Frank Bruinsma, Eric Pels, Piet Rietveld, Hugo Priemus, Bert Van Wee (Auth.), Dr. Frank Bruinsma, Dr. Eric Pels, Prof. Dr. Piet Rietveld, Prof. Dr. Hugo Priemus, Prof. Dr. Bert Van Wee (Eds.)-Railway

Railway Development
Impacts on Urban Dynamics

Frank Bruinsma Eric Pels
Hugo Priemus Piet Rietveld
Bert van Wee
Editors
Railway Development
Impacts on Urban Dynamics
Physica-Verlag
A Springer Company
Dr. Frank Bruinsma Prof. Dr. Hugo Priemus
Dr. Eric Pels Prof. Dr. Bert van Wee
Prof. Dr. Piet Rietveld Delft University of Technology
Free University of Amsterdam Faculty of Technology,
Department of Spatial Economics Policy and Management
De Boelelaan 1105 P.O. Box 5015
1081 HV Amsterdam 2600 GA Delft
The Netherlands The Netherlands
fbruinsma@feweb.vu.nl h.priemus@tbm.tudelft.nl
apels@feweb.vu.nl g.p.vanwee@tbm.tudelft.nl
prietveld@feweb.vu.nl
ISBN 978-3-7908-1971-7 e-ISBN 978-3-7908-1972-4
DOI 10.1007/978-3-7908-1972-4
Library of Congress Control Number: 2007936545
c 2008 Physica-Verlag Heidelberg

This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting,
reproduction on microfilm or in any other way, and storage in data banks. Duplication of this publication
or parts thereof is permitted only under the provisions of the German Copyright Law of September
9, 1965, in its current version, and permission for use must always be obtained from Physica-Verlag.
Violations are liable to prosecution under the German Copyright Law.
The use of general descriptive names, registered names, trademarks, etc. in this publication does not imply,
even in the absence of a specific statement, that such names are exempt from the relevant protective laws
and regulations and therefore free for general use.
Production: LE-TEX Jelonek, Schmidt & Vckler GbR, Leipzig
Cover-design: WMX Design GmbH, Heidelberg
Printed on acid-free paper
987654321
springer.com
Preface
The revitalisation of railway station areas has been a hot issue in policy
making in Europe in recent years. The introduction of two new railway
systems - the high speed train and light rail systems - has pushed forward
the redevelopment of not only railway stations themselves, but also of their
- in many cases deteriorated - direct surroundings. Examples of revitalisa-
tion of railway station areas can be found throughout Europe, for example
in countries like England (Liverpool), France (Euralille), Germany (Berlin)
and the Netherlands (Amsterdam South Axis). The dynamics of revitalisa-
tion of railway station areas can best be understood by combining the in-
sights of several disciplines; economics, and spatial sciences being the
most important disciplines. This book therefore addresses the subject from
a multi-disciplinary perspective.
Railway station area redevelopment projects try to cope with the in-
creased lack of interoperability and interconnectivity between infrastruc-
tures and boost the local/regional economy by creating a new high quality
multifunctional urban landscape at the same time. At present, the initial
outcomes with the transformation of deteriorated railway station areas into
modern mixed office, shopping and residential areas stems hopeful, but
what are the prospects of those redeveloped areas?
Most of the papers contained in this volume were presented at the spe-
cial session Railway stations and urban dynamics of the 45th conference
of the European Regional Science Association (ERSA) held in Amster-
dam, the Netherlands during August 2005. There was a special reason for
this session during the conference. In the Netherlands several scientific re-
search programs, co-funded by the Dutch government, were directed to-
wards this topic of railway station development and urban dynamics.
These programs were Connect-NWO program Railway stations: inter-
faces between railway network developments and urban dynamics, Tran-
sumo (TRANsition SUstainable MObility) and the BSIK programs Next
generation infrastructures and Innovative Land Use. In those programs
researchers of several universities participated to assure a multidisciplinary
perspective. The conference offered a ideal platform to disseminate the re-
sults of the research programs to a broader audience. Moreover, the special
session was open for presentations from authors not involved in one of the
vi Preface
programs. By this the scope of the research extended from a Dutch to a

European perspective. Financial support of the programs made it possible
to organise the special session and to edit this manuscript. This support is
gratefully acknowledged. Sue Davis has assisted in improving the lan-
guage of non-English speaking authors. We thank her for efficient and
conscientious work.
Amsterdam July 2007
Frank Bruinsma
Eric Pels
Hugo Priemus
Piet Rietveld
Bert van Wee
Contents
Preface
1 The impact of railway development on urban 1

dynamics
Frank Bruinsma, Eric Pels, Hugo Priemus,
Piet Rietveld and Bert van Wee
Part A: Setting the scene
2 Urban dynamics and transport infrastructure: 15

Towards greater synergy
Hugo Priemus
3 Station areas as nodes and places in urban 35

networks: An analytical tool and alternative
development strategies
Luca Bertolini
4 HST stations and urban dynamics: Experiences 59

from four European cities
Peter Pol
5 What makes a city: Urban quality in Euralille, 79

Amsterdam South Axis and Rotterdam Central
Jan Jacob Trip
6 New Key Projects for station redevelopment in 101

the Netherlands
Stan Majoor and Dick Schuiling
7 A multidisciplinary approach of railway station 125

development: A case study of s-Hertogenbosch
Gert-Joost Peek and Erik Louw
viii Contents
Part B: Evaluation studies
8 Ex ante evaluation of railway station 147

development projects: Issues still to be solved
Piet Rietveld and Bert van Wee
9 Multicriteria analysis of a high-speed railway 171

station area development project
Ron Vreeker
10 Cost-benefit analysis of railway station area 191

development: The case of Amsterdam South Axis
Carel Eijgenraam and Ioulia Ossokina
11 Measuring the WTP for shopping facilities 213

around railway stations
Thomas de Graaff and Caroline Rodenburg
Part C: High-speed rail and urban dynamics
12 The impact of high-speed railway developments 237

on office locations: A scenario study approach
Jasper Willigers
13 The effect of railway stations on office space 265

rent levels: The implication of HSL South in
station Amsterdam South Axis
Ghebreegziabiher Debrezion and Jasper Willigers
14 Regional high-speed trains on the Svealand line: 295

Evaluation of effects
Oskar Frodh and Bo-Lennart Nelldal
15 Rail pricing and the supply of complementary 315

commercial goods
Eric Pels, Erik Louw and Jan Jacob Trip
Contents ix
Part D: Light rail and urban dynamics
16 Rail system development and urban 337

transformations: Towards a spatial decision
support system
Enrica Papa, Francesca Pagliara and Luca Bertolini
17 Rail-transit and real estate values in a polycentric 359

city: A theoretic simulation approach
Daniel Gat
18 Land use variables in trip generation models: 385

The case of the light rail transit in Tel Aviv
Avigail Ferdman, Daniel Shefer and Shlomo Bekhor
Contributors 415
1 The impact of railway development on urban
dynamics
Frank Bruinsma, Eric Pels and Piet Rietveld
Department of Spatial Economics, Free University, Amsterdam,

the Netherlands
Hugo Priemus and Bert van Wee
Faculty of Technology, Policy and Management, Delft University of

Technology, the Netherlands
1.1 Introduction
This book addresses the role of railways in urban development. Our central
aim is to inquire into how the renaissance of railways since the end of the
20th century (especially the development of high-speed rail and light rail
links) will affect European cities. The analyses are carried out with special
attention given to the broader institutional environment of the railway sys-
tem, including the shift toward privatised railway companies, internation-
alisation, the occurrence of market and government failures in land mar-
kets, and private-public partnerships in the development of railway station
areas.
Spatial developments over the past decades indicate a trend towards a
more diffuse settlement pattern usually defined as urban sprawl. In a num-
ber of countries national policies were set out to explicitly counter sprawl,
but in most, sprawl has become pervasive (see Brueckner 2000 for a dis-
cussion on the seriousness of urban sprawl). An important instrument used
to counter urban sprawl has been the development of new towns with ade-
quate rail connections to major cities. This has indeed led to the focussed
development of residential areas in the regions, but it had a sometimes
negative impact on the growth of the existing larger cities.
2 Frank Bruinsma, Eric Pels, Hugo Priemus, Piet Rietveld and Bert van Wee
A new turn in the development of the railway system is observed at the

end of the 20th century. In many countries large investments are made to
improve the quality of railway systems, in particular the construction of
high-speed rail connections (Vickerman 1996; Haynes 1997). Congestion
on the road network makes rail more competitive. In addition, capacity
shortages in the aviation network lead to an interest in high-speed rail links
as an alternative to feeder flights, which subsequently also reinforces the
position of railways for international trips.
These railway developments offer opportunities for the larger cities ex-
periencing periods of decline. Railway related issues, be it the develop-
ment of terminals for high-speed rail, construction of high quality office
areas near railway stations, or the introduction of light rail systems, have
become important themes in policies to revitalise these cities (cf. Bertolini
and Spit 1998; van den Berg and Pol 1999). The essence of these plans is
that railway stations are not considered merely as nodes where people
change from one vehicle to another, but also as places where spatial con-
centrations of high value activity are recognised as having a positive im-
pact on cities. The mobilisation of resources for these plans is, however, an
important problem. The development of real estate near stations is increas-
ingly seen as a negotiation process in which both public and private actors
play certain roles; this is especially the case in the development of strategic
real estate projects to revitalise central urban areas (Spaans 2000).
Public authorities often play two roles: the classic public role (destina-
tion plan, urban planning, building permit) and more and more, the role of
entrepreneurial partner in the development of real estate. Both roles essen-
tially differ and are seldom distinguished in a transparent way. Although
the dynamics of the railway sector has improved its role, in most transport
markets it is still limited, so that investments in rail improvements may be
considered as risky. The problem is the extent to which it is possible to use
the transport function of railway nodes such that positive spill-overs can be
realised for development opportunities in the cities concerned.
The central question of this book therefore is how the renaissance of
railways since the end of the 20th century (especially the development of
high-speed rail links and light rail systems), interacted with urban dynam-
ics.
1.2 Historical review
According to Sjoberg (1960), pre-industrial cities in various areas of the

world and at different times had much in common, despite their dissimi-
The impact of railway development on urban dynamics 3
larities in location, structure and architecture. They were relatively small

especially compared with the metropolis of today and compact, generally
built no more than three or four storeys high. The pre-industrial transport
system left its mark: the town became the domain of pedestrians, carts,
pets, beasts of burden, and horse-riders. Where towns were sited on wa-
terways, from an early date, ships were used to move people and goods.
Ships and carts carried mainly goods; not until later did we see the advent
of the horse-drawn coach, which was designed to carry the relatively well-
off. During the Middle Ages towns and transport infrastructures were well
integrated: while there were some differences between through traffic and
local traffic, the variations in street layouts and speeds were not substan-
tial. Horses, donkeys and other beasts of burden were usually stabled on
the ground floor of houses. The residential and work functions of buildings
were still integrated, as was also customary on farms in rural areas (Vance,
1966). The speed of traffic was very slow, both within and outside towns.
On long journeys travellers and animals needed adequate rest periods in
order to replenish with food and drink. Towns often developed at the inter-
sections of a number of transport routes, perhaps only roads, as with many,
but also roads and waterways.
The rapidly growing railway network later played a vital role in urban
growth. Smith (1998) argues that the arrival of the railways was the great-
est instrument of change in the layout and nature of the British 19th century
town. One reason for this was that the railway companies had, atypically,
been granted permission to acquire property by compulsory purchase. In
Germany too, the routes of the railways to a large extent determined the
development of many cities (Mikus 1966). Turnock (1998) notes that, as a
result of industrialisation, the railway played a key role in the process of
geographical change during the 19th century. In 1848 a Royal Commission
ordered that railways no longer be allowed to penetrate Central London
(Smith 1998); consequently, from 1863, the Metropolitan Railway pro-
vided a transport service within the capital, but under ground, running be-
low existing roads for much of its length.
Stations were often (but not always) sited just outside town centres to
avoid overly disrupting the urban layout. Routes of railways that cut
through built-up areas were sited almost without exception in areas of
cheap housing where the lowest paid workers lived. As the amount of traf-
fic increased (especially road, but later air traffic), urban dwellers found it
to be more of a nuisance than a convenience.
Urban conurbations burgeoned and traffic jams accumulated, especially
in and around towns. Many areas of European cities were demolished in
the 1960s to make way for motorways. This neither solved the parking
problem nor eased the congestion on access roads; moreover, citizens be-
came increasingly aware of the various harmful external effects of road

traffic, such as noise, vibration and air pollution (Banister 2002). In the
1970s physical planners shifted their attention away from the stations in
city centres towards suburban railway stations. At that time employment in
the service sector, in particular business services, was increasing rapidly at
the expense of employment in industry. Suburban stations were to become
the focal points of employment for the service sector. The direct surround-
ings of those suburban stations were suitable as office locations, and thus
were expected to relieve traffic pressure on the inner cities. However, the
service sector was not particularly charmed by suburban railway station
locations that lacked parking space; they preferred locations near and well
connected to urban ring roads with large areas for parking (Korteweg
1992).
1.3 Current issues
As Marchetti (1987) has shown, new modalities and new types of infra-
structure were, over time, added in waves. In some cases the share of the
modal split accounted for by traditional modes of transport fell dramati-
cally, as was the case with the towed passenger boat and the horse-drawn
coach. The more the main highways outside towns were absorbed into ring
roads, the farther away airports were sited from city centres, and the more
railway stations became hemmed in by urban expansion (thus giving them
an increasingly central position in the town), the more difficult it became
to transfer from one mode to another. The lack of interoperability and in-
terconnectivity of infrastructures increased: this was due partly to a lack of
coordination between urban planners (physical planning and property de-
velopment) and infrastructure developers (including transport policy), for
which different central and decentralised government departments were re-
sponsible. A revival of the attention given to central stations arose during
the 1980s. On the one hand, the existing central railway station had dete-
riorated over the past decade and needed to be revitalised. The renewed in-
terest in the central station was, on the other hand, boosted by the introduc-
tion of two new rail systems high-speed rail and light rail that partly
utilised existing rail and station infrastructure. These two new rail systems
were added towards the existing rail services or better, respectively up-
graded and downgraded regular rail service. The development of stations
where travellers can transfer between a mix of regular train, high-speed
train, light rail, regional and urban buses offered new opportunities to
combine the redevelopment of the railway station and the whole station
area at once. At present, the initial outcomes with the transformation of de-
teriorated railway station areas into modern mixed office, shopping and
residential areas, seems to have been successful. This revival of railway
stations and urban dynamics is the central topic of our book.
1.4 Future perspectives
What are the likely prospects of those redeveloped station areas? In our
view, future prospects seem quite positive; ongoing changes in the sectoral
structure of the economy from industry towards office sector activities
support the redevelopment of station areas into high density multifunc-
tional land use areas in which work, residential and shopping functions
combine with transport functions. Often deteriorated and mainly residen-
tial, railway station areas are being transformed into modern places to
work, live and consume. Business services such as financial and insurance
services in particular, have recently relocated towards redeveloping station
areas and tend to attract other business services, including lawyers and real
estate agents. Regardless of the opportunities that modern ICT facilities of-
fer, such as the internet and teleconferencing, in the knowledge based
economy there is an apparent ever increasing desire for face-to-face con-
tact between managers in order to exchange ideas and build relationships.
Not only have firms shown a growing interest in urban areas, but also resi-
dents, attracted by the new boost in the quality of urban living environment
conditions, seem to be re-urbanising. The high quality dwellings recently
developed within the cities, cultural amenities, and public spaces espe-
cially attracts high income, one-person households. There seems in recent
years to be an improving match between the needs of new lifestyles and
the evolving urban quality of life conditions. To support such forms of ur-
ban dynamics, well connected multimodal transport systems need to be of-
fered. International, (air transport and high-speed rail), national (heavy rail
and highways) and regional/local (light rail, car and public transport)
transport networks need to offer interconnectivity. High-speed rail and
light rail will play an important role in connecting this multi-layered inte-
grated transport network. Stations clearly offer opportunities to reinforce
urban dynamics within the urban fabric.
1.5 Outline of the book
The book is divided into four parts. The first part comprises a multidisci-
plinary introduction to the relation between railway development and ur-
ban dynamics. Part B is dedicated to evaluation methodologies used to
measure the impact of railway development on the urban economy. The
third part provides analyses which measure the spatial and economic im-
pacts of high-speed rail on urban development. And finally, part four is
devoted to a similar treatment of light rail systems
1.5.1 Part A: Setting the scene
In this part of the book the context of the complex interaction between
railway station development and urban dynamics is provided from a mul-
tidisciplinary perspective. Engineers, (transport) planners, sociologists,
(urban) economists and scientists in public administration all contribute to
the knowledge accumulation on the interrelationship between railway de-
velopment and urban dynamics. What can we learn from their multidisci-
plinary scientific efforts?
Before entering into greater detail, Priemus provides a general introduc-
tion on the relationship between infrastructure development and urban dy-
namics in chapter 2. His main theme is the (lack of) synergy between ur-
ban development and the development of transport networks in Europe,
and the role played in this by various types of policies. The question is
whether this synergy has grown or declined in and around European cities
in recent decades, and what policy tools are available and how they could
be applied to increase synergy.
After this introduction towards the main topic of the book the bivalent
character of stations is discussed. They are nodes in transportation net-
works as well as city places. In chapter 3 Bertolini depicts stations as
nodes in urban networks. His node-place model provides an analytical
framework to penetrate the dynamics of station area development. He in-
troduces two ideal typical approaches, a coordinated development and a
reconnecting development approach to reorient the development of the
urban system towards the railway network. He concludes that the specific
form of development is less important than the degree of consistency ap-
plied between land use and transport policies.
Connection to the High-Speed Train (HST) network can be a powerful
incentive to invest in the redevelopment of station areas. High order eco-
nomic activities can, in particular, be attracted to the direct vicinity of HST
stations. Whether this actually occurs depends on how cities react to this
opportunity and on strengthening weak location factors of the station areas.

Depending on the type of cities, other investment strategies and strategic
networks can be expected and are observable. Efforts to develop station
areas are complicated by the large number of public and private actors re-
quired to be involved. Therefore, a precondition for a balanced and well-
timed station development is sufficient organising capacity among the ac-
tors. Pol studies this organising capacity of high-speed rail station areas in
chapter 4. Results are supported by experiences of four cases from the cit-
ies Amsterdam, Munich, Lille, and Rotterdam.
Urban quality is of particular importance since high-speed rail stations
are most commonly developed as national or even international business
centres able to attract major office firms. To what extent does a high-speed
rail connection contribute to the image of an urban area? In chapter 5 Trip
deals with the issue of urban quality of high-speed rail stations areas.
Based on Floridas ideas on quality of place, the developments of the high-
speed station areas of Lille, Amsterdam and Rotterdam is discussed. Apart
from the high-speed connection, he analyses the quality of the urban struc-
ture, architecture, functional diversity, quality of functions, and public
space. He concludes that a high-speed rail connection contributes to the
image of a development project, even more so when it is located in a city
with no great reputation of its own.
Given the large number of public and private parties involved and the
long time span of redevelopment of station areas the continuation in
the decision-making process is a critical success factor. Majoor and Schuil-
ing discuss the decision-making process of high-speed rail stations in the
Netherlands in chapter 6. They define two important policy challenges for
railway station area development: the creation of integrated spaces (con-
tent dimension) and the organization of collective action in a situation of
fragmentation (process dimension). Based on an analysis of six high-speed
railway station development projects, they conclude that the challenge for
station development is not a lack of potential, good planning intentions or
imaginative ideas, but lies more in decision-making and the institutional
fragmentation that has to be conquered.
To round out the multidisciplinary part of the book, Peek and Louw in
chapter 7 examine the redevelopment of station (areas) from the perspec-
tive of engineers and architects, transportation and traffic planners, soci-
ologists, and urban economists, respectively, to identify the bivalent char-
acter of stations. To apply such a multidisciplinary approach they extend
the functional place-node model of Bertolini (chapter 3) by the spatial lo-
cation-network dimension. In their model they combine the content-based
and process management aspects to provide a broader perspective on the
development potential in combination with the conditions determining

whether the potential will be realised.
1.5.2 Part B: Evaluation studies
In the first part of the book insights are provided on a number of perspec-
tives from which the interrelationship between railway station develop-
ment and urban dynamics can be studied. Given the large volumes of pub-
lic investments needed in the case of railway station development projects,
major projects will be decided only after a thorough ex-ante evaluation of
each project. Thus in part B we concentrate on evaluation methodologies.
Rietveld and van Wee set the stage in chapter 8 with a discussion about is-
sues still to be resolved in current ex-ante evaluation methodologies. They
first inquire into the issue of the extent to which travel time savings or
travel cost reductions due to infrastructure investments completely capture
the welfare gains. In addition, the relationship between transport cost
changes and land prices, experiential benefits and place value benefits of
railway stations is discussed. They conclude that the ignored benefit cate-
gories might be substantial and can significantly influence the cost-benefit
ratios of rail projects.
In chapters 9 (Vreeker) and 10 (Eijgenraam and Ossokina) the multi-
criteria analysis (MCA) and cost-benefit analysis (CBA) methodologies
are discussed, respectively. To illustrate the evaluation techniques in both
chapters, the Amsterdam South Axis high-speed railway station develop-
ment project is taken as a case study. In chapter 9 Vreeker uses multi crite-
ria decision analysis (MCDA) to evaluate the decision problem from the
perspective of different actors involved, and analyses which aspects they
deem important. Based on the established perspectives, he identifies
groups of actors for which interests seem to cluster. In chapter 10 Eijgen-
raam and Ossokina apply CBA techniques to evaluate the welfare effects
of combining different land uses by comparing the benefits of multifunc-
tional land use with the costs that their creation may entail. They conclude
that, due to a lack of knowledge on effects of changes in the pattern of land
uses, the outcomes of CBA evaluations of such projects have to be inter-
preted with rather large uncertainty margins.
One of the disadvantages of CBA is that, in the case of absence of mar-
kets, it is problematical to translate impacts into monetary terms. Consider-
ing absence of markets one could think, for instance, of environmental im-
pacts or the impact of multifunctional land use. In the case of MCA it is
difficult and disputable to assign weights to such impacts. In chapter
11 De Graaff and Rodenburg address this problem by applying willingness
to pay techniques for the valuation of various shopping facilities at the

