8 Clusters and Competitive Advantage
One of the main purposes of the present study is to add to knowledge of
and the literature on clusters in the developing world by examining the
phenomenon of clustering in Turkey, which is classified by the World Bank
as a middle-income developing country.
Defining clusters: industrial districts, networks and clusters
Although the variety of clusters makes it difficult to define the concept
precisely, there is no shortage of definitions in the literature. According to
Hill and Brennan (2000, p. 66), for instance, a cluster is ‘a geographic con-
centration of competitive firms or establishments in the same industry that
either have close buy–sell relationships with other industries in the region,
use common technologies or share a specialized labor pool’. This is similar
to the definition adopted by Rosenfeld (1995, 2000), who sees a cluster as a
geographically bounded agglomeration of related firms that together are able
to achieve synergy. Redman (1994, p. 37) includes institutions as well and
defines clusters as a ‘pronounced geographic concentration of production
chains for one product or a range of similar products as well as linked insti-
tutions that influence the competitiveness of these concentrations’.
For ‘industrial districts’, a term that is sometimes used interchangeably
with ‘clusters’ in the literature, Pyke and Spengenberger (1990, p. 2) provide
Ludhiana
Light engineering
Delhi
Cotton hosiery
Okhla
Garments
Agra
Leather footwear
Morvi
Roof tiles
Tiruppur

Cotton knitwear
Bangalore
IT/software
Calcutta
Cotton hosier
y
Kanpur
Leather footwear
Figure 1.3 Examples of highly concentrated industries in India
Introduction 9
the following definition: ‘[Industrial] districts are geographically defined pro-
ductive systems, characterized by a large number of firms that are involved
at various stages, and in various ways, in the production of a homogeneous
product’. According to them, small and often family-owned firms, innova-
tiveness and entrepreneurial spirit, interfirm cooperation and flexible productive
networks are common features of such districts. Another definition comes
from the new industrial districts (NIDs) literature, which focuses on the set
of locational characteristics implied by flexible specialization: a district is a
spatially concentrated cluster of sectorally specialized firms, with a strong
set of forward and backward linkages, a common cultural and social back-
ground linking economic agents and creating a behavioural code, sometimes
explicit but often implicit, and a network of public and private supporting
institutions (Rabelotti, 1995). Biggiero (1999), on the other hand, defines indus-
trial districts as ‘regional hyper-networks’, or ‘networks of firm networks’.
Similarly, industrial districts are defined as a network of small and
medium-sized enterprises within geographically defined production systems
(Asheim, 1994). Meanwhile for Brusco (1990, pp. 14–15), industrial districts
comprise ‘a cluster of firms producing something which is homogeneous in
one way or another, positioning themselves differently on the market. Thus,
the district could be defined as being a cluster, plus a peculiar relationship

amongst firms.’
As a final example, Markusen (1996a) sees an industrial district as a spa-
tially delimited area of trade-oriented activity with a distinctive economic
specialization. Markusen classifies ‘sticky places’ (districts that demonstrated
resilience in the postwar period in advanced industrialized countries) into four
broad categories: Marshallian industrial districts (with an Italian variant); hub
and spoke districts (such as Boeing in Seattle and Toyota in Toyota City);
satellite industrial platforms (such as the US Research Triangle Park); and
state-anchored industrial districts (such as a military base or a university
influence on the development of some districts). According to Markusen, many
localities exhibit elements of all four models. Silicon Valley, for instance,
can be considered an industrial district in electronics, but it also has several
important hubs (including Hewlett Packard and Stanford University) as well
as hosting large branch plants of US, Japanese, South Korean and European
companies (including IBM, Hyundai, Samsung and NTK Ceramics). Further-
more it is the fourth largest recipient of military contracts in the country.
In Markusen’s view, therefore, it is wrong to concentrate solely on Italian-type,
small-firm industrial districts as ‘sticky places’ are the complex product of
multiple forces, including corporate strategies, industrial structures, profit
cycles, state priorities and local and national politics.
From the definitions reviewed above, two broad questions emerge: is it a
necessary condition for a cluster to be a small-firm agglomeration, and is it
a necessary condition for a cluster to be geographically concentrated? These
questions have to be answered in order to differentiate clusters from industrial
10 Clusters and Competitive Advantage
districts and networks. Amin and Thrift’s (1999) analysis of two districts –
Santa Croce in Tuscany and the City of London, the former specializing in
leather shoes and bags and the latter in financial services – is of special rele-
vance in respect of the first question. Santa Croce is a successful small-firm
district that targets the fashion-conscious end of the market. Meanwhile the

Marshallian structure of the City of London has undergone considerable
changes since the 1960s in that large corporations – including many multi-
nationals – now have offices there. Importantly, however, the concentration
of financial services has persisted, which – along with many other examples,
such as car production in Detroit – signals that geographic concentration is
not peculiar to small and medium-sized enterprises. Another interesting
study that reinforces this point is Saxenian’s (1994) work, which compares
the high technology cluster on Route 128 near Boston with Silicon Valley in
California. The former is dominated by vertically integrated, large companies
that prefer to maintain control over their technology and innovations. This
is manifest in the fact that companies along Route 128 are located in large,
self-contained, campus-like areas. In contrast Silicon Valley has a history of
independent, ‘garage-based’ entrepreneurs. It is necessary to remember at this
point that the definitions of industrial districts cited above usually emphasize
the predominance of small and medium-sized enterprises, although it is not clear
in the literature whether this is a necessary condition for industrial districts.
With regard to the question of geographic concentration, many researchers,
starting with Marshall (1949), have cited geographical proximity as a key
characteristic of districts (for example Pyke and Spengenberger, 1990; Asheim,
1994; Van Dijk, 1994). According to Asheim (1994, pp. 93–4), ‘what distin-
guishes an industrial district from other industrial agglomerations with strong
external economies such as the Perrouxian development poles or the Japanese
just-in-time production systems is precisely the existence of agglomeration
economies’. The latter is therefore a common feature of industrial districts
and geographic clusters, but not necessarily of networks. Instead a network
is defined in more general terms as a set of high-trust relationships that are
usually contractual and explicit. ‘In contrast to clusters, networks are generally
based on a group of firms with restricted membership and specific, often
contractual, business objectives . The members of the network choose each
other; they agree explicitly to co-operate in some way’ (Brown and McNaughton,

2002, p. 27). Based on this definition, it can be argued that network relations
are usually more cooperative in nature than are relations among cluster par-
ticipants, for which competitive forces are also emphasized. For example Porter
(1998) points to the intense rivalry in clusters. According to him, clusters offer
transaction cost advantages without imposing the inflexibilities of vertical
integration or the management challenges of creating and maintaining formal
linkages such as networks, alliances and partnerships (ibid., p. 214). What
can be deduced from the above is that a network usually involves explicit
and formal links among firms that are often cooperative in nature. Whether
Introduction 11
or not these are necessary conditions for networks, however, is again not
entirely clear. What is clear is that firms in a network are not necessarily tied
to the same location, while a cluster is a form of network that is situated in
a particular geographic location. Following this rationale, clusters can be seen
as ‘localized networks’ (Van den Berg et al., 2001) involving geographically
concentrated firms from a particular sector with links that can be both
cooperative and competitive in nature. Defined as such, it appears that clus-
ters are a form of network, whereas industrial districts are a form of cluster.
Although there are certainly network relations amongst firms that are not
geographically restricted, the focus of the present study is on the role of the
local environment in shaping competitive advantage in geographically con-
centrated industries. In this regard Porter’s (1998, 2000) definition exactly
matches the purposes of this study, and hence we shall define clusters as
geographic concentrations of interconnected companies and institutions in
a particular field (Porter, 1998, p. 197).
11
They include specialist suppliers,
specialized infrastructure, other service providers and associated institutions
(including universities, standards agencies and trade organizations), and also
extend to customers and firms in related industries.

