Journal of Management Studies 46:3 May 2009

doi: 10.1111/j.1467-6486.2008.00815.x

Demand- and Supply-Side Agglomerations:

Distinguishing between Fundamentally Different
Manifestations of Geographic Concentration

Brian T. McCann and Timothy B. Folta

Purdue University; Purdue University and EM Lyon

abstract Agglomeration research investigates the benefits that firms receive from locating
in close geographic proximity. Despite a substantial surge in interest in this topic over the past
20 years, a lack of distinction among unique manifestations of spatial concentrations of similar
firms threatens continuing progress in this stream of research. We argue that agglomerations
of related firms that draw benefits from the supply-related externalities of increased access to
specialized labour, specialized inputs, and knowledge spillovers are fundamentally different
from those that draw benefits from heightened demand realized through reduction in
consumer search costs. Extending agglomeration theory, we explicate the differences between
these distinct phenomena, discuss how the nature of key theoretical relationships varies across
these agglomeration types, and demonstrate significant implications for research. We discuss
how the differences affect a host of theoretical relationships and empirical research decisions.

INTRODUCTION
The last 20 years have seen a dramatic increase in attention to the potential for firms
locating in close geographic proximity to benefit from positive feedback, or externalities.
Such externalities occur when the ‘net benefits to being in a location together with other
firms increase with the number of firms in the location’ (Arthur, 1990, p. 237). Geo-
graphic concentration of economic activity occurs broadly; well-known examples include
high-technology industries (e.g. California’s Silicon Valley), automotive manufacturing
(e.g. Detroit), and local retail services (e.g. automobile dealers and mall food courts). As
Brown (1989, p. 451) observes, ‘the clustering of similar firms is a truly universal trait,
ranging from the hamburger alleys and automobile rows of most American cities to the
pronounced clusters of goldsmiths and banana sellers in the periodic markets of the third
world’.
Consistent with an emerging stream of literature in this area (e.g. Martin and Sunley,
2003), we take issue with the identification of nearly every geographic assemblage of
Address for reprints: Brian T. McCann, Purdue University, Krannert School of Management, 403 W. State
Street, West Lafayette, IN 47907, USA ([email protected]).

© Blackwell Publishing Ltd 2009. Published by Blackwell Publishing, 9600 Garsington Road, Oxford, OX4 2DQ, UK
and 350 Main Street, Malden, MA 02148, USA.
 Demand- and Supply-Side Agglomerations 363
firms as a ‘cluster’, befitting of the externalities that accompany agglomeration. Indeed,
among agglomerated firms there are stark differences in the likelihood, the extent, and
the nature of such externalities depending on the underlying mechanisms driving the
agglomeration. Our purpose is to highlight that increased precision in the understanding
of such differences yields insight into important questions that have thus far remained
uninvestigated. Different causal mechanisms manifest themselves in very different
agglomeration phenomena. These differences across types of agglomerations have sig-
nificant implications for how we interpret prior literature on agglomeration and how we
develop theory and empirical studies around agglomeration. We contribute to this
literature in three important ways.
First, we follow a research tradition that has identified that clusters come in different
forms. However, our intent is not to develop a new taxonomy of agglomerations. Instead,
we contribute to this literature by investigating the implications of agglomeration differ-
ences across agglomeration types. In particular, we argue that differences in the nature
of externalities bears upon fundamental issues such as the geographic area spatial
concentrations occupy, the complexity and variety of agglomeration membership, the
size and scope of the externalities available to firms, and the degree of competition, as a
few examples. Prior efforts have not thoroughly investigated the differences across types
of agglomerations.
Second, while a number of important perspectives have illuminated the important
link between geography and firm location and performance, including organizational
ecology (e.g. Baum and Haveman, 1997; Baum and Mezias, 1992), international busi-
ness (e.g. Cantwell and Iammarino, 2001; Head et al., 1995), strategic management (e.g.
Porter, 1998b; Sorenson and Baum, 2003), and entrepreneurship (e.g. Cooper and Folta,
2000; Stuart and Sorenson, 2003) among others, we draw from Marshall’s (1920) earlier
work that has emphasized the difference between supply-side versus demand-side
agglomeration. While Marshall crafted this simple typology, he failed to develop the
implications inherent in many of the issues identified above. Our emphasis on Marshall’s
work is appropriate given that much of the prior management literature (e.g. Porter,
1998a; Pouder and St. John, 1996; Saxenian, 1994) focuses on geographical concentra-
tions of related firms whose co-location generates positive externalities (‘Marshallian
externalities’), and this research stream has popularized the term ‘cluster’ to refer to this
phenomenon. This literature, however, fails to fully appreciate a key distinction that
appeared in Marshall’s work. Marshall was careful to elaborate on how different mecha-
nisms might explain different types of geographic agglomerations of firms, and his work
can be used to clarify why agglomeration exists for today’s biotechnology firms versus
automobile dealers. Yet, management research does not frequently distinguish between
these types of agglomerations and their underlying causes. Instead, it recognizes agglom-
erations in a wide range of industries under the ‘cluster’ label, including semiconductors
(Almeida and Kogut, 1999), biotechnology (Prevezer, 1997), hotels (Canina et al., 2005)
and restaurants, car dealers, and antique shops (Porter, 2000). Tying our arguments to
the distinct causal mechanisms first outlined by Marshall (1920) that drive spatial con-
centration, we question whether every agglomeration of related firms is the result of
similar agglomeration externalities such that they should be considered identical
phenomena. In particular, we argue that a fundamental distinction exists between

© Blackwell Publishing Ltd 2009

364 B. T. McCann and T. B. Folta
agglomerations of producers of industrial goods that benefit from supply-related Marshallian
externalities and sellers of consumer goods and services that benefit from demand-related Mar-
shallian externalities. A literature stream that fails to consistently recognize this distinc-
tion and, more importantly, one that fails to investigate the implications of the distinction
between these two fundamentally different types of related-firm agglomeration hampers
our quest to build a clear understanding of the agglomeration phenomenon.
Third, we elaborate on why these differences have significant implications for research
and practice. We discuss how these differences have important conceptual implications
and impact a wide variety of research decisions, including dataset selection, measure-
ment of agglomeration benefits, and operationalizations of agglomerations. These issues
have not been clearly demarcated in prior research, yet are fundamental to understand-
ing whether there are payoffs to agglomeration. Our hope is that this effort will contrib-
ute towards the accumulation of a more coherent theoretical and empirical record of the
general agglomeration phenomenon and its unique manifestations.

PRIOR LITERATURE
A pursuit towards explaining the spatial concentration of economic activity has a long
tradition of scholarly work (e.g. Hotelling, 1929; Marshall, 1920). Our argument to
distinguish between two specific types of agglomeration fits well within a stream of
research that has already established a number of important distinctions. We first review
the clear distinctions that already exist in the literature and then highlight the area in
which we think further work is necessary, namely the difference between supply- versus
demand-related agglomerations of related firms.

Prior Distinctions among Types of Agglomeration

The first key issue in distinguishing among geographic concentrations of industry is
whether related or unrelated activity is being investigated. Agglomeration can generate
benefits for two very distinct reasons, depending on whether the agglomeration consists
of specialized or diverse industrial activity. Spatial concentrations of unrelated firms may
benefit from urbanization economies. These externalities flow from the geographic concen-
tration of aggregate economic activity, such as cities. Firms benefit from urbanization (or
diversification) externalities because industrial diversity fosters fertilization of ideas across
industries. Jacobs’ (1969) work, The Economy of Cities, is so closely associated with this
stream of research that these benefits are commonly called Jacobs’ externalities. The
operation of urbanization economies has been predominantly studied by economists
interested in the development, growth, and operation of urban areas.[1] Diversity among
firm product offerings can also serve as an explanation for concentrations of unrelated
retail firms, as in a shopping mall.
As our particular interest lies in the geographic concentration of related firms, we devote
the remainder of our effort here.[2] Marshall (1920, p. 268) argued that the chief explana-
tion for industrial localization flowed from the unique physical conditions of particular
areas. Examples of resources that might draw firms to a particular area include unique raw
materials, specialized climate, concentrations of customers for retail firms, transport

© Blackwell Publishing Ltd 2009

 Demand- and Supply-Side Agglomerations 365
nodes, cultural facilities, universities, ‘star scientists’ in the biotechnology industry
(Zucker et al., 1998), and institutions that facilitate access to scarce resources (Meyer and
Nguyen, 2005). The key point in distinguishing this explanation is that while location
benefits exist, the source is exogenous to the firms, and those benefits do not increase
as more firms locate in the area. Despite Marshall’s belief in the prevalence of this
explanation for industrial localization, Ellison and Glaeser (1999) found that 80 per cent
of the clusters they studied could not be explained by an identifiable exogenous natural
advantage.
Hotelling’s (1929) work on location choice offers perhaps the best-known explanation
for similar firms choosing to locate together. In his model, firms end up concentrated at
the centre of a linear market as a result of profit-maximizing location choice, his principle
of minimum differentiation. Hotelling’s model was necessarily highly stylized, and
research that followed this work (e.g. Eaton and Lipsey, 1979; Graitson, 1982; Hay,
1976; Prescott and Visscher, 1977) and relaxed a number of the main assumptions of the
model demonstrated that agglomeration was not necessarily an equilibrium outcome.
This work showed that Hotelling’s model does not provide a general explanation for
spatial concentration, concluding that ‘once the assumptions are relaxed very slightly in
the direction of realism, Hotelling’s model predicts that no two firms should be clustered
together’ (Eaton and Lipsey, 1979, p. 422).
Spatial concentrations of related firms may instead be explained by the presence of
endogenous localization or specialization economies. Marshall (1920) enumerates four sources
of agglomeration externalities for geographic concentrations of related firms: (1) greater
access to specialized inputs; (2) greater access to specialized labour; (3) knowledge/
technology spillovers; and (4) access to greater demand. The first three categories benefit
spatially concentrated firms by providing better access to the supply of important
resources while the last category benefits agglomeration members by increasing the
demand for their products or services.

