Buzz Face-To-Face Contact and The Urban Economy PDF

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/5213298
Buzz: Face-To-Face Contact and the Urban Economy
Article in Journal of Economic Geography · August 2004

DOI: 10.1093/jnlecg/lbh027 · Source: RePEc
CITATIONS READS
1,533 1,229
2 authors:
Michael Storper Anthony J. Venables

University of California, Los Angeles University of Oxford
145 PUBLICATIONS 14,442 CITATIONS 240 PUBLICATIONS 22,075 CITATIONS
SEE PROFILE SEE PROFILE
Some of the authors of this publication are also working on these related projects:
The Rise and Fall of Urban Economies: Lessons from San Francisco and Los Angeles View project
Worlds of Production View project
All content following this page was uploaded by Anthony J. Venables on 21 May 2014.
The user has requested enhancement of the downloaded file.

Paper to be presented at the DRUID Summer Conference 2003 on
CREATING, SHARING AND TRANSFERRING KNOWLEDGE.
The role of Geography, Institutions and Organizations.
Copenhagen June 12-14, 2003
BUZZ:
FACE-TO-FACE CONTACT
AND THE URBAN ECONOMY
Michael Storper
University of California, Los Angeles (UCLA)
Institute of Political Studies (IEP, Sciences-Po, Paris)
and London School of Economics
storper@ucla.edu
Anthony J. Venables
London School of Economics and Centre for Economic Policy Research
a.j.venables@lse.ac.uk
http://econ.lse.ac.uk/staff/ajv/)
Abstract:
This paper examines the benefits from and consequences of face- to- face contact (F2F).
Benefits are derived from the efficiency of F2F as a mode of communication: from its use to
overcome incentive problems in working partnerships: and from its role in socializing
individuals in professional networks. The consequences are that agglomeration forces remain
strong, even in world where the spatial costs of information transmission are low.
Keywords: Agglomeration, face-to-face contact, urban economics

JEL codes: R0.
December 19, 2002
BUZZ:
FACE-TO-FACE CONTACT AND THE URBAN
ECONOMY
Michael Storper and Anthony J. Venables*

Storper: University of California, Los Angeles (UCLA); Institute of Political Studies (IEP,
Sciences-Po, Paris); and London School of Economics (storper@ucla.edu).
Venables: London School of Economics and Centre for Economic Policy Research
(a.j.venables@lse.ac.uk, http://econ.lse.ac.uk/staff/ajv/)
Contents
1. AGGLOMERATION REMAINS STRONG
2. THEORIES OF AGGLOMERATION AND CITY GROWTH
2.1 Forward and backward linkages:
2.2 Labor markets:
2.3 Technological spillovers, learning, and the "creativity" of cities
3. FACE-TO-FACE CONTACT
3.1 F2F contact as a technology of communication
3.2 Trust and incentives in relationships
3.3 Socialization and being in the loop
4. THE BUZZ OF THE CITY
4.1 F2F, spillovers and synergies
4.2 Buzz and F2F contact compared to other types of coordination
5. BUZZ CITIES
Earlier versions of this paper were presented at the International Seminar on Economy and Space, Faculty of
Economics, Federal University of Minas Gerais (FACE/UFMG), Centre for Regional Development and
Planning (CEDEPLAR), Ouro Preto, Minas Gerais, Brazil, December 6-7, 2001; as keynote address to the Third
International Congress on the Dynamics of Proximity, Paris, December 2001; and to the Center for
Globalization and Policy Studies, UCLA, March 2002. Revised, December 19, 2002
*Corresponding author:
A.J. Venables
Dept of Economics
LSE
Houghton Street
London WC2A 2AE,
a.j.venables@lse.ac.uk
0
December 19, 2002
BUZZ:
FACE-TO-FACE CONTACT AND THE URBAN
ECONOMY
Abstract:
This paper examines the benefits from and consequences of face- to- face contact
(F2F). Benefits are derived from the efficiency of F2F as a mode of
communication: from its use to overcome incentive problems in working
partnerships: and from its role in socializing individuals in professional networks.
The consequences are that agglomeration forces remain strong, even in world
where the spatial costs of information transmission are low.
Contents
2.1 Forward and backward linkages:
2.2 Labor markets:
3.1 F2F contact as a technology of communication
3.3 Socialization and being in the loop
4. THE BUZZ OF THE CITY
4.1 F2F, spillovers and synergies
5. BUZZ CITIES
Keywords: Agglomeration, face-to-face contact, urban economics

JEL codes: R0.
0
One of the established facts of economic geography is that the force of agglomeration
remains strong, even though transportation and communication costs continue to decline.
There is substantial evidence, for example, that the transportation and communications
improvements that occurred in the United States from the late 18th to mid-19th centuries,
including the construction of railroads, canals and the telegraph, were accompanied by
increasing urban concentration of economic activity, not its reduction (Pred, 1966, 1974;
Fishlow, 1965; DeVries, 1984; Hall, 1998; Teaford, 1986). Relative city sizes remained
stable over the 20th century in the United States (Black and Henderson, 1998), and this
pattern of stability is true of other advanced countries such as Japan and France (Eaton and
Eckstein, 1997). Moreover, there is mostly persistence of the same activities in the same
cities; only a few industries really change their geographical centers or entirely abandon
them, once they are initially locked into a location (Brezis and Krugman, 1997; Storper and
Walker, 1989; Henderson, 1997). Thus, US industrial location patterns at the three-digit
level from the mid-nineteenth century onward have been remarkably stable (Kim, 1995;
Dumais et al, 1997). Currently, about forty percent of US employment is located in counties
constituting just 1.5% of its land area, and the geographical density of employment in many
sectors has actually increased in recent years (Kim, 2002). It has also been estimated that in
the United States, 380 localised clusters of firms employ 57% of the total workforce and
generate 61% of the nation's output and fully 78% of its exports (Rosenfeld, 1996). Other
researchers, using more conservative measures, still find that 30% of the US workforce is
accounted for by localised employment clusters (Porter, 2001).
1
This is why having an adequate theory of the geographical concentration of economic
activity is so important, and why hypothesizing a straightforward relationship between
transport costs and agglomeration forces is at best incomplete. Transport costs are one
manifestation of the more general problem of the geography of transactions between
economic agents. With every round of progress in the technology of transportation, the scale
of production and the degree of predictability and substantive complexity of transactions
continue to make costs of certain kinds of transactions sensitive to distance, and give rise to
localized concentrations of economic activity (Leamer and Storper, 2001).
Until quite recently, the study of transactions focused on physical exchange of goods
(especially intermediate goods) and people (spatial labor markets). Reflections have more
recently come to concentrate on immaterial transactions, involving the transmission and
exchange of information, knowledge and ideas, especially in light of the ascendancy of the
internet. Thus, for many analysts the motor force of agglomeration is now the creation and
communication of ideas, knowledge and information (Hall, 1998).
Yet there remain significant gaps in what economists and geographers have to say
about immaterial transactions and their contributions to urbanisation. One such gap has to do
with a key component of many immaterial transactions: face-to-face (F2F) contact.
Curiously, most of the recent literature on agglomerations and cities avoids direct
consideration of F2F contact.1 In what follows, we specify what it is about face-to-face
contact that is essential to contemporary economic transactions. Building on this, we argue
that the effect of a great deal of F2F contact, which we call "buzz," is an important enough
force to contribute significantly to the agglomeration of economic activity and persons, in an
age where both physical transportation costs have declined and the ability to transmit
information over long distances has increased so dramatically. In so doing, we shall show
1
One recent attempt can be found in Leamer and Storper, 2001.
2
that F2F contact is essential to the modern economic process more generally, as a key form
of coordination of economic agents.
Table 1 presents an inventory of current theories of why economic activities and people
might agglomerate in cities. They are grouped according to three principal reasons: (1)
backward and forward linkages of firms, including access to markets; (2) the clustering of
workers; and (3) localized interactions which promote technological innovation. Different
versions of these three basic forces are presented in Table 1, in the interest of capturing the
various functional versions of each principal line of theory.
2.1 Forward and backward linkages
The geography of forward and backward linkages of firms is shaped by the interaction of
increasing returns to scale with spatial transaction costs. Increasing returns cause firms to
concentrate production in few locations, and the presence of significant spatial transactions
costs induces them to locate close to markets and suppliers (see for example, Fujita,
Krugman and Venables 1999). Uncertainty and information are important additional factors
that play a role in spatial transactions costs. For example, transactions costs in highly
specialised inputs and outputs, such as high quality or innovative products, may be high
because transactions are irregular or occur in small volumes. Searching and matching for
suppliers and customers is important in these activities, as it is for many services. The
information needed to procure specialised inputs is complex, and in many cases cannot be
fully standardized or codified. This reinforces the advantages of the urban environment, in
which there is a dense set of suppliers and demanders for goods and services.
3
Table 1: AGGLOMERATION AND URBANISATION: THEORIES
Area of Causal Causal mechanism Outcome Limitations from

