Professional Documents
Culture Documents
edited by
David F. Batten
Cristoforo S. Bertuglia
Dino Martellato
Sylvie Occelli
"
~.
Learning, innovation, and urban evolution / edited by David F. Batten ... [et al].
p. efi.
Includes bibliographieal references and index.
ISBN 978-1-4613-7083-3 ISBN 978-1-4615-4609-2 (eBook)
DOI 10.1007/978-1-4615-4609-2
1. Cities and towns--Effect of technological innovations. 2. City planning. 3.
Information society. 4. Technology--Social aspects. 5. Technology-Economic
aspects. 1. Batten, David F.
HTI66.L4262000
307.76--dc21
00-040546
Preface VB
Acknowledgement ix
Index 299
The Editors
The editors wish to acknowledge the generous support of the National
Research Council of Italy (CNR) for providing the financial backing
(contracts nos. 97.00185.PF74 and 97.001 86.PF74) which made the
production of this volume possible.
PART I:
LEARNING AND KNOWLEDGE EXCHANGE
Section A:
1.1 PREAMBLE
Learning and innovation have a dual role in urban evolution. On the one
hand, they possess a cosmopolitan dimension in the sense that many creative
exchanges permeate cultural borders and rapidly take on a global
significance. This highly interactive part of the learning process is a
collective one, as mentioned above. It is often regarded as a public good
because it is available to all who can engage in it. On the other hand,
creativity has a private dimension, sometimes even a confidential dimension,
associated with the skills of an individual who produces outstanding kinds of
work - such as music, poetry, scientific ideas or technical innovations. This
private aspect introduces a personal dimension to creativity and innovation,
one that is intimately associated with the local environment.
The complex interplay between the local (or private) dimensions of
learning or innovation, and a city's collective (or public) propensity for
change, is at the core of urban evolution. It also highlights an emerging
paradox. As the literature on innovation and innovation diffusion
accumulates, our individual knowledge about its implications for urban
evolution appears increasingly fragmented, often irrelevant. The complexity
of the issues involved demands a special kind of analysis: a collective,
participatory, dynamic approach. Yet this introduces greater uncertainty
owing to its sensitivity to initial conditions, the importance of path-
dependence and the existence of multiple outcomes.
In a series of earlier books, closely-related notions such as cosmo-
creativity, technogenesis and communication networks have been explored. 3
The point of departure for the collection of papers in this volume is that they
conceive of innovation in an urban context as a broader set of interactions
between the various evolving parts of a modem society. Innovation is seen
as a product of human knowledge and learning, both collective and
individual, and learning is viewed as an interactive, dynamic process. In
other words, the chapters herein adopt a socio-dynamic systems approach.
In order to adopt such an approach, it is helpful to make a distinction
between those contributions which examine the dynamic processes of
learning and knowledge exchange (Part I), and those which examine the
nature of innovation as well as its economic and technological impacts on
urban evolution (Part II). This break-up emphasises the fact that learning and
knowledge acquisition are precursors to any kind of innovation. In short,
invention precedes innovation.
Also implied in this order is a progression from a set of theoretical
papers, through to an exploration of a set of dynamic processes, then finally
to an examination of a variety of impacts that follow from these dynamic
processes. This natural sequence: theory ~ processes ~ impacts provides a
The opening chapter of Part I, by Mario Cimoli and Marina Della Giusta,
provides a survey of the various evolutionary theories of innovation and
economic change. The authors follow Dosi (1996) in demanding a
consideration of dynamics, the presence of microfounded theories, the
assumption of bounded rationality and heterogeneity among agents, as well
as the recognition of the continuous appearance of novelty, the view of
collective interactions as selection mechanisms, and the consideration of
aggregate properties as emergent phenomena. In so doing, they develop an
alternative view of behaviour and learning on the part of firms. Their ideas
turn out to be embedded in broader national and regional 'systems of
innovation' which account for the persistent differences in technological
capacity. The importance of the institutional dimension for evolutionary
theories of innovation is thereby demonstrated.
The next chapter, written by David Batten, explores the theme of
emergence and co-evolutionary learning in urban development. It is argued
that cities and brains, as living organisms whose patterns of development
display traits typical of complex dynamic systems, have much in common.
For example, because they are both subject to co-evolutionary learning, each
can produce spontaneous unexpected outcomes referred to as 'emergent
behaviour'. The author examines these collective processes of emergence
and co-evolution, looking in particular at the interaction patterns within and
between individual groups of economic agents. One of his key conclusions is
that even very small changes in the behavioural ecology of a city's
inhabitants can have profound effects on its economy and its urban
landscape.
The aim of the chapter by Dimitrios Dendrinos is to state the effects of
technological innovation on metropolitan development as an example of the
application of the mathematics of nonlinear dynamics. The chapter shows
that the process of innovation can be regarded as an unexpected,
exogenously induced perturbation of paths, which can lead to phase
transitions or divergences in the dynamic trajectories. A substantive message
of the chapter is that technological innovation is likely to lead to further
Learning, Innovation and Urban Evolution: An Introduction 5
The final section of the volume explores some of the impacts of advances
in telecommunications and transport on the evolutionary path of
metropolitan development. It opens with a chapter by Cristofaro Bertuglia
and Sylvie Dccelli who begin by reflecting on the concept of innovation,
including the question "How new is new?" Various taxonomies of
innovation are examined, and distinctions made between the systemic and
the substantive perspectives. The authors argue that one weakness of such
taxonomies is their preoccupation with the 'hard' (or tangible) components
of innovation at the expense of the 'soft' (or intangible) components. The
latter are associated with changes in the pattern of interactions and
restructuring of relationships, and have strong links to dynamic processes of
knowledge exchange, learning and experience. Their chapter concludes with
an agenda for future research which concentrates more heavily on the softer
components of innovation.
Andrew Gillespie and Ranald Richardson look critically at the potential
of telematics to overcome the problems of peripheral areas. A review is
made of policy approaches to telematics in the UK, designed to stimulate
economic development in rural areas. The authors point out that whereas
early policies emphasised infrastructure provision, more recent policy has
focused on market access. This was the main constraint on smaller firms
wishing to innovate through telematics. Non-metropolitan areas can
doubtless benefit from telecommunications designed to overcome distance,
but rural telematic policies thus far have, by and large, failed to deliver the
expected economic benefits. In order to compete successfully in a global
knowledge society, it would appear that rural firms require people-oriented
skills and market knowledge which normally can be found only in larger
urban areas.
As many existing studies have been keen to show, our use of
telecommunications can serve as a substitute for transport. However, it is
also recognised that telecommunications have an indirect effect in
stimulating the demand for mobility. The final chapter, written jointly by
Ennio Cascetta and Bruno Montella, examines the impact of new types of
communication systems (including telecommuting, teleconferencing, and
teleshopping) on urban transport. They pay particular attention to the
competitive and complementary effects on urban mobility and transport
demand. This is a highly topical research area where, despite the significant
8 Batten D.F., Bertuglia C.S., Martellato D., Occelli S.
REFERENCES
Andersson, A.E., Batten, D.F., Kobayashi, K., Yoshikawa, K., eds. (1993) The Cosmo-
Creative Society. Springer, Berlin
Arthur, B. (1994) Increasing Returns and Path Dependence in the Economy. The University
of Michigan Press, Ann Arbor
Batten, D., Thord, R. (1989) Transportationfor the Future. Springer, Berlin
Batten, D., Casti, J., Thord, R., eds. (1995) Networks in Action. Springer, Berlin
Bertuglia, C.S., Fischer, M.M., Preto, G., eds. (1995) Technological Change. Economic
Development and Space. Springer, Berlin
Bertuglia C.S., Lombardo S., Nijkamp P., eds (1997) Innovative Behaviour in Space and
Time. Springer, Berlin
Dosi, G. (1996) Opportunities, Incentives and the Collective Patterns of Technological
Change. Economic Journal
Graham, S., Marvin, S. (1996) Telecommunications and the City. Routledge, London
Hall, P., Hay, D.G. (1980) Growth Centres in the European Urban System. Heinemann,
London
Hepworth, M., Ducatel, K. (1992) Transport in the Information Age: Wheels and Wires,
Belhaven Press, London, New York
Koestler, A. (1964) The Act of Creation. Pan Books, New York
Nijkamp, P., Pepping, G, Banister, D. (1996) Telematics and Transport Behaviour, Springer,
Berlin
Young, W. (1987) Transport. Communication and Urban Form: Issues and Policies (2
volumes), Monash University, Melbourne
4 In addition to the many references cited in the Cascetta-Montella chapter, further studies
are reported in Young (1987), Batten and Thord (1989), Hepworth and Ducatel (1992),
and Nijkamp et al (1996).
Chapter 2
2.1 INTRODUCTION
For an exhaustive list of studies on this theme, see Rothwell and Walsh (1979) and
Combs, Saviotti and Walsh (1987).
14 Cimoli M., Della Giusta M.
A variety of concepts have recently been put forward to define the nature of innovative
activities: technological regimes, paradigms, trajectories, salients, guideposts, dominants
designs and so on. More crucially, there is a considerable overlap in these concepts, in
that they try to capture a few common features of the procedures and direction of
technical change (for a discussion and references, see Dosi 1988). The rates and direction
of technical change are shaped by the dominant paradigm and their disruption is
correlated with radical changes in paradigms. Freeman and Perez (1988) propose the
notion of techno-economic paradigm, and changes in this are caused by a combination of
interrelated product and process, technical, organisational and managerial innovations
involving an increase in potential productivity for all or most of the economy. In their
view a new paradigm emerges only gradually, since the world continues to be dominated
by an old paradigm. The new paradigm therefore begins to demonstrate its comparative
advantage initially only in few sectors. The new key factors have to satisfy three criteria:
being in rapidly growth, having pervasive applications and falling costs. The current
dominating information technology paradigm clearly possesses all these features, as
previously did the electrical equipment and chemical technology-based paradigm in the
interwar period and the mechanical paradigm associated with the industrial revolution.
Innovation and Patterns of Learning 15
These assets are accumulated and refined over time, making them
impossible to copy. Such characteristics confirm that industrial R&D reflects
a private and a public dimension of knowledge and technology (Nelson
1981). They also facilitate a distinction between technology and
information3 . The latter spreads easily among firms, whereas the former
includes
"tacit and specific knowledge that cannot be written down in blueprint
form and cannot, therefore, be entirely diffused either in the form of
public or proprietary information" (Dosi 1988, p. 1131).
The fact that tacit knowledge is embodied in individuals is particularly
important, and plays a major role in explaining the nature of the impact of
science on technology (see Pavitt 1991; Batten, Kobayashi and Andersson
1989).
Technological paradigms also define basic artefacts which are
progressively modified and improved over time. These artefacts can be
described in terms of some fundamental technological and economic
characteristics. In the case of an aircraft, for example, the basic attributes can
be described not only in terms of the inputs and production costs, but also in
terms of technological features such as wing-load, take-off weight, speed,
refuelling distance, etc. Of interest is the fact that technical progress often
displays patterns of order and invariance in terms of these product
characteristics. Technological invariance has been be found in
semiconductors, agricultural equipment, and cars, as well as in micro-
technological studies.
Technological trajectories may be associated with the progressive
realisation of innovative opportunities associated with each new paradigm.
In principle, this can be measured in terms of the changes in the fundamental
techno-economic characteristics of artefacts and production process4 . Nelson
and Winter (1982) define 'natural trajectories of technical progress' as those
paths which shape the direction in which problem-solving activities move
and which posses a momentum of their own. In this sense, a trajectory
represents the normal problem-solving activity determined by a paradigm
(Dosi 1988).
The core ideas involved in the notion of technological trajectories are:
See also the importance of co-specialised assets described in Teece, Pisano and Shuen
(1990) and Teece (1998), and to which we return later.
4 The interpretation of technical change and a number of historical examples can be found
in pioneering works on the economics of technical change by Chris Freeman, Nathan
Rosenberg, Richard Nelson, Sidney Winter, Thomas Hughes; Paul David, Joel Mokyr,
Paolo Saviotti and others; for a partial survey see Dosi (1988).
Innovation and Patterns of Learning 17
1. that each particular body of knowledge (i.e. each paradigm) shapes and
constrains the rates and direction of technological change irrespective of
market inducements;
2. that one can observe regularities and invariance in the pattern of technical
change that hold under different market conditions (e.g. different relative
prices) and whose disruption is correlated with radical changes in the
knowledge base (in paradigms);
3. that technical change is partly driven by repeated attempts to cope with
the technological imbalances which it creates.
(a) in general, one or a few best practice techniques dominate the others at
any point in time (irrespective of relative prices);
(b) different firms may be characterised by their adoption of better or worse
techniques;
(c) the observed aggregate dynamics of technical coefficients in each
particular activity is the joint result of the process of imitation/diffusion
of existing best-practice techniques, the search for new ones, and market
selection amongst heterogeneous agents;
(d) changes of best-practice techniques over time trace out reasonably regular
trajectories, both in the space of input coefficients and in the space of the
core technical characteristics of outputs.
'\
XI
Figure 2.1 The distribution of technological coefficients
Innovation and Patterns of Learning 19
5 He stresses that the type of technical progress taken into consideration in his own, and in
Atkinson and Stiglitz's analysis is Hicks' neutral technological progress, i.e. purely
labour-saving or capital-saving technical progress is not considered. The effects of
unexpected price shocks on productivity are described at the aggregate level, and a
possible interpretation of the productivity slowdown could be the presence of a
continuous adaptation to a fast-changing environment. The latter would compel firms to
produce with techniques they have not yet learnt to exploit efficiently, or old techniques
that they master efficiently but are inferior to the new ones.
20 Cimoli M., Della Giusta M.
model for neglecting the supply side, i.e. the profitability perceived by
producers of innovation. His own model depicts the pace of diffusion of an
innovation as being determined by both supply and adoption constraints.
This ensures a balanced diffusion path which is determined by an adjustment
gap (i.e. the difference between saturation output level and the initial rate of
demand) and dynamic elements in demand and capacity growth. Thus the
diffusion process becomes the force determining the pace and direction of
technical change. A similar line of reasoning lies at the heart of the model by
Silverberg et al (1988), in which innovation diffusion, diversity of
technological capabilities, business strategies and expectations, are formally
incorporated into a theory of the evolutionary patterns of industries and
nations.
The dynamics in all the above models are founded on learning
mechanisms within firms. This learning is associated with the ways in which
new production factors are introduced and how firms learn to use them. In
other words, the dynamics depends on Arrow's process of learning-by-
doing. Adoption and learning substantially modify the cost functions faced
by individual firms, and thus their productivity dynamics. An interesting
example of this learning-by-doing approach is the work by Gurisatti et al
(1997). They discuss the patterns of diffusion of microelectronics-based
technical change in machine tools employed in some Italian metal working
firms. By interviewing mechanical engineers, they obtained a description of
the process of innovation which consisted of radical improvements (the
installation of new machines) followed by long phases of endogenous
improvements that substantially improved the whole process. The gains in
productivity resulting from the latter often exceeded those from the
installation of the new machines. Moreover, the authors found that the
diffusion of new machines among firms takes a considerable period of time,
and that large variations exist between firms due to differences in their
technological and organisational capabilities.
A graphical description of this process can be integrated with another
described by Dosi (1984). Because unit costs decrease along a
technologically determined learning curve, human skills and abilities clearly
display a cumulative character (see Fig. 2.2). This also helps to explain how
the set of points representing an industry recomposes itself (as we depicted
in Fig. 2.1). Furthermore, it is interesting to note that the existence of
increasing returns to adoption can be explained purely by the cumulative
nature of the learning-by-doing process.
Thus far, the question of paradigms, trajectories or equivalent concepts
has been discussed at a micro-technological level. We have argued that a
paradigm-based theory of innovation and production seems to be highly
consistent with observed patterns of technical change, microeconomic
Innovation and Patterns of Learning 21
WorkinEf-_ _...,
hours
Time
Figure 2.2 A learning curve (adapted from Dosi 1984 and Gurisatti
et a11997)
Among these sectors tnere exist input-output linkages which give rise to
a wide set of externalities and interdependencies based upon the
communality of knowledge bases, complementarities, and technological
spillovers. Such untraded technological flows are essential not only for the
technological development of the firms involved, but also for the
development of the whole industry. Fig. 2.4 portrays some sources of
technological linkages among the sectors described in the dynamic
taxonomy.
I Specialised supp6ers
~uipment
~~- Traditional
capabilities and performance, and the role played by the NIS as the
representative 'arm' of (public and private) institutions. Knowledge flows
are triggered by individuals and their organisations, and thus we find that a
collection of institutions lies at the heart of the whole system.
.
Macro-system
Pei4form ance
...
NIS
Competence Performance
i
. .
NIS
Figure 2.5 A point in the technological capabilities vector defining
the state of an economic system
Clearly, most of the observations at the national level are too aggregate to
allow a deeper understanding of the effect of the innovative process on
specific locations. To understand metropolitan development, we must focus
on the appropriate locational units of analysis. Key factors will be those
36 Cimoli M., Della Giusta M.
systems that are geographically and institutionally localised. One might ask,
for example, whether those metropolitan institutions and capabilities that are
supporting technical change are sufficiently responsive and adaptive to
adjust to whatever underlying economic changes that emerge from market
interactions. Conversely, we may ask whether they are sufficiently creative
to shape innovation and economic performance. In other words, local
institutions and networks of individuals need to be both proactive and
reactive as the principal agents of innovation.
All of the above is clearly an oversimplification of reality. The vector of
capabilities and the matrix of the NIS must be of compatible dimensions, and
there still exists the problem of defining and measuring such components.
Measurement is further complicated by the presence of many informal
relationships between organisations, institutions and sets of actors 7. In the
classical literature, such relationships are treated as 'externalities'. Thus
further investigation is needed to provide a solid foundation of support for
our chosen structural representation. Nonetheless, we believe that the
structure can help to understand those mechanisms by which a NIS may
determine the success and failure of technological progress, and thus the
positioning of the vector of technological capabilities (see Fig 2.6).
The NIS matrix is closely related to the "institutional matrix which
supports and sustains the activities of innovating firms" (proposed by
Metcalfe, (995). Although it has been depicted as a linear representation, it
should be borne in mind that this is only for the purpose of illustration. The
evolutionary foundations which account for the characteristics of national
systems of production and innovation dictate that firms are repositories of
knowledge, nested in various networks of linkages with other firms and
other non-profit organisations. Such networks enhance the opportunities for
each firm to improve their problem-solving capabilities. An even broader
notion (at a wider level of aggregation) tends to map microeconomic
behaviours into a set of social relationships, rules and political constraints
(Granovetter 1985).
.
1ewI ...... (ill IIIIIbt,
lei..,md...-.
1)'tIaII,~ .,~..-dI
ICitIxle IIId II!F-iIJl),
riP system, inIIituIcI (0RlI), rn1iIIIy iD
nwbt~ r-m inmIIhs (MRIs),.
mcdwipno lpIICiestirdaiiDlland
IDdateriIlIlld niIina 1II1J- bee. R.ctD. %GNP.
Indr poIIcia: (AETI), c:enillc:aud
Dilday ecpmcil\ll'e in
IDizcxuJ vs taeIIdl and tcdlnoIogy
R.ctD.lllio of pi~ vs
YCIIical poUcies, aaaaAlims(CRTOs),.
pubIie R.ctD. iIMcnaII ill
iotcrwnion II lIDdIloIosic:aI infmnetim ICi.all:lCJaIb, FDI
seaaiaI and Dian
level,
c:aIIn:S (TIQ),
apmSlliOlllIUppCl1ioa
.a. iqIOIU ofcaptal
pdI
~OIID1l scico;e I*b (0SSPs).
IecIIDoIccY and FDI ICdJodo&y lI1DSfcr.
IIOWI imlitucicm (11'11)
''''
matril:) :. - ,~~
~ . .
8 According to Carlsson and Stankiewicz (1995) and Edquist (1997) the systems of
innovation "may be supranational, national or subnational (regional and local), and at the
same time they may be sectoral within any of these geographical demarcations".
Innovation and Patterns of Leaming 39
Acknowledgements
A number of people made helpful comments on this essay. The authors particularly
wish to thank the following, without implicating them in any way in the opinions
expressed: Giovanni Dosi, Paolo Guerrieri, Sanjaya Lall, Francesco Luna, and
Franco Malerba. The research on which this chapter is based benefited from the
support of a 40% grant from MURST, Crescita Endogena nelle Aree Arretrate
(Italian Ministry for University and Scientific and Technological Research).
40 Cimoli M., Della Giusta M.
REFERENCES
Amsden, A (1989) Asia's Next Giant: South Korea and the Last Industrialisation, Oxford
University Press, New York, Oxford
Amsden, A, Hikino, T. (1993) "Staying Behind, Stumbling Back, Sneaking Up, Soaring
Ahead: Late Industrialisation in Historical Perspective". In International Convergence of
Productivity with Some Evidence from History, Baumol, W., Nelson, R., Wolff, E., eds,
Oxford University Press, Oxford
Amsden, A, Hikino, T. (1994) Project Execution Capability, Organisational Know-How, and
Conglomerate Corporate Growth in Late Industrialisation. Industrial and Corporate
Change 3:111-47
Arthur, W.B. (1989) Competing Technologies, Increasing Returns and Lock-In by Historical
Events. Economic Journal 99( I): 116-31
Arthur, W.8., Ermoliev, Y., Kaniovski, Y. (l987) Path Dependent Processes and the
Emergence of Macro Structures. European Journal ofOperational Research 30: 294-303
Atkinson, AB., Stiglitz, I.E. (1969) A New View of Technological Change. Economic
Journal 79 (3): 573-78
Audretsch, D.B. (1996) Technological Regimes, Industrial Demography and the Evolution of
Industrial Structures. IIASA Working Paper 42, April 1996
Bardhan, P. (1996) The Political Economy of Development Policy: An Asian Perspective.
Development Thinking and Practice Conference, Sept 3-5, Washington DC
Batten, D.F., Kobayashi, K., Andersson, A. E. (1989) "Knowledge, Nodes and Networks: An
Analytical Perspective." In Knowledge and Industrial Organisation. Andersson, A. E.,
Batten D.F., Karlsson, C., eds, Springer, Berlin, 34-46
Bell, M., Pavitt, K.L.R. (1993) Technological Accumulation and Industrial Growth: Contrasts
between Developed and Developing Countries. SPRU, University of Sussex, Sussex
(mimeo)
Breschi, S., Malerba, F. (1997) "Sectoral Innovation Systems: Technological Regimes,
Shumpeterian Dynamics and Spatial Boundaries." In Systems of Innovation, Edquist, C.,
ed, Pinter, London Washington
Calion, M. (1991) Technological Development and Adoption Networks, Conference on
Constructive Technology Assessment, Twente, The Netherlands
Cantwell, AI. (\ 991) "The Theory of Technological Competence and its Application to
International Production." In Foreign Investment, Technology and Growth, McFetridge,
D.G., ed, University of Calgary Press, Calgary
Carlsson, 8., Stankiewicz, R. (1995) "On the Nature, Function and Composition of
Technological Systems." In Technological Systems and Economic Performance: The Case
ofFactory Automation, Carlsson, 8., ed, Kluwer, Dordrecht
Chandler, AD. (1990) Scale and Scope: The Dynamics of Industrial Capitalism. Harvard
University Press, Cambridge Mass
Chandler, AD. (1992a) Corporate Strategy, Structure and Control Methods in the United
States during the 20th Century. Industrial and Corporate Change I: 263-84
Chandler, AD. (1992b) Organisational Capabilities and the Economic History of the
Industrial Enterprise. Journal of Economic Perspectives, 79-100
Chandler, AD. (1993) "Learning and Technological Change: the Perspective from Business
History." In Learning and Technical Change, Thomson, ed, MacMillan, London
Cimoli, M. (1988) Industrial Structures, Technical Change and the Modes of Regulation in
the Labour Market. DRC Discussion Paper 60, SPRU, University of Sussex, Sussex
Innovation and Patterns of Learning 41
Cimoli, M., Dosi, G. (1988) "Technology and Development: Some Implications of Recent
Advances in the Economics of Innovation for the Process of Development." In Science,
Technology and Development: A Critical Perspective. Wad, A., ed, Westview Press,
Boulder Colorado
Cimoli, M, Dosi, G. (1990) "The Characteristics of Technology and the Development
Process: Some Introductory Notes." In Technology Transfer in the Less Developed
Countries, Chatterji, M., ed, Macmillan, London
Cimoli, M., Dosi, G. (1995) "Technological Paradigms, Patterns of Leaming and
Development. An Introductory Roadmap." Journal of Evolutionary Economics 5 (3) (also
Working Paper 83, IIASA, Laxenburg, Austria)
Constant, E. (1985), The Turbojet Revolution, Johns Hopkins University Press, Baltimore.
Coriat, B. (1990) Penser l'envers. Bourgeois, Paris
Dahlman, C., Sananikone, O. (1990) Technological Strategy in Taiwan: Exploiting Foreign
Linkages and Developing Local Capabilities, mimeo
Dahlman, C., Westphal, L. (1982) "Technological Effort in Industrial Development: A
Survey." In The Economics of New Technology in Developing Countries. Stewart, F.,
James, 1. eds, Frances Pinter, London
David, P.A. (1989) Path-Dependence: Putting the Past into the Future of Economics.
Discussion Paper, Stanford University, Stanford
Della Giusta, M. (1996) Education and Technological Capabilities in Industrialisation.
Master's Thesis, University of Reading, Reading
Dierickx, I., Cool, K. (1989) Asset Stock Accumulation and Sustainability of Competitive
Advantage. Management Science 35 (3), September
Dosi, G. (1984) Technical Change and Industrial Transformation. MacMillan, London New
York
Dosi, G. (1988) Sources, Procedures, and Microeconomic Effects of Innovation." Journal of
Economic Literature 26: 1120-71
Dosi, G. (1997) Opportunities, Incentives and the Collective Patterns of Technological
Change. Economic Journal 107: 1530-57
Dosi, G., Freeman, c., Nelson, R., Soete, L. (1988) Technical Change and Economic Theory,
Pinter, London
Dosi, G., Giannetti, R., Toninelli, P.A. (1993) Technology and Enterprise in a Historical
Perspective. The Clarendon Press, Oxford
Dosi, G., Kaniovski, Y. (1994) On Badly Behaved Dynamics. Some Applications of
Generalised Urn Schemes to Technological and Economic Change. Journal of
Evolutionary Economics 4: 93-123
Dosi, G., Lovallo, D. (1995) Rational Entrepreneurs or Optimistic Martyrs? Some
Considerations on Technological Regimes, Corporate Entries, and the Evolutionary Role
of Decision Biases. IIASA Working Paper 77, August
Dosi, G., Marsili, 0., Orsenigo, L, Salvatore, R. (1993) Learning, Market Selection and the
Evolution oflndustrial Structures. IIASA Working Paper
Dosi, G., Marsili, 0., Orsenigo, L., Salvatore, R. (1995) Leaming, Market Selection and the
Evolution oflndustrial Structures. Small Business Economics 7:411-36
Dosi, G., Nelson, R. (1993) "Evolutionary Theories in Economics: Assessment and
Prospects." In Market and Organisation: The Competitive Firm and its Environment,
EECffempus Programme, LATAPSES, Nice, France
Dosi, G., Orsenigo, L. (1994) "Macrodynamics and Microfoundations: an Evolutionary
Perspective." In Economics of Technology. Granstrand, 0., ed, Elsevier Science
Dosi, G., Pavitt, K., Soete, L. (1990) The Economics of Technical Change and International
Trade, Harvester Wheatsheaf Press, London, New York
42 Cimoli M., Della Giusta M.
Schumpeter, J.A. (1943) Capitalism. Socialism and Democracy, Harper and Row, New York
Silverberg, u.; Dosi, G., Orsenigo. L. (1988) Innovation, Diversity and Diffusion: A Self-
Organisation Model. The Economic Journal 98: 1032-54
Simon, H.A. (1955) A Behavioural Model of Rational Choice. Quarterly Journal 0/
&onomics
Teece, D. (1998) Strategic Technology and Public Policy. Selected Papers, Edward Elgar,
London
Teece, D.J., Pisano, G., Shuen, A. (1990) Firm Capabilities, Resources, and the Concept of
Strategy. University of California at Berkeley (mimeo)
Teitel, S. (1981) Towards an Understanding of Technical Change in Semi-Industrial
Countries. Research Policy 10: 127-47
Teitel, S. (1984) Technology Creation in Semi-Industrial Economies. Journal 0/ Development
Economics 16,: 39-61
Teitel, S. (1987) "Towards Conceptualisation of Technological Development as An
Evolutionary Process." In Structural Change. Economic Interdependence and World
Development. Dunning, J., Mikoto, U., eds, Macmillan, London
Teubal, M. (1984) "The Role of Technological Learning in the Exports of Manufactured
Goods: the Case of Selected Capital Goods in Brazil", World Development, 12: 349-865.