Distinguishing between Supply-Side and Demand-Side Agglomerations

The literature on Marshallian externalities is voluminous, and our purpose here is not a
detailed review of that literature, so we limit ourselves to a brief overview.[3] On the
supply side, geographic concentration draws greater numbers of providers of specialized
inputs, granting access to services that firms could otherwise not afford individually. In a
similar fashion, the firm has greater access to specialized labour as population size grows.
Labour markets deepen as skilled workers follow concentrated opportunities, and search
costs for qualified employees become lower. Finally, knowledge and technology may spill
over as social networks develop among the competing firms facilitating the transfer of
information, expertise, and contacts (Almeida and Kogut, 1997; Inkpen and Tsang,
2005; Pouder and St. John, 1996). Many of the more significant and widely cited works
on agglomeration (e.g. Krugman, 1991a, 1991b; Porter, 1998a, 1998b; Saxenian, 1994)
focus their attention on supply-related externalities.
On the demand side, agglomerations of similar firms reduce consumer search costs
(Stahl, 1982; Stuart, 1979), as spatial concentration fosters the discovery and evaluation
of a variety of offerings across different firms. Reduced search costs for consumers

© Blackwell Publishing Ltd 2009

366 B. T. McCann and T. B. Folta
increases the probability that consumers will visit the agglomeration for their purchase
relative to isolated locations, leading to higher demand for co-located firms. This exter-
nality is particularly relevant when products require visual inspection by consumers prior
to purchase (Stahl, 1982) and in cases of high product heterogeneity (Fischer and
Harrington, 1996). While this stream of the agglomeration research is not as voluminous
as its supply-related counterpart, research in demand-side externalities holds a rich
tradition, and interest in the topic remains strong.
For the purposes of this paper, we reserve the use of the term ‘cluster’ or supply-side
agglomeration (SSA) to refer to groups of geographically proximate firms that draw
benefits from supply-side externalities – SSAs provide access to specialized labour and
inputs and enable the more efficient transfer of tacit knowledge across firms in a region.
We refer to demand-side agglomerations (DSA) as groups of geographically proximate
firms that draw benefits from demand-related externalities – DSAs provide access to
customers who seek to reduce search costs.
One might argue that this distinction has been recognized since Marshall’s work
and little value will be gained by emphasizing and explicating these differences. We
acknowledge that research in agglomeration mostly treats supply- and demand-related
externalities as two discrete literatures. Indeed, many scholars are careful to explicitly
note that their work applies to only one area. As just two examples, both Shaver and
Flyer (2000) on the supply-side and Chung and Kalnins (2001) on the demand-side are
careful to restrict the theoretical scope of their articles to particular categories of exter-
nalities. Nevertheless, while most research has been careful to demarcate supply-side
agglomeration from demand-side agglomeration (as noted above), we are concerned
that there may be increased confusion among these phenomena as we move forward.
This concern stems from two reasons. First, one of the most widely cited definitions of
agglomeration confounds the different types of externalities. The definition by Porter
(1998b, 2000) describes clusters as an interconnected web of focal firms, suppliers,
supporting institutions, related-industry firms, and customers but also claims that ‘clus-
ters occur in many types of industries, in smaller fields, and even in some local indus-
tries such as restaurants, car dealers, and antique shops’ (Porter, 2000, p. 18). Our work
will demonstrate the inconsistency in these two descriptions. Second, some recent work
implies an intermingling of the two literatures. As just one example, Canina et al.
(2005, p. 567) discuss demand-related differentiation spillovers but build this argument
on the foundation of a supply-related agglomeration study (Shaver and Flyer, 2000).[4]
Finally, while we agree that Marshall and much of the ensuing work recognize the
distinction, little effort has been made to explicate the theoretical and empirical implications
of these differences. A more nuanced understanding of the differences between supply-
side and demand-side agglomerations would help clarify the literature and enable
research on these unique manifestations of agglomeration to proceed unimpeded by a
confounding of the two.
To emphasize their distinct characteristics, we devote the majority of the paper to
casting SSAs and DSAs as ideal types, conceptually derived constructs that deliberately
accentuate certain characteristics and not necessarily something that is found in empiri-
cal reality (Doty and Glick, 1994; Weber, 1949). Typologies (such as ours) create useful
heuristics and provide a systematic basis for comparison.[5] A central drawback to

© Blackwell Publishing Ltd 2009

 Demand- and Supply-Side Agglomerations 367
typologies is that they are often neither exhaustive nor mutually exclusive, which is the
reason that before we discuss the characteristics of these ideal types, we devote the next
section to a discussion of how the SSAs and DSAs might overlap and be interdependent.

Intersecting Supply-Side and Demand-Side Agglomerations

A glance at the economic landscape suggests that there are many instances where
demand and supply-side agglomerations appear to overlap geographically. For example,
we might observe one or several demand-side agglomerations of automobile dealerships
or hotels within a supply-side agglomeration of biotechnology firms. In this section we
discuss several ways in which geographically overlapping SSAs and DSAs might be
interdependent, and not mutually exclusive.
One form of interdependency might flow from one type of agglomeration spurring
development of the other. For example, we imagine that a growing supply-side agglom-
eration of biotechnology firms and suppliers (along with their employees) creates demand
for the types of consumer goods offered in demand-side agglomerations (e.g. hotels or
auto dealers). A subset of these consumer goods firms attracted to the region could decide
to agglomerate within the region. A growing supply-side agglomeration can thus create
conditions under which demand-side agglomerations may (but not necessarily would)
form. The potential for demand-side agglomerations to spur supply-side agglomerations,
however, seems only likely to occur in special circumstances. We also believe it unlikely
that the supply- and demand-related factors we have discussed interact directly to
accentuate each other. For example, we see no causal mechanisms by which individual
search costs would be further reduced because of firms’ improved access to specialized
labour, specialized inputs, or knowledge spillovers.
Another form of interdependency may reside at the firm level, as firms in the region
might benefit simultaneously from both supply and demand factors. Two examples
illuminate our logic. Business service clusters are spatial concentrations of business
service activities such as financial services, advertising, accounting, and the like that are
concentrated in the central districts of leading large cities (Coe et al., 2007, p. 144). One
recent survey of companies within the London financial services industry provides
evidence that business service cluster firms benefit from the supply-side externalities of
specialized labour, specialized inputs, and knowledge spillovers (Pandit et al., 2004). The
study also observes that the firms benefit from ‘being close to market-leading customers’,
which certainly implies a demand-related benefit. A second example is a hub-and-spoke
district, or a region ‘where a number of key firms and/or facilities act as anchors or hubs
to the regional economy, with suppliers and related activities spread out around them
like spokes of a wheel’ (Markusen, 1996, p. 302). Markusen (1996) clearly argues that
supply-related externalities such as increased access to specialized labour and specialized
inputs providers are important factors in hub-and-spoke districts. And, similar, to the
case of business service clusters, input providers may gain a demand benefit by locating
close to their customers.
Despite these likely interdependencies, we note that our focus in this paper is on
endogenously created externalities. Agglomeration externalities occur when the ‘net
benefits to being in a location together with other firms increase with the number of firms

© Blackwell Publishing Ltd 2009

368 B. T. McCann and T. B. Folta
in the location’ (Arthur, 1990, p. 237). While it is an empirical question, we contend it is
unlikely that a particular business service cluster or hub-and-spoke district will benefit
from both supply- and demand-related externalities. To put it differently, it seems more
likely that the benefits to locating near customers do not increase as more product/service
providers locate in a region. Demand-related externalities exist when growth in the
number of providers causes larger numbers of customers to be attracted to the agglom-
eration. This occurs most often, as we have described, when product/service heteroge-
neity exists and consumers wish to inspect products prior to purchase. This condition
seems mostly inapplicable to the types of products and services offered by the provider
firms in business service clusters and hub-and-spoke districts. In neither case is there a
clear argument that growing numbers of providers make the agglomeration a more
attractive place for customers to purchase. Instead, the customers are already there and
they attract the product/service providers rather than the reverse. We contrast this with
the case of common DSA firms that we have described earlier such as shoe stores,
automobile dealers, and hotels. As more of these firms join an agglomeration, consumer
search costs are reduced, more consumers are drawn to the agglomeration, and the net
benefits to being in that location do increase as more firms locate within the agglomera-
tion. Thus, we believe that any demand-related benefits that accrue to product/service
provider firms within a business service cluster or hub-and-spoke district are not increas-
ing with the number of provider firms; rather, they flow from an exogenous source,
namely locating near a concentration of customers. That is, we believe demand exter-
nalities do not appear to be a relevant factor in these agglomerations unless it is the case
that consumers wish to visit the firms to investigate or inspect their products and services
prior to purchase. Despite our doubt regarding the potential for these benefits, it is an
empirical question worthy of study.[6]
Finally, although we have argued that neither business service clusters nor hub-and-
spoke districts are examples of agglomerations benefiting from both supply and demand
externalities, we have not ruled this possibility out theoretically. Demand-related exter-
nalities flow from a reduction in consumer search costs and potentially accrue to firms
that offer products that consumers wish to inspect prior to purchase. Supply-related
externalities potentially accrue to firms that require specialized labour/inputs or incor-
porate knowledge into their products. One potential case of firms who might benefit from
both demand and supply externalities is the venture capital industry. We imagine that a
group of spatially concentrated venture capitalists, such as those located in the Silicon
Valley area, might benefit from the concentration of specialized labour and the avail-
ability of knowledge spillovers. Furthermore, to the extent that firms seeking venture
capital wish to personally investigate the venture capital firms prior to transacting with
them, the spatial concentration would lead to a reduction of search costs and a corre-
sponding demand increase for the agglomerated venture capital firms. While it is possible
to imagine unique instances of particular firm types that might benefit from both of these
sources, the literature in this research stream and empirical observation does not appear
to support the broad existence of this type of interdependency. Agglomerations of firms
that offer products that consumers wish to visually inspect tend to be found in industries
in which specialized labour, specialized inputs, and knowledge spillovers are not
particularly relevant. Firms in industries benefiting from supply-side agglomeration

© Blackwell Publishing Ltd 2009

 Demand- and Supply-Side Agglomerations 369
externalities tend not to offer products that consumers wish to visually inspect prior to
purchase. Overall, we believe that although any given agglomeration of firms may benefit
from multiple kinds of agglomeration externalities, the dominant externalities accruing
to those firms will be associated with a particular ideal type. For example, clustered firms
in the Silicon Valley area seem to benefit most from Marshallian supply-side externali-
ties. In contrast, spatially concentrated auto dealers likely benefit most notably from
Marshallian demand-side externalities. Thus, we will proceed with comparing the two
types in their pure form.
Before moving to the details of our arguments, we offer two final clarifications. First, we
note that agglomeration explanations are not necessarily industry-specific. The reasons for
a spatial concentration of firms within a certain industry may vary depending on the
context. For example, a number of authors have studied agglomeration externalities in the
hotel industry (Canina et al., 2005; Chung and Kalnins, 2001; Kalnins and Chung, 2004).
In many cases, demand-side externalities explain hotel agglomerations, but in other cases,
a natural resource (such as a beach) may better explain the geographic concentration.
Second, we acknowledge that we are not the first to attempt categorization of related-
firm agglomerations. They have been distinguished along a number of dimensions,
including whether they are focused around a core technology vs. a core industry (St. John
and Pouder, 2006) and whether inter-firm and institutional networks are present or
absent (Rocha and Sternberg, 2005). Markusen (1996) identifies four distinct types of
agglomerations of industrial goods’ firms based on the characteristics of firms that
constitute the agglomeration.[7] We acknowledge the utility and value of these distinc-
tions but believe that they fail to fully appreciate a distinction that exists at a more
fundamental theoretical level, namely the very different casual mechanisms underlying
supply-side and demand-side agglomerations.
In the section below, we elaborate upon these fundamental distinctions between these
two pure forms of agglomerations. We also speculate as to whether the externalities from
these two types of agglomeration will persist over time and are robust to changes in
industrial growth, regional growth, and macro-economic growth. Overall, the topics
covered represent key research areas in the agglomeration literature, and our intent is to
explicate how answers to these research questions will vary substantially depending on
the type of agglomeration being investigated. The underlying causal mechanisms of each
type of agglomeration lead to marked differences in the nature of theoretical relation-
ships in each. Our conclusions are summarized in Table I.

