Deconstructing Clusters:

Chaotic Concept or Policy Panacea?

Ron Martin
Department of Geography
University of Cambridge, UK
E-mail: [email protected]
and
Peter Sunley
Department of Geography
University of Edinburgh, UK
E-mail: [email protected]

Acknowledgements Earlier versions of this paper were presented at the

Regional Studies Association Conference on Regionalising the Knowledge Economy
London, 21 November 2001; at the Cluster Workshop, University of Lund,
Sweden, 5 April, 2002; at the High Technology Small Firms One-Day Clusters
Conference, Manchester Business School, 18 April 2002; and at the Department of
Geography, University College London. Many valuable comments were made
on those occasions. We are grateful also to two anonymous referees for helpful
suggestions.

Forthcoming in Journal of Economic Geography, 3,1

(6 June 2002)

1
 Abstract
Over the past decade, there has been growing interest in local industrial
agglomeration and specialisation, not only from economic geographers but also
from economists and by policy-makers. Of the many ideas and concepts to have
emerged from this new-found focus, Michael Porter’s work on ‘clusters’ has
proved by far the most influential. His ‘cluster theory’ has become the standard
concept in the field, and policy-makers the world over have seized upon Porter’s
cluster model as a tool for promoting national, regional and local
competitiveness, innovation and growth. But the mere popularity of a construct
is by no means a guarantee of its profundity. Seductive though the cluster
concept is, there is much about it that is problematic, and the rush to employ
‘cluster ideas’ has run ahead of many fundamental conceptual, theoretical and
empirical questions. Our aim is to deconstruct the cluster concept in order to
reveal and highlight these issues. Our concerns relate to the definition of the
cluster concept, its theorisation, its empirics, the claims made for its benefits and
advantages, and its use in policy-making. Whilst we do not wish to debunk the
cluster idea outright, we do argue for a much more cautious and circumspect use
of the notion, especially within a policy context: the cluster concept should carry
a public policy health warning.

Key words: Business location Clusters Porter Chaotic concept Cluster

empirics Cluster theory Cluster Policy The cluster ‘brand’

2
 “When I use a word”, Humpty Dumpty said in a rather scornful
tone, “it means just what I choose it to mean - neither more nor
less” (Lewis Carroll, Through the Looking Glass, 1872).

“For an idea ever to be fashionable is ominous, since it must

afterwards be always old-fashioned” (George Santayana, Winds of
Doctrine, 1913).

1. Introduction: Clusters and the Reassertion of Location

In recent years there has been a growing interest in the role of location in
the global economy. Some have argued that globalisation is rendering the
significance of location for economic activity increasingly irrelevant (O’Brien,
1992; Cairncross, 1997; Gray, 1998). Others, however, espouse the opposite view,
that globalisation is actually increasing rather than reducing the importance of
location, that it is promoting greater regional economic distinctiveness, and that
regional economies rather than national economies are now the salient foci of
wealth creation and world trade (Ohmae, 1995; Coyle, 1997, 2001; Krugman,
1997; Porter, 1998; Scott, 1998, 2001; Fujita, Krugman and Venables, 2000). Thus,
as the business economist Michael Porter puts it:

In a global economy – which boasts rapid transportation, high

speed communications and accessible markets – one would expect
location to diminish in importance. But the opposite is true. The
enduring competitive advantages in a global economy are often
heavily localised, arising from concentrations of highly specialised
skills and knowledge, institutions, rivalry, related businesses, and
sophisticated customers (Porter, 1998 c, p.90).

At the same time, it is alleged, increasing global economic integration itself

leads to heightened regional and local specialisation, as falling transport costs
and trade barriers allow firms to agglomerate with other similar firms in order to
benefit from local external economies of scale (Krugman, 1991, Fujita, Krugman,
Venables, 2000), which in their turn are thought to raise local endogenous
innovation and productivity growth (see Martin and Sunley, 1998). For these
and other related reasons, it has become fashionable within certain academic and
policy circles to talk of the ‘re-emergence of regional economies’ (Sabel, 1989),
the ‘localization of the world economy’ (Krugman, 1997) and the rise of a ‘global
mosaic of regional economies’ (Scott, 1998).

3
 One of the most influential – indeed, the most influential - exponent of this
emphasis on economic localisation is Michael Porter, whose notion of industrial
or business clusters has rapidly become the standard concept in the field.
Moreover, Porter has not only promoted the idea of ‘clusters’ as an analytical
concept, but also as a key policy tool. As the celebrated architect and promoter of
the idea, Porter himself has been consulted by policy makers the world over to
help them identify their nation’s or region’s key business clusters or to receive
his advice on how to promote them. From the OECD and the World Bank, to
national governments (such as the UK, France, Germany, the Netherlands,
Portugal, and New Zealand), to regional development agencies (such as the new
Regional Development Agencies in the UK), to local and city governments
(including various US states), policy-makers at all levels have become eager to
promote local business clusters. Nor has this policy interest been confined to the
advanced economies: cluster policies are also being adopted enthusiastically in
an expanding array of developing countries (see Doeringer and Terka, 1996;
Schmitz, 2000; World Bank, 2000). Clusters, it seems, have become a world-wide
fad, a sort of academic and policy fashion item.
The more so because the concept has become increasingly associated with
the so-called ‘knowledge economy’, or what some have labelled the ‘New
Economy’. Norton (2001), for example, argues that the global leadership of the
US in the New Economy derives precisely from the growth there of a number of
large, dynamic clusters of innovative entrepreneurialism. In the US, Porter is
himself leading a major policy-driven research programme to “develop a
definitive framework to evaluate cluster development and innovative
performance at the regional level” in order to identify the ‘best practices’ that can
then be used “to foster clusters of innovation in regions across the country”
(Porter and Ackerman, 2001; Porter and van Opstal, 2001). Likewise, the OECD
(1999, 2001) sees innovative clusters as the drivers of national economic growth,
and as a key policy tool for boosting national competitiveness.
But the mere popularity of a construct is by no means a guarantee of its
profundity. Our argument here is that seductive though the concept is, there is
much about it that is problematic, in that the rush to employ ‘cluster ideas’ has

4
run ahead of many fundamental conceptual, theoretical and empirical questions
(Held, 1996; Steiner, 1998). Whilst it is not our intention to debunk the cluster
idea outright, we do argue for a much more cautious and circumspect use of the
notion, especially within a policy context. We begin by asking why it is that
Porter’s notion of ‘clusters’ has gate-crashed the economic policy arena when the
work of economic geographers on industrial localization, spatial agglomeration
of economic activity and the growing salience of regions in the global economy,
has been largely ignored.

2. Why ‘Clusters’?
As Porter admits, the idea of specialised industrial localisation is hardly
new. As is well-known, Alfred Marshall, writing at the end of the nineteenth
century, included a chapter in his Principles of Economics (1890) on ‘the
concentration of specialised industries in particular localities’. His
characterisation of these local concentrations of specialised activity was cast in
terms of a simple triad of external economies: the ready availability of skilled
labour, the growth of supporting and ancillary trades, and the specialisation of
different firms in different stages and branches of production.
A century later and Porter’s neo-Marshallian cluster concept has burst on
the scene. Its origins can be traced to his earlier work in the late-1980s and early-
1990s on national competitive advantage and international competitiveness, in
which he argued that the success of a nation’s export firms depends on a
favourable national ‘competitive diamond’ of four sets of factors: firm strategy,
structure and rivalry; factor input conditions; demand conditions; and related
and supporting industries. The more developed and intense the interactions
between these four sets of factors, the greater will be the productivity of the
firms concerned (Porter, 1990).
Porter then argued, and this has since become his key theme, that the
intensity of interaction within the ‘competitive diamond’ is enhanced if the firms
concerned are also ‘geographically localised’ or ‘clustered’. In his view, the
geographic concentration of firms in the same industry is “strikingly common
around the world” (1990, p. 120). More specifically, he suggests that a nation’s

5
 most globally competitive industries are also likely to be ‘geographically
clustered’ within that nation. Hence, what originally started out as a way of
decomposing a national economy, the competitive diamond as a group of
interlinked industries and associated activities, has become a spatial metaphor,
the cluster as a geographically localised grouping of interlinked businesses. The

Figure 1

Porter’s Competitive Diamond of Local Industrial Clustering (Based

on Porter, 1998, Ch 10).

Firm Strategy and Rivalry

Vigorous competition among
locally-based rivals

Factor Input Conditions Local Context Demand Conditions

Local labour, capital and natural Local environment Sophisticated and
resources; that encourages demanding local customers
physical, administrative, appropriate forms Specialised local demand
information and technological of investment and Customer needs that anticipate
infrastructures; sustained those elsewhere
specialised inputs upgrading

Related and Supporting

Industries
Presence of capable locally-
based suppliers and competitive
related industries

competitive diamond is the driving force making for cluster development, and
simultaneously the cluster is the spatial manifestation of the competitive
diamond (Figure 1). The systemic nature of the diamond produces local
concentration of the leading rival firms, which in its turn magnifies and
intensifies the interactions between the factors. Hence, according to Porter (1990,
p. 157), “The process of clustering, and the intense interchange among industries
in the cluster, also works best where the industries involved are geographically

6
concentrated”(emphasis added). There is then an obvious affinity between
Porter’s schematic ‘competitive diamond’ of local business clustering and
Marshall’s ‘triad’ of external economies of industrial localisation.
But Porter’s cluster notion is not the only rediscovery and reinvention of
Marshall’s ideas to have taken place in recent years. For the past two decades or
more, economic geographers have devoted considerable effort to studying local
industrial specialisation, spatial economic agglomeration and regional
development, and to identifying the economic, social and institutional processes
involved. They too have invented a whole series of neologisms to capture and
represent the spatial form and nature of local business concentrations, including:
‘industrial districts’, ‘new industrial spaces’, ‘territorial production complexes’,
‘neo-Marshallian nodes’, ‘regional innovation milieux’, ‘network regions’, and
‘learning regions’ (see for example, Scott, 1988; 1998; Amin and Thrift, 1992;
Harrison, 1992; Harrison, Kelly and Grant, 1996; Markusen, 1998; Asheim, 2000).
Not only is this corpus of work by economic geographers largely ignored by
Porter (and by other economists who have recently discovered geography, such
as Paul Krugman), in total contrast to his cluster concept their ideas have
singularly failed to have any major impact on policy-makers. Why then has his
work proved so fashionable and influential while that of economic geographers
has not? Why have some economic geographers themselves started to use
cluster terminology in preference to their own (for example, Pinch and Henry,
1999; May, Mason and Pinch, 2001; Scott, 2001; Keeble and Wilkinson, 2000;
Keeble and Nachum, 2002)?
One possible reason is that, from the beginning, Porter has rooted and
promoted his cluster concept within an overarching focus on the determinants of
‘competitiveness’ (of firms, industries, nations, and now locations). This
resonates closely with the growing emphasis given by politicians and policy-
makers to the importance of competitiveness for succeeding in today’s global
economy. Porter’s avowed aim is to inform companies, cities, regions and
nations how to compete on the world stage, and the undoubted lure of his
cluster concept is that it sits well with the current preoccupation with micro-
economic supply-side intervention, and especially with the policy imperatives of

7
raising productivity and innovation (Porter, 1996; 1998b,c; 2000a, b, c). As an
alleged key determinant of competitiveness, Porter’s clusters have inevitably
attracted considerable interest, particularly given the emphasis he is currently
assigning to geographical industrial clusters in promoting the competitive
advantage of the US economy (Porter and Opstal, 2001). Economic geographers’
work on industrial localisation and regional agglomeration, on the other hand,
has tended to be more diffuse in its aims, and much less concerned with core
issues such as the performance, productivity and competitiveness of firms.
A second, and related, reason could be the way in which Porter has
conveyed his ideas on clusters. His discussion is framed directly in terms of the
economics of ‘business strategy’, and not in terms of the sorts of more general
theoretical debates and concepts - such as ‘post-Fordism’, ‘flexible specialisation’,
‘modes of regulation’, and so on - found in economic geography. The latter do
not chime easily with, or translate readily into, practical business and policy
strategy. In contrast, Porter’s explicit goal “is to develop both rigorous and
useful frameworks for understanding competition that effectively bridge the gap
between theory and practice” (1998a, p.2). Cluster theory, he argues, is “not only
a tool for managers, but also a microeconomic – based approach to economic
development for governments that is closely tied to actual competition” (op cit,
p. 7).
At the same time, in line with this goal, his easy ‘business- and policy-
friendly’ writing style, at once both accessible and common-sense, is undeniably
seductive, and is quite different from the more ‘academic’ discursive approach
that characterises much economic geography writing. Reinforcing this, there can
be little doubt that the popularity of Porter’s cluster concept, compared to
economic geographers’ work on similar notions, derives in large part from his
celebrated international profile as a business economist. This reputation,
combined with his self-confident, authoritative and proselytising style, lends his
cluster concept an apparent authenticity and legitimacy that policy-makers have
found difficult to resist. In contrast, economic geographers have had much less
influence on business policy: indeed the shaping of public policy has,