Teubal, M. (1987) "Innovation and Development: A Review of Some Work at the
IDBIECLNUNDP Programme." In Technology Generation in Latin American
Manu/acturing Industries. Katz, 1., ed, Macmillan, London
Thomson, R. (1993) Learning and Technical Change. Macmillan, London
Utterback, 1., Abernathy, W. (1975) A Dynamic Model of Process and Product Innovation".
Omega Vol.3
Verspagen, B. (1990) "Localised Technological Change, Factor Substitution and the
Productivity Slowdown." In New Explorations in the Economics o/Technological Change.
Freeman, c., Soete, L., eds, Pinter, London
Vincenti (1992) What Do Engineers Do and How Do They Know It? Johns Hopkins
University Press, Baltimore
Wade, R. (1990) Governing the Market: Economic Theory and the Role o/Government in
East Asian Industrialisation. Princeton University Press, Princeton, New Jersey
Walsh, V. (1984) Invention and Innovation in the Chemical Industry: Demand Pull or
Discovery Push? Research Policy 13: 211-34
Winter, S. (1982) "An Essay on the Theory of Production." In &onomics and the World
Around It. Hymans, S.H., ed, University of Michigan Press, Ann Arbor
Winter, S. (1984) Shumpeterian Competition Under Alternative Technological Regimes.
Journal 0/Economic Behaviour o/Organizations 5: 287-320
Zysman, 1. (1994) How Institutions Create Historically Rooted Tr*ctories of Growth.
Industrial and Corporate Change 3: 243-83
Chapter 3
David F. Batten
The Temaplan Group, Melbourne, Australia
3.1 INTRODUCTION
The celebrated astronomer, Carl Sagan, reminded us all some time ago
that:
"the pattern of evolutionary causality is a web of astonishing complexity;
the incompleteness of our understanding humbles us," (Sagan 1980)
Almost forty years earlier, the noted economist Joseph Schumpeter
delivered a similar cautionary message:
"The process of social life is a function of so many variables, many of
which are not amenable to anything like measurement, that even mere
diagnosis of a given state of things becomes a doubtful matter, quite apart
from the formidable sources of error that open up as soon as we attempt
prognosis." (Schumpeter 1942)
Causality remains a mystery in the case of urban development. Like most
problems in evolutionary causality, the difficulty is that too many variables
are involved. With the recent hindsight of nonlinear dynamics, we are
beginning to realise that a city's future path of development cannot be
predicted. If it cannot be predicted, it certainly cannot be optimised. Most of
the world's oldest cities have grown haphazardly, bit by bit in response to
the needs and opportunities of the moment. No compact, lawlike description
of their evolution can be found. Some continue to grow, others have fallen
from grace. The causes of one's survival and another's demise are difficult
46 Batten D.F.
1950 occupied a unique position in that nation's urban hierarchy. The size
distribution of cities corresponded closely to a rank-size relationship. If we
plot population size against rank for every urban centre in the United States,
then the resulting downward-sloping curve is roughly a straight line on a
logarithmic plot. This simply means that the size of each centre is inversely
proportional to its population rank.
~
~ 250
CI'j~
~i /00
~~s
so
Q..
C~
~ 10
I
/ 5 10 2S /00 ZOO 500 ~OOO 2pOO
R.ANK
Figure 3.1 Rank-size distribution of urban centres in the United States,
1950 (Source: Vining 1955)
48 Batten D.F.
Zipf produced similar plots for many geographical regions in the 1920s
and found the same kind of regularity. Subsequent research has confirmed
that this rank-size rule relationship also seems to hold outside the United
States. In a large comparative study of thirty-eight countries at varying levels
of economic development, Berry (1961) showed that thirteen of these
countries had city-size distributions that conformed to the rank-size rule (see
Table 3.1). Later studies have also confirmed that rank-size distributions
exist in some of the nations of South-East Asia (e.g. Sendut 1966). The
regularity expressed by the straight lines in many of these logarithmic plots
of size versus rank, with slope near unity, is known as Zipfs Law.
But what about the twenty-five countries in Table 3.1 whose distributions
did not follow the rank-size rule? Fifteen exhibited a primate distribution, in
which one or two very large cities dominated and there was an absence of
medium-sized cities. The rest revealed patterns that could be seen as
intermediate between rank-size and primate. In these cases, one or two cities
were rather dominant, but the overall pattern was generally closer to rank-
size than to primate.
Emergence and Co-Evolutionary Learning 49
Is there a simple explanation for the differences? The table itself does not
suggest any obvious explanation. While many of the lesser developed
nations (like Guatemala, Mexico, Peru and Thailand) do have primate
patterns, others (like EI Salvador and India) do not. Conversely, a number of
advanced European nations (like Austria, Sweden and the Netherlands) seem
to display a primate pattern rather than a rank-size distribution. On the basis
of Berry's table alone, we feel obliged to conclude that numerous factors
probably account for the differences - factors such as stage of economic
development, changes in political status, length oftime urbanised, and so on.
Before we accept such a conclusion, however, two observations warrant
our attention. First, there is no convincing argument to support the arbitrary
choice of national frontiers as the most appropriate way of defining a system
of cities. Quite different results can be achieved after taking cultural and
linguistic borders into account. More importantly, an historical analysis of
flows and interaction patterns between all candidate cities is a more reliable
way of defining a truly interactive system of cities. This method would
suggest that the towns and cities of Malaysia and Singapore should be
regarded as part of a single distribution of cities which, when treated in this
way, is approximately rank-size (see Sendut 1966). Furthermore, in Japan
one finds a rank-size distribution of cities once Tokyo's chief rival is seen to
be the multi-centred Kansai or Keihanshin conurbation (see Ginsburg 1988;
Batten 1995).
Second, there is no reason to expect that very different systems of cities
will converge to the same kind of city-size distribution at any single point in
time. After all, we are comparing the outcomes of complex evolutionary
processes which have commenced from different initial conditions and are
evolving in different directions at different speeds. Some settlement systems
are already well-developed and strongly interactive. Others remain largely
immature and only weakly interactive. The very fact that one system
displays a rank-size distribution, while another remains primate, merely
confirms that we are dealing with different classes of settlement systems.
Some of these are dynamically stable. Others may be relatively unstable.
25'000
RANK
Figure 3.2 Stability of the rank-size distribution of urban centres in the
United States, 1790-1990
1. Under what conditions can a nation's economic, political and social life
be regarded as complex?
2. Could the rank-size pattern be a sufficiently strong attractor (in the case
of the United States) that the individual towns and cities self-organise in
ways which ensure that the system as a whole preserves its rank-size
structure over time?
Let us deal with the second question first. There is evidence to support
the idea that the rank-size pattern is an attractor. Despite the high degree of
macro-stability shown in Fig. 3.2, the relative position of individual cities in
the U.S urban system has varied considerably. Between 1870 and 1950, for
example, Los Angeles leapt ahead 314 places to reach the top five. Among
those cities with a population of more than 100,000 in 1950, 14 enjoyed a
relatively stable ranking but another 10 experienced changes of more than
100 places. This is something like a forest whose contours never seem to
change, but whose individual trees do. Perhaps the towns and cities self-
organise into an equilibrium pattern or rank-size 'ecology'.
Similar macro-stability can also be found in several city-size distributions
in Eastern Asia. Indonesia, Japan, Malaysia-Singapore, South Korea and
Taiwan have more-or-Iess preserved their rank-size distributions over the
last fifty years, despite the fact that many individual towns and cities have
changed their rankings markedly. Although Zipf's law implies this stability
in the number of cities within a given range of population sizes, it cannot tell
us why a particular city has a certain number of inhabitants. All the
observations are of a collective or statistical nature.
What is so magical about the rank-size distribution that it can command
such rigid adherence? The rank-size distribution is emergent in the sense that
it is not an obvious consequence of the underlying dynamic rules governing
the behaviour of the individmll elements (i.e. towns and cities).
Mathematically speaking, however, Zipf's Law is nothing more than a
specific kind of power law. Power laws express a quantity N as some power
of another quantity s:
simply a power law with exponent -1. Thus the problem of explaining the
observed statistical features of complex systems such as rank-size
distributions of cities can be seen as the problem of explaining the
underlying power-law relationship.
City size distributions are not the only phenomena which can be
expressed as straight lines on a double logarithmic plot. Various catastrophic
events, fractals and noise also exhibit this pattern (see Bak 1996). Such
coupled dissipative phenomena cannot be understood by studying them
within a time frame which is short compared with the evolutionary process
which created them. For example, the laws governing earthquakes cannot be
understood just by studying earthquakes occurring in a single human
lifetime.
A special feature of coupled dissipative systems is that they evolve
naturally towards a self-organised critical state. To illustrate the basic idea of
self-organised criticality in an interactive system, consider a simple pile of
sand. Suppose we start from scratch and build our pile by randomly adding
sand, one grain at a time. The pile will grow and the slope will increase until
eventually it reaches a critical value. If more sand is then added, it will slide
off. Alternatively, if we had started off from a situation where the pile had
been too steep, it would have collapsed until it reached the same critical
state. By 'self-organised', we mean that the system evolves naturally to this
critical state, without detailed specification of the initial conditions. The
critical state is an attractor for the dynamics. It is just barely stable with
respect to further perturbations.
As the pile is built up, the characteristic size of the largest sandslide (or
avalanche) grows, until at the critical point there are avalanches of all sizes
up to size of the whole system. Once this critical point is reached, the system
stays there. This distribution of avalanches in a sandpile also happens to
obey Zipf's law. Perhaps we should be thinking of cities as being formed by
avalanches of human migrations (see Fujita 1996)? More importantly, we
might ask whether certain nation's city-size distributions evolve
spontaneously towards a critical state and later return to it, even if perturbed
at some later stage.
Self-organised systems start from disordered initial conditions and tend
to move to more highly ordered behaviour. This highly ordered behaviour
exhibits surprisingly simple regularities, such as rank-size distributions. One
can imagine with city-size distributions that there might be many potential
outcomes depending on initial conditions. But a handful of these potential
outcomes - namely those self-organised ones in which city sizes roughly
Emergence and Co-Evolutionary Learning 53
Some urban networks persist for many centuries. In New York City, for
example, the arrangement of many of the major streets dates back to the
seventeenth century, the stock exchange to the eighteenth century, the
waterworks to the nineteenth, and the electrical power system to the
twentieth. In the seventeenth century, you travelled between Brooklyn and
Manhattan across the East River by ferry. In the nineteenth century, new
technology allowed the construction of a suspension bridge across the river.
It was built precisely at the site of the ferry terminal. When newer
technology permitted the construction of a tunnel under the river, it too was
built in exactly the same place. This reinforcement of existing circulatory
systems by upgrading the technology of critical links is very much like the
pattern of biological evolution. It also provides some insights into
technological choices which permit the city to function more or less
continuously over many centuries.
Given the vagaries of political decision-making and the costs of altering
large infrastructure networks, it is rather easy for a city to 'lock-in' to a
particular arterial system which was perceived at the time to possess a socio-
economic advantage (see Arthur 1994). Lock-in can occur as sequential
decisions carve out an advantage which a city (or more correctly a city
government) finds it impossible to escape from at a later date. In the
nineteenth century, for example, Philadelphia was a more important port
than New York. The opening of the Erie Canal then tipped the scales in
favour of New York. Although the Erie has been more of a tourist attraction
than a serious transportation route for more than a century, New York still
dominates the USA's urban hierarchy. We shall never know whether
Philadelphia might have achieved that status if the Erie Canal had never
materialised. New York moved ahead and has never looked back. In cases
such as these, history really matters.
The growth of New York versus Philadelphia highlights the importance
of circulation and interaction patterns within and between cities. Circulation
54 Batten D.F.
Obviously each inhabitant cannot interact with all the others, so the
population's interactive potential falls well short of the figures in the right
hand column. But the key message to be learnt from this is more subtle.
Once the ratio of interactions per inhabitant passes a critical threshold, then
the qualitative impact of that population's interaction patterns can change
unexpectedly. The following simple experiment helps to illustrate the
unusual nature of these critical thresholds of interaction.
@
0
CD 0
@
@
@
/
17
Figure 3.3 The crystallisation of connected webs. (As the ratio of edges
to nodes exceeds 0.5, most points become connected; once it
exceeds 1.0, closed pathways begin to emerge)
Emergence and Co-Evolutionary Learning 57
~
~
~ 400~----------------------------------~
~
h..
t1 300
~
'~"
-..../
200
100
to
I\j
(;j
o
o 0.5 1.0 1.5
RATIO OF TJlREAJ)S/BUTTONS
Note that the curve is S-shaped or sigmoidal. The size of the largest
cluster of buttons increases slowly at first, then rapidly, then slows again as
the ratio of threads to buttons increases further. Were there an infinite
number of buttons, then as the ratio of threads to buttons passed 0.5, the size
of the largest component would jump discontinuously from tiny to
enormous. The steep part of the curve would become more vertical than it is
in Fig. 3.4. This is typical of a phase transition, just like when separate water
molecules freeze into a block of ice.
Now let's return to our urban population. Consider what might happen if
some residents start to interact more frequently with certain other residents.
As the number of interactions increases, clusters of 'like-minded' residents
begin to emerge spontaneously. Like-minded residents do not know in
advance who their closest friends and allies may be. These kinships can only
emerge after-the-fact. Furthermore, such like-minded clusters may do more
than simply interact among themselves. In order to pursue their common
interests more widely, eventually they may link up with other like-minded
clusters; thereby creating even larger clusters. Sounds familiar, doesn't it?
People behaving like buttons and threads!
In fact, the socialisation process whereby common interest groups
emerge and develop is analogous to the toy example discussed in the
previous section. People form clusters (e.g. political parties, unions, clubs,
teams) in order to pursue their common interests. These clusters represent
decision-making units which can play a powerful role in the shaping of
urban development. But because such clusters must take their decisions
collectively (e.g. by preference voting), sometimes the outcomes can be
quite different from those intended or expected by some of their members.
We shall refer to these unpredictable outcomes as emergence.
For the purposes of urban research, it is helpful to think of a city as a
complex social system which results from a large number of interactions
between its inhabitants. The defining characteristic of this complex urban
system is that some, perhaps even many, of its collective properties cannot
be predicted simply from our individual knowledge of these underlying
interactions. The whole urban system is greater than all of its parts! Some
collective outcomes can be predicted, but others cannot. They only emerge
as byproducts of the collective experience.
Complex systems breed emergence. For example, a car is a complex
system because its ability to move cannot be predicted simply by knowing
the individual rules which govern its parts (axles, gears, carburettors, etc.). A
car's motion is an emergent property - a process that it can carry out only by
Emergence and Co-Evolutionary Learning 59
3.4 LEARNING-BY-CIRCULATING
easier. If some enterprising city merchants felt that the risks and costs of
travel over longer distances were acceptable, they would visit more distant
places to sell their goods more profitably. Such a quest for more distant
markets would be more compelling once local demand had been satisfied. If
these early merchants reported profitable trading ventures, then other city
merchants would be more likely to follow their lead and eventually engage
in trade.
RATE OF CHANGE
IN FA.RMEIlS
..RATE OF CHANGE
IN Clry MERCHANTS
RA TE OF CflANGE
IN FARMERS
~c'-----------------------~~--------~Er
RATE OF CHANGE
IN CITY MERCHANTS
Figure 3.6 Catastrophic change as trade costs decrease (Source: Mees 1975)
SPECIALISATION
INCREASING
CIRCULATION AND ,. AND INCREASING
"-
RETURNS TO
TRADE
SCALE
II'
,
IMPROVEMENTS TO
SELECTIVE
TRANSPORT AND ~---~
GROWTH IN URBAN
COMMUNICATION
POPULATIONS
NETWORKS
(3.3)
where PI = the unit purchase price in city 1, P2 = the unit sale price in city 2,
t\2 = the unit costs of transporting and protecting the goods between city 1
and city 2, and r\2 = the unit profit accruing from the sale. For any journey
deemed reasonably safe, the merchant's problem thus reduces to that of
estimating the profitability, r\2, of trading between candidate pairs of cities.
From the viewpoint of today's information-intensive world, this may
look like a relatively simple task (although, because of segmented markets
and the complicated dynamics of price movements, the pairwise comparison
of prices is no easy task even nowadays). But when the Medieval merchants
faced it almost one thousand years ago, networks were unreliable and
communication was no easy matter. At least three decision problems posed
great difficulties:
The knowledge and experience needed to solve the first two decision
problems could only be generated by a series of trading journeys. In other
words, the entrepreneurial skills of a successful travelling merchant could
only be acquired through learning-by-circulating. The idea of circulating
over longer distances clearly appealed to the Italians, especially to the
Venetians. Given the modest size of local demand, long-distance trading was
seen as a natural way to combine trading for profit with the Venetians'
acknowledged skills as seafaring navigators. The incentive to exploit their
maritime prowess was stronger than elsewhere, because farming was not a
realistic alternative.
The early fish-eating, marsh-dwelling Veneti gathered and processed salt,
then sought other markets where it might be sold for profit. One such market
was found in Constantinople. As the final decade of the tenth century began,
Venice prospered under the strong hand of the statesman, warrior and
diplomatic genius, Pietro Orseolo II. In less than a year, he negotiated tariffs
with Basil II in Constantinople that were more favourable than any that
Venice had previously enjoyed (Norwich 1977). By effectively reducing the
total transaction costs (tn) for Venetian goods sold in Constantinople, the
Doge established a major competitive advantage for Venetians trading in the
Greek city.
Imagine the thoughts of an enterprising Venetian merchant who knew
that salt was highly regarded by both Muslim and Christian alike. He also
knew that Constantinople supplied many luxury goods such as silk cloth,
gold and silver plate, carved ivory, jewellery, and semi-precious stones. It
also produced more pedestrian commodities such as linen, cotton cloth, and
armaments. Sicilian grain could be bought in Palermo and sold for profit in
many other ports. This merchant had the following bright idea: (1) to buy
salt in Venice for sale in Palermo and Constantinople, (2) to buy a few
luxury goods in Constantinople for sale in Venice or Palermo, and (3) to buy
grain in Palermo for sale in Venice. Thus the more enterprising Venetian
merchants added to salt exports the business of general trade (Jacobs 1969).
In addition to the three decision problems mentioned earlier, this merchant
also needed to charter a suitable vessel, chart the most judicious route and
decide on the quantities to be carried (see Fig. 3.8).
The Venetian merchant's willingness to search for new markets for some
luxury goods bought in Constantinople was not just an act of great courage.
It also revealed a natural talent for sniffing out entrepreneurial profit. An
unfamiliar product is valued by purchasers much as gifts of nature or
pictures by old masters (Schumpeter 1934). Often its price may be
determined without regard to the actual cost. Despite the innumerable
difficulties of such trading ventures, clearly the rewards could be exceptional
(although unpredictable).
66 Batten D.F.
()
J
Salt @ P1
t1.2
~ ~:()
Silk@ P2
t2.3
Grain @ P3
How might our enterprising Venetian merchant have come to terms with
such a complicated situation? According to modern psychology, human
beings have always resorted to both deductive and inductive reasoning. The
left side of our brain presides over rational, analytical and critical thinking.
But the scope of our deductive powers is limited. The right side is mainly
responsible for pattern recognition, intuition, sensitivity and creative insights
(see e.g. Bower and Hilgard 1981). When situations become too complicated
or ill-defined, human nature dictates that we tum more to our inductive
abilities. We resort to intuition or search for patterns which, once recognised,
can help us to simplify the problem. We use our intuition or recognisable
patterns to build temporary mental models to work with, carry out localised
deductions based on these mental models, and then act on such models.
When he first set sail for Constantinople, our Venetian merchant's profit
expectations may have been a little unrealistic. He may have enjoyed mixed
success on the first few trips owing to the many uncertainties associated with
such a novel and risky undertaking. But once he had collected a reasonable
diary of information about the prices prevailing in all three cities, his ability
to make the necessary pairwise price comparisons would have improved
greatly. More accurate risk assessment would have come with the benefit of
68 Batten D.F.
3.5 CO-EVOLUTION
The complexity associated with urban evolution arises partly from the
interdependencies which exist between individual learning, collective
behaviour and the design of physical systems to facilitate behaviour. It is
also related to the stage (or epoch) of development which a city or system of
cities has reached. For example, the innovative behaviour of circulating
merchants was a very positive developmental factor for the system of
Medieval cities. It promoted product and knowledge exchanges, urban
specialisation and productivity improvements in many places at a time when
Europe was ripe for such catalytic changes. Although some negative
externalities were doubtless generated, their cumulative impact would not
have been excessive. On balance, Europe was ready to enter a new epoch of
communication, growth and prosperity.
The innovative behaviour of drivers in today's cities may be seem as an
opportunistic response to a system which has attained quite a different stage
of development. Jamming transitions are symptomatic of a heavily
congested urban system where negative externalities have reached
unacceptable levels. Those urban arteries which have served cities so well
over so many centuries are finally showing signs of distress. Under these
conditions, driver behaviour is often motivated more by desperation than
inspiration! Because levels of pollution generated are at their highest under
stop-start conditions, there is a need for some major qualitative
improvements to the system. On balance, such conditions correspond to a
mature system of cities nearing the end of a prosperous epoch of
development. Their future is uncertain.
Emergence and Co-Evolutionary Learning 71
Logistical networks are those systems in space which are used for the movement of
commodities, information, money and people.
72 Batten D.F.
relatively low. In the face of multiple candidate states, the question of how a
particular state mayor may not be reached is of particular importance. This
says nothing about how systems of cities might evolve interdependently.
Acknowledgments
An unfinished draft version of this chapter was prepared while the author was
visiting the Politecnico di Torino. For providing generous support to facilitate these
visits, sincere thanks are. due to Professor Cristoforo S. Bertuglia and the National
Research Council (CNR) of Italy. The author is also grateful for the comments
provided by the audiences attending his lectures at the Politecnico di Torino, the
University of Bologna and the University of Venice. Specific comments from
Dimitrios Dendrinos, Britton Harris, Dino Martellato, Sylvie Occelli, Aura Reggiani
and Angela Spence are also acknowledged.
REFERENCES
Andersson, A.E. (1985) Presidential Address: The Four Logistical Revolutions. Papers of the
Regional Science Association, 59: 1-12
74 Batten D.F.
Dimitrios S. Dendrinos.
Urban and Transportation Dynamics Laboratory, University of Kansas, USA
4.1 INTRODUCTION
The notions of 'development' and 'evolution' are not equivalent, although they are used
interchangeably in this chapter. In Dendrinos D.S. and Mullally H. (1985), the definition
of 'development' is given as: a dynamic path of a dynamic system given an initial
condition and a fixed set of parameter values in phase space, whereas 'evolution' is
defined as a phase transition in phase space.
78 Dendrinos D.S.
(4.1)
for a full understanding of these linear dynamics, their rules and their
solutions, is absent from nonlinear dynamics.
Consider for example the case of the differential equation, also given in
Tufillaro et al:
(4.2)
This is a discrete time model with relative population stock dynamics and
one dimension, as it contains a single state variable, x. The only parameter
[a] employed by this model, captures all relevant factors from the
environment within which the state variable x has been normalised. As a
resurt, the logistic prototype is not unlike the Malthusian model of
exponential growth, where all complicated environmental factors have been
included (discounted) in the exponential growth rate parameter. In (4.3), a
replication (or reproduction) function is presented, where a future time
period's relative (normalised) stock size is a deterministic function of the
current relative stock size. Time-delay effects can be easily incorporated in
these discrete dynamics, but this is not necessary for our purposes.
Whether this is simply an interesting mathematical tool, or in any way
reproduces in model form some real dynamic event of relative population
growth and decline is a question far deeper than academic speculation can
probe. We shall revisit the issue later, after examining certain key qualitative
ingredients of this simple mathematical formulation. This prototype presents
us with only three distinct qualitative dynamics, well analysed in the
mathematical literature, as one scans the linear parameter space within a
bifurcation diagram:
I. in the area of the parameter space O<a<3 a time independent fixed point
solution x* occurs (with extinction being the attractor outcome in the
range O<a<l; the solution x* = 0 is always a solution to the quadratic
map, albeit unstable beyond a = 1).
11. precisely the moment when the environmental parameter [a] becomes
equal to three, the fixed point is transformed (through a flip bifurcation)
into an unstable equilibrium (repeller) and a stable period-2 limit cycle.
In effect, as the scanning of the linear parameter space continues in the
bifurcation diagram and when a= 3, a cascade of stable period-doubling
cycles sets in at geometrically decreasing intervals given by one of the
Nonlinear Dynamics, Innovation and Metropolitan Development 83
have the same dynamic effects, i.e. they result in a change in a model's
parameter(s) or its initial state.
We make a distinction between basic and marginal innovation, in order to
be able to say something about them, their possible origins and their likely
effects. Both types of innovation can be assumed to first appear randomly at
a particular point in time and space (a metropolitan area), and then to diffuse.
It should be noted, however, that this is a controversial assumption.
Simultaneous or independent appearance of a basic innovation at different
~ocations is also possible to imagine. In any case, they are both construed as
being predominantly exogenous to such setting's on-going dynamics. Put
differently, they are not the likely outcome of internal in-place metropolitan
dynamics, but rather an initial shock to these dynamics. They generate what
Jacobs (1969) called 'new work' and, in the case of basic innovations, what
Schumpeter (1934) described as 'creative gales of destruction'.
a) Basic innovations
is quite some way away from the two stable periodic solutions Xl * and X2*
given respectively by:
Assume now a shock of lesser magnitude, but one which still results in a
bifurcation. Whereas the qualitative effect would be the same if the fast
dynamics are perturbed by changing the environmental parameter [a] from a
Nonlinear Dynamics, Innovation and Metropolitan Development 87
value of, say, 2.99 to 3.01, the quantitative effects are relatively minor in
difference: the fixed point x*= 0.665 ... when a = 2.99 is quite close to both
stable periodic solutions x\*= 0.7005 ... and X2*= 0.6339 ... when a =3.0l.
Thus, proximity to a bifurcation point is substituted by severity of the
exogenous shock in the ensuing phase transition: close proximity implies
less severity as far as the simple logistic prototype is concerned. What this
implies is that a basic innovation associated with a large size jump in a
model's parameter might have the same effect as one with a smaller size
change, depending upon the location of the system's dynamics in the
bifurcation diagram in reference to critical (bifurcation) points.
A large change in parameter value might be due to a fundamental basic
innovation. For example, the advent of electricity might have been more
fundamental than (and must come before) the invention of the microchip,
and thus the former might be depicted by a larger magnitude change in the
environmental parameter than the latter. Under some sequence in parameter
jumps, including timing and magnitude, one might structure the underlying
causes of socio-spatial evolution. The other cause might be attributed to a
series of perturbations in initial conditions to be discussed in the next
subsection. However, both conclusions regarding proximity to a bifurcation
point and impacts of small or large changes in parameter values upon
solutions (end-states) are by no means general conclusions applicable to the
whole gamut of nonlinear dynamics. This is another strIking example of the
locality involved in the analysis of these dynamics.
But beyond differences in end-state, phase transitions due to bifurcations
also imply that the path to the end-states is very different: the type of
oscillatory movement towards the final equilibrium value under fixed point
dynamics is quite different from that involved in a motion converging to
either a stable 2"-period cycle or to quasi-periodic and non periodic
solutions.
Here, a noteworthy technical point should be made. Insofar as the
number of iterations is concerned, another interesting feature emerges from
the above quadratic map related example. Given some starting condition
x(O), the number of iterations (i.e. discrete time periods) necessary to
converge to any arbitrarily chosen neighbourhood of the fixed point
(solution) is relatively small at parameter values further away from
bifurcation points. The number of iterations increases almost exponentially
fast as one draws closer to critical borders in parameter space.
Put differently, the closer one is to a bifurcation point in the quadratic
map, the more difficult it becomes to recognise a solution (for example, to
distinguish fixed point behaviour from a stable two-period motion, a stable
two-period from a stable four-period motion, etc). A period of uncertainty
prevails at the borders of phase transitions as to the final outcome of a basic
88 Dendrinos D.S.
innovation, the more so the closer to the border one is. Clean breaks, i.e.
innovations which involve great changes in parameter values may be more
easily susceptible to detection (since they leave more distinct fingerprints),
than innovations causing small shocks in parameter values in the bifurcation
diagram.
As parameter change implies a considerable shock to metropolitan
dynamics, it must be the result of an expenditure of resources. As different
parameter changes may correspond to different speeds, they will necessitate
different levels of resource expenditure. The two could well be correlated,
and constitute a problem of potential research interest.
b) Marginal innovations
arenas, changes and novelty are constantly present and usually surrounded
by an element of mild surprise.
Art and architecture, fashion and law are fields where these marginal
innovations have been vividly abundant at all points in space and in all time
periods throughout human history. Almost by necessity, socio-spatial
systems feed from these abundant, albeit minor, innovations. Examples
include the novelty in sculptures and buildings, clothing and legal statutes.
Plans and policies for metropolitan growth are further cases in point. Some
of them may be endogenously produced within a metropolitan setting, but
others are likely to be subject to outside influence. Exogenous creativity is
unquestionable in the case of art and architecture, fashion and design, but it
is doubtful for law and planning, these areas of human activity being very
much the product of endogenous metropolitan dynamics.