12
As a final note, in many
of the definitions provided in the literature, clusters are implicitly seen as
dynamic, successful and competitive. This brings us to the focal point of
this study: the theorized link between clustering and competitiveness.
Clusters and competitiveness
The debate on the competitiveness of locations begins with the fundamental
question of whether all clusters are successful. Surprisingly this question has
received little attention in the literature. One notable exception is Amin (1994),
who investigates the attributes of successful versus unsuccessful clusters
based on two Italian case studies: Santa Croce in Tuscany and Stella in
Naples. Santa Croce is a competitive cluster where fashionable leather shoes
and bags are made, while Stella is a footwear cluster in an area with high
unemployment and widespread poverty. Interestingly, Amin argues that
many of the characteristics of the Stella firms are similar to those which
have made the clusters in Third Italy so successful. For example the firms’
owners are master craftsmen, the production process can respond quickly to
changing market signals, the existence of family businesses and community
ties permits labour flexibility, the lack of job opportunities mean cash savings,
and agglomeration and product specialization attract buyers and sellers of
raw materials and machinery. According to Amin, however, the specialist
shoe makers in Stella are not competitive since they have not formed them-
selves into a locally networked economic system (ibid., p. 62). Instead, and
despite their agglomeration, they are isolated from each other so there are no
exchanges of ideas, spin-offs and economies of scale through specialization.
Rather the artisans of Stella carry out ‘the tasks of the whole corporation
12 Clusters and Competitive Advantage
internally, but without any scale advantages or resources’ (ibid., p. 62). In
contrast Santa Croce has a typical Marshallian industrial structure, with
entrepreneurial, institutional and social interdependencies.

According to Amin, firms in successful clusters act like a collective brain,
although what contributes to this and why unsuccessful clusters persist are
not made clear in his analysis. What is clear is that the same conditions that
are associated with internationally competitive clusters are also typical of
some rather unsuccessful ones. The main reason why these are not covered
in the literature is simply that they have had little publicity. The answer to
the question posed at the beginning of this section is therefore ‘no’ – not all
clusters are competitive. The following subsections will discuss alternative
views on how and why some clusters manage to become competitive while
others do not.
Flexible specialization: a sure route to competitiveness?
A stream of research called ‘the flexible specialization approach’ focuses on the
organizational features of the regional economy and highlights the embedded-
ness of economic relations in broader social and political contexts. The main
argument (Piore and Sabel, 1984) is that there has been a transition from
Fordism to post-Fordism over the past two decades and a new post-Fordist
landscape has emerged: new industrial spaces containing small firms with
specialized, flexible production.
13
This perspective has fuelled studies on the
adoption of flexible manufacturing techniques and the link between industrial
organization and agglomeration. Proponents of flexible specialization argue
for vertically disintegrated and locationally fixed production, derived from
examples such as Silicon Valley (high-tech), Third Italy (semirural) and
Hollywood (inner city) (Amin and Thrift, 1999). In light of his study of indus-
trial districts in Emilia-Romagna, Capecchi (1990) argues that the definition
of industrial districts should include flexible specialization as a necessary
condition, together with the presence of small and medium-sized enterprises
(SMEs).
There are, however, strong criticisms of the view that flexible production is

a sure route to competitiveness. According to Amin and Robins (1990, p. 199),
for instance, this view represents ‘a kind of anti-Fordist utopia’ and ‘an
imposing orthodoxy’, since ‘we are being asked to believe that the very laws
of capitalist development are becoming, as it were, Marshallian (as opposed to
Fordist)’. In their view Fordism has far from disappeared, and the tendency
for localized agglomerations is in fact paralleled by a countervailing tendency
for transnational networks. It is therefore multinational corporations that
are the real shakers and shapers of the world economy. According to this
rationale, it is possible that clusters such as ‘Santa Croce will come to perform
only specific tasks in an internationally integrated value-added chain, thus
risking a shake out of firms dependent upon tasks no longer performed
locally’ (Amin, 1994, pp. 59–60). In response to Amin and Robins’ (1990)
Introduction 13
criticisms, Sabel et al. (1990, p. 230) state that proponents of flexible special-
ization have never claimed that all agglomerations can be viewed as flexible
production systems. Sabel et al. cite the Prato cluster as an example: this
cluster ‘has survived, even flourished, on the ashes of a number of crises in
its history. The current crisis – competition by large firms in some markets
served by the district – might simply push the district into doing what it has
done several times in the past: move up the price-performance curve by
specializing in higher quality items and leave the middle-range products to
the large firms’ (ibid., p. 234). Thus the argument that larger corporations
and multinationals are of determining importance is regarded as overrated.
As suggested by Bellini (1996), the experience of Third Italy seems to offer
irrefutable evidence of the possibility of an alternative, socially progressive
path of capitalist growth, providing an intellectual base for a number of micro-
interventions at the territorial level. In the early 1990s, however, there was
a shift in the focus of research when some potentially negative effects of
flexible production were brought to light. Harrison (1994), for instance,
pointed to the adverse effect that corporate flexibility might have on labour

practices in light of evidence that the use of child labour and the exploitation
of immigrants were becoming more common in some small and medium-sized
enterprises in Third Italy as global competition intensified. Malizia and Feser
(1999, p. 224), on the other hand, underlined that cooperation between
contracting firms did not necessarily imply an even playing field between
partners.
It is possible to bring the transaction costs approach to bear on the flexible
specialization debate. This perspective focuses on the activities of the typical
firm and presumes that if transaction costs are high the firm will choose to
internalize its operations. Also, in a region that is relatively underdeveloped
the firm may have no choice but to handle most of its basic functions in-house
since the market may not be large enough for other companies to focus
exclusively on producing the intermediate inputs or services needed. The
flexible specialization approach, on the other hand, builds on the fact that
there are dynamic external economies in districts, and the related benefit is
manifested in reduced costs, enhanced productivity and superior innovation.
These opposing views are likely to continue to compete in shaping not only
the debates in academia but also the organization and location of production.
To conclude, the flexible specialization perspective has a normative
dimension in that industrial districts have effectively been defined as places
where the dominant industries employ flexible production methods and are
highly competitive. On the methodological side, what this means is that few
researchers have conducted studies on the incidence of flexibly specialized
clusters that are struggling in terms of performance. In the absence of such
empirical studies, one might gain the impression that the type of industrial
structure and organization highlighted in the flexible specialization literature
is a guaranteed route to sustained competitiveness. This is of course a rather
14 Clusters and Competitive Advantage
limited picture (Malizia and Feser, 1999, p. 235). In the United States, for
instance, there are strong, resilient clusters that are not flexibly specialized