8
unfortunately, taken something of a back seat in the discipline’s research agenda
(Markusen, 1998; Martin, 2001).
But a third, and equally important reason for its rapturous reception is the
very nature of the ‘cluster concept’ itself. Porter’s cluster metaphor is highly
generic in character, being deliberately vague and sufficiently indeterminate as
to admit a very wide spectrum of industrial groupings and specialisations (from
footwear clusters to wine clusters to biotechnology clusters), demand-supply
linkages, factor conditions, institutional set-ups, and so on, while at the same
time claiming to be based on what are argued to be fundamental processes of
business strategy, industrial organisation and economic interaction. Rather than
being a model or theory to be rigorously tested and evaluated, the cluster idea has
instead become accepted largely on faith as a valid and meaningful ‘way of
thinking’ about the national economy, as a template or procedure with which to
decompose the economy into distinct industrial-geographic groupings for the
purposes of understanding and promoting competitiveness and innovation. The
very definitional incompleteness of the cluster concept has been an important
reason for its popularity (Perry, 1999): clusters have “the discreet charm of
obscure objects of desire” (Steiner, 1998, p. 1). However, although the
definitional and conceptual elasticity of the cluster concept can be seen as a
positive strength, in that it permits a wide range of cases and interpretations to
be included, we consider it to be problematic. The concept has acquired such a
variety of uses, connotations and meanings that it has, in many respects, become
a ‘chaotic concept’, in the sense of conflating and equating quite different types,
processes and spatial scales of economic localisation under a single, all-
embracing universalistic notion.

3. A Chaotic Concept?
A major source of ambiguity is that of definition. Because Porter’s definitions
are so vague, in term of geographical scale and internal socio-economic
dynamics, this has allowed different analysts use the idea in different ways to
suit their own purposes (see, for example, the multiplicity of interpretations used

9
in the World Congress on Local Clusters, OECD-DATAR, 2001). The result is
conceptual and empirical confusion.
The dramatist Alan Bennett tells the story of how his aged mother once
looked at sheep and said “I know what they are, but I don’t know what they’re
called”(Bennett 1994, p. 127). The situation in the cluster literature seems to be
the reverse: we know what they’re called, but defining precisely what they are is
much more difficult. In his own work, Porter has defined clusters as:

Geographic concentrations of interconnected companies,

specialised suppliers, service providers, firms in related industries,
and associated institutions (for example, universities, standards
agencies, and trade associations) in particular fields that compete
but also co-operate (Porter, 1998, p. 197).

Thus, there are two core elements in Porter’s definition. First, the firms in a
cluster must be linked in some way. Clusters are constituted by interconnected
companies and associated institutions linked by commonalities and
complementarities. The links are both vertical (buying and selling chains), and
horizontal (complementary products and services, the use of similar specialised
inputs, technologies or institutions, and other linkages). Moreover, most of these
linkages, he argues, involve social relationships or networks that produce
benefits for the firms involved. Hence,

A cluster is a form of network that occurs within a geographic

location, in which the proximity of firms and institutions ensures
certain forms of commonality and increases the frequency and
impact of interactions (1998a, p.226).

The second fundamental characteristic, therefore, is that clusters are

geographically proximate groups of interlinked companies. Co-location encourages
the formation of, and enhances the value-creating benefits arising from,
networks of interaction between firms.
The obvious problem raised by these cluster definitions is the lack of clear
boundaries, both industrial and geographical. At what level of industrial
aggregation should a cluster be defined, and what range of related or associated
industries and activities should be included? How strong do the linkages
between firms have to be? How economically specialised does a local

10
concentration of firms have to be to constitute a cluster? There is no explicit
reference in Porter’s definitions that clusters are economically specialised entities
in the Marshallian sense, yet all of his examples are, often very narrowly so. In
addition, at what spatial scale, and over what geographical range, do clustering
processes (inter-firm linkages, knowledge spillovers, rivalry, business and social
networks, and so on) operate? What spatial density of such firms and their
interactions defines a cluster? The difficulty is not just that the boundaries of
clusters, as Porter admits, are ‘continuously evolving’, as new firms and
industries emerge and established ones shrink or decline. More fundamentally,
the definition itself seems intentionally opaque and fuzzy.
Cluster boundaries, according to Porter (1998a, p.204), “rarely conform to
standard industrial classification systems, which fail to capture many important
actors in competition as well as linkages across industries …. Because parts of a
cluster often fall within different traditional industrial or service categories,
significant clusters may be obscured or even go unrecognised”. He refers to the
400-firm medical devices cluster in Massachusetts, which he says has long
remained all but invisible, buried within larger and overlapping standard
industry categories. In part then, defining the boundaries of clusters appears to
be about deriving a detailed reclassification of industries that more accurately
reflects the range of specialised economic activity. But then a cluster is also about
linkages within and between such specialised activities, about tracing the supply
chains supporting what is seen as the ‘core’ activity of the cluster. So as Porter
admits,

Drawing cluster boundaries is often a matter of degree, and involves

a creative process informed by understanding the most important
linkages and complementarities across industries and institutions
to competition (1998a, p. 202; emphasis added).

He suggests that “the strength of ‘spillovers’, and their importance to

productivity and innovation determine the ultimate boundaries”; that “cluster
boundaries should encompass all firms, industries and institutions with strong
linkages”, whereas ”those with weak and non-existent linkages can safely be left
out” (1998a, p.202). Exactly how the ‘strength’ of different sorts of linkages and
spillovers should be measured, and where the cut-off between ‘strong’ and

11
‘weak’ ties falls, are, however, issues that are left unspecified. The existence of
clusters, appears then, in part at least, to be in the eye of the beholder – or should
we say, creator.
And how does the requirement of ‘geographical proximity’ enter into the
equation? Although throughout his work on clusters Porter emphasises the
critical role of ‘geographical proximity’ in the formation, performance and
identification of clusters, the term is never defined with any precision. Indeed, it
appears to be highly and ridiculously elastic, for he suggests in fact that clusters
can be found at almost any level of spatial aggregation: “They are present in
large and small economies, in rural and urban areas, and at several geographic
levels (for example nations, states, metropolitan regions, and cities)” (1998a,
p.204); their geographical scope can even encompass “a network of neighbouring
countries” (1998a, p.199). To make matters worse, “the appropriate definition of
a cluster can differ in different locations, depending on the segments in which
the member companies compete and the strategies they employ” (1998a, p. 205).
Such geographical licence has given authors unlimited scope in their definition
and application of the concept (see Table 1).
At one extreme, the term has been used to refer to national groups of
industries and firms that are strongly linked (in terms of traded
interdependencies), but dispersed over several different locations within a
country, with no obvious major geographical concentrations (this was in fact
Porter’s original use of the ‘competitive diamond’). At the other extreme, the
term is used to refer to a local grouping of similar firms in related industries
within a highly spatially circumscribed area - such as the media cluster in Lower
Manhattan, New York (Porter, 1998, p. 205), or the film and media cluster in
Soho, London (Nachum and Keeble, 1999). In between, Porter refers to ‘regional
clusters’, such as the California agribusiness cluster, and the Massachusetts
medical devices cluster. He lists some 60 of these in the US (Porter, 1998.p.229),
although in most cases the clusters are far from being state-wide. Elsewhere,
‘clusters’ and ‘regions’ are used interchangeably (Baptista and Swann, 1998;
Enright 1996, 2001). Thus Enright (2001, p.2) claims that “Regional clustering is

12
 Table 1. Clusters: The Confusion of Definitions
(Some Examples Drawn from the Cluster Literature)

Porter (1998, p. 199) “A cluster is a geographically proximate group of interconnected companies

and associated institutions in a particular field, linked by commonalities and complementarities”.

Crouch and Farrell, (2001, p. 163) “The more general concept of ‘cluster’ suggests something looser:
a tendency for firms in similar types of business to locate close together, though without having a
particularly important presence in an area.”

Rosenfeld (1997, p. 4) “A cluster is very simply used to represent concentrations of firms that are
able to produce synergy because of their geographical proximity and interdependence, even though
their scale of employment may not be pronounced or prominent.”

Feser (1998, p. 26) “Economic clusters are not just related and supporting industries and institutions,
but rather related and supporting institutions that are more competitive by virtue of their
relationships.”

Swann and Prevezer (1996, p. 139) “Clusters are here defined as groups of firms within one
industry based in one geographical area.”

Swann and Prevezer (1998, p. 1) “A cluster means a large group of firms in related industries at a
particular location”.

Simmie and Sennett (1999a, p. 51) “We define an innovative cluster as a large number of
interconnected industrial and/or service companies having a high degree of collaboration, typically
through a supply chain, and operating under the same market conditions.”

Roelandt and den Hertag (1999, p.9) “Clusters can be characterised as networks of producers of
strongly interdependent firms (including specialised suppliers) linked each other in a value-adding
production chain.”

Van den Berg, Braun and van Winden (2001, p. 187) “The popular term cluster is most closely
related to this local or regional dimension of networks … Most definitions share the notion of
clusters as localised networks of specialised organisations, whose production processes are closely
linked through the exchange of goods, services and/or knowledge.”

Enright (1996, p. 191) “A regional cluster is an industrial cluster in which member firms are in close
proximity to each other.”

found in virtually every advanced economy and increasingly in developing

countries as well”. The confusion is further heightened by other studies that
equate clusters and ‘cities’ (see for example Swann, 1998, page 63). Still others
suggest, more specifically, that the proximity inherent in a cluster extends up to
a ‘range of fifty miles’ (May et al, 2001), but such demarcations are obviously
quite arbitrary.

13
 The problem is that geographical terminology is used in a quite cavalier
manner, depending it seems, as Porter himself admits, on what the aim of the
exercise is, or the client or policy-maker for whom the analysis is intended. The
key weakness is that there is nothing inherent in the concept itself to indicate its
spatial range or limits, or whether and in what ways different clustering
processes operate at different geographical scales. We are not suggesting that the
cluster concept should refer to a particular pre-specified geographical size or
scale; but to use the term to refer to any spatial scale is stretching the concept to
the limits of credulity, and assumes that ‘clustering processes’ are scale-
independent. If the same externalities and networks that typify clusters do
indeed operate at a whole variety of spatial scales, this surely weakens the
empirical and analytical significance of the cluster concept.
This lack of geographical precision and consensus is further compounded
by the vague typologies of cluster types and evolutionary paths that have been
proposed. Porter suggests that clusters “vary in size, breadth, and state of
development” (1998a, p. 204). Some clusters consist primarily of small and
medium sized firms (he cites the Italian footwear and North Carolina home
furniture clusters). Other clusters contain both small and large firms (he gives
the German chemical cluster as an example). There are university-centred
clusters and clusters with no university connections; clusters of traditional
industries and clusters of high-technology industries. There are nascent clusters,
new clusters, established clusters and declining clusters. Other authors have
sought to construct typologies based on the evolution of clustering processes.
Rosenfeld (1997), for example, distinguishes three types. ‘Working’ or
‘overachieving’ clusters are ‘self-aware’ and produce more than the sum of their
parts. Latent or ‘underachieving’ clusters present opportunities that have not yet
been fully exploited. ‘Potential’ clusters have some of the key conditions but lack
some inputs and critical mass. This latter type is particularly problematic, since it
becomes difficult to exclude almost any firm from a ‘potential’ cluster, especially
when policy-makers are eager not to be left out of the cluster promotion game
(indeed, many supposed clusters are ‘aspirational’ or the product of ‘wishful
thinking’). In practice, there are probably very few firms that do not have

14
horizontal or vertical links (co-operative or competitive) of some sort with other
loosely-defined ‘geographically proximate’ firms. Does this mean that virtually
every firm could be considered part of a ‘potential’ cluster? Typologies that
employ categories such as ‘embryonic’, ‘latent’, and ‘potential’, come close to
incorporating almost all firms in clusters of one type or another, and as such
become virtually meaningless. Equally problematic is the tendency to devise
typologies that relate specifically and only to the particular set of clusters being
studied, with little or no intention to discern elements or features that might be
of wider relevance.
The proliferation of cluster typologies may well be a genuine attempt to
recognise the diversity of cluster forms and cluster development. But appeals to
such diverse forms, sizes, stages of development, emergence, depth, breadth,
level of aggregation and the like, is equally an indication that the cluster concept
is something of a chaotic one. Porter sees the Italian industrial districts as one
form of cluster just as high-technology areas such as Silicon Valley are another.
What he calls clusters, French analysts refer to as ‘local production systems’ (see
OECD-DATAR, 2001). Recently, drawing on a survey of European examples,
Crouch et al (2001) see ‘empirical clusters’ as one of three types of local
production system, distinct from industrial districts on the one hand and what
they refer to as the ‘networked firm’ on the other. In contrast, in his discussion
of the role of regions in the recent competitive resurgence of the US economy,
Best (2001) uses the terms industrial districts and clusters interchangeably. And
so the confusion goes on. Classification is of course an important stage of
theorising and analysis. But to be meaningful and useful, typologies need to be
based on in-depth comparative analyses of cluster profiles and processes
(Markusen’s (1996) typology of industrial districts provides some pointers in this
regard). Despite the vast and still expanding literature on clusters, however,
there has been little detailed work of this kind.