Within a nonlinear dynamic context, the most appropriate way of
accounting for these marginal innovations (in either products or processes) is
by exercising the other freedom one obtains through the use of dynamic
modelling: i.e., by altering initial conditions in a setting's dynamics. The
source of changes in the initial perturbation could be attributed to either
actions by agents within the metropolitan setting in question, i.e. these
changes are endogenous to a particular metropolitan area as they originate
within a spatial setting, or they could be exogenously imposed.
Actions by individual agents almost always contain an element of
innovation. As a consequence of the individual actions constantly affecting a
metropolitan area, many of the plethora of marginal innovations will cancel
each other out, but a few will accentuate the perturbation. Obviously, the
vast majority of individual human activities are so minor as to be
inconsequential, not even disturbing the initial conditions. But some will
transcend the confines of routine activity. Since individual or collective
agents' action is what fuels these lesser innovations, the motives behind
them can be found in speculative agent behaviour, a subject to be further
explored in Section 4.3. .
Dynamic analysis generally considers initial conditions to be of
paramount importance when in the phase space there are separatrices
delineating areas of different dynamic behaviour (possibly involving areas of
distinct chaotic attractors and fractals). Nonlinear dynamics clearly indicate
that sensitivity to initial conditions matters particularly when a system is in
either quasi-periodic or non periodic (chaotic) motion, and vice versa. This is
the reason why, as already stated, chaotic regimes in nonlinear dynamics are
characterised, among other requirements, by extreme sensitivity to the initial
perturbation.
Phase transitions follow parameter changes, i.e. basic innovations occur
when bifurcation points are present. But phase transitions are also possible
90 Dendrinos D.S.
when starting conditions cross over the separatrices in phase space due to the
push of marginal innovations push. In addition, alteration of initial
conditions (i.e. marginal innovation) could have significant effects when the
dynamics are quasi-periodic or chaotic. As a result, within this framework
exists the potential for two types of marginal innovations: those more
important ones, which guide dynamics in crossing over separatrices, and
those less important ones which, if the dynamic system finds itself in nOn
periodic regimes, might cause it to divert to different end-states.
Effects due to diversion can be shown quite easily. Continuing with the
example from the quadratic map, using a pocket calculator with only 11
significant digits approximation and assuming that a = 3.9 when x(O) = 0.1,
the 5-time period sequence for the state variable is: 0.351, 0.8884161,
0.38661843972, 0.924864025, 0.27101318509. By slightly altering the
initial state to x(O) = 0.1001, the same time horizon sequence is:
0.351311961, 0.8887782815, 0.38552064646, 0.92388846269,
0.2742424277, and at only the ninth iteration have the corresponding values
become 0.5379486459 and 0.4911126596.
Such divergency effects, due to extreme sensitivity to starting values,
brings about the problems of truncation, approximation and estimation
(calibration) of either initial conditions or parameter values in view of an
error accumulation. Is it possible to record in a reliable manner dynamic
paths at any level of approximation and for a given time horizon and
accurately reproduce them? Is it possible to statistically estimate parameters
when confronted with extreme sensitivity conditions (i.e. in regimes of quasi
or non periodicity), independently of how extensive the time series are?
One could speculate that social systems might in effect have built-in
processes to correct errors due to approximations in dynamic series of social
behaviour! A high frequency in political processes (frequent elections) might
be viewed in such a light. Beyond this, we do not intend to add further here
to these potentially important questions, except to point out that innovation
might be feeding from the inability to record, estimate and reproduce such
processes accurately (if at all).
Besides differences in end-state configurations, phase transitions imply
that the way towards an end-state is significantly altered. Otherwise, changes
in initial conditions become irrelevant as far as the end-state is concerned. It
may, however, have a mild effect upon the number of iterations and hence
the time which elapses from the initial perturbation to the terminal time
period. This has been referred to as the' butterfly effect', and was first stated
by Lorenz (1963), who cited the case where the flapping of the wings of a
butterfly in Brazil causes the weather to change a week later in New York
City.
Nonlinear Dynamics, Innovation and Metropolitan Development 91
Here, I should like to make a final point about innovations and nonlinear
dynamics. Innovations, whether basic or marginal, are in effect both
92 Dendrinos D.S.
Relative dynamics, along the lines of the quadratic map, express the size
of the state variables in normalised terms and with reference to a broader
environment. Normalisation thus automatically ties up the local
(metropolitan level) dynamics with the global (environmental) ones by
expressly modelling the local level's share of the total (be they shares of
population, income, wealth, etc). The broader the environment over which a
locality's variables have been normalised, the more global is the source of
impacts affecting the locality.
A number of principles seem to underlie relative dynamics, particularly
in relation to inter and intra-metropolitan dynamics. A few of these
principles are pertinent here. Smaller shares identify the larger environment
within which a locality's state variables are normalised. An environment
which is more extended in space in turn implies a locality more open to
outside influences. The smaller a locality with regards to its effective
environment, the more likely it is that the locality's dynamics may be
turbulent. Larger settings are less likely to be subject to chaotic dynamics
since the more open a setting is to outside influences, i.e. the more
interactions there are with its environment or the larger its environment, the
more likely it is that the setting's aggregate as well as disaggregate
metropolitan dynamics are unstable.
Dynamic instability implies that a setting (as a whole or in parts) is
driven either to extinction or to a far larger size. Instability is manifested,
among other things, in the rank size distribution of settings, where a very
few very large (and wealthy) ones coexist with a very large number of very
small (and poorer) settings. It is also found in the incessant movement of
metropolitan settings up and down the urban hierarchies, as well as in the
continuous change at the top of urban hierarchies.
These principles, together with a few others connected with the
magnitude of interaction, are tied to some basic findings associated with
nonlinear dynamic systems, Oendrinos (1992). A number of these are based
on conclusions reported initially by May (1974).
Nonlinear Dynamics, Innovation and Metropolitan Development 93
In view of these principles, one might expect the smallest and the largest
metropolitan areas to be far more vulnerable to exogenously imposed basic
innovations as they are exposed to more severe impacts than are the settings
occupying the middle ranges of urban hierarchies. Although larger settings
(due to scale effects) are more likely to develop buffers to soften the impact
of exogenous shocks, giant metropolitan agglomerations may be simply too
open to global forces for their size. Medium size settings tend to be more
diversified in their employment base relative to their size than either the
smaller ones or the giant ones, although there may be always exceptions.
Medium size metropolitan areas may principally consist of those moving up
or down the hierarchy. According to these hypotheses, globalisation (i.e.
world-wide interactions) must clearly be viewed as a destabilising factor in
metropolitan dynamics, whether of an inter or intra-metropolitan character.
Given this likely reaction to an outside environment by both small and
large metropolitan settings, the question is: does environmentally instigated
innovation increase the chances for instability? If the answer is yes, then one
should also expect that innovation (basic or marginal) fuels further increases
in spatial (and possibly sectoral) disparities, i.e. dualism, since dynamic
instability in metropolitan development leads to disparities.
To address such a question, it is necessary to have a broader dynamic
framework is than that given by the quadratic map or the mathematical
ecology of May. Dendrinos and Sonis (1990) have formulated the dynamics
of a 'universal map', where the term universal is used to point out that its
dynamics could potentially be inclusive of a vast variety of qualitative
dynamic effects, far richer than the ones discussed earlier and found in the
simple logistic prototype (which emerges as a special case of these universal
dynamics).
Analytical as well as numerical experimentation with this universal map
through computer simulation seems to indicate that dynamic stability is the
overwhelming outcome in the universe of parameter space in low
dimensional cases (where the number of state variables is no greater than
four). But small islands of instability do exist in that sea of calm dynamics,
and there are also islets of turbulent motion. The greater the system's
dimension (and the greater the number of parameters), the more likely it is to
randomly experience stable dynamics. In fact, turbulent dynamics are
eccentric events.
These findings stand in sharp contrast to those reported by May, and
demonstrate the fact that, because no general theory of nonlinear dynamics
exists, the analytical framework one decides to adopt more or less dictates
the qualitative findings one obtains - and maybe vice versa.
One might conclude that this is a cynical view of science, particularly
social science. However, a more careful look at the implications of this
94 Dendrinos D.S.
dictum leads to some important and deep conclusions regarding social reality
and social action. Social reality is perceived by social actors who act on the
basis of these (possibly coexisting and multiple) perceptions, thus affecting
it. Although some of these perceptions may have been way off the mark, this
is irrelevant. What matters is that actors act on such perceptions, thus
disturbing dynamic paths. Innovation is triggered by such actions - the more
unusual and unexpected the actions and perceptions, the greater the impact.
In effect, as evolution is tied here to disturbance of paths, one may conclude
that the existence of multiple and severely varying perceptions is a necessary
condition for socio-spatial evolution!
Creative actions by individual social actors (agents for change) are
beyond the norm and the repetitive routine of human activity. New work
requires the realisation of actions which are at great variance from the
average; in other words, outliers are at the centre of innovation. Creativity in
innovation, novelty and surprise, are the sine qua non not only for
metropolitan development but, more broadly, human development in
general.
A final point to this section. Perturbation of the initial conditions of
metropolitan dynamics, which is the effect of marginal innovation, requires
far fewer resources than those required by a basic innovation, because it is
simply much easier to perturb trajectories than change them.
How one deals with innovative action of the marginal type in
metropolitan development is the subject of the next section.
could lay down a foundation for analysing metropolitan (and broader socio-
spatial) development. Using traditional evolutionary arguments of
competition, one would expect the winner of a speculative (guessing) game
to amass more resources than others and thus have more to spend on
speculative actions in the future. This would imply that, in an evolutionary
context, competition would favour those who tend to guess right, either
because they possess the right information or due to favourable chance (good
luck). Consequently, one would expect expectations markets to eliminate
those who tend to guess wrong (either because they did not make informed
choices or due to adverse chance), as they eventually exhaust the resources
allowing them to speculate. Under this evolutionary argument of
competition, markets would only comprise of winners!
Obviously, this is not possible, as there are at least two fundamental
flaws to this evolutionary argument:
l. winning requires that there be losers, and in fact there are always far
more losers than winners at any point in time. The argument would
require then that, although there must be winners, those who win are not
necessarily the same all the time, or even most of the time, as life cannot
afford to have only a few that guess consistently correctly. Thus, an
evolutionary argument concerning the competition for right guesses
requires that a co-operative association exists between a few winners and
many more losers, and that the roles be constantly shifting among
individuals, commuting between these two groups;
ii. the term 'informed' can never be granted to a judgement ex ante but ex
post Jacto. It is only the results of speculative action which indicate
whether a judgement was informed or not.
Instruments for speculative action include, but are not limited to, the
availability and purchase of commodities, stocks and bonds, as well as
derivatives of these tangible products (commodities, resources) and financial
Nonlinear Dynamics, Innovation and Metropolitan Development 97
a few actors can have a significant effect on the performance of the markets.
Vice versa, a market action may trigger individual responses which, if
coming from influential actors, can have catastrophic effects (in the
mathematical sense of the word) on the market.
A few of these imperfections will be discussed at greater length in the
context of metropolitan markets. In spite of these and other criticisms, the
use of derivative instruments offers considerable advantages which we shall
also touch upon later. Firstly, we shall provide a few basic introductory
definitions for two instruments, options and futures, for the reader who is
unfamiliar with them.
Options and futures are legal contracts offering their holders certain
purchasing rights under strict expiry conditions. Those rights are acquired in
appropriate derivatives' markets and at prices determined by free and
continuous trading. Markets for options and futures may be formal or
informal, i.e. explicit and regulated, or implicit and unregulated. Futures are
contracts which offer their holder the right (but not the obligation) to
purchase a prespecified quantity of a stock (commodity or resource) in
futures markets, at a prespecified future point in time at a price which is
determined currently (at the time when the contract was purchased). In
effect, the market (presumably a large number of individuals) speculates on
the future price of the underlying issue by discounting for all perceived
factors which could possibly affect the price at that time in the future.
At all points in time between the point of purchasing the contract and the
its expiry or selling date, the holder maintains a current balance (in the form
of either credit or debit) on the value of the contract. Thus, the exposure to
risk could be considerable, as relatively large quantities of the underlying
issue (stock, bond, commodity or resource) is contracted.
Options are contracts giving the holder the right (but again not the
obligation) to purchase (or sell) a prespecified quantity of an issue (stock,
bond, commodity or resource) at a prespecified point in time in the future
and at a preset price (referred to as the striking price). The price of the
contract is determined currently on the appropriate option's market. There
are numerous types of option (call vs. put options, U.S. vs. European
options, etc), but for the purpose of this exposition, these differences are not
relevant. The sellers of these contracts are governed by equivalent
conditions. How and what transaction costs enter into these markets is
largely irrelevant here, although it is recognised that the magnitude and type
of transaction costs involved could potentially affect the performance of
these markets.
Nonlinear Dynamics, Innovation and Metropolitan Development 99
Since the most one might lose by participating in an options market is the
price of the acquired options contract(s) (usually a fraction of the value of
the contracted issue), the exposure to risk is much less than for futures
contracts. On the other hand, the time horizons over which futures are
contracted are far longer than those for options. So options, generally,
simply pick up the value (premium) of the remaining time before the
option's expiry date.
Introduction of innovation within derivatives markets may have profound
effects upon these markets. For instance, a decrease in the reaction times to
information diffusion among actors or to trading may offer opportunities for
greater efficiency, but also traps for exaggerated responses.
Derivatives result from the need for risk management. Firstly, because of
the durability of large scale public capital investments, the viability of
alternatives (courses of action) can be sorted out in the derivatives markets
before costly public projects are undertaken. Projects are often built before
obtaining signals concerning potential failure. The planning literature is full
of accounts of unfinished projects or schemes which ended in disaster,
giving testimony to failed planning.
A second reason for the existence of derivatives is that they present
opportunities for those who hold large quantities of an underlying
commodity, resource or any primary financial instrument (stocks, bonds) to
hedge against undesired price movements. The presence of options and
futures markets is a form of insurance. In a metropolitan scene, this concern
for insurance mostly affects land holdings (although by no means not
exclusively). In brief, the existence of options and futures contracts markets
for public plans could offer the opportunity for greater flexibility in deriving
more socially optimum plans or policies more formally and effectively.
On the other hand, criticisms of derivatives' markets abound. For one
thing, these markets tend to be dominated by a few actors-participants, and
monopolistic or oligopolistic conditions are commonplace. Bandwagon
feedback dynamics are widespread, as minor actions or erroneous
information can result in major price swings. As options (or futures) contract
prices increase, the volume of contracts demanded tends to increase too, in a
pattern of trading where time is at a premium. This interdependency between
volume and price can lead to a mass phenomenon of panic buying. Similarly,
as contract prices decrease, the same frenzy can occur, causing prices to
decline yet further as contracts holders decide to exit from the markets as
fast as they entered it.
100 Dendrinos D.S.
In the final part of this section, we address the role of plans for
metropolitan development. It is argued that there is a greater need to employ
options on metropolitan resources (including land, water, air, the physical
urban environment, and even its historical landmarks) as well as on
metropolitan infrastructure (housing, utilities, transportation network, public
services etc.). Above all, there is a need to derive an options markets for
formal plans.
Options on metropolitan and regional resources is an instrument for
investment already in place in the US, as well as in other Western countries.
They are usually employed in the form of "rights", although in a strict sense
rights and options (or futures) are not identical. However, the differences
between them, which are mostly of an accounting nature, are of no interest
here. Land transactions represent the most widely used forum for their use.
Four examples of rights already widespread in the US are the transfer of
'development rights', 'water rights', 'pollution rights' and 'congestion
rights'. Although these rights behave in an option-like manner, because of
their thin markets they are not traded like options in formal options markets.
In this difference lies the potential for significant innovation, but also the
risk of pitfalls.
In such formal markets, the effects of a large set of metropolitan
development policies could be fully and effectively evaluated. Policies might
be traded in a more formal public forum than the largely informal ones
currently used for metropolitan policy markets. A number of ways of
promoting greater and more systematic public participation and optimisation
(or sub optimisation) in the planning process, as advocated by Harris in this
book, might be more fully addressed in this context. The question of
innovation of this kind in the approach to drawing up metropolitan plans and
policy-making would require much deeper consideration than is possible in
102 Dendrinos D.S.
the confines of this chapter. It constitutes in fact a call for further research as
it is clear that there is a need to address the economic, financial, legal,
institutional, social and other problems associated with setting up these
formal options markets. Such issues, although extremely interesting, are not
of course the main subject of this chapter or book.
a) Plans as Options
Economic and social (or, more broadly, cultural) activity settles, i.e.
locates at specific points in space and time, and thereafter develops as a
result of the spatio-temporal choices exercised by the human population.
Metropolitan areas themselves are the outcome of concrete choices exercised
by a myriad of individuals and collectives when confronted with alternatives
in space, in a much broader context than that envisioned by Tiebout (1956).
When the population of a metropolitan area grows, or its economic
output and wealth increases, or its resources appreciate in value relative to
other competing metropolitan areas, this implies that the city's fortunes in
options markets looks brighter and its option's price is on the move. On the
other hand, when a metropolis declines (in terms of population, employment
or wealth) or is abandoned by its population, this simply means that its
option price has collapsed in metropolitan growth-options markets within a
national or global context.
Marginal innovations, occurring through individual actions, which either
set in motion the construction of cities or disturb their dynamic paths, for the
better or the worse, are central in metropolitan development and evolution.
Individual actions manifest themselves in highly interactive, albeit
imperfect, options and futures markets. It therefore seems of some
importance to investigate the specifics involved (qualitatively and
quantitatively) in derivatives markets if we are to understand better the
process of metropolitan development.
104 Dendrinos D.S.
4.5 CONCLUSIONS
legal, institutional, economic, financial, social and other aspects of such plan
options markets would need to be addressed. These represent another set of
issues requiring further research, especially the question of how the
hypotheses presented could be translated into the form of a mathematical
model.
Acknowledgements
This chapter was prepared while the author was a Visiting Scholar at the Politecnico
di Torino under a grant from the Italian National Research Council (CNR). An
earlier draft of this chapter was used as the basis for Class Lectures at the
Department of Regional Development at Turin Polytechnic. Support from Professor
C.S. Bertuglia was instrumental for the realisation of this project and is gratefully
acknowledged. Helpful comments were received from Professors Britton Harris,
David Batten and Dino Martellato, as well as from Sylvie Occelli from IRES and
Angela Spence. The usual caveats, however, apply.
REFERENCES
Bertuglia, C.S., Fischer, M.M., Preto, G., eds (1995) Technological Change, Economic
Development and Space. Springer-Verlag, Berlin
Cimoli, M., Dosi, G. (1995) "Technological Paradigms, Patterns of Learning and
Development: An Introductory Road Map". Evolutionary Economics 5: 243-68
Dendrinos, D.S. (1992) The Dynamics of Cities: Ecological Determinism, Dualism and
Chaos. Routledge, London
Dendrinos, D.S. (1996) Chaos: Challenges from and to Socio-Spatial Form and Policy,
Discrete Dynamics in Nature and Society
Dendrinos, D.S., Mullally, H. (1985) Urban Evolution: Studies in the Mathematical Ecology
of Cities. Oxford University Press, Oxford
Dendrinos, D.S., Sonis, M.(1990) Chaos and Socio-Spatial Dynamics. Springer, New York
Devaney, R.L. (1989) An Introduction to Chaotic Dynamic Systems. Second Edition,
Addison-Wesley, New York
Feigenbaum, M.J. (1980) Universal Behavior in Nonlinear Systems. Los Alamos Science 1: 4-
27
G1eick, J. (1987) Chaos: Making a New Science. Viking, New York
Jacobs, J. (1969) The Economy of Cities. Random House, New York
Lorenz, E.N. (1963) Deterministic Nonperiodic Flows. Journal of Atmospheric Sciences 20:
130-41
May, R. M. (1976) Simple Mathematical Models With Very Complicated Dynamics. Nature
261: 459-67
Penrose, R. (1989) The Emperor's New Mind. Oxford University Press, Oxford
Poincare, H. (1957) New Methods of Celestial Mechanics. (Reprinted from 1899 French
Edition), Dover
May, R. (1974) Complexity and Stability in Model Ecosystems. Princeton University Press,
Princeton
Schumpeter, J.A. (1934) The Theory of Economic Development, Harvard University Press,
Cambridge
106 Dendrinos D.S.
5.1 INTRODUCTION
their presence. A different set of events at the outset could have steered the
locational pattern to a different outcome, so settlement history is crucial.
Because of the existence of multiple equilibria, minor changes in the values
of the critical parameters may generate dramatic changes in the equilibrium
spatial configuration. This suggests that historical events can account for
current industrial patterns, and that circular causation generates a snowball
effect that leads manufacturing firms to be locked-in to the same region for
long periods of time (Arthur 1994).
Very little statistical data are available in Japan for the locational
behaviour of research laboratories, especially those belonging to private
companies. While data on the total number of research laboratories in each
Prefecture are available, they are highly aggregated and therefore unsuitable
for micro behavioural analyses. Moreover, the 'research laboratories'
category may include various other research-related centres and laboratories,
along with the research divisions of companies in the manufacturing
industry. Thus, in this study, the database was constructed by extracting
research laboratory data from the Directory of National Experimental and
Research Organisations, published by the Science and Technology Agency.
In this way, the relationships between these research laboratories and their
parent companies are transparent. In addition, details about the relationships
between the locations of research laboratories, the functions of the parent
companies, and the locations of their factories were ascertained from the
Survey of National Industrial Factories (Nikkan Industrial Newspaper). Due
to the restriction of our focus to laboratories that perform research functions,
we have examined those industries with high R&D investments, such as the
electrical and electronic, mechanical, chemical, medical, and automobile
industries. From these industries, we have chosen the one hundred and forty
two research laboratories of companies with listed stocks, on the basis that
they furnished sufficient data for the intended observation.
114 Kobayashi K, Kunihisa S. and Fukuyama K
Table 5.1 shows the number of research laboratories in the sample which
have located in each area since 1950. The number in brackets shows the
percentage of the total sample number. It can been observed that there is a
strong tendency for research laboratories to concentrate in large cities,
especially the Tokyo Metropolitan Area, which includes the Ibaragi,
Saitama, Chiba, and Kanagawa Prefectures. The total number of new
laboratories set up reached its peak during the period 1961-1970.
Total 34 21 19 35 3 30 142
% total sample (23.9) (14.8) (13.4) (24.6) (2.1) (21.1) (100)
120
100
80
60
40
20
120
100
80
60
40
20
Total
o General Mechanical
Industry
~~ '. Automobile Industry
ro
.Q +
'n
.c \
U \
Chemical Industry
Tohoku Shif:j,kansen; {
/ Ibaragi Pref.
I
Tochigi Pref. I.
I
I
I
I
o more than 20
o 10-20
o 5-10
less than 5
ml\ior expressways
Shinkansen railways
(5.1)
where vi} is the deterministic part of the utility function which is common to
all laboratories, &; is the stochastic part of utility that fluctuates
stochastically due to unobservable factors, and ()'= (Ot, ,Om! are
parameters to be estimated. Now, the probability Pi)' that point / is chosen
from amongst the possible location points in Ai can be given as:
p~.y = Prob{V.n.y~ ,
V;~ for all j in Ai} (5.2)
(5.3)
let the attributes vector of point Bij E Ai' (}xij' be expressed as follows (by
utilising the vector ~ij that represents the estrangement from the average
attribute of the arbitrary possible location point in choice set Ai, Xi:
(5.6)
The Knowledge-Intensive Nature of Japan's Urban Development 121
(5.7)
(5.8)
(5.9)
(5.11)
The data set used for model estimation was gathered from eighty research
laboratories located throughout the Tokyo Metropolitan Area (including
Tsukuba). All data were pooled for parameter estimation. The attributes of
the chosen location were described using data drawn from the specific site in
that year when the laboratory concerned was actually located. Those
locations which were not chosen were represented by the average values of
the corresponding alternatives in each location. In calculating the average
value of the attributes of the Tokyo Metropolitan Area (option A2 ), the
whole area was divided into six zones, and the average attributes of these
areas was assumed to represent the average locational attributes of each
location. Alternative models were estimated according to different
combinations of explanatory variables. The desired estimations are
summarised in Table 5.3, which has the largest likelihood ratio among the
estimates satisfying the desired conditions. The likelihood ratio and hit ratio
are 0.806 and 95.4%, respectively, and therefore the model fits the data well.
The selected explanatory variables are: 'the number of researchers', 'land
price', 'travel time to Tokyo', and 'population density'. In calculating the
number of researchers, the researchers in the public and private sectors have
been included. 'Population density' also includes those working in both
public and private sectors. Hence, the variables 'travel time to Tokyo',
'number of researchers' and 'population density' can be partially controlled
by policy initiatives. However, the basic feature of the estimated results is
that the attractiveness of Tsukuba is regulated by endogenous variables
which serve as proxies for agglomeration economies. In other words, it can
be seen from Table 5.3 that the agglomeration itself furnishes the conditions
for further agglomeration.
The Knowledge-Intensive Nature of Japan's Urban Development 123
points that help agents co-ordinate their spatial decisions. Reshaping the
urban landscape therefore requires major changes in agents' expectations
(Krugman 1991). Although there is a great deal of flexibility in the choice of
location a priori, there is also a strong rigidity of spatial structure once the
process of urbanisation has started. These factors could explain why many
planned research cities have failed to develop successfully. Governments
have often withdrawn their support from key urban activities before some
critical mass could be reached (Fujita and Thisse 1996).
1 r-------------------------------~----~~
Number of national
09 r---------~~~~~~~~----~~----~~~~----~
research institutions (2)
08 r---------------------~_=--~~~----------- r----~
01 r-------------------------~~~--------.----------~
06 r-----------------------~~~---------------------~
os r-------~~------~--~~--~----~~----------~
03 r-------------__~-:--~--~----------------------~
02
Ol ~~------~~----------L-------------------------~
o - A --"-'-_
1975 1980 1985 1990
(I), (2), (3) are nonnalised such that their 1990 values = 1.0
5.6 CONCLUSIONS
REFERENCES
Arthur, W. B. (1994) Increasing Returns and Path Dependence in the Economy. Ann Arbor,
The University of Michigan Press, Michigan
Batten, D. F., Kobayashi, K., Andersson, A.E. (1989) "Knowledge, Nodes, and Networks: An
Analytical Perspective." In Knowledge and Industrial Organization, A. E. Andersson et al.
(eds. ), Springer-Verlag, Heidelberg Berlin
Florax, R. (1992) The University, A Regional Booster? Avebury, Aldershot
Fujita, M., Krugman, P. (1995) When is the Economy Monocentric? Von Thtinen and
Chamberlin Unified. Regional Science and Urban Economics 25: 505-28
Fujita, M., Ogawa, H. (1982) Multiple Equilibria and Structural Transition of Non-
Monocentric Urban Configurations. Regional Science and Urban Economics 12: 161-96
Fujita, M., Thisse, J.-F. (1996) Economic Agglomeration. Discussion Paper 430, Institute of
Economic Research, Kyoto University, Kyoto
Kobayashi, K., Batten, D.F., Andersson, A. E. (1991) The Sequential Location of Knowledge-
Oriented Firms Over Time and Space. Papers in Regional Science 70: 381-97
Krugman, P. (1991) History Versus Expectations. Quarterly Journal of Economics 106: 651-
67
Krugman, P. (1993) First Nature, Second Nature, and Metropolitan Location. Journal of
Regional Science 33: 129-44
Lucas, R. E. (1988) On the Mechanics of Economic Development. Journal of Monetary
Economics 22: 3-22
McFadden, D. (1978) "Modelling the Choice of Residential Location." In Spatial Interaction
Theory and Residential Location, Karlqvist, A. et ai, eds, North-Holland, Amsterdam, 75-
96
Markusen, A., Happ, P., Glasmeier (1986) High Tech America, The What, Where. and Why of
the Sunrise Industries. Allen & Unwin, Boston
Marshall, A. (1920) Principles of Economics. 8th edition, Macmillan. London
Matsuyama, K. (1995) Complementarities and Cumulative Process in Models of Monopolistic
Competition. Journal of Economic Literature 33: 701-29
Oakey, R. P. (1984) High Technology Small Firms, Innovation. Frances Pinter, London
Premus, R. (1982) Location of High Technology Firms and Regional Economic Development.
US Government Planning Office, Washington DC
Romer, P. M. (1986) Increasing Returns and Long-Run Growth. Journal of Political
Economy, 94: 1002-37
Chapter 6
Martin J. Beckmann
Technical University. Munich. Germany
6.1 INTRODUCTION
Publications are equally accessible in all locations: the cost and time
difference in the mailing of journals are not significant. Moreover, research
is essentially a footloose industry, tied neither to 'resource deposits' for raw
material inputs, nor to markets of a large consuming public. It would be
erroneous, however, to conclude from this that location does not matter and
that all places are equally suitable for knowledge production.
As a first observation, we note that the researcher invariably requires an
organisation to provide support, since only organisations can afford the
infrastructure of libraries, advanced computers, data banks and
communication links, not to mention the secretarial and research assistance.