DIFFERENCES IN SUPPLY-SIDE AND DEMAND-SIDE

AGGLOMERATIONS
Fundamental Differences
We believe that the different causal factors driving externalities in SSAs and DSAs lead
to differences in fundamental features such as industrial diversity, geographic size, and
network interconnectedness/relationship complexity. The first two characteristics of
industrial diversity and geographic size are focused upon definitional concerns similar
to those echoed by Martin and Sunley (2003, p. 10) who argue that ‘the obvious

© Blackwell Publishing Ltd 2009

370 B. T. McCann and T. B. Folta
Table I. Supply-side versus demand-side agglomeration summary

Dimension SSAs DSAs

Primary sources of • Specialized labour, specialized inputs, • Heightened demand flowing from
agglomeration and knowledge spillovers reduced consumer search costs
benefits
Industries that form • Most likely found in industries that are • Most likely found in retail and
SSAs and DSAs knowledge-intensive and require high consumer services industries
innovation rates (e.g. biotechnology) • Industries should be characterized by
• May have a common technology core heterogeneous products that require
but span across a number of industries visual inspection by the consumer
Composition • Focal firms, suppliers, and supporting • Primarily horizontally related firms
institutions (universities, VCs, • Focal firms
consultants, etc) – both horizontally and • Perhaps firms offering complementary
vertically related firms goods
• Related industry firms, producers of
complementary products, and primary
customer groups
Relationship • High • Low
complexity • Large number of both vertical and • Realizing agglomeration benefits does
horizontal relationships between not require any intra-DSA relationships
members to exist
• Knowledge flows across these • Where evidence of inter-firm
relationships, a relatively complex relationships does exist, relationships
process tend to involve exchange of explicit
• Knowledge that flows is often tacit information
Geographic • No requirement of immediate physical • To deliver reduced search costs to
space adjacency to realize benefits consumers, firms must locate within a
• Can extend over a fairly wide reasonable distance of each other
geographic range • Thus, groupings should be tighter
Source of • Congestion costs and rising input prices • Congestion costs and rising input prices
diseconomies less likely to develop as quickly more likely to develop quickly
• Adverse selection concerns may be • Adverse selection concerns not balanced
balanced by ‘better’ firms benefiting by ‘better’ firms benefiting more
more • Increased localized competition is a
• Increased localized competition less of a major concern
concern
Life cycle • Should take longer to initially develop • Benefits are relatively simple and should
• Establishing complex relationships takes not take long to establish (although it is
time not clear how many firms do you need
• Likely to take longer to reach to make a DSA)
diseconomy stage • Likely to reach diseconomy stage more
• Overall, life cycle might thus be longer quickly
• Overall, life cycle might thus be shorter
Temporal changes • Growing • Waning
in value • As successful competition continues to • Falling transportation costs and
become more dependent on knowledge increased information availability make
and innovation, cluster value should it less costly to gather information about
continue to increase heterogeneous products

© Blackwell Publishing Ltd 2009

 Demand- and Supply-Side Agglomerations 371
problem raised by these cluster definitions is the lack of clear boundaries, both indus-
trial and geographical’. Our focus on network interconnectedness and relationship
complexity is directly related to the key role that interconnections appear to play in the
clustering phenomenon as reflected by Tallman et al. (2004, p. 261), who describe a
cluster as ‘group of firms tied together by geographical co-location and complex social
interaction’.

Industrial composition of agglomerated firms. The first dimension we consider is the composition
of firms in the agglomeration. Specifically, we consider whether supply- and demand-
side agglomerations might differ with regard to industrial diversity within the agglom-
eration, from both a horizontal and vertical perspective. We consider SSAs and DSAs,
respectively.
1. SSA composition. Given that one of the key agglomeration economies in clustering is
access to specialized inputs, it is unsurprising that clusters should require the presence of
suppliers to generate those benefits. Nearby suppliers can also contribute to knowledge
spillovers. Specialized labour will be further attracted to the area by the presence of
suppliers and companies in related industries who are likely to require skill sets at least
somewhat similar to those required by the focal firms. One key type of specialized labour
often present in technology-related clusters may be ‘star scientists’. Both Audretsch and
Stephan (1996) and Zucker et al. (1998) found evidence that the location of new biotech
firms was associated with the presence of prominent scientists in the area. Supporting
institutions provide specialized inputs as well, including technical, financial, and strategic
advice. For example, Feller (1997) described the role of the Advanced Research Program
in Texas, which helped develop a high-technology cluster around the city of Austin. Folta
et al. (2006) described a variety of institutions that support biotechnology clusters.
Overall, clusters consist of both firms in a focal industry and a variety of suppliers of
related products, services, and infrastructure.
To the extent that knowledge spillovers and specialized labour/inputs are not
industry-specific, we should expect clusters to include not just firms in the focal industry
but also those in vertically-related industries that might be involved in the industry’s
vertical scope of activities. This view is consistent with Maskell (2001), who argues that
clustered firms draw advantages from both the vertical and horizontal dimensions of the
cluster. To the extent that the supporting institutions are also engaged in horizontally-
related industries, it is possible that clusters might be quite industrially diverse.
2. DSA composition. In contrast to clusters, DSAs generate agglomeration economies by
reducing consumer search costs, and all that is required for a reduction in these search
costs is a grouping of similar firms from which customers can choose. Neither suppliers
of specialized inputs nor supporting institutions must be present, as they bear no rela-
tionship to the reduction of search costs. Canina et al. (2005, p. 567) describe demand-
side agglomerations purely as the ‘result of firm heterogeneity that attracts more
consumers to an area simply because the area has a wider variety of firms from which to
choose’. This fact is even clearer to see when we consider one example. Fischer and
Harrington (1996) observe that shoe stores tend to co-locate in the Baltimore area. They
speculated that women’s shoes, in particular, were characterized by a lack of standard-
ized products, making price comparisons meaningless without visual inspection, thereby

© Blackwell Publishing Ltd 2009

372 B. T. McCann and T. B. Folta
encouraging repeated search and comparison shopping. Under such conditions, DSAs
create value. These geographic concentrations of shoe retailers, however, existed without
any supporting institutions located geographically proximate. Suppliers, such as shoe
manufacturers, showed no tendency to co-locate with the retailers either. The firms
themselves are a sufficient condition for generating the agglomeration benefit.
We do anticipate a potential role for producers of complementary goods in a DSA.
Their specific role in this phenomenon is a currently unaddressed area in the literature.
While their presence might not be a necessary condition, they may generate a benefit when
they are present. We imagine that a higher concentration of restaurants might heighten
the demand faced by a focal hotel within a DSA; however, the theoretical mechanism is
distinct from a reduction in consumer search costs. While complementary goods pro-
ducers may enhance the attractiveness of the agglomeration, DSAs do not require the
other organization types that populate clusters.

Connections among agglomerated firms. The second dimension we consider is the connections
among firms in the agglomeration. Specifically, we consider whether SSAs and DSAs
might differ with regards to the intensity of interaction and the nature of knowledge that may be
conveyed across agglomerated firms.
1. SSA relationships. We begin by acknowledging that not all scholars agree on the
necessity of knowledge spillovers as an explanation for supply-side agglomerations. For
example, Krugman (1991a, p. 497) offers a formal model to explain agglomeration and
notes that ‘What is particularly nice about this result is that it requires no appeal to
elusive concepts such as pure technological externalities’. Krugman (1991b) further
argues that many highly localized industries (carpets in Dalton, Georgia or jewellery
producers in Providence, Rhode Island) are not characterized by high levels of knowl-
edge or technology. Yet, even in non-knowledge-based explanations for SSAs, intercon-
nections between firms play a key role. Focal firms access specialized inputs from their
relationships with supporting institutions and suppliers. As Krugman (1991b, p. 51)
notes, ‘there will be both backward and forward linkages that provide an incentive to
concentrate production’.
We cannot deny, however, the omnipresence of the role of knowledge spillovers in
research on supply-side agglomerations. Empirical evidence indicates that geographic
proximity does enhance knowledge spillovers. Jaffe et al. (1993) provide evidence con-
sistent with localization of knowledge spillovers by examining patent citation data. They
found that citations were five to ten times as likely to come from the same metropolitan
area as control patents. Audretsch and Feldman (1996) showed that knowledge-
dependent industries tend to be more spatially concentrated. Rosenthal and Strange
(2001) noted evidence suggesting that knowledge spillovers contribute to agglomeration
at the local level using a broad sample of 459 manufacturing industries. Rosenkopf and
Almeida (2003) also found evidence supporting geographic localization of knowledge
using patent citation data from the semiconductor industry. Finally, Sammarra and
Biggiero (2008) found that firms within Rome’s aerospace industrial cluster shared
technological, market, and managerial knowledge through innovation networks.
Geographic proximity facilitates the transfer of tacit knowledge across firms. In con-
trast, spatial concentration provides little advantage for the sharing of explicit knowledge,

© Blackwell Publishing Ltd 2009

 Demand- and Supply-Side Agglomerations 373
which transfers over longer distances more easily. Given its uncertain and highly
contextual nature, tacit knowledge is best transmitted via rich media like face-to-face
meetings (Daft and Lengel, 1986) and through frequent repeated contact. Keeble and
Wilkinson (1999) discuss three mechanisms by which geographic co-location facilitates
the transfer of tacit knowledge: employee mobility across firms within the cluster; inter-
actions between customers and suppliers of equipment; and the birth of new organiza-
tions from existing firms, research laboratories, and universities. Inkpen and Tsang
(2005) note how informal social gatherings, meetings, and interactions that occur within
the cluster foster the development of personal relationships that in turn help create
network ties between spatially concentrated firms. ‘Proximity helps the formation of
network ties and facilitates interfirm and especially interpersonal interactions through
which knowledge is exchanged’ (Inkpen and Tsang, 2005, p. 156). Chen (2003) found
that areas rich in such network resources tended to draw foreign direct investment
activity from expanding Taiwanese electronics firms.
Tallman et al. (2004) argue that firms in clusters share component and architectural
knowledge (Henderson and Cockburn, 1994) related to the particular technology
employed in the cluster, but what truly distinguishes clusters is the existence of cluster-level
architectural knowledge. This complex construct ‘represents understandings developed
at the regional cluster level through the routinization of the network of interactions,
interdependencies, and common interests among the members’. Firms’ interactions lead
to a shared understanding of ‘reciprocity, reputation, interdependencies, and the other
relational aspects of the knowledge cluster’ (Tallman et al., 2004, p. 265). This shared
understanding facilitates the transfer of product-level component and architectural
knowledge within the cluster by increasing each constituent firm’s absorptive capacity
(Cohen and Levinthal, 1990) and slows knowledge leakage out of the cluster. In a similar
view, Storper (1995, p. 206) argues that the key feature of clusters is the presence of
‘untraded interdependencies’, which include ‘labour markets, and “conventions”, or
common languages and rules for developing, communicating and interpreting
knowledge’.
Overall, then, SSAs are comprised of a web of multiple relationships. Each firm may
share some relationship with each of the other related firms, in addition to its relation-
ships with suppliers of specialized inputs, complementers, related-industry firms, and
supporting institutions. When we consider that institutions other than the cluster focal
firms likely possess inter-connections as well, we begin to see a cluster as the complex,
interconnected web of organizations depicted in Figure 1. The potential for the sharing
of tacit knowledge across firms within networks supported by interpersonal relationships
contributes to the complexity of this web of relationships.[8]
2. DSA relationships. We have argued that a demand-side agglomeration consists of
fewer firm types than a cluster. That, in and of itself, implies a lower level of complexity
within DSAs. We believe further that the nature of connections between these firms results
in lower complexity. First, the existence of relationships between focal firms is not a
necessary condition for reduction of consumer search costs. A group of hotels or auto
dealers who never speak to each other and share no information among themselves still
receive the benefits of demand-side agglomeration externalities. Thus, DSAs may exist
with no interconnections.