4. What Sort of Theory for What Sort of Cluster?

All of which begs the issue of the status of ‘cluster theory’. Porter’s
‘clusters’ are constructs. They are as much analytical creations as they are

15
objectively real phenomena. They have no essential self-defining boundaries,
whether in terms of inter-sectoral, inter-firm linkages, information networks, or
geographical reach. The notion is so generic that it used as a sort of cover term
to refer to a whole assortment of types and degrees of specialised industrial
localisation (for example, see Porter, 2001). Little wonder, then, that cluster
theory is in a similar state of confusion.
According to Porter:

A variety of bodies of literature have in some respects recognized

and shed light on the phenomenon of clusters, including those on
growth poles and backward and forward linkages, agglomeration
economies, economic geography, urban and regional economics,
national innovation systems, regional science, industrial districts
and social networks. … Overall most past theories address aspects
of clusters or clusters of a particular type. (1998a,p. 207).

While eclecticism can be a virtue under certain circumstances, forging a

theoretical synthesis out of this list of perspectives would seem a dubious
endeavour. Yet this is what Porter tries to do. He sees his task as “embedding
clusters in a broader and dynamic theory of competition that encompasses both
cost and differentiation and both static efficiency and continuous improvement
and innovation, and that recognises a world of global factor and product
markets” (op cit, p.208 ). Porter’s theory of competition is not simply about cost
advantages and factor inputs, but also about ‘strategic positioning’ by
companies, that is choosing activities that are different from and superior to
those of rivals. Essentially, Porter sees a cluster as a self-reinforcing system that
stimulates the competitive strategies of the firms in the cluster and hence the
‘competitiveness’ of the cluster itself. He then argues that these processes
depend in part on personal relationships, face-to-face communication and social
networks (social capital) so that “cluster theory bridges network theory and
competition”(Porter, 1998, p. 226). He goes even further: “Clusters offer a new
way of exploring the mechanisms by which networks, social capital and civic
engagement affect competition” (op cit, p. 227). What is being proposed here,
therefore, is nothing less than a general theory of clusters and their socio-
economy.

16
 Yet three questions immediately arise. First, just how far can the full
complexity of economic, social and institutional factors and processes alleged to
underpin cluster formation, development, and success, be reduced to or
subsumed within an overarching concept of ‘competitiveness’? Second, to what
extent is it possible to construct a universal theory of cluster formation, dynamics
and evolution capable of covering the wide range of cluster types and processes
thought or argued to exist, without degenerating into superficial generalities of
the sort that have surrounded industrial districts (see Amin, 2000)? And third,
just how far does Porter’s cluster theory really illuminate the socio-insitutional
processes that are alleged to be so important in cluster formation and dynamics?
In Porter’s work the notion of ‘competitiveness’ is used to link a variety of
conceptual scales: the individual firm, the industry, the regional or local business
cluster, and the nation. For Porter, firms compete, clusters compete and nations
compete. He talks of the ‘competitiveness of locations’ (see Porter, 1998). At the
heart of Porter’s ‘theory of competitiveness’ is his longstanding idea of
‘competitive strategy’, which posits three generic strategies that companies must
follow to establish a lead in their market: differentiation (of product or service),
cost leadership, and focus strategy (focusing activities on the needs of specific
segments of the market). The role of clusters in this theory is that through
concentrating the interaction between the elements of the ‘competitive diamond’
they enhance all three aspects of strategy. But while clustering may well enhance
the competitiveness of firms, is this the same thing as talking about the
‘competitiveness of clusters or locations’? Locations obviously do not develop
competitive strategies in this sense (though many policy-makers seem to believe
they should).
In any case, Porter’s approach to competition and competitive strategy are
far from universally accepted within the business economics, industrial
organisation and management studies fields (O’Malley and Vanegeraat, 2000;
Jacobs and De Jong, 1992). Several authors have criticised Porters’ three generic
competitive strategies as being too superficial, for lacking specificity, for being
difficult to measure, and for not being as independent of one another, or as
universally applicable, as Porter assumes (Miller, 1992; Segev, 1997). As Buckley,

17
Pass and Prescott (1988) have shown, the notion of competitiveness is highly
complex and varies with the economic scale at which the concept is being used.
Indeed, yet other economists view the very notion of ‘competitiveness’ with
extreme scepticism. They argue that while the term may have meaning at the
level of the firm, it becomes increasingly more problematic as we move up the
scale of economic aggregation (see, for example, Krugman, 1994, 1996: Turner,
2001). According to these authors, nations and regions do not compete with one
another in the way that firms do, and the analogy between a company and a
nation or region is false.
The problem is reinforced, because Porter uses the terms ‘competitive
advantage’, ‘competition’ and ‘productivity’ interchangeably. In fact, recently he
has directly equated regional competitiveness with regional productivity (Porter,
2002). Certainly, productivity is a key index of regional and local economic
performance, and both the measurement and determinants of regional
productivity are important topics of enquiry. However, equating
competitiveness with productivity is to invite tautology and ontological
confusion: is a region more competitive because it is more productive, or is it
more productive because it is more competitive? Furthermore, a simple
productivity view of regional competitiveness obscures the complex nature of
competition itself. As Klein (2001) convincingly argues, Porter ‘s notion is highly
restrictive and fails to recognise the several different modalities that competition
can assume.
Yet a further complication is that Porter’s ‘competitiveness theory’ of
clusters is founded on the assumption that the important clusters are those
orientated to external trade. He estimates that these account for about 32 percent
of total US employment (Porter, 2002). This is consistent with the view of other
observers (such as Krugman, 1996, and Turner, 2001) that in today’s highly
urbanised world, the bulk of production serves local demand. Unfortunately,
Porter confusingly refers to this non-traded aspect of regional economies, which
he estimates makes up 67 percent of US total employment, as ‘local clusters’
(thereby implying that no less than 99 percent of the US economy is clustered!).
Not only is it misleading to use the term ‘local clusters’ to describe non tradable

18
local services and activities that are found in every urban area, it is also unclear
what is meant by competitiveness in relation to such activities. While the
productivity or more accurately the productivity growth, of local nontradable
activities is critical for local wealth and prosperity, as Krugman (1997) argues,
this has nothing to do with ‘competitiveness’.
If there are difficulties with clusters in Porter’s theory of competitiveness,
the generality of his ‘cluster theory’ adds to the problem. Clusters vary
considerably in type, origins, structure, organisation, dynamics, and
developmental trajectory, yet Porter’s theory is supposedly intended to fit all. It
is not clear, however, whether this is because it is assumed that all clusters can
be explained in the same way, despite their diversity, or because the highly
general nature of the theory is intended to cover all eventualities, allowing
analysts to pick and choose different elements to suit different types of cluster.
The difficulty is that Porter’s cluster model actually combines ideas from quite
different perspectives - from agglomeration theory to social network theory -
some of which are complementary and others much less so. As a result,
empirical observations of clusters and clustering can then be interpreted in quite
different ways, thereby buttressing a generalised notion of the benefits of
clustering by conflating elements for which there may actually be little evidence
with elements to which the evidence more directly relates.
In response to this shortcoming, Gordon and McCann (2000), argue for
distinguishing three main cluster models (theories). The first is the ‘pure
agglomeration economies’ model, which they trace from Marshall through to
modern urban economic theory, and which emphasises the external economies
of geographical concentration (see also, Bellflame, Picard and Thisse, 2000). The
second is what they call the ‘industrial complex’ model, in which clusters are
seen primarily as the spatial counterparts of the input-output models of regional
economics, as geographical concentrations forged by inter-firm trading links and
the minimisation of transactions costs. And the third is the ‘social-network’
model, which as the term suggests interprets clusters mainly in terms of strong
local networks of inter-personal relations, trust and institutionalised practices.
Gordon and McCann argue that

19
 Defining analytically which of these types is the dominant
structural characteristic of a particular cluster (or set of clusters) is
essential, in order to be able to discuss their performance
empirically, and to determine what more general lessons may be
drawn from that (2000, p. 515).

However, while helpful this tripartite theoretical schema is not

unproblematic. For one thing it fails to specify the particular circumstances,
economic and spatial, under which one theoretical model should be more
applicable than another. To be convincing, cluster theory ought to be able to
specify a priori how different sorts of cluster are likely to develop under different
conditions. Otherwise, explanation is reduced to a ‘best-fit’ exercise on a case by
case basis. Second, as Gordon and McCann themselves acknowledge, these three
theoretical models are ideal-types. By their very nature, ideal-typical models
never fit reality exactly, and in this case it is difficult to think of a pure
agglomeration economies cluster, a pure industrial complex cluster, or a pure
network cluster. While Gordon and McCann admit that a given cluster may
contain elements of more than one model, nevertheless they insist that

Contrasts in the policy implications of these three ideal types of

cluster make it particularly important to avoid confusing features
of one with those of another, even though elements of each may co-
exist in particular situations (2000, p. 528)

In reality, such co-existence is likely to be the rule. Indeed, what are social
networks in this context, other than a particular form of external economy
associated with agglomeration?