Even Karl Marx, denied a professorial chair, had to immerse himself in the
British Museum to write Das Kapital! Organisations also provide the
opportunity for constant interaction with fellow researchers, who serve as
sources of both reference information and critical opinion. Moreover, the
synergy of teaching and research can be achieved only in a teaching
institution, such as a university.
Secondly, people of scholarly and scientific talent, especially at the
higher levels, show distinct preferences in their choice of residential
location. As their 'consumption' tends to include cultural services, such as
good schools for their children, musical and theatrical events, exhibitions,
and so on, they will opt where possible, to live in places where these are
available, which almost invariably implies a metropolitan area. Students, too,
will be more attracted to metropolitan areas with varied cultural facilities
than to remote places in the hinterland. It is not easy for organisations
located in inaccessible places to draw the specialised personnel needed to
build up a scientific community of renown. The knowledge network of
scholars/scientists is thus imbedded in a network of institutions located, for
the most part, in a network of metropolitan cities.
..
90
80
70
60
~ ----------------
40
2D
10
o~--~~--~--~--~--~~
1900 1910 1920 1930 l!I4n 1!l5(J 1960 I!I7O
MATHEMATICAL APPENDIX
B1 =L;,K
max ,Kj j
b.L~Kf1K~
1 1 1 J
-wL.
1 1
-cK
1 1
-c IJ.. K.J (6.1)
(6.2)
aBj- --yL'
0 -- - TaKPKY-1
. J -c .. (6.3)
aK. 1 1 IJ
J
Dividing:
K = ---.!...
y c
_I (6.4)
KJ fic IJ..
Thus the relative utilisation of outside knowledge decreases with the cost
ratio, cij/cj which in tum means a decrease with distance rjj (say), since cjlrij)
is an increasing function. This must result in a loss of output, i.e. of research
achievement.
Interurban Knowledge Networks 135
REFERENCES
Andersson A., Persson O. (1993) Networking Scientists. Annals ofRegional Science 27:11-21
Beckmann, M. (1994) On Knowledge Networks in Science: Collaboration Among Equals.
Annals of Regional Science. 28:233-42
Beckmann, M., Persson O. (1996) Scientific Collaboration as Spatial Interaction. Institute for
Futures Studies, Stockholm
Crane, D. (1972) Invisible Colleges. University of Chicago Press, Chicago
Katz, 1.S. (1994) Geographical Proximity and Scientific Collaboration. Scientometrics, Vol.
31(1): 31-43
Persson, O. (1991) Regional Collaboration in Science. Data on Nordic Coauthorships 1988-
1990. CERUM Working Paper 16
Persson, 0., Beckmann O. (1995) Locating the Network oflnteracting Authors in Scientific
Specialties. Scientometrics 33(3): 35\-66
Price, D. de Solla (1963) Little Science. Big SCience, Columbia University Press, New York
Stewart, 1.Q. (1950) The Development of Social Physics. American Journal of Physics, 18(5):
239-53
Chapter 7
Britton Harris
Department of City and Regional Planning, University ofPennsylvania, USA
7.1 INTRODUCTION
The coming twenty-first century will be one of great challenge and also
of danger for humanity. The population of the world may more than double
its present level of over five billion. The pressure of population on resources
will greatly increase the danger of starvation and pestilence. Such dangers
will interact with social unrest and political instability, and all of these
together will add to the dangers of war and of environmental catastrophe.
The scene in which most of this drama will be played out will differ in
many respects from the present world. While now far less than half of the
world population lives in cities, by 2050 far more than half will do so. Urban
populations will increase by a factor of more than five, and it will be difficult
to provide resources for infrastructure, transport, and daily living, with the
result that overcrowding and pollution may continue and increase.
Some hope may however be seen in the intrinsic nature of present
technological growth and change. The two most important sectors are the
biological sciences and the information and computing technologies. Both
sectors are relatively environmentally benign in providing a technology
which does not consume vast amounts of power, nor materials which deplete
the environment, or produce residues which poison it. These technologies
can redirect consumption in environmentally less harmful directions, and
provide at the same time other benefits.
Up until now, the world hope for controlling population growth and
increasing the level of world citizenship has rested mainly on a slow
transitional process of social, economic, demographic, and urban-rural
change. In largely urban societies, birth rates have fallen and levels of
literacy have risen. Exceptional rural cases like the Indian state of Kerala and
13 8 Harris 8.
Fifty years ago, at mid-century, the world was recovering from war, and
there was a widespread atmosphere of optimism and growth, particularly in
the USA, with a neo-enlightenment view that rational action could overcome
many social problems. These and other factors conspired to speed the
development of a new understanding of metropolitan function and its
changes. We will try to review these developments in a way which may
serve at least three purposes: we will show concretely how urban analysis
changed and what potentialities grew out of this change; we will present this
in a way which emphasises the innovative nature of scientific progress itself;
and we will show in the end how the science of urban analysis has defined
some of its own limits.
The presentation will take the form of a roughly chronological review of
the more important episodes in this development.
The computer was probably the most important single factor in assisting
this scientific revolution. The first general-purpose digital computer was
invented during the war and used for ballistic computations. The first
commercially available computer, the UNIV AC, was produced in 1950. The
first major use of computers in urban analysis was in network analysis by the
Chicago Area Transportation Study in about 1958 (CATS 1959).
Auxiliary disciplines included in particular Operations Research, with its
emphasis on man-machine systems, and Computer Science, dealing with
algorithmic methods and optimisation along with other tools. Disciplines
like statistics and survey research were revitalised.
Transportation planning introduced the first large scale metropolitan
analysis. Its new methods were developed in the course of a struggle by the
US Bureau of Public Roads to achieve a methodology with which to plan the
already burgeoning system of Interstate highways and urban expressways in
the US. Some elements of this set of methods (many originating in CATS)
were the following:
The' gravity model', which had been in development and use for about
ten years, became a central feature of many urban analyses; contradicting
conventional economic theory, it proposed and showed that similar
people in similar circumstances behaved dissimilarly (for example, not
everyone travelled to the closest available job). Much earlier, the
sociologist Stouffer (1940) had developed an explanation in the idea of
'intervening opportunities' , which was applied in a new form by
Schneider at CATS.
The 'shortest path through a maze' was an algorithm of utmost
importance to all spatial analysis, solving the problem of realistic
measurement of spatial separation between all pairs of localities in a
metropolis. The method was found by Moore (1957) and based on the
Innovation and Urban Planning 141
The high visibility of transportation planning, and the vast scale of its
research efforts energised the whole field of metropolitan locational analysis
in several ways. It showed by example how large-scale urban research could
be mounted. It provided data and methods which facilitated new research.
And by virtue of its shortcomings, real or imagined, it led practitioners and
academicians to undertake novel investigations.
Rough empirical generalisations, like the observed negative exponential
decline of density with distance from the Central Business District (CBD),
inspired statisticians like Colin Clark to prodigies of ex post generalisation,
and others to look for more specific causal explanations. This led to a
prolonged effort to improve understanding of location in general and
housing markets in particular.
models by Mills (1972) and by the staff of the NBER modelling project,
among others. These models all tended to suffer from the difficulty that
linear programming does not 'mix people together', but populates each
subarea with a single household type, or usually at the most two types.
The connection between transportation planning and land-use planning went
beyond the transfer of computer methods and data, with an implicit
recognition of interaction, as in Mitchell and Rapkin, and more explicit
connections showing how accessibility shapes land-use (Hansen 1959).
In the US models called PLUM (Projective Land Use Model) and BASS
(Bay Area Simulation Study) adopted Lowry's methods, sometimes using
actual transportation times to measure separation. The location of 'basic
industry' was made at least partially endogenous. In the UK, similarly
modified Lowry Models became an important implement in the analysis
needed for structure planning.
Putman (1976, 1982) in the US, along with Echenique (see Echenique and
Owers 1994) in the UK, recognised and emphasised the reciprocal
interrelationship between transportation conditions and land use change.
They showed how different transport plans affected land-use development
differentially, thus indirectly affecting transport demand as well, and hence
how transport-dependent land-use models were necessary for transportation
planning. Putman's models - PLUM, DRAM, ITLUP, and EMPAL -
follow the modified Lowry model tradition, and are widely used in US
transport planning agencies. Echenique extended the Lowry tradition in
ways to be discussed, and enjoyed a different type of success in Europe and
Latin America.
In the seventies there was a new round of creative activity, followed in the
eighties by a process of mostly incremental improvements and fine tuning.
The seventies started with a strong attack on modelling (Lee 1973), which
based itself on experiments preceding many of those discussed here. Lee
disregarded these later experiments and the self-critical attitudes of the
model developers, thus utterly miscalculating the future of modelling. The
main effect of his paper was to weaken the connection between theory and
practice and delay the professional use of models outside of transport
planning.
Wilson (1970) found a new derivation of the gravity model and initiated a
more complete new style of locational modelling through the use of so-
called 'entropy maximisation', an idea closely linked with thermodynamics,
but yielding new insights into optimisation in location theory. These new
models were also related to the measurement and interpretation of consumer
and producer surpluses, and hence of welfare. In many cases they depended
on a more careful use of constraints than the pure Lowry model, producing
144 Harris B.
tend to imply clearing markets for land, labour, goods, and services; also,
network equilibrium is a pseudo-market in congestion avoidance. Under
these market conditions, and in the absence of externalities, the competition
amongst locators and users of facilities tends to produce an optimal
distribution of activity within any subsector, subject to the locational
decisions of all other locators or sectors and subsectors.
As may be seen from the diversity of criteria for location, separating
activities by sector or function has been practical and attractive to the
disciplines of analysis. At the same time, it removes some of the most
important externalities from the individual models and puts them in the
realm of sector interaction. This means however that we cannot
automatically be sure that the joint location of all groups will result in a
system optimum. For obvious reasons, optimising all of location involves
taking account of externalities, economies of scale, and indivisibilities.
These can, more unusually, impeach the optimality of each sectoral model in
its own sphere, since these factors favour the appearance of local optima.
It is exactly problems of this type which generate so-called market
failures, meaning that purely economic rules may lead to solution which may
be good but not necessarily the best possible. These failures arise in cases of
indivisibilities, externalities, economies of scale or agglomeration, public
goods, and monopolies. Planners have realised the existence of exactly these
difficulties for years, and mathematical programmers had been frustrated by
the difficulty of solving intransigent problems like that of the travelling
salesman. The discovery of NP-completeness thus in a sense unified the
common difficulties of economics, planning, and computer science.
The reader should be struck by the fact that nearly every paragraph of the
discussion in this section has contained a capsule description of one or more
creative acts of invention in theory or its application to analysis. It is not
surprising, however, that these discoveries may have led to (or resolved) a
situation in which a major new jump would be required to overcome a
newly-identified major obstacle. In facing the unknown, we are mostly
confined to speculation until the exploration has gained momentum, and it is
this spirit that we will examine the territory which now seems beyond a fully
rational limit to engaging in rational action.
outcome of the solution of another, and may then invalidate the first result.
While a rejection of optimising can be fully supported on the grounds that it
is unthinkable to try to find truly optimal solutions to these three problems
jointly, there are still other very important considerations which militate
against formal optimisation ..
The first of these depends on the nature of the planning process itself.
Optimising methods require a full knowledge of the possible decisions: this
is the policy space. Characteristically planners (and designers in general)
tend, in the very process of planning, to invent new measures or actions
which might be taken. Any such invention, if effective, would require a
complete reconsideration of all optimising work done previously. If on the
other hand, optimising commitments were permitted to rule out the invention
of actions, and to enforce their banishment to an earlier stage, much of the
possible gains from invention in planning would be given up.
A second, and overriding, consideration is the fact that most optimising
processes have limited sets of criteria, and that these do not match the
breadth of the criteria which will emerge in discussions in the process of
public choice. It is widely recognised that there are mUltiple bases of choice
which are hard to combine in a single model, and this gives rise to methods
of multi-criterion decision-making, but this is beside the point. We contend
that if several local optima are collected from an optimising process, the one
which is formally the best will often not be judged the best by the users of
the process, because they will use a larger and partly inaccessible set of
criteria and change the ordinal ranking of the cases based on the formal
criteria.
This line of thought underscores the fundamental reason for preserving
less than optimal solutions to a problem and providing both planning staffs
and the decision-making public (or its representatives) with alternatives. The
means by which alternatives can be encountered are not immediately
obvious, and the locus of creative actions in the planning process is obscure,
so we need to discuss innovative planning in the light of both existing
practice and optimising procedures. We will see that in a qualified sense,
much professional practice contains the same elements as formal optimising.
Generally speaking, both creativity and optimising are combinatorial
processes. They require the discovery and selection of good or superior
combinations of decisions from a very large (and in a practical sense,
unbounded) universe of possible decision sets. Since most of these sets are
vastly inferior to a thin layer of those with superior performance, the
identification and selection of good sets is important and may be non-trivial.
We start with the somewhat arbitrary distinction between inventing
elements of a plan, and finding or inventing good combinations of elements.
The elements of a plan may be physical constructs or social measures, and in
Innovation and Urban Planning 149
some basic sense they are themselves combinations of even more elementary
components. Examples which come readily to mind are: the skyscraper
(which in tum depended on the invention of structural forms and of the
elevator or lift); the limited access, divided, multilane highway - and its
extension into a system of such highways; and Federal mortgage guarantees
in the US, which incubated the post-war suburbs. I consider a particularly
charming example to be the decision taken in New York, San Francisco,
Prince Edward Island, and a few other places with toll water crossings of
suitable configurations, to collect tolls in only one direction but at a higher
level. This decision has saved many millions of dollars in collection costs
and in eliminating user delays - yet its discovery has no precedents, no
history that I can discover, and no reason why it should not have been
invented much earlier.
Obviously engineering, architecture, planning, and policy abound with
such examples of inventing elements in the process of planning. Indeed, the
same thing occurs in the course of building scientific theories and
experiments, and in writing, painting, and other creative arts. This process
needs to be pursued in planning, but will be done at the expense of orderly
predefined optimising procedures. Most particularly, new measures may
create new difficulties in predicting the impact of plans which embody them,
and this is a problem for planning analysis. Because the invention of new
measures is so obscure, and its optimum-seeking characteristics so ill-
defined, we reluctantly forego further discussion of it at this point.
There is a handful of standard optimum-seeking procedures which when
fully developed can become actual optimising methods. We discuss the most
important of these in relation to planning practice.
The most fundamental optimising procedure, and in many ways the least
creative, is the application of successive improvements to a complete
configuration up to the point where no further improvements are available.
This is called hill-climbing and is a method of finding a peak in
multidimensional space. In discrete space, methods of the calculus (dear to
engineers) may not be available, and there is no analogy to finding Mount
Everest on a clear day by triangulation, because the peaks and their ranges
are not visible.
Successive improvements are often tested by swapping elements (or
locators), and swapping is regarded as a fundamental procedure for discrete
problems. It is the method which, for example, permits linear programming
to proceed inexorably toward a guaranteed optimum. This guarantee depends
on convexity, and planners should know that most of their problems are non-
convex. Thus there is always a danger that effort is being applied to
improving the wrong plan. This calls our attention to the fact that most
improvement methods must start with a trial plan (formally called a feasible
150 Harris B.
solution), and in the case of non-convexity, many such starting plans might
be used. The selection of these starting points is crucial, since a blind and
random selection may waste resources by testing many starts whose
outcomes are all alike, while a better-informed set of starts, designed by a
knowledgeable planner, may suffer from some bias which eliminates
unusual but possibly strong contenders.
Two points are at issue here. The first relates more to the previous
discussion of science in planning: the use of market clearing models to
approach or achieve equilibrium models in locating many populations,
determining densities, and specifying interaction patterns implicitly makes
many planning decisions, and can provide guiding indicators such as land
prices and congestion for changing them or making new ones. The original
starting plans and their improvement can thus be conducted with a focus on a
smaller number of higher level decisions.
The second issue is the multiplicity of plans based on different starting
points. In the extreme case, where there is good reason to have confidence in
a single start, the process is greatly simplified. The Chicago Area
Transportation Study began with an analysis of "The Optimal Spacing of
Expressways" which guided an initial plan, and some forty additional
modified plans were generated in succession, largely without any of the
backtracking which would indicate difficulties in the improvement process.
A process like this does not generate alternatives for later examination, since
the problem is treated as if it were convex and only a single start were
required.
Such certainty is often encountered among planners, and may from time
to time be subject to severe criticism. In the CATS case, for example, plans
with alternative levels of mass transportation should probably have been
considered. Their omission reflects the bias of the sponsors of the study,
which was transmitted to the staff and which generated some conceptual
errors in the analytic procedures. In general, methods which generate
alternatives, or require their introduction, also generate cognitive dissonance
amongst planners and politicians. Despite the repeated mention of
alternatives in planning theory discussions, the issue of their sources is
largely unexamined, as is the startling absence of them in planning practice.
We feel that the creative exploration of alternatives in planning, in a
fashion which avoids the prevalence of failures and promotes the discovery
of configurations which are very different in structure but very similar in
value - is worthy of much study and encouragement.
We finally turn to a cluster of other optimising methods which show how
planners and many others try to avoid alternatives. We shall try briefly to
exam me how these methods might be used to advantage in a creative
process.
Innovation and Urban Planning 151
none is available, backtracks toward the root of the decision tree until a node
is found with options still available.
We note first that this procedure requires at least one complete plan to
establish a comparative value or cost. This could be the existing plan for a
city similar to the one under study. Later complete but improved plans can
arise successively in the branch and bound procedure, tightening the bounds
at each step. Complete solutions which are later discarded could be
preserved as alternatives, but usually are not. Second, we observe that the
great difficulty with this method is finding ways to estimate the
consequences of incomplete plans. Doing so may be very costly or uncertain
under available estimating methods, while failing to do so leads to an
unconscionable expansion of the number of combinations explored, and the
time and resources expended.
Nevertheless, planners often implicitly follow this procedure in planning.
Its efficiency is improved by placing the decisions with the most
consequences, or the highest net costs, early in the decision process. As
before, all of the decisions should be strategic, with the ultimate aim of
exploring the details through the use of equilibrium models to optimise
subsystems by developing their performance in detail. In ordinary planning,
the decision whether to explore further from a given combination is taken
judgementally, on the basis of intuition rather than complete analysis. This is
the point at which planning procedures can vary most widely when operating
in this mode. Very tight and self-assured methods of judgement will leave
only one option leading out of each node, and the procedure reduces to the
greedy algorithm. Different bounding criteria will lead to different solutions,
even in the greedy case: this is then a new means of generating alternatives.
As we repeatedly notice, strong bases of judgement introduce bias, but weak
ones grossly extend the planning process.
This sketch of the broad relationship between creative but professional
planning and optimising has, I think, demonstrated the importance of further
articulating planning method. Complete optimisation is practically, if not
conceptually, impossible. I have for instance not dealt in detail with the
problem of different interests and value systems, since I feel that this is a
problem not in optimising, but in civil government and conflict resolution. It
affects not the outcomes or consequences of decisions and thus of plans, but
their evaluation and acceptability; accommodating these differences is
another, but subsidiary, reason for preserving alternatives. Given the
practical obstacles to optimisation, we must fall back on a professional but
judgmental process, where the greatest difficulties are presented by bias and
single-mindedness. But there is an additional problem, which is the lack of
imagination and the inability to innovate, which must also be overcome.
Some of the problems in all of these areas will be related to the scope and
Innovation and Urban Planning 153
style of public participation in the next section, and will be approached more
generally in the last section.
In the context of planning, these last two operating functions require the
most creative capacity and lead to the most intensive computing. I like to
think of widespread public planning with citizen participation as a kind of
parallel creativity, with people and organisations providing millions of
parallel processors by analogy with parallel computation. Here again, the
real problem down the road is the programming and co-ordination of the
processors, so as to make their contributions mutually reinforcing and
effective, and not just a jumble of diverse but undirected activity.
In looking toward this future there are many aspects, attitudes, and
practices which need to be reviewed, rethought, and reorganised before the
whole process can be made coherent and effective.
The analytic community needs to make its models more open and
understandable to the users and the public. (I avoid the use of the term
'transparent', which actually means 'invisible when looked through'. The
surface of a model should be transparent, but the interior should not; this
reverses the usual practice, in which the surface hides the working parts.)
Such understanding is usually quite possible at the level of concepts,
provided the terminology is demystified - as in translating 'negative price
elasticity' into 'increasing reluctance to pay higher prices for a given
commodity'. However, not all these terms survive their translation into an
operating model, and the validity of the implementation is a matter of quite
proper concern to both professional and lay members of the using
community. There is some urgency in pursuing the development of a set of
protocols for such validation of models in a non-superficial way - that is, by
content rather than by immediate results.
The analytic process, once agreed upon, needs to be greatly opened and
speeded up, and to some degree simplified. If we imagine a member of the
public or groups of the public and clients at a planning discussion, we ought
to be able also to imagine that a new proposal could be analysed with the use
of a computer model in a few minutes, and surely in less than an hour.
Unfortunately, what tends to happen is that either proposals of alternatives
are analysed by planners off the top of the head, or the process of interaction
Innovation and Urban Planning 157
is drawn out interminably, and interest in it falls toward zero or is turned into
antagonism. Instant expertise is not a characteristic of planning
bureaucracies, and failed negotiations are a not always desired end of the
public process!
In fact, the availability of rapid means of analysis could be beneficial to
the planning process even within agencies. Currently, the tedium of
elaborating and evaluating alternatives leads to an unseemly rush to
judgement, and to the premature rejection of novel ideas. An imaginative
and agile planning process, with active staff participation, requires adequate
and neutral means of exploring new suggestions, rather than authoritarian or
uninformed ends to discussion.
We have already mentioned the need for public access to data, models,
and computing facilities. With the development of processes of the kind
envisaged, it is apparent that public advocacy will often require expert
technical support. Such support has already been used in civil suits against
certain practices in transportation planning in the US. Over the long haul,
such analytic support for public advocacy will become a fixture of the
profession, will be established internally in large public organisations, and
will become an acquired skill among some members of these organisations.
It is hardly necessary to add that major accommodations will be required
by both the analytic and the planning professions - within their practices and
home agencies, in their professional contacts with each other, and in their
relations with the public. More respect will have to be offered to others in
place of professional assurances and disciplinary hauteur. And new activities
will have to be invented and undertaken, such as the public explanation and
justification of modelling, the preservation of planning alternatives, and the
activities of public education and advocacy in general.
It may by now be apparent that there are many different types and levels
of innovation in planning - in the analytic process, in the development of
plans, and in securing the participation, contribution and trust of the public.
In our concluding section, therefore, we will take a very brief look at the
attitudes and processes which tend to support these changes.
some of the nature of invention. We will, rather, look at the components and
sources of invention.
It is sometimes said of invention, and especially in well-trodden fields
like urbanism, that there is 'nothing new under the sun'. We will avoid
debate by conceding that in general there are no new things, but that there
are new combinations of old things. (A safety-pin is no more than a bent
piece of wire.) Finding new combinations, subject to constraints, is not easy;
as we have seen above, a major aspect of design is its combinatorial nature,
and th~s design is a part, or form, of invention. Science and technology
constitute another part, and without analysing them we will note that they
may impinge on urban innovation from outside of planning, but not without
the invention of new modes of organisation and action in the city, without
which they would not become integrated into its life.
In principle, we could invent new combinations of enormous variety, but
the constraints on innovation tend to curb the exuberance of such
inventiveness. These constraints are of the general nature of workability,
usefulness, affordability, a benign nature, moral acceptability, and beauty (or
more aptly and practically, non-ugliness). Some constraints, like legality,
may be susceptible to change outside the process of planning and design.
These overlapping and sometimes conflicting constraints are indicative only;
the process of planning takes them, and others, into account directly and
indirectly. Insofar as planning and design are purposeful (and this is most
often the case - but that is a different paper!), they accord with the maxim
that necessity is the mother of invention. We may be looking for the father.
In complex systems like cities, the problem does not invent the solution,
any more than in medicine the disease invents the cure, or even in difficult
cases, the diagnosis. The necessity which is thought to provoke invention at
the most makes only a suggestion, and most often this is - at least in its
simplest form - wrong. This is because all inventions and innovations have
indirect effects which may defeat the original purpose. This is particularly
true in the reverse case, where invention is the mother of necessity - that is,
where something is desired and adopted because it is new and fulfils a need
only just now revealed, and with effects only partially explored. This
imperfection of untried novelty is a principal source of conservatism in daily
life, and of the distrust of major innovations and technological fixes.
Borrowing often takes the place of invention, in whole or in part. Cities
can borrow technology from other fields, and can do so creatively and
imaginatively. Borrowing from the experience of other nations and other
cities is particularly apt, but is subject to the need for adaptation to variations
in history, resources, terrain, climate, culture, and so on. This kind of
borrowing can be especially useful because only a few cities are at the
cutting edge of urban development, and hence most other cities can still
Innovation and Urban Planning 159
learn from them. This option is closed if development is too rapid, or if the
leading cities are taken as an object lesson in an undesired future.
We have already seen how analysis and design can borrow extensively
from the social sciences, and from the 'managerial sciences' such as
operations research and computing. Here the transfer of knowledge and
practice is two-edged. Some of these disciplines are privileged in the realm
of invention because they are a step removed from practice and not
constrained in the ways we have discussed. Their inventions may therefore
be more advanced than others which are so to speak home-cooked, but they
may not be suitable for innovation in urban analysis, planning, or function.
Finally, invention which is not purely inspirational (and such invention
may not exist) probably depends most extensively on an extended form of
borrowing - namely metaphor. A more pedestrian form of metaphor is
analogy, and an even more pedestrian form of analogy is simulation. All
three of these might be thought of as models, either of the urban system or of
the planning process, but there is a difference of major importance. Models
are judged on the basis of their realism or accuracy, while metaphors are
judged by their suggestive power; that is, metaphors have enough realism to
sustain a connection, and enough novelty to suggest that our view of reality
might be improved.
The most powerful metaphors are those which inject the most novelty
into our views of a phenomenon, without losing a sense of realistic fitness.
The best inventors probably have the widest knowledge of distant things of
all kinds, and a distinctive ability to make the connections, see the relevance,
and resolve the dissonance by redefining the nearer objective of the
metaphor. Metaphors have to do with bringing together disparate forms, or
discovering hidden but congruent forms, in the widest sense of the word
'form'. There are underlying metaphors in writing, painting, architectural
design, and scientific and technological innovation. Often, the borrowing by
one field from another is largely metaphorical.
An important source of metaphor in any activity which involves
systematic patterns, or their denial, is mathematics. 'Pure' mathematics is a
cornucopia of systematic expressions of form; many of these are not
invented for any practical purpose, and may lie unused for many decades.
But most of these forms ultimately find some application in the real world,
starting as a metaphor and evolving into a theory which has to be tested.
Conversely, many real world findings stimulate mathematical invention,
because they are uncomfortable without a metaphor supporting their
systematic understanding.
Dealing in urban form, and in the formal statement of urban relations,
metaphors from mathematics must be treated with both enthusiasm and
caution. They have an intrinsic regularity and beauty which supports their
160 Harris B.
potential power, but they may not fit the relational purposes of a metaphor.
Since no systematic representation of complex system can be mounted
without computer support and mathematical statements of its inner
relationships, this dilemma is important to planning analysts and planners in
everything they undertake.
This gallop through the nature of creative analysis and design has left us
with a daunting sketch of the knowledge and skills which must be brought to
these activities. Let us conclude by listing them:
This list is indicative and not exhaustive. Its scope is beyond the
complete grasp of most if not all individuals, and compels us to recognise
once again that planning and analysis are collaborative activities - as well as
admitting above all that the public belongs in that collaboration as a full
partner.
