Research Policy, 16,
2-4, pp. 87-99.
FAGERBERG, J. and VERSPAGEN, B. (1996), Heading for divergence? Regional growth in Europe
reconsidered, Journal of Common Market Studies, 34, 3, pp. 431-448.
FAGERBERG, J., VERSPAGEN, B. and von TUNZELMAN, N. (eds.) (1994) The Dynamics of
Technology, Trade and Growth. Aldershot: Edward Elgar.
FAGGIAN, A. and MCCANN, P. (2004) Human Capital Flows and Regional Knowledge Assets: A
Simultaneous Equation Approach, University of Reading, mimeo.
FELDMAN, M. and MARTIN, R. (2004) Jurisdictional Advantage, paper presented at the Conference on
“Regionalisation of Innovation Policy”, Berlin, 4-5 June.

27
FREEMAN, C. (2002) Continental, national and sub-national innovation systems – complementarity and
economic growth, Research Policy, 31, 2, pp. 191-211.
FREEMAN, C. (1995) The National System of Innovation in Historical Perspective, Cambridge Journal
of Economics, 19, 1, pp. 5-24.
FREEMAN, C. (1987) Technology Policy and Economic Performance. London: Pinter.
FRITSCH, M. (2001) Cooperation in regional innovation systems, Regional Studies, 35, pp. 297-307.
GORDON, I.R. and MCCANN, P. (2000) Industrial Clusters: Complexes, Agglomeration and/or Social
Networks?, Urban Studies, 3, pp. 513-532.
GUERRIERI, P. and IAMMARINO, S. (2002), Vulnerabilità e regioni nell’Unione Europea: un esercizio
sul Mezzogiorno italiano, Italian Journal of Regional Sciences, 2, pp. 5-28.
HÄGERSTRAND, T. (1967) Innovation Diffusion as a Spatial Process. Chicago: University of Chicago
Press.
HOWELLS, J. (1999) Regional Systems of Innovation?, in D. ARCHIBUGI, J. HOWELLS and J. MICHIE
(Eds) Innovation policy in a global economy, pp. 67-93. Cambridge: Cambridge University Press.
HOWELLS, J. (1996) Tacit knowledge, innovation and technology transfer, Technology Analysis &
Strategic Management, 8, pp. 91-106.
IAMMARINO, S. (2005) Heterogeneity and path-dependency in regional innovation trajectories. The
Italian historical divide, SPRU-University of Sussex, in progress.
IAMMARINO, S., JONA LASINIO, C. and MANTEGAZZA, S. (2004) Labour productivity, ICT and
regions. The revival of the Italian dualism?, SEWP no. 127, SPRU, University of Sussex.
IAMMARINO, S., PRISCO, M.R. and SILVANI, A. (1999) The geography of production and innovation:
how regional “styles” play in the global scenario, Regional Science Review, 18, pp. 31-45.
IAMMARINO, S., PRISCO, M.R. and SILVANI, A. (1996) La struttura regionale dell’innovazione,
Economia e Politica Industriale, 89, pp. 187-229.
KLINE, G.J. and ROSENBERG, N. (1986) An overview of innovation, in R. LANDAU and N.
ROSENBERG (Eds) The Positive Sum Strategy: Harnessing Technology For Economic Growth, pp.
275-306. Washington, D.C.: National Academy Press.
LAMBOOY, J.G. and BOSCHMA, R.A. (2001) Evolutionary economics and regional policy, The Annals
of Regional Science, 35, pp. 113-131.
LUNDVALL, B.Å. (1996) The social dimension of the learning economy, DRUID Working Papers no. 96-
1, Aalborg University.
LUNDVALL, B.Å. (1992) National systems of innovation: Towards a theory of innovation and
interactive learning. London: Pinter.
LUNDVALL, B.Å. (1988) Innovation as an Interactive Process: From User-Producer Interaction to the
National System of Innovation, in G. DOSI, C. FREEMAN, R. NELSON, G. SILVERBERG and L.
SOETE (Eds), Technical Change and Economic Theory. London: Pinter Publishers.
MALECKI, E.J. (1983) Technology and Regional Development: a survey, International Regional Science
Review, 8, pp. 89-125.
MALECKI, E.J. (1977) Firms and innovation diffusion, Environment and Planning, A9, pp. 1291-1305.
MALERBA, F. (1993) The National System of innovation: Italy, in R.R. NELSON (Ed.).
MANSELL, R. and STEINMUELLER, W.E. (2000) Competing Interests and Strategies in the Information
Society, in Mobilizing the Information Society: Strategies for Growth and Opportunity. London:
Oxford University Press.
MARKUSEN, A. (1996) Sticky Places in Slippery Space: A Typology of Industrial Districts, Economic
Geography, 72, pp. 293-313.