Porter argues that ‘social embeddedness’ – the existence of facilitative

social networks and social capital – is crucial for the successful functioning
and upgrading of clusters. Moreover,

…cluster theory also provides a way to connect theories of

networks, social capital and civic engagements more tightly to
business competition and economic prosperity… Cluster theory
helps to isolate the most beneficial forms of networks .. [and] may
reveal how network relationships form and how social capital is
acquired (Porter, 1998a, p. 227)

20
Despite this claim, however, the social dimensions of cluster formation and
cluster dynamics remain something of a black box in Porter’s work. While he
stresses, for example, the importance of local social networks for the production
and flow of information and knowledge within clusters, these processes are
conspicuously under-theorised in his cluster model. And even in his case study
examples, there is little explicit empirical investigation of these social and
knowledge networks, which more often than not are simply inferred from the
presence of particular formal and informal institutions within a cluster.
The problem of conceptualising and empirically analysing knowledge
networks and other ‘soft’ socio-cultural-institutional features of clusters and
spatial economic agglomerations is not, of course, confined to Porter’s work.
There is in fact an increasing tendency to explain cluster formation and
development in terms of local knowledge and ‘collective learning’ (see, for
example, Pinch and Henry, 1999; Hassink, 1997; Steiner and Hartman, 1998,
2001; Keeble and Wilkinson, 2000; Maskell, 2001). The argument is that in a
globalised economy the key resources for competitiveness depend on localised
processes of knowledge creation, in which people and firms learn about new
technology, learn to trust each other, and share and exchange information. The
emphasis is on the role of ‘tacit’ as against ‘codified’ knowledge, in that the
former is viewed as being especially dependent on localised face-to-face contacts
and spillovers. Indeed, the assumed link between localisation and tacit or
informal, uncodified knowledge is now almost accepted axiomatically (Breschi
and Lissonie, 2001). And according to Leamer and Storper (2001) in the new
information economy not only is the role of tacit knowledge increasing, this in
turn is accentuating the spatial agglomeration and localised specialisation of
economic activity.
This local knowledge ‘cluster theory’ itself faces several difficulties. First,
despite the numerous assertions that ‘tacit’ knowledge is the key to business
success, this remains an unsubstantiated and obscure proposition . Not only are
the distinctions between different forms of knowledge less clear cut and more
fluid than binary divisions such as formal and informal, codified and tacit,
suggest (Amin and Cohendetet, 1999; Breschi and Lissonie, 2001; Breschi and

21
Malerba, 2001), it is too simplistic to argue that a given form of knowledge is
inevitably linked to one form of geographical socio-economic organisation
(clusters) or any one scale of social relationships. Secondly, many accounts refer
to localised tacit knowledge without making clear precisely what it is, or how it
acts as a source of competitive advantage. More helpfully, Lawson and Lorenz
(1999) argue that the key form of tacit knowledge may actually be embedded in
firm routines, which guide a firm’s innovativeness, problem-solving and
adaptability. Ironically, however, the problem here is that the cluster literature,
including Porter’s own approach, lacks any serious analysis or theory of the
internal organization of business enterprise (Best and Forrant, 1996). Instead, it
emphasises the importance of factors external to firms and somehow residing in
the local environment. In too many accounts local ‘territorial learning’ is
privileged, yet what this process actually is remains ambiguous and its
interactions with firm-based learning are left completely unexamined (Hudson,
1999). Related to this, and as May et al (2001) rightly argue, given the current
fashion for non-economic explanations, cluster studies often assume that
‘institutional thickness’ refers to non-firm institutions rather than examining the
key institutions of firms and labour markets.
A further, and in our view fundamental, limitation of the current state of
‘cluster theory’ is that it abstracts clusters from the rest of the economic
landscape, so that they often appear as isolated and self contained entities
(Breschi and Malerba, 2001). Two things are missing. On the one hand, what is
needed is a cluster theory that situates cluster development within the dynamics
and evolution of industry and innovation more generally. In effect, Porter’s
approach is to delimit clusters and then to analyse them as if they are isolated
islands in the economy. Not all firms in a given sector of activity need be
clustered, and we should also consider the evolutionary trajectories and
interdependencies of firms outside clusters as well as those inside clusters. In
one of the very few attempts to address this issue, Pouder and St John (1996)
draw on a range of theoretical ideas (from institutional evolution, organisational
ecology, management cognition, as as well as standard agglomeration
economies) to construct an evolutionary model of the development of clustered

22
and non-clustered firms. They argue that the economies of agglomeration that
initially draw firms together into clusters eventually erode. The competitive
strategies of firms in clusters, which are initially highly innovative compared to
firms outside clusters, tend to converge (for example through mimetic and
normative isomorphism) and to be less innovative over time because cluster
firms define their field of competition as the cluster to which they belong, rather
than as the wider external industry. This restricted collective perspective gives
rise to competitive ‘blind spots’ which limit cluster firms’ innovative potential,
strategic positioning, and ability to anticipate and react to industry–wide shocks.
Non-clustered firms tend to be less constrained and potentially remain more
adaptable to sudden system-wide changes. In effect, these authors sketch out a
theory of cluster formation, growth and decline, set against the background of
the development of the wider industry as a whole. The very networks of
interdependence that were a source of strength in the early phase of cluster
formation and growth are hypothesised to become, over time, sources of inertia
and inflexibility, relative to the firms outside the clusters. While Porter does refer
to potential problems of cluster decline, Pouder and St John’s theory assumes
that relative if not absolute decline is an inherent systemic feature of cluster
dynamics.
By the same token, cluster theory provides a highly partial view of
regional development. Not only are clusters only one possible form or source of
regional economic growth, they tend to be analysed as if they are separate from
wider processes of regional development. Yet while their cluster theory
represents an advance over Porter’s in this respect, even their approach fails to
consider the dynamics of cluster formation and obsolescence within a more
holistic theory of uneven regional development. This is where, potentially,
economic geographers could make a significant contribution; but unfortunately
they seem to have all but abandoned their former interest in theorising the
development of the economic landscape as a whole in order to focus too on
particular types of region economy (invariably successful regional economies) in
isolation from the wider inter-regional system as a whole.

23
4. Selective Empirics and the Cluster Creation Game
Obviously, a vaguely defined and theorised concept does not lend itself to
easy or precise empirical delimitation. In fact, in most applications the
geographical mapping of clusters is surprisingly unsophisticated and stylistic.
Whilst Porter’s diagrammatic ‘flow diagrams’ of particular clusters of (up-stream
and down-stream) interlinked activities are often detailed, his ‘cluster maps’ are
extraordinarily simplistic and unexplained (for example, see his map of regional
clusters of competitive US industries in Porter, 1998a, p. 229). There is no agreed
method for identifying and mapping clusters, either in terms of the key variables
that should be measured or the procedures by which the geographical
boundaries of clusters should be determined. Hence different authors use
different types of data and different methods to identify them empirically, with
the result that varying claims are made for how many clusters exist and what
their geographies are. For example, while Porter (op cit.) identifies and maps
some sixty significant clusters in the US, according to the Secretary-General of
the OECD, the US contains no less than 380, producing some 60 percent of the
country’s output (Johnston, 2001). Yet again, while Porter identifies a mere
handful in the UK, others claim to identify several dozen (see Crouch and
Farrell, 2001).
Empirical methodologies and ‘mapping’ strategies vary considerably (see
Table 2). At one extreme are the ‘top–down’ national mapping exercises that
utilise selective types of data to identify, on an industry by industry basis,
particular important localisations of specialised activity or linked activities. At
the other extreme are ‘bottom -up’ approaches that are only concerned with
identifying clusters in a particular regional or local area, often in a highly
qualitative, impressionistic way. In between are all sorts of combinations. Even
top-down studies take different forms. Following Porter’s (1990) original major
work on national competitiveness, an initial stage in constructing clusters
adopted by some authors is to identify first those national ‘core’ industries that
are ‘globally competitive’, usually defined in terms each industry’s market share
of world exports, or of world value added. National input-output tables are then
constructed to determine the nature and extent of the trading linkages based

24
Table 2: Varieties of Cluster and the Cluster Measurement Problem
Cluster Concept Conceptual/ Empirical Ease of Empirical
Definitional Methodology Measurement Support
Depth
Co-location Shallow Top-Down Easy to Measure Indirect Evidence
(Quantitative)
Co-location and
Technological Proximity

Input-Output Table and

Complementarities

Co-location and Superior

Performance

Marshallian Externalities

Network
Firms

Explicit Collaboration

Informal Knowledge Deep Bottom Up Hard to Direct Evidence

Spillovers Measure
(Qualitative)

Adapted from Swann (2002)

around these ‘core’, globally competitive industries. Essentially, this approach to

defining ‘national industry clusters’ seeks to subdivide the economy by forming
industrial groupings (clusters) linked by particularly strong or distinctive
supply-demand transactions. In other accounts, attention is focussed on some
other subset of activities, whether they are globally competitive or not (e.g. high-
tech industries). In yet others, all of a nation’s industries are examined (see Miller
et al, 2001; Crouch et al, 2001). Since inter-industry trade data are rarely available
for sub-national geographical areas, in most top-down approaches the cluster
mapping exercise itself typically reduces to the mapping of regional or
subregional level data on employment and/or number of businesses, or value
added, in order to estimate the local significance of the industry clusters being
investigated (see Feser and Bergman, 2000).
The drawbacks of such analyses are obvious. Recall Porter’s arguments
that clusters typically cut across the sort of standard industrial sector
classifications used to collect employment, output and related business data. Yet
most top-down (and other) studies have no option but to use data based on such

25
classifications. To compound the problem, census type geographical data on
industrial employment and business populations are collected on the basis of
pre-given administrative and political units - such as metropolitan areas and
states in the US, and standard regions, local authority areas in the UK, or NUTS
regions in the EU - which may bear no close relationship with the geographical
boundaries or reach of clusters, however the latter are defined. Moreover, even
with a given national context, the size of such regional or local data collecting
units can vary substantially. If the spatial units are too large, they can overbound
and obscure local clusters. On the other hand, since the degree of local economic
differentiation and specialisation tends to increase as the size of geographical
units decreases, the use of small area data may exaggerate the number and
significance of clusters.
Given the problem of establishing the precise boundaries and
composition of clusters, many studies take an easier route. That is, they take
relatively large scale geographical units, such as states and regions, and make
the highly contestable assumption that sectoral employment totals for these units
provide a direct measure of the strength of cluster development in each location.
For example, Baptista and Swann (1998) examine the introduction of
manufacturing innovations across UK regions and conclude that a firm is more
likely to innovate if located in a region where the presence of firms in its own
industry is strong, as measured by employment, so that clustering furthers
innovation (see also Baptista, 2000; Beaudry et al, 1999). But to what extent can
the strength of regional employment in a sector be taken as evidence of the
existence of a cluster in that region? In response, Baptista and Swann (1998)
assert that agglomeration benefits and externalities become stronger when the
geographical level of analysis is reduced, so that using relatively large scale
regional data biases against the relevance of spillovers and externalities. This
may be so, but this surely does not mean that clusters necessarily exist in regions
where employment in any one sector is high. A high regional employment total
could surely just as easily reflect the presence of several large, dispersed and
unconnected employers within that region. There is also no agreement on what
degree of spatial concentration of an industry or industry group constitutes a

26
cluster. Many studies employ location quotients to measure relative spatial
concentration, and high values of location quotients are taken to indicate the
presence of clusters. But although these make some adjustment for the varying
size of areal units, they do not of themselves discriminate between the presence
of a large number of small or medium sized inter-linked firms, and a large single
firm employing the same overall number of workers. Associated data on the
geographical distribution of individual businesses by size and sector are clearly
essential. In addition, how much greater than unity does a location quotient
have to be to indicate the existence of a cluster (see Miller et al, 2001)? In some
studies, the statistical acrobatics employed to map ‘significant’ clusters are
complex. (see for example Ellison and Glaeser 1997). In other studies, the ‘rules’
used to distinguish clusters are highly arbitrary (for an example, see Crouch and
Farrell, 2001).
The extensive methodologies of top-down mapping exercises can at best
only suggest the existence and location of possible clusters: they provide a
shallow, indirect view of clusters . They can not provide much if any insight into
the nature and strength of local inter-firm linkages (traded and untraded),
knowledge spillovers, social networks and institutional support structures
argued to be the defining and distinctive features of clusters. Thus a common
tendency is to identify clusters in a piecemeal way and then deduce their benefits
and effects from the co-variation and co-location of selected variables. As
Hanson (2000, p. 481) explains, “The externalities that contribute to spatial
agglomeration, such as spillovers between workers, learning across firms, or cost
and demand linkages between local industries, are difficult to observe. We are
left to infer their existence from the covariance of observed variables such as
wages, employment and output.” This process of inference has produced some
very mixed and inconclusive results, and even when covariance is found it is
never precisely clear exactly what type of externality is responsible. For example,
there has been a long-running and unresolved debate on whether localisation
economies are conducive to higher productivity and employment growth, firm
entry and innovation; or whether in fact urbanisation economies are more
important and beneficial ( Henderson et al, 1995; Glaeser et al, 1992; Feldman,

27
2000). Indeed, while Porter believes that urbanisation economies are declining in
importance, others suggest that innovative clusters are driven by precisely these
economies (Simmie and Sennett, 1999).
In view of these difficulties, it is perhaps not surprising that many studies
give up on the idea of identifying clusters directly. Instead, they tend to rely on
loose ad hoc, ‘bottom up’ means of identification (Doeringer and Terkla, 1995).
Some analysts simply ask local economic agencies to supply lists of ‘local
clusters’ in their area which are then studied in more detail (for example see Van
Den Berg et al, 2000). In many of these instances, however, what are claimed to
be clusters often turn out, on closer empirical inspection, to be small and only
loosely connected collections of similar or related firms, and sometimes have
more to do with local policy aspirations than with realities on the ground.
Indeed, some cluster enthusiasts appear to eschew any prior empirical
identification of clusters at all, on the assumption that latent clusters are out
there everywhere if only their constituent businesses, institutions, and agencies
realised it. Thus, for example, according to Cluster Navigators Ltd (one of the
increasingly numerous cluster consultancies and promotional bodies):