Amsterdam South Axis high-speed railway station development project.
We have already mentioned that, instead of a new transport modality or

transport network, the railway network was revitalised by the introduction
of two rail systems to the regular rail system. On the one hand there was an
up-scaling of the regular rail services by the introduction of the high-speed
train with a national and even international service area, and on the other
hand there was a down sizing by the introduction of light rail with an ur-
ban/regional service area. The remaining two parts of the book will deal
with the impacts of both rail systems on urban dynamics. Part C is devoted
to high-speed rail and Part D to light rail.
1.5.3 Part C: High-speed rail and urban dynamics
High-speed rail networks add a strong international dimension to railway

stations. As a consequence the direct surroundings of these stations were
also redeveloped as locations for international business firms. High-speed
railway stations were intended to become attractive areas with high quality
offices, services and residential areas that would compete with other top
European locations to attract international head offices. Willigers re-
searches this issue in chapter 12 at the level of cities connected to the high-
speed rail network, and at the level of sites near high-speed rail stations by
a model for determining potential accessibility and a model for the location
choices of office decision-makers for such locations. He uses a scenario
study approach to clarify how the attractiveness of cities and sites depends
on the choices made for cities and stations connected to the high-speed rail
network, the level of train services and the fare system applied. He con-
cludes that high-speed rail has a considerable effect on the attractiveness of
locations, especially at the intraregional level.
In chapter 13 Debrezion and Willigers also focus on the impact of ac-
cessibility. The value of proximity to a railway station is capitalised on the
value of properties. However, the capitalisation process can take different
forms across different property types. They calculate the impacts of dis-
tance to the high-speed rail station on office space rent levels by means of
a hedonic spatial autocorrelation model. The study shows the relevance of
railway accessibility as accounted for by proximity and rail service quality
measures for office space rents. However, the range at which railway ac-
cessibility will have a meaningful effect on the rent levels is limited to a
reasonable walking distance. This might be expected due to the fact that
office locations in general lay on the departure part of rail trips.
Apart from impacts on individual high-speed railway stations, the net-

work effect is important. Fridh and Nelldal study the mobility impacts of
the construction of a regional high-speed railway line, the Svealand line,
between Stockholm and the interior of Sweden on the regional economic
development in chapter 14. They applied ex-ante and ex-post case study
research based on questionnaires and interviews, to investigate the impact
of the changes in supply of public transport services on regional imbal-
ances in the housing and labour market in the Stockholm-Mlaren region.
Due to the reduction in travel time, the opening of the Svealand line con-
tributed to the regional integration for daily commuting over medium dis-
tances.
Finally, an organisational issue is dealt with in chapter 15 by Pels, Louw
and Trip, in which they model the demand for railway services as well as
commercial activities. It is assumed that demand for railway services and
commercial activities are correlated. They show the welfare economic im-
pact of different organizational structures of the rail sector. For instance,
when the railway company also operates the commercial activity, overall
welfare is high, compared to the case of separate operations of railway ser-
vices and commercial activities. Furthermore, competition between loca-
tion(s), for instance the independent development of stations and surround-
ing areas, will result in welfare losses.
1.5.4 Part D: Light rail and urban dynamics
Light rail systems provide relatively fast and frequent services within an
urban agglomeration and have at most a regional coverage. Due to fast and
frequent service, light rail stations often become focal points of urban re-
development, in considering residential as well as office functions. The
node-place model of Bertolini, discussed in chapter 3, is applied to the
light railway system of Naples in chapter 16 by Papa, Pagliara and Berto-
lini, to analyse the impacts of the transport system on the urban structure.
They find a decentralisation of residents from the centre to suburban areas
and an above city average increase of all sorts of real estate values near
new light rail stations. The application of the node-place model indicates a
connectivity increase in the urban system and an increase in network hier-
archy among the station areas. Finally, the node-place model shows that
the transit network also has a structuring effect on the urban system.
The impact of a light rail system on the residential real estate market is
studied by a theoretic simulation approach by Gat in chapter 17. He pre-
sents an urban polycentric model that is able to deal with automobile con-
gestion as well as transit and park-and-ride. His findings support the idea
that transit introduction is likely to raise real estate values, but are condi-
tional with regard to sparkling inner city redevelopment.
Lastly, Ferdman, Shefer and Bekhor in chapter 18 analyse the impact of
density and diversity: land use mix in trip generation for the light rail sta-
tion surroundings in the Tel Aviv metropolitan area. Their findings
strongly suggest that density and diversity (land use mix), play a major
role in determining zonal trip generation. Factors such as population den-
sity, job density, dependency ratio, and the proximity of ones residence to
the work place affect daily trip generation trends.
References
Banister D (2002) Transport Planning. Spon, London

Berg L van den, Pol P. (1999) The High Speed Train Station and Urban Revitali-
sation. Erasmus University, Rotterdam
Bertolini L, Spit T (1998) Cities on Rails: the Redevelopment of Railway Station
Areas. Spon, London
Brueckner J (2000) Urban Sprawl: Diagnosis and Remedies. International Re-
gional Science Review 23:160-171
Haynes KE (1997) Labor Markets and Regional Transportation Improvements: the
Case of High Speed Trains. The Annals of Regional Science 31:57-76
Korteweg PJ (1992) Kantoorruimte en Segmentering, VOGON Journaal 1:17-21
Marchetti C (1987) On Transport in Europe: the Last 50 Years and the Next 20.
IIASA, Laxenburg
Mikus W (1966) Die Auswirkungen eines Eisenbahnknotenpunktes auf die
Geographische Struktur einer Siedlung: am Speziellen Beispiel von Lehrte
und ein Vergleich mit Bebra uns Olten/Schweiz. PhD-thesis, Geographisch
Institut der Albert Ludwigs Universitt, Freiburg
Sjoberg G (1960) The Pre-industrial City. Free Press, New York
Smith D (1998) The Railway Mapping of British Towns. The Cartographic Jour-
nal 35:141-154
Spaans M (2000) Realisatie van Stedelijke Revitaliseringsprojecten: een Interna-
tionale Vergelijking. Delft University Press, Delft
Turnock D (1998) An Historical Geography of Railways in Great Britain and Ire-
land. Ashgate, Aldershot
Vance JE (1966) Housing the Worker: the Employment Linkage as a Force in Ur-
ban Structure. Economic Geography 42:294-325
Vickerman (1996) High Speed Rail in Europe: Experience and Issues for Future
Development. The Annals of Regional Science 31:21-38
Part A: Setting the scene
2 Urban dynamics and transport infrastructure:
Towards greater synergy
Hugo Priemus

2.1 Introduction
Both urban planning and infrastructure development are highly path-

dependent: it is difficult to begin with a clean slate in and around a city
unless there has been an earthquake, flood or warfare. Towns always have
to adapt their development to givens such as landscape, nature areas, con-
tours, the courses of rivers, lakes and seas, heritage sites, and historic set-
tlement patterns. In many cases the pattern of agricultural land division is
reflected in the layout of urban building (Amsterdam is a well-known ex-
ample). A major advantage of this path-dependency is that each town is
unique and has its own special cultural and landscape qualities.
The main theme of this chapter is the synergy (or lack thereof) between
urban development and the development of transport networks in Europe,
and the role played in this by various types of policy. The question is
whether this synergy has grown or declined in and around European cities
in recent decades, in which the modal split has changed. We then discuss
what policy tools are available to increase this synergy and how they could
be applied.
The hypothesis in this contribution is that the synergy between urban
development and the development of transport networks has in effect dete-
riorated since the advent and spread of the car and the airplane in Europe.
Policies have until recently neglected this point. There is every reason to
strengthen the interrelationship, as it improves the functioning of both the
urban regions and the transport networks.
This contribution focuses on European urban culture, taking a relatively
large number of examples from Dutch cities and regions. Here we are see-
16 Hugo Priemus
ing the development of poly-nuclear urban regions (urban networks and

network cities), a trend which holds much promise for a better synergy be-
tween urban development and the development of infrastructure networks.
This poly-nuclear regional development appears to be a more general
European phenomenon (Meijers 2005).
We shall not consider the relatively young American cities in what fol-
lows: although the concept of Transit Oriented Development is becoming
more popular in the USA, the unchecked growth of suburbs has produced
an almost entirely car-dependent urban system, and the differences be-
tween the central city and the suburbs have intensified over the years by
such factors as exclusionary zoning in the latter. Only a small number of
cities (such as New York, Washington DC, San Francisco, and Portland)
have reasonably good public transport systems. The modal split in the
USA is increasingly dominated by car and air travel, even more than in
Europe (EU 2004) as will be shown in section 2.4.
Positive examples of a satisfactory synergy between city and infrastruc-
ture are also provided by monocentric cities such as London and Paris.
Both cities developed extensive metro systems at an early stage that main-
tained accessibility to the city centre and then intensified the interaction
between city centre, suburbs and region; their metro systems have there-
fore become essential to their survival. In other European cities (like
Strasbourg, Nantes, Grenoble, and Karlsruhe) we are seeing the develop-
ment of light rail, integrating the classic urban and regional public trans-
port systems (Priemus and Konings 2001). One factor behind these trends
is the increasing tendency for the urban system to develop across munici-
pal boundaries: housing markets, labour markets and mobility markets are
increasingly operating on a regional scale. As a result both urban patterns
and infrastructure networks are becoming more and more regional.
Section 2.2 presents historical notes on the development of the railways
in Europe. Section 2.3 deals with the background of the current revival of
railway networks and station areas in Western Europe. Section 2.4 dis-
cusses the changes in modal split since 1970 and the recent increase in the
share of high-speed rail transport. Section 2.5 analyses the case of the
Netherlands as a country in which transport networks and urban patterns
diverged in the 19th and 20th centuries. Section 2.6 sets out the prospects
for a better integration of infrastructure networks and urban patterns, illus-
trated by some recent developments in urban regions. Finally, section 2.7
suggests a number of policy instruments to improve the synergy between
urban dynamics and transport infrastructures (Priemus et al. 2001). In
Europe and other parts of the world better integration between urban dy-
namics and transport infrastructure is vital if we want to improve both the
urban economy and the quality of the environment.
Urban dynamics and transport infrastructure: Towards greater synergy 17
2.2 Some historical notes on the development of the

railways
If we want to know what impact modern high-speed train (HST) stations

have on urban dynamics, we need to learn from the experience of building
railways and railway stations in the 19th century. Richards and MacKenzie
(1986 p 1) comment that There is perhaps no more potent or dramatic
symbol of the Industrial Revolution than the railways. Railways in fact go
as far back as the 17th century. Before 1830 wagons on railways were
moved by horsepower, gravity and, occasionally, manpower. They usually
carried bulk goods: coal, ore, stone, and timber. There was no question of
an integrated network prior to 1830, merely isolated lines to and from
mines, quarries and in many cases ports. Turnock (1998 p 55) describes a
common formula: loaded wagons should proceed safely by gravity
while horses returned the empties to the pits.
The breakthrough for the railways came at the beginning of the 19th cen-
tury, when they were conquered by the steam engine, an invention linked
particularly to the name of George Stephenson. From 1830 onwards began
the construction of a highly ramified network of public railways using lo-
comotion. Passenger transport now developed alongside that of goods, first
of all in the United Kingdom, where the 19th century is referred to as the
Railway Age. Coal traffic remained of supreme importance in the UK,
however, and it continued to stimulate the construction of new railway
lines right through the First World War (Turnock 1998).
The building of railway stations by no means always resulted in the
towns receiving a boost to their growth; Simmons (1986 pp 16-17) even
mentions a number of (rather exceptional) cases where the opening of a
station heralded a decline in the urban population. In many other cases
passenger transport by rail and the urban population grew in parallel, but
causes and effects were not clear. Everywhere industrialisation entailed
migration to the cities, as it was there that industrial employment was in-
creasing at the expense of jobs in the rural areas, which fell sharply as a re-
sult of rises in agricultural productivity. The development of rail travel
presumably played, at most, a conditioning role here (Turnock 1998 p
187): Arguably the impact of railways was greatest in the case of the ur-
ban areas. There was an increase in the number of towns and their average
size; and the transfer of population from the countryside to the towns
The railways, moreover, linked up the city centre with the suburbs, many
of which presumably owed their growth partly to the rail link with the cen-
tre.
18 Hugo Priemus
More stations were built in the cities, often in the form of terminuses
around a central urban area. Lines to big cities were opened by more than
one railway company. Often a central hub would develop there, and in
some cases, terminal stations had to be moved. The hub would usually
keep clear of the commercial heart of the town in order to minimise disrup-
tion.
The original idea was that railways and highways would link up with
one another. As early as 1800 J. Anderson urged that railways be built at
the side of turnpikes, following the established alignments apart from oc-
casional diversions around the edge of hills and some construction of tun-
nels and viaducts (Turnock 1998). As the network grew, the rail system
enhanced geographical cohesion: as Turnock remarked (1998 p 27)
[R]ailways contributed not only to national unity but also to regional co-
herence. The older canal system had only contributed to coherence within
regions; only later did the road network exceed the rail network in creating
geographical coherence on a regional, national and international scale.
2.3 Towards a new dynamic in railway networks and

station areas
In most Western European countries nowadays railway companies are

leading a troubled existence. Traditionally they are public monopolies
which are being made independent with considerable difficulty, and in a
few cases are completely or partly privatised. The train has only a modest
share in the modal split of passenger transport, while the passenger car and
the airplane are still gaining ground. Stations and station areas have, for a
long time, shared in the malaise of the railway sector. Privatisation in the
British railway sector was introduced in 1985. In the Netherlands the rail
sector was de-merged and privatised during the 1990s. At that time the op-
eration of station buildings and the development of station areas was in the
hands of separate limited liability companies; this marked the beginning of
a new dynamic in station areas in the Netherlands. In many other European
countries we see similar developments.
A second factor that has reinforced this dynamic has been the develop-
ment of the European HST network that started in France. Greater num-
bers of countries, such as Germany, Spain, Italy, Sweden, and the Nether-
lands are connected to this network. We expect that the stations where
high-speed train stops will have to cope with an increasing flow of passen-
gers, partly of high socio-economic status, will bring purchasing power
into the station building and its vicinity, thus reinforcing both the urban
and the transport functions of the station (Bertolini 1996, 1998; Bertolini
and Spit 1998; Serlie 1998; Zweedijk and Serlie 1998).
There is a third consideration that markedly reinforces the previously
mentioned factors: the construction of light rail connections and metro sys-
tems which improve public transport systems at the local and regional lev-
els (Raad voor Verkeer en Waterstaat 1996; Priemus and Konings 1999,
2000, 2001; Priemus 2004). Benchmark cities with elaborate underground
systems are of course Paris and London. Light rail systems were intro-
duced in a number of French and German urban regions. The relatively old
stations in urban areas serve as central nodes for these public transport sys-
tems which connect networks with each other at various scale levels.
Not only do developments in the transport sector mark a new dynamic
phase in the existence of urban station areas; the recovery of the urban
economy in advanced economies is also relevant (Boelhouwer et al. 1995;
Priemus 2001). This brings us to the fourth factor. The structural shift from
employment in industry to employment in business and personal services
is putting the cities back on the map. The rapid development of informa-
tion and communication technology raises the productivity of production
processes and supports consumer demand, thus leading to more opportuni-
ties for creative activities and knowledge development (Florida 2002). Cit-
ies provide a more attractive environment for activities of this kind rather
than for large-scale industrial processes which, to an increasing extent, are
moving to rapidly developing low-wage countries.
Finally, it seems that in certain respects the city is gaining ground as a
residential area. Surveys of households residential preferences in Western
Europe show that the suburban residential environment is still the most
popular. However, a significant minority of 25% to 30% prefer to live in a
metropolitan environment close to urban facilities and in a relatively cen-
tral location (Spaans et al. 2004). This trend makes the station area attrac-
tive as a residential environment, and supports the recent policy shift in the
Netherlands from compact city to network city (Govers et al. 1999;
Bontje 2003; Van den Burg and Dieleman 2004). The network city is poly-
nuclear and has several nodes, with stations contributing significantly to
the accessibility of the urban centres. At a slightly higher scale level, a few
network cities together form an urban network, such as Brabantstad and
Randstad Holland in the Netherlands. Urban networks seem to be a typi-
cal European pattern, popular in the British Midlands, the Ruhr Area and
the Flemish Diamond.
The new dynamic of stations and station areas means a considerable op-
portunity for investments for both the private and the public sector.
Broadly speaking, government authorities are responsible for the im-
provement of transport infrastructure and public space. There is also plenty
20 Hugo Priemus
to coordinate in the area of spatial economics. For the private sector there
are potential profits to be gained in various areas: land and property devel-
opment in the station area and possibly the operation of some public trans-
port connections.
2.4 Changes in modal split: The growing share of high-

speed rail transport
In this section we deal with the question: How is high-speed rail transport
in the EU developing since 1990 amidst a changing modal split? First we
will analyse the changing modal split since 1970. Between 1970 and 2000
the share of rail transport in the modal split of passenger transport in the
old EU (15 member states) fell from 10.4% to 6.3%. During the same pe-
riod the share of tram and metro fell from 1.6% to 1.0%. The bus also lost
ground: from 12.7% to 8.3%. The winners in this period were the passen-
ger car (from 73.8% to 78.5%) and the aeroplane (from 1.6% to 5.9%)
(Table 2.1). Although the passenger car has become increasingly dominant
in the EU since 1970, and although the market share of the airplane gained
ground in the modal split, railway transport also increased slightly in abso-
lute terms, in particular in highly urbanised areas.
Table 2.1. Performance by mode for passenger transport, EU-15, 1970-2000, se-
lected years (pkm, horizontal %)
Passenger Cars Bus & Coach Railways Tram & Metro Air Total
1970 73.8 12.7 10.4 1.6 1.6 100.0
1980 76.1 11.8 8.4 1.2 2.5 100.0
1990 79.0 9.3 6.7 1.0 4.0 100.0
Source: EU 2004
In 2002 the share of the passenger car in the Netherlands (81.5%) was
above the EU average (78.8%). The train had a higher market share in the
Netherlands than in the EU (8.1% against 6.2). In the Netherlands the bus
lags behind the EU average (4.1% against 8.3%); this is also true for the
tram plus metro (0.8% against 1.0%).
Table 2.2 shows that the shares of the passenger car and air transport
have also grown substantially in the Netherlands since 1990. The bus has
meanwhile lost market share. Since 1990 the shares of both tram and rail
services rose slightly; this is due mainly to the introduction of the free pub-
lic transport pass for students.
Table 2.2. Development of passenger transport in the Netherlands, 1990-2002, se-

lected years (1000 mio pkm)
1990 2000
Passenger Cars 137.30 141.10
Bus & Coach 13.00 7.50
Tram & Metro 1.26 1.43
Railways 11.06 14.76
Air Transport 4.08 9.59
Source: EU 2004
Although we devote our attention in this chapter to the redevelopment

of railway station areas, we must bear in mind that in the Netherlands and
the EU both the railways and the other public transport services are losing
ground. A couple of bright spots are the increase in light rail transport and
in particular the rapid development of high-speed rail transport in
Europe, which is discussed in the remainder of this section.
In the European Union high-speed rail transport has increased signifi-
cantly since 1990. Between 1990 and 2003 passenger transport by high-
speed train increased almost fivefold (Table 2.3).
Table 2.3. High-speed rail transport, EU-15, 1990-2003, selected years (1000 mio
pkm).
Year x 1000 mio pkm
1990 15.2
1995 32.8
2000 59.1
2003 70.5
Source: EU 2004
Table 2.4 shows how a number of countries participated in high-speed

rail transport in 2003. France has a dominant position in European high-
speed rail transport, followed by Germany, Italy and Spain (see also Table
2.5). In the Netherlands high-speed rail transport began only recently on
existing tracks that do not as yet allow for very high speeds. This will
change in 2007 when the new dedicated HST line connecting Amsterdam
and Belgium will open (Visser and Bentvelsen 1991; TCI 2004a, 2004b).
The United Kingdom was not represented in the 2003 data.
A considerable number of high-speed lines are under construction in
Europe (see Table 2.6). Step by step, the isolated high-speed tracks will be
integrated in order to form a true European high-speed rail network. Sev-
eral nodes in this network may develop a strategic function. Within a de-
22 Hugo Priemus
creasing share of the train in the modal split, the role of the high-speed
train in Europe will increase in the next decades.
Table 2.4. High-speed rail transport in EU countries, 2003 (1000 mio pkm).
Country x 1000 mio pkm
France 39.6
Germany 17.5
Italy 7.4
Spain 2.5
Sweden 2.3
Belgium 0.9
Netherlands 0.2
Finland 0.1
EU 15 70.5
Source: EU 2004
Table 2.5. Length of dedicated high-speed rail network per country, EU-15, 2002
(km)
France 1395
Germany 687
Spain 377
Italy 259
Belgium 135
Total EU15 2853
Source: EU 2004
Table 2.6. High-speed lines under construction in the EU, (2004)

Belgium Liege-German border 33 km
Belgium Antwerp-Dutch border 38 km
Netherlands Amsterdam/Schiphol-Belgian border 120 km
Germany Nuremberg-Ingolstadt 88 km
Spain Madrid-Lerida 481 km
France TGV Est 302 km
Italy Milan-Bologna 196 km
Italy Bologna-Florance 77 km
Italy Turin-Novara 92 km
Italy Rome-Napels 220 km
Sweden Sdertlje-Linkping 140 km
Sweden Nyland-Ume 190 km
United Kingdom Ebbsfleet-London (St. Pancras) 38 km
Total 2015 km
Source: EU 2004
2.5 Divergence of transport networks and urban patterns

in Europe
European cities experienced a sharp demographic and economic growth

during the second half of the 19th century and after World War II in par-
ticular. In the 19th century and at the beginning of the 20th century, mainly
migration from rural areas to towns and cities promoted the growth of the
latter, whereas after World War II a major role was played by natural
growth and subsequent foreign immigration. In the early days the growth
of each town was mostly concentric: a town developed from a single cen-
tre by means of peripheral urban growth, topography and landscape per-
mitting. The beginning of the 20th century in Great Britain witnessed the
inception of the garden city movement, which broke with the rampant
growth of the big cities (Howard 1903). After Welwyn a few garden cities
were developed at some distance from the central city, on a relatively
small scale and with close interaction between city and countryside: less
bricks and mortar, more open space. As Great Britain too needed to meet a
rapidly rising demand for housing, it was not enough to build charming,
small-scale villages, but a number of substantial new towns were also built
at a distance from the city centres. This approach crossed over to other
countries; in the Netherlands a growth centre policy was drawn up and im-
plemented starting in the 1970s (Faludi and Van der Valk 1990, 1995).
This was a reflection of British new town policy which was adopted in a
number of European countries and maintained in the Netherlands until the
end of the 1980s.
In most European cities providing access for road traffic has proved to
be a difficult task. In the 1960s ambitious demolition schemes were carried
out in many European towns and cities, designed partly to make room
for car traffic. This trend met with intense opposition from many residents;
it badly damaged the spatial quality, the environmental quality and pleas-
antness of the town or city, and problems of congestion shifted to the mo-
torway slip roads.
In the 1970s housing renovation and preservation of the residential func-
tion of town centres became the top priority in the Netherlands (Koffij-
berg 2005). Traffic calming methods gained in popularity in most towns.
The first home zones (woonerven) were created in residential areas and
pedestrian areas were sprouting up in town centres. This too initially gave
rise to protest, but people soon realised that most of the pedestrian areas
worked well. Many cities in Europe followed this track and began to im-
prove the quality of open space in their urban areas. But it proved exceed-
24 Hugo Priemus
ingly difficult to integrate transport networks and urban patterns at the

same time.
In 1988 the Dutch Fourth Spatial Planning Memorandum (Ministerie
van VROM 1988) was published; close on its heels came the Fourth Spa-
tial Planning Memorandum Extra (VINEX, Ministerie van VROM 1990)
which broke with growth centre policy and introduced its diametric oppo-
site: compact city policy. New housing estates were to be built on large,
concentrated house building sites, the so-called VINEX sites, in and ad-
jacent to towns. Hundreds of thousands of homes were built on these sites,
especially between 1995 and 2005. A problem with the VINEX sites is that
80% of the new houses are single-family homes in a not-particularly high
urban density of 30-35 dwellings per hectare, and the new estates are not
connected to a high-quality regional public transport network. Neither
growth centre policy nor VINEX policy have contributed much to the syn-
ergy between urban development and infrastructure networks. In general,
no nodes were developed where infrastructure networks and urban net-
works coincide and support each other.
In a growing number of European cities cars were now increasingly dis-
couraged from entering towns and cities. Multi-storey or underground car
parks were built, but this only marginally increased total parking capacity
in the town centres. Many of these parking garages show no profit and are
partly subsidised by local governments. Dutch VINEX policy (Ministerie
van VROM 1990) introduced ABC location policy, under which parking
space around major central railway stations was even deliberately re-
stricted. The aim of this policy was to promote public transport and dimin-
ish car traffic in town centres. The former was unsuccessful, the latter suc-
cessful, but the question is whether the towns whose car access had been
greatly reduced really had anything to crow about. ABC location policy
severely truncated thinking in terms of multimodal transport systems
(Priemus 2000). It became almost impossible to transfer from train to car
(and vice versa) in town centres; instead, people were re-routed to much
smaller park-and-ride stations in the suburbs, or even further outlying ar-
eas. Again, there was a lack of synergy between transport networks and
urban patterns.
Ausubel and Herman (1988) wrote a noteworthy book entitled Cities
and their vital systems. Their edited collection of articles analyse the vari-
ous types of urban infrastructure. In the book Craig (1988 pp 222-32) dis-
cusses the relationship between airports and cities: The so-called landside
problems of the air transport system create massive and largely un-
researched issues of () how to connect the air transport system to other
modes of transport. The airport can be regarded as essentially the location
at which autos meet aircraft, and these intermodal connections are gener-
ally ignored. We also wonder when airports will begin to be recognised