© Blackwell Publishing Ltd 2009

374 B. T. McCann and T. B. Folta

Supporting
Institution

Supplier Supplier

Related Related
Industry Industry
Firm Firm

Firm Firm
(Industry A) (Industry A)

Supporting Supporting
Institution Institution

Firm Firm
Pool of Specialized Labour (Industry A) (Industry A)

Related Related
Industry Industry
Firm Firm

Supporting
Supplier Institution Supplier

Figure 1. SSA configuration

Notes: Supply-side agglomerations include a web of multiple relationships supported by a pool of specialized
labour. In addition to the focal firms, SSAs often include supplier firms, supporting institutions, and firms
from related industries. The arrows between firms represent the inter-connections that may exist at both the
firm and employee level among the focal firms as well as among the other firms within the cluster. These
interconnections help create the potential for the sharing of tacit knowledge across firms within networks
supported by interpersonal relationships.

While we have argued that interconnections are not necessary, we do acknowledge

that they may exist. For example, Kalnins (2006) notes two interconnections that may
exist between geographically proximate hotels. First, some hotels form networks across
which they make referrals. Second, managers jointly participate in ‘call-arounds’ in
which hotels call each other and share information on current price and occupancy
levels.[9] What is noteworthy about both of these practices is the nature of the information
being shared. In DSAs, it appears that if information is shared, it tends to be more of an
explicit nature (e.g. price and occupancy information) as opposed to the explicit and tacit
knowledge that flows between organizations within clusters.[10]
In a separate paper, Kalnins and Chung (2006) find evidence of another type of
connection between hotels that can lead to inter-firm resource sharing. However, the
authors note that these network relationships only develop when there is some underly-
ing affiliation between the firms (e.g. a common ethnic background of the hotel owners)
and the resources shared are not knowledge-based but rather tangible goods like furni-
ture and capital.

© Blackwell Publishing Ltd 2009

 Demand- and Supply-Side Agglomerations 375

Complement Complement
Firm Firm

Firm Firm
(Industry A) (Industry A)

Pool of Common Customers

Firm Firm
(Industry A) (Industry A)

Complement Complement
Firm Firm

Figure 2. DSA configuration

In comparison to an SSA, the number of organizational types in a DSA is lower – while firms offering
complementary products may exist, their presence is not necessary. Rather than pulling from a pool of
common specialized labour, DSA firms pull from a pool of common customers. The dotted lines between
DSA firms capture the fact that whether interconnections exist between these firms is questionable. Fur-
thermore, to the extent that inter-firm relationships do exist, they tend to involve the transfer of relatively
simple information or tangible goods.

Our discussion of demand-side agglomerations results in a much simpler composition

as we depict in Figure 2. The number of organization types is lower and whether
interconnections exist between these firms is questionable. Even when inter-firm rela-
tionships exist, they will tend to involve the transfer of relatively simple information or
tangible goods. Hence, DSA relationship complexity should be relatively low.[11]

Size of geographic area of agglomerated firms.

1. SSA size. We turn first to the issue of the geographic limits of clusters. The essential
question in defining SSA boundaries is over how far of a distance will the benefits of
specialized labour, inputs, and knowledge spillovers extend. As a first observation, a
requirement of immediate physical adjacency appears unnecessary to realize efficiencies.
The case for specialized labour and inputs flowing across relatively wide geographic
boundaries is straightforward. Firms can draw labour from a relatively wide area assuming
that labour is relatively mobile. Similarly, clustered firms can access specialized inputs

© Blackwell Publishing Ltd 2009

376 B. T. McCann and T. B. Folta
from a fairly broad geographic area as long as the transportation infrastructure of the
particular area is sufficiently developed. In support of these views, Rosenthal and Strange
(2001) found that proxies for labour market pooling and reliance on manufactured inputs
positively influenced agglomeration at levels of geography extending state-wide.
The externality that may most constrain the size of supply-side agglomerations is
knowledge spillovers. There is relatively strong evidence that knowledge spillovers are
stronger over shorter distances. For example, Jaffe et al. (1993) found that knowledge
spillovers appeared to be most evident for firms in the same Metropolitan Statistical Area
(MSA), even though their evidence suggests that spillovers exist at the MSA, the state,
and the country level. Orlando (2004) provided evidence that knowledge spillovers
within the same three-digit SIC extended only through a 50 mile radius. Rosenthal and
Strange (2001) found that spillover benefits were constrained to the tighter geographic
area of the zip code level. Overall, the weight of the evidence suggests that knowledge,
especially if it has tacit components, diffuses most effectively over areas significantly
smaller than states or countries.
2. DSA size. The key issue that may constrain the size of demand-side agglomerations
is the requirement that the consumer physically travel to the location of the business in
order to inspect and purchase the product or service. The smaller geographic size of the
DSA, the lower the search costs for the consumers. Casual observation verifies the tight
boundaries of DSAs; when we see a group of automobile dealers that constitute an auto
mall, they tend to be located within blocks (not tens of miles) of each other. It is this level
of closeness that distinguishes a DSA, and we tend to find these agglomerations described
as ‘malls’ or ‘districts’ (as opposed to the ‘regions’ that characterize SSAs).
While what constitutes a ‘close’ distance will vary depending on the ability and
willingness of consumers to travel, we can at least say that it is improbable that this distance
will extend to the geographical boundaries seen in clusters. The location of a demand-side
agglomeration may also impact the ability and willingness of consumers to travel and
therefore the range of DSA boundaries. In areas in which travel is more difficult, we might
expect that DSAs will be more tightly packed. The required spatial tightness is likely
further related to the nature of the industry and the size of the purchase, with consumers
being willing to expend greater effort for larger, less frequent purchases (like automobiles);
however, consumers are unlikely to expand their search area to cluster-level distances
(such as state-wide) to evaluate such purchase alternatives. Operationalizations of the
geographic range of DSAs in previous research support a more limited view of their reach
with multiple distinct areas identified within the boundaries of a city (Baum and Haveman,
1997; Fischer and Harrington, 1996) or by studying these agglomeration types at the zip
code level (Chung and Kalnins, 2001; Kalnins and Chung, 2004).
To sum up our argument to this point, we have elaborated on fundamental differences
between SSAs and DSAs, emphasizing the presence of complementary industries, the
degree of relationship complexity, and geographic size. Our perspective thus far has been
fairly static, while in reality, agglomerations evolve over time (e.g. Saxenian, 1994). In the
section which follows, we distinguish between supply-side and demand-side agglomera-
tions on issues around evolution – their dynamic character. As theory in this realm is less
developed, the distinctions we make are somewhat more speculative than the previous
section.

© Blackwell Publishing Ltd 2009

 Demand- and Supply-Side Agglomerations 377
Differences in Evolution
In considering whether types of agglomeration differ in their evolution, we build from
theory presented by Pouder and St. John (1996), who suggested that geographic ‘hot
spots’ might have life cycles where net agglomeration benefits differ over time. Since their
work did not consider how net agglomeration benefits might differ between SSAs and
DSAs, we do so below to further accentuate the different implications of these two types
of agglomerations. We also consider how the net benefits of SSAs and DSAs might vary
with changes in the macro-economic environment over time.

Diseconomies of agglomeration. Some recent work (Pouder and St. John, 1996; Prevezer,
1997; Shaver and Flyer, 2000) has begun to note the potential for diseconomies of
agglomeration, particularly in later stages of a cluster’s evolution as the cluster grows
larger. While economies of agglomeration suggest increased benefits to members with
each additional firm, diseconomies of agglomeration suggest decreased benefits to
members with each additional firm. Folta et al. (2006) provide some support for the
presence of diseconomies in biotechnology clusters, finding that beyond a certain size
inflection point, the rate of patenting actually declined with each additional firm in a
cluster. We focus on four potential sources of diseconomy – congestion costs, input
prices, adverse selection, and local competition – and for each we provide a brief
explanation and analyse the relevance of the factor in each type of agglomeration.
1. Congestion costs. As areas draw more organizations, density increases, leading to what
some authors have labelled congestion costs (Prevezer, 1997). Congestion costs are a
function of the number of firms agglomerating within a region and the carrying capacity
of the area. The specific congestion costs with which we are concerned are those related
to increased traffic and transportation costs, making it difficult for employees and
customers to navigate within a region.
Such costs are consequential in SSAs because they may affect quality of life of
employees and potential employees by raising commute times, potentially lowering the
region’s ability to attract labour. They may also discourage the personal interaction
necessary to actualize knowledge spillovers. Congestion costs are relevant in demand-
side agglomerations because they have a direct effect on one of the key externalities
associated with agglomeration – it negatively influences customers trying to reduce
search costs. As search times and costs increase, shopping within the DSA becomes less
attractive to consumers, reducing the demand-heightening benefits of agglomeration.
Of course, some, if not most, of these costs may be due to urbanization and would
have occurred regardless of whether the agglomeration developed and prospered.
Accordingly, it is important to distinguish between the absolute level of congestion costs
and the marginal level of congestion costs attributable to agglomeration. The former will
be tied to the density of people and all economic activity in the region. The latter will
be tied to the marginal increase in density and economic activity in the region attrib-
utable to the agglomeration. In practice, it is difficult to calculate marginal congestion
costs. Nevertheless, we conceptually focus on marginal congestion costs and believe
they are likely to more rapidly develop in demand-side agglomerations than in supply-
side agglomerations.