Our experience is that extensive analysis is not required to identify

initial arenas [ie clusters] for collaborative engagement [ie between
cluster firms and policy-makers]. …The time for detailed analysis
and systematic cluster benchmarking is after initial engagement has
been obtained, not before (www.clusternavigator.com, emphasis
added).
This putting of the promotional cart before the analytic horse is understandably
attractive to cluster consultancies and public policy-makers eager to enter the
cluster promotion game. No doubt many (and perhaps Porter himself) would
argue that excessive (academic) analysis only leads to policy paralysis; that the
detailed structure and workings of a cluster will become obvious soon enough
once we begin to think about an activity in cluster terms. But to our minds, such
arguments require an a priori faith, and compound the difficulties surrounding
the cluster concept.
All of which renders many of the claims about the superior performance
of clusters of dubious validity. Clusters, it is argued, raise the productivity,
innovativeness, competitiveness, profitability and job creation of their

28
constituent firms, of the geographical areas in which the clusters are located, and
thence of the wider national economy. But what is the evidence for these claims?
Some advocates assert that the economic advantages of clusters have already
been empirically demonstrated (Baptista, 1998; 2000). Even the Secretary General
of the OECD states that it has been shown that being located in a cluster raises
the profitability of firms on average ‘between two and four percent’ (DATAR-
OECD, 2001, p. 8). A more detached review suggests that the evidence is
incomplete, and that far more detailed comparative research needs to be carried
out. Much of the evidence used in support of the superior performance
argument is anecdotal and based on success stories about particular locations
(Malmberg, 1996), and there are few extensive studies which document how
common and important clustering is within particular industries (Malmberg and
Maskell, 1997), or studies that carefully compare similar firms inside and outside
of clusters.
The evidence of a positive association between clustering and innovation
is not consistent. One of the few detailed studies that exist, of metalworking
across the US by Harrison and others (1996), actually found no evidence that
firms in local concentrations adopted new technologies more rapidly than their
more geographically dispersed or isolated counterparts. Likewise, in their study
of the impact of clusters on firm growth and innovation for a range of industries
across Europe, Beaudry, Brechi, and Swann (2000) found the results to be
ambivalent. While firms located in clusters that were strong in their own
industry tended to grow faster and have higher propensities to innovate, firms in
clusters that were strong in other industries did not have faster growth and
innovation rates. Moreover, it has not been conclusively shown that regions
based on specialised clusters consistently enjoy a higher rate of innovation and
economic growth (Segal Quince and Wicksteed, 2001; Steiner, 1998; Rodriguez-
Pose, 2001). At the very least, the case that clusters invariably boost business
performance and local economic development is not conclusively proven (Best,
2001).
Far from being the general rule and the key missing link in local
competitive advantage, the benefits realised from geographical clustering appear

29
to be specific to certain industries at certain stages of development in certain
places, and are only realised under particular conditions (Glasmeier, 2000). For
example, Audretsch and Feldman (1996) examined the distribution of
commercial innovative activity across the US and concluded that the propensity
to cluster is itself greatest in industries with a high dependence on new economic
knowledge, as captured by industry and university R and D, and skilled labour.
Indeed, the dominant view is now that clustering is most significant in sectors
that are crucially dependent on tacit or informal knowledge, often in pre-
commercialisation stages (Audretsch, 1998; Keeble and Wilkinson, 2000). But
even within high-technology and knowledge-based activities, the significance of
clustering has been found to be variable and produced by different processes.
Thus, studies of computing in the US and the UK find that employment growth
is promoted by own sector employment within particular states, and that firm
entry is positively associated with regional employment in only a few subsectors,
mainly hardware and components (Baptista and Swann, 1999). A similar study
of US biotechnology, however, concludes that firm employment growth has been
positively associated with own-sector employment but that firm entry in this
case is attracted by the strength of the science base more than by own sector
employment with a particular state (Swann and Prevezer, 1996; Prevezer 1997).
The common message from studies of biotechnology and aerospace in the UK is
that there are signs of clusters only in certain parts of these industries (Prevezer,
1997; Shohet, 1998; Beaudry; 2000). There is no evidence that firms in some
biotechnology sectors (for example related to food, chemicals, and agriculture)
attract each other, possibly because large firms are able to absorb and internalise
any knowledge spillovers. To complicate things further, a study of UK financial
services, using the same methodology, found that own sector employment has
promoted both employment growth and firm entry, while other-sector
employment negatively affects firm growth due to congestion effects (Pandit et
al, 1999).
Given that there are so many different ways of identifying clusters, their
advocates can always counter any disappointing results or criticism of their
findings by insisting that the cluster boundaries or their economic outcomes

30
have been incorrectly and inappropriately specified and measured. Ultimately, it
seems that it is impossible to definitively support or reject clusters with empirical
evidence, as there are so many ambiguities, identification problems, exceptions
and extraneous factors.

5. Cluster Policy: Hard Targets or Fashion Labels?

Despite these weaknesses, few other ideas can begin to rival the current
popularity of the clusters notion amongst local economic practitioners and
national and regional policy communities: “It is difficult to identify another
equally obscure concept that appeals to such a broad spectrum of academic
disciplines, professions and even lay people” (Bergman, 1998, p. 92). Cluster
policies fit in well with a growing trend towards the decentralisation of policy
responsibility and a focus on the indigenous potential of localities and regions
(Temple, 1998). Enright and Ffowcs-Williams (2001) argue that cluster policies
should be delivered by the level of government most closely matched to the
geographical extent of the cluster (assuming that this can be known). But given
that there is so little empirical work which conclusively demonstrates that
clustering actually produces increased local economic prosperity, this
extraordinary popularity begs explanation. As there is no agreed and shared
definition of what a cluster is, it is hardly surprising that there is no one single
model for such policy, and cluster labels are often attached to quite different
sorts of policies.
In terms of recommendations on best practice, however, there is a
consensus that cluster promotion policies are unlikely to succeed in creating
clusters ab initio (Schmitz and Nadvi, 1999). Rather, they should somehow
attempt to build on the potential already present in a particular economy. While
this seems sensible advice, it begs the question of what agencies should do if
they lack the basis of embryonic clusters. Such a question is not just hypothetical,
since it has particular relevance to debates on cluster promotion in lagging
peripheral regions in developed economies as well to a wide range of developing
countries. No convincing answers are given in the cluster literature. A typical
response is to argue that there are few if any regions that have no cluster
potential, however limited that might be.

31
 The standard rationale for cluster policies is they can help promote the
supply of those local and regional public goods which are absent due to market
failure (OECD, 1998; Scott, 1998). Four main varieties of such goods are usually
identified. First, cluster policy emphasises the benefits of creating co-operative
networks and encouraging dialogue between firms and other agencies. In such
networks, it is claimed, firms can exchange information, pool resources, design
collective solutions to shared problems and develop a stronger collective
identity. Thus some cluster policies start by appointing brokers and
intermediaries to organise these dialogues. This, it is suggested, can also yield a
better co-ordination both between public and private agents and between
different public agencies (Lagendijk and Charles, 1999). Second, and related to
this, cluster policies often involve collective marketing of an industrial specialism,
based on place marketing and raising awareness of the region’s industrial
strengths. Cluster policy typically represents a relatively cheap form of regional
policy, but it is one that is able to raise the public relations profile of particular
economies. Third, it is also argued that cluster policy should aim to provide local
services for firms such as financial advice, marketing and design services. A key
recommendation is that local service provision should be targeted on particular
industrial specialisms so as to ensure that it meets specific local needs. For
instance, a widespread aspiration for some high-tech initiatives is to develop
links with relevant university research facilities and further the
commercialisation of their research. Fourth, it is argued that cluster policies
should identify weaknesses in existing cluster value chains and attract investors
and businesses to fill those gaps and strengthen demand and supply links. In
some cases, marketing strategies should target particular types of investor who
will add key pieces to the cluster jig-saw.
It is by no means our intention to argue that all of these measures are, in
themselves, misguided and of no benefit to local and regional economies.
However, what is dubious is whether setting and attempting to implement such
policies within a cluster framework actually improves their effectiveness and
outcomes. In many cases it appears that the cluster framework is either
unnecessary or even constraining. The decentralised promotion of local
indigenous economic potential certainly does not depend on a cluster approach.

32
There are many types of network policy which promote information sharing
between firms which do not depend on a cluster framework and remit (see, for
example, Cooke and Morgan, 1998). Indeed as Rosenfeld (2001) argues, what has
happened in actual fact is that many local authorities have backed into cluster
initiatives from pre-existing network support programmes.
The first major problem inevitably encountered by such initiatives is how
their boundaries should be drawn. Which firms should be left out? How far up-
stream and down-stream of the ‘core’ cluster activity should policies extend?
There is a fundamental tension between the public policy desire to include as
many firms as possible and the notion that policy interventions can be more cost
effective if they are targeted in some way. But if the policy is too targeted, then it
starts to look like old style industrial policy and too close to the discredited
notion of ‘picking winners’. The tension is nicely captured in this ambiguous
advice by Enright and Ffowcs-Williams (2001, p. 5): “A policy on clusters should
aim to provide services that all firms merit access to, whether they are clustered
or not, but in a more targeted fashion.” In the case of generic services, which
would benefit all firms, it would seem preferable to drop the cluster framework
all together. Moreover, is it really wise to exclude certain firms from
‘institutional dialogues’, particularly when the future course of local industrial
and technological change is so hard to predict and previously marginal firms can
become key nodes in the local economy? Indeed, partly in response to this,
Danish industrial policy has shifted to using the less exclusive notion of
‘resource areas’ (see Drejer et al, 1999). The linkages in wider and more diverse
networks may well be weaker but they may provide greater long-run local
adaptability (Grabher, 1993).
What is clear is that, strictly speaking, most cluster policies do not identify
working clusters, but rely instead on more immediately and statistically visible
broad industrial sectors. Policy makers are clearly under pressure to find clusters
in as many regions as possible for fear of offending some regional interests. The
Sainsbury (1999) report on biotechnology clusters in the UK, for example, finds
established biotechnology clusters around Cambridge and Oxford, but also
identifies earlier clusters in the North West, Surrey, Sussex, Kent, the North East,
the North West, North Yorkshire, Wales, London, Central Scotland, Wales; that

33
is, in almost all regions of the country, even though in many regions the
‘clusters’ are small and lack the inter-firm linkages and spillovers that are held to
be key cluster features. Likewise, in the DTI’s more recent general investigation
of business clusters across the UK (Miller et al, 2001) there is an obvious political
tension between mapping significant industry clusters wherever these happen to
be on the one hand (and many are in South East England), and ensuring an even
spread between the various Regional Development Agency areas, on the other.
It is hard to avoid noticing the similarity in the types of clusters
distinguished by many cluster promotion programme (Rosenfeld, 2001). To
some extent this no doubt reflects the enormous practical difficulty of identifying
working clusters and the inevitable ambiguities and complexities involved.
Faced with this difficulty, many public authorities resort to using the same set of
cluster consultancy companies (flying ‘cluster-makers’), who are commissioned,
often at considerable public expense, to rapidly produce a cluster decomposition
of the relevant local or regional economy. We are doubtful that such quick
diagnoses are able to identify weak links in local value chains, understand the
relevant spillovers and knowledge flows, detect how different industries are
developing and anticipate the necessary service requirements. More generally,
how should public authorities distinguish market failures in service provision? If
firms are not paying for particular services, does this indicate a lack of
information or a shared fear of free-riders; or is it because the services are not
really needed? Another important ambiguity in the literature is whether policy
should aim to differentiate between different clusters and, if so, in what ways.
On the one hand, Porter (1995; 1996; 1998a) recommends that policy makers
should not try to discriminate between clusters; rather, he suggests that ‘it is not
what you do but how you do it’ that matters, and favouring some clusters is
“bad economics”. On the other hand, as we have already noted, Porter (1998a)
argues in a neo-Keynesian manner that it is outward-oriented (export-based)
clusters, rather than those supplying local demand, that are the primary long-run
source of economic growth and prosperity. Other authors insist that policy
makers should distinguish between clusters according to their growth potential
(Fisher and Reuben, 2000), so that responses to survival clusters of micro-scale
enterprises with limited competitive potential should be distinctive (Altenburg

34
and Meyer-Stamer, 1999). A related issue is whether inter-firm collaboration is
appropriate in all industries, and to what extent policies can promote both
competition and rivalry, at the same time as furthering firm collaboration (see
Enright. 1996). How are public cluster sponsors supposed to know when, and on
which issues, to encourage knowledge sharing and exchange, and when to urge
knowledge retention and proprietary secrecy in order to sustain innovation
incentives?
In the context of these uncertainties and questions, it is increasingly
evident that policy makers, seeking safety in numbers perhaps, tend to be drawn
to promoting similar varieties of ‘high-technology, knowledge-based’ clusters.
But, even assuming that the new knowledge economy has a coherent meaning, it
is unlikely that all regions can rely on the same, high-skill, knowledge-intensive
sectors. As Keep and Mayhew (1999, pp. 57-8) have argued

While it is relatively straightforward to aspire to the high

performance vision when it is applied to some sectors of the
economy, it becomes a very much greater challenge when the focus
shifts to the economy as a whole. Policy-makers need to guard
against the dangerous tendency to seize upon leading edge practice
located within a particular sector or competitive environment and
then to assume that this can be, or indeed is being generalised
across the entire range of economic activity.