as the new city centres, as they become central places for more and more
socio-economic activity.
It generally became clear that urban areas and the air transport system
are not very well connected, thus highlighting yet another dimension aris-
ing in the lack of synergy between transport networks and urban patterns.
This contribution focusses on European urban culture, taking a relatively
large number of examples from Dutch cities and regions. Here we are see-
ing the development of poly-nuclear urban regions (urban networks and
network cities), a trend which holds much promise for a better synergy be-
tween urban development and the development of infrastructure networks.
This poly-nuclear regional development appears to be a more general
European phenomenon (Meijers 2005).
The relationship between airports and the centres of some cities has
been altered in a number of ways. Amsterdam Schiphol has rapidly devel-
oped into an airport city, with no residents but a growing workforce and an
enormous number of visitors every day. Where airports are situated close
to the city, increasing numbers of passengers board and disembark at air-
ports. Here a fast, reliable, safe, and environmentally friendly link with the
city centre becomes increasingly important. Taxis and buses have tradi-
tionally provided seamless transport between airports and city centres
convenient for passengers with plenty of luggage who may be unfamiliar
with the area, but a formula with major capacity limitations and environ-
mental drawbacks in the long-run. Interesting developments can be seen in
Shanghai (the Maglev rail link between the airport and the city), Kuala
Lumpur and Tokyo (fast rail links between the airport and the city), and
Amsterdam, with its high-speed rail link to the South Axis and/or to Am-
sterdam Central. The fact remains, however, that most European cities
have poor interconnections between the various modes (air, rail, car,
tram/metro/light rail) and their central areas are inaccessible to the modes
with the highest share in the modal split: air and car. Although the synergy
between the urban pattern and infrastructure networks has improved in
various places, we must conclude that in most other cities this relationship
has deteriorated in recent years.
2.6 Towards better integration of infrastructure networks

and urban patterns
The development of the mononuclear city into the poly-nuclear urban re-
gion (Batten 1995; Kloosterman and Lambregts 2001; Van der Burg and
26 Hugo Priemus
Dieleman 2004; Meijers 2005) by definition means the development of

more urban centres, and more nodes connecting the urban pattern and the
infrastructure networks together (Ministerie van VROM 2004). Not only is
housing suburbanising, so is employment, with the development of secon-
dary urban centres (edge cities) (Garreau 1991). What is interesting in this
connection is the study of nodes carried out by the Province of South Hol-
land (Werkgroep Knooppunten 1999) which looked into i.) specific func-
tion mixes for each node and ii.) specific transport functions. This offers
new opportunities for areas around stations. Special attention is being paid
to the areas around those where the high-speed train will stop (Pol 2002):
here major urban impacts are expected, as has previously been the case in
Japan (Amano et al. 1991) and France (Newman and Thornley 1995). The
three main stations in Paris (Nord, Lyon and Montparnasse) are near to
maximum capacity, and in 2005 the SNCF (French Railways) put forward
the idea of an HST station in the centre of the city, an underground station
beneath Place de la Rpublique, closely linked to the metro system. The
Netherlands will soon have no fewer than seven HST stations within a
small area: Amsterdam Central or Amsterdam South Axis, Schiphol, The
Hague Central, Rotterdam Central, Utrecht Central, Breda and Arnhem.
With such a high HST station density, the time-saving of international rail
transport is small, and there is little reason to expect a major boost in urban
development.
The idea of the poly-nuclear urban region is a promising perspective
when it comes to answering the following question: How can the synergy
between the urban pattern and infrastructure networks be enhanced, taking
into account the substantial investments which the existing urban pattern
and infrastructure networks have already demanded?
In general, we would recommend the following approach. First of all it
is worthwhile to plot the existing urban pattern, including additions and
extensions already planned, and the existing infrastructure networks. The
first question that can be posed is whether passengers can transfer from
one mode to another, and if so, how. In many cases the system will need to
be redesigned at the regional level, creating or improving transfer points:
From air to car or train (and vice versa);
From car (on trunk roads outside built-up areas) to
rail/metro/tram/light rail (and vice versa);
From rail to metro/tram/light rail (and vice versa).
In the Netherlands it is particularly important to chart the network of cy-
cle routes and provide every railway station with a cycle shed for sufficient
numbers of bicycles, as the share of the bicycle is relatively high in the
Randstad: 22.6 30.2% of all work trips in 1990 (Clark and Kuijpers-
Linde 1994).
Once the redesign mentioned above has been carried out, the nodes need
to be highlighted and classified. An airport is by definition a node, as is a
railway station. It is also important to plot the metro/tram/light rail stops
and zoom in on the nodes where passengers can transfer from one mode to
another. In many cases the transport node function will need to be en-
hanced. Apart from the transport function, the function mix in and around
each node is significant: housing, offices, hotels, restaurants, bars, educa-
tional and cultural facilities, etc. When analysing the functioning of and
prospects for HST station areas, Bertolinis approach (1996, 1999) is use-
ful, as it distinguishes between node value (transport value) and place
value (functional value) (Bertolini and Spit 1998 and chapter 3). In this
approach it is essential that the transport function and function mix of each
node be in balance with each other. As a rule of thumb, the more passen-
gers per day who transfer and get on and off at a node, the more reason
there is to provide a rich and varied function mix. It is important here to
not tar all stations and station areas with the same brush. Various authors
have developed station typologies. Richards and MacKenzie (1986 pp 160-
184) identify the following station types: royal stations, commuter stations,
pilgrim stations, seaside stations, country stations, and finally necropo-
lis stations. This classification by main user category and location seems to
not be generally applicable. Simmons (1986) classifies them on the basis
of the size and nature of the towns they serve: London, great provincial cit-
ies, major towns, railway towns, ports and docks, watering-places, minor
towns, and rural England and Wales. Nowadays it makes sense to classify
stations by their size and site on the network (along a route, at a three-way
fork, at an intersection, etc.). In general, it is important that urban nodes be
considered as interfaces between transport networks and urban functions,
each with its own specific characteristics.
The approach offered by Bertolini (1996, 1999; Bertolini and Spit 1998)
can result in the function mix around certain nodes being enriched and in-
tensified, and also the differences in mix between nodes being intensified,
so as to avoid duplication, to promote complementarity and to do greater
justice to the urban functions. What this approach produces is a (re)design
for the network city, and where a number of cities are involved, usually at
a higher level of scale, a (re)design for an urban network.
This contribution focusses on European urban culture, taking a rela-
tively large number of examples from Dutch cities and regions. Here we
are seeing the development of poly-nuclear urban regions (urban networks
and network cities), a trend which holds much promise for a better synergy
between urban development and the development of infrastructure net-
28 Hugo Priemus
works. This poly-nuclear regional development appears to be a more gen-

eral European phenomenon (Meijers 2005).
Policy of the European Commission is increasingly important; however,
it does not pay sufficient attention to the synergy between urban develop-
ment and infrastructure policy: the Commission is responsible for infra-
structure policy at the European scale, but not for spatial policy or urban
policy. As a result, the European Commission cannot provide synergy and
integration between infrastructure policy and spatial policy.
The high-speed rail network has recently been developed in Europe as
part of the Trans-European Network (TEN), in part as an alternative to
continental flights over relatively short distances. It is important for high-
speed trains to stop at the major airports in order to ensure interconnectiv-
ity between air routes (e.g. intercontinental) and rail routes (continental)
(Givoni and Banister forthcoming). Whether this will result in a substantial
substitution of air travel by rail travel (as intended from the start) is doubt-
ful, as the low fares charged by price-cutters such as Easyjet and Ryanair
have upset the original substitution calculations. However, concerted atten-
tion will be devoted to nodes in the future spatial development of city re-
gions; there are four basic types of intermodal transport nodes:
Airports (sometimes including an airport city, as in the case of Schip-
hol);
Railway stations (including a concentration of urban functions);
Park-and-ride stations (with emphasis on multimodal transfer,
usually sited outside town centres);
Ports (generally with a high proportion of goods traffic and/or
recreational passenger transport).
It is important for car drivers to have access to parking areas near the
home or office (in many cases even on the premises); in this case pricing
needs to be used to ration demand. In highly urbanised areas (e.g., Manhat-
tan and Hong Kong) non-stacked parking will be too expensive. An inte-
grated system of user charges has to be introduced for both moving traffic
and parked cars, with facilities for paying in the same quick and reliable
way throughout the country (preferably throughout the EU). We are still a
long way from achieving this ideal. The present time consuming, widely
varying collection systems confront car drivers and public transport pas-
sengers with continuous annoyances. In general, there are still ample op-
portunities for travel information systems (navigation systems in cars, pub-
lic transport information systems) to contribute to the smooth running of
urban transport systems.
A major task for the future is to make towns and cities accessible to traf-
fic. Multi-storey and underground car parks need to be built in and around
town centres to provide access for a selection of car drivers who are will-
ing and able to pay the price. Wherever possible the capacity of urban
trunk roads will need to be increased still further, but there are obvious
limitations to this. Cars will moreover increasingly have to be parked on
the periphery, at park-and-ride stations, from which high-quality public
transport shuttles visitors to the town centre; this model has been applied
fairly consistently in Bristol, for example. More facilities will be added to
these park-and-ride stations, e.g., filling stations, car washes, car repair fa-
cilities, florists, gift shops, cafes, and meeting rooms. In London the city
centre has been made car-free, but it remains easily accessible, thanks
partly to the metro system. If a city has a high-quality, high-capacity pub-
lic transport system that is safe and secure, the city example can be fol-
lowed elsewhere.
At present the hot topic is indeed the safety and security of public trans-
port systems. Since the attacks in Madrid and London, there has been
strong interest in the external safety of metro systems. In general, our
transport systems are highly based on trust; any car can carry a car bomb
on board; any rucksack can contain explosives. There would seem to be no
alternative to defining certain subsystems within the general, trust-based
transport system where systematic checks are carried out, viz. airports, air-
craft, ocean-going vessels, high-speed trains, and possibly underground
metro systems, as the effects of a disaster there can be catastrophic.
Checks could involve metal detector portals, CCTV systems, surveillance
staff, and possibly military personnel. This expensive prospect could be a
good reason to bring public transport systems above ground wherever fea-
sible, make the vehicles noticed, which could then encourage more people
to use public transport as well as highlight the visibility of the integration
of the city and the transport infrastructure. Modern light rail systems pro-
vide attractive examples. In the mix of underground and above ground
transport infrastructures the best compromise will have to be found be-
tween transport safety and transport security.
When it comes to the dynamic development of a network city and an
urban network we need to look not only at the relationship between the
transport infrastructure and urban property (Priemus et al. 2001) but also,
of course, at the quality of public space and the quality of green-blue net-
works (areas for nature and recreation). It is important to town dwellers
that these networks have high biodiversity and open-air recreation facilities
not too far from their homes. Although essential, this topic will not be
elaborated in this chapter.
30 Hugo Priemus
2.7 Policy tools to enhance the synergy between urban

dynamics and transport infrastructures
We have already mentioned a few tools that can help improve the relation-
ship between urban dynamics and transport infrastructure. In this section
we examine them systematically. The two related research questions dealt
with in this section are: what policy instruments could promote the synergy
between urban dynamics and transport infrastructure, and how can they be
applied?
First, a comprehensive spatial policy is important, aiming above all at
good horizontal coordination between property development and infra-
structure networks at the regional level and good vertical coordination be-
tween local, regional and national/international spatial policy. Next, con-
sistent attention needs to be paid to the development of infrastructure
networks, both rail and road, ensuring that interconnections between the
networks (air, road, light rail/tram/metro/bus) are rigorously promoted. It
is then important to selectively and actively develop or redevelop nodes,
always aiming in line with suggestions made by Bertolini (1996, 1999)
for a good balance between functional and transport value. In some cases
the function mix of railway station areas needs to be enhanced, thus pro-
moting the economic vitality of towns and enabling public transport to
cover more of its costs. To ensure that town centres remain accessible in
the long-term, a considerable improvement in public transport in the urban
region is needed: in addition to underground backbones, the development
of light rail systems is an interesting option, enabling urban and regional
public transport to be integrated.
The tools mentioned here can be applied as part of a spatial policy
aimed at developing and strengthening network cities and urban networks.
Designs need to cover different levels of scale so that European infrastruc-
ture networks, poly-nuclear urban regions, regional and local infrastruc-
ture, and property development strengthen one another.
Lastly, pricing is an important tool if we are to increase the actual ca-
pacity of infrastructure networks and achieve better utilisation of the
available parking facilities, both in urban centres and at nodes. In some
cases a price may be charged for access to certain attractive areas, as
shown by the introduction of the congestion charge in Londons city cen-
tre. Privatisation of urban infrastructures and space can, however, contrib-
ute to the splintering of urbanism (Graham and Marvin 2001).
The general aim should be to preserve the increased value (value cap-
turing) of nodes due to greater accessibility in order to improve the quality
of the urban space and/or co-fund infrastructure links between the nodes.
Forms of public-private partnership where risks and benefits are shared

are promising here: if suitable arrangements are made, there can be a
marked improvement of urban public transport and substantial savings on
government subsidies.
Synergy between urban dynamics and transport infrastructure is not
something that develops automatically; a mix of policy tools is required.
Spatial policy, infrastructure policy and pricing policy need to be com-
bined. Only by applying such a mix can better synergy between urban dy-
namics and transport infrastructure be reached.
Acknowledgements
This contribution was written in the framework of the Delft Center Next
Generation Infrastructures and the BSIK-programme Next Generation In-
frastructures, co-funded by the Dutch government.
References
Amano K, Toda T, Nakagawa D (1991) The Rapid Transportation System and the
Socioeconomic Restructuring of Japan. In: Brotchie J, Batty M, Hall P, New-
ton P (eds) Cities of the 21st Century: New Technologies and Spatial Systems.
Longman Cheshire, New York, pp. 3947
Ausubel JH, Herman R (eds) (1988) Cities and Their Vital Systems. Infrastruc-
ture: Past, Present, and Future. Advisory Committee on Technology and Soci-
ety, National Academy Press, Washington DC
Batten DF (1995) Network Cities: Creative Urban Agglomerations for the 21st
Century. Urban Studies 32:313327
Bertolini L (1996) Nodes and Places: Complexities of Railway Station Redevel-
opment. European Planning Studies 4:331345
Bertolini L (1998) (Her)ontwikkeling van Stationslocaties. Stedebouw en Ruim-
telijke Ordening 79/4:49
Bertolini L (1999) Spatial Development Patterns and Public Transport: the Appli-
cation of an Analytical Model in the Netherlands. Planning Practice & Re-
search, 14:199210
Areas. Spon, London/New York
Boelhouwer PJ, Kruythoff HM, Priemus H (1995) Beleid voor de Grote Stad in de
Toekomst. Delft University Press, Delft
Bontje M (2003) A Planners Paradise Lost? Past, Present and Future of Dutch
National Urbanization Policy. European Urban and Regional Studies 10:135
151
32 Hugo Priemus
Burg A J van der, Dieleman FM (2004) Dutch Urbanisation Policies: from Com-
pact City to Urban Network. TESG 95:108116
Clark WAV, Kuijpers-Linde M (1994) Commuting in Restructuring Urban Re-
gions. Urban Studies 31:465483
Craig T (1988) Air Traffic Congestion: Problems and Prospects. In: Ausubel JH,
Herman R (eds) Cities and Their Vital Systems. Infrastructure: Past, Present,
and Future. Advisory Committee on Technology and Society, National Acad-
emy Press, Washington DC, pp 222232
European Commission, Directorate-General for Energy and Transport (2004) EU
Energy and Transport in Figures: Statistical Pocketbook 2004. Office for Of-
ficial Publications of the European Communities, Luxembourg
Faludi A, Valk A van der (1990) De Groeikernen als Hoeksteen van de
Nederlandse Ruimtelijke Planningsdoctrine. Universiteit van Amsterdam,
Amsterdam
Faludi A, Valk A van der (1995) Rule and Order: Dutch Planning Doctrine in the
Twentieth Century. Kluwer, Dordrecht
Florida R (2002) The Rise of the Creative Class. Basic Books, New York
Garreau J (1991) Edge City: Life on the New Frontier. Doubleday, New York
Givoni M, Banister D (forthcoming) The Role of the Railways in the Future of Air
Transport. Transportation Planning and Technology
Govers B, Brand-van Tuijn R, Leusden R van (1999) Van Stad naar Netwerk:
Kansen voor het Openbaar Vervoer. Rooilijn 32:285290
Graham S, Marvin S (2001) Splintering Urbanism: Networked Infrastructures,
Technological Mobilities and the Urban Condition. Routledge, London / New
York
Howard E (1903) Garden Cities of Tomorrow. Faber & Faber, London
Kloosterman RC, Lambregts B (2001) Clustering of Economic Activities in Poly-
centric Urban Regions: the Case of the Randstad. Urban Studies 38:717732
Koffijberg J (2005) Getijden van Beleid: Omslagpunten in de Volkshuisvesting.
Over de Rol van Hirarchie en Netwerken bij Grote Veranderingen. DUP Sci-
ence, Delft
Meijers E (2005) Polycentric Urban Regions and the Quest for Synergy: is a Net-
work of Cities More Than the Sum of the Parts? Urban Studies 42:765781
Ministerie van Volkshuisvesting, Ruimtelijke Ordening en Milieubeheer (VROM)
(1988) Vierde Nota Ruimtelijke Ordening. Ministerie van VROM, The Hague
Ministerie van VROM (1990) Vierde Nota Ruimtelijke Ordening Extra (VINEX).
Ministerie van VROM, The Hague
Ministerie van VROM (2004) Nota Ruimte: Ruimte voor Ontwikkeling. Minis-
terie van VROM,The Hague
Newman P, Thornley A (1995) Euralille: Boosterism at the Centre of Europe.
European Urban and Regional Studies 2:237246
Pol PMJ (2002) A Renaissance of Stations, Railways and Cities: Economic Ef-
fects, Development Strategies and Organisational Issues of European High-
Speed Train Stations. PhD thesis, DUP Science, Delft
Priemus H (2000) Dutch Experience with Compact City Policy and ABC Location
Policy. In: Brunsing J, Frehn M (eds) Stadt der Kurzen Wege. Zukunftfhiger
Leitbild oder Planerische Utopie? Institut fr Raumplanung, Dortmund, pp

105111
Priemus H (2001) Recent Transformations in Urban Policies in the Netherlands.
In: Graafland A, Hauptmann D (eds) Cities in Transition. 010 Publishers, Rot-
terdam, pp 388403
Priemus H (2004) Light Rail: Backbone of European Urban Regions. In: Beuthe
M, Himanen V, Reggiani A, Zamparini L (eds) Transport Developments and
Innovations in an Evolving World. Springer, Berlin, pp 255273
Priemus H, Konings JW (1999) Stadsgewestelijk Openbaar Vervoer. Infrastruc-
tuur, Transport en Logistiek 27, Delft University Press, Delft
Priemus H, Konings JW (2000) Public Transport in Urbanised Regions: the Miss-
ing Link in the Pursuit of the Economic Vitality of Cities. Planning Practice &
Research 15:233245.
Priemus H, Konings R (2001) Light Rail in Urban Regions: What Dutch Policy
Makers Could Learn From Experiences in France, Germany and Japan. Jour-
nal of Transport Geography 9:187198
Priemus H, Nijkamp P, Banister D (2001) Mobility and Spatial Dynamics: an Un-
easy Relationship. Journal of Transport Geography 9:167171
Priemus H, Visser JGSN (1995) Infrastructure Policy in the Randstad Holland:
Struggle Between Accessibility and Sustainability. Political Geography
14:363377
Raad voor Verkeer en Waterstaat (1996) Knooppunten in Openbaar Vervoernet-
werken. Advisory Council for Transport, Public Works and Water Manage-
ment, The Hague
Richards J, MacKenzie JM (1986) The railway station: a social history. Oxford
University Press, Oxford/New York
Serlie Z (1998) Stationslocaties in Vergelijkend Perspectief. PhD thesis, Universi-
ty of Utrecht, Utrecht
Simmons J (1986) The Railway in Town and Country: 18301914. David &
Charles, Newton Abbot/London/North Pomfret
Spaans M, Maat K, Konings R, Meijers E, Kersloot J, Priemus H (2004) Potenties
voor Wonen nabij Knooppunten. Bouwfonds Ontwikkeling, Hoevelaken
Tijdelijke Commissie Infrastructuurprojecten (TCI) (2004a) Grote Projecten Uit-
vergroot: een Infrastructuur voor Besluitvorming. Sdu Uitgevers, The Hague
TCI (2004b) Reconstructie HSL-Zuid: de Besluitvorming Uitvergroot. Sdu Uit-
gevers, The Hague
Visser JGSN, Bentvelsen TGM (1991) De Hoge-snelheidstrein in de Randstad:
Beleid en Besluitvorming Omtrent een Grootschalig Infrastructuurproject.
Delft University Press, Delft
Werkgroep Knooppunten (1999) Knooppunten in het Netwerk van de Zuid-
vleugel: een Verkenning. Provincie Zuid-Holland, The Hague
Zweedijk A, Serlie Z (1998) Een Knoop-Plaats-Model voor Stationslocaties.
Geografie 7:3537
3 Station areas as nodes and places in urban
networks: An analytical tool and alternative
development strategies
Luca Bertolini
Department of Geography, Planning and International Development