REFERENCES
Alonso, W. (1964) Location and Land Use. Harvard University Press, Cambridge Mass
Anas, A. (1975) The Empirical Calibration and Testing ofa Simulation Model of Residential
Location. Environment and Planning A 7: 899-920
Anas, A. (1987) Modeling in Urban and Regional Economics. Harwood Academic
Publishers, Chur, Switzerland
Innovation and Urban Planning 161
Carroll, J.D. (1955) Defining Urban Trade Areas. Traffic Quarterly 9: 149-61
CATS (1959) Final Report Volume 1. Chicago Area Transportation Study, Chicago III
Cook, S.A. (1971) The Complexity of Theorem Proving Procedures. Proceedings of3rd ACM
Symposium on the Theory of Computing, ACM, 151-58
Echenique, M., Owers, J., eds. (1994) Research into Practice: The Work of the Martin Center
in Urban and Regional Modeling. Environment and Planning B (Special Issue) 21:5
Goldner, W. (1971) The Lowry Model Heritage. Journal of the American Institute of
Planners 37: 100-10
Hamburg, J.R., Creighton, J.R. (1959) Predicting Chicago's Land Use Pattern. Journal of the
American Institute of Planners 25: 67-72
Hansen, W.G. (1959) How Accessibility Shapes Land Use. Journal of the American Institute
ofPlanners 25: 73-6
Harris, B. (1963) Linear Programming and the Projection of Land Uses. Penn-Jersey Paper
No. 20, Philadelphia
Harris, B. (1964) A Model of Locational Equilibrium for Retail Trade. Institute for Urban
Studies (mimeo), University of Pennsylvania, Philadelphia PA
Harris, B., Wilson, A.G. (1979) Equilibrium Values and Dynamics of Attractiveness Terms in
a Production-Constrained Spatial-Interaction Model. Environment and Planning A, 10:
371-88
Herbert, J. S., Stevens, B. H. (1960) A Model of the Distribution of Residential Activity in
Urban Areas. Journal of Regional Science 2: 21-36
Hill, D.M. (1965) A Growth Model for the Boston Region. Journal of the American Institute
ofPlanners 31: 111-19
Huff, D.L. (1964) Defining and Estimating a Trade Area. Journal ofMarketing 28: 4-38
Ingram, G.K., Kain, J.F., Ginn, J.R. (1972) The Detroit Prototype of the NBER Urban
Simulation Model, National Bureau of Economic Research, New York
Koopmans, T.C., Beckmann, M.J. (1957) Assignment Problems in the Location of Economic
Activities. Econometrica 25: 53-76
Lakshmanan, T.R., Hansen, W.G. (1965) A Retail Market Potential Model. Journal of the
American Institute of Planners, 31: 134-43
Lee, D.B. (1973) Requiem for Large-Scale Models. Journal of the American Institute of
Planners 39: 163-78
Lowry, I.S. (1964) A Model of a Metropolis, RM-4035-RC, The Rand Corporation, Santa
MonicaCA
McFadden, D. (1973) "Conditional Logit Analysis and Quantitative Choice Behavior." In
Frontiers in Econometrics, Zarembka, P., ed, Academic Press, New York, 105-42
Mills, E.S. (1972) Urban Economics, Scott Foresman, Glenview III
Mitchell, R.B., Rapkin, C. (1954) Urban Traffic: a Function of Land Use. Columbia
University Press, New York
Moore, E.F. (1957) The Shortest Path Through a Maze. International Symposium on the
Theory of Switching, Harvard University, Cambridge Mass
Putman, S.H. (1976) The Interrelationships of Transportation Development and Land
Development. U.S. Department of Transportation, Federal Highway Administration,
Washington DC
Putman, S.H. (1982) Integrated Urban Models: Policy Analysis of Transportation and Land
Use. Pion, London
Stouffer, S.H. (1940) Intervening Opportunities: a Theory Relating Mobility and Distance.
American Sociological Review 5: 845-57
Wheaton, W.C. (1974) Linear Programming and Locational Equilibrium: The
Herbert-Stevens Model Revisited. Journal of Urban Economics 1: 278-87
162 Harris B.
Wilson, A.G. (1970) Entropy in Urban and Regional Modeling. Pion, London
Wingo, L. Jr. (1961) Transportation and Urban Land Use. The Johns Hopkins Press,
Baltimore MD
PART II:
INNOVATION AND ITS SPATIAL IMPACTS
Section C:
8.1 INTRODUCTION
across different types of urban region, and (ii) how these novelties are
diffused and relocated in a system of urban regions. The study also analyses
how an urban region may remain an initiator or early follower, or may lose
its leading role over time.
Three alternative approaches to the explanation of hierarchical
specialisation dynamics are outlined and compared. First, a model of
location dynamics is formulated on the basis of the theory of central place
systems (CPS-model). Second, specialisation theories are examined with a
focus on dynamic location advantages (DLA-model). Third, a type of
product cycle model is introduced with special emphasis on product vintages
(PV -model). Each of these three theoretical models helps to explain specific
aspects of observed location dynamics. The analysis shows under what
conditions predictions from the CPS-model become inconsistent with
predictions from the DLA-model and PV-model.
resources will be harvested (Moroney and Walker 1966; Smith 1975). For
production whose location is not based on proximity to natural resources, the
DLA-model can easily be transformed into the PV class of models. These
models assume that a high proportion of new products are initiated or
imitated (at an early stage) in a leading region. As production expands, the
products become standardised and techniques are routinised, then relocation
begins to takes place. However, the number of followers is expected to be
limited, since growing economies of scale prevent full decentralisation to
other regions. According to the PV-theory, the location is assumed to be
supply-based also at this follower stage of a product cycle Vernon 1960,
1966; Andersson and Johansson 1984).
The dynamic version of the CPS-model is based on a similar logic to that
of the PV-model, with standardisation and routinisation being the driving
forces. However, in the CPS-theory, density of demand is the governing
principle (Beckmann 1958; Camagni, Diappi and Leonardi 1986). Hence, as
production is started at lower levels in the hierarchy, all regions with a
sufficient demand density are likely to become hosts of the activities which
'filter down'. In the long term, all existing activities may be challenged and
replaced by a new substitute. The decline of the obsolete activity is also
assumed to initiate in the leading region. In this sense, the CPS and PV-
models have the same overall description of long-run substitution.
As a consequence of the assumptions for the dynamic CPS and PV-
models outlined above, a set of sub-hypotheses can be formulated. For
example, activities with both a high ,u-value and fast growth are assumed to
be non-routinised and refer to non-standardised products. They are also
assumed to be R&D and knowledge intensive. Activities with a low ,u-value
are assumed to be routinised and price competing, with a strong tendency to
reduce the labour input coefficient.
options arise. Both the CPS-model and PV-model make similar predictions
about where new activities should be expected to appear.
In this section a description is also given of the differences between the
models. Basically, the CPS-model refers to activities which gradually
become more standardised and routinised by reducing average input costs as
well as the set-up costs. The PV-model refers to activities and products for
which routinisation means that knowledge-oriented input coefficients
decrease much faster than other input coefficients.
Section 8.3 shifts the perspective from interregional conditions to
processes and structural features within each urban region. By focusing on
the economic milieu and the associated location attributes of a region, an
analysis is made of the specialisation pattern. In subsection 8.3.2 a model of
sectoral change is introduced to clarify under what conditions the region can
succeed in renewing its economy and under what conditions its activity
pattern ages and leads to decline. There is also a discussion about the
dynamics of a region's location attributes.
In Section 8.4 the leader-follower assumption is introduced and applied
to an empirical analysis of changes in a system of urban regions in Sweden
and Norway consisting of municipalities and groups of municipalities. The
empirical investigation focuses on the spatial concentration of declining
product vintages. In Section 8.5 the perspective is reversed. Here the
knowledge intensity of expanding product groups is examined. In particular,
the location attributes are characterised with regard to initiating regions, as
well as early and late followers. In Section 8.6 a set of conclusions and
conjectures is presented about the leader-follower model in a Europe-wide
context.
(8.1)
where d k+1 signifies the distance from a city of rank k+ 1 to its demand-
associated cities of rank k, A the friction parameter referring to deliveries
from supply sites of rank k+ 1 to markets of rank k, and where A may have a
specific value for each commodity.
Given the urban system described above, a feasible location for a market-
oriented firm (which supplies a given type of commodity) depends on the
cost function which applies at a given point in time for the production of this
commodity. To describe this, let F denote the fixed costs, including the non-
variable part of development costs. Consider also the variable costs which
may here be approximated by a proportional cost component cQ, where c is
a fixed coefficient and Q denotes output or supply. This yields:
C=F+cQ (8.2)
Given this cost function a firm supplying the commodity in question can
locate in a central place k, if it can break even there. For a given price p this
is equivalent to the following condition:
170 Forslund U.M. and Johansson B.
Next, let p* denote the price which maximises the aggregate profit
function:
(8.4)
which describes the maximum profit a single supplier can obtain. With n
suppliers, the function changes to trk (p) = (p - c)f(p)Mk / n - F. Given
the profit condition in (8.4) one can specify a set of industries or supply
activities which locate in a central place of rank k. This set consists of all
sectors (commodity groups) j such that:
This type of set will form a hierarchy which reflects the city-size
hierarchy.
We shall study the dynamics of this hierarchy by introducing a technical
change process. The change of production techniques in an industry is
described by the technique vintage index r, which is close to 0 for a new
process and approaches 1 as the technique matures. Because of this, the cost
function in (8.2) has to be respecified to read:
, = F, / x, + c,
(8.5)
c, = p,a + m,b
where for a certain product, , denotes total cost per unit output, c, the
associated variable cost, F, the fixed cost which may vary between locations,
x, the production or supply of the product in region r. Moreover, (8.5)
specifies two input coefficients where a signifies the necessary input of
routine resources and b the necessary input of development and knowledge
activities. The two price variables p, and m, are location specific.
Let us first consider a case where the product is standardised and
produced with a routinised technique. Region r will then have an advantage
if routine resources, including labour with standard skills, are available at
low prices, i.e. when p, is low. Moreover, the supply of land and suitable
infrastructure may be such that F, is smaller in region r than in alternative
locations.
We consider next a case where the product is non-standardised and still
in a development phase. It may be a new product or a product for which each
delivery is customised. Then F, is assumed to decrease as a region acquires
higher customer, supplier and import intensity. In particular, the average
price of knowledge resources is assumed to be lower in regions with a high
R&D and knowledge intensity as well as accessibility to specialised,
development-oriented suppliers. In order to remain development-oriented, a
region must be able to keep a balance between the diversity of the economic
174 Forslund U.M. and Johansson B.
milieu and its specialisation on few activities. The European urban regions
which in the period 1960-1980 were specialised in shipbuilding, all seem to
have lost the quality of balance in their economic milieu during this period,
although some may have found new specialisations later. Another strong
impression is that automobile specialisation, as observed in Wolfsburg,
Goteborg or Turin, tends to bring about a similar impoverishment of the
economic milieu.
The above examples may serve to illustrate location advantages which
are based on the economic milieu and which may add to or subtract from the
location properties which follow from the size of an urban region. In
particular, the economic milieu is influenced by the composition of activities
in the region. Hence, we should expect the dynamics of location advantages
to be non-linear, with both increasing returns to scale and increasing returns
to scope. Because of this, a region with a lower rank sometimes manages to
develop into a system-wide supplier in a specialised field.
Let a product be identified by its product group i and its vintage index
0~~~1, where a low B-value indicates either a young product or an early
vintage, and a high B-value indicates an late vintage. The former refers to
non-standardised products which require large shares of knowledge or
development resources in the production process. High T-values therefore
refer to routinised (and usually automated) production and interaction
activities, whereas low T-values represent non-routinised activities. Late
product vintages are assumed to use smaller shares of knowledge resources.
With these input categories the cost function in (8.2) can be extended as in
(8.6), which is an alternative specification of the cost function in (8.5). When
focusing on a particular product group, we can suppress the associated
index, i, and consider the following cost function which refers to product
e,
vintage which is based on technique T, and which is applied in a rank-k
region r = r(k):
where ,( T,B) denotes total cost per unit output, C,( T,B) is the associated
variable cost, F,(T,B) refers to fixed cost associated with technique T, a(T,B)
represents the input coefficient (delivered output per unit) referring to
routine input resources, and b( T,B) represents the input coefficient referring
to development resources or, in other words, input resources for knowledge
activities or knowledge production. The input coefficients are assumed to be
Product Vintages and Specialisation Dynamics 175
determined by the selected technique and are not dependent on the location.
On the other hand, the two price variables, Pr and OJr, are location specific,
where Pn denotes the price of routine inputs and OJ,. denotes the price of
development, knowledge-intensive resources.
In this subsection we study how the knowledge intensity develops as the
technique-vintage and product-vintage indexes increase. For these analyses
we construct the variable f3( .,8) = b( .,8) f a( .,8) or simply f3 = bfa, which is
called the knowledge-routine ratio.
In order to bring out the essential aspect of ageing product vintages, we
that.=
assume that the process of technical change is driven by the change in () so
1(8). Following the product cycle model in Johansson and
Andersson (1996), we assume that the routinisation of production increases
as the product becomes more standardised, i.e. as the ()..value gets larger. We
consider two stages of the product cycle development represented by two
pairs of input coefficients (b O, an) and (b I , a I ), with b g =b('f(Og and
d =a( 'f(~, where g = 0,1. Let the initial technique (b O, an) refer to a
young product and the second to a more mature one so that:
(8.7)
In Remark 2 we consider two regions, where the first has rank k and the
second rank h, and k> h. We examine the variable cost, cqg =p q a g + OJ q b g ,
for q = {k, h} and g E {O,l}. The region with the higher rank, k, is assumed to
have lower prices (costs) of knowledge resources and higher prices of
routine resources - in comparison with the level h region. The following
remark is self-evident from the definitions given in (8.6)-(8.7).
Remark 2: Let k > h, Pk > Ph and OJk < OJ h Consider a particular product
which is initially located in a region with rank k with < c2 c2 .
Moreover,
assume that the technique develops as specified in (8.7). Then a shift in
variable cost advantage occurs iff the knowledge-routine ratio reduces in
such a way that:
The described variable cost advantage of rank h regions does not imply
that such regions offer a feasible location. The size of demand must also be
large enough at this level of the urban hierarchy. In order to be a feasible
location, a rank h region, r = r(h), must satisfy the condition.
1r; =(p - c; )f(p) - FfM r > O. This observation leads to Remark 3.
176 Forslund U.M. and Johansson B.
Remark 3: Consider the assumptions in Remark 2 and assume also that the
price shifts to pi < pO when the knowledge-routine ratio shifts from pO to
pl. A rank h region becomes a feasible location if (pI - c! )f(pl) ~ F I M h
Such a change in technique contradicts the rank-size order of the CPS-model
ifinaddition (pl-c!)f(pl)<FIM k fork>h.
The described contradiction may well to occur. It implies that the
commodity is delivered from a region at a lower level to regions at higher
levels in the hierarchy. Observe that the effect of the new technique is (i)
ck >c! and (ii)i</.
The contradiction occurs if ;r! = (pI - c! )f(pl) - F I M k < O. When
such a contradiction occurs it becomes necessary to shift model framework.
In particular, the distribution system cannot be restricted to the hierarchy
given by the CPS-model, we have to consider the entire urban system U.
Because of this, we also have to refer to a network of deliveries from region
r to all other regions, where drs denotes the distance between rand s in this
network. Reformulating (8.1 ???) so that it refers to the introduced network
system, the potential demand for a particular product delivered from region r
can be defined as follows:
With this formulation there may be a large set of regions at rank k which
are feasible locations for the technique indexed by 1. In order to obtain
predictions about likely location patterns, a probability distribution therefore
has to be superimposed.
Product Vintages and Specialisation Dynamics 177
The previous section examined how the economic milieu in each region
affects the interregional location dynamics. The ambition in the present
section is to show how the economic milieu itself is influenced by the
location process. In this way, Section 8.3 investigates the same process as in
Section 8.2, but with a specification of zones within each individual region.
In this subsection, the five aspects of a region's economic milieu (from
Section 8.2.2) are considered: (i) infrastructure, (ii) knowledge and R&D
intensity, (iii) customer intensity, (iv) supplier intensity and (v) import
intensity. The infrastructure of the urban region as a whole as well as of
individual zones may facilitate the improvement of (ii)-(v). However, the
critical factor is the dynamic pattern of the location of activities within the
region. If the capacity of the built environment is taken up by routinised
activities related to late product vintages, then there is limited space for new
activities. Moreover, in such a situation the customer and supplier intensity
will be oriented towards late product vintages and not early ones. Over time,
such a situation will make the economic milieu gradually less suitable for
innovation activities.
Let us consider the zones in an urban region. The capacity and quality of
a zone is denoted by Z and the value added of the activities in the zone is
denoted by Y. A frequent problem is that the activities which have in the past
had a prosperous period tend to remain in the same zone even when the
value added or profit they can generate continue to fall (due to routinised
techniques and standardised products). In order to assess this type of
dynamic problem the following simple model is introduced:
Y(O)=G(Z,O) (8.10)
where Cq(Z) signifies the overall production cost per unit value added in a
zone of type q, which is endowed with the infrastructure level Z. Given this
specification, a set of conclusions are specified below. They are based on the
vintage theory assumptions introduced in Section 8.2, together with the
assumptions introduced above in this subsection. They also refer to the
structures depicted in Fig. 8.1. We should note the following:
(i) the B-zone has by assumption an advantage in production with low ()-
values, in the sense that GB(ZB,OBGB(ZB,OA), which implies that
1C B(OB) > 1C B(OA);
(ii) the B-zone is by assumption superior to the A-zone for early product
vintages (low ()) in the sense that GB(ZB,(}BGA(ZA,(}B). If
CB(ZB,OB) is not too high relative to CA(ZA,OB), this yields
1C B(OB1C A(OB) ;
(iii) for certain routine activities one may expect that the A-zone is superior
to the B-zone, in the sense that GA(ZA ,OA) > GB(ZA ,OA). However,
irrespective of this productivity condition, the cost level in the A-zone
means that 1C A(0 A) > 1C B(0 A) ;
(iv) by assumption, the A-zone has an advantage as a site for routine
activities, in the sense that G A (Z A, 0 A ) > G A(Z A ,0 B), which yields
1fA((}A) >1fA((}B).
Product Vintages and Specialisation Dynamics 179
In Fig. 8.1, z! is optimal for low O-values and zA is optimal for high 0-
values. Hence, for non-routine activities and young products with a low B-
value, the B-zone is superior to the A-zone. The reversed situation obtains
for routine activities and old products with a high B-value. In particular, one
can observe from the figure:
8-zone with a G
A-zone with a
lowe lowe
y y
z z
y A-zone with a
B-zone with
high e c y high e
J.-_--G
G
z z
Figure 8.1 Value added and cost in A -zones and B-zones for high
and low B-values
This implies that when the B-value increases for a given production,
profits will fall and become negative if it remains located in a B-zone. A
relocation to an A-zone is then a rewarding alternative. Another option is to
relocate these activities to an A-zone outside the urban region. The major
lesson is that exit of routinised activities from B-zones is necessary in order
to provide new space for the introduction of new and development-oriented
activities (with low B-value).
180 Forslund U.M. and Johansson B.
This subsection continues the approach in subsection 8.3.1, but now all
A -resources and all B-resources in an urban region are aggregated.
Moreover, all activities with a low B-values are grouped together and called
J-activities. In the same way, all activities with a high B-value are grouped
together under the label I-activities. Hence, this subsection transforms the
micro approach of Section 8.2 by analysing broad categories of products
which are identified as clusters or groups of products. We assume that
individual but associated product cycles can be expected to develop in
parallel clusters, with a base in the same urban region or in a small set of
interconnected regions. When such a specialisation pattern obtains, we shall
use the term product group specialisation of a region. Such a group may
then be characterised by its average vintage index.
Let I denote routine-oriented, standardised activities, and let J denote
non-routinised, knowledge-oriented activities. The output from activities of
type I is called a type I product, and the output from the second type of
activities a type J product. Moreover, let A and B represent the routine-
oriented resources and the knowledge resources, respectively, in a given
region and let these resources comprise both market resources and other
parts of the economic milieu. We also need two pairs of technical
Product Vintages and Specialisation Dynamics 181
QijAx i shows how the A-resources available for activity} shrink and
aul1x i shows how the A-resources available for activity i shrink as the
activity level Xi increases by Lhj; i = I, J and} = I, J (8.l3)
byLlx, shows how the B-resources available for activity} shrink and
(8.15)
where the functions f(x[) and f(x J ) specify the speed of adjustment, given
the size of each type of activity at each point in time. These adjustment
factors will express self-reinforcing adjustments if df / dx > O. In many
modelsfisexpressedas f(x[)=y[x[ and f(xJ)=YJxJ' where Y[ andYJ
are two positive constants (e.g. Fisher and Pry 1971; Batten and Johansson
1989).
The model in (8.15) is designed to be compatible with product cycle
assumptions as expressed in Remarks 1-3. Hence, the condition
b ll / all > b l / ai' must be satisfied, which implies that the knowledge-
routine ratio is higher in non-routine than in routine activities. Moreover, if
equal activities disturb each other less than unequal activities, one obtains
aij ~aii and bij ~bii.
The change process in (8.15) is characterised by competitive exclusion,
and the time profile of both growth and decline has a sigmoid form. It is also
evident from the equations in (8.15) that if x I is not declining, or if the
process is very slow, then the entry of non-routine activities will be delayed
or completely obstructed.
Thus, given that the resources A and B remain unchanged over time, the
output from type J activities can, with sufficient demand, continue to grow if
activities of type I continue to reduce their input consumption of the same
resources. The latter is in this case equivalent with a gradually declining
output of type 1. It should be observed that if such a process is constrained
by the available supply of B-resources, it will generate an excess supply of
A-resources. This follows from the specification of the knowledge-routine
ratios above. Hence, when studying long time periods, expansion and
contraction of the two categories of resources must be contemplated. This
Product Vintages and Specialisation Dynamics 183
,
ECONOMIC MILIEU OF THE
Evolution of:
(1) A-resources
and
..- URBAN REGION
(1) Expansion or decline of
J-activities
(2) B-resources
(2) Transformation of J-activities
to I-activities
(3) Expansion or decline of
I-activities
,
Exit of 'old' activities
assumed to occur much more often in the leader regions than elsewhere.
Moreover, they serve as examples to other regions and demonstrate that the
new activities are both feasible and profitable. According to the theory
introduced earlier, novel activities and still-developing, non-standardised
product vintages are frequently initiated and promoted in leader regions. In
order to leave space for new activities, standardised and routinised activities
must gradually be removed from the leading region.
The approach described above was applied to the set of municipalities in
Sweden and Norway. In many cases, the functional region was composed of
three or more municipalities. The results presented below have been
aggregated into the following categories: (i) the set of cities in the Malardal
region (including Stockholm), (ii) Oslo-Akershus (Norway's leading region),
(iii) all Swedish regions outside the Malardal region, (iv) all Norwegian
regions including Oslo-Akershus. The result of the empirical analysis is that
the Malardal region (M-region) has a leading role in the Sweden-Norway
context. (In this study, the M-region comprises the county of Stockholm and
the municipalities of Uppsala, Enkoping, Habo, Vasteras, Koping, Arboga,
Eskilstuna and Strangnas.)
The first step was to measure for each municipality the relative share of
the employment in each activity. Next, the average share of each sector was
calculated for the set of all Swedish regions. For sector j, this share is
denoted by S j The corresponding value for region k is denoted by S jk By
comparing such shares in the M-region, S jM , with the corresponding share
in Sweden as a whole, one may classify sectors into sets with high, medium
and low concentration in the M-region. Let P j = S jM / S j and define:
The class limits between high, medium and low concentrations, as well
as between fast and medium growth, have been selected to generate a sharp
division between the corresponding clusters. In the empirical analysis,
sectors are cross-classified according to their concentration in the M-region
186 Forslund U.M. and Johansson B.
and their growth rate in the rest of Sweden (all municipalities outside the M-
region). In a similar way, the M-region concentration is classified against
sectoral growth in the Oslo-Akershus region, and against Norway as a
whole.
Sectors with a high ,u-value will have a fast growth in the rest of Sweden.
These sectors will have a high concentration also in the Oslo-Akershus
region, since it is a leading region in Norway. However, the concentration
value for most of these sectors is higher in the M-region than in the Oslo-
Akershus region. This follows from the assumption that the M-region is a
leading region for both Sweden and Norway.
Sectors with a low ,u-value are expected decline in the rest of Sweden, in
Norway, and in Oslo-Akershus.
Medium (~O) H+ M+ L+
Negative H- M- L-
Sources: Forslund (1996); Forslund and Johansson (1995); National Bureau of Statistics in Sweden and
Norway.
Table 8.3 reveals that during the 1980s Norway had a larger number of
sectors with declining employment than Sweden. However, the table still
supports the hypothesis that a low concentration in the M-region predicts
decline among Norwegian regions. Moreover, of the sectors with a positive
growth in Norway, 79 percent have a high ,u-value and only 21 percent a low
,u-value. In summary, we find that 72 percent of all sectors in Norway are
placed on the diagonal of the matrix. Standard i-measures for Tables 8.2
and 8.3 show that we cannot reject the hypotheses stated above. Also, when
the growth pattern in the Oslo-Akershus regions is examined in the same
way as in Table 8.3, it proves to be a follower of the M-region.
Number of sectors in 4 9 31 44
the rest of Sweden
Number of sectors in 5 \3 38 56
Norway
Product Vintages and Specialisation Dynamics 189
Table 8.5 Average education shares (as %) in 1992 in Sweden for sectors
with a low fl value
Are the university shares reported for 1992 in Table 8.5 particularly low?
Table 8.6 provides an answer. It shows that sectors with a high concentration
in the M-region and fast expansion in the rest of the country had a much
higher university share in 1992 than all sectors in Table 8.5. Moreover, even
for sectors with a medium concentration in the M-region, the university
share is almost three times higher than in sectors with a low concentration.
All this suggests that the knowledge intensity is relatively low in sectors
with a low ,u-value.
(H++) Sectors with high (H+) Sectors with high fl- (M++) Sectors with medium
fl-value and fast growth value and medium growth fl-value and fast growth
s/(l-s) t
I
Stockholm
It can be seen that all classes of sectors with a high ,u-value have twice as
large a share of employees with pre-university education than sectors with a
low ,u-value. They have around 2-3 times the share of employees with some
form of university education. Sectors with a high ,u-value have as much as
10-15 times the share of employees with post graduate studies than sectors
with a low ,u-value. Equation (8.18) describes a regression of a sector's
concentration (,u-value) and its share of employees with some kind of
university education, denoted by u. The association between fl and u is quite
strong, and indicated by a high t-value in brackets (8.18).
important second stage in the empirical analysis presented here. Such a step
forward could also be combined with a specification of attributes of the
economic milieu.
Another extension of the empirical investigation presented would be to
include a larger set of European countries. Since the data requirements are
fairly modest (employment figures with a fine sector specification), such an
extension is quite feasible. An interesting challenge would be to identify a
set of leading regions in Europe, examining to what extent they lead in
different fields, and investigating the relationship between dynamic
specialisation and regional distribution of economic milieu attributes. This
effort could also shed new light on existing pictures of the economic
geography of Europe. Would the leading region be located in the famous
'corridor' from the London region, curving down through Randstad,
following the Rhein, passing the Swiss domain and ending in the Milan
region? Other major central places may include Paris, Munich, Berlin and
Brussels. Secondary leading regions will include Rome, Florence, Vienna,
Helsinki, Stockholm, Oslo-Akerhus, Copenhagen, several British
metropolitan regions and still others. Moreover, could one apply the same
model to describe the technology diffusion and the location dynamics for
lower levels in the urban hierarchy of Europe?
Finally, it should be stressed that the PV-model has both an interregional
and an intra-regional message. The interregional aspect comprises the
relations between technical change, product development and location
dynamics in a multi-regional setting, where regions are characterised by
their economic milieu. But the intra-regional aspect is equally important. In
this case, the economic milieu is influenced by composition of economic
activities in the region, and hence the entry and exit of activities is part of the
ongoing formation of a region's economic milieu. If obsolete activities exit at
slow pace, the milieu characteristics deteriorate and there is not enough
space for new activities to enter. Moreover, a significant policy aspect is that
the economic milieu can evolve as the result of development and design of a
region's infrastructure.
REFERENCES
Beckmann, M.J. (1958) City Hierarchies and the Distribution of City Sizes. Economic
Development and Cultural Change, VI(3): 243-48
Beckmann, M.J., McPherson, l (1970) City Size Distribution in a Central Place Hierarchy:
An Alternative Approach. Journal ofRegional Science 10: 25-33
Beckmann, M.J. (1996) The Location of Market Oriented Industries in a Growing Economy,
Paper presented at the 5th World Congress of the RSAI in Tokyo, May
Bos, H.C. (1965) Spatial Dispersion of Economic Activity. Rotterdam
Camagni, R., Diappi, L., Leonardi, G. (1986) Urban Growth and Decline in a Hierarchical
System - A Supply-Oriented Dynamic Approach, Regional Science and Urban Economics
16: 145-60
Cheshire, P.C., Gordon, I., eds (1995a) Territorial Competition in an Integrating Europe.
Avebury, Aldershot
Cheshire, P.C., Gordon, I. (1995b) "Change and Spatial Specialisation within the South East's
Economy." In Territorial Competition in an Integrating Europe, Cheshire, P., Gordon, I.,
eds, A vebury, Aldershot, 65-97
Cheshire, P.e., Hay, D.G. (1989) Urban Problems in Western Europe. An Economic Analysis,
Unwin Hyman, London
Ciccone, A., Hall, R.E. (1996) Productivity and the Density of Economic Activity, American
Economic Review 86: 54-70
Erickson, R.A, (1976) The Filtering-Down Process: Industrial Location in a Non-
Metropolitan Area. Professional Geographer, XXVIII
Fisher, lC., Pry, R.H. (1971) A Simple Substitution Model of Technological Change.