Nevertheless, such generalisation or replication is precisely what Porter

appears to be currently advocating. His present cluster mapping project across
the US is expressly aimed at identifying ‘best practice’ in successful clusters as a
‘blueprint’ for promoting innovativeness and competitiveness across regional
America as a whole. While successful clusters may well hold lessons for policy
practice elsewhere, the idea that there are cluster ‘blueprints’ that can be readily
implemented in quite different local economic, social and institutional contexts is
highly debatable. Cluster policy texts also say little about how inter-regional,
distribution issues should be approached. If one region implements a new
cluster strategy should this be allowed to undermine established clusters
elsewhere or should regional clustering somehow be made complementary (DTI,
2001)? Porter (1996) argues that regional authorities should avoid “smokestack
chasing” as using tax incentives and subsidies to bid against each other to attract

35
key inward investors produces a zero-sum competition. Yet, in practice, cluster
policies are surely likely to encourage just such a process.
What is striking is that in much of the literature on cluster policies, there
is no real reason why place marketing and the advertising of industrial
specialisms really needs to be tied to a ‘cluster’ label of doubtful relevance and
content. There is no reason why co-ordination between different policies and
groups should be handicapped with a confusing cluster framework, or why the
provision of demand-led, productivity enhancing services to firms would be
improved by setting up some imagined cluster boundaries. Moreover, there are
also several potential dangers associated with promoting clusters (see Table 3).
First, cluster policy may sponsor an exaggerated view of the extent to which firm
performance is determined by local context. For instance, Porter (1998a) claims,
“The presence of clusters suggests that much of competitive advantage lies
outside a given company or even outside its industry, residing instead in the
locations of its business units”(p. 198, see also Steiner, 1998, p. 4). Internal and
external advantages are clearly not independent, but if a company suffers from
poor management, culture and practices it is hard to believe that it can rely on
the competitive advantage of its location. While supporting institutions and a
networked semi-public sphere may often be necessary for innovative and
dynamic firm performance, such factors are unlikely to be sufficient.
Local and regional specialisation also represents a risky strategy (Perry,
1999). The risk of decline and profound instability in specialised regional
economies is well known and its relevance has been underlined by the recent
downturn in Silicon Valley (. Economic landscapes are littered with local areas of
industrial specialisation that were once prosperous and dynamic but have since
gone into relative or even absolute decline. Porter himself argues that the causes
of decline can be both internal to the cluster and the result of radical changes in
external conditions, such as technological discontinuities. Internal decline will be
rapid if the cluster suffers from the lock-in of established ways of thinking and
doing things, or technological isomorphism (through the normative or mimetic
behaviour of firms) occurs. And there are some authors who suggest that a that
a reliance on local face-to face contact and tacit knowledge do indeed make local
networks of industry especially vulnerable to lock-in (Amin and Cohendet,

36
 Table 23: Clusters have Costs as well as Benefits
Claimed Advantages Potential Disadvantages
Higher innovation Technological isomorphism
Highergrowth
Higher innovation Technological
Labour isomorphism
cost inflation
Higher productivity Inflation of land and housing costs
Increased profitability Widening of income disparities
Increased competitiveness Over-specialisation
Higher new firm formation Institutional and industrial lock-in
High job growth Local congestion
and environmental pressure

1999). Clustered firms may be capable of incremental and continuous

improvement within certain parameters, but may be unable to adapt to radical
shifts in technology and product (see Pouder and St John, 1996; Loasby, 1998).
There is clearly a danger that by encouraging highly specialised localisations of
industry, cluster policies may actually reduce innovation rates. The evidence
from urban economics suggests that, for many industrial sectors, innovation is
associated with location in a diversified urban base. As Duranton and Puga
(2000: 553) write “The link between innovation and diversity seems fairly robust,
so that highly innovative clusters cannot be bred in previously specialised
environments.” None of the swelling ranks of cluster adherents seem to explain
how policies should respond to these dangers. Fritz et al (1998) argue that policy
makers should see local industrial specialisation as having a ‘risk-return trade
off’, plotting the risk of ossification against the higher returns gained from
clustering (a point made some time ago by Conroy, 1975, in his study of regional
specialisation versus regional diversification). In practice, constructing such
‘trade-offs’ is obviously very far from straightforward.
A further risk related to clusters concerns that of localised inflation and
‘overheating’. The cluster literature itself tends to downplay the importance of
cost-based competition. To quote Steiner (1998, p. 4): “The existence of clusters is
the decisive element for the competitiveness of regions and nations, not cheap
land, labour, or energy, nor even high subsidies and low social costs, nor even
high technology strong and leading industries” . Other more cautious sources

37
express concern about the effects of clusters on local costs. For example, the
DETR (2000) notes that the growth of industrial concentrations tightens the
labour market, leads to increased congestion and puts pressure on the housing
stock, destroying the features conducive to development. It adds, “Firms with
lower margins may be forced out of the area and workers on lower incomes,
perhaps working in essential services or sectors which support the cluster, find it
more difficult to find affordable housing” (ibid.).
The question of the impact of cluster growth on other local sectors of
activity is unresolved. The cluster orthodoxy would presumably be that as costs
rise in the cluster, less productive firms are either put out of business or move
away. In Porter’s (1998a, p. 245). view “Rising local wages and profits reflect
economic success. This means that less skilled and less productive activities
should move to other locations” (his emphasis) However, if these firms are labour
intensive this would have a large negative impact on the local labour market.
There is no theoretical guarantee that the high-productivity growth firms will be
able to absorb the excess labour. Those working within the ‘core’ cluster firms in
a locality may enjoy high living standards and rising wages as a compensation
for the growing congestion, while those in non-core activities have to make do
with inferior real wages and living standards, or have to move away. It cannot,
therefore be assumed that the promotion of one or several clusters within a
regional economy leads to balanced economic development, or greater
competitiveness, or greater well-being across the entire region. Rather the
outcome will depend on how the cluster affects the costs and employment of
others sectors and localities (Venables, 1996).
Given these potential disadvantages, it would seem more advisable for
local and regional authorities to concentrate on encouraging productivity
improvements in all local firms, as well as improving their business
environments, without necessarily committing to a cluster mind-set. The danger
of a cluster-based approach to policy is that it detracts from the need to take a
more holistic view of regional development. It is likely that dynamic regions
will produce networks even without government incentives (Rosenfeld, 2001).
Furthermore, even cluster enthusiasts find it enormously difficult to point to any
examples of deliberate cluster promotion programmes that have been

38
unambiguously successful. Given the consensus that public programmes that
attempt to directly create and steer clusters are likely to be ineffective, some
argue that cluster policies should concentrate on accommodating the formation
of new firms and investing in education and support infrastructure (Breschi and
Malerba, 2001). But, once again , there are no clear grounds for tying such
measures to confusing cluster frameworks As Porter (1998a) himself rightly
emphasises, many of the most significant influences on industrial development
stem from the way in which national regulatory frameworks influence the
demand for sophisticated products, the course of industrial innovation, and
levels of entrepreneurship (see also Miller et al, 2001). Such regulation, along
with the quality of the economic and social infrastructure, may well represent a
better focus for policy makers’ attention. At its best the current policy
preoccupation with cluster strategies looks like a fad for a fairly imprecise and
flexible label for differing combinations of measures. Like many fashion labels
which are not accompanied by a high quality product, the cluster approach
could go out of fashion as quickly as its popularity has mushroomed.

6. Conclusions: The Cluster Brand?

Constructing a critical review of clusters is a difficult task. There are now
so many different varieties of clusters and so many confusing claims about their
theoretical basis, form, identification and significance that the concept is
peculiarly elusive and hard to pin down. The feeling that there must be ‘more to
it than this’ is endemic. The cluster literature is a patchy constellation of ideas,
some of which are clearly important to contemporary economic development
and some of which are either banal or misleading. But there are two key
limitations which we wish to emphasise in conclusion. First, a concept so elastic
as the cluster can not provide a universal and deterministic model on how
agglomeration is related to regional and local economic growth. At present the
siren of universalism is pulling the cluster concept into shallow waters. It is
being applied so widely that its explanation of causality and determination
becomes overly stretched, thin and fractured. Second, and related to this,
economic geographers and other regional analysts have long been aware that
just because there is an association between some high-growth industries and

39
various forms of geographical concentration does not mean that this
concentration is the main cause of their economic growth or relative success. The
empirical case for clustering remains in its infancy and repeatedly makes the
mistake of jumping from particular associations to general causality and
applicability.
But this heterogeneity and chaos is only half the story. It would be
tempting to conclude that the notion of clusters has no real significance. Yet this
is clearly at odds with the enormous policy popularity of the notion and the
generous tolerance granted to the idea by a usually critical academic community.
The answer to this paradox may lie in the way in which the cluster concept has
been marketed by Porter and other enthusiasts as a brand, rather than just another
intellectual product. Just as large corporations use branding to distinguish their
products from other largely indistinguishable products (for example see Klein,
2000), so the ‘cluster brand’ has been much more successful despite the existence
of many similar theories and policy recommendations of industrial
agglomeration. It has certainly attracted many more ‘buyers’ in the policy and
practitioner markets than its rivals. The reason appears to lie not so much in the
theoretical or empirical superiority of the concept, but in the way in which it has
been closely tied to a set of positive images and associations. Again, just as
branded products are ultimately image-based so that consumers come to
associate them with rewarding lifestyle experiences, the cluster brand at its core
is based on am image of a high-productivity, knowledge-rich, decentralised,
entrepreneurial and socially progressive economy within the reach of local
policy-markers (a regional version of the American Dream, perhaps?). In this
way, it conforms with, and reinforces, the ideological appeal of the “new
regionalism” (see Lovering, 1999). And just as brands are not confined to
particular products but can exploit synergies between them, so we should expect
the cluster idea to act as an umbrella brand for many different things. The core
meaning of clusters lies more in this image than in a coherent and carefully
defined set of ideas.
In short, Porter’s cluster idea displays all of the key features needed for
such a metaphor to assume the power of a successful ‘brand’ (or even myth – see

40
Harfield, 1998). First, the metaphor must accord with some strong, if not always
clearly defined, aspirations – in this case promoting innovation and
competitiveness. It must be expressed in language sufficiently flexible as to
permit a wide range of interpretations – in this instance, the hybrid nature of the
cluster concept. The metaphor must have authority - here Porter’s ‘expert
knowledge’ on competition and business strategy. It must be capable of
continual and consistent re-invention and re-application – Porter’s cluster
concept is itself the latest stage in his evolving theory of competitive advantage,
now being actively applied to the latest phase of economic development , the so-
called new knowledge economy. And the language of the metaphor must allow
the possibility of providing practical action – the cluster as a policy tool.
It is as if, in effect, Porter has applied his theory of competitive strategy –
of ‘strategic positioning’ – to his cluster idea itself. Clever positioning and
marketing of the cluster idea have been extremely influential in selling it to
policy-makers the world over. In adopting the cluster idea, policy-makers
purchase the ‘Porter brand’, and in doing so serve to reinforce the brand’s
prominence. What this implies, of course, is that given the power of the ‘cluster
brand’, academic critiques such as this are unlikely to have much of an impact on
the concept’s popularity. It is perhaps only as the actual limits to ‘brand-based
cluster policy making’ emerge, along with the marshalling of careful evaluative
research, that the grip that Porter’s cluster idea currently exerts on analytical and
policy circles will lessen. As Santayana reminds us (in the quote at the beginning
of this paper), fashionable ideas tend to share one thing in common: they all
eventually become unfashionable,

41
Bibliography

Altenberg, T. and Meyer-Stamer, J.(1999) How to Promote Clusters: Policy

Experiences from Latin America, World Development, 27, 9, pp. 1693-
1713.