Studies, University of Amsterdam, the Netherlands
3.1 Introduction: Why stations?
In cities around the world, railway stations are increasingly the focus of in-
tegrated transport and land use development efforts, whether under the la-
bel Transit Oriented Development (TOD) as in North America, or more
plainly, as (re)development of and around railway stations as in Europe
and elsewhere (Bertolini and Spit 1998; Cervero 1998, 2004; Van den
Berg and Pol 1998; Dittmar and Ohland 2004; Dunphy et al. 2005). A
combination of heterogeneous, interrelated factors converge in determin-
ing this upsurge of station area-related urban projects.
A first factor triggering station area projects are the new development
opportunities provided by transport innovations, such as the expansion of
high-speed railways systems (particularly in Europe and Asia) or light rail
systems (as most notably in North America and Western Europe), but also
by the generalised transfer of distribution and manufacturing activities
away from station areas and towards more peripheral locations or new,
dedicated freight interchanges.
A second factor is the ongoing privatisation process or at least the shift
towards greater market-orientation of transportation, and most notably,
railway companies. One consequence of privatisation is that transportation
infrastructure and service providers are increasingly seeking ways to re-
capture the accessibility premium they help to create. Characteristically,
this implies the development of commercial activities within stations and
redevelopment of land above or around stations. Many Asian cities have a
36 Luca Bertolini
long tradition in this respect, but the trend has been expanding in Western
Europe and North America as well.
Third is a wish to boost the competitive position of cities as places to
live, work and consume through new large-scale urban projects. Many of
these projects, typically showing a dense mix of office, retail, leisure, and
housing, are located around highly accessible places such as main railway
stations. High-speed railway station areas in European cities in particular
have been the theatres of many such initiatives in recent years. A last, but
not least, factor and most notably in North America, is mounting concern
about the sustainability of sprawling and car-dependent urbanisation
patterns. The integrated development of railway networks and land around
the nodes of those networks is seen as a way towards a more public trans-
port and non-motorised, modes-oriented, concentrated urbanisation pat-
tern. The arguments for this shift are not merely environmental (reduction
of pollution, greenhouse emissions, land consumption, etc.); many local
governments and citizens also see it as a condition for the development of
a much needed mobility alternative for metropolises rapidly approaching,
but not yet experiencing, total traffic gridlock.
For all of its perceived potential, the integration of transport and urban
development at station areas is also a very complex undertaking. The
growing flows of people passing through stations are a direct result of the
increasingly open nature of the urban system: of people living in one place,
working in a second and spending their free-time in yet a third, but also of
business relationships requiring exchanges of persons based in distant lo-
cations, or of equally extensive spatial patterns of movement generated by
different types of consumption. The coincidence of different spatial scales
(in the most extreme case from the global scale of High-Speed Train
(HST) destinations to the locale of the station neighbourhood) is mirrored
by the presence of a broad range of users (from the cosmopolitan busi-
nessperson to the drifting homeless).
Station areas are, ambivalently, both nodes and places (Bertolini
1996). They are (or may become) important nodes in both transport and
non-transport (e.g. business, consumption) networks. Conversely, station
areas also identify a place, a both permanently and temporarily inhabited
area of the city, a dense and diverse conglomeration of uses and forms ac-
cumulated over time, which may or may not share in the life of the node.
Accordingly, a multifarious array of both node- and place-based actors
crowd station area development processes, of which the local government
and the railway company are characteristic. Depending on the local con-
text, other actors will also have a decisive role. These include different
levels of the public administration, different transportation companies and
market actors: developers, investors, end-users. Furthermore, and particu-
Station areas as nodes and places in urban networks 37
larly at station areas set in dense, historically stratified urban districts, local
residents and businesses may also have a significant stake. The objectives
of this heterogeneous array of actors are often conflicting and at best un-
coordinated.
Any successful development strategy for station areas needs to deal with
this complex, node-place dynamics. Insight into the characteristics of such
dynamics, its underlying factors, and the possibilities of influencing it,
would seem a necessary precondition for effective action. The information
should be sophisticated enough to be true to the complex reality, but plain
enough to be understood by those involved in the strategy development
process. This is a combination of qualities that is, however, seldom met,
and not just in the case of station areas. Most information on the function-
ing of spatial systems tends to be either too sophisticated to be understood
by policy makers, or too simple to add to already existing insights (Lee
1973, 1994; Vonk et al. 2005).
The aim of this chapter is to help fill this information gap by explicitly
linking exploration of the development dynamics to exploration of poten-
tial development strategies. This is done in two steps. In the first part of
the chapter, I introduce an analytical tool to enhance insight in the devel-
opment dynamics of station areas in urban networks, a node-place model,
based on Bertolini (1999). Its workings are illustrated by an application to
station areas in the Amsterdam and Utrecht urban regions. In the second
part I show how the gained insight can help characterise and discuss alter-
native development strategies. A more traditional coordinated develop-
ment approach typified by the Stockholm case is contrasted with an
emerging reconnecting developments approach typified by the examples
of Karlsruhe, Naples and Rotterdam-The Hague. A provisional balance of
the proposed view is made in the conclusion.
3.2 Characterising the development challenge
3.2.1 A node-place model
The node-place model introduced in Bertolini (1999; Figure 3.1) provides

an analytical framework to penetrate the dynamics of station area devel-
opment. The model freely builds on elaborations of the transport land use
feed back cycle (e.g., Manheim 1974; Hanson 1995; Wegener and Frst
1999; Meyer and Miller 2001).
38 Luca Bertolini
Node
Unbalanced Stress
node
Balance
Dependence Unbalanced
place
Place
Figure 3.1. A node-place model
The underlying idea is that improving transport provision in a location

(in the model, its node-value, the y axis in Figure 3.1) will, because of im-
proved accessibility, create conditions favourable to the further intensifica-
tion and diversification of land uses there. In its turn, intensification and
diversification of land uses in a location (or increase in its place-value, the
x axis in Figure 3.1) will, because of growth in the demand for connec-
tions, create conditions favourable to the further development of infra-
structure there. The emphasis on conditions is important, as it implies a
distinction between existence of a development potential and its actual re-
alisation, which will by and large, depend on factors other than transport
and land use. As we will see, realisation of the potential may or may not
occur, and development can take different directions.
Five ideal typical situations can be distinguished in the model. Along
the middle line are balanced locations, where node and place values are
equally strong. At the top of the line are areas under stress. Here the in-
tensity and diversity of both mobility flows and urban activities is maxi-
mal. This indicates that the potential for land use development is highest
(strong node) and that it has been realised (strong place). The same can be
said about the potential for transport development. However, these are also
locations where the great concentrations of flows and area-based activities
mean that there is an equally great chance of conflicts between multiple
claims on the limited space and that further development might become in-
creasingly problematic. At the bottom of the middle line is a third ideal
typical situation, represented by the dependent areas.
The struggle for space is minimal here, but the demand for transporta-
tion services from area residents, workers and other users, and the demand
for urban activities from travellers are both so low that supply can be held
in place only by the intervention of other factors such as peculiarities in the
topography of the area or in the morphology of the transportation net-
works, external subsidies, etc.
Finally, two unbalanced situations can be identified. On one side, at
the top left of the diagram, are the unbalanced nodes, areas where trans-
portation supply is relatively much more developed than urban activities
(think for instance of a newly opened out-of-town railway station). On the
other side, at the bottom right of the diagram, are the unbalanced places,
where the opposite is true (consider an historic, relatively difficult to ac-
cess urban neighbourhood).
The latter two are particularly interesting location-types. We can assume
that, following the transport land use feedback cycle, they will show a
strong tendency to move towards a more balanced state. However, and
crucially, this could always happen in two radically different ways. An
unbalanced node could either increase its place-value (for instance by at-
tracting property development) or decrease its node-value (perhaps
through reduction in the level of transportation services). A reverse reason-
ing can be applied to an unbalanced place: either the level of connection
will be increased or a lower density, and possibly qualitatively different
functional mix, will be developed. The emergence of unbalanced nodes
and places, either as a deliberate policy move, or as the result of autono-
mous trends, can be seen as an essential factor in the development of the
urban transport and land use system: without unbalanced situations, there
will be no change at all. At the same time, the fact that the system can re-
act in different ways means that different, or even divergent development-
paths are possible.
40 Luca Bertolini
3.2.2 Development paths
With the help of the node-place model both up- and down-grading proc-
esses of either single locations or of entire urban networks can be identi-
fied. To illustrate this, let us now consider an application of the model to
station areas in the Amsterdam and Utrecht urban regions in the Nether-
lands (Figures 3.2 to 3.5). For our purposes here, the node and place di-
mensions have been translated into a node- and a place-index, each com-
bining different variables by means of a multicriteria analysis (MCA). The
node-index is a measure of the accessibility of the station area. Intensity
and diversity of transport supply are the key criteria here. The index com-
bines accessibility by train (number of directions served, daily frequency
of services, amount of stations within 45 minutes of travel), by bus, tram
and subway (number of directions, daily frequency), by car (distance from
the closest motorway access, parking capacity) and by bicycle (number of
free-standing bicycle paths, parking capacity).
Figure 3.2. The Amsterdam and Utrecht urban regions in 1997 and 2005.
1,0 AC
0,8
AS
Ha
0,6
AA AZ
Du AL
Sch
Za
Hd AB
0,4 We AM AR
DZ
AV
Di KB
0,2 Ab
0,0
0,0 0,2 0,4 0,6 0,8 1,0
Legend
Amsterdam AA Amsterdam AR Diemen Di Koog KB
Amstel RAI Bloemwijk
Amsterdam AB Amsterdam AS Diemen-Zuid DZ Schiphol Sch
Bijlmer Sloterdijk Airport1
Amsterdam AC Amsterdam AV Duivendrecht Du Weesp We
CS Vlugtlaan
Amsterdam AL Amsterdam AZ Haarlem Ha Zaandam Za
Lelylaan Zuid/WTC
Amsterdam AM Abcoude Ab Hoofddorp Hd
Muiderpoort
1. No place data available
Figure 3.3. Application of the node-place model to station areas in the Amsterdam
region, 1997.
The place-index is a measure of the intensity and diversity of activities

in the station area. The station area has been defined as the surface in-
cluded within a walkable radius of 700 metres from the main pedestrian
entrance to the station. The variables are the number of residents in the
area, the number of workers per each of four economic clusters (re-
tail/hotel and catering, education/health/culture, administration and ser-
42 Luca Bertolini
vices, industry and distribution) and the degree of functional mix (for
methodological and technical details see Zweedijk 1997; Serlie 1998).
1,0 UC
0,8
0,6
0,4
Dr
UO Wo
0,2 UL Br Ho Bu
DD Bi Ma
HR Vl
0,0
0,0 0,2 0,4 0,6 0,8 1,0
Legend:
Bilthoven Bi Driebergen Dr Utrecht CS UC Woerden Wo
Breukelen Br Hollandse HR Utrecht Lunet- UL
Rading ten
Bunnik Bu Houten Ho Utrecht Over- UO
vecht
Den Dol- DD Maarssen Ma Vleuten Vl
der
Figure 3.4. Application of the node-place model to station areas in the Utrecht re-
gion, 1997
In the Amsterdam urban region (Figure 3.3) there are both examples of
station areas under stress, of dependent locations and several unbalanced
nodes and places. Patterns of development in the 1997-2005 period seem
to confirm the characterisation of the different locations: for instance, un-
der stress Central Station (Amsterdam CS) has been struggling with the
great complexity of further development there, and has accordingly, lost
some of its supremacy; dependent Amsterdam Vlugtlaan has been closed
down because the station was deemed not economically viable by the rail-
way company; and it is the unbalanced station areas that have shown the
greatest development dynamics. Both striking upgrading processes and

downgrading processes have taken place here.
Legend:
Amsterdam AA Amsterdam AR Diemen Di Haarlem HS
Amstel RAI Spaarnw.
Amsterdam AB Amsterdam AS Diemen-Zuid DZ Koog KB
Bijlmer Sloterdijk Bloemwijk
Amsterdam AC Amsterdam AV Duivendrecht Du Schiphol Sch
CS Vlugtlaan Airport1
Amsterdam AL Amsterdam AZ Haarlem Ha Weesp We
Lelylaan Zuid/WTC
Amsterdam AM Abcoude Ab Hoofddorp Hd Zaandam Za
Muiderpoort
1. No place data available
Figure 3.5. Application of the node-place model to station areas in the Amster-
dam region, 1997-2005.
44 Luca Bertolini
Examples of the former are the developments of the station areas of

Amsterdam Sloterdijk and Amsterdam Bijlmer, which have gained on both
node and place index, but also Schiphol Airport, with a marked gain on the
node index (and no data available for the place index). An example of
downgrading is Amsterdam Zuid, which has lost significantly on both
node and place index. New unbalanced situations have also appeared in
this area of the diagram, as in a number of station areas sharp gains on the
place index have been matched by much smaller gains, or even losses on
the node index1.
In the Utrecht urban region (Figure 3.4) there is even less variety of sta-
tion area types. The system is dominated by a single location (Utrecht CS).
The rest follows at great distance; also in the Utrecht region, development
patterns since 1997, while not quantified, seem to confirm the characterisa-
tion of locations. Development efforts have been mainly directed at
Utrecht CS. However, the difficulty of transformation there has meant that
not much has been achieved on the ground. At the same time, little if any
development activity has touched the many dependent station areas in
the region.
From a system perspective, most interesting are the cumulative patterns
of development at the level of the two urban regions (compare Figures 3.3
and 3.4). The Amsterdam region shows a clustering in the centre of the
balance line, the Utrecht agglomeration a clustering at the bottom, with
the sole exception of Utrecht Central Station. This is a crucial difference.
As already contended, areas at the very top of the middle line can offer
high opportunities for development, but may also instigate the most intense
conflicts. The (relative) borders of growth will be reached there before it is
reached in areas with lower node and place values. It then becomes impor-
tant to have in supply alternative areas with adequate public transport con-
nections if growth is not to be diverted to areas not well-connected to the
railway network. Amsterdam has those alternative station areas; Utrecht
does not. As a result, in Amsterdam the polycentric, public transport-
oriented pattern is being reinforced (Figures 3.2 and 3.5), while in Utrecht
development has concentrated in peripheral areas with poor public trans-
port access (Figure 3.2).
1
The author wishes to thank Angelique Klinkers for help in updating the origi-
nal analysis to 2005.
3.2.3 Conditions for change
Which factors might be behind these different development paths? In a

situation where many broader context factors are similar (consider eco-
nomic development, cultural orientation, planning regimes, etc.), a striking
difference between Amsterdam and Utrecht is that between the morphol-
ogy of the two railway networks: a combination of a radial and a tangential
network in Amsterdam and a strongly radial network in Utrecht (Figure
3.2). The more articulated railway network in Amsterdam seems to have
been an essential condition in allowing the emergence there of a public
transport-oriented polycentrism. This fact supports the frequent plea for
the development of public transport orbitals (Hall and Ward 1998). How-
ever, it also stresses how the shaping of conditions conducive to a particu-
lar urban transport and land use development pattern is a long-term, only
partially controllable process.
The present network morphology in Amsterdam is the result of a very
long chain of decisions and actions, often unknowingly or unwillingly con-
tributing to the final result (Poelstra 2003). This is especially the case with
the development of its most crucial piece: the railway tangents. Decisions
and actions contributing to this development include land reservations for
a never materialised railway freight line around the city as early as the
beginning of the 20th century, and the opening starting in the 1970s and
profiting from those rights-of-way of new links to connect Schiphol air-
port to the national railway network. The desirability of locations along the
newly emerged railway tangents was further and decisively reinforced by
the realisation envisaged since the 1960s and implemented since the
1970s of a motorway ring along the same routes. Intriguingly, most if not
all aforementioned developments were unrelated to any deliberate attempts
to develop a multi-modal, polycentric urban system in Amsterdam and the
region. The only major transport infrastructure that consciously sought to
support such a pattern is the metro ring line running along the railway
tangents and the motorway ring, and connecting all the major peripheral
development centres, which opened as late as 1997.
If it took Amsterdam so long, including a not-insignificant portion of
chance, a more than legitimate question is also whether Utrecht should,
and could, pursue a similar development pattern, all of its theoretical ad-
vantages notwithstanding. The gap between model and reality may be such
in that case that an alternative, and possibly more incremental strategy, is
plausible. Intriguingly, this is the approach adopted by the Utrecht regional
transport plan now under discussion, where an incremental development of
public transport tangents is proposed. The idea is to begin with bus ser-
vices during rush hour, and move later to regular bus lines, leaving the op-
46 Luca Bertolini
tion of building dedicated infrastructure open and dependent on the evolu-

tion of actual demand (Bestuur Regio Utrecht 2003).
What is true of the differences between Amsterdam and Utrecht is, ar-
guably, even truer when the transfer of solutions between cities in different
national or even continental contexts is contemplated. This recognition
should lead to serious questioning of much literature on sustainable devel-
opment. How much of a solution is, for instance, Transit Oriented Devel-
opment of the type being proposed by Calthorpe (1993) and long applied
in European cities as Stockholm and Copenhagen (see next section) for
car-dependent cities in North America? Havent the United States and
Canada gone too far in dispersing land use? And how about exploding, re-
source-poor cities in the developing world? Will they ever have enough
means to develop the extensive public transport networks required by such
a model? The point is to not rule out that the answers could still be in the
affirmative. However, the great complications and the very long-term na-
ture of any transition to a significantly different transport and land use pat-
tern should be more carefully appreciated than in much current debate. At
the same time, greater effort should be exerted in understanding the poten-
tial for more incremental, more situation-specific, less orthodox strategies.
Examples of these are successful public transport-based approaches in cit-
ies as diverse as Ottawa in Canada, Curitiba in Brazil, Karlsruhe in Ger-
many, or Mexico City (all discussed in Cervero 1998).
Both network and area specific conditions and historical contingencies
are responsible for the different development paths described above. From
a strategic point of view, the crucial question is if a (re)orientation of urban
development towards the railway network can also be more proactively
supported, and how. In order to find an answer, alternative development
strategies are compared in the next section. With a focus on experiences in
Europe since the post-war period, two possible approaches can be distin-
guished, respectively a coordinated development and a reconnecting de-
velopments approach. They are illustrated by means of the examples of
Stockholm (coordinated development), and Karlsruhe, Naples and Rotter-
dam-The Hague (reconnecting developments).
3.3 The search for effective network development

strategies
3.3.1 Coordinated development: Stockholm
Stockholm, Sweden is a classic example of public transport-oriented de-

velopment; in many ways it has literally anticipated the principles being
propagated in much of the recent literature (e.g., Calthorpe 1993; Rogers
1997; Hall and Ward 1998) and strongly reminiscent of established transit
metropolises elsewhere (Copenhagen, Munich, Tokyo, or Singapore
discussed in Cervero 1998). At the heart of Stockholms strategy (Figure
3.6) is a radial subway system emanating from the central city, and new
towns built along the lines. Importantly, the latter not only has a residential
but also an employment function, favourable to an efficient bi-directional
use of the public transport network at all times. The Stockholm model is
highly successful in terms of public transport use. With a public transport
share of 28% of all trips (METREX, 2006) and of 44% of home to work
trips (Office of Regional Planning and Urban Transportation 2006) the
Stockholm region is one of Europes most public transport oriented me-
tropolises.
Contributing to these results are not just the general principles sketched
above, but also their consequent application at every spatial scale. Public
transport systems (bus, tram, train) all converge at subway stations, thus
allowing seamless connections. Public transport fares are kept low while
car use is made relatively less attractive (for instance, by expensive, lim-
ited parking facilities). On the land use side, both densities and functional
mix are highest around stations. It is not merely a matter of quantity: sta-
tion areas are the very centres of urban life, especially in the new towns.
Here retail and services are concentrated and public events are held, while
the carefully manicured open spaces are natural meeting points for inhabi-
tants and visitors alike.
The Stockholm approach rests on quite unique institutional conditions,
including a strong role of the central city in the metropolitan region, land
largely in public hands, and an urban development long led by the public
housing sector. While these conditions are important, it should be noted
that not dissimilar development patterns have also emerged in very differ-
ent institutional contexts (for example, market-led Tokyo). However, and
quite crucially, Stockholm shares with these other contexts the great conti-
nuity of the strategy, which has by now been consistently applied for about
60 years.
48 Luca Bertolini
Source: adapted from Cervero (1998)
Figure 3.6. Stockholm, the essential ingredients of the development strategy.
All past successes notwithstanding, the Stockholm model presently also

shows significant tensions, which are essentially due to two developments.
The first is the sheer expansion of the urban system, meaning that dis-
tances are becoming too long to be covered by subway within an accept-
able time. The second development is the increasing amount of mobility
demand between new towns (rather than between new towns and the cen-
tral city); this demand is difficult to cater to with public transport, and cer-
tainly with such a high capacity, low flexibility system as the subway. Po-
tential solutions are being explored; a light rail orbital is being developed
to cope with tangential relationships (albeit serving the central city fringe
rather than new towns). A high-speed train system traversing the entire
Mlaren region to the east of Stockholm has been developed to cope with
the up-scaling of the urban system (see chapter 14 for an evaluation of this
line). Third, the focus of land use policies has shifted from the design of
new urban areas to the densification and diversification of existing ones,
following the motto build the city inwards (Office of Regional Planning
and Urban Transportation 2006). It is too soon to say whether these provi-
sions will be enough. The outcome remains uncertain, not lastly because it
has also become increasingly difficult to coordinate development effec-

tively from one centre in such an enlarged, multi-nuclear urban system.
In conclusion, while successful, the coordinated development approach
also shows emerging tensions (at least in its monocentric, strongly hierar-
chical version), and above all appears dependent on very specific local
conditions, most importantly including multiple decade policy continuity.
Other successful examples of coordinated development in Europe (such as
Copenhagen and Munich) also support this last point. What decision to
take then in the large majority of European cities without such a long-
standing tradition? An alternative approach seems to be emerging there,
which takes spatial and institutional fragmentation as its starting point, is
more incremental, and attempts to achieve greater transport land use inte-
gration combining top-down and bottom-up initiatives (so-called multi-
level governance). This second approach is entitled reconnecting devel-
opments and is much less formalised in practice as well as theory, but it
deserves more recognition, as it better fits the present conditions of many
European cities. Three examples will be discussed below: Karlsruhe in
Germany, Naples in Italy and the Rotterdam-The Hague area (or the South
of the Randstad) in the Netherlands.
3.3.2 Reconnecting developments: Karlsruhe, Naples and

Rotterdam-The Hague
In the 1990s, when Karlsruhe realised it needed a more public transport-

oriented development strategy, it was already a heavily decentralised urban
system, with the central city accounting for only 300 000 of the 1,1 million
inhabitants of the metropolitan area, and a fragmented public transport sys-
tem lacking clear hierarchies. In this context the greatest weakness of pub-
lic transport was its poor connectivity, both internal (connectivity between
different public transport systems) and external (connectivity between pub-
lic transport and places of activity). Due to limited resources and institu-
tional constraints, a grand plan for a whole new system was, however,
not a feasible option. However, on the positive side, there was an extensive
railway network with much under-utilised capacity. The solution was
eventually found in a strategy consisting of three main elements (Figure
3.7): i.) a limited number of new rail links improving the internal connec-
tivity of the public transport system, ii.) new stations improving its exter-
nal connectivity, and iii.) new light-rail vehicles combining the features
of trams (more limited disruption, best suited to the intra-urban sections)
and trains (relatively high-speed, best suited to the extra-urban sections).
On the institutional side and as in Stockholm, the core municipality of
50 Luca Bertolini
Karlsruhe has had a pivotal role in promoting this metropolitan strategy.

Differently than in Stockholm, this was not so much attributable to its rela-
tive weight or dominant position, but rather through the effective employ-
ment of its ownership of the urban and regional public transportation
agency.
Source: adapted from Jefferson & Khn (1996)
Figure 3.7. Karlsruhe, the essential ingredients of the development strategy

(schematic, not to scale).
This strategy has been highly successful in terms of modal shift, with
Karlsruhe being one of the few German cities where public transport use is
growing, and most importantly largely at the expense of the car (60% of
those commuting by public transport are former car drivers: VBK and
ABG 1996).
The second example comes from a very different context: Naples, Italy.
Here too, however, the situation has parallels with Karlsruhe in that it was
one of great network fragmentation (both physically and in fare structures
and timetables) and disconnection between urban developments and public
transport systems. Also here an extensive railway network inherited from
the past was regarded as an opportunity to develop more integrated sys-

tems. The main elements of the adopted transport strategy remind us of the
approach in Karlsruhe as well: strategic links, more stations (from 45 to
98) and more interchanges (both rail-rail: from 5 to 18, and rail-car: from 0
to 16: Cascetta 2000). Currently this strategy is being up-scaled to the
whole of the Campania region. Fare and timetable integration has been the
first step, already resulting in a 35% passenger increase in 2001-2004
(Pagliara and Cascetta 2006). More proactively than in Karlsruhe, how-
ever, a land use strategy matches the transport strategy. Its essence is a re-
focus of vast urban redevelopment efforts of the metropolitan area around
the stations of the railway network (Figure 3.8). The institutional context
also differs, in that strong leadership and coordination between municipal-
ity and region has been crucial (made possible by broader institutional re-
forms in Italy, and most notably the direct election and increased auton-
omy of the mayor of the municipality and the governor of the region).
Source: adapted from Papa (2004)
Figure 3.8. Naples, the essential ingredients of the development strategy (central
section only).
52 Luca Bertolini
The third and final example comes from the South of the Randstad (Rot-
terdam-The Hague area), in the Netherlands. The South of the Randstad
has been growing in a highly uncoordinated way in the last 30 or so years.
While developments have ironically, followed nationally mandated com-
pact city principles, a very uneven distribution of new built up areas in fa-
vour of secondary centres in the region, coupled with the very short dis-
tances between different centres, has resulted in extensive decentralisation
of activities and diffusion of mobility patterns. The current accessibility
and sustainability problems and a general lack of coherence of the urban
system are associated with these developments and in need of urgent atten-
tion.
The emerging strategy resembles that described for Naples and is
strongly reminiscent of the strategic opportunism (a term borrowed from
Frank le Clercq, personal exchange) already noted in the Naples and
Karlsruhe cases. In the South of the Randstad the trigger for identifying a
potential solution has been the realisation that capacity be freed up on the
national railway network (due to the development of new dedicated high-
speed and freight links and the dismissal of short-range passenger services)
to develop a new regional public transport network. As in Karlsruhe and
Naples, a limited number of strategic links (rail and bus) improving the in-
terconnectivity of the network and new stations and better suited to the
dispersed nature of development, are constituent elements of the unfolding
approach (Figure 3.9). Less visibly but not less importantly, higher fre-
quencies, integration with local rail (tram, metro) and road-based public
transport services, also including the car system mainly via carefully lo-
cated park-and-ride facilities are also a part.
On the land use side the strategy shows direct parallels with Naples, but
adds a systematic note by demonstrating the amount and type of spatial
program to be developed at each station-node with its relative position in
the emerging multi-modal regional transportation network. A node-place
model based on Bertolini (1999; see above) has been applied for the pur-
pose (Platform Zuidvleugel 2003). The strategy is now entering the im-
plementation phase, which, in its manifestation, also aptly typifies the re-
connecting developments approach: not so much a grand plan but a
limited number of station-pilots selected on the basis of real, short-term
development opportunities identified together with local stakeholders (mu-
nicipalities and private developers). The institutional dimension also has
its specificities: leadership has been taken by the Province of South-
Holland, in close collaboration with the main municipalities of Rotterdam
and The Hague, and intense interaction with such crucial stakeholders as
smaller municipalities, the national government, and the national railway
company.
Source: adapted from Platform Zuidvleugel (2003)
Figure 3.9. Rotterdam-The Hague, the essential ingredients of the development

strategy.
3.3.3 Comparing the two strategies
The coordinated development approach has been very successful in the

(few) European cities able to coherently maintain it throughout the second
half of the 20th century. Next to Stockholm, the examples of Copenhagen
and Munich can also be cited. However, even in these classic examples,
de- and re-centralisation processes are challenging the approach, as shown
in the Stockholm case. Even more importantly, the coordinated develop-
ment approach appears not to be an option for European cities that are only
now addressing the problem. The hypothesis advanced at the end of this
discussion is that, next to other aspects, a determinant factor of the viabil-
54 Luca Bertolini
ity of a development strategy is the phase of a city in its urban lifecycle.