Technological Forecasting and Social Change 3: 75-88
Forslund, U.M. (1996) Industrial Location - Interregional Leads and Lags in Sweden and
Norway, Working Paper No 1, Swedish Institute for Regional Research, Ostersund
Forslund, U.M., Johansson, B. (1995) "The Mlilardalen: A Leading Region in Scandinavia
and Europe?" In Territorial Competition in an Integrating Europe, Cheshire, P.C.,
Gordon, I., eds, Gower, Aldershot, 3-27
Johansson, 8., ed (1996a) Dynamics in Metropolitan Processes and Policies, RR-86-8, IIASA,
Laxenburg, Austria
Johansson, B, ed (1996b) Spatial Dynamics and Metropolitan Change, RR-86-9, IIASA,
Laxenburg, Austria
Johansson, 8. (1987) "Technological Vintages and Substitution Processes." In Economic
Evolution and Structural Adjustment Batten, D.F., Casti, J., Johansson, 8., eds, Springer-
Verlag, Berlin, 145-65
Johansson, 8. (1993) "Economic Evolution and Urban Infrastructure Dynamics." In The
Cosmo-Creative SOCiety - Logistical Networks in a Dynamic Economy, Andersson, A.E.,
Batten, D.F., Kobayashi, K., Yoshikawa, K., eds, Springer-Verlag, Berlin, 151-75
Johansson, B., Karlsson, C. (1987) "Processes ofIndustrial Change: Scale, Location and Type
of Job." In Urban and Regional Labour Markets, Nijkamp. P., Fischer, M.M., eds, North-
Holland, Amsterdam, d 133-79
Johansson, B., Nijkamp, P. (1987) "An Analysis of Episodes in Urban Event Histories." In
Spatial Cycles. van den Berg, L., Klaassen, L.H., eds, Gower, Aldershot, 43-66
Johansson, B., Andersson, A.E. (1996) A Schloss Laxenburg Model of Product Cycle
Dynamics, mimeo, JOnkoping International Business School
Johansson, 8., Wigren, R. (1996) "Production Milieu and Competitive Advantages." In
Infrastructure and the Complexity of Economic Development, Batten, D.F., Karlsson, C.,
eds, Springer-Verlag, Berlin, 187-212
Marshall, A. (1920) Principles of Economics. Macmillan, London
196 Forslund U.M. and Johansson B.
Moroney, J.R., Walker, J.M. (1966) A Regional Test of the Heckscher-Ohlin Theorem.
Journal ofPolitical Economy
Nijkamp, P., ed (1990) Sustainability of Urban Systems - A Cross-National Evolutionary
Analysis of Urban Innovation, Avebury, Aldershot
Smith, B. (1975) Regional Specialisation and Trade in the UK. Scottish Journal of Political
Economy
Tinbergen, J. (1967) The Hierarchy Model of the Size Distribution of Centres. Papers of the
Regional Science Association, 20: 65-80
Vernon, R. (1960) Metropolis 1985, Harvard University Press, Cambridge Mass
Vernon, R. (1966) International Investment and International Trade in the Product Cycle.
Quarterly Journal ofEconomics, 80: 190-207
Chapter 9
Stefano Magrini
University Ca' Foscari, Venice, Italy
9.1 INTRODUCTION
The renewed attention over the past decade given to the theoretical issues
of economic growth has been accompanied by a growing body of empirical
analyses aimed, in particular, at confirming the existence of a process of
convergence across national and regional economies. A substantial part of
this empirical literature is made up of cross-sectional and panel data
regression analyses that focus on the behaviour of a representative economy,
making the implicit assumption that it is moving along a steady path of
growth.
Although these studies are developed within the framework of the
traditional neo-classical model of growth, many authors do not consider
cross-sectional and panel data regression analyses of convergence a valid
test for this theory (Romer 1993, 1994; Fagerberg 1994; Cheshire and
Carbonaro 1995; Barro and Sala-i-Martin 1995; Sala-i-Martin 1996). Indeed,
the typical finding that economic systems are converging at a stable annual
rate of2 per cent (among others Barro 1991; Barro and Sala-i-Martin 1991,
1992, 1995; Holtz-Eakin 1992; Armstrong 1995a, 1995b; Sala-i-Martin
1996) is consistent with many other explanations of the growth process. It
could, for example, be equally consistent with both the evolutionary and
endogenous growth theories. Other sources of empirical evidence on the
determinants of economic growth need to be utilised therefore in evaluating
the relative merits of the different approaches. Several findings (Lucas 1988;
Barro and Sala-i-Martin 1995; Pavitt and Soete 1982; Fagerberg 1987, 1988;
Acs et a/1992, 1994; Feldman 1994; Audretsch and Feldman 1996; Coe et
198 MagriniS.
al 1997 among others) suggest that the study of economic growth cannot be
separated from the study of technological change and its determinants. For
this purpose, cross-sectional regressions can still represent an useful tool,
provided that they are used to shed light on the role played by different
factors in the growth process, rather than to analyse whether the poor
economies are converging towards the richer ones.
Within the traditional neo-classical model, technological change is
interpreted as a purely exogenous phenomenon and thus no economic
explanation of its evolution is given. By contrast, the evolutionary approach
has developed a framework in which technological change is explained by
the action of economic agents. Much of the theoretical work within the
evolutionary tradition has relied primarily on appreciative theory, i.e. on less
abstract, more descriptive modelling. In response, recent 'endogenous
growth' contributions have tried to codify some of the fundamental elements
of the evolutionary view within the formal modelling tradition of
mainstream economics. As a result, although differing profoundly in many
ways, both frameworks interpret technological progress either as the by-
product of other economic activities or as the intentional result of research
efforts carried out by profit-seeking agents. They therefore consider human
capital and innovation as fundamental elements in the explanation of the
process of economic growth.
One of the aspects that has received insufficient attention from
endogenous growth theorists is the relationship between technological
progress, knowledge spillovers and space. Fagerberg emphasises that:
"appreciative theorising often describes technology as organizationally
embedded, tacit and cumulative in nature, influenced by the interaction
between firms and their environments, and geographically localised"
(Fagerberg 1994, p. 1170).
Tacit knowledge, in particular, being the non-written personal heritage of
individuals or groups is naturally concentrated in space. Moreover, because
of its personal nature, tacit knowledge spills over space essentially through
direct, face-to-face contacts. It is important therefore to explicitly analyse the
geographical dimension of these spillovers since it is a feature of knowledge
creation and transmission that has been entirely neglected in formal theories
of endogenous growth.
The present empirical analysis of the role of research and development
(R&D) activities in regional growth is based on a theoretical model already
developed elsewhere (Magrini 1997, 1998). The main elements will be
briefly recalled in the following section. Section 9.3 presents the empirical
The Impact of Research Activities on the European Urban System 199
results for a data set of 122 major European Functional Urban Regions
(FURs) over the period 1979-1990, and in Section 9.4 we draw some
conclusions.
The present analysis adopts the set of functional regions derived by Hall and Hay (1980)
and described in the Data Appendix. As argued by Cheshire and Hay (1989), Cheshire
and Carbonaro (1996), Cheshire et at (1996) and Magrini (1999), the very nature of
regional economic disparities means that any empirical study on the subject must take
space into consideration and opt for a definition of the region which centres on the spatial
sphere of socio-economic influence. Since the functional links between spatial units are
limited by space, functional regions seem to be a suitable choice, as they take account
explicitly ofthe distance factor.
200 Magrini s.
creating' regions. Since research activities tend to make a more intensive use
of human capital than manufacturing activities, the relative concentration of
research in one location leads to a parallel concentration of human capital.
Moreover, since wages for human capital tend to be higher than wages for
unskilled labour, the relative concentration of human capital in one region
implies that the average level of per capita income in these 'knowledge
creating' regions will be higher than in 'manufacturing' regions.
The model also offers a possible interpretation of the effects of European
integration on the disparities in per capita income. Indeed, the process of
integration which has characterised recent European history seems able to
foster the long term growth in the economic system, by determining a
reduction in the cost of physical distance. At the same time, however, given
that the reduction in the cost of distance has been achieved primarily through
a reduction in the travel time between locations (there have been few
improvements in cultural and institutional homogeneity, so a high degree of
heterogeneity still characterises the European system), the price to pay could
be an increase in regional differentials. In such a situation, even though
integration may reduce existing gaps in regional levels of technological
competence in research, disparities in per capita income are likely to widen.
In other words, the integration process might determine the emergence of a
new steady-state equilibrium characterised by a further concentration of
research activities in the regions which were already relatively specialised in
research. During the transition towards this new equilibrium, per capita
income growth rates differ across regions. While the adjustment takes place
through the reallocation of unskilled labour and human capital, the average
per capita income in the more innovative, relatively research-intensive
region grows at a faster rate than in the other region.
The fundamental equation of the model describes the activity of each
regional research sector. Let us consider an economic system made up of
two regions, i and j, in which the cost of moving from region one region to
the other is equal to dij. The flow of new knowledge, i.e. the number of new
designs created in region i at any point in time is given by:
(9.1)
where Hr; is the level of human capital employed in the research sector of
region i, and 0; is the level of technological competence characteristic of the
research sector located in region i. A is the number of intermediate inputs
existing in the system and the overall level of abstract knowledge created so
far (and available to all researcher due to aspatial knowledge spillovers). As
far as the spatial spillovers of technological tacit knowledge are concerned,
Hr/ reflects the size of the intra-regional spillovers, whilst the term
The Impact of Research Activities on the European Urban System 201
Hr} d;ll Pij represents the extent of the inter-regional spillovers of tacit
knowledge that benefit the research effort in region i and originate from the
interaction with the research sector located in region j. These are functions of
the level of human capital existing in the other region, Hrj, weighted by the
cost of physical distance, dij' and a measure of the potential technological
benefit from interaction, pij. As these spillovers result primarily from the
physical interaction between researchers, their size is inversely related to
(the cost of) physical distance. However, the 'catch-up' argument
(Gerschenkron 1962; Abramovitz 1986) emphasises the potential benefit that
can be enjoyed by technologically less advanced economies from the
interaction with economies closer to the technological frontier, due to the
possibility of imitating technologies already developed elsewhere. It is
assumed that the potential technological benefit accruing to researchers
located in one region from the interaction with the researchers of the
technologically more advanced region is an increasing function of the
relative local technological competencies in research ~echnological leader / 80ther
region and a decreasing function of the cost of the physical distance. The
measure of the potential technological benefit accruing to region i from the
interaction with region j can therefore be represented by:
(9.2)
where
8 if 8; > Ii}
{
Ii technological leader = ~ if 8; < Ii}
quality of space available for innovative activities, the structure of the local
industrial sector, and both the system-wide and local macro-economic
setting.
The aim of the present section is to provide a simple test for the model's
predictions. In particular, we study the fundamental determinants of
economic growth in the 122 major European FURs. The period chosen, from
1979 to 1990, conforms to two fundament requirements. Firstly, it is long
enough to allow for cyclical movements to occur around the growth trend.
Secondly, this is a period in which the European system underwent
important steps in its process of economic integration.
The dependent variable is the growth rate of per capita GDP in each
FUR. The formula for the growth rate is the traditional logarithmic
transformation of the ratio of regional per capita GDP at the beginning and
end of the study period:
R&D No
R&DSI Yes 90 minutes
R&DS2 Yes 110 minutes
R&DS3 Yes 115 minutes
R&DS4 Yes 120 minutes
R&DS5 Yes 125 minutes
R&DS6 Yes 130 minutes
R&DS7 Yes 150 minutes
presence of coal mining can adversely affect local growth prospects for a
considerable period of time, even after this industry has ceased to account
for a substantial share of employment. Consequently, the influence of the
coal industry is taken into account through a dummy variable related to the
coincidence of the area of the FUR with a coal field as defined in the Oxford
Regional Economic Atlas (1971).
The second variable reflects the extent of port activities, measured in
terms of the amount of sea-going freight handled in 1978. Dramatic new
developments in transport technology and, in particular, the introduction of
containerisation and roll-on roll-off ferries, have greatly reduced the
attractiveness of ports as locations for processing activities. During the study
period, this is likely to have negatively affected all ports in relation to their
size. However, the transformation in the industry is likely to have led to a re-
organisation of traffic flows and, therefore, to an increase in the degree of
competition among existing ports. Large ports, thanks to their economies of
scale, may have taken advantage of the process of re-organisation and
increased their share of traffic at the expense of smaller ports. As a result,
the relationship between port size and growth of per capita income in the
region could be quadratic rather than linear.
Another relevant feature of the local industrial structure concerns the
relative importance of service sector due to the role played by the variety of
business services in providing firms with market, financial and commercial
knowledge. This factor is measured as the percentage share of employment
in service activities over employment in services and manufacturing in 1980,
and is expected to be positively related to per capita GDP growth.
An interesting question related to the local industrial structure is whether
local ability to innovate is promoted by industrial specialisation, due to intra-
industry spillovers, rather than by industrial diversity and inter-industry
spillovers. The underlying theoretical model adopted here does not provide
any indication of the relative importance of these two possibilities, but
leaves the question open to empirical investigation. The degree of
specialisation of the regional economies is therefore measured on the basis
of employment data for nine industrial sectors. 3 The employment in each
regional sector is expressed as a percentage of the total industrial
employment in the region. After ranking the sectors by size, the index of
regional specialisation is calculated as the ratio between the average
percentage of employment for the four smallest regional sectors over the
average percentages of employment for the four largest ones. The index,
therefore, ranges between 0 and 1, these two extremes indicating
respectively specialisation and diversity in the regional industrial structures.
Yi _Yj
Yi Yj
(9.4)
dij
where dij represents the road distance between regions. Moreover, to avoid
problems of definitional correlation with the dependent variable, the growth
rates are calculated over the period 1979-1985. This variable is introduced
into the analysis in order to take into account the spatial adjustment between
neighbouring FURs. As explained by Cheshire (1979), adjacent local labour
markets tend to interact primarily through the adjustment of commuting
patterns (see also Evans and Richardson 1981; Burridge and Gordon 1981;
Gordon and Lamont 1982; Gordon 1985). Relatively rapid growth of per
capita GDP in one FUR will tend to attract additional in-commuters from
surrounding FURs. The impact of this mechanism is twofold. The first effect
The Impact of Research Activities on the European Urban System 207
-1 1n (Yi'1990
11 YiI979
J
- - =ao+a E i1979+Si1990
"
(9.5)
where E1979 is the vector of explanatory variables just described. The results
of the OLS cross-sectional estimation of these equations have been given in
Table 9.2.
4 A dummy variable for the Italian Mezzogiorno was also introduced. However, the
inclusion of this variable proved to add no explanatory power to the model.
S The FURs included in this variable are Alicante, Cordoba, Granada, Malaga, Murcia, and
Seville.
N
0
00
Table 9.2 The determinants of per capita GDP growth in the FURs
~
~
(JQ
2 3 4 S 6 7 8 :
Constant 0.00718 0.00692 0.0070S 0.00743 O.OO72S 0.00690 0.00713 0.00736 ~
(0.52) (O.SO) (0.51 ) (0.S4) (0.S3) (0.50) (0.52) (0.S3)
R&D 0.00011
(2.74)
R&DSI 6.6ge-S
(2.81 )
R&DS2 6.02e-S
(2.90)
R&DS3 S.92e-S
(3.01 )
R&DS4 S.6Se-S
(3.00)
R&DSS S.1ge-5
(3.00)
R&DS6 4.86e-S
(2.98)
R&DS7 4.34e-S
(2.9S)
University 3.81e-7 4.24e-7 4.23e-7 4.23e-7 4.24e-7 4.17e-7 4.0ge-7 4.15e-7
(2.22) (2.47) (2.47) (2.48) (2.48) (2.44) (2.40) (2.43)
Coal -0.00296 -0.00331 -0.00339 -0.00336 -0.00336 -0.00336 -0.00339 -0.00338
(-2.31) (-2.57) (-2.63) (-2.62) (-2.62) (-2.62) (-2.64) (-2.63)
(continues)
Table 9.2 The determinants of per capita GDP growth in the FURs (continued)
(l)
(-2.53) (-2.62) (-2.52) (-2.52) (-2.54) (-2.56) (-2.60) (-2.60) '"
!:l
n
Agriculture 8.9ge-4 9.26e-4 9.46e-4 9.55e-4 9.58e-4 9.54e-4 9.50e-4 9.4ge-4 ::r
(4.01) (4.09) (4.16) (4.21 ) (4.21) (4.20) (4.19) (4.18) :>
n
....
Agriculture 2 -3.35e-5 -3.43e-5 -3.48e-5 -3.50e-5 -3.51e-5 -3.50e-5 -3.50e-5 -3.50e-5 ~:
....
(-5.28) (-5.38) (-5.44) (-5.48) (-5.49) (-5.48) (-5.47) (-5.47) G
'"0
Density -0.00936 -0.00952 -0.00957 -0.00954 -0.00957 -0.00960 -0.00963 -0.00959 ::t
....
::r
(-3.99) (-4.06) (-4.09) (-4.09) (-4.10) (-4.12) (-4.12) (-4.11) (l)
tTl
SDG 0.19645 0.20122 0.20366 0.20424 0.20538 0.20554 0.20437 0.20251 ~
...
0
(4.93) (5.04) (5.10) (5.13) (5.15) (5.15) (5.13) (5.09) '"d
(l)
Whereas the first version of the model (shown in the first column) makes
use of the variable on research activity without considering spatial spillover
effects, these effects are allowed for in all the other estimated versions. The
results appear quite robust in all versions. The R 2 values range between 0.58
and 0.59, a satisfactory level for a large cross sectional data set. All the
expectations on the signs of the coefficients are met, and all coefficients are
generally highly significant.
The first important result is that not only do all the coefficients for the
variables reflecting the role of R&D activities on per capita GDP growth
have the expected positive sign, but they are also highly statistically
significant. Despite all the caveats concerning the measurement of this
activity expressed in the previous section, this is nonetheless an encouraging
result.
The comparison between the results on the R&D variables for the
different models gives some insight into the role of spatial knowledge
spillovers. Indeed, we should note that the inclusion of the spillovers in the
R&D variable determines a generalised improvement in the regression
results. Both the R 2 values and the level of statistical significance of the
coefficients for research activities are generally enhanced when these spatial
effects are accounted for. At the same time, the statistical significance of the
other variable closely related to innovation activity, the number of university
staff, is also substantially improved by the inclusion of these effects. All
these results could therefore be interpreted as supporting the view that spatial
spillovers of knowledge are an important feature of innovation activities.
Concentrating on the versions of the model that allow for these effects
(columns 2-8), it is possible to analyse the effect of space on the strength of
interaction between research sectors of neighbouring FURs. As explained in
the previous section, these variable are calculated by considering different
distance ranges for the inter-regional spillovers. The distance ranges
considered here vary from a minimum of 90 minutes to a maximum of 150
minutes. The best version of the model, both in terms of the regression
R 2 and of the t-ratios for the 'research activity' and the 'university'
variables, corresponds to the R&DS3 variable (column 4), which allows for
interaction between regional research sectors located within a range of 115
minutes. In other words, the strength of the inter-regional interaction
between researchers appears to reach its strongest level when the researchers
are on average less than 115 minutes apart. Moreover, the strength of the
inter-regional spillovers declines slowly with increases in the distance range
up to 125 minutes, whilst it appears to decline more rapidly for larger
distances.
Given these results, we now concentrate on the fourth version of the
model. The coefficient for the index of specialisation of the local industrial
structure is negative and significant at the 1% level. This suggests that, other
The Impact of Research Activities on the European Urban System 211
things being equal, those regions that were characterised by a higher degree
of specialisation in their industrial structures at the beginning of the study
period have grown faster. With regard to the nature of the intra-regional
spillovers, this suggests that, in aggregate terms, intra-industry dynamic
externalities have been more effective than inter-regional dynamic
externalities in stimulating per capita GDP growth.
The role of port activities on regional growth according to the estimated
coefficients is described in Fig. 9.1. The regression results seem to support
the view, expressed in the previous section, that the relationship between
port size and regional growth is quadratic. However, a closer look shows that
the minimum point of the curve is reached for an amount of trade just
exceeding 120 million tons. The only port which at that time was handling
more than 100 million tons was the port of Rotterdam (with 259 million
tons), whilst the second port in terms of goods handled was Marseilles with
93 million tons.
0.002
-0.004
-0.006
'Port2' and 'Port22 " the variables on port activity which exclude the
observation for Rotterdam, are both statistically non significant, thus
rejecting the hypothesis of a quadratic form for the influence of the other
European ports on regional growth. On the contrary, when the relation
between port activity and growth is assumed to be linear (version 10), the
coefficient is negative and highly significant.
4 9 \0
Constant 0.00743 0.00851 0.00823
(0.54) (0.62) (0.60)
R&DS3 5.92e-5 5.72e-5 5.76e-5
(3.01) (2.91) (2.97)
University 4.23e-7 4.16e-7 4.16e-7
(2.48) (2.46) (2.47)
Coal -0.00336 -0.00340 -0.00340
(-2.62) (-2.66) (-2.67)
Port -9.54e-5
(-2.43)
Porr 3.9Ie-7
(2.20)
Port2 -6.15e-4 -7.30e-4
(-0.82) (-2.47)
Port2 2 -1.78e-5
(-0.17)
Service 0.01630 0.01649 0.01658
(2.07) (2.12) (2.15)
Specialisation -0.03907 -0.03979 -0.03967
(-2.52) (-2.59) (-2.59)
Agriculture 9.55e-4 9.43e-4 9.44e-4
(4.21) (4.19) (4.22)
Agriculture 2 -3.50e-5 -3.48e-5 -3.48e-5
(-5.48) (-5.49) (-5.51)
Density -0.00954 -0.00942 -0.00947
(-4.09) (-4.02) (-4.09)
SDG 0.20424 0.20356 0.20343
(5.13) (5.14) (5.16)
NFGrowth 1.02832 1.01503 1.01784
(6.98) (6.90) (6.99)
DSE -0.00954 -0.00942 -0.00947
(-4.09) (-4.02) (-4.09)
0.006
10 20 % share
-0.004
multiplier test, Kelejian and Robinson's test for spatial error, and Anselin's
test for spatial lag (for details see Anselin 1988, 1994). In each case, the
tests were based on both distance matrices used in the derivation of the
variables of the model. All diagnostics excluded the presence of
specification problems with either of the two preferred versions of the
model.
9.4 CONCLUSIONS
The results of the regression analyses lend some support to the main
predictions of the theoretical model outlined in Section 2. These results can
be summarised as follows. Firstly, research activities appear to play an
important role in the process of regional growth. Indeed, the coefficients for
the variables measuring regional research efforts were always positive and
highly significant. Secondly, by considering different specifications of the
spatial interaction between researchers, it has been possible to find evidence
supporting the existence of spatial spillovers of knowledge. Inter-regional
spillovers of knowledge appear to reach their maximum effect within a
radius of about 115 minutes. Thirdly, it has been possible to identify several
factors which affect the regional growth rate of per capita GDP by shaping
the local level of technological competence in research. One of these appears
to be the existence of universities. Indeed, universities contribute to the
regional research effort both directly, in their role as research centres, and
indirectly, as part of the regional infrastructure providing new human capital.
Data limitations did not allow us to analyse these effects separately.
Nonetheless, according to the results of the empirical analysis, the
combination of these effects has a significant positive impact on regional
growth. Finally, another interesting outcome concerns the controversy on the
relative importance of intra-industry and inter-industry dynamic spillovers in
promoting growth. An index of the degree of sectoral specialisation of
regional industrial specialisation has been used to shed light on this widely
debated issue. The results indicate that, during the period 1979-1990,
European regions characterised by a higher degree of sectoral specialisation
grew faster than regions with a more diverse industrial structure. In other
words, intra-regional dynamic spillovers appear to have been more
successful than inter-regional dynamic spillovers in fostering regional
economic growth.
Finally, it is necessary to sound a note of caution in connection with the
problem of data limitations since these have, in some instances, led to the
use of rather rough measures for the variables required by the theoretical
model. This is the case, in particular, for the variable relating to research
The Impact of Research Activities on the European Urban System 215
DATA APPENDIX
6 Defined as those with a core city greater than 200,000 inhabitants and a total population
greater than a third of a million inhabitants.
216 Magrini S.
This variable reflects the number of academic staff (full-time plus part-
time) employed in universities and other further education institutions during
the 1976-1977 academic year. The sources of data on employment are:
REFERENCES
Abramovitz, M. (1986) Catching Up, Forging Ahead, and Falling Behind. Journal of
Economic History 46: 385-406
Acs, Z.1., Audretsch, 0.8., Feldman, M.P. (1992) Real Effects of Academic Research:
Comment. American Economic Review 82: 363-67
Acs, Z.1.,Audretsch, 0.8., Feldman, M.P. (1994) R&D Spillovers and Recipient Firm Size.
Review ofEconomics and Statistics 100: 336-40
Anselin L. (1988) Spatial Econometrics: Methods and Models. Kluwer Academic Publishers,
London
218 Magrini S.
Dino Martellato
University Ca' Foscari, Venice, Italy
10.1 INTRODUCTION
The answers to these questions bear essentially on the same arguments and
depart considerably from the early analyses of FDI.
It is recognised that a large number of economic factors are involved in
the decision to undertake FDI. In the explanations, a distinction can usually
be made between those based on theories assuming perfect markets and
theories which assume some degree of market imperfection. But whatever
the nature of the market, it would seem fair to say that the most common
factor behind FDI is the search for a cheaper - if not the cheapest - location
or way of serving the market (whether the foreign market, the home market
or the general market).
Here, I intend to focus on one single factor, namely the real exchange
rate, which was already considered by Aliber (1970, 1971) to be a powerful
factor behind a great many FDI decisions. However, the risk connected with
exchange rate fluctuations can be covered by hedging, so why should we
bother with this aspect? The point is that the exchange rate risk can be
hedged only partially. Real exchange rate deviations from the purchasing
parity equilibrium are considerable and appear to be one of the most
important factors influencing the expected revenue in many FDI initiatives.
Indeed, the role of real exchange rates has been revised more recently by
Caves (1988), Froot and Stein (1991) and others.
More specifically, I wish to consider the role of exchange rate for the
following reasons:
it is a fact that real exchange rates are not as stable as generally thought;
by managing nominal exchange rates, a country can manage its real
exchange rate.
a higher level of real exchange rate may be very positive for some
exporting sectors.
the current and expected level of the exchange rate is relevant for many
types of capital transfer and thus relevant for technology transfer.
The divergence between the current level of the real exchange rate of a
currency and its level in equilibrium is directly related to the strength of the
currency.
The nominal exchange rate is a strategic factor in the determination of the
competitiveness of a region (directly, through its effect on the real exchange
rate and indirectly, through FDI and innovation, the exchange rate drives the
competitive position of any region).
While the first three points do not need any special comment here, the
remaining three deserve some explanations. They wiIl be dealt with in the
following sections, but only after having considered the relationship between
FDI and the possibility of technology spillovers.
Is the Real Exchange Rate Relevant for Determining Innovation Potential? 223
old and mature products. From a certain point of view, Krugman's model is
a development of that presented in a celebrated article written by Dornbusch,
et al (1977) twenty years ago. Using a Ricardian model, in which goods can
be produced anywhere by labour only, the authors show how tastes,
technology and size jointly determine the equilibrium wage and the efficient
production specialisation of different countries. Their model takes into
account a technology transfer, but no new products. In additional, it is
assumed that income, but not capital, may be transferred from one country to
another. Krugman's contribution therefore provides those elements of
analysis which are lacking in the Dornbusch-Fisher-Samuelson model, i.e.
new products, relocation, capital transfer and technological diffusion.
In a recent research report, it was pointed out by the World Bank (1997)
that in the 1990s FDI has been used not so much a way for multinational
corporations to indulge in 'tariff jumping', as a way of competing. Currently,
FDI benefits in particular from factors such as:
(i) In the first case, the creation of a foreign subsidiary implies export
substitution, which may occur in two situations:
because the parent firm wants the production of its foreign subsidiary to
replace current domestic production. If the outside production is for the
outside market, FOI leads to a reduction in exports, i.e. a trade
contraction. However, a revaluation of the home currency would also
imply a reduction in exports, thus FOI is a way of keeping the market
share;
if the outside production is both for the domestic market and the general
market, there is a switch from exports to imports. This leads, at least
partially, to a case of trade diversion. An increase in the real exchange
rate in the country where the foreign subsidiary has been located (or a
revaluation of the home currency, which is equivalent) yields a
reduction in production costs (imports) and hence higher profits.
(ii)ln the second case, the creation of a foreign subsidiary does not imply
that there is export substitution. This case of horizontal extension is
explained by the pursuit of profits by direct penetration in a foreign and,
arguably, growing market. In this situation, an increase of the real
exchange rate of the country where the foreign subsidiary has been
located could imply a reduction in profits, as the market could be best
served with imports from outside.
(i) A backward extension occurs when the early stages of the production
process are decentralised. In this case imports of intermediate products
increase and the outflow of foreign currency increases, but a devaluation
of the foreign currency, as before, implies lower costs and thus higher
profits for the parent firm.
(ii)A forward extension occurs when the end stages of the production
process are decentralised. In this case, the parent firm sells intermediate
products at a lower price, thus increasing profits in the foreign subsidiary.
A devaluation of the currency would imply higher profits for the parent
firm, but lower ones for the subsidiary.