explanation agent and for agglomeration for the evidence
behavior economy
I. Input-output Firms: search for Proximity to markets and Clustering close to Many kinds of
linkages: suppliers, suppliers. markets (final goods or agglomerations don't
backward and suppliers search Transport costs. intermediate goods). display a high level of
forward for clients. Search/matching costs. Specialisation and local intermediate trade.
economies of scale
II. Thick labor Specialised Search costs, risk sharing, Efficiency gains from Applies only to part of
markets for workers search reduction in ‘hold-up’. better matches. Increased urban labor market,
specialised skills for jobs; firms incentives to acquire much of which is
search for labor specialised skills traditional employment
skills. relation.
II. Urban labor Workers :seek to Clustering implies Lifetime earnings and Probably applies to very
markets as places maximise lifetime opportunity for labour skill growth greater in big highly skilled and
where careers are career market turnover, labor markets highly specialised
formed and possibilities and increasing learning. industries.
people "learn" wages Does not identify
precisely how people
learn and role of
proximity.
III. People: Networks: circulation of Efficiency and innovation Stars and specialists
Technological information specialists or "stars" gains should have well-
spillovers within high turnover, large structured long-distance
between firms, numbers labor market networks, unless
within industry information strongly
tacit. Doesn't explain
spillovers between
industries.
III. Goods (trade): Markets: goods "carry" Specialised goods Time lag to long-
Technological information information. Very provide first-mover distance circulation of
spillovers specialised goods circulate advantages to those in goods is very short, so
initially in a limited first contact with them. spillovers are not
geographical space localised
III. People, firms, Diversity and its corollary, Diversified economy Definitions are vague,
Technological objects, the unexpected, lead to should be more creative causal mechanism not
learning and environment inventiveness and and productive clear. Hence difficult to
creativity: Jacobs creativity. measure or verify
III. People and firms Proximity of firms Clustering should be Evidence mixed.
Technological share generates linkages. more dynamically Theory is suggestive,
learning: "atmosphere" Secondary consequence is productive because more but vague on
Marshallian development of networks innovative explanation of exactly
that spread knowledge how "atmosphere"
functions or why it is
necessary.
2.2 Labor markets
Search/matching dynamics underlie the dynamics of urban labor markets, where linkages
concern specialised skills rather than specialised intermediate outputs. Firms may require
4
specialised workers, but because of output fluctuations they are not in a position to make
long-term commitments to them. Access to a large pool of specialized talent reduces the need
to hoard such workers during down-turns. Likewise, if workers are to invest in specialised
skills, they will either require long-term employment relations, or the possibility of rapid and
efficient search and rehire in a context of high turnover (Rotemberg and Saloner 2000; Jayet,
1983). These labor market agglomeration forces are largely resistant to falling transport
costs. Though daily commuting fields have extended greatly, fixed costs associated with
workers’ relocation remain high, creating a premium on matching with firms in a fairly
narrow geographical area.
The third group of explanations found in Table 1 concern technological innovation. The
notion that the city is a locus of inventiveness goes back to observations by Adam Smith,
figures prominently in Marshall (1919), and was renewed by Jane Jacobs (1969). There is
some fragmentary but fairly convincing evidence that cities -- both big diversified ones and
specialised ones -- are centers of innovation in the production of ideas and knowledge and
their commercialisation (Feldman and Audretsch, 1999; Jaffe, Trachtenberg and Henderson,
1993). The notion frequently adduced to explain these facts is that spatial proximity must
somehow improve flows of information upon which innovators depend, creating
technological "spillovers”. This is a thorny and difficult area of theory and measurement, but
one that may contain the most promising explanations for why agglomeration continues to be
such a powerful force.
It is promising because technological spillovers are the consequences of immaterial
transactions that are ‘weightless’, and should have marginal transport costs close to zero.
Many intellectual products are amenable to procurement at a distance - the design in Detroit,
5
advertising in New York and strategy in Chicago. Yet, although the clients of specialized
intellectual firms are sometimes far-flung, their intermediate suppliers and strongest
competitors usually are not. Firms producing specialized immaterial products tend to cluster
tightly together in financial ‘districts’ and downtown office buildings such as the City of
London. Furthermore, it is common for them to locate branch offices in major cities near the
clients for their ideas, suggesting that the ‘shipping’ of an intellectual product may be quite
costly (Leamer and Storper, 2001).2
Nonetheless, theory is quite spotty when it comes to accounting for the precise
mechanisms behind knowledge spillovers. One avenue of inquiry has to do with the
circulation of knowledgeable workers between firms. This enhances the ability of these firms
to recombine knowledge, imitate best practices, and otherwise improve their products. For
example, in Glaeser’s (1999) model of learning, people can absorb knowledge from contact
with more skilled individuals in their own industry, and the number of probable contacts an
individual makes is an increasing function of city size. Large cities therefore facilitate
learning, and are particularly attractive for highly-talented young people who have large
potential returns from learning.
Yet these approaches are not themselves very precise about causal mechanisms.
Knowledge workers often have well developed long distance networks, used for professional
contact and recruitment. Inter- and intra-firm long-distance technological cooperation is
common nowadays (Darby and Zucker, 2002). Indeed, these long-distance networks are
perhaps better developed in science-based industries than in traditional industries, yet it is
2
There is also a branch of theory which holds that technological spillovers occur in
geographical space via trade in goods. Goods themselves contain information which is
transmitted by their trade, and this helps others to become more innovative, via imitation and
competition. This is a force for agglomeration only in the comparatively rare cases in which
products are slow to circulate outside a restricted local space.
6
precisely in certain science-based innovative sectors that we see the greatest concentration of
"star scientists” and the most intense relationships with the leading innovative firms (Zucker
and Darby, 1998).
It is equally hollow to claim that technological spillovers are simply an unintended by-
product of the high levels of turnover found in places such as Silicon Valley, the City of
London, or Hollywood. The hypothesis is that knowledge simply "rubs off" on those around
them. But if this is the case, we must identify how rubbing off occurs and the motivations
people have for engaging in it.
This critique would not be complete without considering three additional notions
about the relationship between creativity and cities. Jacobs (1969) advanced the idea that
cities enjoy an advantage because of their economic and social diversity. This diversity,
because it is highly packed into limited space, facilitates haphazard, serendipitous contact
among people. Florida (2002) updates Jacobs to argue that the diversity found in
cosmopolitan cities facilitates “creativity” because of the openness of their networks and their
greater resistance to hide-bound tradition. But in neither formulation do we learn how
cosmopolitanism reduces the risk of sclerosis, nor how diversity is actually used by economic
agents.
Another recent line of reflection concentrates on the nature of the information which
is transmitted informally or locally. It is said to have a tacit component because it is difficult
to set down in blueprints or to codify completely. The communication of this complex
information is most efficiently carried out in a restricted geographical space. However, little
has been said precisely about the ways that short-distance contact overcomes the difficulties
associated with tacitness.
In another vein, Alfred Marshall, one of the main inspirations for contemporary
students of the "industrial district" (referring to the spatial concentration of many competitors,
7
as well as important parts of their input supply structures), also suggested the importance of
direct and unplanned contact between economic agents (Becattini, 2000). In studying the
textile districts of Lancashire a century ago, Marshall advanced two very different ideas. In
some passages, he made much of the fact of localized concentrations of competitive suppliers
as the source of efficiencies deriving from spatial concentration. But in other passages, he
thought about the qualitative dimensions of this concentration, culminating in his famous
observation that "the secrets of industry become no mysteries....they are....in the air."
(Marshall, 1919; Storper, 1997). Marshall's story is different from that of Jacobs’, in that it is
social belonging to a specialised producer community which diffuses the "secrets" of
industry, not the kind of cosmopolitan and haphazard city life described by Jacobs. Numerous
attempts have been made to transform his notion into a theory of networks underlying
contemporary industrial districts. However, the network seems overly restrictive as a theory
of this process, because virtually all the voluminous and rich descriptive evidence suggests
that sometimes people know each other in these places, but sometimes they don't, and that
even when they do, they are often careful not to share information that could be directly
pirated by their competitors. Most importantly, if they do interact through networks, what do
their interactions consist of and what are their incentives for undertaking such interactions?
In sum, none of the theories of technological or information spillovers in local space
that we have reviewed thus far, provide a real analysis of face-to-face contact. They refer to
structures and circumstances that are supposed to facilitate or necessitate close contact
between persons but not to why individuals engage in the action of contact, nor what they do
in this form of encounter.
8
The encounters to which we refer are, of course, face-to-face contacts. Table 2 summarises
the key things that happen via F2F contact. The first two rows of Table 2 define the
advantages of F2F as a communication technology. Communication in an F2F context is