There is in Europe a huge difference between the rapid urban growth of the
post-war boom and the much slower pace of the last decades. In these
more mature phases, cities that have not completely adopted a coordinated
development strategy seem poised to gain much from the reconnecting de-
velopments strategy illustrated by Karlsruhe, Naples and the Rotterdam-
The Hague region. In so doing, they may also be able to address problems
arising from the classic, strongly hierarchical monocentric model typified
by Stockholm, and thus learn to cope with an urban reality increasingly
characterised by multiple, unstable hierarchies, and multi-directional
flows, both in the spatial and institutional sense. In the meantime, identify-
ing the salient distinguishing features of the two approaches may help to
focus the debate. A tentative attempt in this direction is carried out in Ta-
ble 3.1.
Table 3.1. The two ideal typical approaches compared

Coordinated development Reconnecting develop-
ments
Spatial scale Urban Regional
Spatial hierarchy Vertical, strong Horizontal, weak
Urban morphology Monocentric Polycentric
Infrastructure morphology Radial (first phase); Organic grid
Radial with tangents (sec-
ond phase)
Transport system Mono-modal (rail) Multi- and inter-modal
(rail, bus, car)
Urban growth rate Fast (as in post war Europe) Slow (as in contemporary
Europe)
Institutional hierarchy Strong Weak
(core-municipality domi-
nated)
Inter-institutional interac- Limited Extensive
tion (across scales, sectors,
domains)
Policy discourse Growth management Sustainable development;
Metropolitan competitive-
ness
Planning philosophy Comprehensive, plan-drivenIncremental, project-driven
(strategic opportunism)
Temporal horizon Long-term Short-term
3.4 Conclusions
Proactively supporting a re-orientation of urban-regional development to-

wards the railway network and its nodes requires addressing the complex,
urban-region-wide node-place dynamics of stations areas. The node-place
model introduced in the first part of the chapter can be seen as an analyti-
cal tool to help gain insight into the dynamics, raise questions about its un-
derlying factors, and explore the scope for intervention. Application of the
model to the Amsterdam and Utrecht urban regions has, for instance,
shown how the more articulated railway network of the Amsterdam region
may have both supported and been reinforced by a polycentric, station
area-focussed urban development pattern. At the same time, an enquiry
into the way this network emerged has warned against a too-direct transfer
of the strategy to other contexts. It has been contented that, in other situa-
tions including Utrecht, a lighter, incremental approach towards enhanced
network articulation may be better suited.
Further elaborating on this discussion, in the second part of the chapter,
two alternative, ideal-typical ways of linking the development of the urban
system to the development of the railway network were discussed: a coor-
dinated development approach, typified by Stockholm, and a reconnect-
ing developments approach, illustrated by Karlsruhe, Naples and Cam-
pania, and the Rotterdam-The Hague area. While the coordinated
development approach has a long and successful history in some Euro-
pean cities, the examples illustrating the reconnecting developments ap-
proach are still in their early stages. It is too soon to determine whether
they will achieve their expected trend-breaking status. Whatever the out-
come, the variety of solutions they document suggests that it is not the spe-
cific form of development which is crucial here, but rather it is the degree
of consistency between land use and transport policies that matters most.
The content of these policies can, and should be different, as they respond
to specific and continuously evolving local conditions (consider the new
challenges posed by poly-centrism and up-scaling of the urban system in
Stockholm).
The above means that the universal models on which some current litera-
ture tend to focus (e.g., Calthorpe 1993; Rogers 1997; Hall and Ward
1998) should not be embraced uncritically, and that the diversity of ap-
proaches central to other studies (e.g., Cervero 1998) could be a more use-
ful source of inspiration. Identifying a workable and working strategy
seems to require understanding what works in a given, historically-grown
situation, with all its unique constraints (such as a strongly decentralised
urban system) but also its opportunities (an existing under-utilised national
56 Luca Bertolini
railway network). Furthermore, in all contexts there seems to be both a

need and a possibility to identify relatively marginal changes (such as a
few extra links) that may achieve system-wide radical changes (as a much
more integrated network). This is in any event the hope, at least already
supported by some evidence (consider the modal shifts in Karlsruhe and
Naples), indicated by the last three cases outlined above.
References
Berg L van den, Pol P (1998) The European High-Speed Train and Urban Devel-
opment. Ashgate, Aldershot
opment. European Planning Studies 4:331345
Bertolini L (1999) Spatial Development Patterns and Public Transport: the Appli-
cation of an Analytical Model in the Netherlands. Planning Practice & Re-
search 14:199210
Areas. E & FN Spon, London
Bestuur Regio Utrecht (2003) Regionale Verkeers- en Vervoesplan. Bestuur Regio
Utrecht, Utrecht
Calthorpe P (1993) The Next American Metropolis: Ecology, Community, and the
American Dream. Princeton Architectural Press, Princeton
Cascetta E (2000) La Metropolitana nel Progetto della Mobilit di Napoli, In:
Spadea, R (ed) La Metropolitana di Napoli. Nuovi Spazi per la Mobilit e la
Cultura. I: Electa, Napoli
Cervero R (1998) The Transit Metropolis; a Global Inquiry. Island Press, Wash-
ington DC
Cervero R (ed) (2004) Transit-Oriented Development in the United States: Ex-
periences, Challenges and Prospects. TCRP Report 102, Washington DC
Dittmar H, Ohland G (eds) (2004) The New Transit Town: Best Practices in Tran-
sit-Oriented Development. Island Press, Washington DC
Dunphy R, Cervero R, Dock F, Mc Avey M, Porter D (2005) Development
Around Transit. Urban Land Institute, Washington DC
Hall P, Ward C (1998) Sociable Cities: the Legacy of Ebenezer Howard. John
Wiley & Sons, Chichester
Hanson S (ed) (1995) The Geography of Urban Transportation. The Guilford
Press, New York/London
Jefferson C, Khn A (1996) Multimodal LRT Vehicles: a Development for the
Future? In: Recio J, Sucharov L (eds) Urban Transport and the Environment
II. Urban Transport 96, Barcelona
Lee DB (1973) Requiem for Large-Scale Models. Journal of the American Plan-
ning Association 39:163178.
Lee DB (1994) Retrospective on Large-Scale Urban Models. Journal of the
American Planning Association 60:3540.
Manheim ML (1974) Fundamentals of Transportation Systems Analysis, Volume

1: Basic Concepts. MIT Press, Cambridge
Meyer MD, Miller EJ (2001) Urban Transportation Planning, Second Edition.
McGraw-Hill, New York.
METREX (2006) Region of Stockholm. www.Eurometrex.org/Docs/eAtlas/
STOCKHOLM eAtlas.pdf
Office of Regional Planning and Urban Transportation (2006) The Stockholm Re-
gion: Regional Planning and Development Review. Stockholm County Coun-
cil, Stockholm
Pagliara F, Cascetta E (2006) Integrated Land Use and Transport Planning: the
Metro Systems Project of Naples and Campania. Paper presented at the con-
gress: Transportation planning: A policy design challenge? Amsterdam
Papa E (2004) Land Use Transportation Planning Policies: the Naples 100 Sta-
tions Plan and the Case of Municipio Station. Paper presented at the AESOP
2004 congress, Grenoble
Platform Zuidvleugel (2003) De Stedenbaan. Provincie Zuid-Holland, Den Haag
Poelstra H (2003) Eerst Infrastructuur, Dan Beleid. In: Dienst Ruimtelijk Orde-
ning Gemeente Amsterdam (ed) Stadsplan Amsterdam 1928-2003. Nai uit-
gevers, Rotterdam, pp 118129
Rogers R (1997) Cities for a Small Planet. Faber & Faber, London
Serlie Z (1998) Stationslocaties in Vergelijkend Perspectief. Master Thesis, Uni-
versiteit Utrecht, Utrecht
VBK, ABG (1996) Report 95. Verkehrsbetriebe Karlsruhe and Albatal Verkehrs-
Gesellschaft mbH, Karlsruhe
Vonk G, Geertman S, Schot P (2005) Bottlenecks Blocking Widespread Usage of
Planning Support Systems. Environment and Planning 37:909924
Wegener M, Frst F (1999) Land-Use Transport Interaction: State of the Art. In-
stitut fr Raumplanung, Dortmund
Zweedijk A (1997) Knoop of Plaats? Naar een Operationalisering van het Begrip
Stationslocatie. Master Thesis, Universiteit Utrecht, Utrecht
4 HST stations and urban dynamics:
Experiences from four European cities
Peter Pol
Department of Applied Economics, Erasmus University Rotterdam,

the Netherlands
4.1 Introduction
Connection to the High-Speed-Train (HST) network is, for many cities, a

momentous occasion towards redeveloping their station areas. Many local
authorities have drawn up ambitious plans for the redevelopment of the
zones at and around HST nodes. The expectation is that these redevelop-
ments will enhance the accessibility and economic attractiveness of metro-
politan centres, and will attract new urban activities and residents (Bonna-
fous 1987; Plassard 1991). To the actors involved, the advent of the HST
can be an opportunity to renew and strengthen the urban economy, to
change the modal split in favour of more environmental-friendly modali-
ties, and improve the image of the metropolitan centre and its urban re-
gion.
In this chapter we focus on two types of cities: Cities in Transition
(CiTs) and International Service Cities (ISCs). The former are often old
manufacturing or port cities aiming to diversify their economic structures
through efforts to attract new economic activities and inhabitants. CiTs are
e.g., Lige, Lille, Liverpool, and Rotterdam. ISCs have a competitive edge
in the international service and knowledge economy, because of their high-
grade (international) facilities, attractiveness and accessibility.1 Examples
include London, Paris, Amsterdam, and Munich. These two archetypes of
cities have been defined because we think they are relevant towards better
understanding the situation of HST cities in the European context. Ours is
1 See Beaverstock et al. 1999. They would classify the ISCs from this chapter
as Gamma World Cities and the CiTs as cities with evidence of world city forma-
tion.
60 Peter Pol
not an exclusive distinction, but the cities along European HST lines can
often be categorised with the help of these archetypes (Pol 2002). This
chapter will refer to the results of four case studies: two ISCs, Amsterdam
and Munich, and two CiTs, Lille and Rotterdam (Figure 4.1).
Amsterdam
Rotterdam
Lille
Munich
Source: Pol, 2002.
Figure 4.1. Map with the case studies and HST lines
The central question of this chapter is: how do urban actors anticipate
the effects of the advent of the HST? From this the following sub-
questions are derived: i.) what are the conditions for economic benefits for
cities linked to the HST network; ii.) what are important organisational is-
sues; and iii.) what do we see in practice? The chapter is built on several
international comparative urban studies (van den Berg and Pol 1998, 1999;
van den Berg et al. 2001; Pol 2002). For each of the case studies dealt with
here, we carried out 10 to 15 in-depth interviews with key actors.
We begin by analysing the conditions for the economic effects of the
HST. Organisational issues of station redevelopments, structured by the
four themes: vision, support, leadership, and strategic networks, are then
investigated; the four cases are briefly discussed and rounded out in our
conclusions.
4.2 Conditions for economic effects of HST linkage
The HST expands the relevant regions of cities because community dis-
tances are enlarged by faster transport. As a result, the HST encourages the
HST stations and urban dynamics: experiences from four European cities 61
formation of new economic gravity centres, because transport benefits are

largest in the direct vicinity of HST stations. Many of the newly attracted
firms and residents prefer settling closest to the station in the HST station
area which offers the greatest gains in accessibility. Schtz (1996) defines
in this respect the primary development zone, a zone within five to 10
minutes reach of the HST station. According to Schtz, the largest effects
of the advent of the HST can be expected in the primary development
zone. It is indeed here that most travel time can be saved because, in prin-
ciple, the traveller requires no complementary transport. The location at-
tractiveness of the station surroundings depends on its accessibility via
other transport infrastructure such as highways, airports and conventional
public transport. If these transport systems are mutually well-connected,
the station surroundings can become highly attractive locations for firms
and residents who attach great value to high intermodal and international
accessibility. Locating close to this area will help to enhance the maximum
communication distance for e.g, two-earner households (who realise
greater freedom in finding jobs in a larger relevant region), and for interna-
tional firms that can serve a larger market.
Moreover, by its proximity to the HST network, the area profits directly
from its improved status as a location. That is why particularly in the pri-
mary development zone new firms and residents can be attracted, and
where increases in land and real estate values are also expected. In addi-
tion, at more remote distances from the HST station (at secondary and ter-
tiary development zones), effects of the advent of the HST are noticeable,
though smaller, than at primary development zones. This can be seen for
instance in the case of Lille, where many private and public investments
were attracted to the inner city after the new HST station was completed
(van den Berg and Pol 1998).
The HST connection may affect the regional economy in two ways by
either acting as a catalyst or by having a facilitating role. A catalysing role
is achieved when the connection to the HST network draws new activities
to an urban region in transition and thus causes the economy to burgeon.
By a facilitating role, we mean its impact on cities with a prosperous local
economy, which requires new infrastructure to accommodate their eco-
nomic growth. The connection to the HST network signals a momentum
for urban actors to further invest in station areas. The net effects of an HST
connection depend on whether these investments fit adequately to per-
ceived discrepancies regarding the station area and its surroundings as well
as precipitate new opportunities. Research results show that the HST has
spatial effects that can be positive when used explicitly and are directed by
city planners (Schtz 1996; Troin 1995). Conversely, when no additional
investments are being made, cities are more likely to face no spatial
62 Peter Pol
changes related to the advent of the HST, or worse, can experience back-
wash effects.
Urban actors will normally invest in four elements of station develop-
ments: the node, the place (Bertolini and Spit 1998), spatial quality, and
image (Pol 2002). Investment in the node means upgrading the terminal
and optimising the pivotal function and the complementary accessibility.
Investment in the place means creating space, infrastructure and super-
structures for new urban activities. Investment in spatial quality has to do
with the configuration and appearance of the public space, the embedded-
ness in the surrounding urban fabric, and the architecture of the buildings.
Investment in the image means explicitly using the advent of the HST and
the related investments to improve the image of the urban region and the
status of the urban area.
All four elements will normally be aspects of station development
strategies. The element that weighs heaviest in a given city depends on ex-
pectations elicited by the connection to the HST network, and on the spe-
cific features (opportunities and problems) of a station area. The assump-
tion is that, notably, investments in spatial quality and image enhancement
contribute to the distinguishing features of an urban region, because they
have often the largest impact on the mental map of urban actors. If these
features are given sufficient attention, a city can have an edge on its com-
petitors in attracting new economic activities and residents.
4.3 Organising capacity regarding station developments
Efforts to develop station areas are complicated by the relatively large

number of actors involved, in particular local and national government, the
railway company, investors, other business firms, and residents. Each has
its own interests in the development of station areas, and goals they aim to
accomplish by the station development. Their interests may coincide, but
may also just as easily generate conflict.
All actors involved have certain resources necessary to develop station
areas. These are the means (in the broadest sense) which actors can con-
tribute to station development. Resources can be divided into four catego-
ries: finance, land, knowledge, and proceedings (decision-making, partici-
pation). The power that they can exercise during station development
mainly depends on these stakeholder resources.
When we consider the many actors involved in the redevelopment of
station areas, sufficient organising capacity is a conditio sine qua non. The
main actor has to cooperate with the others in order to achieve the aims of
the station redevelopment. This key actor in the development process

must, as much as possible, do justice to the interests of all relevant actors,
and make optimum use of the resources available (van den Berg et al.
1997 pp. 8-14). In view of this, we define organising capacity as the ability
of those responsible to use the advent of the HST as a springboard to con-
vene all involved partners, to jointly generate new ideas, to formulate and
implement strategies that respond to fundamental developments, and to
create conditions for sustainable urban economic growth.
According to van den Berg et al. (2003), key elements of organising ca-
pacity are vision, support, leadership, and strategic networks (Figure 4.2).
In the next sections these elements are subsequently explored. The explicit
introduction of the metropolitan context in Figure 4.2 reflects the necessity
for specific knowledge of local circumstances.
General metropolitan context
Vision Leadership
Support Strategic
networks
Based on: van den Berg et al. 2003
Figure 4.2. Key elements of organising capacity
4.4 Vision of an advisable station development
A sound comprehensive vision of the future of an urban area is a necessary

requirement for the development of a consistent strategy for station devel-
opment. A comprehensive vision is a prerequisite for the integration of
different aspects and the prevention of inconsistencies in development
strategies (van den Berg et al. 1997 p. 13). Such a vision should be based
on a rigorous analysis of the problems and opportunities of the area.
A vision should also be capable of acquiring broad support. To that end
it will have to be sufficiently realistic and reflect a cognizance of the op-
portunities and problems of the urban region. Conversely, it must do jus-
64 Peter Pol
tice in a balanced way to the interests and targets of all stakeholders, and
thus ultimately has to be a coherent and impartial framework of objectives
and strategies for the development of an urban area. The vision can, once
accepted as the common reference framework, help guide the relevant ac-
tors in their spatial behaviour.
4.5 Support for station development plans
No matter how valuable a project might be for sustainable urban develop-

ment, the lack of support from those directly involved or interested, par-
ticularly politicians and residents, may likely reduce its chances for suc-
cessful implementation (van den Berg et al. 1997 p. 13).
Political support is an important precondition for organising capacity,
since political relations and financial pre-conditions greatly affect the op-
portunities for initiating and implementing new policies. Political decision-
makers must be aware of the importance of the proposed development in
order to take a positive decision regarding both the area development as
such as well as its public financing. The better the area development
matches the policy priorities already selected by the government levels
concerned, the more political support will be mustered.
Since station areas are moreover normally surrounded by built-up areas
interspersed with dwellings, support from residents for redevelopment
plans is also crucial. If station development plans are not fully articulated
to the local population, the development of the station area could be de-
layed. The importance of the station development may be obvious to the
actors directly involved (project partners), but failure to communicate the
idea to residents may also reduce its chances for successful implementa-
tion.
4.6 Necessity for strong leadership
We believe that strong, decisive leadership by governmental, business or

other organisations is essential to initiate and implement adequate policies.
Every organisation or project needs a leading actor to initiate, continue and
complete it. The assumption is that the leadership of key actors contributes
substantially to the successful design, development and implementation of
projects. Leadership is a necessity, whether relying on specific competen-
cies (the position in the administrative hierarchy, financial capabilities,
specific know-how or other powers) or on the charisma of public or private
individuals who successfully drive the project (van den Berg et al. 1997
p. 12).
John and Cole (1999) developed a typology of leadership according to
two dimensions (Figure 4.3): One is the continuum between responsive
and proactive leaders. The other is the difference between leaders who
generate capacity and those who are self-regarding in a narrow sense. The
continuum is captured by Stones (1989) distinction between power over
and power to. The former is about the narrow exercise of power (a strong
hierarchical attitude); the latter is about exercising power by involving all
relevant actors (a cooperative, network attitude).
power to
consensual visionary
facilitator
responsive proactive
caretaker city boss
power over
Source: John and Cole 1999
Figure 4.3. Typology of leadership roles in the new urban governance
In a city where the actors involved expect a catalyst effect from the ad-
vent of the HST and where the necessary investment measures are taken in
a pro-active way with various public and private actors, there is likely to
be a visionary leader. We find a good example in the city of Lille, which
has made proactive use of the momentum of the advent of the HST. On the
contrary, a city that takes a wait and see attitude, and involves no other
actors in its policy, probably has a caretaker type of leader.
4.7 Strategic networks
The success of urban projects increasingly depends on the presence of re-

gional strategic networks between government, private companies and
other institutional structures. The presence of close-knit, homogeneous and
active coalitions is fundamental to the initiation of regional-economic dy-
66 Peter Pol
namics. The directing role of the government (governing) changes to a co-

managing role (governance). Strategic networks can be conceived as pat-
terns of interaction between mutually-dependent actors that evolve around
policy problems or projects. A network consists of the total relations be-
tween (public and private) organisations, institutions and persons, the rela-
tions being marked by a degree of two-way dynamics (van den Berg et al.
1997 p. 11).
A strategic network can compel actors to interact in a structural way and
is a means of arriving at a strategy for the station area that has relatively
broad support, does justice to the variety of social constructs, and takes ac-
count of the discrepancies pointed out by the actors involved. Robentraub
and Helmke (1996) argue that strategic partnerships occur in response to
geographically and technologically bound opportunities, potentials and
conditions. A region has to meet certain preconditions for economic
growth to stimulate actors to develop new strategies and policies in re-
sponse to opportunities and/or problems. These groups will then coordi-
nate their activities and affect, if not dictate, development patterns. In this
chapter, strategic partnerships spring from the momentum created by the
advent of the HST to European cities. In some cities this momentum is fur-
ther strengthened by relatively bad economic circumstances, for instance, a
high unemployment rate due to declining industrial activity (predomi-
nantly in CiTs).
Several types of strategic networks can be identified. We made a dis-
tinction based on regime literature (Stone 1989; Brindley et al. 1989; Di-
gaetano and Klemanski 1993, pp. 59-60). Important criteria in this respect
are the leading actor and the dominating aim of the partnership (based on
the developed vision). From this we can distinguish three types of possible
strategic networks for station developments:
Market-led pro-growth networks predominantly focus on gener-
ating long-term profits from the urban areas in which they plan
to invest. In these networks, private actors play a leading role.
Their strategy, often in partnership with local government, is to
provide the necessary infrastructure to attract economic activity,
residents and/or visitors, and thus generate profits;
Local government-led pro-growth networks focus on stimulat-
ing local economic development. The strategy of the leading lo-
cal authority, often in partnership with a railway company and
private partners, is to invest proactively in urban areas and en-
courage, promote and subsidise private investment through
grants, loans, land sites, and other inducements;
Balanced-growth networks try to encourage a balanced eco-

nomic growth in urban areas. These networks, which are nor-
mally led by local authorities, use planning and other local re-
sources and invest proactively in the urban region to achieve
spatial balance within the targeted urban region. The goals of
this strategy may be a spatially balanced poly-nuclear develop-
ment and a reduction of pressure on central urban areas. Fur-
thermore, they strive to balance growth and quality of life. Bal-
anced growth networks also attempt to exact social benefits
from urban growth. The expansion of affordable housing and
improvement of neighbourhood services may figure promi-
nently in these networks (Digaetano and Klemanski 1993).
In the empirical part below we discuss whether the aforementioned
types of strategic networks can be found in the chosen cases.
4.8 The Central Station in the city of Rotterdam
Urban context: The port city of Rotterdam (595 000 inhabitants, see Table
4.1) can be considered as a City in Transition (CiT). Though a medium-
sized city, the Port of Rotterdam is a global player with a strong market
position. Rotterdam authorities are keen to support the diversification of
the economy by stimulating promising clusters, such as health care, the
creative sector, media, and tourism. Most policy attention still focuses,
however, on port-related activities. HST link: In Rotterdam the HST al-
ready stops at the Central Station (CS). As from 2007, the city will be
linked to dedicated HST infrastructure. The central station area is being
redeveloped to improve in particular its capacity and spatial quality. The
new station will be completed in 2010.
Aim of the HST developments: The HST developments in Rotterdam,
are expected to have a catalysing effect on the urban economy, which aim
to attract new urban activities and residents. Under the initial auspices of
the Ministry of Housing, Spatial Planning and Environment, the city has
drawn up ambitious redevelopment plans for the station area. The Master
Plan for the redevelopment of the CS area, presented in April 2001,
showed elements that contribute to aims broader than merely promoting
local growth. Several sustainable urban development goals were also to be
achieved; for instance, the development of a diversified urban district
around the railway station, realising attractive (public) urban space, and re-
lated urban developments in the wider urban surroundings.
68 Peter Pol
Table 4.1. Key data of case-studies

City type City Inhabitants Dedicated HST stations
2001 HST-line (in italics: focus of study)
International Amsterdam 735 000 2007 Amsterdam Zuid, Cen-
Service Cities traal Station Schiphol
Munich 1.228 000 ? Hauptbahnhof, Pasing,
airport, Ostbahnhof
Cities in Lille 200 000 1994 Gare Europe, Gare de
Transition Flandres
Rotterdam 595 000 2007 Centraal Station
Source: Urban Audit II (www.urbanaudit.org); Pol (2002)
However, the lack of financial means to realise the ambitious plans and
local political changes since 2002, have together instigated new develop-
ment plans for the station area. These plans focus particularly on the de-
velopment of a new station terminal (and no longer on the entire station
area). By so doing, given the financial restrictions, more quality can be
achieved in the station terminal, which is in fact an important showpiece of
the city. The idea is that, with a high-quality station terminal, more private
investments can be lured. In the follow-up stage, the relatively high ambi-
tions for the area development are to be achieved by inviting private inves-
tors to create sound plans that meet the citys demands.
Actors involved: The city of Rotterdam is initiator and coordinator of the
redevelopment of Rotterdam CS. Because the municipality itself cannot
fund the entire process, it counts on other actors for supplementary re-
sources, in particular, national and regional government (Table 4.2). In the
follow-up stage, private investors are to play a greater role in (financing)
the redevelopment of the surrounding area.
Table 4.2. Actors involved in HST station area development

City type City Public Semi-public Private
International Amsterdam City, National gov- NS, Prorail Investors
Service Cities ernment
Munich City, Regional gov- DB
ernment
Cities in Lille City, Regional gov- SCETA, Chambers Banks
Transition ernment of Commerce
Rotterdam City, National and NS, Prorail 2nd stage:
Regional government Investors
Strategic network: The local authority in Rotterdam plays the leading

role and the development plans have been relatively ambitious for a CiT;
they appeared to focus on sustainable urban development in the long-term.