Amin, A. (2000) Industrial Districts, Ch 10 in Sheppard, E. and Barnes, T.

(eds) A Companion to Economic Geography Oxford: Blackwell, pp. 149-168.

Amin, A. and Cohendet, P. (1999) Learning and Adaptation in Decentralised

Business Networks, Environment and Planning D: Society and Space 17,
pp. 87-104.

Amin, A. and Thrift, N.J. (1992) Neo-Marshallian Nodes in Global Networks,

International Journal of Urban and Regional research, 16, pp. 571-587.

Asheim, B. (2000) Industrial Districts: The Contributions of Marshall and

Beyond, Ch 21 in Clark, G. L., Feldman, M. and Gertler, M. (Eds) The
Oxford Handbook of Economic Geography, Oxford: Oxford University
Press, pp. 413-431.

Audretsch, D. (1998) Agglomeration and the Location of Innovative Acitivity,

Oxford Review of Economic Policy, 14, 2, pp. 18-29.

Audretsch, D. and Feldman, M. (1996) R&D Spillovers and the Geography of

Innovation and production, The American Economic Review, 86, 3, pp.
630- 640.

Baptista, R. (1998) Clusters, Innovation and Growth: A Survey of the

Literature, in Swann, G.M.P., Prevezer, M. and Stout, D. (Eds) The
Dynamics of Industrial Clustering: International Comparisons in Computing
and Biotechnology, Oxford: Oxford University Press, pp. 13-51.

Baptista, R. (2000) Do Innovations Diffuse Faster within Geographical

Clusters? International Journal of Industrial Organization, 18, pp. 515-535.

Baptista, R. and Swann, P. (1998) Do Firms in Clusters Innovate More?

Research Policy, 27, pp. 525-540.

Baptista, R. and Swann, P. (1999) A Comparison of Clustering Dynamics in

US and UK Computer Industries, Journal of Evolutionary Economics, 9,
pp. 373-399.

Beaudry C. (2000) Entry, Growth and Patenting in Industrial Clusters: A

Study of the Aerospace Industry in the UK, Manchester Business
School WP 413.

42
Beaudry, C., Breschi, S. and Swann, P. (2000) Clusters, Innovation and
Growth: A Comparative Study of European Countries, Working Paper,
Manchester Business School.

Bennett, A. (1994) Writing Home, London: Faber.

Bergman, E.M. (1998) Industrial Trade Clusters in Action: Seeing Regional

Economies Whole, in Steiner, M. (Ed.) Clusters and Regional
Specialisation: On Geography, Technology and Networks, London: Pion, pp.
92-110.

Best, M. and Forrant, R. (1996) Creating Industrial Capacity: Pentagon-led

Versus Production-led Industrial Policies, in J. Michie and J. Grieve
Smith (Eds.) Creating Industrial Capacity: Towards Full Employment,
Oxford: Oxford University Press, pp ????.

Best, M. (2001) The New Competitive Advantage: The Renewal of American

Industry, Oxford: OUP.

Breschi, S. and Lissoni, F. (2001) Localised Knowledge Spillovers versus

Innovative Milieux: Knowledge ‘Tacitness’ Reconsidered, Papers in
Regional Science, 80, pp. 255-273.

Breschi, S. and Malerba, F. (2001) The Geography of Innovation and

Economic Clustering: Some Introductory Notes, Industrial and Corporate
Change, 10, 4, pp. 817-833.

Brown, R. (2000) Clusters, Supply Chains and Local Embeddeness in Fyrstad,

European Urban and Regional Studies, 7, 4, pp. 291-306.

Buckley, P.J., Pass, C. and Prescott, K. (1988) Measures of International

Competitiveness: A Critical Survey, Journal of Marketing Management, 4,
pp. 175-200.

Cairncross, F. (1997) The Death of Distance, London: Orion Business Books.

Cohen, S. and Fields, G. (1999) Social Capital and Capital Gains in Silicon
Valley, California Management Review, 41, 2, pp. 108-130.

Conroy, M. (1975) Regional Economic Diversification, New York: Praeger.

Cooke, P. and Morgan K. (1998) The Associational Economy, Oxford: Oxford

University Press.

Council on Competitiveness (2001) US Competitiveness 2001: Strengths,

Vulnerability and Long-Term Priorities (www.compete.org)

Coyle, D. (1997) The Weightless World: Strategies for Managing the Digital
Economy, London: Capstone.

43
Coyle, D. (2001) Paradoxes of Prosperity: Why the New Capitalism Benefits All,
London: Texere Publishing.

Crouch, C. and Farrell, H. (2001) Great Britain: Falling through the holes in
the Network Concept, pp. 161-211 in Crouch, C., Le Galés, P., Trogilia,
C. and Voelzkow, H. (2001) Local Production System in Europe: Rise or
Demise? Oxford: Oxford University Press.

Crouch, C., Le Galés, P., Trogilia, C. and Voelzkow, H. (2001) Local Production
System in Europe: Rise or Demise? Oxford: Oxford University Press.

DETR (2000) Planning for Clusters, London: HMSO.

Doeringer P.B. and Terka, D.G. (1996) Why Do Industries Cluster? In Staber,
U., Schaefer, N. and Sharma, B., (Eds.) Business Networks: Prospects for
Regional Development, Berlin: Walter de Gruyter, pp. 175-189.

Drejer, I. Kristensen, F. and Laursen, K. (1999) Cluster Studies as a Basis for

Industrial Policy: The Case of Denmark, Industry and Innovation, 6,2, pp.
171-190.

Duranton, G. and Puga, D. (2000) Diversity and Specialisation in Cities: Why,

Where and When Does it Matter? Urban Studies 37,3, pp. 533-555.

Ellison, G. and Glaeser, E.L. (1997) Geographic Concentration in US

Manufacturing Industries: A Dartboard Approach, The Journal of
Political Economy, 105, pp. 889-927.

Enright, M. (1996) Regional Clusters and Economic Development: A Research

Agenda, in Staber, U., Schaefer, N. and Sharma, B., (Eds.) Business
Networks: Prospects for Regional Development, Berlin: Walter de Gruyter,
pp. 190- 213.

Enright, M. (2001) Regional Clusters: What We Know and What We Should Know,
Paper presented at the Kiel Institute International Workshop on
Innovation Clusters and Interregional Competition, 12-13 November.

Enright, M. and Ffowcs-Williams, I. (2001) Local Partnership, Clusters and

SME Globalisation, OECD Workshop paper, www.oecd.org.

Feldman, M. (2000) Location and Innovation: The New Economic Geography

of Innovation, Spillovers, and Agglomeration, in Clark, G.L., Feldman,
M. and Gertler, M. (Eds.) Oxford Handbook of Economic Geography,
Oxford: Oxford University Press, pp. 373-394.

Feser, E.J. (1998) Old and New Theories of Industry Clusters, in Steiner, M.
(1998) (Ed.) Clusters and Regional Specialisation: On Geography, Technology
and Networks, London: Pion, pp. 18-40.

44
Feser, E.J. and Bergman, E.M. (2000) National Industry Cluster Templates: A
Framework for Regional Cluster Analysis, Regional Studies, 34, 1, pp. 1-
20.

Fisher, E. and Reuben, R. (2000) Industrial Clusters and SME Promotion in

Developing Countries, Commonwealth Trade Enterprise Paper Number
3, London: Commonwealth Secretariat.

Fritz , O., Mahringer, H. and Valdenama, M. (1998) A Risk-oriented Analysis

of Regional Clusters, pp. 181-191 in M.Steiner (ed.) Clusters and Regional
Specialisation: On Geography, Technology and Networks, London: Pion.

Fujita, M. Krugman, P. and Venables, A. (2000) The Spatial Economy: Cities

Regions and International Trade, Cambridge, Mass: MIT Press.

Glaeser, E., Kallal, H., Scheinkman and Shleifer, A. (1992) Growth in Cities,
Journal of Political Economy, 100, 6, pp. 1126- 1152.

Glasmeier, A. (2000) Economic Geography in Practice: Local Economic

Development Policy, pages 559-579 in Clark, G., Feldman, M. and
Gertler, M. (Eds.) The Oxford Handbook of Economic Geography, Oxford:
Oxford University Press.

Gordon, I.R. and McCann, P. (2000) Industrial Clusters: Complexes,

Agglomeration and/or Social Networks? Urban Studies, 37, 3, pp. 513-
532.

Grabher, G. (1993) The Weakness of Strong Ties: The Lock-in of Regional

Development in the Ruhr Area, pp. 255-277 in The Embedded Firm: On
the Socio-economics of Industrial Networks, London: Routledge.

Gray, J. (1998) False Dawn: The Delusions of Global Capitalism, London: Granta
Books

Hanson, G. (2000) Firms, Workers, and the Geographic Concentration of

Economic Activity, in Clark, G.L., Feldman, M. and Gertler, M. (Eds.)
Oxford Handbook of Economic Geography, Oxford: Oxford University
Press, pp. 477-494.

Harfield, T (1998) Strategic Management and Michael Porter: A Postmodern

Reading, Electronic Journal of Radical Organisation, 4, 1. (

Harrison, B. (1992) Industrial Districts: Old Wine in New Bottles? Regional

Studies, 26, pp. 469-483.

Harrison, B, Kelley, M. and Gant, J. (1996) Innovative Firm Behaviour and

Local Milieu: Exploring the Intersection of Agglomeration, Firm Effects,
and Technological Change, Economic Geography, 72, pp. 233- 258.

45
Hassink, R. (1997) Localised Industrial Learning and Innovation Policies,
European Planning Studies, 5, pp. 279-282.

Held, J. R. (1996) Clusters as an Economic Development Tool: Beyond the

Pitfalls, Economic Development Quarterly, 10, 3, pp. 249-261.

Henderson, V., Kuncoro, A. and Turner, M. (1995) Industrial Development in

Cities, Journal of Political Economy, 103, 5, pp. 1067-1090.

Hill, J. and Naroff, J.L. (1984) The Effect of Location on the Performance of
High Technology Firms, Financial Management, Spring, pp. 27-36.

Hudson, R. (1999) ‘The Learning Economy, the Learning Firm and the
Learning Region’: A Sympathetic Critique of the Limits to Learning,
European Urban and Regional Studies, 6, pp. 59-72.

Jacobs, D. and De Jong, M. (1992) Industrial Clusters and the

Competitiveness of the Netherlands, De Economist, 140, 2, pp. 233-252.

Johnston, D. (2001) Opening Remarks in World Congress on Local Clusters,

OECD-DATAR, Paris, pp. 7-9.

Keeble, D. and Wilkinson, F. (2000)(Eds.) High-Technology Clusters, Networking

and Collective Learning in Europe, Aldershot: Ashgate.

Keeble, D.E and Nachum, L. (2002) Why do Business Service Firms Cluster?
Small Consultancies, Clustering and Decentralisation in London and
Southern England, Transactions of the Institute of British Geographers,
(forthcoming).

Keep, E. and Mayhew, K. (1999) Towards the Knowledge-driven Economy:

Some Policy Issues, Renewal, 7, 4, pp. 50-85.

Klein, J. (2001) A Critique of Competitive Advantage, Paper presented at the

Critical Management Studies Conference, University of Manchester,
July 2001.

Klein, N. (2000) No Logo, London: Flamingo.

Krugman, P. (1991) Geography and Trade, Cambridge, Mass: MIT Press.

Krugman, P. (1994) Productivity and Competitiveness, Appendix to Chapter

10, in Peddling Prosperity, New York, W.W. Norton, pp. 268-280.

Krugman, P. (1996) Competitiveness: A Dangerous Obsession, Chapter 1,

and Myths and Realities of US Competitiveness, Chapter 6, in Pop
Internationalism, Cambridge, Mass: MIT Press, pp, 3-24 and pp. 87-104.