communication on many levels at the same time -- verbal, physical, contextual, intentional,
non-intentional. Such multidimensional communication is held by many to be essential to the
transmission of complex, tacit knowledge. For example, social psychologists argue that
creativity results from several different ways of processing information at one time, including
not only the standard deductive way but analogical, metaphorical, and parallel methods as
well (Bateson 1973; Csikszentmihalyi 1996). These different means of communication are
mutually enriching, and lead to connections being made that cannot be had through strictly
linear perception and reasoning. An extension of this is that the full benefits of diversity and
serendipity are only realized through these multiple levels of communication.
Linguists such as Searle (1969) and Austin (1962) develop another aspect of
communicational analysis, arguing that "language is behavior" and F2F dialogue is a complex
socially-creative activity. In a similar vein, sociologists such as Goffman (1959) and
Garfinkel (1987) show that the interaction which comes from co-presence can be likened to
being on stage, playing a role, where the visual and corporeal cues are at least as important to
knowing what is being "said" as are the words themselves. In this regard, face-to-face
communication does not derive its richness and power merely from allowing us to see each
other’s faces and to detect the intended and unintended messages that can be sent by such
visual contact. F2F communication is a performance, a means to information production and
not merely to more efficient exchange. In this performance, speech, intentions, role-playing
and a specific context all come together to raise the quantity and quality of information which
can be transmitted.
9
Table 2: FACE-TO-FACE CONTACT
FUNCTION CAUSE PROCESS OUTCOMES
F2F is a communication F2F speech performance Parallel sending of many Efficient communication
technology: enables high frequency kinds of information about under uncertainty, especially
Communicating/ exchange message and intentions. High tacit knowledge
Transmitting speeds allow exploration of
complex themes.
F2F is a communication F2F speech and body Decoding through parallel Acting or responding
technology: language performance processing of many things correctly under uncertainty
understanding/ and context when a message is intended.
receiving/observing
F2F reduces incentive and F2F performance in relation Co-presence is an Ability to trust and bond
coordination problems: to intentions and truthfulness investment of time; a where messages and their
aligns commitments forfeitable bond if content is inherently
relationship not pursued. uncertain
Also, parallel processing
about intentions = precursor
to trust
F2F creates incentives: F2F performance as display “Rush:” Bio-physiological Higher individual
the motivation to imitate and effects of competition and productivity, creativity,
compete desire: generate more and inventiveness, energy
better effort
F2F allows identification of F2F allows screening Signaling that one can judge Efficient partnering/
partners according to prior according to shared codes. allows one to "be in the matching enhances
socialisation loop." Once in the loop, one productivity of groups
has to judge correctly, again
The second row of Table 2 refers to the advantages of F2F contact for the receiver.
F2F enables a rich palette of ways of receiving and interpreting multidimensional
information, especially through analog as well as parallel processing. Such capacities are
relevant not only to intentional, deliberate F2F contact, as in relationships where people
decide to see each other; they are also available in the case of just "being there" in a random
or semi-random fashion, bumping up against people and contexts in a complex, diversified
environment. The receiving possibilities that we have via F2F contact allow us to go beyond
network-dominated communication channels and intentional contacts.
The third row of Table 2 refers to the notion that co-presence -- being close enough
literally to touch each other -- allows visual “contact” and "emotional closeness," the basis for
building human relationships. This in turn may reduce incentive and coordination problems
10
that arise in economic relationships. With tacit knowledge there is always residual uncertainty
and hence the need to minimise the incentives for one agent to free ride or manipulate the
other. These moral hazards exist when the inherent degree of reliability of a message is low.
They can sometimes be reduced through improvements in the transparency or clarity of the
information itself or in how well it can be verified. But in other cases they require shaping a
relationship between the interested parties. Such a relationship can be developed either by
aligning incentives or by developing trust. F2F contact is a solder for this type of trust, in two
ways. On the one hand, because F2F is an efficient technology of sending and receiving, it is
much more difficult to be insincere in F2F encounters; on the other, F2F contact is quite
costly, and investment in F2F time signals commitment to the relationship.
The fourth row of Table 2 shows another dimension of the incentive effects of F2F
contact, which goes beyond verbal or visual communication. Psychologists have shown that
the search for pleasure is a powerful motivating force in behavior. Certain kinds of pleasure
are linked to pride of status and position: we imitate others, try to do better than them and
derive pleasure from succeeding at so doing. When we make an effort, and are on the route to
success, there is a bio-physical “rush” that pushes us forward. However, all pleasure quickly
recedes as it blends into the preceding “normal” state, and it is only by once again changing
this state that pleasure is found again. The search for such pride of status and position is thus
a strong motivation which must be continuously renewed (Scitovsky, 1976). F2F contact
provides the strongest, most embodied signals of such desire and can generate the rush that
pushes us to make greater and better efforts. It is thus no surprise that even with the
sophisticated computer monitoring that can be carried out on employee performance today,
that very few workplaces, which are -- essentially -- centers of F2F contact, have disappeared.
It is not just that it is easier to monitor employees when they are present, it is also that such
11
presence is motivating, because it contributes to desire, imitation, and competition, and the
fear of shame from failure (Scitovsky, 1976; Kahneman, et al, 1998).
In the final row of Table 2, we note what it is that permits individuals to enter into
certain kinds of communication processes in the first place: their prior socialisation to belong
to a group of potential partners, and the identification of such persons as potential partners.
We only spend time working with people who satisfy certain criteria. Some of these are set
by formal screening mechanisms (eg education) and others by informal screening
mechanisms. The ability to screen partners is a central theme of the concept of socialisation,
at the heart of sociological theory. It refers to the production of the individual as a social
being who develops specific capacities to signal to others that she belongs to a certain world,
and hence elicits from others the recognition of belonging.3 This is because socialisation is
central to the collective activity of producing and sharing the "codes" by which particular
kinds of interaction are structured (Coleman, 1990). Thus, it enables individuals to show that
they have certain criteria of judgement, which in turn signal to others that they belong to the
same social world. Signaling that one can judge enables one to get into and stay "in the
loop", by being able to judge what is going on in that loop, what counts and what doesn't
count. Socialisation is inevitably achieved in large measure through face-to-face contact, from
family, schooling, and the social environment in one's community and workplaces.
In the following sections, we deepen the themes in Table 2 and propose some
analytical models of the effects of F2F contact in the economy: the efficiency of F2F as a
communication technology (3.1); the use of F2F to overcome incentive problems in the
formation of working partnerships (3.2); and the dynamics of socialization and being ‘in the
loop’ (3.3).
3
The concept of socialization belongs upstream of the economist's notions of human capital, screening and
selection, because it is concerned with the generation of initial capacities for action, selection and discrimination,
not merely to their rational deployment.
12
3.1 F2F as a technology of communication
Codifiable information has a stable meaning which is associated in a determinate way with
the symbol system in which it is expressed, whether it be linguistic, mathematical, or visual.
Generally speaking, codifiable information is cheap to transfer because its underlying symbol
systems can be widely disseminated through information infrastructure, sharply reducing the
marginal cost of individual messages. Acquiring the symbol system may be expensive or slow
(language, mathematical skills, etc), as may be building the transmission system, but using it
to communicate information is cheap. Thus, the transmission of codifiable information has
strong network externalities, since once the infrastructure is acquired a new user can plug in
and access the whole network.
By contrast, the transmission of uncodifable information may have very limited
network externalities, since the successful transmission of the message depends on
infrastructure that is largely committed to one specific sender-receiver pair. Uncodified
information is only loosely related to the symbol system in which it is expressed. This
includes much linguistic, words-based expression (the famous distinction between "speech"
and "language"), particularly what might be called "complex discourse" (Searle, 1969). For
example, one can master the grammar and the syntax of a language without understanding its
metaphors. This is also true for some mathematically expressed information, and much visual
information. If the information is not codifiable, merely acquiring the symbol system or
having the physical infrastructure is not enough for the successful transmission of a message.
Bateson (1973) refers to the "analog" quality of tacit knowledge: communication between
individuals which requires a kind of parallel processing of the complexities of an issue, as
different dimensions of a problem are perceived and understood only in relation to one
another.
13
F2F encounters provide an efficient technology of transaction under these
circumstances by permitting a depth and speed of feedback that is impossible in other forms
of communication. As organizational theorists Nitin Nohria and Robert Eccles (1992: 292)
point out:
‘...relative to electronically-mediated exchange, the structure of