(Porter, 1998, 2000). Such clusters have been understudied, as have ones that
are flexibly specialized but not competitive. Storper (1999) thinks that although
the flexible specialization debate is theoretically powerful, empirical investi-
gation covering a wider sectoral base is needed to determine whether or not
the experience is specific to Italy. In Storper’s view there are deep historical
roots associated with the Italian districts that are difficult to generalize to
other competitive cultures, such as the Anglo-American ones.
Overall, although it is by no means certain that localized flexible special-
ization is a sure route to competitiveness (Malizia and Feser, 1999, pp. 237–8),
it is undeniable that the flexible specialization perspective has enabled us to
develop a more sophisticated understanding of clusters, where not only markets
and industries but also industrial organization, interfirm business relations
and the social situation can be instrumental in success (see Chapter 2 for a
discussion of this issue from the viewpoint of the management literature).
An approach that shares a common thread with the flexible specialization
perspective is the socioeconomic approach, which emphasizes the specific roles
played by the social situation and politics. We shall consider this in the fol-
lowing subsection.
The parts played by the social situation, trust and politics
This perspective is concerned with social, cultural and institutional influences
on the competitiveness of clusters. Brusco (1996), for instance, argues that the
importance of knowledge accumulation exceeds that of capital accumulation.
According to him, two types of knowledge are of particular importance in
this respect; namely codified knowledge (scientific and technical knowledge
in scientific journals, technical reviews and textbooks, whose conventions
and language are universal and known to the scientific community) and local
or tacit knowledge (embedded in the minds, imagination and skill of people
who live side by side and swap news and experiences when working
together). The latter type of knowledge is acquired by seeing how other
people do things, a process that is better managed in a local system (ibid.)

Becattini and Rullani (1996, p. 167) attribute special importance to a local
system’s ability to integrate codified and contextual knowledge, and argue
that this makes the cluster concept essentially socioeconomic and thus
a ‘disciplinary hybrid’, combining economics, sociology, geography and indus-
trial organization. In a similar vein, and building on the fact that phenomena
that persist over time possess some internal logic that cannot yet be explained in
full, Becattini (1990) calls for a ‘socioeconomic notion’ that will link neo-
classical, Marshallian and Marxian thinking. According to him, ‘better equi-
librium in analysis cannot be reached without the direct contribution of the
non-economists’ (ibid., p. 38). Only then can we understand what leads to the
construction of a strong image that evokes feelings of identification and
Introduction 15
belonging, ‘giving substance to expressions such as “the response of Prato”
to the lira devaluation’ (Becattini and Rullani, 1996, p. 172). In contrast to this
line of thinking, Lissoni (2001) points to the possibility that even in clusters
dominated by small and medium-sized firms, knowledge may be highly codi-
fied and firm-specific, rather than flowing freely throughout the cluster.
The idea that economic action is embedded in the structures of local social
relations paved the way for a related body of work on the role of social capital
in reducing transaction costs and facilitating network formation (Granovetter,
1985). According to this literature, close social networks and thus proximity
can help social capital to develop (Brown and McNaughton, 2002). A lead-
ing scholar in this field, Putnam (1993), argues that areas with low levels of
social capital, which in the case of Italy are concentrated in the south, have
slower rates of economic development than those with high levels of social
capital, which are concentrated in the central and northern parts of the country.
According to Putnam, repeated interaction and trust enhance social capital,
and it is likely that there will be a high demand for law enforcement in areas
with low levels of social capital as a result of heightened distrust. Of course it is
possible that other communities will exhibit a different type of social capital

(Flora and Sharp, 1997). Cohen and Fields (1999), for example, examined
social capital networks in Silicon Valley and found that the understanding
of social capital influenced by Putnam’s (1993) ideas (which refer to the
complex of local institutions and relationships of trust among economic actors
that evolve from unique local cultures) does not fit the situation in Silicon
Valley. Specifically, ‘networks of civic engagement’, which Putnam sees as
facilitating the activities of politics, production and exchange, have played
little role in Silicon Valley. For instance it is not possible to say that business
relations in Silicon Valley are embedded in family structures, given that Silicon
Valley is a world of strangers and newcomers. Moreover there is no deep
history to speak of. For Cohen and Fields, Silicon Valley is an economic
space built on a very different kind of social capital, where the pursuit of
economic objectives relates specifically to innovation and competitiveness
and there is virtually nothing in the history of Silicon Valley to connect
these networks of innovation to civil society. The high incidence of lawyers,
accountants and auditors is presented as an indicator of the limited degree
of informal, familial and communitarian trust in Silicon Valley, while the
rapid turnover of employees reveals a commitment to innovation rather
than to any particular company. Cohen and Fields conclude that there is trust
in Silicon Valley, but it is of a specific kind. This commercially valuable and
performance-focused trust is the building block of Silicon Valley’s ‘particular
brand of social capital’.
Another dimension is the role of work and politics in the competitiveness
of clusters. According to Brusco (1996), competitiveness and worker partici-
pation are closely linked. He argues (rather dramatically) that the clusters
in Emilia-Romagna show that cluster firms have at least solved some of the
16 Clusters and Competitive Advantage
key problems of large companies, that is, ‘How to involve workers and
indeed all production people generally in the productive process, how to
secure the participation of workers and technicians, how to storm world

markets with products that are the creation not only of the hands but also of
the heads and hearts of those that have made them’ (ibid., p. 154). Trigilia’s
(1990) emphasis is on the related issue of local political subcultures. Following
the institutional perspective, he examines two locales with similar socioeco-
nomic but different political structures: the ‘red’ Valdelsa (furniture and
glass) and the ‘white’ Bassano (shoemaking). The results of Triglia’s research
show that not only social components such as the extended family and the
local community, but also specific political components such as industrial
relations and the activities of governments in respect of the Catholic and
communist subcultures have played a part in the continuing competitive
success of the clusters studied. Regardless of its being ‘red’ or ‘white’, a political
movement that actively defends the collective interests of the local society
seems to matter.
Innovative milieu and untraded interdependencies
14

Schumpeter (1934) stressed the importance of the past trajectory of a locale
as a sign of its future innovative capacity. Accordingly the likelihood of the
next wave of an innovation to take place in its original area of development
is quite high. In this regard, Lagendijk and Charles (1999) call for a distinction
to be made between scholars who emphasize the role of networking in a
particular sociocultural context, as captured in the term ‘innovative milieu’,
and those who adopt the more institutional concept of ‘regional innovation
systems’. For the former, the growth of a locally embedded innovation system
is essential in shaping the social routines and strategies of actors in the
regional economy, whereas the latter pay more attention to the development
of and interaction between specific technology-oriented organizations such
as universities and research centres (ibid., p. 129). Borrowing from Storper
(1997, p. 16) however, there are a large number of universities around the
world but ‘there is a much smaller number of Silicon Valleys’, which sug-