The strategic network involved could therefore be characterised as a bal-
anced-growth network. The risk however, was that these aims could not be
achieved because of limited financial resources. The solution is being
sought through a staging process: first to invest in a high quality station
terminal and, after its completion, to develop the surrounding area, at
which time private investors in particular are to be attracted.
4.9 Euralille in the city of Lille
Urban context: The municipality of Lille (200 000 inhabitants) is the core
of the Communaut Urbaine de Lille (CUdL, over 1M inhabitants). Lille
is an illustrative example of a European CiT; it has evolved from an indus-
trial to a modern economy with knowledge-intensive, service-providing
activities. In the 1970s the decline of companies in the metal and textile
sectors caused a flood of unemployment. Nowadays, the urban region of
Lille is one of the most export-oriented regions of France and trades heav-
ily with the UK. Growth industries in the region include tourism, commu-
nication and international services.
HST link: By deciding to run the route for the high-speed line from Paris
via the Channel Tunnel to London through Lille, the national government
boosted its chances for development. The Lille authorities decided to retain
the terminus station, Gare de Flandres, and built a new through station
nearby: Gare dEurope. HSTs from Paris stop at Gare de Flandres,
while Gare dEurope, opened in 1994, is a stopping place for some of the
Eurostar trains running between Paris and London and between Brussels
and London. A new multi-functional urban area, Euralille, was developed
in the zone between the two stations.
Aim of the HST developments: The arrival of the HST was expected to
attract new economic activities and visitors. As the station development
process advanced, emphasis shifted from the creation of a sound transport
node towards the development of a location fit to attract new urban activi-
ties. The strategy to develop Euralille as a grand projet was inspired by
the conviction that this was the momentum needed to reactivate the re-
gions economy.
Actors involved: The prominent actors behind Euralille are the local au-
thorities: the city of Lille and the CUdL, representing 80 municipalities.
Pierre Mauroy, as mayor, president of the CUdL and former Prime Minis-
ter, was the leading force behind the HST integration process in Lille. He
used his political influence, first to get Lille connected to the HST net-
70 Peter Pol
work, and then to bring all public actors involved and several private par-
ties to a consensus on the station development. To that end, a public-
private partnership, a Socit dEconomie Mixte was founded, in which
public authorities accounted for 53%. The remaining shares were in the
hands of banks, the Chamber of Commerce, and a subsidiary company of
the French Railways (SCETA).
Compensation programs were staged to obtain wide regional support for
the station development process. Several municipalities in the Lille region,
which opposed the project, were given compensation because they feared
negative effects from the planned Euralille. The large municipalities in the
Lille region moreover received shares in the Euralille project. Furthermore,
large projects were developed in other cities to convince citizens that Eu-
ralille was not undertaken at the expense of other public investments (Pol
2002 p. 70).
Strategic network: The partnership for the development of the station
area of Lille can best be characterised as a local-government-led pro-
growth type of network. Most protagonists of the process are public actors,
with the local government in the lead. The explicit bet was that the advent
of the HST, in combination with the station development, would have a
catalysing effect on the urban region strong enough to induce (interna-
tional) private investments. Although the focus was on local economic
growth at and around Euralille, under pressure from other local govern-
ments in the Lille-region the partnership has acquired features of a bal-
anced growth network. As the process advanced, attention was given to the
question of how other parts of the region might share the advantages of the
HST development in Lille.
4.10 The South Axis in the city of Amsterdam
Urban context: Amsterdam (735 000 inhabitants) can be considered as an

International Service City (ISC) and attracts above proportional numbers
of European head offices of international firms. Financial services, lawyers
and consultants, etc., are located in the city centre in particular. Neverthe-
less, the importance of the economic function of the historical city centre is
diminishing for reasons of limited accessibility and, more importantly, a
lack of expansion possibilities for firms. A shift of offices to the South
Axis can be observed, especially for international financial firms.
HST link: Amsterdam will be connected to the new HST line to Paris by
2007. There will be three HST stations in the Amsterdam region, Schiphol
Airport, CS and the South Axis. The area around the South Axis station
will be the most important HST development location for the region of
Amsterdam.
Aim of the HST developments: The South Axis is strategically situated,
immediately on the ring motorway, close to Schiphol airport, and close to
a public transport junction. When the financial companies ABN-AMRO
and ING decided to establish their headquarters there, the development of
plans for the South Axis gained momentum. The creation of an HST sta-
tion will merely facilitate the current fast growth of the South Axis. The
HST developments will strengthen the locational attractiveness of the
South Axis, but they are not considered as a necessary condition for the
thriving developments taking place there.
Actors involved: With respect to the node of the South Axis develop-
ment, the municipality of Amsterdam has a leading role to play, but for the
surrounding development zone the dominant stakeholders are private in-
vestors. They are likely to invest mainly in the place, from which they can
expect a sizeable long-term return. As the process advances, the munici-
pality of Amsterdam has, partly under pressure of the local population, as-
sumed a more important role in the development of the South Axis, with
the intention of achieving a balanced development of functions and paying
attention to spatial quality. Moreover, as in the Rotterdam case, the na-
tional government in Amsterdam has an important role in the co-financing
of infrastructure investments.
Strategic network: In first instance, private actors dominated the devel-
opment of the Amsterdam South Axis. The development network during
that time could be called pro-growth and market-led. However, during the
process, increasing attention was given to spatial quality and a balanced
distribution of functions. The development of the South Axis originally
sprang from market pressure rather than from a comprehensive vision of
city development. Only at a later stage did the municipality of Amsterdam
try to check the unstructured development and so devised a comprehensive
urban plan for the area. The South Axis, which at first was destined to be-
come predominantly an office zone, has subsequently evolved during its
planning into a multi-functional urban zone, requiring measures to enhance
the quality of life and the perceived social security. That has happened un-
der pressure from the local population, but also through progressive in-
sights into the important role which the South Axis will play in the urban
structure of the Amsterdam region. The conclusion is warranted, therefore,
that the network governing the South Axis station development has, over
time, acquired features of a balanced growth network.
72 Peter Pol
4.11 The Central Station in the city of Munich
Urban context: Munich (1.228 000 inhabitants) can be considered as an In-

ternational Service City (ISC); its economy is highly diversified and
knowledge-intensive. The city has a very modern industry and hosts head-
quarters and research bases of key companies such as Siemens and BMW.
The inner city of Munich is the economic gravity centre of city and region.
The majority of the socio-economic and cultural activities are located in
the city centre. Until recently, because of heavy pressure on the inner city
by activities and transport flows, the Munich City government explicitly
stimulated the de-concentration of activities in order to achieve a more
balanced polycentric urban structure.
HST link: Munich will be at the junction of two major Trans-European
HST lines, the Brenner-Linie and the Magistrale for Europe. The for-
mer will run from Munich northward by Nrnberg to Berlin, and south-
ward through the Brenner to Bologna, Milan, Firenze, and Rome; the latter
will link Munich westwards by Stuttgart and Karlsruhe with Strasbourg
and Paris, and eastwards by Vienna with Budapest. Along both axes many
sections of the tracks still need upgrading to be fit for a high-grade rail
connection. In the near future Munich will be connected to the dedicated
HST line to Nrnberg.
Aim of the HST developments: The linkage to new HST services is not
expected to much affect the prospering regional economy. Improved pub-
lic transport could, however, help achieve a change in the modal split and
reduce car use. New rail services are therefore expected to improve the
quality of life in the Munich region rather than initiate new economic
processes. The new services can facilitate the current urban Kompakt-
Urban-Grn policy (concentrated urban development with high quality of
life) by stimulating the concentration of activities around collective trans-
port junctions.
In Munich the redevelopment of the central station environment com-
prises plots of land west of the CS, the almost complete rebuilding of the
CS, a tunnel beneath the city, the restructuring of the Ostbahnhof, and the
redevelopment of railway land around this East station in Munich.
Deutsche Bahn (DB, the German railway company) presented in 1996 its
vision of a new station in Munich. It foresaw the demolition of the surface
tracks and the laying out of a four-track tunnel. The Hauptbahnhof would
become an underground through-station with tracks at 35 metres below
street level. Furthermore, the station building would be transformed into a
multi-functional commercial centre of 300 000m2. The draft plan gave rise
to heavy opposition; the general view was that this concept would not
benefit Munich. It was too grand in relation to the existing urban fabric;
the rails would be too deep underground, and there would be far too many
shops (totalling 20% of the existing number in the inner city).
In 2001 the City of Munich, DB and the Bavaria region selected an al-
ternative redevelopment of the CS: a combination of a through-station and
a terminus. The three actors wanted to achieve rather ambitious ends with
the station development, in particular to improve the pivotal function of
the station, create space for new urban activities, and generate positive ef-
fects for the quality of the urban environment.
Actors involved: The key actors involved in the development of the sta-
tion environment are the City of Munich, the Bavarian regional govern-
ment and DB.
Strategic network: The DB in Munich played an initiating role in the re-
development of station areas. The DB aimed at a self-supporting station
development to be financed largely with revenue gained from land and real
estate. That was indeed one reason why the DB initially directed its plans
to the large-scale commercial development of the station area, which im-
plied a market-led pro-growth network. Notably, under pressure from the
local population, an alternative plan had to be drawn up, led by the local
authority, allotting less money to investment in the place (particularly re-
tail activities) and giving more thought to spatial quality and investment in
the node. The local partnership thus assumed some characteristics of a bal-
anced growth network, but has thus far failed to implement its develop-
ment plans due to lack of sufficient investment funds.
4.12 Conclusions
The linkage to the HST network implies, among others, larger catchment
areas and larger economic relevant regions for the cities involved. As a
consequence, particularly in the primary development zones around the
HST stations, more high-order economic activities can be attracted.
Whether this actually occurs depends on how cities react to this opportu-
nity and on the strengthening of weak location factors of the station areas.
Depending on the type of city, other investment strategies and strategic
networks can be expected and observed.
We identified two urban archetypes: Cities in Transition (CiTs) and In-
ternational Service Cities (ISCs). The former are expected to anticipate a
catalysing economic impact of the advent of the HST on the urban region;
the latter a facilitating economic impact. Depending on the behaviour of
the actors involved and the envisaged aims, we observe that, for these city
74 Peter Pol
types, certain strategic networks for station developments arise. They are
indicated in Table 4.3 and explained below. In this table and in the text, we
distinguish an early station development stage (t0, i.e. the planning stage)
and a late station development stage (t1, i.e. the realisation stage).
Table 4.3. Strategic networks governing the development of station areas

City type Aims local govern- Strategic networks European cases
ment
International Indifferent Market-led pro-growth Amsterdam t0
Service Cities Munich t0
Sustainable urban Balanced growth Amsterdam t1
development Munich t1
Cities in Local economic Local-government-led Lille t0
Transition growth pro-growth
Sustainable urban Balanced Growth Lille t1, Rotterdam
development
t0 = early development stage (planning stage); t1 = late development stage (realisation
stage)
Through cooperation in strategic networks, in addition to local author-

ity, other actors involved can exert influence on station development
strategies. This influence will depend especially on their financial contri-
bution and other resources, and on their ability to respond in a proactive
way to new opportunities. When a relatively weak local government is the
case, other actors may earlier dominate the station development process.
The local authority has in all cases played a prominent role in the station
development but with varying influence. The conclusion from our cases is
that, with CiTs, local authority plays an essential part in starting up the
planning process, but becomes increasingly dependent on other authorities
for its implementation. With ISCs, the situation is often the reverse; plans
are initiated or stimulated by private actors, but as the process advances lo-
cal authority tries to tighten its grip.
In Lille and Rotterdam CiTs, station plans have been initiated by the lo-
cal authorities, which subsequently failed to realise them on their own.
Lille sought and found support from the national government, regional au-
thorities and banks, and Rotterdam needed support from the regional and
national government. In the ISC of Amsterdam, private actors were most
active in initiating the area development near the South Axis; in Munich,
the (independent) national railways had an important stake in the planning.
With the two ISCs the local authority became increasingly involved in the
planning process by trying to achieve a more balanced urban growth and
attempting to gain sufficient social support.
A local-government-led pro-growth network could be found in Lille

(during period t0). For this CiT, with the advent of the HST and the station
development, the aim is to attract new urban activities. That happens in
uncertain circumstances. The returns to be realised are normally lower than
in ISCs; investments are therefore expected predominantly from the public
sector. One disadvantage of a pro-growth network, however, is that its fo-
cus is on generating growth within a limited area (often only within the
primary development zone); and it gives less attention to a more balanced
and sustainable growth within the wider urban region.
Market-led pro-growth networks could be expected to occur particularly
in ISCs, in which the local authorities do not act in a proactive way. Be-
cause there are relatively high land rents in these regions, relatively high
returns can be generated in the primary development zone with reasonable
certainty. Indeed, we notice in t0 a dominance of market parties in the sta-
tion development processes of Amsterdam and Munich. They focus pre-
dominantly on maximising profits from the place developments.
A balanced-growth network, normally led by the local authority, tries to
achieve a certain spatial balance of the economic development within an
urban region, and is likely to use the momentum of the advent of the HST
to raise the quality of the built environment. For those targets to be feasi-
ble, the urban regions concerned are mostly those that have already
reached a critical level of prosperity, i.e. the ISCs.
In two ISCs in our investigation, Amsterdam and Munich, it was possi-
ble to observe a shift from dominant market parties towards greater influ-
ence by the local authorities who aimed for a more balanced growth within
the urban region. Thus, a shift from market-led pro-growth (in t0) to bal-
anced growth networks (in t1) could be detected. The consequence was a
tendency to favour more costly plans, which led in the Munich case to the
postponement of its redevelopment plans.
A balanced growth network for the development of an HST station area
is less likely to occur in a CiT. The main reason for this is the relatively
high investment risk involved. Local authorities developing a sound long-
term vision might discern the need to invest in a balanced urban region.
However, to invest in a high quality of life without having achieved suffi-
cient local economic growth does impose high risks. In the long-term, such
a policy might pay off, provided there has indeed been sufficient local dy-
namics. Otherwise the local authorities might be confronted with a rela-
tively high debt position because their investments were not compensated
by sufficient revenues from local growth.
As the process advanced in the CiT Lille, ways were sought to increase
regional support for the intended station development by compensation
programs, distributing Euralille shares to and developing large projects
76 Peter Pol
in neighbouring cities. This allowed other parts of the region to benefit

from the Euralille development. Therefore, for the sake of political sup-
port, the network in Lille acquired features of a balanced growth network
(in t1).
The station development partnership in Rotterdam showed features of a
balanced growth network. The level of ambition for the station redevelop-
ment has been relatively high for a CiT. There was, however, a serious risk
of substantially downgrading these ambitions because of financial conces-
sions. However, until now, by staging up the process first developing the
station terminal, and later on the area development the city of Rotterdam
could largely hold onto its strategy in order to realise a high-quality zone.
To support a sustainable economic development of HST cities, it is nec-
essary to have sufficiently broad objectives pursued by the station-
development network. They should be wider than merely the improvement
of the station terminal, and should include elements such as the develop-
ment and spatial quality of the surrounding station area, and of other re-
lated urban parts. Then, in principle, more urban spread effects of the ad-
vent of the HST can be expected, because raising urban attractiveness is an
important precondition for attracting talented people and high-end eco-
nomic activities. Investing in a balanced way in the HST station area can
thus be instrumental to luring public and private investment over a longer
time-frame and in a multitude of places within the urban region involved,
and can therefore be a condition for sustainable urban growth catalysed by
the HST.
References
Beaverstock JV, Smith RG, Taylor PJ (1999) A Roster of World Cities. Cities
16:445 458
Berg L van den, Braun E, Meer J van der (1997) Metropolitan Organising Capac-
ity: Experiences with Organising Major Projects in European Cities. Ashgate,
Aldershot
Berg L van den, Pol PMJ (1998) The European High-Speed Train and Urban De-
velopment: Experiences in Fourteen European Urban Regions. Ashgate, Al-
dershot
Berg L van den, Pol PMJ (1999) The High-Speed-Train Station and Urban Revi-
talisation: A Comparative Analysis of Station Areas in the Cities of Amster-
dam, Antwerp, Berlin, Cologne, Dortmund, Lille, Munich, Rotterdam, and
Utrecht. Euricur, Rotterdam
Berg L van den, Meer J van der, Pol PMJ (2001) Maatschappelijk-Economische
Evaluatie van de Herontwikkelingsplannen van het Rotterdamse Centraal-
Stationsgebied. Euricur, Rotterdam
Berg L van den, Meer J van der, Pol PMJ (2003) Social Challenges and Organis-
ing Capacity in Cities: Experiences in Eight European Cities. Ashgate, Alder-
shot
Bertolini L, Spit T (1998) Cities on Rails: The Redevelopment of Railway Station
Areas. E & FN Spon, London/New York
Bonnafous A (1987) The Regional Impact of the TGV. Transportation 14:127
137
Brindley T, Rydin Y, Stoker G (1989) Remaking Planning: The Politics of Urban
Change in the Thatcher Years. Unwin-Hyman, London
Digaetano A, Klemanski JS (1993) Urban Regimes in a Comparative Perspective:
The Politics of Urban Development in Britain. Urban Affairs Quarterly 29:54
83
John P, Cole A (1999) Political Leadership in the New Urban Governance: Britain
and France Compared. Local Government Studies 25:98 115
Plassard F (1991) Transport and Spatial Distribution of Activities. Round Table
85, ECMT, Paris, pp 43 74
Speed-Train Stations. DUP Science, Delft
Robentraub MS, Helmke P (1996) Location Theory: A Growth Coalition and a
Regime in the Development of a Medium-Sized City. Urban Affairs Review
31:482 507
Schtz E (1996) Stadtentwicklung durch Hochgeschwindigkeits-verkehr, Konzep-
tionelle und methodische Anstze zum Umgang mit den Raumwirkungen des
schienengebunden Personen-Hochgeschwindigkeitsverkehr, Dissertation,
Universitt Kaiserslautern
Stone C (1989) Regime Politics Governing Atlanta 1946-1988. University Press,
Lawrence, Kansas
Troin JF (1995) Rail et amnagement du territoire. Edisud, Aix-en-Provence
Urban Audit II, www.urbanaudit.org
5 What makes a city: Urban quality in Euralille,
Amsterdam South Axis and Rotterdam Central
Jan Jacob Trip
OTB Research Institute for Housing, Urban and Mobility Studies,