face-to-face interaction offers an unusual capacity for
interruption, repair, feedback, and learning. In contrast to
interactions that are largely sequential, face-to-face interaction
makes it possible for two people to be sending and delivering
messages simultaneously. The cycle of interruption, feedback
and repair possible in face-to-face interaction is so quick that it
is virtually instantaneous.’
This echoes the findings of sociologist Erving Goffman (1982) that "a speaker can see how
others are responding to her message even before it is done and alter it midstream to elicit a
different response."
The distinction between codifiable and noncodifiable messages comes up implicitly in
the economics literature on ‘search’ goods and ‘experience’ goods (Nelson, 1974). A
‘search’ good has a transparent value, evident upon initial inspection. An ‘experience’ good
has a nontransparent value that depends on the user and that is experienced slowly over time.
Markets that match faceless buyers and faceless sellers can mediate the exchange of search
goods, but the exchange of experience goods requires trust, understanding and in some cases,
long-term relationships. What we can call ‘experience information’ always involves some
direct human contact.
The modern information economy is based on projects in which individuals come together to
acquire and exchange information. Typically the later stages of such a project -- writing the
report, executing the transaction, or constructing the investment -- involve codifiable
14
information. It is the earlier stages where information is more fluid. Is the project a good
idea? Should one approach be followed or another? Answering these questions requires that
partners in the project undertake research and share their results. Often neither the inputs nor
the outputs of this research are observable. Thus, a partner can conscientiously research the
project, or simply free-ride, hoping that other members of the team will do the work. F2F can
play important roles in mitigating these incentive and free-rider problems.
The first point is simply that it is easier to observe and verify a partner’s behavior in
an F2F situation. Humans are very effective at sensing non-verbal messages from one
another, particularly about emotions, cooperation, and trustworthiness. Robert Putnam
(2000:175) notes that "it seems that the ability to spot non-verbal signs of mendacity offered a
significant survival advantage during the course of human evolution." Psychologist Albert
Mehrabian (1981:iii) notes that "our facial and vocal expressions, postures, movements and
gestures," are crucial; when our words "contradict the messages contained within them, others
mistrust what we say -- they rely almost completely on what we do."
Second, F2F promotes the development of trust. Any message may be understood but
not believed. There are strong questions of intentionality at work in communication.
Knowing the intentions of another actor enables us to decode the practical consequences of
what they are expressing to us (Husserl, 1968). Speech and action are tightly interrelated, but
speech does not automatically reveal to us what another person intends to do (Searle, 1969).
The credibility of a message thus often depends on mutual trust. Trust in turn depends on
reputation effects or on multi-layered relations between the parties to a transaction that can
create low-cost enforcement opportunities (Gambetta, 1988; Lorenz, 1992). Trust can be
substituted by enforceability through internalization, but the latter cannot solve the problem
of ambiguity, i.e. when complete contingent contracts are impossible (Williamson, 1985).
15
Trust also comes from the bonds that both parties establish to guarantee the
truthfulness of the message. One important economic bond is the time and money costs of
co-presence (schmoozing), which can far exceed the direct costs of sending the message.
These costs, like advertising expenses (Klein and Leffler, 1981), amount to a sunk cost that
enhances the validity of the message. Also like advertising, there is a special incentive to
continue to invest in the relationship in order to maintain the value of the relational asset that
was created by earlier encounters (absent a second date, the value of the first date disappears).
To create a relationship bond, the costs must be substantial and transparent. E-mail,
paradoxically, can be so efficient that it destroys the value of the message. The e-mail
medium greatly reduces the cost of sending a message, somewhat reduces the cost of
receiving the message, and it makes the costs mostly nontransparent. The low costs and the
nontransparency greatly limit the value of the relationship bond. A return receipt only
means that the recipient has opened the message, but the sender cannot be sure that enough
attention has been devoted to it to absorb the content. Thus, for complex context-dependent
information, the medium is the message. And when this is the case, the most powerful such
medium for verifying the intentions of another is direct F2F contact.
Game theoretic analysis provides a way of drawing out some of the incentive issues
that arise when information is fluid and actions are not observable. To illustrate, suppose that
two people are considering undertaking a joint project, but they are uncertain about its
ultimate value or quality. All they know, initially, is that the project is either good, yielding
final payoff A, or bad, yielding zero; they both attach the same prior probability, D, to the
project being good. The game has two stages. The first involves acquisition of information
about the quality of the project, and the second involves the decision of whether or not to
undertake the project, and then its implementation.
16
At the first stage the two individuals can undertake research independently and obtain
a signal of whether the project is good or bad. The signal obtained by player i may be
favorable, gi, or unfavorable, bi. However, the signals are not accurate – a good project can
send out a signal that it is bad, and vice versa. By expending effort, ei, each player (i = 1, 2)
can improve the quality of the signal received (details are given in appendix 1).
At the second stage of the game players truthfully reveal their signals to each other.
Using standard Bayesian techniques they use their combined information to compute the
probability that the project is good; this probability is higher the more good signals have been
received and the more effort has been expended, improving the quality of their signals. They
then decide whether or not to proceed. Proceeding further costs C and yields payoff A if the
project turns out to be good, and zero otherwise; we assume AD = C, so (prior to research) the
project yields zero expected surplus.
The incentives faced by individuals and the equilibrium outcomes are illustrated in
Figure 1. The axes are the effort levels of the two players, and the lines OA and OB divide the
space up into three regions. Between OA and OB effort levels are such that players will, at
the second stage, choose to go ahead with the project only if they have both received good
signals, {g1, g2}. However, below OB player 1 is putting in so little effort relative to player 2
(and hence 1's signal is so unreliable) that they proceed if 2 has a good signal and 1 a bad one
{g2, b1}. Similarly, above OA they proceed with signals {g1, b2}. The curves labeled EU1 are
expected utility indifference curves for player 1, increasing to the right, and kinked where
they cross lines OA and OB. The best response function for player 1 to each effort level e2 is
given by the bold solid lines, e1 = R1(e2). We see that if e2 is very low, then player 1 will
ignore 2's signal and put in a constant amount of effort (in the region to the left of OA).
Conversely, if e2 is high enough, player 1 will free-ride, putting in zero effort (in the region
below OB). At intermediate levels of e2 player 1 puts in a positive level of effort, decreasing
17
in e2. Just as the solid bold lines are the best responses of player 1 to 2's effort levels, so the
dashed bold lines (their reflection around the 45o line) give the best responses of player 2 to
1's effort levels.
As illustrated in Figure 1, this game has three Nash equilibria, labeled ES, E1 and E2,
occurring where the best response functions of the two players intersect. ES is symmetric, and
involves both players putting in equal amounts of effort. E1 and E2 are equilibria where
player 1 (respectively 2) exerts no effort; but given this, it is privately optimal for the other
player to put in effort to the level illustrated. This free-riding means that little information is
gathered, and at these equilibria more projects are undertaken than at ES, the proportion of
failing projects is larger, and aggregate returns lower. Player 1 prefers E1 to E2, but if neither
player does research there is no expected surplus from the project.
As there are three equilibria, what can be said about which will occur? One standard
way of resolving this issue is based on randomization. Each player assigns a probability to
playing each of the equilibrium effort levels, and the equilibrium values of these probabilities
can be computed. Thus, randomization between E1 and E2 gives a mixed strategy equilibrium
with positive probabilities that both players put in effort, that just one does, or that neither
does so.
F2F contact – a meeting between the players – can play two distinct roles in selecting
the symmetric equilibrium. First, an F2F meeting prior to the start of the game may allow
players to coordinate on this equilibrium. It is quite difficult to go into a meeting maintaining
a commitment to put in no effort. This is partly because of the inherent simultaneity of the
meeting: the two players are placed in a situation where neither has a mechanism to commit
to making no effort. And it is partly because of the psychological effects of F2F contact;
participants want to be highly esteemed by others and this is likely to be fostered by
18
cooperation rather than conflict. With F2F it is thus difficult for one player to maintain the
position that he will put in no effort and free ride on the other.
A second role that an F2F meeting can play arises if an intermediate stage is added to
the game. In the game above, the first stage was individual research and the second stage the
sharing of results and decision whether or not to incur the cost (C) of proceeding with the
project. Now, suppose instead that players, after they have done their private research, have
to decide whether or not to attend a meeting. Attending the meeting has a real cost and,
crucially, each makes the decision of whether or not to attend on the basis of her own
information: it is in the meeting that information is shared and the decision on whether or not
to go ahead with the project is taken. How does this change the situation? If the meeting cost
is high enough then players who have done no research (as well as those who have received
an unfavorable signal) will not find it worthwhile to attend the meeting. Doing nothing is no
longer privately profitable, because one has to pay a cost (that of attending the meeting)
before obtaining the partner's information, and the cost is not worth paying given the original
information.
In terms of Figure 1 each player’s indifference curves lose the sections running along
the axis, so EU1 is instead extended to the right from point a. Points E1 and E2 cease to be
equilibria, and best response functions now have a single intersection at point ES where both
players have positive effort levels. The meeting therefore reduces the set of equilibria to the
unique one at which both players make an effort.
This analysis, while highly stylized, formalizes two different possible roles that F2F
meetings may have. One is as a form of preplay communication to coordinate on one of the
possible equilibria. The other is as a way of increasing the cost of free-riding; a player who
19
makes no effort will not find it worthwhile to attend the meeting, and so cannot make a
positive return from the project.4
3.3: Socialisation and being in the loop.
Even if we admit, on the basis of the above argument, that F2F is an efficient technology of
transacting, it is nonetheless very costly, not least because it is time consuming. We do not
have the luxury of F2F encounters with the entire world, so need to screen out the people with
whom we want to interact. Yet this screening is complex because much of what is most