© Blackwell Publishing Ltd 2009

378 B. T. McCann and T. B. Folta
Congestion within supply-side agglomerations is created not only by the spatial
concentration of the focal firms but also by the co-location of suppliers, supporting
institutions, and related-industry firms which drive externalities. Concern regarding
overcrowding remains a notable feature of some of the most prominent SSAs around the
world. As one example, a recent survey of Silicon Valley executives rated traffic conges-
tion and transportation issues as the second most important cost-of-living challenge
facing Silicon Valley employees (Silicon Valley Leadership Group, 2007). Similar con-
cerns can be noted in Cambridge, England. The relationship between the number of
firms and the development of congestion costs in SSAs, however, is weakened by the
relatively wide geographic limits that provide relief from congestion. The view that
supply-side agglomerations have a large carrying capacity is consistent with research
suggesting that they can extend to areas as broad as regions or states. The belief that
congestion may not become a major issue until a cluster grows especially large is also
supported by prior research noting congestion being a particular concern for ‘the densest
agglomerations’ (Fingleton et al., 2005, p. 284) or in ‘older clusters’ (Prevezer, 1997, p.
259). We also note that the examples discussed above are noteworthy in that while
absolute congestion costs may be high, the marginal congestion costs attributable to
agglomeration may be low because of the high levels of other economic activity there and
already dense population. Even if firms are concerned about absolute levels of congestion
costs, they may find relief by locating on the periphery of the region, if they can avoid a
loss in reputation and other costs from such a move.
While congestion costs have received attention in studies of supply-side agglomera-
tions, they have not featured prominently in demand-related work. Despite this inatten-
tion, we tend to believe that congestion effects will develop more quickly in DSAs due to
their lower carrying capacity resulting from their tighter geographic boundaries. Firms in
DSAs also have less flexibility to avoid congestion by locating on the periphery. Location
on the periphery of a DSA may significantly decrease attractiveness that is heavily
influenced by a central location. Furthermore, such relief may be unattainable if there
are limited options for expansion, such as in a highly urbanized area. In general, we
believe the potential for the more rapid development of marginal congestion costs to be
significant for demand-side agglomerations; in contrast, these costs will not become
consequential until supply-side agglomerations are very mature.
2. Rising input prices. Input prices for factors such as workers, material, services, and land
are relevant for any firm whether they are located in an agglomeration or not, but these
prices are particularly consequential in SSAs and DSAs because of their close relation-
ship to sources of agglomeration benefits. In clusters, access to specialized labour and
supply inputs is one key rationale for locating in a cluster, but increases in localized
demand for these factors would lead to rising input prices in the absence of any coun-
tervailing supply increases. Porter (2000, p. 22) argues that demand increases, however,
are met by equivalent increases in supply as labourers and suppliers move into the region:
‘the availability of specialized personnel, services, and components, and the number of
entities creating them, often is far greater at clusters than elsewhere despite the greater
competition’. This increased supply reduces upward price pressure.
Supply of at least one important input in demand-side agglomerations, however, is not
quite so elastic. The key rationale for locating in a DSA is to position the firm in a

© Blackwell Publishing Ltd 2009

 Demand- and Supply-Side Agglomerations 379
geographically attractive location for customers, and the acquisition costs of that attrac-
tive location will represent a significant concern for these firms. Localized demand for
land within the DSA area increases, but the supply of land cannot increase, unlike the
supply of labour and specialized inputs. This limited availability of attractive land within
the relatively tight geographic boundaries of the DSA will drive up price.
Overall, holding all else constant, rising input prices may be a less of a concern in
SSAs, at least until they become particularly large or dense, while these changing prices,
especially for land, may be more consequential in DSAs. Further lessening the relative
potential impact of rising input prices is the flexibility of new cluster members in deciding
where exactly to locate within the cluster given it broad base. In contrast, firms desiring
to join a demand-side agglomeration are more limited in how far from the centre of the
DSA they can locate.
3. Adverse selection. Shaver and Flyer (2000) raise another interesting source of potential
diseconomy as regions of co-locating organizations grow larger. They first note that
contributions to agglomeration externalities are asymmetric. For example, a large, highly
innovative firm with substantial R&D spending may create more knowledge spillovers
than a smaller, less innovative rival. Given that the ‘better’ firms have the most to lose
from locating in close geographic proximity to rivals, Shaver and Flyer (2000) contend
they have the most incentive to avoid co-location. This may lead to a growth cycle
characterized by adverse selection. Weak firms, who have the most to gain from
co-location, will be more drawn to the agglomerating region, while stronger firms, who
have the most to lose, will be more drawn to isolated locations. As this process plays out,
the average quality of organizations in the region will decline. To avoid this negative
cycle of adverse selection, there must be a way in which strong firms might benefit more
than weak firms.
Strong firms will benefit more in SSAs if they have relatively better access to beneficial
knowledge spillovers. If strong firms with large knowledge stocks also have higher
absorptive capacity, the ability to identify, assimilate and exploit knowledge from other
organizations (Cohen and Levinthal, 1990), they might be able to gain more knowledge
than they lose. Instead of adverse selection dominating, clusters might more reasonably
consist of strong firms with high absorptive capacity, creating a ‘win–win’ situation for
all.
We expect, however, that adverse selection pressures will be more salient in demand-
side agglomerations. Chung and Kalnins (2001) find that asymmetric contributions to
demand-side externalities exist, demonstrating an association between the presence
of large and high-resource organizations and agglomeration benefits.[12] Kalnins and
Chung (2004) further find that these asymmetries are associated with a cycle of adverse
selection in hotel location choices. High-resource entrants seek out DSAs of other
high-resource establishments but avoid DSAs of low-resource establishments. Over time,
high-resource firms will agglomerate with their counterparts; however, this concentra-
tion attracts low-resource firms. As more low-resource firms co-locate, the area becomes
increasingly unattractive to future high-resource entrants who will then prefer to locate
in isolation to avoid the low-resource firms. This ultimately will result in a preponderance
of low-resource firms in the original DSA. Holding all else constant, the potential for
adverse selection seems to be a larger concern in DSAs than in SSAs because firm types

© Blackwell Publishing Ltd 2009

380 B. T. McCann and T. B. Folta
that contribute more to the creation of demand-related externalities do not have the ability
to extract more benefits. In contrast, strong contributors to supply-related externalities
may also have a greater ability to extract benefits.
4. Increased localized competition. A number of studies (Baum and Mezias, 1992; Hannan
and Carroll, 1992; Ingram and Inman, 1996) find that organizations compete more
intensely with geographically proximate competitors. Agglomeration increases the
number of local organizations, which translates into increasing competitive pressures to
the extent that competition is truly localized. We consider two aspects of competition.
The first, which we shall refer to as product–market competition, refers to firms that
produce close substitutes to the products or services offered by a focal firm. Second, we
consider perceptions of competition, that is, who the focal firm perceives as its rivals.
Regarding product–market competition, we question the extent to which clustered firms
compete more heavily in output markets with geographically proximate rivals than with
isolated rivals. First, products produced in supply-side agglomerations tend to be differ-
entiated and produced by specialized labour with specialized inputs, which reduces
substitutability. Second, the opportunity space of SSA firms typically extends beyond
local boundaries, and that space is equally pursued by rivals inside and outside the
cluster. A biotechnology or semiconductor firm in Silicon Valley does not just compete
with other Silicon Valley firms. That is, the physical location of the firm does not have
a significant impact on the substitutability of products among firms. We note that many
of the localized competition studies have occurred in empirical settings less relevant to
SSA research, such as hotels (Baum and Mezias, 1992), day care centres (Baum and
Singh, 1994), and rural cooperative banks (Lomi, 1995). Finally (Porter, 2000, p. 18)
notes that many cluster members ‘are not direct competitors but rather serve different
segments of industries’.
In contrast, firm physical location does have a significant impact on the substitutability
of products and services offered by DSA firms, and we therefore expect that localized
product–market competition will play a more prominent role in demand-side agglomera-
tions as opposed to SSAs. Geographic limits in output markets are very relevant in DSAs.
Consider the typical products that appear in demand-side agglomerations (e.g. women’s
shoes, automobiles, hotels). The likely market reach of a particular firm is the surrounding
geographic area extending radially from the focal firm. That is, a women’s shoe retailer or
hotel in Boston does not compete with a women’s shoe retailer or hotel in New York. Firms
who join a DSA will have higher degrees of overlap in market reach while isolated firms
will exhibit lower levels of overlap. Hence, localized product–market competition will be
a salient source of diseconomy in DSAs. This is consistent with Fischer and Harrington’s
(1996) view that the main cost to joining a demand-side agglomeration is that the close
proximity of firms intensifies price competition.
The above comparison of the relative impact of localized competition in SSAs and
DSAs should not be interpreted as an argument that localized competition is irrelevant
in clusters. As we noted above, competition can also be discussed in terms of the perceptions
of focal firms. Prior research does indicate that firms may form mental models of
competition that focus on more proximate rivals. For example, Porac et al. (1989)
describe how Scottish knitwear manufacturers tend to focus on other Scottish manufac-
turers as their competition. Pouder and St. John (1996) contend that the higher salience

© Blackwell Publishing Ltd 2009

 Demand- and Supply-Side Agglomerations 381
of the actions of proximate rivals can lead to homogenized competitive outlooks over
time within a cluster. Firms’ mental models become biased towards local competitors
and directed away from competitors outside the region, eventually leading to ‘insularity,
sluggishness, and failed attempts to innovate’ (Pouder and St. John, 1996, p. 1209). While
perceptions of competition clearly matter, we see no obvious reasons why the impact of
perceived competition should be stronger or weaker within SSAs or DSAs.
With no obvious difference in the effect of perceived competition and a more negative
effect of product–market competition in DSAs (driven by the key role geography plays in
the substitutability of products/services), localized competition would appear to be a
more significant diseconomy force in demand-side agglomerations.

The life cycles of SSAs and DSAs. The life cycle of an agglomeration is related to at least two
factors: how quickly the benefits are realized initially and how fast diseconomies start to
affect the agglomerating firms. We believe that SSAs should take longer (i.e. require
more firms) to initially develop, primarily due to the more complex nature of clusters.
Benefits flow from a complex web of inter-relationships among focal firms, related firms,
suppliers, customers, and supporting institutions. Such a web will take time to develop.
We have also argued above that clusters are generally less prone to rapid development
of diseconomies.
In contrast to clusters, we expect that DSA benefits will materialize rather quickly. No
complex web of relationships must form – all that is needed is a grouping of firms that
consumers recognize and patronize. Diseconomy forces, however, likely affect DSAs
more quickly than they do SSAs.
The above discussion addresses how the value of SSAs and DSAs differs as they grow
larger and as they age. This discussion treats time in a relative sense in that age is
measured relative to the start of the agglomeration. But, we also believe that agglom-
eration benefits will change over time when time is viewed in more of an absolute sense.
We next ask the question of how the value of SSAs and DSAs has changed over the past
say 30 years and how it might continue to change in the future.