In the preceding analysis it has been pointed out that there are many
factors which impinge on FDI, but it has also been argued that the current
228 Martellato D.
and expected levels of the real exchange rate are able to determine the rate of
return on investments and thus influence many types of FOI. It has also been
argued that capital transfer has an impact on a region's innovation potential,
although it is not clear if technology spillovers are necessarily an outcome of
FOI. Furthermore, it is clear that trade flows and FDI are correlated. In this
section, I wish to show that a sustainable condition for a capital importing
region implies a trade surplus. Accordingly, we investigate the macro-
economic conditions which tend to favour both a trade surplus and a foreign
capital inflow.
As the region's trade balance can be explained, other things being
constant, by its competitive position, i.e. by the real rate of exchange and the
innovative content of its products, we need to focus on the relationship
between the trade balance and the net debt position of the region, as it is
obvious that this will be modified by FOI. Particular attention is therefore
given to the relationship between trade balance and capital mobility. To do
so, we adopt a very simple model that has also been used to highlight the
regional impact of the EMUs and which directly involves the relationship
between trade and debt. The resulting equation has some interesting
properties. First of all, it illustrates the role of the rate of return and that of
the real exchange rate in determining the long run or sustainable ratio
between the trade balance and the net financial position. Furthermore, it
shows quite clearly that the trade balance and the region's debt position are
cyclical in nature.
There is an obvious link between the investment/savings gap and the
change in the stock of net foreign assets owned by an economic agent. When
personal savings are insufficient, the agent must raise extra funds. This
constraint holds true for whatever set of agents (whether these are found in a
region or in a country). It reads:
(10.1)
where (t) is the year of reference. If the savings gap is permanent, the
regional net debt will be piling up. Let us assume that this condition holds
for a while and that the region's net debt position grows at an exponential
rate equal to (g):
(10.2)
(10.3)
where r is the average rate of interest and the expression in brackets is the
trade balance. By combining the last two equations, we obtain the most basic
relationship of this section, that is:
(10.4)
This relationship makes the growth rate (g), the interest rate (r) and the
trade balance (M - X)I mutually consistent for a given net stock of external
debt (DI). From this equation it is easy to derive a relationship in which the
rate of change (g) or the stock (D1+1) is related to the real exchange rate and
the rate of interest.
An equivalent equation says that the net stock of external debt is equal to
the current one, conveniently capitalised, plus the trade balance:
(10.5)
e = aP' / P (10.6)
where the external and internal price indices of exportable goods are
respectively ( P * ) and (P). These also depend on external demand Y * and
internal demand Y in a standard way:
e=a-b-l>O (10.9)
This equation gives the feasible current growth rate of the net debt, given
the average rate of return (r), the stock (D) and the current trade balance. It
can be used to describe a regional cycle of trade and capital transfer. The
stock of net external debt (credit) may be at any of the points of the straight
line in Fig. 10.1, but not all these points are equivalent from an economic
point of view. In the figure there are three pairs of symmetric conditions
which we shall now comment upon.
87
f6\.
Sustainable
conditions
Unsustainable
conditions
In CASE 1, a net external debt is associated with a trade deficit (in the
symmetrical CASE 6, a net external surplus is associated with a trade
surplus). Clearly, both these conditions cannot be sustained for long.
In CASE 2, the net external debt is increasing because a trade surplus
only partially offsets negative interest payments. In CASE 5, a net external
credit position is paired by a trade deficit and positive interest payments
again with only partial compensation.
CASE 3 represents that of a region where the net debt decreases as the
trade surplus outperforms the negative interest outlays. Case 4 is
symmetrical, in that the stock of net credit decreases over time as the trade
deficit outperforms the positive interest payments.
The dynamics of the stock of debt is given by equation (10.11) slightly
reformulated as:
A more complete way of looking at the same cycle is shown in Fig. 10.2
where the trade deficit and the stock of net debt are simply plotted against
time. The figure indicates all six cases, but some of them (where the curves
cross) have to be considered non sustainable since they may be maintained
for very short time periods only. The system is able to move from one stage
to the next through a suitable change in the rate of return on capital and the
real exchange rate.
The first conclusion that can be drawn from the preceding analysis is that
a region can modify its competitive position in many ways, but - in the
present instance - the main ways would be a) by increasing the rate of return
of capital or b) by increasing the real exchange rate (here I refer to the
expression within the inner brackets of the last two equations). This latter
can be done in two ways: by undertaking a competitive devaluation
(increasing a) or through competitive deflation (i.e. by controlling
production costs and increasing export prices, but by a lesser amount).
It is interesting to note that the first alternative, a), is by no means in
conflict with either of the last two alternatives, but that these last two may be
in conflict with each other. There is however another possibility. This
implies a squeeze on domestic demand and employment, but - at least in
theory - no government would be keen to deliberately pursue it.
More in general, the conclusion is that a region has to improve its macro-
economic fundamentals as well as the internal micro-economic conditions
(e.g. infrastructure and human capital supply, social environment, etc.).
However, since R&D activities are capital intensive and probably have
232 Martellato D.
strong scale economies, small countries and regions may not be well suited
to competing in research activities and would possibly do better limiting
themselves to importing new technologies from outside. In this case,
innovation may not be the result of domestic R&D investments, but the
consequence of strong macro-economic conditions, if these favour an inflow
ofFDI.
The argument we tried to develop in the first part of the chapter was that
by encouraging inward FDI, and by using a suitable combination of the three
alternatives mentioned above 9. a region can benefit from technology
spillovers, faster innovation and increased capacity in the industries with
exportable goods. A further striking conclusion is that by entering EMU, a
region is obliged to exclude exchange manoeuvres and to rely on the other
two policy options a) or b), as well as internal demand restraint.
eo
,l 60
I
I
" l
f I
jl .~~~~[JJIIIII~JI.II~III'1 I;])
20
~
11"1 o ~
20
-40
Peiods
- Trade balance ~ Stock ri net det:t
9 Alternative a) is by no means in conflict with alternative b). A high exchange rate may
sustain the investment return and encourage the inflow of capital.
Is the Real Exchange Rate Relevant for Determining Innovation Potential? 233
REFERENCES
Aliber, R.Z. (1970) "A Theory of Direct Foreign Investment." In The International
Corporation: A Symposium. Kindleberger, C.P., ed, The MIT Press, Cambridge Mass
Aliber, R.Z. (1971) The Multinational Enterprise in a Multiple Currency World. In The
Multinational Enterprise, Dunning, 1.H., ed, The MIT Press, Cambridge Mass
Caves, R.E. (1971) International Corporations: The Industrial Economics of Foreign
Investment. Economica 38: 1-27
Caves, R.E. (1982) Multinational Enterprise and Economic Analysis. Cambridge University
Press, Cambridge
Caves, R.E. (1988) Exchange Rate Movements and Foreign Direct Investment in the United
States; Harvard Institute of Economic Research, Discussion Paper 1383, Harvard
University
Coe, D., Helpman, E., Hoffmaister, A. (1995) North-South R&D Spillovers. Discussion Paper
1133, Centre for Economic Policy Research, London
Dornbusch, R., Fischer, S., Samuelson, P.A. (1977) Comparative Advantage, Trade, and
Payments in a Ricardian Model with a Continuum of Goods. American Economic Review
67 (5): 823-39
Fitoussi, J.-P. (1995) Le Debat Interdit, Monnaie, Europe, Pauvrete. Editions Arlea, Paris
Froot, K.A., Stein, 1., Stein, C. (1991) Exchange Rates and Foreign Direct Investments: An
Imperfect Capital Markets Approach. Quarterly Journal of Economics 106: 1191-217
Grossman, G.M., Helpman, E. (1991) Innovation and Growth in the Global Economy. The
MIT Press, Cambridge Mass
Krugman, P. (1979) A Model oflnnovation, Technology Transfer, and the World Distribution
of Income. Journal of Political Economy 87 (2): 253-66
Martellato, D. (1999) "Trade and Investments in the EMU Regions." In Spatial Dynamics of
European Integration. Regional Policy Issues at the Turn of the Century. Fischer, M.M.
and Nijkamp, P., eds, Springer, Berlin
World Bank (1997) Global Development Finance, Washington DC
Section D:
11.1 INTRODUCTION
2. what outcome is likely for the different types of area (i.e. different kinds
of region) affected by the above processes?;
3. what kind of phenomena should be observed and studied in a 'spatial'
analysis of NIT?
I 1/ III III IV
Period 178011850 185011890 189011940 194011980 Since 1980
Main features Mechanisation Steam engines Electrical and Fordist mass Information
heavy production and tele-
engineering communica-
tions
Main sectors Textiles, Steam Electrical Cars, Computers,
iron-working, engines, machinery, armaments, capital goods,
water power steamships steel ships consumer telecomm-
durables, unications,
petro- optical fibres
chemicals
Infrastructure Canals, roads Railways, Electrical Highways, Digital
world cables airlines, communica-
shipping airports tions,
networks,
satellites
Efficiency
Demand pressures,
Continuousl y improvements,
Incremental socio-cultural factors,
distributed over ! All productivity gains, medium run
innovation I technological
ti me and sectors quality
opportunities
imDrovements
R&D activities
Structural changes,
Discontinuously in firms,
Radical limited and
distributed over universities, ns long-run
innovation localised economic
time and sectors government
impacts
laboratories -- ~-
Legend
Adaptable areas Areas in which there is a high resource potential
and firms with creative (entrepreneurial) behaviour
Adaptive areas Areas dominated by standardised production in which
firms are sensitive to locational costs and factor prices
Non-adapted areas Areas with an insufficient adjustment performance
and stagnating firms
But even the most recent literature, where the two analytical perspectives,
the systemic and the substantive, are being brought together to build sounder
approaches to the interpretation of the relationships between innovation and
urban development, i.e. what some authors call "the social construction of
technology" (Graham and Marvin 1996), there are difficulties with the very
concept of innovation as far as explanations of urban evolution are
concerned.
244 Bertuglia c.s., Occelli S.
We argue that these difficulties are due to the fact that the concept of
innovation - commonly understood in general terms as a 'thing' (i.e. a
product, an artefact, a technique, an idea or even a behaviour) which is new
because it is qualitatively different from the existing form - needs to be
extended and refined.
First of all, we need to recognise that an innovation is made up of two
main components, the 'hard' and the 'soft' part, which although closely
interrelated are nonetheless distinct:
The above concept of innovation is very general and can be seen to hold
for the different kinds of innovation (including products, artefacts,
techniques, or simply ideas). It is, of course, particularly relevant for the so-
called New Information Technologies (NIT), i.e. those technologies derived
from the convergence of telecommunications and computing, which
characterise the 'new informational society' (Castells 1989). The point worth
Impact of the New Information Technologies 245
emphasising here is that the soft component represents the most outstanding
aspect of innovation today. Although still largely unnoticed, it constitutes the
'live wire' connecting most of the wide-ranging issues in the current debate
on innovation.
The two components of innovation also have deep implications for the
way in which we interpret the nature of innovation and identify the agents
likely to trigger and/or diffuse innovation. As far as the nature of innovation
is concerned, it seems to be increasingly difficult, as noted by some authors
(see Barnett 1953), to put forward any explanation which depends
exclusively on the properties of the innovation itself, since these are context
dependent. In this respect, for example, the distinction between basic and
derived innovations becomes blurred and meaningless. Only when seen in
terms of its hard component can innovation be 'referenced' to its
characteristics, its spatial location and the timing of its appearance.
When the soft component is taken into account, the interaction patterns
come to the fore. As innovation implies re-organisation and reshaping, the
interaction patterns are changed through a whole range of behavioural (i.e.
goal seeking, explorative, imitative, hyper-selective) processes. These can
act at both the individual and collective level and affect spatial systems in a
number of different ways which all feed its evolution. In this respect, it
appears that innovation is a permanent feature (a process) of system
evolution. Moreover, everyone is likely to be an agent of innovation. By
identifying the ordinary agent, whether an individual, a household, an
organisation or any collective society, as the main repository of innovation, a
less elitist view is posited.
Finally, the awareness of the twofold components of innovation with
their corollary implications will also significantly affect our ways of
conceiving the relationships between innovation and urban development. It
is the ability to continuously adjust, modify and create new interpretations of
these relationships (i.e. the mental models of the future course of the world,
see Silverberg 1988; Andersson et at 1993b) which moves the agents'
actions and ultimately shapes the direction of urban development.
Adoption perspective
Consumer innovation
Demand aspect of diffusion
Processes bv IVhich it occurs
a. The thesis advanced in this chapter is that major potential for urban
development is to be found in the 'soft component' associated with the
use of innovation. This consists of the continuously modified interaction
patterns, which define what is usually understood as the enabling role of
innovation. The soft component makes a wide range of opportunities
accessible and beneficial to individual urban agents as well as collective
agents. How to create and make use of these opportunities represents the
challenge of planning the city of tomorrow. A corollary of this thesis is
that the soft component is primarily an 'individual experience' on the part
of ordinary agents (whether an individual, a firm, a household, an
institution or a planning department) encountered in performing their
daily activities which occur in built places located in geographical space.
b. Whereas innovation has to be considered an intrinsic feature of system
evolution, whose appearance can be casual and unexpected, the
environmental conditions (i.e. the economic, social, spatial, cultural and
institutional context) are largely responsible for its viability. This holds at
the micro (individual) level as well as at the macro (collective) level. The
existing spatial structure as well as the built environment and cultural
milieu, therefore, are likely to play an important role in this respect.
248 Bertuglia c.s., Occelli S.
Within the above framework, we shall now briefly discuss some aspects
which could become major topics of a future research agenda dealing with
the relationships between innovation and urban development. These can be
listed under the headings: a) accessibility, b) people and places, c)
representations of innovation.
a) Accessibility
example, Castells 1989; Bertuglia and Occelli 1995; Batty 1995; Capello and
Nijkamp 1996; Graham and Marvin 1996). A major feature emphasised in
these studies is that the spatial and temporal re-organisation induced by NIT
allows socio-economic processes to be 'co-evolutionary', thus increasing the
degree of freedom in the evolution paths of urban systems. This also
enhances the planning possibilities, requiring the formulation of 'creative
visions' of likely futures, as well as careful assessments of the social benefits
and costs involved.
An urban system, i.e. a region, is however - no matter what kind of
description (iconographic, linguistic or formalised) we use to define it (Preto
1995) - an entity in which the space of flows (i.e. the tangible and intangible
relationships determining the system interactions) coexists with a space
consisting of places (i.e. the human settlements located in a built, natural and
cultural environment). What ultimately matters is the path of changes made
possible by the co-evolutionary role of NIT. We therefore need to make an
assessment of the outcome of lock-in and/or bifurcation processes which are
liable to occur, and an assessment of the capacity ofthe various agents of the
system to cope with changes.
As system innovative ness has its roots in the ability of individual agents
to exploit the soft component of an innovation and seize the resulting
opportunities (Salomon 1996), a major dimension to be looked at in a future
research agenda is what can be called 'the local dimension of the urban
action space'. By that we simply mean that improvements achieved through
innovation should be an endogenous. and permanent driving force in the
behaviour of the various urban actors and the functioning of the system.
To give an example, we mention three domains where such a drive would
seem likely to produce relevant transformations. The first is transportation
where, besides the complementary/substitution issue between physical
movement and virtual interaction, also the logistical organisation of
transportation modes and user services are deeply affected by innovation
(see, Hepworth and Ducatel 1992). A second domain concerns the
management of local development as the outcome of a multitude of
corporate and institutional decisions (Bennett and McCoshan 1993). A major
area in which innovation can have a significant role is the enhancement of
the capacity of local administrations and 'governance' to respond to the
global socio-economic changes as well as to deal with everyday
commitments. The third domain concerns the conservation of the built
environment as having a value of its own. This is a multi-faceted topic which
is attracting growing interest, especially in the most developed countries, as
a result of increasing sensitivity to the historical heritage consolidated in the
built landscape of human settlements (Coccossis and Nijkamp 1995). To be
Impact of the New Infonnation Technologies 251
c) Representations of innovation
means for sharing the experience gained elsewhere, and offers support in
learning about 'how to learn about' urban processes.
REFERENCES
Amin, A., ed (1994) Post-Fordism. Blackwell, Oxford
Andersson, A.E., Batten, D.F., Kobayashi, K.,Yoshikawa, K., eds (1993a) The Cosmo-
Creative Society. Springer, Berlin
Andersson, A.E., Batten, D.F., Kobayashi, K., Yoshikawa, K. (1993b) "Logistical Dynamics,
Creativity and Infrastructure." In The Cosmo-Creative Society. Andersson, A.E., Batten,
D.F., Kobayashi, K., Yoshikawa, K., eds, Springer, Berlin, 1-16
Barnett, H.G. (1953) Innovation: The Basis of Cultural Change. McGraw-Hili, New York
Batten, D., Casti, 1., Thord, R., eds (1995) Networks in Action. Springer, Berlin
Batty, M. (1995) The Computable City. Keynote Address at the Fourth International
Conference on "Computers in Urban Planning and Urban Management", Melbourne, July
11-14
Bennett, R.1., McCoshan, A(1993) Enterprise and Human Resource Development: Local
Capacity Building. Chapman, London
Bertuglia, C.S., Fischer, M.M., Preto, G. (eds.) (1995) Technological Change, Economic
Development and Space, Springer, Berlin
Bertuglia, C.S., Lombardo, S., Nijkamp, P., eds (1997) Innovative Behaviour in Space and
Time, Springer, Berlin
Bertuglia, C.S., Lombardo, S., Occelli, S. (1995) "The Interacting Choice Processes of
Innovation, Location and Mobility: A Compartmental Approach." In Technological
Change, Economic Development and Space. Bertuglia, C.S., Fischer, M.M., Preto, G., eds,
Springer, Berlin, 118-44
Bertuglia, C.S., Occelli, S. (1995) Transportation, Communications and Patterns of Location,
in Technological Change, Economic Development and Space, Bertuglia C.S., Fischer
M.M., Preto G., eds, Springer, Berlin, 92-117
Brotchie, 1., Batty, M., Hall, P., Newton, P., eds, (1991) Cities of the 21st Century: New
Technologies and Spatial Systems. Longman, London
Brotchie, J, Hall, P., Newton, P. (1986) The Transition to an Information Society. In Brotchie
1., Hall P., Newton P., eds, The Spatial Impact of Technological Change. Croom Helm,
London, 435-51
Brown, L.A. (1981) Innovation Diffusion: A New Perspective. Methuen, London
Burns, L.D. (1979) Transportation, Temporal and Spatial Components of Accessibility.
Lexington Books, Lexington Mass
Capello, R., Nijkamp, P. (1996) Telecommunications Technologies and Regional
Development. The Annals of Regional Science 30: 7-30
Castells, M. (1989) The Informational City. Blackwell, Cambridge
Coccosis, H., Nijkamp, P. (1995) Planning for Our Cultural Heritage. Avebury, Aldershot
Freeman, C., Perez, C. (1988) "Structural Crises of Adjustment: Business Cycles and
Investment Behaviour." In Technical Change and Economic Theory. Dosi, G., Freeman,
c., Nelson, R., Silverberg, G., Soete, L., eds, Pinter, London, 38-66
Gaffard, J.L. (1990) Innovations et changements structurels, Revue d'Economie Politique 3,
325-82
Giaoutzi, M.(I990) "Telecommunications Infrastructure and Regional Development." In
Infrastructure and the Space-Economy, Peschel, K., ed, Springer, Berlin, 116-30
Impact of the New Information Technologies 253
Graham, S.D.N., Dominy, G.R. (1991) Planning for the Information City: The U.K Case,
Progress in Planning 35, 169-248
Graham, S., Marvin, S. (1996) Telecommunications and the City. Routledge, London
Hepworth, M., Ducatel, K (1992) Transport in the Information Age. Belhaven Press, London
IRES (1995) Un 'analisi dell 'accessibifita in Piemonte. Studio di supporto alia valutazione
delle politiche del piano regionale dei trasporti. Occelli, S., Gallino, T., eds, Quaderni di
Ricerca IRES 74, Turin
IRES (1996) Telecomunicazioni e imprese: if casa del Piemonte. Lanzetti, R., Antonelli, C.,
Rizzello, S. (eds.) Quaderni di Ricerca IRES 80, Turin
Kobayashi, K., Sunao, S., Yoshikawa, K.(I993) "Spatial Equilibria of Knowledge Production
with Meeting-Facilities." In The Cosmo-Creative Society. Andersson, A.E., Batten, D.F.,
Kobayashi, K, Yoshikawa, K., eds, Springer, Berlin, 219-44
Martellato, D.(1993) "Reti di interazione e tecnologia della informazione. Un'analisi dei loro
effetti sulla localizzazione." In Innovazione e trasformazioni della citta: Teorie, metodi e
programmi per if mutamento. Lombardo, S., Preto, G., eds, Collana AiSRre, Franco
Angeli, Milan, 130-47
Mela, A.(1991) Innovation, Communication and Social Structures. Sistemi Urbani 13: 97-108
Metcalfe, 1.(1988) 'The Diffusion of Innovation: an Interpretative Survey." In Technical
Change and Economic Theory. Dosi, G., Freeman, C., Nelson, R., Silverberg, G., Soete L.,
eds, Pinter, London, 560-89
Miles, I., Robins, K (1992) "Making Sense of Information." In Understanding Information,
Business, Technology and Geography. Robins, K (ed.) Belhaven, London, 1-26
Occelli, S., Gallino, T. (1992) Un'analisi dell'accessibilita in Piemonte: note metododologiche
ed applicazione empirica di alcuni indicatori, Atti della XlII Conferenza Italiana di Scienze
Regionali, Ancona, 337-68
Preto, G.(I995) "The Region as an Evolutive System." In Technological Change, Economic
Development and Space Bertuglia, C.S., Fischer, M.M., Preto, G. (eds.), Springer, Berlin,
257-75
Rabino, G.A., Occelli, S. (1997) Re-thinking Urban System Modelling: New Features and a
Proposal. Paper presented at the ICCS Conference, Cortona, 22-25 September
Robins, K. (ed.) (1992) Understanding Information, Business, Technology and Geography.
Belhaven, London
Robins, K., Gillespie, A. (1992) "Communication, Organisation and Territory." In
Understanding Information, Business, Technology and Geography. Robins, K, ed,
Belhaven, London, 157-64
Salomon, I. (1986) Telecommunications and Travel Relationships: A Review. Transportation
Research A, 20A: 223-38
Salomon, I. (1996) Telecommunications, Cities and Technological Opportunism. The Annals
of Regional Science 30, 75-90
Steiner, M.(1990) "How Different Are Regions? An Evolutionary Approach to Regional
Inequality." In Infrastructure and the Space-Economy. Peschel, K., ed, Springer, Berlin,
294-316
Chapter 12
12.1 INTRODUCTION
In recent years, there has been increasing recognition of the role that
telecommunications could play in facilitating rural development. In part, this
recognition is based on technical advances within telecommunications
networks, particularly their digitalisation, which is opening up a range of
possibilities for combining communications with computing to create new
'telematics' services and applications. Beyond this technological impetus,
there is also recognition that profound structural changes in the advanced
economies - embodied in sectoral shifts from industrial to post-industrial or
informational actIvities, and in shifts from centralised 'Fordist'
organisational structures towards more decentralised and flexible 'post-
Fordist' forms - are reducing or eliminating many of the distance-related
constraints upon processes of economic development in remote or rural areas
(Gillespie et aI1994).
In the North American context, for example, Hudson and Parker (1990)
see the combination of telecommunications advances and structural
economic changes as having the potential to remove the 'distance penalty'
under which rural America has long laboured:
"In the provision of physical goods and services, rural areas could
compete across barriers of distance and geography if they had a natural
resource advantage. In the provision of information goods and services,
reliable telecommunications infrastructure can make geography and
distance irrelevant" (p.195).
There are a number of ways in which telematics innovations could,
potentially at least, contribute to overcoming the rural 'distance penalty'.
These include:
telematics... are far from being realised. The principal reason is user-
resistance"(llbery et al1995, p.66).
The suggestion that user-resistance lies behind the observed phenomenon
of rural under-utilisation is undoubtedly part of the explanation. The
adoption of telematics is a form of innovation likely to be suppressed in
traditional or conservative approaches to management, which we might
reasonably expect to characterise many firms in rural areas. Further,
however, the expectation that firms in rural areas should be making more use
of telecommunications services because of their remote location is based on
a misconception of the role of telecommunications; telecommunications is
not primarily a tool for overcoming distance, but rather for managing
complexity. Firms with simple transactional structures, with relatively stable
markets and which operate from single sites, will in consequence have
relatively limited 'need' for telecommunications, regardless of the location
from which they are operating. By contrast, firms with complex transactional
structures, with rapidly changing markets or which operate from multiple
sites, will display high levels of 'need' to telecommunicate, again regardless
of the particular locations from which they operate. The propensity to
telecommunicate is thus dependent primarily on the characteristics of the
firm, not on its location or its geographical 'distance' from centres of
economic activity. We would thus contend that the observation that firms in
rural areas make less use of telecommunications than expected is, at least in
part, explained by the nature of the firms concerned; simply, they are making
as much use of telecommunications as they need to, given the requirements
stemming from their transactional structures and forms of organisation.
For such reasons, rural areas with broadly similar geographical
characteristics in terms of remoteness and distance from cities can display
markedly different telecommunications usage profiles, due to variations in
the types of firms operating from them. A study of telecommunications
usage in rural areas in Scotland, for example, revealed a marked contrast
between the relatively high use of data communications in firms in Dumfries
and Galloway, a remote rural region with few large towns, and the much
lower use evident in the Borders region, a less remote rural region with an
urbanised core (Gillespie et al 1994). The explanation for the difference in
usage lay in the nature of the firms present in each area. In Dumfries and
Galloway, a polarised enterprise structure was apparent, with some large
branch plants making intensive use of data communications to connect them
into global corporate networks; in contrast, in the Borders region the firms
were mainly medium-sized, locally-owned and operating from single or a
few local sites, and data communications was little used. Significantly,
however, their limited use oftelematics cannot be taken to imply a constraint
on their activities, or a barrier to accessing remote markets. Indeed, the firms
Telematics Innovation and the Development of Non-Metropolitan Areas 259
improving data transfer services within the area, and making them
available at local call rates;
up-grading 65 exchanges to support ISDN services;
establishing a Network Services Agency to provide value-added services
and access to national databases.
The rationale for the initiative was to defend existing jobs, to allow new
data-intensive businesses to be set up, to encourage existing firms to expand
into new ventures, and to improve the area's attractiveness for inward
investment. Of these objectives, only the latter has really borne fruit, with a
number of back offices and telephone call centres having decided to locate in
the region. The take-up of advanced communications services by existing
firms has been limited, and the infrastructure remains massively under-
utilised (Richardson and Gillespie 1996).
Although the Highlands and Islands Initiative was not based on the belief
that inadequate infrastructure was holding back the region's development,
the decision to invest in infrastructure 'ahead of demand' was clearly
intended both to prevent future bottlenecks developing and stimulate new
262 Gillespie A. and Richardson R.
rural SMEs of such telematics applications and services, with the specific
aim of helping them gain access to remote markets, will often have
disappointing or limited results: Many of the firms concerned will have
products or services which are not competitive outside the local market, and
others will need to put in place mechanisms for providing them with direct
presence in remote markets if they wish to compete effectively.
Ever since Toffler (1980) put the electronic cottage at the centre of his
'third wave' society, those concerned with the development of rural areas
have attached considerable importance to the prospects for rural teleworking.
The limited and relatively narrow range of employment opportunities in
rural areas could, it is suggested, be transformed by the availability of new
teleworkingjobs which are not tied to any particular location.
There are, however, a number of problems with this vision. Firstly,
teleworking does not really provide new jobs, but rather a new tele-mediated
way of undertaking existing jobs, or types of jobs. A pre-requisite for
participation in teleworking is, therefore, the possession of the types of skills
and competencies which can be amenable to teleworking. Many rural
residents will not possess such skills. Secondly, many teleworking jobs are
not as locationally footloose as is often supposed. For employed teleworkers
(as opposed to those who are self-employed), survey evidence from Britain
suggests that they spend on average a quarter of their time on the employer's
premises and a further quarter elsewhere, for example on clients' premises
(Huws 1993). As a consequence, they are most likely to be found in
suburban and urban areas, where they are close to their employers and
clients. A recent study on the prospects for partially home-based teleworking
in Britain concluded that
"while a minority of rural areas offer considerable potential for the
development of this form ofteleworking, the majority, including all those
categorised as 'deep rural', do not" (Huws et a11996, p. 33).
More locational flexibility is likely for those who are self-employed or
free-lance teleworkers. Though there are still constraints due to the need for
access to clients, Huws et al (1996) conclude that this type of teleworking
has the highest potential for rural areas, as result of life style choices which
are favouring attractive rural areas.
For such reasons, teleworking in rural areas is likely to be primarily a
feature of in-migrating professionals seeking a particular quality of life. This
is certainly the pattern in the Highlands of Scotland, which have proved
successful in attracting teleworkers to locate in often deeply rural
environments (Richardson and Gillespie 1996). Significantly, these
266 Gillespie A. and Richardson R.
12.8 CONCLUSIONS
Our contention in this paper is that rural telematics policies have, by and
large, failed to deliver the expected benefits to rural economic development.