gests that there must be other necessary conditions for the development of
innovative clusters (Brown and McNaughton, 2002).
An innovative milieu supports the development of such conditions. Spe-
cifically, based on the concept of knowledge spillovers in clusters, the milieu
approach focuses on the ways in which a local socioeconomic network creates
favourable conditions for innovation and competitive capacity: ‘local factors
such as business services, public support, infrastructures, skilled labor and
venture capital must be successfully woven together in order to sustain and
support innovation’ (McDonald and Vertova, 2002, p. 45). A closely related
concept is the ‘learning region’, in which a collective learning process by
firms takes place via social and business networks and becomes embedded
in the region (Camagni, 1991). This can also be linked to knowledge creation
Introduction 17
and the importance of tacit and codified knowledge in this process, as dis-
cussed above. According to this approach, globalization has triggered a shift
in the sources of competitive advantage towards innovation and thus
knowledge-based economic activity. In turn, knowledge-intensive economic
activities have a high propensity to cluster within a geographic region since
knowledge is generated and transmitted more efficiently in a local system
(Audretsch, 1998).
According to Storper (1999), this approach is paralysed by the circularity
involved in its analysis: innovation occurs because of a milieu, and a milieu
exists in regions where there is innovation. Instead, what really generate
region-specific assets and thus govern the sustainability of a cluster’s com-
petitiveness are ‘untraded interdependencies’. Storper argues that a region
can be seen as ‘a nexus of untraded interdependencies’ among three systems:
the labour market, the input – output system and the knowledge system.
A process of becoming specific that is in operation in such regions, imply
that ‘there is only one Silicon Valley if one wants to be “in the know” for
the most advanced innovations in semiconductor technology’ (ibid., p. 213).

The same logic can be extended to explain why some clusters persist and
maintain their competitiveness over time. That is, there are webs of user–
producer relations and untraded interdependencies, and localization of
these is frequent. The region is then key in the supply architecture for learn-
ing and innovation (ibid., p. 214). The literature, however, is inconclusive
about the conditions that lead to the emergence of an innovative milieu or
untraded interdependencies in a locale (see Chapter 2 for the management
literature’s view on how clusters foster innovation).
Is it just an accident of history? Path dependency and the lock-in
phenomenon
This approach investigates how and why a cluster emerges in a given location
and looks at the conditions associated with its subsequent development.
Understanding how certain standards or technologies persist despite the
fact that they might not be optimal has been the special concern of some
researchers (David, 1985; Arthur, 1985). Regarding the initiation of a cluster,
several authors (Marshall, 1949; Myrdal, 1957; Scott, 1988; Krugman, 1991a)
emphasize the role of ‘historical accident’. That is, the initial pattern may
simply be an accident of history, and once established a self-reinforcing loop
might occur (Martin, 1999). Thus the initial pattern becomes locked into an
area for economic and sociocultural reasons.
15
Once a locale has become a
centre of activity the lock-in effect comes into operation, and even if exogenous
circumstances change (perhaps reducing the attractiveness of the site) economic
agents may not want to move away and forgo the benefits of agglomeration.
History, then, might be a determining factor in the spatial pattern of economic
activity. A connected idea is related to sunk costs and how these bear upon
18 Clusters and Competitive Advantage
the spatial configuration of the firm, and thus upon the geography of
economic activity (Clark and Wrigley, 1997).

Another dimension of the issue is how the location of an agglomeration is
determined, given that it is typical for many locations to be candidates for
hosting an agglomeration. In this regard the possibilities range from pure
accident and the existence of small initial differences (Henderson, 2000), to
chance events that are partly shaped by the conditions prevailing in the
business environment offered by a location (Porter, 1990). Porter argues that
even inventions, which might change the competitive prospects of an
industry as well as the location in which it is concentrated, are more likely
to occur in places that are conducive to their development, the conditions
of which are specified in the diamond framework (see Chapter 2 for a
detailed discussion of this). Hill and Brennan (2000), on the other hand,
argue that it matters little why an industry locates in a certain place; what is
crucial is whether or not the conditions prevailing in the earliest stage of its
development begin to generate cluster economies. In short, despite the fact
that many studies conclude that ‘history matters’ (Arthur, 1986), it is not
clear which industries will become locked in and which will not, and hence
whether history does or does not matter (Enright, 1990). Rauch (1993), for
instance, asks, whether history matters only when it matters a little.
Negative externalities and the possible dissolution of clusters
This final subsection considers a rather overlooked aspect of the competi-
tiveness of clusters: negative externalities. So far we have concentrated on
the positive externalities associated with agglomeration. Pulling in the opposite
direction, however, are forces of dispersion. The main concern in this respect
has traditionally been the impact of congestion on firm costs and performance
(Brown and McNaughton, 2002, p. 24). The effects of an increase in the cost
and supply of immobile factors (rents and labour costs in particular) might
be serious enough to cause firms to relocate. There is also a possibility that
a cluster might be locked into the rules and routines of the past, thus missing
any new chances of upgrading. What might be even more damaging is that
such a situation, which Porter (1998) calls ‘groupthink’, might be exacerbated

by a tendency to exclude newcomers or outsiders, as suggested in the social
capital literature (Brown and McNaughton, 2002, p. 25).
It has been argued that there is a tendency for clusters to disperse as they
grow (ibid.) Relatedly, Dei Ottati (1994) emphasizes the destructive aspects
of price competition, which are more evident in a cluster environment than
elsewhere. She suggests that formal institutions should intervene to prevent
competition from degenerating into destructive forms. Otherwise the result-
ing disequilibrium might produce a chain reaction, and if the destructive
forces are sufficiently intense and prolonged a change of organizational model
(that is, dissolution of the cluster) might be unavoidable. Yet another force
that might contribute to the dissolution of clusters is the globalization of
Introduction 19
economic activities in general and the impact of multinational enterprises
in particular (Amin and Robins, 1990).
Overall, what the above discussion reveals is that very different circum-
stances, both economic and non-economic, might have an impact on the
structure and competitiveness of clusters. Accordingly the foundations of
success are rather different in, say, the clusters of Baden-Württemburg,
where there is ‘a Darwinian competitive pressure’, than in the clusters of
Third Italy, where there is a significant role for non-economic factors as well
(Staber, 1998).
16
The various explanations of why a cluster might become
competitive and sustain its competitiveness include those related to the
organization of production (for example flexibly specialized small and medium-
sized enterprises), the social and political context (for example the part
played by trust, social capital and institutions) and the existence of relations
within the cluster that pave the way for innovation, learning and untraded
interdependencies. There is also a possibility that both the initiation and
the subsequent development of a cluster might be an accident of history, with

the cluster then being locked into the region. Finally, the opposite forces of
dispersion should also be taken into account. None of these explanations,
however, can explain fully why certain clusters are competitive while others
are not. Specifically, contrary to what the flexible specialization approach
envisages, there are competitive clusters that are not flexibly specialized and
flexibly specialized clusters that are not competitive, as discussed above. On
the other hand, although the part played by the social and political situation
in general and extensive collaboration among cluster participants in par-
ticular might be important to the success of some clusters, such as those in
Third Italy, this cannot explain the success of prominent clusters such as
Silicon Valley. Similarly, the explanations that tie the competitiveness of
a cluster to innovation, learning and untraded interdependencies do not
sufficiently clarify why some clusters manage to become innovative and/or
develop untraded interdependencies while others do not. Hence despite the
fact that the numerous approaches in the literature have improved our
knowledge of the clustering of economic activity, we still lack a comprehensive
theory that can explain the competitiveness of clusters in full. One likely
contributor to a more complete understanding of the competitiveness of
clusters is a relative latecomer to the area; namely the unique perspective
offered by the management discipline, which is the subject matter of the
next chapter.
20
2
Clusters in the Management Literature
This chapter provides a review and discussion of recent debates on clusters
in the management literature in order to complete the background on clusters
presented in the previous chapter. An additional purpose of the chapter is to
highlight the unique contribution made by the management literature in
this regard. As Porter’s (1990, 1998) approach directly parallels the central
concern of this study due to its focus on the link between clusters and com-