Temporal changes in the value of SSAs and DSAs. We finally consider how the value of locating
within a supply- or demand-side agglomeration relative to locating in isolation may
change over time. We offer several factors related to technological advancement that
imply increasing value for SSAs (relative to isolation) over time and decreasing value for
DSAs.
With the increasing influence of technology on the ability to conduct business over
distances, a natural first inclination might be to believe that SSAs have become less
important over time. Falling physical and information transport costs related to techno-
logical advance may have diminished locational influences on competition; however, a
host of new influences increase the importance of cluster-based benefits in the increas-
ingly global and knowledge-based competitive environment.
In regard to the issue of specialized labour, Krugman (1996, p. 195) notes that ‘it
seems undeniable that over the past 20 years the advanced nations have experienced
technological change that is strongly biased in favor of skilled workers’. This bias is driven
not by a change in what goods and services are being produced but rather a change in

© Blackwell Publishing Ltd 2009

382 B. T. McCann and T. B. Folta
the means of production. So not only are specialized workers becoming increasingly
important but so are specialized inputs to the production process. Competitive advan-
tage today requires higher productive use of inputs – greater and more efficient use of
specialized labour and inputs.
Hand in hand with the need to increase productivity is rising pressure to innovate.
Audretsch (2003, p. 30) posits that the key to successfully responding to globalization
pressures in Europe and North America is to shift economic activity into ‘knowledge-
based economic activity’. Facilitating innovation and increasing the production and
utility of knowledge are one of the main theorized benefits of supply-side agglomerations.
The concentration of knowledge-generating activities within SSAs can also serve as a
key source of entrepreneurial opportunities to spur new venture creation (Audretsch and
Keilbach, 2007). As successful competition becomes more dependent on knowledge and
innovation, the knowledge-related benefits of clusters increase as well.[13]
In contrast to extant theorizing regarding the increasing value of clusters as technolo-
gies advance, little work has considered how technology may affect the value of DSAs. As
an initial foray into this issue, we believe that at least two factors related to advancing
technology have the potential to impact the value of DSAs relative to isolated locations.
Since the value of being in a demand-side agglomeration is directly related to the
magnitude of consumer search costs, any exogenous changes that reduce or eliminate the
costs associated with searching may reduce the relative advantage of DSAs. We consider
two such changes associated with declining search costs over time: improvements in
transportation and growth in information-gathering alternatives.
As it becomes less costly to travel to visually inspect goods, search costs fall. Thus, as
transportation infrastructure improves we expect that firms may realize less relative
benefit from agglomerating. Consider a hypothetical consumer in a market made up of
three dispersed stores and three stores located within an agglomeration. If the consumer
wishes to search across three stores and she incurs travel time of 20 minutes to visit a
location (either one of the dispersed stores or the agglomeration), including the agglom-
eration in her search reduces total travel time from 60 minutes to 20 minutes, a net
saving of 40 minutes, which represents the advantage of the agglomeration over the
dispersed locations. If improvements in transportation reduce travel times to 10 minutes
per location, the net advantage of the agglomeration falls to 20 minutes (30 minutes to
visit three dispersed locations versus 10 minutes to visit the agglomeration). The relative
advantage of the agglomeration is reflected in the differential time savings, which is
decreasing in travel time. Chung and Kalnins (2001) argue that such differential trans-
portation costs are a key reason why demand-related agglomeration effects are stronger
in rural areas than in urban areas. Ingram and Inman (1996) make a similar argument
in their study of hotels in the Niagara Falls area to help explain the finding that distance
from the Falls matters less over time.
A second exogenous change affecting consumer search costs over time is the growth of
alternative means of gathering information. Search costs obviously drop dramatically if
goods can be inspected without having to physically visit each store. Internet technology
allows customers to virtually inspect the products that firms have to offer, thus permitting
detailed comparisons without the need to ever travel further than the short walk to the
nearest computer. While virtual search may not perfectly replace physical search,

© Blackwell Publishing Ltd 2009

 Demand- and Supply-Side Agglomerations 383
especially given the potential experiential characteristics of the search and consumption
process, it does provide a partial substitute that potentially reduces the benefits of
physical co-location. Consideration of the hotel industry, where DSAs are common,
reveals the potential for the internet to reduce search costs. Information on particular
hotels, including property and room photos is all accessible online, and hotel ratings by
other consumers further facilitate product comparisons.
Another typical category of DSA firms, automobile dealers, faces similar trends.
Manufacturer websites allow consumers to view new cars in nearly any configuration
imaginable. Dealer sites offer photos of actual on-site inventory. As larger quantities and
higher quality of information becomes available online, the need to travel from dealer-
ship to dealership declines. Some buyers even complete a purchase transaction without
ever visiting a dealership.
Overall then, as economies advance and transportation and information-gathering
options increase, consumer search costs may be less influenced by localization. Hence,
the value of locating within a demand-related agglomeration versus an isolated location
may decline over time and continue to do so with further technological progression. The
relationship between technological advancement and value of DSAs also suggests that
differences may exist in the value of DSAs in advanced versus developing economies.
The ability to search wider alternatives is lower in developing economies because of the
lack of transportation alternatives and the less developed transportation infrastructure.
Thus, the advantage of a DSA over isolated locations may become more pronounced.
DSA value in developing economies versus advanced economies may be further
enhanced by the lack of electronic alternatives to physical search. Consumers in less
developed economies have lower ability to replace search via physical travel/inspection
with search via virtual inspection.

DISCUSSION AND IMPLICATIONS

In this paper, we have sought to make the record more clear about the differences
between demand- and supply-side agglomerations. Given our argument that these two
manifestations of the agglomeration phenomenon have unique sources of externalities,
which generally do not intertwine, we are concerned that researchers studying agglom-
eration are improperly generalizing theoretical implications to both SSAs and DSAs,
when in fact, their theory is likely to apply to only one. If we are to develop sound theory
and empirical research around agglomeration and the benefits to geographic location,
we believe it is important for scholars not only to recognize and understand these
differences but also to account for these differences in the design and implementation of
their theoretical and empirical research. Our arguments are applicable to a number of
research streams. Whether research on agglomeration is being conducted from an
organizational ecology, international business, strategic management or entrepreneur-
ship perspective, it is essential for scholars to appreciate the differences between these two
phenomena. To elucidate the importance of the distinctions we have made thus far, we
offer some additional implications for conducting theoretical research summarized in
Table II and also introduce some implications for empirical research in the future
summarized in Table III.

© Blackwell Publishing Ltd 2009

384 B. T. McCann and T. B. Folta
Table II. Additional issues for future theoretical research

SSAs DSAs

What type of firms might benefit • Firms least able to attract • Firms least able to attract
most from agglomeration? specialized labour and input demand on their own
• Firms least able to internally
generate knowledge
What types of firms might • Large firms because they • Large firms because they attract
generate larger externalities? provide greater attraction for greater numbers of customers
labour and input providers? • Upscale firms because they
• Small firms because they are signal positive aspects of
more open and innovative agglomeration (e.g. safety of a
generating larger knowledge hotel location)
spillovers? • Other attributes that contribute
to higher demand spillovers
What is the form of the • Non-linear relationships likely • Non-linear relationships likely
agglomeration–performance • Positive externalities may • Positive externalities may
relationship? require greater time to develop develop fairly quickly
• Diseconomies may not develop • Diseconomies may develop
until SSAs become particularly earlier
large or dense
What complementary theories • Network theory • Transaction costs
might help further understand • Knowledge-based view • Industrial organization
agglomeration? • Others focused on social • Others focused on efficiency
interaction

Additional Implications for Future Theoretical Research

Understanding firm asymmetries. A fundamental issue to scholars of management is how
heterogeneity among firms might impact the relationship between agglomeration and
performance. We believe that recognizing the different underlying causal mechanisms
driving the agglomeration–performance relationship in SSAs and DSAs will influence
theoretical predictions on the role of heterogeneity. A first consideration is whether some
firms might benefit more from agglomeration than others. It strikes us that the difference
in the types of externalities generated in DSAs and SSAs may significantly bear upon this
question. For example, Shaver and Flyer (2000) argue that weaker firms benefit more
from clustering leading to adverse selection; however, the characteristics that define
‘weak’ firms may vary across these two agglomeration types based on the source of
externalities. In DSAs, weak firms may be those who are least able to attract demand on
their own. In contrast, weak firms in SSAs might be those who are least able to attract
specialized labour or specialized input providers and generate knowledge on their own.
Further research is necessary to identify what firms might benefit most from different
types of agglomeration, and we believe that the differing nature of the underlying
externalities will impact the direction of this research.

© Blackwell Publishing Ltd 2009

 Demand- and Supply-Side Agglomerations 385
Table III. Issues for empirical analysis

SSAs DSAs

What type of industries • Specific industries focusing on • Specific industries focusing on

might be suitable for industrial goods: examples include consumer goods where physical
studying agglomeration? software, biotechnology, visits are common: examples
semiconductors, manufacturing, include hotels, restaurants, auto
etc dealers, antique stores, flea
markets, etc
How might the benefits • Improved innovation • Higher revenue
of agglomeration be • Lower transaction costs for • More customers
observed at the securing employees
firm-level? • Improved access to private equity
• Improved access to strategic
partners
At what unit of analysis • Metropolitan Statistical Area • Zip code
should agglomeration • Region • Identified regions/neighbourhoods
effects be observed? • State within metropolitan areas
• Use of clustering algorithm based • Use of clustering algorithm based
on distance on distance
How might agglomeration • Number of firms in SSA for • Number of firms in DSA for
effects be captured in industry industry
multivariate regression • Number of firms in SSA for • Number of firms in DSA for
techniques? complementary industries complementary industries
• Number of specialized employees • Revenues in DSA for industry
in region • Revenues in DSA for
• Revenues in SSA for industry complementary industries
• Revenues in SSA for
complementary industries
What contingent • SSA size squared • DSA size squared
relationships • SSA age • Infrastructure that may influence
might influence • Presence of complementary customer search costs
agglomeration effects industries in SSA • Presence of complementary
for firms? • Firm social connectivity industries in DSA
• Firm absorptive capacity • Firm reputation among customers
• Firm reputation for attracting • Degree of competitive behaviour
quality employees • Firm attributes such as size,
• Availability of private capital in low-cost strategy, etc
cluster
• Firm age

Heterogeneity may also exist in how much individual firms contribute to agglomera-
tion externalities, and again we believe that the distinct causal mechanisms play a role.
In DSAs, it appears that large firms and upscale firms contribute more (Canina et al.,
2005; Chung and Kalnins, 2001), perhaps due to their greater demand-generating
abilities. In SSAs, it is less clear who the differential contributors are. Shaver and Flyer
(2000) suggest that large firms may contribute more due to their higher ability to attract

© Blackwell Publishing Ltd 2009

386 B. T. McCann and T. B. Folta
specialized labour and input providers, but this assertion has not yet been directly tested.
Some empirical evidence (e.g. Rosenthal and Strange, 2003) suggests that it may be small
firms who contribute most to the creation of supply-side externalities, a finding that
would be consistent with the idea that small firms are more open and innovative,
generating greater knowledge spillovers. The key point is that our conclusions about the
role of firm heterogeneity in the agglomeration–performance relationship are inextrica-
bly linked to the differences in the theoretical mechanisms we have highlighted in this
research.

Understanding the form of the agglomeration–performance relationship. Researchers should also be

aware that relationships between the number of firms in an agglomeration and perfor-
mance may not be simply linear. The form of the non-linearity might also vary across
the two agglomeration types given the differences we have discussed. The operation of
diseconomies and establishing the point at which they become relevant appears to be a
promising area for future research. The role of technology as a moderating influence
may also impact the agglomeration–performance relationship. While it seems possible
that increasing technology hurts DSAs as it makes consumer search easier, thereby
reducing the demand-based advantages of agglomeration, its impact on SSAs is less
straightforward. On one hand, technology advances may facilitate communication and
reduce the need for proximity to actualize knowledge spillovers. On the other hand, it
tends to push competition to become more knowledge-based and to require more use of
specialized labour and inputs.