valuable about potential partners is their tacit knowledge. Much of such knowledge can only
be successfully communicated as metaphor (Nisbet, 1969), whose meanings are highly
culture and context-dependent (Lakoff and Johnson, 1980). Michael Polanyi (1966:4) noted
that "we know more than we can tell," suggesting that tacit and metaphorical knowledge is
deeply embedded in specific contexts.
Thus, to find partners without engaging in excessively costly direct contact with the
entire world, potential partners need to ‘know’ each other, or have a broad common
background, acquired through socialisation. Most importantly, they need to be able to verify
the prior socialisation of potential partners. There is thus something upstream of direct and
random F2F contact that allows people to increase the probability of forming partnerships
with appropriate people.
One possible solution is the formation of professional associations, based on formal
certification and some institutionalised screening mechanism, such as professional
examinations. Where this is not possible (as in cases where the profession’s performance
4
Notice that we have not assumed that the meeting can agree upon and enforce the
optimal work program; enforcement is impossible because effort is unobservable. A meeting
plays the role of narrowing the set of equilibrium (self-enforcing) outcomes.
20
criteria cannot be codified and hence institutionalised) an alternative is the development of an
informal network: being ‘in the loop’ or in the in-group. Group membership serves as a label
of quality, so members can increase their probability of success by pairing with other group
members.
What is the informational basis of such a group? Where certification of individuals’
ability or effort is not possible there has to be open, although not necessarily costless,
membership to all. However, once in, members cease to be anonymous, knowing who is in
the group, observing the performance of members, and in turn being observed by other
members. This information is used to maintain the quality of the group. At its simplest, a
record of failure is used as the basis for expulsion from the group; group members are
continually judging and being judged, and know exactly who is ‘in’ and who is ‘out’.
In some professions this information structure can be maintained across wide
distances. Academics (perhaps above all others) have national or international in-groups, in
which members are continually judging each other, and membership of which is conditional
on continuing success. In other fields, we think that such an information structure can only be
maintained within a restricted geographical area. In such fields as fashion, public relations,
and many of the arts (including cinema, television, and radio), though there are international
networks “at the top,” in the middle of these professions, networks are highly localised,
change rapidly, and information used by members to stay in the loop is highly context-
dependent. In parts of the financial services and high technology industries, local networks
intersect with long-distance contact systems. In almost anything relating to business-
government relations, networks have a strongly national and regional cast. Even in the
internationalised academic world, there are localised nodes of interactions within universities,
and within internationalised networks: these are people who come together to brainstorm,
open up new questions, carry out pre-paradigmatic research, develop new directions. All of
21
these kinds of activities require “in groups,” (which may be temporary), where competence is
high. By examining the ways such an in-group forms, we can then realise how important a
technology of communication such as F2F contact would be to such group formation.
If such an in-group forms, then what are its characteristics? First, it will contain a
higher than average proportion of able people. Continuing membership is driven by
continuing success, so if the probability of success is increasing in the ability of the partners
surviving members of the group will on average be relatively able. Second, members of the
group will (conditional on their ability) have higher earnings than outsiders, because they are
matching with (on average) higher quality people. Third, members of the group will work
harder than outsiders. The earnings differential creates an incentive to stay in the group, and
the probability of staying in is increased by hard work. Finally, although initial access to the
group is open to all, there may be an entry cost, perhaps in the form of time and effort to
become known as deserving to belong to the group. Even if this is the same for people of all
abilities it will still have a greater deterrent effect for the less able because their income
differential from being in the group is less. This is a further self-selection mechanism that
reinforces the difference in the ability composition of the in-group relative to outsiders.5
It is worth outlining a formal model of group formation to illustrate the relationship
between these forces. Suppose that the population is of size 1, with exogenous death and
birth rate of * per period. There are two types of individual, high ability and low ability,
subscripted by H and L, and the proportion of high ability in the population is . The size of
the in-group is endogenously determined and denoted N. The proportion of this group that is
of high ability (also endogenous) is denoted :I, while the proportion of outsiders that are high
5
Of course, if the entry cost is lower for more able people then further screening effects
come into play.
22
ability is :O, so where we use superscripts I and O to denote variables
for insiders and outsiders respectively.
In each time period all individuals match into pairs to undertake a project. Matching
takes place within each group, but is otherwise random. The success or failure of a project
depends on the ability of the two partners and the effort they put in, so the probabilities of
success for projects with two high ability partners, a high ability partner and a low ability
partner, or two low ability partners takes the following forms:
(1)
These probabilities depend on an exogenous component, DHH > DHL > DLL, and on the effort of
the individuals. An individual’s effort is is denoted for a high ability person inside the
group, etc, and affects probability through an increasing concave function, f ( ). Thus, a
project is more likely to be successful if undertaken by high ability and harder working
individuals.
If a project undertaken by members of the in-group fails, then both the participants in
the project are ejected from the group with probability (.6 Since the probability of failing
depends on one’s partner, and partners are selected randomly from members of the group, the
ejection probabilities for high and low ability people, 0H and 0L are given by,
(2)
6
We take this probability to be exogenous. A more complex model might have the ejection
probability depending on a past history of success and failure, rather than just success or
failure on the last project.
23
Thus, the probability of an individual in the group matching with a high ability person is :I; a
partnership with two high ability individuals fails with probability (1 - ), and a
partnership with one high and one low ability individual fails with probability (1 - ) , etc.
The size and skill composition of the in-group can now be determined. The number of
able people in the in-group, , evolves according to differential equation
(3)
The first term is the flow of able people going into the group. This consists of births, *,
proportion of whom are able, and proportion 8H of whom choose to enter the group (this
proportion may be unity, and is discussed below). The second term is the number of high
ability people who are ejected plus the number who die. Similarly, for low ability people,
(4)
In steady state these expressions are zero, giving the numbers of high ability and low ability
people in the in-group, and , as:
(5)
Equations (1) - (5) give the base case model, and its solution is seen most easily
through a numerical example, details of which are given in appendix 2. Suppose that all
individuals have the same effort levels ( ) and all members of the
population start off in the group, 8H = 8L = 1. Column 1 of table 3 reports the size of the in-
group (about 15% of the population) and the proportion of the in-group that is able, :I, this
exceeding the proportion of high ability people in the rest of the population, :O. As a
consequence people in the group are better off than outsiders, as indicated by the present
24
value of expected utility, V (see below for definition). Notice also that value to being in the
in-group is greater for high ability than low ability individuals, . The
reason is simply that they expect the benefits of group membership to last longer – they are
less likely to be involved in a failing project and to be ejected.7
This basic structure is amplified by two further sorts of behavior. The first is to let
individuals chose the level of effort they put in, and the second is to add a cost of entry to the
group, so that individuals will self select whether to enter or not. Adding these choice
variables requires that we evaluate individuals’ payoffs as a function of their behavior.
Suppose then, that a project either succeeds, yielding payoff 2" or fails, yielding 2$, and the
payoff is split equally between the two partners. The expected payoff to a high and low
ability person inside or outside the group takes the form,
(6)
where we subtract the cost of effort. The present values of payoffs are,
(7)
Outsiders, in the first row of equation (7), simply get instantaneous utility discounted at rate
*. Insiders (second row) get additional utility for as long as they stay in the group. The
probability of being ejected works like an increase in the discount rate, reflected in the
denominators of the final terms in these expressions.
7
The example is constructed with DHH - DHL = DHL - DLL, so a good match has the same effect
on probability for high and low ability people. Adding super-modularity would reinforce this
result, although it could be reversed by sufficient sub-modularity.
25
It is now straightforward to derive individuals’ optimal effort levels. An increase in
effort has a direct cost, and also changes probabilities of success (through equation (1)), hence
changing utility (equations (6)); insiders also benefit from a reduced ejection probability
(equations (2)). First order conditions give, for outsiders and insiders respectively,
(8)
We see that insiders work harder than outsiders because they fear ejection from the group.
This effort effect is greater for more able people because .
Finally, do new entrants to the labor force initially enter the in-group or not – how are
are 8H and 8L determined? We assume that all individuals can enter the group, but that entry
has a cost – perhaps the cost of working in a more expensive city, or of time invested in
building initial contacts with the group. We model this as a fixed cost c that varies across
individuals. Indexing members of the population by z, the cost takes the form,
. The proportions of high and low ability individuals who initially
enter the in-group are obtained by finding the marginal entrant for whom the fixed cost equals
the expected premium to being in the group, so:
(9)
For reasons of space we do not present formal analysis of this model, but instead
illustrate its results from numerical simulation. We have already discussed the first column of
table 3, and the remaining columns allow for endogenous choice of effort and costs of entry
to the group. The effects of allowing effort to be endogenous are given in column 2 of the
table. People in the group work harder than outsiders and since high
ability group members put in more effort than low ability ones. The effect is to increase
26
group size as failure probabilities are reduced, and to increase that proportion of the in-group
that is high ability, :I. Column 3 gives the effect of there being a cost of entering the group.
Although this cost is the same for high and low ability people the return to being in the group
is greater for high ability people. Thus, in this example, 81% of high ability people enter,
compared to just 25% of low ability. The final column gives outcomes with endogenous
effort and entry costs. This case compounds the previous effects, giving the highest value of
:I, the proportion of the in-group that is high ability.
Table 3: Group formation: ( = 0.8.
BASE EFFORT COSTS OF EFFORT AND