petitiveness, the discussion will pay special attention to this approach and
the literature it has spawned, after considering some recent studies in the
general management literature that focus on different aspects of the issue.
An overview
As with the management literature in general, the literature on organization
theory is becoming richer in respect of cluster-related studies. Neoinstitutional
theory, for instance, offers a distinctive explanation of why organizations
cluster and argues that to be perceived as legitimate, organizations must
conform to the rules and requirements in a given institutional environment.
According to this rationale the organizations in a common institutional
environment converge in response to similar regulatory and normative pres-
sures or as they copy successful organizations (Baum and Haveman, 1997).
Meanwhile ecological models predict that the addition of an organization to
a population is likely to have stronger competitive effects on neighbouring
organizations than on those further away (Hannan et al., 1995). In a related
work, Hannan and Freeman (1977) outline a model in which competition
among the members of an organizational population is localized with
regard to size. They infer that organizations of different sizes use different
strategies and structures, and similarly sized organizations compete most
intensely. However Baum and Haveman’s (1997) analysis of location decisions
by the Manhattan hotel industry provides evidence in support of a combined
perspective in which hoteliers locate new hotels sufficiently close to
established hotels that are similar in terms of price and therefore benefit
Clusters in the Management Literature 21
from agglomeration economies, but differ in terms of size, thus avoiding
localized competition and creating complementary differences. In this
study, information externalities and the reduction of consumer search costs
are seen as the primary reasons for clustering. Ecological models of localized
competition have much in common with Hawley’s (1950) model of competi-
tive processes, which asserts that localized competition eventually leads to

differentiation since less fit organizations are transformed through functional
or territorial differentiation as a result of competitive pressures. Similarly,
according to White (1981), entrepreneurs try to avoid invading not only
competitors’ product/client niches but also their geographic niches. This
approach parallels a line of work from the area of marketing, which particularly
focuses on the competitive effects of brand location in product space
(Hauser and Simmie, 1981). In fact marketers have developed techniques
such as multidimensional scaling to find ‘ideal points’ for new product
entries (Baum and Haveman, 1997).
Another stream of research is the study of networks and interorganizational
relationships. These mostly concentrate on buyer–supplier relationships
and pay little attention to location. Studies that compare ‘global Toyotaism’
with ‘global Fordism’ (for example Fujita and Hill, 1999) are a notable excep-
tion in this regard. Their main conclusion is that global Toyotaism localizes
more of the production process than does global Fordism. The just-in-time
(JIT) supply system used by Japanese car makers encourages suppliers to locate
nearby to avoid the risk of delivery delays (Van Dijk, 1994). A frequently cited
example of this is Toyota City, an immense complex of interlinked suppliers
that have grown up around Toyota’s main assembly plants. A similar situation
exists in the case of Japanese automotive factories in North America in that
component companies tend to be located within a short driving distance
of the factories. ‘Just-in-time and in one place’, therefore, is a common
phenomenon, but is not necessarily inevitable. Hudson (1992) argues that
a historical analysis of the patterns of investment by Japanese automotive
companies in North America versus those in Europe provides an illustrative
case in this regard. The fact that American car producers have internalized
a considerable proportion of component production has left their Japanese
counterparts with little choice but to ensure that new component companies
are set up to supply them, and it is rational for these companies to be
located in close proximity. However a very different pattern is observed in

Europe, where the majority of the main European component producers
are significant international players and their production is rationalized on
a European scale. Therefore it is no surprise that new Japanese assembly plants
have plugged into the existing system. The introduction of JIT production
organization, therefore, may or may not lead to the spatial concentration of
production, depending on the specific historical circumstances and the cost
of coordination and transportation, which may be high enough to necessitate
close proximity (Malizia and Feser, 1999, p. 233).
1
22 Clusters and Competitive Advantage
Paralleling the network literature in general, the study of local networks,
which has only recently secured a prominent place in the research agenda
(Porter, 1998), focuses on trust and identity. In this literature the functioning
of a positive feedback loop in a local network is summarized as follows: the
more that people trust one another the more they strengthen their group
(network) identity, and the more they perceive themselves as a group (network)
the more they trust one another (Biggiero, 1999). Geographic proximity
reinforces this process (Porter, 1998). The transaction costs approach has
brought additional insights in this regard. According to some researchers
(for example Storper, 1999), agglomeration is the result of efforts to minimize
transaction costs. Specifically, the assertion that vertical integration has
transaction cost advantages over the market form is challenged by the cluster
approach since clusters are seen as having an organizational form that lies
in between markets and hierarchies, given that some disadvantages of the
market form are diminished by geographic proximity. The cluster environ-
ment, for instance, is conducive to reputation building since the cluster
participants usually live in the same area and are likely to have direct social
and economic exchanges over a long period of time, thus making it possible
to monitor their behaviour. Visser (1999) argues that spatial clustering
promotes the development of networks by lowering transaction costs in at

least two ways: the high density of related economic activities facilitates the
screening and selection of business partners on the basis of local information
and established reputations, and proximity between agents facilitates the
monitoring of behaviour and enforcement of contracts. In these circumstances,
local business customs are developed to such an extent that violating them
may result not only in the withdrawal of cooperation by other parties but
also in social sanctions (Dei Ottati, 1994).
2
The notion of clusters as an organizational form relates to the connections
between location and industrial organization, which are just beginning to
be explored (Enright, 1990). In a related work, Steinle and Schiele (2001)
argue that clustering is more relevant for industries with a divisible production
process and a transportable product. Another interesting piece of research is
that by Saxenian (1994), who provides a historical analysis of the development
of Silicon Valley and Route 128. This comparison of the two clusters shows
that Silicon Valley has exhibited resilience when facing challenges, whereas
Route 128 has been rather slow to respond. The main reason for this, according
to Saxenian, is related to industrial organization. Specifically, Silicon Valley
is a network-based industrial system whereas Route 128 is dominated by
a small number of relatively integrated corporations. There is a risk that this
conclusion might be misinterpreted as it seems to suggest that we should
rethink the large firm as an organizational form. It is therefore necessary to
point out that Saxenian’s emphasis is on the importance of local networks
rather than the size of firms in a cluster: ‘The contrasting experiences of Silicon
Valley and Route 128 suggest that industrial systems built on regional networks
Clusters in the Management Literature 23
are more flexible and technologically dynamic than those in which experi-
mentation and learning are confined to individual firms’ (ibid., p. 161). She
also acknowledges that there might be large as well as small firm variants of
network-based systems, as the Japanese experience suggests. Another note-