Delft University of Technology, the Netherlands
5.1 Introduction
Many HST station areas take shape as what we may call international
business centres: locations that aim to attract the offices of, in particular,
internationally-oriented producer services. Euralille, for example, was de-
veloped explicitly as a European Business Centre; in Amsterdam, interna-
tional banks did in effect initiate the development of the South Axis before
local authorities joined in. The HST may facilitate the development of an
international business centre mainly for two reasons. First, it provides ad-
ditional transport facilities, which are important as knowledge-intensive
activities still very much depend on face-to-face contacts. Second, it pro-
vides an image that suits international business. Neither is indispensable:
exemplary centres of international business such as La Dfense or Canary
Wharf can do without. But for cities of a somewhat smaller calibre, the
HST is considered a must-have.
International business requires large amounts of modern, efficient office
space, which often results in large-scale, monotonous and rather schematic
areas (again, consider La Dfense). Many of these fail to provide the qual-
ity and metropolitan atmosphere required for high-end locations. The types
of firms that locate there tend to be quite sensitive to the quality and status
of their offices. Quality of the urban environment for instance, architec-
ture, urban design, and public space may actually pay by means of in-
creased real estate revenues despite the additional investments it requires
(Rowley 1998; UCL 2001). Another line of thought relates a more diverse
set of urban quality or quality of life issues to urban competitiveness in the
long-term (Kresl 1995; Segedy 1997; Gospodini 2002). In recent years,
80 Jan Jacob Trip
Richard Florida has been the most notable advocate of this idea. Building
on the work of Jane Jacobs (1961) in particular, Florida (2002a) states that
advanced service economies are driven by a specific creative class. This in
turn is attracted and retained by certain characteristics of the day-to-day
urban environment, which Florida defines as quality of place.
Quality of place as defined by Florida entails a set of qualities that col-
lectively make a city an attractive place of residence for the creative class.
It may roughly be divided into three components (Kloosterman and Trip
2006 pp. 3-4):
whats there? (the built environment, specific amenities, third
spaces for informal meetings);
whos there? (diversity of people, tolerance);
whats going on? (street life, buzz).
Some of these qualities are fairly elusive and relevant only on an urban
or regional scale; some are more concrete and related to urban design. Ac-
cordingly, quality of place is hard to reproduce or plan, and as far as it can
be planned for, requires a distinct long-term perspective: first, simply be-
cause it requires time to evolve; second, because in view of the shared
public-private development and control of HST station redevelopment pro-
jects, quality of place necessitates a collective action which lasts long after
the immediate development phase.
Floridas book seemingly has the characteristics of an urban growth
manual. It has been popular almost instantaneously among local and re-
gional policy makers in the US and Europe, and increasingly became an is-
sue of debate and, sometimes venomous, criticism (Trip 2006 pp. 4-5). A
main strength of the concept is the explicit link Florida makes between the
competitiveness of the urban economy and a sophisticated perspective on
urban development, covering a broad field of economy, sociology and ur-
ban development. Furthermore, the fact alone that Floridas ideas affect so
many cities economic policies makes them key to discussion.
Thus, the quality of the HST station area is in one way or another rele-
vant to its potential as a high-end business centre. Moreover, HST stations
are often located centrally in the city; they are important as public spaces
which makes it even more important that they be high-quality urban areas,
rather than mere business locations. The focus of this chapter is therefore
on the question of the role of quality, specifically quality of place, in the
development of HST station areas. Three projects are considered along the
TGV Nord from Paris to Amsterdam: Euralille in Lille, which was largely
completed in the early 1990s, and the South Axis in Amsterdam and Rot-
terdam Central in Rotterdam, both of which are now partly under construc-
tion, partly still on the drawing board.
What makes a city: urban quality 81
These projects are considered from the perspective of quality of place;

not because the concept as such is bound to play an important role in these
projects, but because it is currently the most elaborate, theoretically
founded and influential concept of quality of the urban environment affect-
ing the economic performance of cities. Florida hardly provides clear-cut
guidelines for the planning of specific areas although he comes close to
this in his report on the rebuilding of Lower Manhattan (Florida 2002b).
But if his ideas are translated into concrete urban development, they
point to issues that largely resemble the ideas of, particularly, Jacobs.
Hence, the question is which elements of quality of place can be recog-
nised in the development of these projects, and to which extent the concept
is taken into account in the development process.
The analysis of Euralille, the reference case, is based on site visits and
literature. The analysis of the other cases is based on two main sources.
First, an analysis is carried out of the project plans, as presented in plan-
ning documents and on websites. Second, a series of in-depth interviews
have been conducted with key actors involved in the planning process.1 In-
terviewees have been selected on the grounds of their active involvement
in, and estimated influence on, the planning process and their involvement
in the development of the station area as a place, rather than as transport
node. Interviewees include developers (representatives of banks and other
private development corporations), designers (representatives of municipal
planning and design departments, and architects commissioned by public
bodies), and coordinators (representatives of public development corpora-
tions, rather market-oriented in their tasks and attitude). Any mutual dif-
ferences in attitude and interests that may appear are not generally blurred
by differences in background knowledge or jargon, as these are rather
similar for all three groups.
5.2 Euralille
It was the idea of Lilles mayor Pierre Mauroy to bring the TGV to the
centre of Lille and use it as the anchor of a cluster of high-valued service
1 Including representatives of: Benthem Crouwel/Team CS architects;
Bouwfonds Property Development; De Architekten Cie.; Dutch Railways, De-

partment of Commerce; Fortis Real Estate Development; ING Real Estate; Mu-
nicipality of Amsterdam, Development Corporation; Municipality of Amsterdam,
Department of Spatial Planning (2x); Municipality of Rotterdam, Development
Corporation; Municipality of Rotterdam, Department of Urban Planning and Pub-
lic Housing (3x); Zuidas Project Office.
82 Jan Jacob Trip
industries, commerce and leisure. Its main purpose was to improve the
economic position of the city, ailing after the decline of its textile and min-
ing industries (Bertolini and Spit 1998 p. 68). The focus of the project
would be international; it would be built on a site between the existing
Lille Flandres station and the ring road, formerly occupied by fortifica-
tions.
Dutch architect Rem Koolhaas was appointed in 1989 to make the urban
design of what was to become Euralille. As Koolhaas recalls: On reading
the competition brief, I was at first convinced that the idea of a European
Business Centre was a typical example of French megalomania, of which I
was very sceptical. Remember, this was a time when it was still unsure
how great the impact of the TGV would actually be (Koolhaas et al. 1996
p. 51). As it became clear that the project would be feasible, in 1990 the
SAEM (Societ Anonyme dEconomie Mixte), a public-private partner-
ship for the development of Euralille was founded (Spaans 2002 pp. 206-
212).
The initial Euralille consists of three parts (Bertolini and Spit 1998 pp.
78-79): the Cit des Affaires, including the Lille Europe station with the
WTC and Crdit Lyonnais office towers above it, the Euralille Centre,
situated between the two stations, including offices, apartments, a shop-
ping centre, a hotel and other amenities (Figure 5.1), and the Grand Palais
Congress Centre. Other subprojects were added gradually. Table 5.1 shows
the present real estate programme. Due to a crisis in the real estate market
in the mid 1990s, the amount of offices realised in particular, has long re-
mained below schedule. The Cit des Affaires was planned as a row of
skyscrapers above the station, only two of which have at the time of this
writing been built. A second stage, Euralille 2, is anticipated to be com-
pleted by 2010.
The involvement of Koolhaas has been essential for the success of the
project. He made the urban design on which the project is based, and in
later stages supervised the development process. The urban design, and by
this Koolhaas conceptual vision, has been a crucial element in the devel-
opment of Euralille; it fulfilled its obvious function as a framework for the
physical development of the project. Furthermore, it was important for the
image of the project and played a key role in attracting private investors
(Tilman 1994 p. 29).
Despite initial scepticism, the effect of Euralille on the urban economy
is generally considered as positive. Furthermore, the project has significant
symbolic value, contributing much to the image and self-confidence of
Lille as a modern city, and its position as a centre for shopping and tourism
(cf. Bertolini and Spit 1998 p. 84; Spaans 2002 p. 223). In particular, the
position of the inner city itself has been strengthened (Bertolini 2000 p.
471). The disappointments expressed with regard to the effects of Euralille

seem to be caused in part by excessive initial expectations and partly by
externally-driven market developments.
Figure 5.1. The Crdit Lyonnais tower (Christian de Portzamparc) and the Eura-
lille Centre (Jean Nouvel).
Table 5.1. Real estate programme and functional mix in Euralille (current plan-
ning), compared to the South Axis and Rotterdam Central (planning in 2003).
Total (m2) Business (%) Residential (%) Amenities (%)
#
Euralille 801 093 40 21 39
South Axis (dock) 2 362 000 42 45 14
South Axis (without 927 000 46 36 18
dock)
Rotterdam Central 220 000 27 59 14
Source: VROM (2003, p. 61); www.saem-euralille.fr (24 March 2006).
#
Euralille 1 and 2, excluding Grand Palais, including Portes du Romarin area.
Quality of place in Euralille
Euralille is conceived primarily as a part of the international space of

flows, rather than as an extension of Lille. The project as it exists today is
largely the result of Koolhaas idea of the city, which is not primarily
based on context or history, but rather on a generic, functional modernism.
Two themes continue to reappear in Koolhaas writings: the density and
multitude of different uses of the skyscraper, and the highly commercial-
ised space of the shopping centre or the amusement park. Both are evident
in Euralille: the row of towers, the shopping centre and the Grand Palais.
In response to his French criticasters, Koolhaas famously stated that critics
of Euralille as a place hold no grounds against a non-place, and that the
84 Jan Jacob Trip
projects supposed ugliness is irrelevant as long as it succeeds in its objec-

tives (Koolhaas et al. 1996 p. 189-90). Actually, the receptive mind can
easily find beauty in Euralille, however subjective, and it may well be con-
sidered that traditional values are not always satisfactory regarding such an
innovative project. But aside from semantics, even this non-place is in fact
a place for people to work, live, recreate, etc. As such, it has some definite
weaknesses.
A major difficulty was how to integrate this project of a radical mod-
ernisation into the existing city (Koolhaas in: Koolhaas et al. 1996 p. 65;
Koolhaas and Mau 1995 p. 1160; Dovey 1998 p. 89-90). Previously, the
location of Euralille was occupied by a little-used park and even then the
site was considered as an urban void; moreover, the ring road was also al-
ready there at the time. To overcome this double barrier, the Le Corbusier
viaduct was constructed which connects the inner city and the Euralille
Centre with the station and the ring road and the La Madeleine and Saint
Maurice districts beyond. The metro system was also extended to the Lille
Europe station. Other relevant elements of the project include the famous
TGV window, a glass wall between the train platform and the station
square, intended to visualise the relation between the TGV and the city,
and a diverse functional programme with strong emphasis on various
amenities. Public space around the station is furnished with art, water,
trees, and access to shops and catering facilities within the Centre.
However, the result is not satisfactory in every aspect. Figure 5.2 shows
the public space around the station. The pictures, taken on a Saturday af-
ternoon, can hardly be said to show a vivid urban area. A main reason for
this is the shopping centre; with its inward orientation it tends to draw
people from the street, while amenities on the outside are insufficient to
generate any street life (Dovey, 1998, p. 94). To make things worse, the
catering on the square has mostly closed. Furthermore, there remains a
feeling that Euralille is a barrier between the inner city of Lille and the
neighbourhoods beyond the ring road.
5.3 The South Axis and Rotterdam Central projects
The projects involved here, Amsterdam South Axis and Rotterdam Cen-
tral, cannot be separated from their local contexts. For one thing, Amster-
dam is more of a service economy than Rotterdam. Although the cities are
converging in terms of their economic structure, absolute employment
growth is considerably larger in Amsterdam, which benefits from its larger
share of fast-growing producer services (Kloosterman and Trip 2004).
Amsterdam also ranks above Rotterdam in terms of creative class size and
quality of place, particularly regarding the socio-cultural aspects empha-
sised by Florida (Trip 2006). These different contexts affect the develop-
ment and the objectives of the projects.
Figure 5.2. The Le Corbusier viaduct seen from the inner city towards Lille
Europe station (above left); the viaduct crossing the station building, seen from
above the ring road (above right); place Franois Mitterrand in front of the station
(below left) and seen towards the Euralille Centre (below right).
5.3.1 South Axis
Since the 1960s the enduring demand for office space in Amsterdam could
no longer be accommodated within the inner city. Nonetheless, in the early
1990s local authorities found that the city still needed a high-quality office
location. Initially, the intention was to redevelop the embankments of the
River IJ, near the Central Station. But private parties preferred locations
closer to the ring road. In effect, the South Axis, around the southern rail-
way and A10 ring road, rather than the embankments area became Am-
sterdams top-end office location early in the 1990s. The local authorities
then assumed an active role in the development of the area (Ploeger 2004
p. 95-99; Salet in Salet and Majoor 2005 p. 43).
86 Jan Jacob Trip
The project first proceeded with the Masterplan South Axis (DRO
1998). At about the same time, architect Pi de Bruijn was named urban de-
sign supervisor of the South Axis, a function comparable to Koolhaas role
in the development of Euralille. However, whereas Koolhaas could virtu-
ally start from scratch, the rough outline of the South Axis was actually
there when De Bruijn was appointed. Since then, plans have been elabo-
rated gradually, while the focus has shifted from offices to a mixture of of-
fices, apartments, and other functions (DRO 2001, 2004). The aim of the
South Axis project is to create a secondary city centre with a distinct, met-
ropolitan atmosphere; it should be a high-end location for service indus-
tries, competitive on the international level. To succeed in this, the South
Axis should also be an attractive urban district. The project entails numer-
ous subprojects. The focus here is on the area around the WTC and the
Mahler and Gershwin projects south of the railway station where devel-
opment is furthest. These areas are therefore currently most relevant with
respect to the analysis of quality of place. Public debate, in contrast, fo-
cusses on the construction of a 1.2 km long railway and motorway tunnel
(the dock). Plans concerning this area are still preliminary, and it is hard to
say something specific about its quality of place. In the long-term, how-
ever, the dock could have an important effect on the areas quality of
place.
5.3.2 Rotterdam Central
Like Lille, Rotterdam suffered from general industrial decline, and like
Lille it was searching for ways to develop a more service-oriented econ-
omy. The local authorities now focus on attracting higher-income groups
and retaining the middle class. In this respect, the inner city poses a prob-
lem, especially in view of the growing importance of urban quality. Most
of central Rotterdam was destroyed in May 1940 and replaced by an inner
city designed in accordance with strict modernist principles; but what was
considered at the time as modern and efficient now often looks bland and
outdated. In recent years, however, Rotterdams renewed self-
consciousness is increasingly reflected in striking modern architecture, the
context in which Rotterdam Central is currently one of the citys main pro-
jects.
The decision to construct a high-speed railway between Amsterdam and
Paris induced the renewal plans for Rotterdam Central Station, but the pro-
jects aims exceed merely that. The station itself has to be enlarged, all the
more so, as it has to accommodate Randstad Rail, a light-rail to The
Hague. However, the objective was also to improve the quality of the sta-
tion area, and thereby, the attractiveness and dynamics of the inner city as
a whole (Gemeente Rotterdam 2003). As a result of its modernist layout,
Rotterdam is one of the few Dutch cities able to handle large-scale office
development within its inner city. The station area already is an important
office area, but for a large part it is also an unattractive out-of-the-way
place.
William Alsop presented his ambitious Masterplan Rotterdam Central
in 2001: not so much as a detailed architectural design but as a flexible
framework aiming towards the improvement of the station itself, and on
the more abstract objective of upgrading the inner city and creating a met-
ropolitan atmosphere suiting the new economy. Were his ideas accepted,
Alsop might have played a role similar to that of Koolhaas in Lille and De
Bruijn in Amsterdam. Instead, they were considered too expensive and too
extravagant by many, and in 2002 the Masterplan was abandoned (Alsop
2001; Kooijman and Wigmans 2003 p. 321).
The current project is considerably smaller in scope and, accordingly,
much cheaper. The focus is on the station itself, which is now most urgent.
Team CS, a combination of Benthem Crouwel, Meyer and Van Schooten
and West 8 Landscape Architects have been appointed to design the new
station. The development of the surrounding area is expected to occur in a
second stage.
5.4 Quality of place in the South Axis and Rotterdam

Central
To investigate the role of quality of place in the development process of

these projects, in this section a brief inventory is made of i.) elements of
quality of place included in the projects plans, and ii.) actors perceptions
of quality of place. After that, some main elements are further elaborated.
5.4.1 Quality of place in the project plans
The projects plans of the South Axis (DRO 2001, 2004) and Rotterdam
Central (Gemeente Rotterdam 2003, 2005; Team CS 2005) contain several
elements that may be considered to aim at urban quality or, for that matter,
quality of place:
South Axis plans include a diversity of business, residential, shop-
ping and leisure functions. In Rotterdam, the emphasis is on in-
88 Jan Jacob Trip
creasing the residential function (Table 5.1, above), whereas the

plan includes no new, large-scale shopping or leisure facilities;
density in the central area of the South Axis is relatively high, es-
pecially compared to current Dutch practice. In Rotterdam density
is already considerable, but will increase;
the South Axis is planned as a relatively fine-grained grid with
relatively narrow streets, alternated with squares. In Rotterdam the
station area and the inner city are predominantly large-scale, but
newly-developed real estate should be conceived on an apparently
smaller scale. The station and the station square, however, retain a
distinctly large scale;
to increase liveliness in the South Axis functions such as retail and
catering are planned in the streets, rather than in internally-
oriented shopping centres. Also in Rotterdam, these functions
should be included in the streets, even in the now blind walls of
existing buildings;
however preliminary it may presently be, the dock alternative in
which infrastructure is placed at the subterranean level is most
relevant to the quality of place of the South Axis, as it would
largely reduce the negative effects of the infrastructure, particu-
larly spatial fragmentation and noise. Moreover, without the dock,
noise restrictions would, necessarily, strongly limit the residential
function of the South Axis (Table 5.1);
in Rotterdam a main objective is to improve pedestrian comfort.
The noisy and dangerous traffic swarm in front of the station is un-
ravelled, and the square itself is to become the pedestrians do-
main.
These mostly concern aspects directly related to urban design which
suggests that some but not all of the ideas of Florida are taken into ac-
count: the primary idea being the whats there? component. These ele-
ments are important at the level of the project area, but they are most easy
to schedule in urban design. Now we must inquire as to what actors in-
volved in the planning of these projects actually understand by quality of
place.
5.4.2 Actors perception of quality of place
The actors perception of quality of place may be expected to influence the

actual project plans, and vice versa, as the planning process is lengthy and
often iterative. It is hardly surprising then, that the issues mentioned below
largely correspond to the elements found in the actual project plans.
All three groups of interviewees developers, designers and coordina-

tors have articulate ideas on what a city should be and these ideas do not
differ radically from each other. Irrespective of whether these ideas are
based specifically on the concept of quality of place, they certainly to a
large extent correspond with the elements of quality of place relevant in
the development of station areas. Thus, whereas few of the actors inter-
viewed are more than superficially familiar with Floridas ideas concern-
ing the creative class and quality of place, all pay explicit attention to ele-
ments of quality of place, although often under different labels.
Table 5.2 shows the characteristics of quality of place that interviewees,
upon being questioned, mention spontaneously. Like the elements included
in the project plans listed above, those put forward by interviewees seem to
concern primarily the more concrete components of urban quality, closely
related to their sphere of activity. These issues also correspond to the fac-
tors of thriving urban areas emphasised by Jane Jacobs. Rather than refer
specifically to quality of place, they are closely related to the desired urban
or metropolitan climate, which is, as one interviewee states, an expression
of quality of place.
Table 5.2. Main characteristics of quality of place mentioned by interviewees (n

= 14).
Characteristics Frequency
Public functions at street level outside of buildings 7
Quality of public space 7
Functional mixture 6
Liveliness during the day 6
Density 5
Materials applied 5
Clarity of design; extent to which public space can be surveyed 4
Control and maintenance 4
Quality of architecture 4
Accessibility 3
Safety and convenience for pedestrians 3
Nonetheless, some actors, most of whom are designers, are also con-
cerned with the more intangible aspects of quality of place, related to the
questions whos there? and whats going on? This does not necessarily
mean that they are familiar with Floridas work; but it does mean that they
share some of his ideas. Supervisor Pi de Bruijn refers to the South Axis as
a metaphor of modern society, an open society that is founded on local
values (De Bruijn 2005); and although it is not the goal we set ourselves
beforehand, our ambitions come very close to the kind of city that Richard
90 Jan Jacob Trip
Florida regards as the fertile breeding-ground for creative talent with

Jane Jacobs, of course, in the background, who described which qualities a
city needs to produce genuine urban life back in the early 1960s. It is
above all the city where you come across both like-minded people and
people who are strange in every aspect. It is the city in the words of Flor-
ida that is open, diverse, tolerant. It is also the city with a lively street
life, though one that is not confined to the streets themselves, but extends
to public amenities such as the bar (De Bruijn, in: Franke and Verhagen
2005 p. 159). We recognise, however, that these elements of quality of
place are relevant mainly on a higher scale and hardly applicable in a sta-
tion area setting.
Based on the whole of the interviews conducted as well as the project
plans, a number of main issues may be distinguished that are briefly dis-
cussed below: i.) urban structure, ii.) functional diversity, iii.) the quality
of functions, iv.) the quality of public space and v.) architectural quality.
station architect design of

(Rotterdam) public space
supervisor design of general

(Zuidas) urban structure
definition of
municipal plan-
functional
ning department
programme
architecture of
private
commercial
developers
real estate
municipal deve- selection

lopment corp. of users
Figure 5.3. Distribution of actors responsibilities in the South Axis and Rotter-
dam Central projects concerning various aspects relevant to quality of place. Dot-
ted lines represent consultation rather than formal responsibility.
Figure 5.3 shows the distribution of actors responsibilities with respect
to the elements of the development process related to these issues. As a
generalised scheme it does not show all local peculiarities in detail.2 On the
whole, however, public actors largely decide on the quality of public
space, the functional programme and the general urban structure of the
area, while the selection of users is entrusted to the private sector.
2
See Kloosterman and Trip (2006 pp. 8-10) for a more detailed description of
the development process.
i.) Urban structure
Issues such as street pattern, scale and grain are defined in the urban de-
sign made by the municipal planning departments, in the South Axis, to-
gether with the supervisor. The South Axis is largely planned as a grid of
relatively small blocks, separated by relatively narrow streets of about 10m
wide (Figure 5.4). With an estimated floor space index of three to nine in
the central area, density is relatively high (DRO 2004 p. 27). Blocks could
be developed separately, enabling diversity and flexibility. Moreover, a
high building density also implies a high density of people, which with
the narrow streets should contribute to liveliness and street life. It is in-
spired on the Manhattan grid, but also on the inner city of Amsterdam.
Figure 5.4. The station area in Rotterdam (left); relatively narrow streets in the
South Axis (right).
Most of the urban structure in Rotterdam already exists, making it de

facto fixed. The current density is increased substantially, mainly by add-
ing high-rise apartment buildings in later stages of the project (Gemeente
Rotterdam 2005).
Several, mostly public, actors mention the issue of scale. Designers in
both cities want varied, small-scale (tall but narrow) buildings: It con-
cerns a fine grain in public space, the height of blocks, small blocks. The
tendency is to have large blocks developed by one single developer, but
thats not what we want. [] It would be good if individuals could also
develop a building, not just big is beautiful. But that doesnt have to be
possible everywhere. It might be possible on the edges of the area, perhaps
a villa-like development. You can often see something like it in Belgian
cities: a fine grain, very sophisticated. However, they recognise that this
92 Jan Jacob Trip
is hardly possible because investors and developers prefer large blocks and
buildings. The result should be large buildings designed to look small and
varied, providing a suggestion of small scale. Yet, current development in
the South Axis thus far can hardly be called small-scale. Projects embed-
ded in an existing urban area may have an advantage in this respect over
all new development. Rotterdam is an exception to this, as the existing en-
vironment of the station is already quite large-scale (Figure 5.4). More-
over, it is considered to be a distinctively positive feature of the city: Rot-
terdam is a city of large spaces, particularly in the inner city. But many
people long for the old inner cities, for cosiness, for nostalgia instead of
modernity, especially in these fearful times. So, should you rebuild an old
city? Other people find Rotterdam exciting. The tension of the great, the
high-rise buildings, the space that is different from that in other cities; you
must preserve that. You should use the space as a quality.
ii.) Functional diversity
The municipal planning department, and in the South Axis the supervisor,
define the functional programme in close cooperation with private devel-
opers. Table 5.1 (above) shows the building volume and functional pro-
gramme of the projects. The shares of business and residential functions in
the South Axis are about equal in 2003, whereas in 1998 the ratio was 65%
to 21% (DRO 1998), a change which reflects the shift in objective from an
office area to a metropolitan urban centre. The station area in Rotterdam is
presently an important, rather large-scale office location. Hence the focus
here is on increasing the residential function (Gemeente Rotterdam 2005).
A limited number of amenities are planned, merely to facilitate the station
area itself and to enliven the streets surrounding the station. In both pro-
jects public and semi-public amenities are planned deliberately in the
streets, rather than inside a shopping centre. In this respect, Euralille is
considered as an example to be avoided.
Apart from the usual functions offices, residential and amenities and
the specific functions already present, some new amenities are being
planned, particularly in the South Axis. First, a theatre is to be developed
near the RAI exhibition centre, perhaps too far from the central area of the
South Axis to entice people there. Furthermore, a cluster of museums is
planned east of the WTC. Besides these rather large-scale amenities, inter-
viewees emphasise the importance of a wide range of public and semi-
public functions, such as sports facilities, fitness clubs, day care centres,
medical services, etc. These are considered important in making the South
Axis a piece of real city, which could provide its users and inhabitants
with a broad range of facilities. Also, some of these functions may attract
different groups of people to the area and increase liveliness in the eve-
nings: a diversity of functions is here explicitly linked to a diversity of
people. Fitness clubs are mentioned as an example, especially if they were
visible from the outside, rather than hidden. The Reebok fitness centre at
Canary Wharf is mentioned as an example in this respect.
iii.) Quality of functions
Apart from the type of functions, their quality is considered essential to the
overall quality of the area in the long-term. In other words, who are the ac-
tual users of the area? Many firms in the business service sectors are quite
sensitive to the quality of their office environments; the location of a su-
permarket at a prestigious location in the South Axis was actually blocked
by a protesting lawyers office located in the same building, while a delica-
tessen or a bookstore would not have posed a problem. Especially among
developers, it is felt that within the functional programme, certain func-
tions that do not match should not be accommodated beside each other.
Social housing does fit in with urbanity, but not with high-end ameni-
ties. It should not have a dominating negative impact on its neighbours,
but it should be located, for instance, a few blocks away. On a smaller
scale, the question is who decides on whether catering will be a Burger
King or a Michelin-rated restaurant. This issue is emphasised particularly
in the South Axis, with its high ambitions and extremely long construction
period, but it is also an issue of debate in Rotterdam.
In both projects the eventual selection of users is done by developers
and investors. Therefore, the long-term view of market parties seems es-
sential in maintaining the quality of the area. The municipality provides
guidelines and may influence the selection process, for instance, by means
of coordination committees. The general feeling seems to be that a prohibi-
tive approach, while possible in a strictly legal sense, is not carried out. At
present, both public and private interviewees express a strong belief in the
competitiveness of the area and in the effect of self-regulation. One inter-
viewee explains: The rent level and the atmosphere of the area will help
sort things out; you can already see it happening in Mahler. Market parties
select their tenants with this in mind, and they are doing it fairly well. []
It is a matter of coordination and conviction; you can no longer direct and
enforce [these things]. There is a danger that quality requirements may be
weakened in time, but what else could you do? You might establish a
steering committee, but its doubtful whether that would work. What is
better is to start with an in-depth discussion with all the parties about the
ambitions you have, the things you want.
94 Jan Jacob Trip
Nevertheless, in the South Axis (where the issue is most acute) both
public and private actors recognise the reduction of quality requirements as
a potential danger: it is clear that if stagnation were to occur, developers
and investors would, after a certain period, decrease their standards rather
than leave the building unoccupied. Mutual coordination between actors
should prevent this. It is acknowledged that it is important to maintain a
high standard in every part of the project right from the beginning. This
makes actors very dependent on each other: once the quality of the area is
gone, which could happen quite rapidly, it will not be easily regained and
all investors suffer. In the end, however, the robustness of private actors
commitments, as opposed to the supposedly short-term requirements of
economic competition, cannot be known for sure at this time.
iv.) Public space
In both cases the quality of public space is among the most frequently
mentioned issues and includes a variety of factors. When asked to specify
what they thought defines good public space, interviewees mention many
of the issues also listed separately, such as functional diversity, architec-
ture and appropriate scale. Nonetheless, public space as such is clearly
perceived as one of the defining factors of quality of place. This also in-
volves the availability of parks, sittable spaces and various amenities. The
divergent views became clear with respect to the Zuidplein in the South
Axis, and were highly praised by some and despised by others (Figure
5.5). Public actors and publicly commissioned architects are responsible
for the design of public space (streets and squares) and public amenities
(schools, parks, etc.). Maintenance of public space is also handled by pub-
lic parties, although it is suggested that private control (as in Canary
Wharf) would better guarantee the quality of public space.
Figure 5.5. The Zuidplein near South/WTC station.