ENTRY ENTRY COSTS
N 0.184 0.203 0.132 0.224
:I 0.496 0.514 0.797 0.799
:O 0.293 0.283 0.258 0.194
6.07 6.41 6.62 7.19
1.44 1.73 1.56 1.7
5.82 6.07 5.66 5.67
1.32 1.57 1.16 1.17
8H 1 1 0.81 1
8L 1 1 0.25 0.37
Who are the gainers and who are the losers from this process? Rows 4 - 7 give the
present value of individuals’ utilities.8 If no group existed, all high ability individuals would
have the same utility, as would all low ability individuals. Existence of the group creates a
gap between insiders and outsiders, and this gap is larger for high-ability individuals than low
8
Where entry costs are incurred these values are reported for the median individual in the
group.
27
ability, and is greater when effort is endogenous and entry costs cause selection of individuals
initially entering the group. Outsiders are the big losers as refinement of group membership
forces them to make worse matches.9
This analysis above sets out the mechanics of in-group formation and operation, and
also illustrates the incentives for an in-group to form. The gainers from the in-group are the
high ability insiders. However, it is interesting to note that even these individuals do not want
failure of a project to lead to ejection with probability 1. Varying (, it turns out that their
utility is typically maximised at some value between zero and unity. Too low, and the group
is not of high enough average quality; too high, and even high ability insiders face a
significant probability of ejection.
This brings us back to our central points. Part of the buzz of the city is judging, and
putting oneself up to be judged. Judgement matters as group members cease to be
anonymous: if you have been observed to fail then -- with probability ( – you are branded an
outsider, and group members will no longer seek to match with you. The magnitude of ( is, in
many activities, inherently spatial. In a faceless and anonymous world ( = 0, and in-groups
cannot form. F2F contact removes anonymity and, by raising (, creates the possibility of
group formation.
This, we suggest, is the importance of F2F contact. By removing anonymity F2F raises
the probability of good, step-by-step iterative judgements about the abilities of others, where
we cannot know exactly how able or hardworking they are in the beginning; it also makes
individuals better at learning how to signal to others what abilities and effort levels they have;
it provides a vehicle for making this information transparent in a low-cost way; but it also
permits, in previous rounds of F2F contact, certain people to join and stay in. Judgement
matters increasingly as group members cease to be anonymous.. Thus, at any given moment,
9
Although all individuals gain from being allowed to optimise their effort level.
28
the ex-ante and the ex-post revelation of information blend in an unending circle of (1)
joining, staying in or being ejected –> better project completion, leading to (2) old members
–> new joiners –> better project completion, and so on. F2F contact and information flows in
the group act both to improve (although not to perfect) knowledge about the ability of
potential partners, and to sharpen incentives to succeed.
4. THE "BUZZ" OF THE CITY
4.1 F2F contact, spillovers and synergies
F2F contact is thus a highly efficient technology of communication; a means of overcoming
coordination problems in the presence of informational and effort uncertainty; of overcoming
potential problems of incentives in the face of moral hazards created by incomplete
information; and a key element of the socialisation that in turn allows people to be candidates
for membership in ‘in groups’ and to stay in such groups. The effect of such advantages of
F2F contact is to make possible joint projects that otherwise would not take place or that
would be hampered in important ways.
Because F2F contact is essential to certain forms of knowledge development through
joint projects, it is central to a number of kinds of localised knowledge exchange. There is a
large literature suggesting the existence of localized, industry-specific knowledge spillovers
within the science- and technology-based industries (Acz, 2002). These spillovers are held to
function through networks established between firms and industries clustered within regions,
such that university-based scientists are given incentives to innovate through enhanced
probabilities of commercialization. Critically, these processes center on the production of
knowledge which has a strong relational and tacit component, and hence can only be
transferred through learning-by-doing, involving much F2F contact. Subsequently, such
knowledge may become less tacit through commercialization, and then be susceptible to spill
29
over to wider geographical spaces (Darby and Zucker, 2002). In any event, F2F contact
underpins the observed clustering of certain kinds of specialised knowledge producers, and the
associated geographical concentration of talent which has been confirmed by numerous
studies (Glaeser, 1999; Hanson, 2000; Simon, 1998).
But the geographical effects of F2F contact are arguably wider than this. First, certain
sectors have more open, less institutionalized, and higher-turnover networks than others.
These include the fashion, design and culture industries. Their participants, as noted, often
take part in a geographically-concentrated Marshallian milieu. In these cases, informal or
unplanned contact is essential to the processes of evaluation, mutual identification, and
collaboration we have theorized above. Second, there is cross-fertilization between sectorally-
specialized networks, whether in traditional or in science and technology-based industries.
High technology and government have close interactions, for example, and this is why
Washington DC has become a major high-technology region. Design and
entertainment/communications have strong crossover effects in their development of content,
and this is why places such as NY, LA, London and Paris concentrate them together. Higher
education, finance and government are a powerful nexus for production of elites and their
ideas and contact networks. All the mechanisms behind F2F contact which we have described
here are present in the new media industry clusters in New York and San Francisco (Pratt,
2002). In general, then, F2F contact is essential to the force of diversified urban economies.
These inter-network spillover effects were alluded to by Jacobs (1969), in her intuition that
urban diversity is central to certain kinds of economic creativity. Jacobs also called attention
to the unplanned and haphazard aspect of some of these contacts.
These patterns of F2F contact contribute to what we shall call the "buzz" of certain
cities or agglomerations. Buzz is more than the circulation of information in F2F networks. It
results from the externalities of organized F2F processes. Buzz allows people in and around
30
networks to know what is happening; it socializes individuals to allow them to signal so that
others will admit them to intentional face-to-face contacts; and it attracts able individuals to
places where they are likely to pick up on signals that allow them to get into unplanned but
valuable contacts by "rubbing elbows." Thus, buzz is a superadditive form of information
circulation, generating increasing returns for people who are in the buzz, and for the
agglomerations in which they work.
Buzz and F2F contact are adapted to particular circumstances of coordination among agents.
Table 4 shows that there are six principal means of such coordination, depending on the kind
of knowledge that needs to be transacted, and the environment in which the transaction takes
place. By environment of coordination, the vertical axis, is meant the stability of the input and
output pools that are drawn upon. A stable environment is more likely when large numbers
are present in these pools. In the second and third rows we find two kinds of less stable
environment, according to the classical distinction established by Frank Knight (1921):
fluctuations which are amenable to some kind of statistical estimation or prediction, which he
called "risks" and those which are not, which he called true "uncertainty." The horizontal axis
uses criteria reflecting the form of the knowledge used in coordination, ranging from tacit to
transparent.
Notice that some of the reasons for agglomeration which were identified in Table 1
reappear in column one of Table 4. Proximity requirements are high where there is search and
matching for forward and backward linkages (labor and products), and where there is buzz.
Buzz is the superior mode of coordination when two conditions are satisfied: knowledge is
tacit, and the environment is truly uncertain. F2F contact is also central to the functioning of
many networks, when knowledge is tacit and there are significant risks in the environment.
31
Both correspond to cases where socialisation, getting into the loop, and staying in the loop are
vital to successful coordination.
Table 4: MODES OF COORDINATION