worthy finding of her study is that although collaborative practices in Silicon
Valley are pronounced, the region has long been dominated by individual
achievement, pointing to the value of competition as well as collaboration
in this cluster. Finally, the idea that ‘industrial organization matters, although
the type of organization that is most successful in particular places may
vary’ (Malizia and Feser, 1999, p. 236) is perhaps the only point of consensus
in this literature, and the organizational aspects of a cluster of independent
but informally linked firms and institutions, which constitutes a robust
organizational form in the continuum between markets and hierarchies, are
still underexplored (Porter, 2000).
The main contribution of international business scholars (for example
Dunning, 1995, 1998; Rugman and Verbeke, 2000) has been to analyse the
geographical concentration/dispersion of FDI conducted by multinational
enterprises, which may strengthen the location advantages of the countries
in which they operate through spillovers to local networks. A competitive
local cluster is in turn beneficial for the multinationals, since it enables
them to gain access to leading-edge ideas and specialist talents (Bartlett and
Ghoshal, 1991). With regard to new entrants to US industries, for instance,
there is evidence that foreign firms locate new plants in states with a high
concentration of similar activities (Head et al., 1995). When examining the
characteristics of foreign-owned subsidiaries in leading industry clusters,
Birkinshaw and Hood (2000) found that these subsidiaries gradually develop
the same characteristics as other firms in the clusters, since they have to
become insiders in order to reap the associated benefits. According to Rugman
and Verbeke (2000, p. 29), it is not only possible that multinationals might
alter a cluster’s attractiveness but also that they might act as a conduit for
international exchanges and spillovers, and thus contribute to the global
diffusion of knowledge. Finally, since the structure and content of the location
portfolios of multinationals have recently become more crucial to their global
competitive position, more attention needs to be paid to the importance of

location per se as a variable that affects the global competitiveness of firms
(Dunning, 1998, p. 60).
An interesting but overlooked aspect of agglomeration economies has
been addressed by Shaver and Flyer (2000): firms contribute to externalities
in addition to benefiting from them. This suggests that if firms are heteroge-
neous they will differ in the net benefits they receive from agglomeration.
According to Shaver and Flyer, firms with the best technologies, human
capital, training programmes, suppliers and distributors will gain little but
suffer much when their technologies, employees and access to supporting
industries spill over to competitors. Therefore, unlike firms with weak
24 Clusters and Competitive Advantage
endowments, these firms have little motivation to cluster. Shaver and Flyer
found supportive evidence of their arguments by examining the location
choice and survival of new foreign investments in US manufacturing indus-
tries. Specifically, their results show that large entities are less likely to
agglomerate, which in their view explains why some studies have identified
the existence of agglomeration mechanisms but many large-sample studies
have not found a superior performance among agglomerating firms.
One body of research in the strategy literature has investigated how clusters
affect innovation (Bergeron et al., 1998). With regard to the localized character
of innovation in the UK, for instance, Baptista and Swann (1998) have
found that firms in strong clusters are more likely to innovate. Another
interesting piece of research is that by Beal and Gimeno (2001), who have
found that technological knowledge spillovers matter little for US software
firms. What really matters are marketing spillovers in that cluster firms gain
an edge in finding markets and customers, as well as tailoring products for
them. This finding, which at first seems counterintuitive, highlights two
important issues. First, although technological innovations are without doubt
very important, sometimes it is possible to gain access to them relatively
easily, as in the case of new software technologies. Second, a broader under-

standing of innovation is necessary, rather than reducing the concept to
‘scientists in laboratories making breakthrough inventions’ (Porter, 1990).
As Beal and Gimeno’s (2001) study illustrates, innovation in marketing can
be of determining importance as well, and the related information (for
example on industry trends, market niches and customers’ needs) might be
difficult to codify and is thus easier to acquire in a cluster environment.
Following this rationale, it is not unlikely that a non-agglomerated firm will
make a breakthrough innovation but mistime its entry into markets (Yu, 2002).
As the above discussion has revealed, management scholars have addressed
numerous aspects of clustering, including the organization of production,
transaction costs, the nature of localized competition, the role of firm strategies
and marketing, firm-specific perceptions of the costs and benefits of clustering,
the part played by multinationals and the interaction between global and
local networks. The unique approach of management scholars involves
putting the firm at the centre of the analysis and trying to understand the
phenomenon of clustering from that point of view. This micro approach to
the issue has without doubt enhanced our understanding of clusters.
However, despite the growing interest in location-related issues in man-
agement research in recent years, relatively little management thinking is
directly connected to location (Porter, 1998). In fact Porter (1994) argues
that in the strategy literature, the part played by location has been seriously
neglected, particularly in respect of competitive advantage. Meanwhile
profound changes have been taking place in the world economy in that
globalization and the intensification of knowledge have greatly altered the role
of location in competition. Understanding this new role ‘requires embedding
Clusters in the Management Literature 25
clusters in a broader and dynamic theory of competition’ (Porter, 1998, p. 208).
Porter’s (1990, 1998, 2000) work on this has triggered a revival of interest in
the subject and prompted a lively debate in the literature. Apart from
furthering our understanding of clusters, Porter’s contributions are of particular

relevance to this study since they specifically focus on the relationship
between clustering and competitive advantage. The rest of this chapter will
therefore be devoted to a discussion of these contributions and the debate
they have spawned in the literature.
Porter-style geographic clusters
The main focus of Porter’s book The Competitive Advantage of Nations (1990)
was on sources of international competitive advantage at the industry level.
Based on a study of more than one hundred industries in ten countries,
he developed his diamond framework, which rests on the idea that sources
of advantage are local. In this framework, four attributes of the local
environment – factor conditions, demand conditions, related and supporting
industries, and context for firm strategy and rivalry – play a major part in
enabling domestic firms to gain and sustain competitive advantage. These
factors interact with each other to form a complex, mutually reinforcing
system, which makes the resulting advantage very difficult to replicate
elsewhere, and hence more sustainable. The dynamic character of the system
is magnified by the effects of domestic rivalry and the geographic concen-
tration of industry. Domestic rivalry prompts improvements in all the other
determinants, while geographic proximity amplifies the interaction between
the sources of competitive advantage. Pressure and challenge are of particular
importance in the emergence and sustainability of competitive advantage,
and both are driven by intense domestic rivalry and felt more heavily when
firms are in close physical proximity.
In his subsequent works Porter (1998, 2000) argues that the existence of
clusters is a manifestation of diamond theory. In his view the beginnings of
a geographic cluster can often be traced to historical circumstances or an
unusually sophisticated local demand (Porter, 1998). The prior existence of
related industries and/or one or two innovative companies might also provide
the seed for a new cluster, the latter echoing the growth pole approach.
Alternatively a chance event (such as an invention) may create advantages

that foster cluster development. Once a cluster begins to form a self-reinforcing
cycle promotes its growth since talented individuals are attracted by success
stories, specialist suppliers emerge, information accumulates and local insti-
tutions develop specialized training programmes, research facilities and
infrastructure. It can take a decade or more to develop a sustainable competitive
advantage, but it is also possible that some clusters will lose their competitive
edge because external (for example technological discontinuities)
3
and
internal forces (for example ‘group-think’ among cluster participants, which
26 Clusters and Competitive Advantage
can result in an overly inward-looking cluster) cause shifts in relative
competitive positions.
In Porter’s (1994) view the basis of competitive advantage has shifted
from static efficiencies (such as low input costs) to the ability to innovate
and upgrade skills and technology. This has brought about a radical change
in the importance of location in that the capacity to innovate and upgrade
draws heavily on the local environment, and the related advantages are
difficult to tap from a distance. Enduring competitive advantages thus lie
in the local environment and distant rivals cannot easily replicate them. In
other words the microeconomic quality of the local business environment
determines the sophistication with which companies compete.
Clusters affect competition in three broad ways: by increasing productivity,
by driving the direction and pace of innovation (which means future pro-
ductivity growth) and by stimulating the formation of new businesses
(Porter, 1998). Local rivalry boosts productivity since the presence of successful
rivals puts pressure on firms to innovate and break their dependence on less
sustainable sources of advantage – say, basic factors – to which their rivals
also have access. Close proximity facilitates constant comparison, and rivalry
among local firms often transcends the economic to become emotional and