28
MARKUSEN, A. (1985) Profit Cycles, Oligopoly, and Regional Development. Cambridge, MA: MIT
Press.
MORGAN, K. (2004) The exaggerated death of geography: learning, proximity and territorial innovation
systems, Journal of Economic Geography, 4, pp. 3-21.
MORGAN, K. (1997) The learning region: institutions, innovation and regional renewal, Regional Studies,
31, pp. 491-504.
MOULAERT, F. and SEKIA, F. (2003) Territorial Innovation Models: A Critical Survey, Regional
Studies, 37, 3, pp. 289-302.
MUSCARA’, C. (1992) Dal decentramento urbano alla ripolarizzazione dello spazio geografico
italiano. Rome: Società Geografica Italiana.
MUSCARA’, C. (1967) La geografia dello sviluppo. Milan: Comunità.
NELSON, R.R. (ed.) (1993) National innovation systems: a comparative analysis. London: Oxford
University Press.
NELSON, R.R. and ROSENBERG, N. (1993) Technical innovation and national systems, in R.R.
NELSON (Ed.).
NELSON, R.R. and WINTER, S. (1992) An evolutionary theory of economic change. Cambridge, MA:
Harvard University Press.
PATRUCCO, P. (2004) The economics of collective knowledge and technological communication, SPRU-
University of Sussex, May, mimeo.
PERROUX, F. (1950) Economic space: theory and applications, Quarterly Journal of Economics, 1, pp.
89-104.
PRAKASH, A. and HART, J. (1999) Globalisation and Governance. London: Routledge.
PRED, A. (1977) City Systems in Advanced Economies. Cambridge, MA: Harvard University Press.
PRED, A. (1967) Behaviour and Location. Part I. Foundations for a Geographic and Dynamic
Location Theory. Lund: Gleerup.
RANTISI, N.M. (2002) The Local Innovation System as a Source of ‘Variety’: Openness and Adaptability
in New York City’s Garment District, Regional Studies, 36, 6, pp. 587–602.
RODRÍGUEZ-POSE, A. (1994) Socioeconomic Restructuring and Regional Change: Rethinking Growth in
the European Community, Economic Geography, 79, 4, pp. 325-343.
RODRÍGUEZ-POSE, A. (1998) Dynamics of Regional Growth in Europe. Social and Political Factors.
Oxford: Clarendon Press.
SAXENIAN, A. (1994) Regional advantage. Culture and competition in Silicon Valley and route 128.
Cambridge, MA: Harvard University Press.
SCHWERIN, J. (2004) The evolution of the Clyde region’s shipbuilding innovation system in the second
half of the nineteenth century, Journal of Economic Geography, 4, 1, pp. 83-101.
SCHWERIN, J. and WERKER, C. (2003), Learning innovation policy based on historical experience,
Structural Change and Economic Dynamics, 14, pp. 385-404.
SILVANI, A., DE BRESSON, C., BERNI, A. and HU, X. (1993) La localisation regionale des grappes
d’innovation en Italie: Troisième Italie ou Lombardie?, Revue d’Economie Regionale et Urbaine, 2, pp.
244-280.
SIMMIE, J. (2003) Innovation and Urban Regions as National and International Nodes for the Transfer and
Sharing of Knowledge, Regional Studies, 37, 6&7, pp. 607-620.
STEINMUELLER, W.E. (2001) Seven Foundations of the Information Society: A Social Science
Perspective, Journal of Science Policy and Research Management, 16, 1/2, pp. 4-19.

29
STEINMUELLER W.E. (2000) Will New Information and Communication Technologies Improve the
‘Codification’ of Knowledge?, Industrial and Corporate Change, 9, 2, pp. 361-376.
STORPER, M. (1998) The Regional World: Territorial Development in a Global Economy, New York:
Guilford.
STORPER, M. (1992), The limits of globalisation: technology districts and international trade, Economic
Geography, 68, pp. 60-92.
TÖDTLING, F. and TRIPPL, M. (2004), One size fits all? Towards a differentiated policy approach with
respect to regional innovation systems, paper presented at the Conference on “Regionalisation of Innovation
Policy”, Berlin, 4-5 June.
van DER MEER, A., van WINDEN, W. and WOETS, P. (2003) ICT clusters in European cities during the
1990s: development patterns and policy lessons, paper presented at the European Regional Science
Association Conference, August, Jyväskylä, Finland.
VIESTI, G. (2003) Abolire il Mezzogiorno. Rome-Bari: Editori Laterza.
VILANOVA, M. and LEYDESDORFF, L. (2001) Why Catalonia cannot be considered as a Regional
Innovation System, Scientometrics, 50, 2, pp. 215-240.
von TUNZELMANN, N. (2004) Network alignment in the catching-up economies of Europe, in F.
Mcgowan, S. RADOSEVIC and N. von TUNZELMANN (Eds) The Emerging Industrial Structure of
the Wider Europe. London and New York: Routledge.
von TUNZELMANN, N. (2003) Historical coevolution of governance and technology in the industrial
revolutions, Structural Change and Economic Dynamics, 14, pp. 365-384.
ZAMAGNI, V. (1990) Dalla periferia al centro. La seconda rinascita economica dell’Italia (1861-
1990). Bologna: Il Mulino.

30