Proximity requirement in italics
Environment of Knowledge used in coordination

Coordination:
TACIT CODIFIED UBIQUITOUS/
TRANSPARENT
STABLE Bureaucracy, Markets.
Authority.
LOW LOW
RISK: Specialised Organized networks Markets

PREDICTABLE networks for for search/matching supplemented by
FLUCTUATIONS search/ matching (organized projects). networks or
using F2F contact. bureaucracies.
HIGH MEDIUM LOW
REAL Buzz
UNCERTAINTY HIGH
Four additional modes of coordination emerge in Table 4: bureaucracies (i.e.
organisations such as firms or states), markets, organized networks, and mixes of markets and
networks/bureaucracies. The first two are not among the modes of coordination we associate
strongly with agglomeration. Neither bureaucracy nor fully-functioning markets have high
requirements for geographical proximity of the agents who coordinate. In the latter case, the
ubiquity of the information necessary, and the large numbers of demanders and suppliers,
make possible long-distance transactions by radically reducing search costs between the
agents in the system. Bureaucracies are generally used to coordinate when there are
sufficiently large numbers of inputs and outputs to offset the overhead costs of
32
bureaucratisation, and when the information needed to coordinate among agents can be
codified.10 There is an intermediate case, where information is transparent but there are
significant fluctuations, and in these cases, there are likely to be combinations of market-based
and organizational coordination (Williamson, 1985). There are also organised or formal
networks, where information is largely codified but the environment is subject to significant
natural fluctuation. In both these latter cases, there are sometimes pressures for agglomeration
due to the transactions costs associated with fluctuations in the environment. But it is unlikely
that F2F contact is a significant aspect of their matching mechanisms.
The informational and stability properties of interactions in the economy thus lead to
different coordination solutions, among which are F2F contact and its most open-ended form,
buzz.
5. BUZZ CITIES
The emphasis on buzz -- indeed on the city in general as a site of immaterial linkages -- might
seem paradoxical to some, since the advent of broad-band internet communications would
appear, finally, to provide us the means to avoid F2F contact. The Internet has enabled
certain kinds of complex communication to occur at a distance which were previously
constrained by proximity, and some have gone so far as to claim that this is leading to the
"death of distance" (Cairncross, 2001).
The reality is certainly more complicated than this, however. This is because the
history of economic geography consists of a dynamic of coordination over space that is
strongly shaped by permanent tension between two opposing forces. On the one hand, there is
10
Although certain State bureaucracies operate, in reality, through highly changing and
politicised contact networks, in which case it is not the formal bureaucratic rules which are
dominant, but being in the loop (Crozier, 1964).
33
ongoing transformation of complex and unfamiliar coordination tasks into routine activities
that can be successfully accomplished at remote but cheaper locations. This is reflected in the
codification of information, stabilization of meanings, and the reduction of incentive problems
(opening up the possibility of more complete contracting), so that less F2F contact is needed.
Its principal geographical consequence is the tendency toward deagglomeration or dispersion
of production. On the other hand, bursts of innovations create new activities which can only
initially be carried out via complex and unfamiliar coordination tasks. For example, by
increasing reactivity to the market and by cheapening relationships upstream in commodity
chains (and inter-industrial relationships in general), improved transportation and
communications deepen the recombinative possibilities between the different parts of the
economy, making it possible to initiate new product cycles and interrupt old ones. The
increased flexibility and reactivity made possible through transportation improvements
generate new "islands of complexity" in coordination, and thus more demands for F2F contact
and a renewal of the force of buzz (Leamer and Storper, 2001). At the same time, newness and
uncertainty are always under pressure from the tendency to deepen markets, increase scale,
and the ongoing effort to codify information and hence open up the twin possibilities of
replacing F2F contact with other forms of coordination, and as a consequence, to relocate to
cheaper places. At any given moment, these two opposing forces combine in different ways,
according to the sector at hand. There is thus a considerable diversity of geographical patterns
of agglomeration and dispersion from one sector to another.
Any given city is likely to reflect some combination of the different forces of
agglomeration listed in Table 1, according to the mix of sectors comprising its economic
base.11 These forces may operate at the sectoral or the functional level (Puga and Duranton,
11
In addition, this focus on agglomerative forces should not blind us to the other reasons
why cities continue to exist: the locally-generated demand for goods and services consequent
upon the concentration of population (due to agglomeration) gives cities a strongly self-
reproducing existence, and the historical path-dependencies of prior external economies which
34
2001). In diversified city economies, functional agglomerations consist of pieces of different
sectors sharing common input structures and common clients. Buzz cities, we suggest, derive
their agglomeration force from knowledge and information based activities including: (a)
creative and cultural functions (including industries linked to this, such as fashion, design and
the arts); (b) finance and business services; (c) science, technology and high technology and
research; and (d) power and influence (government, headquarters, trade associations, and
international agencies), (Hall, 1998; Scott, 2001). They are not necessarily large, as suggested
by university towns and the industrial clusters found in many European small- and medium-
sized cities. These cities attraction for talent and their efficiency in socializing and hence
reinforcing talent confer important advantages on their participants. Buzz cities continue to
have such force today because they are the places where, more than ever, critical problems of
coordination in the modern economy are resolved through F2F contact and the positive
externalities of buzz are enjoyed by all who are in the loop.
are hard to break down, even when -- were the world to start with a tabula rasa -- other
locational patterns would be optimal.
35
Appendix 1:
Payoffs are expected monetary gains minus ei .
A good project sends out signal that it is good with probability (, and a false signal with
probability 1 – (. A bad project sends out a false signal (that it is good) with probability $i ,
and the true signal with probability 1 - $i , where $i = ( (-2 + ei) - 0.5
. Thus, if no effort is
expended, $I = (.
Parameter values: A = 150, C = 50, D = 1/3, ( = 0.8,
Probabilities are all computed by Bayes theorem and figure 1 is computed in GAUSS.
Appendix 2:
Parameter values: µ bar = 0.33, * = 0.1, " = 1, $ = 0, ( = 0.8, DHH = 0.9, DHL = 0.45, DLL = 0,
c bar = 0.15, c tilde = 1. Effort function:, f(e)= s e0.5 , s = 0.2.
References:
Acz, Z. 2002, Innovation and the Economy of Cities. Cheltenham: Edward Elgar.
Austin, JL, 1962. How to Do Things with Words. Oxford: Clarendon Press.
Bateson, G. 1973, Steps Toward an Ecology of Mind, London: Paladin Press.
Becattini, G, 2000, Il Distretto Industriale: Un Nuovo Modo di Interpretare il Cambiamento
Economico. Turin: Rosenberg & Sellier.
Black, D, and Henderson, JV, 1998, ‘Urban Evolution in the USA,’ London: London School
of Economics, mimeo, Dept of Economics.
Brezis, E and Krugman, P, 1997, Technology and the Life Cycles of Cities,’ Journal of
Economic Growth 2: 369-83
Cairncross, F, 2001, ‘The death of distance 2.0; how the communications revolution will
change our lives’, Harvard Business School Press, Cambridge.
Coleman, JS, 1990, Foundations of Social Theory. Cambridge, MA: Harvard/ Belknap.
Crozier, M. 1964. The Bureaucratic Phenomenon.
Csikszentmihalyi, M, 1996, Creativity: Flow and the Psychology of Discovery and Invention.
New York: Harper Collins.
Darby, M. and Zucker, L.G., 2002, “Growing by leaps and inches: creative destruction, real
cost reduction, and inching up.” Cambridge, MA: NBER Working Paper no. 8947.
36
DeVries, J, 1984, European Urbanization 1500-1800, Cambridge, MA: Harvard University
Press.
Dumais,G; Ellison, G; and Glaeser, E, 1997. ‘Geographic Concentration as a Dynamic
Process.’ Cambridge, MA: NBER Working Paper no. 6270.
Eaton, J. And Eckstein, Z, 1997, ‘Cities and Growth: Theory and Evidence from France and
Japan,’ Regional Science and Urban Economics 27: 443-74.
Feldman, M and Audretsch, D, 1999, ‘Innovation in Cities: Science-based Diversity,
Specialization, and Localized Competition,’ European Economic Review 43: 409-29.
Fishlow, A. 1965, American Railroads and the Transformation of the Antebellum Economy,
Cambridge, MA: Harvard University Press.
Florida, R, 2002, The Rise of the Creative Class, New York: Basic Books.
Fujita, M., P. Krugman and A.J. Venables, 1999 ‘The spatial economy; cities, regions and
international trade’, MIT press Cambridge MA.
Gambetta, D.,ed, 1988. Trust. Oxford: Oxford University Press.
Garfinkel, H. 1987. Studies in Ethnomethodology. Oxford: Blackwell.
Gaspar, J. and E.L. Glaeser, 1998, ‘Information technology and the future of cities’, Journal
of Urban Economics, 43, 136-156.
Gibbons, R. 1998. ‘Incentives in organisations’, Journal of Economic Perspectives, 12, 115-
132.
Glaeser, E.L. 1999. ‘Learning in cities’, Journal of Urban Economics, 46, 254-277.
Goffman, E. 1959.The Presentation of Self in Everyday Life. New York: Doubleday.
Goffman, E. 1982. Interaction Rituals: Essays on Face-to-Face Behavior. New York:
Pantheon Books.
Greenwald, B and Stiglitz, J, 1986, “Externalities in economies with imperfect information
and incomplete markets.” Quarterly Journal of Economics 101: 229-264.
Hall, P., 1998, Cities in Civilization, Oxford: Blackwell.
Hanson, G, 2000, "Firms, Workers and the Geographic Concentration of Economic Activity."
in: Clark, G; Gertler, M; and Feldman, M, eds, The Oxford Handbook of Economic
Geography, Oxford: OUP, 477-494.
Henderson, JV, 1997, ‘Medium-sized cities,’ Regional Science and Urban Economics 27:
583-612.
Husserl, E., 1968. The Ideas of Phenomenology. The Hague: Nijhoff.
Jacobs, J, 1969. The Economy of Cities. New York: Random House.
37
Jaffe, A; Trachtenberg, M and Henderson, R, 1993, ‘Geographic Localization of Knowledge
Spillovers as Evidenced by Patent Citations,’ Quarterly Journal of Economics 63:
577-598.
Jayet, H, 1983, ‘Chômer plus souvent en région urbaine, plus longtemps en région rurale,’
Economie et Statistique 153: 47-57.
Kahneman, D; Diener, E; Schwartz, N. 1998. Understanding well-being: Scientific
Perspectives on Enjoyment and Suffering. New York: Russell Sage Foundation.
Kim, S. 1995, ‘Expansion of Markets and the Geographic Distribution of Economic
Activities: Trends in US Manufacturing Structure, 1860-1987,’ Quarterly Journal of
Economics 110: 881-908.
Kim, S, 2002, "The reconstruction of the American urban landscape in the twentieth century,"
Cambridge, MA: NBER Working Paper 8857.
Klein, B. and Leffler, K, 1995, ‘The role of market forces in assuring contractual
performance,’ in Williamson, O and Masten, S, eds, Transaction Costs Economics,
vol. 1: Theory and Concepts. London: Elgar, 181-207.
Knight, F., 1921. Risk, Uncertainty and Profit. New York: A.H. Kelley.
Lakoff, G. and Johnson, M. 1980. Metaphors We Live By. Chicago: University of Chicago
Press.
Leamer, E and Storper, M, 2001, ‘The Economic Geography of the Internet Age.’ Journal of
International Business Studies (December), and NBER Working Paper no. W8450.
Lorenz, E. 1992. ‘Trust and the Theory of Industrial Districts.’ in M. Storper and AJ Scott,
eds, Pathways to Industrialization and Regional Development, London: Routledge.
Marshall, A, 1919. Principles of economics, London: Macmillan, (8th ed.).
Mehrabian, A, 1981, Silent Messages: Implicit Communications of Emotions and Attitudes.
Belmont, CA: Wadsworth.
Nelson, P. 1974. ‘Advertising as Information.’ The Journal of Political Economy 82,4: 729-
54.
Nisbet, R. 1969. Social Change and History: Aspects of the Western Theory of Development.
Oxford: Oxford University Press.
Nohria, N and Eccles, R, 1992, Networks and Organizations: Structure, Form and Action.
Boston: Harvard Business School Press.
Nonaka, I. 1994. ‘A Dynamic Theory of Organisational Knowledge Creation.’ Organization
Science 5,1: 14-37.
38
Polanyi, M. 1966. The Tacit Dimension. London: Routledge.
Porter, ME, 2001, "Clusters of innovation: regional foundations of competitiveness."
Washington, DC: US Council on Competitiveness
Pratt, A.C. 2002, “Firm Boundaries? The organization of new media production in SF 1996-
98.” London: LSE Dept of Geography, manuscript.
Pred, A. 1966, The Spatial Dynamics of Urban Growth in the United States, 1800-1914,
Cambridge, MA: Harvard University Press.
Pred, A. 1974, Urban Growth and the Circulation of Information, 1790-1840, Cambridge,
MA: Harvard University Press.
Puga, D and G. Duranton, (2001), ‘From sectoral to functional specialisation’, CEPR dp no
2971
Putnam, R. 2000. Bowling Alone; the Collapse and Revival of American Community. New
York: Touchstone.
Rosenfeld, S, 1996. "United States: Business Clusters." IN: OECD, Networks of Enterprises
and Local Development. Paris: Organisation for Economic Cooperation and
Development, Territorial Development Service.
Rotemberg J.J. and G. Saloner 2000, ‘Competition and human capital accumulation; a theory
of inter-regional specialisation and trade’, Regional Science and Urban Economics,
30, 373-404.
Scitovsky, T. 1976. The Joyless Economy: An Inquiry into Human Satisfaction and Consumer
Dissatisfaction. New York: Oxford University Press.
Scott, AJ, 1988, Metropolis: From the Divison of Labor to Urban Form, Berkeley and Los
Angeles: University of California Press.
Scott, AJ, ed, 2001, Global City Regions: Theory and Policy. Oxford: Oxford University
Press.
Searle, J. 1969. Speech Acts: An Essay in the Philosophy of Language. New York: Cambridge
University Press.
Simon, C. 1998, "Human Capital and Metropolitan Employment Growth." Journal of Urban
Economics 43: 223-243.
Storper, M. 1997. The Regional World: Territorial Development in a Global Economy. New
York and London: The Guilford Press.
Storper, M and Walker, R, 1989, The Capitalist Imperative: Territory, Technology and
Industrial Growth. Oxford: Blackwell.
39
Storper, M and Salais, R, 1987, Worlds of Production: the Action Frameworks of the
Economy. Cambridge, MA: Harvard University Press.
Teaford, J. 1986. The Twentieth Century American City: Problem, Promise and Reality.
Baltimore: Johns Hopkins University Press.
Veltz, P. 1996. Mondialisation, villes et territoires: l’économie de l’archipel. Paris: Presses
Universitaires de France.
Williamson, O. 1985. The Economic Institutions of Capitalism. New York: The Free Press.
40
Effort, A e1 = R1(e2)
e1 e2 = R2(e1)
EU1 EU1
EU1
E2 EU1
ES a B
O E1 Effort, e2
Figure 1: Equilibria in a game of information
acquisition and sharing
View publication stats