This is largely because of the mis-specification of the barriers or obstacles to
development. The two main types of barrier which the policies were
designed to overcome are: a) barriers of access to telecommunications
infrastructure, IT facilities and appropriate telematics services, and b)
barriers of access to markets and customers. .
The first barrier has proved to be largely illusory. Policies designed to
tackle the various impedances to access (individually or in combination),
have made it increasingly clear that the real barrier lies not in access, but in
the limited use which rural SMEs make of infrastructures, facilities and
services. While policy-makers have correctly identified the existence of a
'usage gap', the diagnosis of access problems has led, too often, to policies
likely to exacerbate the real problem. Once the supposed supply-side access
problems have been 'solved' through the provision of infrastructure ahead of
demand, the provision of telecottages with IT facilities, or the design of
telematics services targeted at rural SMEs, it becomes even more starkly
apparent that the real barrier is not one of availability, but the lack of usage
of these infrastructures.
Telematics Innovation and the Development of Non-Metropolitan Areas 267
REFERENCES
Biehl, D., ed, (1986) The Contribution of Infrastructure to Regional Development. (2 vols),
Office for Official Publications of the European Communities, Luxembourg
Cronin, FJ., Parker, E.B., Colleran, E.K., Gold, M.A. (1993) Telecommunications
Infrastructure Investment and Economic Development. Telecommunications Policy 17(6):
415-30
Ewbank Preece (1993) STAR - Special Telecommunications Action for Regional
Development: Community-level Evaluation, Executive Summary. Ewbank Preece,
Brighton
Gillespie, A, Coombes, M., Raybould, S. (1994) Contribution of Telecommunications To
Rural Economic Development: Variations on a Theme? Entrepreneurship and Regional
Development 6: 201-217
Gillespie, A, Cornford, 1. (1996) "Telecommunications Infrastructures and Regional
Development". In Information and Communication Technologies: Visions and Realities.
Dutton, W., ed, Oxford University Press, Oxford
Gillespie, A, Richardson, R., Cornford, 1. (1995) Review ofTelework in Britain: Implications
for Public Policy. Report prepared for the UK Parliamentary Office of Science and
Technology, CURDS, University of Newcastle upon Tyne
Gillespie, A, Robins, K. (1991) "Non-universal Service? Political Economy and
Communications Geography." In Cities of the 21'1 Century: New Technologies and Spatial
Systems. Brotchie, J., Batty, M., Hall, P., Newton. P., eds, Longmans, Harlow, 159-70
Gold, 1.R. (1991) "Fishing in Muddy Waters: Communications Media, Homeworking and the
Electronic Cottage." In Collapsing Space and Time: Geographic Aspects of
Communication and Information. Brunn, S.D., Leinbach, T.R., eds, Harper Collins
Academic, London, 327-41
Goldmark, P.G. (1972) Tomorrow we will Telecommute to our Jobs. The Futurist 6: 55-8
Hillman, J. (1993) Telelifestyles and the Flexicity: A European Study. European Foundation
for Living and Working Conditions, Dublin
Hudson, H.E., Parker, E.B. (1990) Information Gaps in Rural America: Telecommunications
Policies for Rural Development. Telecommunications Policy, 14: 193-205
Huws, U. (1993) Teleworking in Britain, Research Series No.18, Employment Department,
Sheffield
Huws, U., Honey, S., Morris, S. (1996) Teleworking and Rural Development. Rural Research
Report No.27, Rural Development Commission. Rural Development Commission, London
I1bery, 8., Clark, D., Berkeley, N., Goldman, I. (1995) "Telematics and Rural Development:
Evidence for a Survey of Small Businesses in The European Union." European Urban and
Regional Studies 2 (1): 55-68
McLuhan, H.M. (1964) Understanding Media: The Extension ofMan. Sphere, London
Parker, E.B., Hudson, H.E., Dillman, D.A, Roscoe, AD. (1989) Rural America in the
Information Age: Telecommunications Policy for Rural Development. The Aspen Institute,
Washington DC and University Press of America, Boston
Parker, E.B., Hudson, H.E. (with Dillman, D.A., Strover, S., Williams, F.) (1995) Electronic
Byways: State Policies for Rural Development Through Telecommunications. The Aspen
Institute, Washington DC
Qvortrup, L. (1989) The Nordic Telecottages: Community Teleservice Centres for Rural
Regions, Telecommunications Policy 13: 59-68
Qvortrup, L. (1994) Advanced Communications and Regional Development: the Case of
Finnmark. Report for the CEC ACCORDE Project
Telematics Innovation and the Development of Non-Metropolitan Areas 269
Richardson, D. (1996) The Internet and Rural Development: Recommendations for Strategy
and Activity: Final Report and Executive Summary. Report for the Food and Agriculture
Organization of the United Nations (http://www.fao.orglwaic... irect/CDDO/)
Richardson, R., Gillespie, A. (1996) Advanced Communications and Employment Creation in
Rural and Peripheral Regions: A Case Study of the Highlands and Islands of Scotland.
Annals of Regional Science, 30: 91-110
Salomon, I. (1988) Geographical Variations in Telecommunications Systems: the
Implications for Location of Activities. Transportation 14: 311-27
Telecottage Association (1997) Telecottage Map: British and Irish Telecottages. Teleworker 4
(5): 20-21
Toffier, A. (1980) The Third Wave. Pan, London
US Congress, Office of Technology Assessment (1995) The Technological Shaping of
Metropolitan America. OTA-ETI-643. us Government Printing Office, Washington DC
Vickerman, R. (1989) "Measuring Changes in Regional Competitiveness: the Effects of
International Infrastructure Investments." Annals ofRegional Science 23: 275-286
Williams, R. (1973) The Country and the City. Chatto and Windus, London
WOLF Consortium (1997) WWW and Internet Opportunities in the Less Favoured Regions.
Project Final Report, WOLF Consortium, Onyx Internet, Middlesborough
Chapter 13
13.1 INTRODUCTION
During the last few decades the considerable increase in mobility, evident
almost everywhere, has derived from a number of socio-economic factors.
Among these, the growth of income and increased involvement in various
urban activities (changes in lifestyle) have played a major role. In the larger
urban areas this has led to widespread congestion, which appears to be
growing steadily in all cities of the Western world. If this current trend in
mobility continues, within a few years the increase in congestion and
pollution in large cities is predicted to exceed the alarm threshold.
The growth in mobility demand which has induced this phenomenon has
been generated by the increasing need, or desire, to participate in a wide
range of activities. However, many of these activities could be carried out
either by means of physical proximity or telematic links. The supply (or
network) of the opportunity for 'exchange', in its various forms, may thus be
seen as consisting of two parallel networks: a physical communications
network and a telecommunications network. This suggests that the
development of telecommunications (TLC) could have a significant effect
upon both urban mobility demand and transportation supply. The issue has
already been raised within the EU, and in May 1994 the Council of Ministers
identified five salient objectives relating to the development of interactions
between transportation and telecommunications. These can be summarised
as follows:
272 Cascetta E. and Montella B.
As will be seen, the more serious the congestion and the higher the
relative transportation costs, the more intense these effects are likely to be.
Substitution effects are generally connected with changes in lifestyle, which
are reflected in mobility needs and in the performance of the transport
system, with effects in both the short and medium-long term. We are already
seeing the need for some physical trips being eliminated as a result of the
performance of certain activities 'at a distance', thanks to the possibilities
offered by telecommuting, teleconferencing, teleshopping and other tele-
assisted activities (telebanking, tele-entertainment, etc.). On the other hand,
the development of telecommunications can also create complementarity
effects, since the increased efficiency of the transportation system made
The Impact ofInnovations in Telecommunications and Information Systems 273
possible by the use of TLC systems tends to generates new mobility demand.
This increase in demand results from the reduction of time previously
consumed in physical trips. It is generally presumed that the 'extra' time
made available will be used to make trips for leisure or cultural purposes and
will therefore modify the whole pattern of mobility, since the trips are likely
to be oriented towards different parts of the city and transport modes.
The application of TLC systems can help to achieve greater efficiency in
both private and public transportation. As regards private transportation,
TLC functions enable the user to adapt his or her departure time and route
choice in accordance with real-time information on the current and predicted
state of the transport system. As far as public transportation is concerned, the
functions can be used by transport authorities to increase the efficiency of
their service and also affect users by improving trip quality. In the former
case, the information channelled through centralised vehicle control systems
has important spin-offs in optimising a series of corporate functions. It
allows more efficient management of operations, as well as of depots and
vehicle maintenance. As regards the users, information systems may be
static or dynamic, and be 'pre-trip', i.e. consulted prior to beginning the trip
(and hence possibly affecting the mode chosen as well as the itinerary) as
well as 'en route'. Increasingly, the information provided relates to all
possible transportation modes available for the proposed trip (including
public and private transportation systems, with possible interchange and
parking facilities). Thus the user is provided with a complete picture of the
alternatives and assisted in the choice of the most appropriate one.
A recent survey of research carried out in this field has been made by
Mokhtarian and Salomon (1997), who show that there have been numerous
investigations of the possible positive and negative effects of TLC
applications, examined at both national and international level. The authors
have drawn up a classification of the studies, dividing them into four
categories based on the scale (macro and micro) and the scope of the
coverage. It emerges however that although some effects have received
particular attention, especially user information systems for road and public
transport systems in Europe and the USA (probably due to their implications
for industry), there are still no experimental results concerning large scale
applications. Some suggestions about the possibility of integrating TLC
options into spatial choice structure have been presented by Ben-Akiva
(Ben-Akiva et aI1996).
The first part of this chapter describes the results of research on the main
substitution effects on mobility demand and the possible complementarity
effects on transport demand and supply. The second part provides a brief
update on the development of mathematical models and calculation
methodologies for evaluating the effects ofTLC on transport systems.
274 Cascetta E. and Montella B.
....
(0) Use transportation
- r-
or
telecommunications?
/
(G) Evaluation of outcome
of communication choice
.. .......
Figure J3. J An integrated framework for communication/transportation
interactions: the supply-demand model
The Impact ofInnovations in Telecommunications and Infonnation Systems 275
4. Organisational
characteristics
~,
of media perceptions
+ +
10. Formation of media preferences
in specific context
11. Situational
constraints
~r
Other indirect impacts on trip structure may occur due to changes in the
trip chains (i.e. sequences of trips starting and ending at home ). At present,
these chains are chiefly conditioned by the journey to work as the primary
purpose (i.e. work + shopping and/or accompanying another person). By
reducing work trips, these chains may be reduced in favour of simpler or
circular trips (home - activity - home). There could also be impacts on
modal choice, with cars being used more for the dispersed destinations and
less for trips to the central zones better served by public transportation.
Lastly, there may be effects on the urban structure itself, due for example to
a reduction in urban centrality. However, an in-depth assessment of this type
of effect, which may in some cases be undesirable, lies beyond the scope of
this chapter.
278 Cascetta E. and Montella B.
customer to be guided within the virtual store, told about special offers and
led to a specific department.
A paper by Batty (Batty 1997) explores many interesting issues
concerning teleshopping, especially as regards the physical interaction
between sales staff and customers. If we assume that the use of teleshopping
services will depend on the quality and user-friendliness of product
information supplied, then its use seem likely to increase.
The large scale development of teleshopping would presumably lead to a
net decrease in the number of physical shopping trips made, especially the
longer ones, as these could be substituted in both the market search and the
purchase phase. It would also be possible to eliminate many trips involving
the distribution of goods to sales outlets, as it would no longer be necessary
to display goods nor hold stocks at local wholesale or retail outlets. Indeed, it
is possible to imagine central warehouses where goods would be stocked for
direct delivery to purchasers.
There appear to be very good prospects for the development of such e-
commerce: with the increasing diffusion of the Internet, sound foundations
have been laid for the spread of teleshopping on a worldwide basis. It
therefore seems likely that a substantial change in the level of trip demand
for shopping (above all for trips outside the city) can be expected in the near
future. This would also affect their distribution, as sales would no longer be
tied to physical structures, but to their virtual counterparts. A simple shop of
a few square meters may well sell quantities of goods comparable to those
sold in a large department store.
As mentioned above, a highly significant implication for urban mobility
would be the transformation in the whole pattern of freight distribution, due
to the modifications in the distribution chain. There would be new nodes,
and different delivery routes used as goods could be transported directly
from the producer to the consumer.
Teleservices cover a wide range of services which can be performed by
means of TLC, including telebanking, tele-entertainment, telemedicine, etc.
Also in this sector, the development potential and consequent impact on
mobility are high. Undoubtedly, many services, especially those which are
quite tedious, are eminently suitable for substitution with teleservices. For
some such services, such as many banking and postal transactions,
telediagnostics, access to data banks and libraries, information and
certification services, etc. facilities already exist and are spreading rapidly.
Finally, as regards freight transportation, TLC may in some cases
completely substitute physical transport. Already a large amount of written
matter, whether single letters or whole documents, is now 'dispatched' bye-
mail rather than sent by post. In other cases, as explained in the example
above, the physical transport of goods may be considerably modified as an
The Impact of Innovations in Telecommunications and Information Systems 281
The model relates the number (f4) of physical trips (in a year) to number
of the telephone transactions (TLC trips) (f)) and contacts through TLC
terminals (f2)' This relationship, calibrated with a survey conducted in 1987,
produces substitution coefficients close to one for both types of distance
interaction. Efforts to arrive at a rigorous quantification of substitution
effects and the formulation of appropriate models are still being undertaken.
The table shows that substitution effects have the greatest relative weight
or impact for the work-work component, closely followed by services. It
seems probable that, with time, the latter will increase until it has the same
relative weight as the former. There are also substantial complementarity
effects in relaton to the work and service activities.
As regards the time horizon, information for the user can be of two types
(Ben Akiva et aI1991):
A systematic review is outside the scope of this chapter and the interested
reader is referred to Papageorgiou (Papageorgiou 1991) and the Proceedings
of the First World Congress on Applications of Transport Telematics and
Intelligent Vehicle-Highway Systems (Ertico 1995), which describe a
number of theoretical and empirical studies of such systems.
complex and the results less consolidated. This does not hold for models
simulating the complementarity effects of ATIS and ATCS systems. These
have been the subject of various research programs in Europe (DG XIII
framework programs such as DRIVE 1,2 and 3), the USA (IHVS) and Japan
(interest being motivated at least in part by the industrial interests involved).
It is worth pointing out (Shofer et at 1993) that decisions regarding the
implementation of ATIS systems should be based on the social as well as the
private benefits expected from the technology. But such benefits will depend
largely on how users respond to such sources of information.
Given the complexity of demand and supply interactions involved, ATIS
systems need to be designed, evaluated and operated on the basis of
sophisticated mathematical models able to simulate the performance of the
transport system. To adequately capture the possible impacts and to evaluate
the potential benefits of an ATIS in specific area, such models must have a
series of specific requisites and, in particular, be able to represent:
The effects on users of real-time TLC information about the state of the
transport system may be simulated on the basis of behavioural theory by
means of random utility models, which are now widely used in transport
systems modelling. Random utility models simulate the choice behaviour of
the transport system user, who is considered to be a 'rational' decision-
maker, i.e. an individual who follows three fundamental behavioural
hypotheses (Ben Akiva and Lerman 1985; Cascetta, 1990):
in making the choice, the user considers a set of available alternatives and
associates a perceived utility with each alternative;
the alternative chosen is that which has the greatest perceived utility;
the utility associated with each alternative can be described as a function
of attributes relating to the alternative (e.g. the level of service and the
decision-maker's socio-economic characteristics).
The Impact of Innovations in Telecommunications and Infonnation Systems 287
The modal choice model supplies the fraction pC (mlodsh) of trips of user
category c who use mode m to reach zone d from zone 0 for purpose s in
time slice h. The modal choice models used almost always have a
behavioural interpretation in practice, as their name would suggest. As
regards the functional form of modal choice models, the Multinomial Logit
is the one most frequently used:
(13.1)
where the systemic utilities for each mode m, Vm , are calculated by using the
level of service attributes of the trip between 0 and d in time slice h and the
socio-economic attributes of the market segment, or individual, c.
Increasing use is being made of Nested Logit modal choice models,
which can introduce correlation structures between perceived utilities of
various mode groups, such as individual modes and public modes, and/or
between various services within the same mode (e.g. slow and fast lines).
288 Cascetta E. and Montella B.
The route choice model supplies the number, p(k/mods), of trips using
route k, by mode m, to go from 0 to d for purpose s. The route choice models
most commonly used are all behavioural models, although hypotheses on
user behaviour differ for road route choice and trips using the public
transportation network. In the case of road networks, it is normally assumed
that the user chooses the route before starting the trip (preventive choice). In
this case, it is assumed that the variables affecting route choice are the
service attributes, known either historically or through real-time information.
However, the level of service information obtained is generally
'negative', or in other words represents a 'cost' for the driver (travel time,
monetary cost). Hence, reference will be made below to the perceived cost
1\
(13.2)
where Ck is the average perceived cost of route k, & is the random residual
and fo d m is the set of 'feasible' paths linking the origin/destination (o/d) pair
with mode m.
Various types of behavioural model may be used to obtain route choice
probabilities. The most commonly used random utility models are the Logit
and Probit. The former is specified as follows:
The Impact ofInnovations in Telecommunications and Information Systems 289
exp[-aC ]
p(k I mods) = k (13.3)
"L..,heiodm exp[- aCh ]
(13.4)
into account user behaviour, defined by the link flow vector. Stochastic
process models, on the other hand, model the evolution of the
demand/supply system as a random process. Conditions may be identified
for which there is a stationary probability distribution, allowing the averages
and dispersion of the main variables to be obtained, such as link flows
(Cascetta and Cantarella 1993).
the public transport system before the trip takes place. This knowledge
ensures that all the combinations 'stop at origin - run - stop at destination'
are seen by the user as primary alternatives. To evaluate the user's average
time saving with a pre-trip information system, generalised path costs were
evaluated. The results obtained show an average potential generalised cost
saving of approximately 25% in small-scale cases.
A different procedure has been defined for evaluating saving in user time
deriving from full real time knowledge of the transport system obtained the
trip, usually at the stop (Biggiero et al 1996). The time saving, in terms of
generalised costs, was calculated through a procedure for obtaining the
service level provided to users in two different scenarios with different levels
of information available. The attributes known to affect choice are total
travel time, waiting time, degree of crowding, and the number of transfers
made. When this information is provided by the ATIS, the user chooses the
set of service which minimise his or her generalised cost (Balogh and Smith
1992; Baneres 1990; Murnaghan 1993). Once the system state is known, two
mechanisms affecting choice come into play: the first involves a reduction in
the weight that the user attributes to waiting time, insofar as it no longer
constitutes an uncertainty (in fact, if known a priori, it may be used
differently); the second involves a trade-off between waiting time and travel
time, i.e. a longer wait at the stop may be preferred in order to travel in a
faster vehicle. Overall, generalised cost reductions of 10-12% were
achieved.
In private transport (car networks), the time-saving effects due to the
introduction of TLC technologies providing drivers with real time
information have been studied since the mid 1980s (Cascetta and Cantarella
1991). An example of an evaluation model which includes driver response to
information, network flow variations and new used paths, is that developed
by Jayakrishnan et al (1994) and known by the acronym DYNASMART
(Dynamic Network Assignment Simulation Model for Advanced Road
Telematics) (Peeta and Mahamassani 1995; Hu and Mahamassani 1995).
This model has the following features:
the ability to model the route choice of drivers with and without access to
information;
responsiveness to dynamic O-D information available to the controller;
the ability to track the locations of all drivers;
the ability to predict the travel time based on the assignment results;
the ability to model real traffic control strategies in the network;
the ability to comprehensively model both freeway and surface street
traffic.
294 Cascetta E. and Montella B.
13.10 CONCLUSIONS
REFERENCES
Anderson, T. (1993) Real-Time Passenger Information for Transit Systems: Proposal for the
Jubilee Line Extension. PTRC Summer Annual Meeting, 21
Arnott R. et al (1991) Does Providing Information to Drivers Reduce Traffic Congestion?
Transportation Research 25A (5): 309-18
Balogh, S., Smith, R. (1992) Real-time Bus Information - The London Transport Route 18
Demonstration. PTRC Summer Annual Meeting, 20
Baneres,1. (1990) Automatic Vehicle Monitoring System for the Bus Network of Barcelona.
URBANI A Congress, Padova Italy
Batty, M. (1997) The Retail Revolution. Environment and Planning B, Planning and Design
24: 1-2
Ben Akiva, M.E., Lerman, S.R. (1985) Discrete Choice Analysis. MIT Press, Cambridge
Mass
The Impact ofInnovations in Telecommunications and Information Systems 295
Ben-Akiva, M., de Palma, A., Kaysi, I. (\991) Dynamic Network Models and Driver
Information Systems. Transportation Research 25A (5) 251-66
Ben-Akiva, M., Bowman, J., Dinesh, G., (\996) Travel Demand Model System for the
Information Era. Transportation 23: 241-66
Bennison, DJ. (1988) Transport/Telecommunication Interactions: Empirical Evidence from a
Videoconferencing Field Trial in the United Kingdom. Transportation Research 22A (4)
291-300
Bernardino, A., Ben Akiva, M., Salomon, I. (1993) Stated Preference Approach to Modeling
the Adoption of Telecommuting. Transportation Research Record 1413: 22-30
Biggiero, L., Di Gangi, M., Montella, B. (1996) A Methodology for Estimating Effects on
Users of Introducing TLC Into Urban Public Road Transport. Applications of Advanced
Technologies in Transportation Engineering, ASCE New York
Boyce, D.E. (1988) Route Guidance Systems for Improving Urban Travel and Location
Choice. Transportation Research 22A (4) 275-81
Button, K., Maggi, R. (1994) Videoconferencing and its Implications for Transport: An
Anglo-Swiss Perspective. Transport Reviews 15(1): 59-75
Cantarella, G.E., Cascetta, E. (1996) Stochastic Assignment to Transportation Networks:
Models and Algorithms. CTR XXV, Montreal 10-11 October
Cascetta, E. (1990) Metodi quantitativi per la pianificazione dei sistemi di trasporto CEDAM,
Padova Italy
Cascetta, E., Cantarella, G.E. (1991) A Day To Day and Within Dynamic Stochastic
Assignment Model. Transportation Research 25A (5): 277-91
Cascetta, E., Cantarella, G.E. (\ 993) Modelling Dynamics in Transportation Networks: State
of the Art and Future Developments. Journal ofSimulation Practice and Theory I: 65-91
Cascetta, E., Nuzzolo, A. (\986) Uno schema comportamentale per la modellizzazione delle
scelte di percorso nelle reti di trasporto pubblico urbano. Proceedings of the 4th PFT CNR
National Congress, Turin, Italy
Cascetta, E., Nuzzolo, A., Biggiero, L. (1992) Analysis and Modelling of Commuters'
Departure Time and Route Choice in Urban Networks. Proceedings of 2nd International
Seminar on Urban Traffic Networks, Capri 5-8 July
Divieti L. (1995) "Trasporti e telecomunicazioni: potenziali interazioni tra spostamenti fisici e
spostamenti immateriali." In Sviluppi della ricerca sui sistemi di trasporto, SlOT, Franco
Angeli, Milan Italy, 185-201
ERTICO (1995) Proceedings of the First World Congress on Applications of Transport
Telematics and Intelligent Vehicle-Highway Systems
Florian, M., Spiess, H. (1989) Optimal Strategies: a New Assigment Model for Transit
Network. Transportation Research 5: 152-74
Garrison, W.L., Deakin, E. (1988) Travel, Work, and Telecommunications: a View of
Electronics Revolution and its Potential Impacts. Transportation Research 22A (4): 239-
45
Hamer, R, Kroes, E., van Ooststroom (1991) Teleworking in the Netherlands: an Evaluation
of Changes in Travel Behaviour. Transportation 18: 365-82
Hamer, R, Kroes, E., van Ooststroom, Kockelkoren, M. (1992) Teleworking in the
Netherlands: Evaluation of Changes in Travel Behaviour - Further Results. Transportation
Research Record 1357: 82-9
Hansen, M., Huong, Y. (1997) Road Supply and Traffic in California Urban Areas.
Transportation Research A 31 (3): 215-18
Henderson, D.K., Mokhtarian, P.L. (1996) Impacts of Center-Based Telecommuting on
Travel and Emissions: Analysis ofthe Puget Sound Demonstration Project. Transportation
Research 01(1): 25-49
296 Cascetta E. and Montella B.
Hu, P.S., Young, 1. (1992) Nationwide Personal Transportation Survey: Summary of Travel
Trends. Report FHWA-PL-92-027. Federal Highway Administration Office of Highway
Information Management, Washington DC, March
Hu, T., Mahamassani, H. (1995) Evolution of Network Flows Under Real Time Information:
A Day To Day Dynamic Simulation-Assignment Framework. Transportation Research
Record, 24-31
Jayakrishnan, R., Hu, T., Mahamassani, H. (1994) An Evaluation Tool for Advanced Traffic
Information and Management Systems in Urban Networks. Transportation Research 2C
(3): 129-47
Kitamura, R., Nilles, 1., Fleming, D., Conroy, P. (1990) Telecommuting as a Transportation
Planning Measure: Initial Results of State of California Pilot Project. Transportation
Research Record 1285: 98-104
Koenig, B. E., Henderson, D. K., Mokhtarian, P.L. (1996) The Travel and Emissions Impacts
of Telecommuting for the State of California Telecommuting Pilot Project. Transportation
Research 4C (1): 13-32
La Bella, A. Silvestrelli, M.A. (1991) Telecomunicazioni e trasporti nella citta del futuro.
Proceedings of the 5th PFT CNR National Congress, Rome
Mahamassani, H. S., Jayakrishnan, R. (1991) System Performance and User Response Under
Real-Time Information in a Congested Traffic Corridor. Transportation Research 25A (5):
293-307
Mahamassani, H., Yen, 1., Herman, R., Sullivan, M. (1994) Employee Attitudes and Stated
Preference Toward Telecommuting: An Exploratory Analysis. Transportation Research
Record 1413
Mokhtarian, P.L. (1991) Telecommuting and Travel: State of the Practice, State of the Art.
Transportation, 18(4): 319-42
Mokhtarian, P.L. (1988) An Empirical Evaluation of the Travel Impacts of Teleconferencing.
Transportation Research, 22A (4): 283-88
Mokhtarian P.L. (1997) Emerging Travel Patterns: Do Telecommunications Make a
Difference? 8th Meeting of the International Association for Travel Behavior Research,
Austin Texas, September
Mohtarian, P.L., Salomon, I. (1997) Modelling the Desire to Telecommute. Transportation
Research 31A: 35-50
Montella, B. (1996) A Methodology for Estimating Effects on Users ofIntroducing TLC into
Urban Public Road Transport. Applications of Advanced Technologies in Transportation
Engineering, ASCE, New York
Moore, A., Jovanis, P.P. (1988) Modelling Media Choice in Business Organizations:
Implications for Analyzing Telecommunications - Transportation Interactions.
Transportation Research 22A (4): 257-73
Murnaghan, P. (1993) Real Time Information at Bus Stops Romance Project Southampton.
PTRC Summer Annual Meeting, 21
Nguyen, S., Pallottino, S. (1986) Assegnamento dei passeggeri ad un sistema di linee urbane:
determinazione degli ipercammini minimi. Ricerca Operativa 38: 207-30
Nilles, 1. M. (1988) Traffic Reduction by Telecommuting: A Status Review and Selected
Bibliography. Transportation Research 22A( 4): 301-17
Papageorgiou, M. (1991) Traffic & Transportation Systems. Pergamon Press, New York
Peeta, S. Mahamassani, H. (1995) Multiple User Classes Real Time Traffic Assignment for
Online Operations: A Rolling Horizon Solution Framework. Transportation Research 3C
(2): 83-98
Salomon, I. (1985) Telecommunications and Travel: Substitution or Modified Mobility?
Journal a/Transport Economics and Policy 19(3): 219-35
The Impact of Innovations in Telecommunications and Information Systems 297
Salomon, I., Koppelman, F. (1988) A Framework for Studying Teleshopping Versus Store
Shopping. Transportation Research 22A(4): 247-55
Salomon, I., Schneider, H.N., Schofer, J.L. (1991) Is Telecommunicating Cheaper than
Travel? An Examination of Interaction Costs in a Business Setting. Transportation 18:
291-318
Schofer, J. Khattak, A., Koppelman, F. (1993) Behavioral Issues in the Design and Evaluation
of Advanced Traveller Information Systems. Transportation Research 1C (2): 107-17
Selvanathan, E.A., Selvanathan, S. (1994) The Demand for Transport and Communication in
the United Kindom and Australia. Transportation Research 28B (1): 1-9
Small, K.A. (1982) The Scheduling of Consumer Activities: Work Trips. The American
Economic Review 73 (3): 467-79
US Department of Transportation (1993) Transportation Implications of Telecommuting US
Government Printing Office
Subject Index
List of Contributors