personal. In fact the motivation-driving nature of a cluster can be amplified
by pride and a desire to look good in the local community. On the other
hand, interactions among cluster participants are likely to reflect longer-term
interests because the easy spread of information can affect reputation and
individuals’ desire for standing in the local community (Porter, 2000). Word
soon spreads if one cluster participant takes advantage of another or provides
shoddy products or services (Enright, 1990). Moreover, since assets and
skills are readily available in the location and would-be entrepreneurs can
benefit from established relationships, clusters are conducive to new business
formation. These factors, together with the presence of local firms that have
achieved success, reduce the perceived risks of entry. Customers’ demands
can also be better exploited in a cluster environment since firms can tune
into customers’ changing needs with a speed that is difficult to match by
companies located elsewhere. Relationships and face-to-face contacts in
a cluster enable companies to learn early about evolving technology and
marketing concepts. Since the opportunities for innovation are more visible
and therefore firms are prompted to act rapidly, clusters can remain centres
of innovation for decades (Porter, 2000).
This approach to clusters has implications both for economic development
(it calls for attention to microeconomic factors in government policy) and
for companies (for instance in respect of a location’s chance of attracting FDI).
Porter (ibid.) argues that the seeds of most clusters germinate independently of
government action and that a market test must be passed before development
efforts are justified. What is required of the government changes as a cluster
develops. In the early stages the government should focus on improving the
Clusters in the Management Literature 27
local infrastructure and eliminating disadvantages. In the later stages, it should
concentrate on removing obstacles to innovation. In short, the government’s
role is to improve the microeconomic foundations for competition, which
will ultimately determine the competitiveness of the cluster. According to

Porter, since knowledge and innovation are key factors in cluster development,
the incidence of clusters tends to increase with economic development. In
developing countries, which usually lack well-developed clusters, diverse
economic activities tend to be concentrated in one or two large cities.
4
Promoting clusters in such countries, Porter (1998) argues, should start with
the basics: improving capital markets, institutions, education and skill levels,
and technological capacity. Only then should cluster-specific investments
be made.
As mentioned earlier, Porter’s ideas on the importance of the local business
environment for creating and sustaining competitive advantage became the
focus of an intense debate in the academic literature.
5
In addition to the ten
countries studied in the original work (Porter, 1990), the approach was
applied to hundreds of industries in more than 40 countries, and the academic
literature on the topic grew to more than 30 book reviews and around 50
published articles (Davies and Ellis, 2000). The following pages will provide
examples of the application of Porter’s approach in different contexts, plus
a review of the major debates in the literature. Of particular relevance for
our purposes are critiques and comments on issues related to geographic
clustering.
Applications and empirical tests
The many applications of Porter’s approach include ones that replicated
his 1990 study in other countries. Some of these replications were conducted
by groups headed by Porter himself (for example in Canada, New Zealand
and several Latin American and Middle Eastern countries). A commonality
among these studies was that they took the value of Porter’s (1990) framework
for granted, and thus the primary purpose of the exercises was to identify
the sources of advantage/disadvantage in particular industries. The results of

the analyses were then used to suggest policy remedies for their weaknesses
and strategies to build upon their strengths. Replications conducted by other
researchers (such as those in Turkey by Öz, 1999, and Greece by Konsolas,
2002) were aimed at determining whether the diamond concept would
work in different contexts.
There have also been empirical tests of different aspects of Porter’s model.
For example O’Donnellan (1994) has tested, among other things, the geo-
graphical concentration hypothesis. The study concludes that in Ireland
there is a general preference among industries for large urban areas such as
Dublin and Cork, and that there is little association between sectoral clustering
and industrial performance. O’Malley and Van Egeraat (2000) have also
investigated the link between industrial performance (in this case in terms
28 Clusters and Competitive Advantage
of growth) and clustering, and concluded that there is limited evidence of
Porter-type clusters in Irish industry and no clear association between the
occurrence of such clusters and the growth of indigenous manufacturing.
Yet another study of Ireland is that by Clancy et al. (2001), which integrates
the principal findings of three Irish case studies on dairies (O’Connel et al.,
1997), popular music (Clancy and Twomey, 1997) and software (O’Gorman
et al., 1997). The overall conclusion is that although various elements in
Porter’s diamond have contributed to the competitive advantage of the
industries studied, there is also support for critics of Porter, especially those
who question the importance of a home base in small, open economies
such as Ireland, as well as the part played by foreign multinationals in
fostering competitiveness. However it should be noted that Clancy etal. (2001)
do not focus on industries that are both competitive and indigenous, which
they themselves acknowledge is a weakness of their study. This is a reasonable
self-critique, and a counterfactual speculation is possible. This does not,
however, change the fact that although it lacks clear examples of Porter-style
clusters, Ireland has performed well economically in recent years. This leaves

two possible explanations. First, critics of Porter may be right that his
approach to clusters is not relevant for a small, open economy. Alternatively,
it may be that Ireland is an exceptional case, with growth occurring from
a low starting point, and it is possible that it will prove difficult to sustain
indigenous growth in the long term unless strong clusters emerge.
In another empirical test, Yetton etal. (1992) have re-examined the analyses
conducted by Porter in Canada and New Zealand and supplemented them
with an original analysis in Australia. They conclude that competitive
industries usually lack strong diamond elements in their home base in these
countries. Similarly, and again focusing on Australia, Ellis and Pecotich (1996)
have found evidence to refute the importance of strong home-base diamond
elements. The test results obtained by Cartwright (1993) in his attempt to
test the whole model in light of the New Zealand finding show that the Porter-
ideal model can be associated with industries that are characterized by
moderate competitiveness and static/declining profitability, rather than ones
characterized by strong competitiveness and growing profitability. A similar
result has been produced from a study on Hong Kong’s competitiveness, in
that Hong Kong’s most competitive industries do not have strong demand
conditions or favourable factor conditions in their home base. Finally, based
on their study in the Netherlands, Jacobs and De Jong (1992) conclude that
the importance of the home base varies from sector to sector.
Having summarized the country applications and specific tests of the
hypotheses put forward by Porter (1990, 1998, 2000), we shall now consider
some regional investigations of Porter-style clustering. In their abovementioned
study of clusters in Ireland, Clancy et al. (2001) look at the geographical
location of dairies, popular music and software. While all three are spatially
concentrated, the dairy cluster is the only one that is both indigenous and
Clusters in the Management Literature 29
very competitive. The authors argue that proximity has facilitated information
flow within this cluster. Interestingly, for the dairy cluster (like many other