Selecting complementary theories to further explain agglomeration. The fundamental differences

between demand and supply-side agglomerations are significant in that they suggest
theories may have different degrees of applicability to both. For example, theories
pertinent to social interaction may be more applicable for studying supply-side agglom-
erations given that firms are more interconnected in their relationships with one another
up and down the value chain. Consequently, we might look to network theory and the
knowledge-based view of the firm for theoretical insight in supply-side agglomerations. In
contrast, theories pertinent to economic efficiency may be most pertinent to demand-side
agglomerations given the close geographic proximity and lack of intense relationship
between firms. Consequently, transaction cost theory and industrial organization may
most adequately characterize competitive decisions in demand-side agglomerations.

Implications for Future Empirical Research

Selecting datasets. Researchers should carefully select industries that are appropriate for
studying either demand-side externalities or supply-side externalities. These are not
likely to be the same industry; the industry choice should be consistent with the specific
theorized externalities.

Measuring the benefits of agglomeration. With the diverse sources of agglomeration externali-
ties, what constitutes an appropriate dependent variable is directly impacted by the type
of agglomeration being studied. With benefits flowing from access to specialized labour,

© Blackwell Publishing Ltd 2009

 Demand- and Supply-Side Agglomerations 387
specialized inputs, and knowledge spillovers, SSAs should offer a relatively broader set of
dependent variables. For example, evidence of higher levels of knowledge creation, such
as patents or product innovations, are appropriate choices of dependent variables in
clusters. We might also expect clustered firms to have cost advantages over isolated firms
if technology is focused on process improvements. With a sole benefit source of height-
ened demand, researchers have a more limited choice of dependent variables when
investigating demand-side agglomerations. Revenue is likely the most relevant measure
of agglomeration advantages for DSA firms.
Operationalizing agglomerations. While SSAs are most often operationalized by counting
firms within a certain geographic area, this relatively simple definition is more relevant
to demand-side as opposed to supply-side agglomerations. As the strength of the SSA
effect is also directly related to the number and quality of supporting institutions, they
should also be considered in operationalizations. No such support structure is necessary
for a DSA of firms, as the benefits are generated by the firms themselves, not the
supporting firms the group draws to the area. Whether particular types of firms contrib-
ute more to externality generation will also impact operationalizations. Finally, we have
argued that the geographic area over which a researcher chooses to sum the number of
similar firms will be quite different for the two phenomena. Boundaries can be more
widely drawn in SSAs but need to be more narrowly drawn in DSAs.

Conclusion
A large body of research has developed in an attempt to explain the role geographic
location plays in economic performance. We have argued that our quest to more fully
understand this relationship will continue to be retarded without a clear distinction
between fundamentally different types of agglomerations. Although they share a
common heritage, the phenomena of demand-based and supply-based agglomeration
are quite different.
Answers to fundamental questions in the agglomeration research stream differ across
SSAs and DSAs. Different types and densities of firms compose SSAs and DSAs, and
they are found in different industries. Moreover, they have dissimilar geographic bound-
aries and disparate levels of relationship complexity. In addition to trait-based distinc-
tions, we have also argued that how benefits are realized fundamentally differs between
supply- and demand-side agglomerations. Sources of diseconomies vary, and we have
speculated that the gains from SSAs are likely to have grown over time while the gains
of DSAs have likely fallen. Finally, we expect that the life cycles of the two types of
agglomeration are different.
Engaging in a rigorous comparison also allows us to offer insights into open research
questions regarding clusters. In particular, this paper responds to the call of authors like
Audretsch (2003, p. 44), who notes that ‘another important but relatively unchartered
area for future research involves the life cycle of spatial units’, and Folta et al. (2006, p.
240), who argue that ‘more work is needed to examine older industries and the possibility
that diseconomies of agglomeration dominate at a certain point’. Our comparisons also
help bring greater clarity to the definition of clusters, a topic of continuing debate
(Martin and Sunley, 2003).

© Blackwell Publishing Ltd 2009

388 B. T. McCann and T. B. Folta
Acknowledging the differences between these two different manifestations of the
agglomeration phenomenon will lead to a more coherent theoretical and empirical
record while also influencing future methodological choices. In this case, one size does
not fit all. A phenomenon that applies in such diverse settings as Silicon Valley to rural
hotels to fruit sellers in third world markets cries out for improved classification and
clarification.

ACKNOWLEDGMENTS
We are grateful for the extensive comments received from General Editor Mike Wright and three anony-
mous reviewers. We also thank Stephen Tallman for his comments and suggestions on an earlier draft.

NOTES
[1] For review of this literature, see Fujita et al. (1999) and Fujita and Thisse (2002).
[2] Agglomeration researchers typically view relatedness in one of two ways. Relatedness may be defined
to encompass situations in which co-locating firms share an underlying technology or product focus not
constrained to a particular industry. For example, Porter (1998b) cites the example of over 400
Massachusetts’ companies focused on medical device products whose industries vary from electronic
equipment to plastic products as an example of a product-focused agglomeration. Relatedness is often
more narrowly viewed as firms within a particular industry (firms which produce goods that are at least
close substitutes) who co-locate. Much of the literature in this area involves studies of spatial concen-
trations within individual industries.
[3] For a detailed review of this literature, see McCann and Folta (2008).
[4] We also note that a colleague who has published extensively on supply-related agglomerations shared
with us that he often receives reviewer comments asking him to inform his work with studies of
demand-related agglomerations.
[5] Typologies are based on conceptual constructions while taxonomies consist of empirical entities that may
be classified into exhaustive and mutually exclusive categories based on observable and measurable
characteristics (Bailey, 1994).
[6] We thank an anonymous reviewer for encouraging us to more carefully consider the potential for
demand-related externalities in business service clusters.
[7] The four types of agglomerations are: (i) ‘Marshallian new industrial district’ made up of a number of
small but similarly-sized firms who exhibit levels of regional cooperation; (ii) ‘hub-and-spoke district’,
in which the regional structure revolves around one or several major corporations in one or a few
industries; (iii) ‘satellite industrial platform’, which is an area composed chiefly of branch plants of
absent multinational corporations; and (iv) ‘state-centred district’ in which a major government tenant
anchors the regional economy (like a state capital, key military or research facility, etc). We would argue
that the first and third are examples of agglomerations that include supply-side externalities; in the
hub-and-spoke district, the hub firms may generate supply-side externalities while the location deci-
sions of the firms along the spokes involve a source of benefits (a major customer) that is not increasing
with the number of spoke firms; the state-centred district also appears to be driven by an exogenous
benefit source of a major customer.
[8] We do note, however, that some scholars do not believe connections among firms within a cluster are
necessary for those firms to benefit from the enhanced knowledge generated within a cluster. Maskell
(2001) argues that firms can accrue knowledge benefits even in the absence of interactions among
related firms. The benefits flow from increased variation of product solutions created by clustered firms,
which competitors capitalize on through simple monitoring of those firms.
[9] Ingram and Roberts (2000) also found evidence of friendships among managers of high-end hotels in
Sydney, Australia. They further speculate that the association between friendships and performance
that they discover may be related to these referral practices.
[10] Kalnins (2006) notes that the relatively fixed capacity of hotels (at least in the short to medium term)
helps facilitate networks and ties between owner/managers of hotels. This industry-specific condition
may help explain why we might not see these relationships in other DSA industries.

© Blackwell Publishing Ltd 2009

 Demand- and Supply-Side Agglomerations 389
[11] We do acknowledge that geographical proximity may enhance the ability of firms to engage in implicit
collusion, which might require more tacit relationships; however, sustaining collusive behaviour would
be difficult as the number of firms within an agglomeration grows.
[12] We do note a concern with Chung and Kalnins’ (2001) study. The authors’ agglomeration hypotheses
are both based on the number of large and chain-affiliated hotels within a market while their
operationalization of these measures is based on fractional measures. Moreover, their operationaliza-
tions fail to account for the overall size of the market. As the literature and Chung and Kalnins’ own
arguments clearly suggest that agglomeration is market invariant and depends on the number rather
than the fraction of large size hotels, we believe the number of large firms instead of the fraction of large
firms offered a better measure. Curiously, in a follow-up study on the Texas lodging industry, they
actually based all their hypotheses on the numbers of low- and high-resource hotels in a market
(Kalnins and Chung, 2004), as we note here. Our citations to Chung and Kalnins (2001) are based on
the assumption that their results would still hold despite these concerns. We thank an anonymous
reviewer for emphasizing this concern.
[13] We do note, however, that some authors would argue that technology advances potentially reduce the
importance of proximity in realizing knowledge spillovers. As a caution against the view that technol-
ogy can replace local ties, we note that knowledge spillovers are not always the result of an act of clear
planning but often occur by happenstance. Firms locating in close proximity to each other increases
social interactions, which results in a higher likelihood of information and knowledge sharing regardless
of whether the interaction was designed to accomplish this purpose. Technology may be able to help
replace the intentional sharing of tacit information but it faces a much greater challenge in recreating
the social conditions that facilitate unintentional sharing.

REFERENCES
Almeida, P. and Kogut, B. (1997). ‘The exploration of technological diversity and the geographic localiza-
tion of innovation’. Small Business Economics, 9, 21–31.
Almeida, P. and Kogut, B. (1999). ‘Localization of knowledge and the mobility of engineers in regional
networks’. Management Science, 45, 905–17.
Arthur, W. (1990). ‘Silicon Valley locational clusters: when do increasing returns imply monopoly?’. Math-
ematical Social Sciences, 19, 235–73.
Audretsch, D. B. (2003). ‘Managing knowledge spillovers: the role of geographic proximity’. In Baum,
J. A. C. and Sorenson, O. (Eds), Geography and Strategy: Advances in Strategic Management. Oxford: JAI Press,
20, 23–48.
Audretsch, D. B. and Feldman, M. P. (1996). ‘R&D spillovers and the geography of innovation and
production’. American Economic Review, 86, 630–40.
Audretsch, D. B. and Keilbach, M. (2007). ‘The theory of knowledge spillover entrepreneurship’. Journal of
Management Studies, 44, 1242–54.
Audretsch, D. B. and Stephan, P. E. (1996). ‘Company-scientist locational links: the case of biotechnology’.
American Economic Review, 86, 641–52.
Bailey, K. D. (1994). Typologies and Taxonomies: An Introduction to Classification Techniques. Thousand Oaks, CA:
Sage.
Baum, J. A. C. and Haveman, H. A. (1997). ‘Love thy neighbor? Differentiation and agglomeration in the
Manhattan hotel industry, 1898–1990’. Administrative Science Quarterly, 42, 304–38.
Baum, J. A. C. and Mezias, S. J. (1992). ‘Localized competition and organizational failure in the Manhattan
hotel industry, 1898–1990’. Administrative Science Quarterly, 37, 580–604.
Baum, J. A. C. and Singh, J. V. (1994). ‘Organizational niches and the dynamics of organizational mortality’.
American Journal of Sociology, 100, 346–80.
Brown, S. (1989). ‘Retail location theory – the legacy of Hotelling, Harold’. Journal of Retailing, 65, 450–70.
Canina, L., Enz, C. A. and Harrison, J. S. (2005). ‘Agglomeration effects and strategic orientations: evidence
from the US lodging industry’. Academy of Management Journal, 48, 565–81.
Cantwell, J. A. and Iammarino, S. (2001). ‘EU regions and multinational corporations: change, stability and
strengthening of technological comparative advantage’. Industrial and Corporate Change, 10, 1007–37.
Chen, T. J. (2003). ‘Network resources for internationalization: the case of Taiwan’s electronics firms’.
Journal of Management Studies, 40, 1107–30.
Chung, W. and Kalnins, A. (2001). ‘Agglomeration effects and performance: a test of the Texas lodging
industry’. Strategic Management Journal, 22, 969–88.