46
Krugman, P. (1996) Pop Internationalism, Cambridge, Mass: MIT Press,

Lagendijk, A. and Charles, D. (1999) Clustering as a New Growth Strategy for

Regional Economies? A Discussion of New Forms of Regional Industrial
Policy in the UK, in OECD (1999) Boosting Innovation: The Cluster
Approach, Paris: OECD, pp. 127-153.

Lawson, C. and Lorenz, E. (1999) Collective Learning, Tacit Knowledge and

Regional Innovative Capacity, Regional Studies, 33, 4, pages 305-317.

Lazerson, M. and Lorenzoni, G. (1999) The Firms that Feed Industrial

Districts: A Return to the Italian Source, Industrial and Corporate Change,
Vol. 8, Number 2, pp. 235-266.

Leamer, E. and Storper, M. (2001) The Economic Geography of the Internet

Age, Working Paper 8450, Washington: National Bureau of Economic
Research.

Loasby, B. (1998) Industrial Districts as Knowledge Communities, in M.

Bellet and C. L’Harmet (eds) Industry, Space and Competition: The
Contributions of Economists of the Past, Cheltenham: Edward Elgar, pp.
70-85.

Lovering, J. (1999) Theory led by Policy? The Inadequacies of the ‘New

Regionalism’, International Journal of Urban and Economic Research, 23, pp.
379-395.

Malmberg, A., (1996) Industrial Geography: Agglomeration and Local Milieu,

Progress in Human Geography, 20, 3, pp. 392-403.

Malmberg, A. (2002) Why the Cluster is Causing Continuing Confusion – Despite

Being Potentially a Core Concept in Economic Geography, Paper presented
at the Cluster Workshop, University of Lund, Sweden, 5 April.

Malmberg, A. and Maskell, P. (1997) Toward an Explanation of Regional

Specialization and Industry Agglomeration, European Planning Studies,
5, 1, pp. 25-41.

Markusen, A. (1998) Sticky Places in Slippery Space, Economic Geography,

Economic Geography, 72, p. 293-313.

Markusen, A. (1999) Fuzzy Concepts, Scanty Evidence and Policy Distance:

The Case for Rigour and Policy relevance in Critical Regional Studies,
Regional Studies, 33, pp. 869-886.

Marshall (1890) Principles of Economics, London: Macmillan.

47
Martin, R.L. (2001) Geography and Public Policy: The Case of the Missing
Agenda, Progress in Human Geography, 25, 2, pp189-210.

Martin, R.L. and Sunley, P. (1998) Slow Convergence? The New Endogenous
Growth Theory and Regional Development, Economic Geography, 74, pp.
201-227.

Maskell, P. (2001) Towards a Knowledge-based Theory of the Geographical

Cluster, Industrial and Corporate Change, 10, 4, pp. 921-943.

May, W., Mason, C. and Pinch, S. (2001) Explaining Industrial

Agglomeration: The Case of the British High-Fidelity Industry,
Geoforum, 32, 3, pp. 363-376.

Miller, P. Botham, R. Martin, R. L. and Moore, B. (2001) Business Clusters in

the UK: A First Assessment, London: Department of Trade and Industry.

Nachum, L. and Keeble, D.E. (1999) Neo-Marshallian nodes, Global

Networks and Firm Competitiveness: The Media Cluster of Central
London, Working Paper 158, ESRC Centre for Business Research,
University of Cambridge,.

Norton, R.D. (2001) Creating the New Economy: The Entrepreneur and US
Resurgence, Cheltenham: Edward Elgar.

O’Brien, R. (1992) Global Financial Integration: The End of Geography? London:

Pinter.

O’Malley, E. and Vanegeraat, C. (2000) Industry Clusters and Irish

Indigenous Manufacturing: Limits of the Porter View, Economic and
Social Review, 31, 4, pp. 55-

OECD (1999) Boosting Innovation: The Cluster Approach, Paris: OECD.

OECD (2001) Innovative Clusters: Drivers of National Innovation Systems, Paris:

OECD

OECD-DATAR (2001) World Congress on Local Clusters, Paris: OECD

Ohmae, K. (1995) The End of the Nation State: The Rise of Regional Economies,
London: Harper Collins.

Pandit, N., Cook, G. and Swann, P. (1999) The Dynamics of Industrial Clustering
in UK Financial Services, Manchester Business School Working paper,
399.

Perry, M. (1999) Clusters Last Stand, Planning Practice and Research, 14, 2, pp.
149-152.

48
Pinch, S. and Henry, N. (1999) Paul Krugman’s Geographical Economics,
Industrial Clustering and the British Motor Sport Industry, Regional
Studies, 33, 9, pp. 815-827.

Porter, M. E. (1990) The Competitive Advantage of Nations, London: Macmillan.

Porter, M. E. (1995) The Competitive Advantage of the Inner City, Harvard

Business Review, 74, pp.61-78.

Porter, M.E. (1996) Competitive Advantage, Agglomeration Economies, and

Regional Policy, International Regional Science Review, 19, 1, pp. 85-94.

Porter, M. E. (1998a) On Competition, Harvard Business School Press.

Porter, M.E. (1998b) Location, Clusters and the ‘New’ Microeconomics of

Competition, Business Economics, 33, 1, pp. 7-17.

Porter, M.E. (1998c) Clusters and the New Economics of Competitiveness,

Harvard Business Review, December, pp. 77-90.

Porter,M. E. (2000a) Locations, Clusters and Company Strategy, in Clark,

G.L., Feldman, M. and Gertler, M. (Eds) Handbook of Economic
Geography, Oxford: Oxford University Press, pp253-274.

Porter, M. E. (2000b) Clusters and Government Policy, Wirtschaftspolitische

Bldtter, 47, 2, pp. 144-154.

Porter , M.E. (2000c) Location, Competition and Economic Development:

Local Clusters in the Global Economy, Economic Development Quarterly,
14, 1, pp. 15-31.

Porter, M.E. (2001) Regions and the New Economics of Competition, in Scott,
A. (Ed) Global City Regions, Oxford: Blackwell, pp. 139-152.

Porter, M.E, (2002) Regional Foundations of Competitiveness and Implications for

Government Policy, Paper presented at the Department of Tyrade ad
Industry Workshop on Regional Competitiveness, DTI, London.

Porter, M. E. and Solvell, O. (1998) The Role of Geography in the Process of

Innovation and the Sustainable Competitive Advantage of Firms, in
Chandler, A., Hagstrom, P. and Solvell, O. (Eds) The Dynamic Firm: The
Role of Technology, Strategy, Organizations and Regions, New York: Oxford
University Press. pp. 440-457.

Porter, M. and Ackerman, F.D. (2001) Regional Clusters of Innovation,

Washington: Council on Competitiveness (www.compete.org).

49
Porter, M.E. and van Opstal, D. (2001) US Competitiveness 2001: Strengths,
Vulnerability and Long-Term Priorities, Washington: Council on
Competitiveness (www.compete.org).

Pouder, R. and St John, C. H. (1996) Hot Spots and Blind Spots: Geographical
Clusters of Firms and Innovation, Academy of Management Review, 21, 4,
pp 1192-1125.

Prevezer, M. (1997) The Dynamics of Industrial Clustering in Biotechnology,

Small Business Economics, 9, pp. 255-271.

Rodriguez-Pose, A. (2001) Local Production Systems and Economic

Performance in France, Germany, Italy and the UK, in Crouch, C., Le
Galés, P., Trogilia, C. and Voelzkow, H. (eds) Local Production System in
Europe: Rise or Demise? Oxford: Oxford University Press, pp. 25-45.

Roelandt, T. and den Hertog, P. (1999) Cluster Analysis and Cluster-Based

Policy Making in OECD Countries: An Introduction to the Theme, Ch 1
in OECD (1999) Boosting Innovation: The Cluster Approach, Paris: OECD,
pp. 9-23.

Rosenfeld, S.A. (1997) Bringing Business Clusters into the Mainstream of

Economic Development, European Planning Studies, 5,1, pp. 3-23.

Rosenfeld, S. (2001) Backing into Clusters: Retrofitting Public Policies, paper

available at www.oecd.org

Sabel. C. (1989) Flexible Specialisation and the Re-emergence of Regional

Economies, in Hirst, P. and Zeitlin, J. (Eds) Reversing Industrial Decline:
Industrial Structure and Policies in Britain and her Competitors, Oxford:
Berg, pp 17-70.

Sainsbury, D. (1999) Biotechnology Clusters: Report of a Team led by Lord

Sainsbury, Minister for Science, London: DTI/TSO.

Schmitz, H. (2000) Does Local Co-operation Matter? Evidence form Industrial

Clusters in South Asia and Latin America, Oxford Development Studies,
28, 3, pp. 323-336.

Schmitz, H. and Nadvi, K. (1999) Clustering and Industrialization:

Introduction, World Development Vol. 27, No.9, pp. 1503-1514.

Scott, A.J. (1998) Regions and the World Economy, Oxford: Oxford University
Press.

Scott, A.J. (Ed) (2001) Global City Regions: Trends, Theory and Policy, Oxford:
Oxford University Press.

50
Shohet, S. (1998) Clustering and UK Biotechnology, in Swann, G.M.P.,
Prevezer, M. and Stout, D. (Eds) (1998) The Dynamics of Industrial
Clustering: International Comparisons in Computing and Biotechnology,
Oxford: Oxford University Press, pp. 194-224.

Simmie, J. and Sennett, J. (1999a) Innovation in the London Metropolitan

Region, Ch 4 in Hart, D., Simmie, J., Wood, P. and Sennett, J. Inovative
Clusters and Competitive Cities in the UK and Europe, Oxford Brookes
School of Planning Working Paper 182.

Simmie, J. and Sennett, J. (1999b) Innovative Clusters: Global or Local

Linkages? National Institute Economic Review, 170, pp. 87-98.

Segal Quince Wicksteed (2001) Study of the Information Technology,

Communications and Electronics Sectors, Cambridge: SQW Ltd.

Steiner, M. (1998) (Ed.) Clusters and Regional Specialisation: On Geography,

Technology and Networks, London: Pion.

Steiner, M. and Hartmann, C, (1998) Learning with Clusters: A Case Study

from Upper Styria, European Research in Regional Science, 8, pp. 211-225.

Steiner, M. and Hartmann, C, (2001) Looking for the Invisible: Material and
Immaterial Dimensions of Clusters, Paper presented at the Regional
Studies Association Annual Conference on ‘Regionalising the
Knowledge Economy’, November 21, London.

Storper, M . (1997) The Regional World: Territorial Development in a Global

Economy, New York: The Guildford Press.

Swann, G.M.P. (1998) Towards a Model of Clustering in High Technology

Industries, in Swann, G.M.P., Prevezer, M. and Stout, D. (Eds) (1998)
The Dynamics of Industrial Clustering: International Comparisons in
Computing and Biotechnology, Oxford: Oxford University Press, pp. 52-76.

Swann, G.M.P. (2002) The Implications of Clusters: Some Reflections, Paper

presented at the Clusters Conference, Manchester Business School, 18
April.

Swann, G.M.P. and Prevezer, M. (1996) A Comparison of the Dynamics of

Industrial Clustering in Computing and Biotechnology, Research Policy,
25, pp. 1139-1157. Swann, G.M.P., Prevezer, M. and Stout, D. (Eds)
(1998) The Dynamics of Industrial Clustering: International Comparisons in
Computing and Biotechnology, Oxford: Oxford University Press.

Temple, P. (1998) Clusters and Competitiveness: A Policy Perspective, in

Swann, G.M.P., Prevezer, M. and Stout, D. (Eds) (1998) The Dynamics of
Industrial Clustering: International Comparisons in Computing and
Biotechnology, Oxford: Oxford University Press, pp. 257-307.

51
Turner, A. (2001) The Competitiveness of Nations: Myths and Delusions,
Chapter 1 in Just Capital: The Liberal Economy, London: Macmillan, pp.
23-50.

Van den Berg, L., Braun, E. and van Winden, W. (2001) Growth Clusters in
European Cities: An Integral Approach, Urban Studies, 38, 1, pp. 186-
206.

Venables, A.J. (1996) Localization of Industry and Trade Performance, Oxford

Review of Economic Policy, 12, 3, pp. 52- 60.

World Bank (2000) Electronic Conference on Clusters, World Bank.

52
53