clusters in Ireland) most of the important downstream and related industries
are foreign-owned. This is linked to a much-debated aspect of Porter’s
approach that will be discussed in detail in the next subsection.
In his application of the diamond framework in London, Hamilton (1991)
provides a historical analysis of the capital and argues that during the past
30 years the area known as the City has been transformed from a predomin-
antly materials-based economy to a mainly information- and financial-based
one. He then investigates London’s international competitiveness with the
help of Porter’s diamond model, which is assumed to be valid and is used as
an organizing framework to examine competitiveness. According to Healey
and Dunham (1994), who apply Porter’s analysis to another British local
economy, Coventry, some changes are required when the diamond is applied
to subnational units since a number of the tools available to the national
government (such as exchange rate control and interest rates) are not available
to regions. Moreover factors of production, especially labour and capital, are
more mobile within than between countries. Having discussed such concerns,
Healey and Dunham use Porter’s approach to investigate why Coventry’s
position changed from relative competitive disadvantage during the 1970s
and early 1980s to relative competitive advantage during the remainder of
the 1980s. They conclude that Porter’s (1990) analysis provides a useful
framework to examine the competitiveness of local economies, since each of
Porter’s four determinants has had some bearing on the change in Coventry’s
competitive advantage.
Kaufman and Gittel (1994) have applied the diamond framework in New
Hampshire (United States) to evaluate Porter’s (1990) hypotheses on the
effect of geographical concentration on industrial performance. To identify
New Hampshire’s competitive industries they used export shares (following
Porter’s methodology) plus two additional indicators: productivity and wage
levels. A historical analysis of New Hampshire’s four leading industries –
fabricated metals, industrial machinery and equipment, electronic and

electrical equipment, and instruments – was supplemented by a survey
that provided interesting finding. When asked whether proximity promoted
rivalry, for example, New Hampshire companies stated that they did not
compete with one another, nor did they generally compete with other
firms in the region. Kaufman and Gittel’s interpretation of this is that
firms in each of the leading industries were operating in different market
niches, thus avoiding direct competition. Similarly, with regard to proximity
to customers, half of the respondents stated that this was of little importance
since the majority of primary products were sold to customers located
elsewhere in the country. However there was some evidence of suppliers
being geographically concentrated. Thus the survey results indicate that
several of Porter’s conclusions do not hold for New Hampshire. Kaufman
30 Clusters and Competitive Advantage
and Gittel’s overall conclusion is that although Porter’s model provides
a useful tool for analysing a region’s economic activity, the New Hampshire
economy differs in some very important ways from the predictions of the
model.
As well as the applications and empirical tests summarized above, other
studies have taken elements of Porter’s approach. For instance Padmore and
Gibson (1998) have designed a model to describe and assess the strengths
and weaknesses of regional industrial clusters using some dimensions of
Porter’s diamond and applying the model at the regional level in British
Columbia (Canada). Dobkins (1996) has modelled locales that produce goods
for trade outside their boundaries, assuming monopolistic competition. This
model was inspired by Porter’s (1990) story of the Montebelluna (Italy) ski
boot industry, in which externalities, economic growth, and historical and
spatial considerations were all present. The most significant contribution of
these two studies has been to put some of Porter’s key ideas into a math-
ematical model. In another study, Shilton and Stanley (1999) trace the survival
and growth of the headquarters of publicly listed firms in the United States.

Their finding supports Porter’s thesis that firms cluster for competitive
advantage. Finally, Doeringer and Terkla (1995) explore the economic foun-
dations of business clusters in the United States and their link to development
and competitive advantage. Drawing on interview-based field research, they
warn that intense local rivalry among firms within production channels can
be counterproductive.
The debate on Porter-style clusters
There have been various criticisms of Porter’s (1990, 1998) approach. Stop-
ford and Strange (1991) criticize Porter’s lack of formal analytic modelling,
while Bellak and Weiss (1993), Dunning (1992) and Grant (1991) question
the originality of the framework. Porter has also been criticized for his
treatment of macroeconomic policy (Daly, 1993), for his failure to clearly
define determinants and several key terms (Grant, 1991) and for paying
insufficient attention to modern trade theory (Bellak and Weiss, 1993) and
the role of national culture (Van den Bosch and Van Prooijen, 1992). His
methodology has also attracted criticism, including his heavy reliance on
world export shares as a measure of international competitiveness (Grant,
1991; Cartwright, 1993; Rugman and D’Cruz, 1993), his inadequate treatment
of relatively less competitive industries (Yetton etal., 1992) and his treatment
of multinationals and foreign direct investment (Rugman, 1991; Bellak and
Weiss, 1993; Dunning, 1993; Hodgetts, 1993; Rugman and D’Cruz, 1993;
Rugman and Verbeke, 1993). Criticisms of his diamond framework relate to
the undue importance attributed to the relationship between domestic
rivalry and international competitiveness (Smith, 1993) and the indirect
role envisaged for the government (Stopford and Strange, 1991; Van den
Bosch and de Man, 1994).
Clusters in the Management Literature 31
Several of the contributors to a volume compiled by the OECD (1999)
have incorporated Porter’s approach to clustering into their analysis. For
example Drejer et al. (1999) assess Porter’s Danish studies, Peneder (1999)

focuses on the case of Austria, and Rouvinen and Yla-Anttila (1999) first
evaluate Steinbock’s (1998) study of Finland and then apply Porter’s
approach to Finland, but with a number of changes.
6
While these studies are
generally positive about Porter’s approach, Davies and Ellis (2000, p. 1189)
offer a rather negative evaluation of Porter’s 1990 book: ‘While it was
enormously rich in its range and scope it fell far short of the claims made
for it.’ According to them, the assertions at the heart of the study can be
refuted because strong diamonds are not in place in the home bases of many
internationally successful industries. Amongst the numerous proposals for
improvement, Grein and Craig (1996) suggest that the diamond should
have three elements instead of four – ‘infrastructure/demand’, ‘competitive
investment’ and ‘education’ – and O’Shaughnessy (1997) adds ‘custom’,
‘history’ and ‘politics’. When considering the South Korean case, Cho (1994)
suggests that the four-cornered diamond should be replaced by a nine-factor
model consisting of four physical factors (endowed resources, business
environment, related and supporting industries, domestic demand), four
human resources (workers, politicians and bureaucrats, entrepreneurs,
professional managers and engineers) and chance.
However many scholars have acknowledged the value of Porter’s approach,
and ‘even the most hostile [have] praised the richness of Porter’s cases’
(Davies and Ellis, 2000). For instance Greenaway (1993) is of the opinion
that Porter’s approach serves as an excellent complement to mathematical
analyses of competitive advantage. Similarly Smith (1993, p. 404) believes
that Porter’s firm-oriented approach represents an original contribution to
development theory: ‘The alternative of simply applying an unalloyed
paradigm of neo-classical analysis plus the assumption that government failure
is always greater than market failure seems weak by comparison.’ Holt
(1998) especially celebrates Porter’s emphasis on local rivalry and considers

that his analysis of how competitive pressure in clusters drive innovation
and prosperity is ‘one of the glories of his approach’. According to Gray
(1991) and Dunning (1992), Porter’s extensive and rich field research has
advanced our knowledge of why corporations in some locations have suc-
cessfully penetrated foreign markets in some product areas but not others,
and of why some countries have been able to attract foreign-owned firms to
participate in some value-added activities but not others. According to
Grant (1991), Porter’s assertion that innovation and upgrading are central
to the creation and maintenance of competitive advantage constitutes a step
towards the reformulation of the strategy model in a dynamic context. One
implication of this is that when determining the sources of competitive
advantage and formulating strategy, it should be remembered that the
resource base of a firm is determined not only by its own past investments