Buzz Face-To-Face Contact and The Urban Economy PDF

Загружено:

Сведения о документе

Оригинальное название

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Buzz Face-To-Face Contact and The Urban Economy PDF

Загружено:

Авторское право:

Доступные форматы

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

Buzz: Face-To-Face Contact and the Urban Economy

Article in Journal of Economic Geography · August 2004

Michael Storper Anthony J. Venables

SEE PROFILE SEE PROFILE

Worlds of Production View project

The user has requested enhancement of the downloaded file.

Copenhagen June 12-14, 2003

Keywords: Agglomeration, face-to-face contact, urban economics

Michael Storper and Anthony J. Venables*

Keywords: Agglomeration, face-to-face contact, urban economics

accounted for by localised employment clusters (Porter, 2001).

activity is so important, and why hypothesizing a straightforward relationship between

manifestation of the more general problem of the geography of transactions between

of production and the degree of predictability and substantive complexity of transactions

localized concentrations of economic activity (Leamer and Storper, 2001).

recently come to concentrate on immaterial transactions, involving the transmission and

communication of ideas, knowledge and information (Hall, 1998).

with a key component of many immaterial transactions: face-to-face (F2F) contact.

consideration of F2F contact.1 In what follows, we specify what it is about face-to-face

contact that is essential to contemporary economic transactions. Building on this, we argue

force to contribute significantly to the agglomeration of economic activity and persons, in an

of coordination of economic agents.

2. THEORIES OF AGGLOMERATION AND CITY GROWTH

various functional versions of each principal line of theory.

2.1 Forward and backward linkages

Area of Causal Causal mechanism Outcome Limitations from

2.2 Labor markets

narrow geographical area.

2.3 Technological spillovers, learning, and the "creativity" of cities

somehow improve flows of information upon which innovators depend, creating

such a powerful force.

It is promising because technological spillovers are the consequences of immaterial

costly (Leamer and Storper, 2001).2

potential returns from learning.

contact and recruitment. Inter- and intra-firm long-distance technological cooperation is

perhaps better developed in science-based industries than in traditional industries, yet it is

and Darby, 1998).

people have for engaging in it.

greater resistance to hide-bound tradition. But in neither formulation do we learn how

is transmitted informally or locally. It is said to have a tacit component because it is difficult

to set down in blueprints or to codify completely. The communication of this complex

associated with tacitness.

social belonging to a specialised producer community which diffuses the "secrets" of

In sum, none of the theories of technological or information spillovers in local space

in this form of encounter.

advantages of F2F as a communication technology. Communication in an F2F context is

non-intentional. Such multidimensional communication is held by many to be essential to the

serendipity are only realized through these multiple levels of communication.

socially-creative activity. In a similar vein, sociologists such as Goffman (1959) and

visual contact. F2F communication is a performance, a means to information production and

F2F enables a rich palette of ways of receiving and interpreting multidimensional

or semi-random fashion, bumping up against people and contexts in a complex, diversified

network-dominated communication channels and intentional contacts.

costly, and investment in F2F time signals commitment to the relationship.

fear of shame from failure (Scitovsky, 1976; Kahneman, et al, 1998).

by formal screening mechanisms (eg education) and others by informal screening

the symbol system in which it is expressed, whether it be linguistic, mathematical, or visual.

to communicate information is cheap. Thus, the transmission of codifiable information has

and access the whole network.

By contrast, the transmission of uncodifable information may have very limited

network externalities, since the successful transmission of the message depends on

infrastructure that is largely committed to one specific sender-receiver pair. Uncodified