© Blackwell Publishing Ltd 2009

390 B. T. McCann and T. B. Folta
Coe, N. M., Kelly, P. F. and Yeung, H.W-C. (2007). Economic Geography: A Contemporary Introduction. Malden,
MA: Blackwell.
Cohen, W. M. and Levinthal, D. A. (1990). ‘Absorptive capacity – a new perspective on learning and
innovation’. Administrative Science Quarterly, 35, 128–52.
Cooper, A. and Folta, T. (2000). ‘Entrepreneurship and high-technology clusters’. In Sexton, D. L. and
Landstrom, H. (Eds), The Blackwell Handbook of Entrepreneurship. Oxford: Blackwell, 349–67.
Daft, R. L. and Lengel, R. H. (1986). ‘Organizational information requirements, media richness and
structural design’. Management Science, 32, 554–71.
Doty, D. H. and Glick, W. H. (1994). ‘Typologies as a unique form of theory building: toward improved
understanding and modeling’. Academy of Management Review, 19, 230–51.
Eaton, B. C. and Lipsey, R. G. (1979). ‘Comparison shopping and the clustering of homogeneous firms’.
Journal of Regional Science, 19, 421–35.
Ellison, G. and Glaeser, E. L. (1999). ‘Evolution of geographic concentration of industry – the geographic
concentration of industry: does natural advantage explain agglomeration?’. American Economic Review,
89, 311–16.
Feller, I. (1997). ‘Federal and state government roles in science and technology’. Economic Development
Quarterly, 11, 283–96.
Fingleton, B., Igliori, D. and Moore, B. (2005). ‘Cluster dynamics: new evidence and projections for
computing services in Great Britain’. Journal of Regional Science, 45, 283–311.
Fischer, J. H. and Harrington, J. E. Jr (1996). ‘Product variety and firm agglomeration’. The Rand Journal of
Economics, 27, 281–309.
Folta, T., Cooper, A. C. and Baik, Y.-S. (2006). ‘Geographic cluster size and firm performance’. Journal of
Business Venturing, 21, 217–42.
Fujita, M. and Thisse, J.-F. (2002). Economics of Agglomeration. Cities, Industrial Location, and Regional Growth.
Cambridge and New York: Cambridge University Press.
Fujita, M., Krugman, P. and Venables, A. (1999). The Spatial Economy: Cities, Regions, and International Trade.
Cambridge, MA: MIT Press.
Graitson, D. (1982). ‘Spatial competition a la Hotelling – a selective survey’. Journal of Industrial Economics, 31,
13–25.
Hannan, M. T. and Carroll, G. R. (1992). Dynamics of Organizational Populations: Density, Legitimation, and
Competition. New York: Oxford University Press.
Hay, D. A. (1976). ‘Sequential entry and entry-deterring strategies in spatial competition’. Oxford Economic
Papers – New Series, 28, 240–57.
Head, K., Ries, J. and Swenson, D. (1995). ‘Agglomeration benefits and location choice: evidence from
Japanese manufacturing investments in the United States’. Journal of International Economics, 38, 223–47.
Henderson, R. and Cockburn, I. (1994). ‘Measuring competence – exploring firm effects in pharmaceutical
research’. Strategic Management Journal, 15, 63–84.
Hotelling, H. (1929). ‘Stability in competition’. Economic Journal, 39, 41–57.
Ingram, P. and Inman, C. (1996). ‘Institutions, intergroup competition, and the evolution of hotel popula-
tions around Niagara Falls’. Administrative Science Quarterly, 41, 629–58.
Ingram, P. and Roberts, P. (2000). ‘Friendships among competitors in the Sydney hotel industry’. American
Journal of Sociology, 106, 387–423.
Inkpen, A. C. and Tsang, E. W. K. (2005). ‘Social capital, networks, and knowledge transfer’. Academy of
Management Review, 30, 146–65.
Jacobs, J. (1969). The Economy of Cities. New York: Vintage.
Jaffe, A. B., Trajtenberg, M. and Henderson, R. (1993). ‘Geographic localization of knowledge spillovers as
evidenced by patent citations’. Quarterly Journal of Economics, 108, 577–98.
Kalnins, A. (2006). ‘Markets: the U.S. lodging industry’. Journal of Economic Perspectives, 20, 203–18.
Kalnins, A. and Chung, W. (2004). ‘Resource-seeking agglomeration: a study of market entry in the lodging
industry’. Strategic Management Journal, 25, 689–99.
Kalnins, A. and Chung, W. (2006). ‘Social capital, geography, and survival: Gujarati immigrant entrepre-
neurs in the US lodging industry’. Management Science, 52, 233–47.
Keeble, D. and Wilkinson, F. (1999). ‘Collective learning and knowledge development in the evolution of
regional clusters of high technology SMEs in Europe’. Regional Studies, 33, 295–303.
Krugman, P. (1991a). ‘Increasing returns and economic geography’. The Journal of Political Economy, 99,
483–99.
Krugman, P. (1991b). Geography and Trade. Cambridge, MA: MIT Press.
Krugman, P. (1996). Pop Internationalism. Cambridge, MA: MIT Press.

© Blackwell Publishing Ltd 2009

 Demand- and Supply-Side Agglomerations 391
Lomi, A. (1995). ‘The population ecology of organizational founding – location dependence and unobserved
heterogeneity’. Administrative Science Quarterly, 40, 111–44.
Markusen, A. (1996). ‘Sticky spaces in slippery space: a typology of industrial districts’. Economic Geography, 72,
293–313.
Marshall, A. (1920). Principles of Economics, 8th edition. London: Macmillan.
Martin, R. and Sunley, P. (2003). ‘Deconstructing clusters: chaotic concept or policy panacea?’. Journal of
Economic Geography, 3, 5–35.
Maskell, P. (2001). ‘Towards a knowledge-based theory of the geographic cluster’. Industrial and Corporate
Change, 10, 921–43.
McCann, B. T. and Folta, T. B. (2008). ‘Location matters: where we have been and where we might go in
agglomeration research’. Journal of Management, 34, 532–65.
Meyer, K. E. and Nguyen, H. V. (2005). ‘Foreign investment strategies and sub-national institutions in
emerging markets: evidence from Vietnam’. Journal of Management Studies, 42, 63–93.
Orlando, M. (2004). ‘Measuring spillovers from industrial R&D: on the importance of geographic and
technological proximity’. Rand Journal of Economics, 35, 777–86.
Pandit, N. R., Cook, G. A. S., Beaverstock, J. V., Pain, K. and Taylor, P. (2004). ‘The clustering of financial
services in London’. 44th Congress of the European Regional Science Association, 25–29 August 2004,
Porto, Portugal.
Porac, J. F., Thomas, H. and Baden-Fuller, C. (1989). ‘Competitive groups as cognitive communities: the
case of Scottish knitwear manufacturers’. Journal of Management Studies, 26, 397–416.
Porter, M. E. (1998b). ‘Clusters and the new economics of competition’. Harvard Business Review, 76, 77–
90.
Porter, M. E. (1998a). On Competition. Cambridge, MA: Harvard Business School Press.
Porter, M. E. (2000). ‘Location, competition, and economic development: local clusters in a global
economy’. Economic Development Quarterly, 14, 15–34.
Pouder, R. and St. John, C. H. (1996). ‘Hot spots and blind spots: geographical clusters of firms and
innovation’. Academy of Management Review, 21, 1192–225.
Prescott, E. C. and Visscher, M. (1977). ‘Sequential location among firms with foresight’. Bell Journal of
Economics, 8, 378–93.
Prevezer, M. (1997). ‘The dynamics of industrial clustering in biotechnology’. Small Business Economics, 9,
255–71.
Rocha, H. O. and Sternberg, R. (2005). ‘Entrepreneurship: the role of clusters theoretical perspectives and
empirical evidence from Germany’. Small Business Economics, 24, 267–92.
Rosenkopf, L. and Almeida, P. (2003). ‘Overcoming local search through alliances and mobility’. Management
Science, 49, 751–66.
Rosenthal, S. S. and Strange, W. C. (2001). ‘The determinants of agglomeration’. Journal of Urban Economics,
50, 191–229.
Rosenthal, S. S. and Strange, W. C. (2003). ‘Geography, industrial organization, and agglomeration’. Review
of Economics and Statistics, 85, 377–93.
Sammarra, A. and Biggiero, L. (2008). ‘Heterogeneity and specificity of inter-firm knowledge flows in
innovation networks’. Journal of Management Studies, 45, 800–29.
Saxenian, A. (1994). Regional Advantage: Culture and Competition in Silicon Valley and Route 128. Cambridge, MA:
Harvard University Press.
Shaver, J. M. and Flyer, F. (2000). ‘Agglomeration economies, firm heterogeneity, and foreign direct
investment in the United States’. Strategic Management Journal, 21, 1175–93.
Silicon Valley Leadership Group (2007). 2007 CEO Business Climate Survey. Available at: http://svlg.net/
press/library/ceosurvey_2007.pdf (accessed 16 March 2008).
Sorenson, O. and Baum, J. A. C. (2003). ‘Editors’ introduction: Geography and strategy: the strategic
management of space and place’. In Baum, J. A. C. and Sorenson, O. (Eds), Geography and Strategy:
Advances in Strategic Management. Oxford: JAI Press, 20, 1–22.
St John, C. H. and Pouder, R. W. (2006). ‘Technology clusters versus industry clusters: resources, networks,
and regional advantages’. Growth and Change, 37, 141–71.
Stahl, K. (1982). ‘Differentiated products, consumer search, and locational oligopoly’. Journal of Industrial
Economics, 31, 97–113.
Storper, M. (1995). ‘The resurgence of regional economies, ten years later: the region as a nexus of untraded
interdependencies’. Journal of European Urban and Regional Studies, 2, 191–221.
Stuart, C. (1979). ‘Search and the spatial organization of trading’. In Lippman, S. A. and McCall, J. J. (Eds),
Studies in the Economics of Search. Amsterdam: North Holland, 17–34.

© Blackwell Publishing Ltd 2009

392 B. T. McCann and T. B. Folta
Stuart, T. and Sorenson, O. (2003). ‘The geography of opportunity: spatial heterogeneity in founding rates
and the performance of biotechnology firms’. Research Policy, 32, 229–53.
Tallman, S., Jenkins, M., Henry, N. and Pinch, S. (2004). ‘Knowledge, clusters, and competitive advantage’.
Academy of Management Review, 29, 258–71.
Weber, M. (1949). The Methodology of the Social Sciences. Glencoe, IL: The Free Press.
Zucker, L. G., Darby, M. R. and Brewer, M. B. (1998). ‘Intellectual human capital and the birth of U.S.
biotechnology enterprises’. American Economic Review, 88, 290–306.

© Blackwell Publishing Ltd 2009