Public space is probably the design element concerned most directly

with the behaviour and well-being of people: with whos there? and whats
going on? It is also closely related to the occurrence of third spaces, the
public and semi-public spaces suitable for informal meetings that is em-
phasised by Florida. Opinions differ on whether these play any role in the
planning process; in particular, developers in the South Axis refer to a
large variety of amenities that may serve as meeting points and locations
of buzz: specific pubs and restaurants, the fitness club, the hotel, the
Mahler square. Parks are also mentioned, the theatre, the metro station,
even Richard Florida could not possibly have imagined this the grave-
yard. The concept of third spaces as such is not evident, however. Particu-
larly developers seem somewhat sceptical about the role of third spaces in
the current design. Conversely, designers mainly stress that good public
spaces and third spaces for that matter, have to evolve and cannot be con-
structed instantaneously, but may be wasted in next to no time.
v.) Architecture
Finally, two related aspects mentioned by all private developers and repre-
sentatives of public development corporations (but hardly by designers)
are quality of architecture and applied materials. Developers are responsi-
ble for the architecture of the commercially developed real estate within
the margins set by the conceived urban plan; public or publicly commis-
sioned designers are responsible for the architecture of public spaces and
public amenities. Architectural quality is considered to imply a variety of
building forms and involve such renowned architects as Michael Graves
and Rafael Violy in the South Axis, who are able to give their buildings a
distinct expression.
The emphasis on applied materials concerns, for instance, stone or
pavement types and the craftsmanship put into buildings and public space,
or what one interviewee calls the semiotics of the area: A beautiful finish
is very important for the quality of the area, of public space as well as of
buildings. For example, you can see that a banking office is a valuable
building when it has a beautiful appearance. When the finish of both build-
ings and public spaces is good, this has a very strong effect on the image
of the area. Architectural quality is considered an important issue in the
South Axis, in view of the explicit ambition to develop a high-quality in-
ternational business location and attract targeted users. Still, an unresolved
problem remains that architectural quality, as far as it implies beauty, is in
the eye of the beholder, and that high-quality architecture, from a profes-
sional viewpoint, does not necessarily appeal to a general audience.
96 Jan Jacob Trip
These issues are mentioned less often in Rotterdam, where private de-
velopment is not in order at the moment. Here, however, it is emphasised
that the entrance of the city, in terms of architecture and detailing, should
be beautiful.
5.4.3 The HST
What about the high-speed train (HST) itself? Actually, it is mentioned

only occasionally as a factor of the quality of the area. Its appreciation dif-
fers, depending on the local context and the citys position in the HST
network. In the case of the South Axis, proximity to Schiphol Airport is
considered the areas most important asset, and the HST will not further
improve this. Accessibility by car is also valued at least as much as the
HST, and in terms of passenger volume, the metro and regional rail trans-
port are far more important. The HST is appreciated, however, for its ef-
fect on the image of the South Axis. As one interviewee states: It would
not be a lethal blow for the South Axis if it did not come; there are already
firms located there at present, after all. But you cannot pretend to be a na-
tional, let alone an international, high-end location if you do not have the
high-speed train. However, even if the South Axis were not to get an HST
connection, Amsterdam definitely would.
The HST in Rotterdam makes more of a difference, as it would signifi-
cantly shorten travel times to Schiphol. The image effect is considered im-
portant here as well; even more so as Rotterdam, compared to Amsterdam,
lacks its own strong image. This is also true for Lille. Moreover, in this
case the transport value of the HST is likely to be even more important,
relatively, due to Lilles more central position in the HST network.
5.5 Conclusions
The development of HST station areas is closely related to the dilemma of

international business location: how to accommodate a large amount of of-
fice space and yet maintain the necessary attractive urban environment?
More than a fringe location, station area development brings this dilemma
back to the heart of the city, making urban quality elaborated here as the
concept of quality of place even more important. The role of quality of
place as such in the development of the projects considered here is limited.
However, many of the measures aimed at urban quality in fact concern
elements of quality of place. Unsurprisingly, these are not the elusive,
socio-cultural aspects, but rather the elements that are relevant on the scale
of the project and relate to urban design: diversity, safety, authenticity, and
beauty. These issues largely concerned with whats there? are explic-
itly, but not always entirely or convincingly linked to the density and di-
versity of people, and liveliness in short, to whos there? and whats go-
ing on? It largely vindicates Jane Jacobs; she was heavily criticised for a
long time, because her ideas were in many respects similar to those now
popularised by Florida.
The HST itself is not primarily mentioned as a factor of quality of place.
Its appreciation differs between the cases analysed here, and moreover, it
has two faces. It is valued as it contributes to the image of a project, par-
ticularly when it is located in a city with no great reputation of its own. Its
appreciation as a means of transport depends on the projects position in
the HST network, as well as the position relative to an international air-
port. The South Axis is not a station-based project in the sense that it is
dependent on, or even induced by, the HST. The HST is considered an
added value, not an essential precondition for the projects success. Rot-
terdam Central and especially Euralille are in effect induced by the HST,
even while the deep motivation of the projects lies largely in the local con-
text. This is the most obvious in the case of Euralille, located rather cen-
trally on an important node, whereas Rotterdam and Amsterdam are much
more peripheral in the European HST network.
More abstract ideas play a role in the background. In this respect, the
conceptual ideas on the city of Koolhaas versus De Bruijn and the design-
ers of Rotterdam Central seem crucial. Euralille is conceived primarily on
the basis of its position in international networks, the South Axis and Rot-
terdam Central as part of existing cities. Unlike Euralille, the South Axis
and Rotterdam Central are largely based on the idea of the traditional street
as the model of urban life. However, other factors affect urban design as
well, such as especially in these station areas the existing infrastructure
pattern. As it is, Euralille largely fulfilled its objectives set on a macro-
level, but it has a limited urban quality on the street level. Perhaps the pro-
ject is saved by its modest scale vis--vis La Dfense and, for an interna-
tional business centre, its relatively small number of offices; ironically the
one aspect in which it did not, for a long time, fulfil its purpose. The ideas
behind the South Axis and Rotterdam Central better match in many aspects
with the essence of quality of place, but they have yet to be proven.
On paper, both the South Axis and Rotterdam Central are way ahead in
terms of urban quality of such archetypical business locations as Canary
Wharf and La Dfense; Euralille is at present already much better. Will it
be sufficient to convince a sceptic observer? Probably not. But it is likely,
on the other hand, that the urban planner will be content and perhaps, in si-
lence, reassured.
98 Jan Jacob Trip
Acknowledgements
I have benefited from discussions with Martin Aarts and Robert Klooster-
man, both of whom I would like to thank. Furthermore, I thank Marjolein
Spaans, who has provided valuable information on Euralille. Finally, I ex-
tend my gratitude to all interviewees. The research on which this chapter is
based is co-funded by NWO-Connekt.
References
Alsop Architects (2001) Rotterdam Central Ontwerp Masterplan. Projectbureau

Rotterdam Central/Alsop Architects, Rotterdam/London
Bertolini L (2000) Planning in the Borderless City: A Conceptualisation and an
Application to the Case of Station Area Redevelopment. Town Planning Re-
view 71:455 475
Areas. Spon, London/New York
Bruijn P de (2005) The Intentional City: Applying Local Values and Choice in a
Global Context. Assuring Civic Quality, Achieving Urban Excellence. Paper
presented at the IFHP Spring Conference, Portland
Dovey K (1998) Multiplicities and Complicities: Signifying the Future at Eu-
ralille. Urban Design International 3/8:89 99
DRO [Gemeente Amsterdam, Dienst Ruimtelijke Ordening] (1998) Masterplan
Zuidas, Amsterdam
DRO [Gemeente Amsterdam, Dienst Ruimtelijke Ordening] (2001) Visie Zuidas;
Stand van zaken maart 2001. Amsterdam
DRO [Gemeente Amsterdam, Dienst Ruimtelijke Ordening] (2004) Visie Zuidas;
Stand van zaken maart 2004. Amsterdam
Florida R (2002a) The Rise of the Creative Class and how its Transforming
Work, Leisure, Community, and Everyday Life. Basic Books, New York
Florida R (2002b) Rebuilding Lower Manhattan for the Creative Age: Implica-
tions for the Greater New York Region. Prepared for the Regional Plan Asso-
ciation and the Civic Alliance, www.creativeclass.org
Franke S, Verhagen E (eds) (2005) Creativity and the City: How the Creative
Economy Changes the City. NAi Publishers, Rotterdam
Gemeente Rotterdam (2003) Programma van Eisen Rotterdam Centraal. Rotter-
dam
Gemeente Rotterdam (2005) Concept Stedenbouwkundig Plan Stationslocatie.
Rotterdam
Gospodini A (2002) European Cities in Competition and the New Uses of Urban
Design. Journal of Urban Design 7:59 73
Jacobs J (1961) The Death and Life of Great American Cities. Random House,
New York
Kloosterman R, Trip JJ (2004) Gestold Modernisme: Een Analyse van de Rotter-

damse Economie vanuit een Postindustrieel Perspectief. In: Becker F, Henne-
keler WR van, Sie Dhian Ho M, Tromp B, Linthorst M (eds) Rotterdam: Het
Vijfentwintigste Jaarboek voor het Democratisch Socialisme. Mets &
Schilt/Wiardi Beckman Stichting, Amsterdam, pp 39 57
Kloosterman R, Trip JJ (2006) Planning for Quality? Assessing the Role of Qual-
ity of Place in Developing High-Speed Railway Stations. Paper presented at
the International Conference on Urban Conditions and Life Chances, Am-
sterdam
Kooijman D, Wigmans G (2003) Managing the City: Flows and Places at Rotter-
dam Central Station. City 7:301 26
Koolhaas R, Mau B (1995) Small, Medium, Large, Extra Large. 010 Publishers,
Rotterdam
Koolhaas R, Nouvel J, Portzamparc C de, Vasconi C, Duthilleul JM (1996) Eu-
ralille: The Making of a New City Centre. Birkhuser, Basel/Boston/Berlin
Kresl PK (1995) The Determinants of Urban Competitiveness: A Survey. In:
Kresl, PK, Gappert G (eds) North American Cities and the Global Economy:
Challenges and Opportunities. Sage, Thousand Oaks/London/New Delhi, pp
45 68
Ploeger R (2004) Regulating Urban Office Provision: A Study of the Ebb and
Flow of Regimes of Urbanisation in Amsterdam and Frankfurt am Main,
1945-2000. University of Amsterdam, Amsterdam
Rowley A (1998) Private Property Decision-Makers and the Quality of Urban De-
sign. Journal of Urban Design 3:151 73
Salet W, Majoor S (eds) (2005) Amsterdam Zuidas European Space. 010 Publish-
ers, Rotterdam
Segedy JA (1997) How Important is Quality of Life in Location Decisions and
Local Economic Development? In: Bingham RD, Mier R (eds) Dilemmas of
Urban Economic Development: Issues in Theory and Practice. Sage, Thou-
sand Oaks/London/New Delhi, pp 56 81
Spaans M (2002) The Implementation of Urban Revitalization Projects. Delft
University Press, Delft
Team CS (2005) Voorlopig Ontwerp Rotterdam Central.www.rotterdamCentral.nl
Tilman H (1994) Vruchtbare Reductie van de Complexiteit: De Stedebouw van
Euralille. De Architect, 12:22 30
Trip JJ (2006) Measuring Quality of Place: A Comparative Analysis of Quality of
Place in Amsterdam and Rotterdam and Its Role in Large-Scale Urban Rede-
velopment. Paper presented at the Planning Research Conference Global
Places, Local Spaces, London
UCL [University College London] (2001) The Value of Urban Design.
CABE/DTLR/ Thomas Telford, London
VROM [Ministerie van Volkshuisvesting, Ruimtelijke Ordening en Milieu]
(2003) Nieuwe Sleutelprojecten in Aantocht; voortgangsrapportage December
2003. The Hague
www.saem-euralille.fr
6 New Key Projects for station redevelopment in
the Netherlands
Stan Majoor and Dick Schuiling

6.1 Introduction
Not only is railway station redevelopment related to urban dynamics, it is

also a dynamic process in itself. The wave of investments in new high-
speed train tracks and (new or upgraded) stations sweeping through West-
ern Europe over the last decades has led to numerous questions regarding
the content of these major urban investments and the processes used to de-
cide upon them. We investigate this issue exactly at this borderline. We
define two important policy challenges for railway station (area) develop-
ment as i.) the creation of integrated spaces: a content dimension related to
the actual physical object of the station and station areas and its use, and
ii.) the organisation of collective action in a situation of fragmentation: a
process dimension related to decision-making and institutional aspects of
station redevelopment.
This distinction is used to critically analyse and evaluate the latest gen-
eration of major station redevelopment projects in the Netherlands, offi-
cially called New Key Projects (Nieuwe Sleutelprojecten [NSP] in Dutch).
The Key Project policy is a modern mainstay of Dutch national spatial
planning. Since the Fourth National Spatial Planning Report of 1988, pol-
icy initiatives under this generic label have been formulated by national
government to foster strategic projects at the local planning level (Schuil-
ing 1996). The second and until now last generation of Key Projects
was announced in 1997. Its focus is entirely on the proposed stations and
the direct surroundings of the future stops of the High-Speed Train (HST)
in the Netherlands.
102 Stan Majoor and Dick Schuiling
At the beginning of the 1990s, the Dutch government presented plans

for two connections to the unfolding pan-European HST network: an HST
South connecting Amsterdam with Belgium (Antwerp Brussels) and an
HST East connecting Amsterdam with Germany (Ruhr area). Taking into
account the considerable delays in decision-making and implementation of
the plans (the HST South will be in full operation only in 2010, while the
HST East as dedicated high-speed track has been cancelled for the time be-
ing), the Netherlands is regarded as one of the latecomers to the party of
western European countries implementing high-speed train infrastructure.
We will not focus in our analysis on the difficulties regarding the infra-
structure between the cities, but rather on the decision-making process of
the designated stations on the two lines. Originally five HST stations were
selected in 1997 as New Key projects by the Cabinet: Rotterdam Central
and The Hague Central on the HST South line, Utrecht Central and Arn-
hem Central on the HST East line and Amsterdam South Axis, possibly on
both lines. Through the intervention of Parliament, a sixth project was
added in 1998: Breda Central on the HST South route (see Figure 6.1). Al-
though many high ambitions have been observed since the mid-1990s, the
implementation of the policy has been rather slow and problematic until
now. In this chapter we first formulate the main challenges for decision-
making on the integrated railway station in section 6.2, and then assess the
development of these future Dutch HST stations in sections 6.3 and 6.4.
6.2 The challenge of integrated railway station

development
In this paragraph we briefly introduce the two most important, and interre-
lated, policy challenges for contemporary railway station redevelopment:
i.) the creation of (new) integrated spaces (content dimension), and ii.) the
organisation of integrated decision-making as a form of collective action in
a situation of fragmentation (process dimension).
6.2.1 New networks, nodes and multiple land use: Creating

integrated spaces
The development of the high-speed train means that over time a new pre-
mium layer of connectedness is being created between the (new) nodes in
the HST network. This has two major consequences for decision-making.
First, on a meta-level, one can observe fierce competition between cities
(and regions) to obtain a node on this emerging network, due to a strong
New Key Projects for station redevelopment in the Netherlands 103
belief among policy makers and planners in the economic and symbolic
potential of a connection, and a subsequent fear of missing the train if not
connected. This often leads to quite politicised, rather than technical, deci-
sion-making processes regarding the actual track and stops of the train, as
our Dutch case study will later reveal.
Figure 6.1. Overview of the HST network in the Netherlands and the six Key
Projects
A second consequence is that the eventual designation of a station as a

future HST stop is often the trigger for large-scale station redevelopment
plans that go much further than merely the technical accommodation of the
arrival of the new trains at the station. It is important, however, to make a
general distinction between the geographical (and functional) locations of
these projects. In some cases (almost) completely new stations located out-
side the traditional urban cores are planned. Notable examples are station
projects in Amsterdam (South Axis) and Lyon (Gare de la Part-Dieu).
However, most projects are (existing) stations located in complex central
city areas, such as Lige (Guillemins), Berlin (Hauptbahnhof), or Rotter-
dam (Central station).
A (future) HST station is often seen as a key device to foster an urban

renaissance, by integrating it with different physical, socioeconomic and
cultural policies (Bertolini and Spit 1998; Bertolini 2000). Plans for station
redevelopment often refer to discourses on intensification, mixing land
uses, and the creation of an urban environment. The policy rationale for
this seems to be in utilising the higher level of connectedness (node-value)
especially if the high-speed train is connected to a variety of other public
and private means of transportation to create more place-value (Bertolini
and Spit 1998). Although the HST itself is primarily only a modest deliv-
erer of passengers compared to other means of transportation at the sta-
tions, its image is often expected to be of great value.
In the case of totally new stations in traditionally non-urban areas, there
is a huge challenge in connecting a quite sudden and partly still artificial
centrality with an urban program. For HST station projects in existing
urban areas (mostly expanded existing stations), the challenge is somewhat
different, since greater care has to be taken about potential difficulties in-
tegrating the project in the environment: complicated schemes of land
ownership, soil contamination, existing buildings and owners, difficulties
with the infrastructure in historic cities, etc. Nevertheless, these locations
offer far more natural opportunities to functionally weave the station (and
its immediate environment) into the existing city.
In both cases revenues from new real estate developments in and near
the station are hard financial necessities to compensate for the expensive
public parts of most projects: the station itself, public spaces and additional
infrastructure. As we will discover, complex sets of arrangements among a
multitude of actors regarding the redistribution of money over profitable
and costing components of the station projects are then a necessity.
Nevertheless, even if expectations for the potential of station redevel-
opment to create mixed and vibrant urban places have been high, the actual
spatial performance of most projects has not yet been very convincing,
according to critical scholars. Although stations themselves are public
spaces par excellence, the results of large-scale redevelopment of station
areas in this respect are sometimes criticised. This is especially true for the
new peripheral station areas. Overviews of a variety of large-scale urban
projects in Europe, which also include station-projects (Moulaert et al.
2003; Salet and Gualini 2007; Trip 2007), indicate that the projects capac-
ity to become integrated urban places is not yet self-evident. Most places
are purely functional, privatised, controlled areas of elite consumption that
feel spatially detached and hardly really urban.
We can therefore derive two major evaluative questions in our content-
dimension for the future Dutch HST stations. The first question pertains to
the HST itself: Does its arrival really create a new premium layer of nodes
with a substantially higher level of connectedness as its proponents ex-

pect? Our second question deals specifically with the development of the
station and station area: Are the opportunities of an HST connection, in the
form of an expected higher node value, capitalised on to create a new cen-
tral mixed high quality environment?
It is important to consider these questions in conjunction with a more
comprehensive understanding of the decision-making process on large sta-
tion projects. Though some actors certainly show ambitions for an integra-
tive program, they operate in fragmented governance settings wherein
some form of collective action has to be organised.
6.2.2 Organising collective action in a situation of

fragmentation: Integrated decision-making?
The complexity of station (area) development is overwhelming, especially

in a situation where ambition regarding integrated spaces is high. Between
the start of the initial ideas and the actual realisation of a project, political
and economic cycles can occur, as well as accompanying changes in goals,
program and phasing. Organising a project demands the integration (in
time and space) of investments of a large number of actors who are often
mutually-dependent. In most cases the goals and means of actors are di-
verse. Most often the nation state (sometimes with an EU subsidy) is re-
sponsible for transportation infrastructure (HST tracks), but is only mar-
ginally interested in the actual development of the station and its direct
area, an issue which is decided in a completely different arena. In most
cases a lower (regional, metropolitan) level of government is responsible
for other forms of infrastructure in the station area, such as regional trains,
light rail, or the metro.
Local authorities frequently take the lead in initiating projects as part of
the urban renaissance discourse mentioned above; mostly they are the
strongest defenders of integrating urban concepts. Although they regularly
hold some trump cards (especially juridical), they certainly depend on oth-
ers to realise these ambitions. Self-evidently, most development around the
station is undertaken by private actors who are not naturally interested in
investing in less profitable omnipresent parts in these projects, since large
public spaces (plazas) and infrastructure facilities are necessary in order to
handle passengers. Since most HST station areas aim to attract high end
national and international businesses, the capriciousness of the office mar-
ket in this segment and the changing positions in the competitive landscape
of projects, are also major uncertainty aspects.
Within the station building itself the situation is often even more com-
plicated. Here a program has to be set up to match the requirements of the
(different and sometimes competing) operators of the infrastructure, more
frequently nowadays (semi) privatised actors, and actors who represent the
public value of the station as an important public space. The role in these
complex processes of the actual users, especially the passengers, is mostly
unclear.
One essential question is how to organise collective action in a situation
of fragmentation: how can a variety of actors with different goals, means
and strategies come to concerted action in such long, costly and techni-
cally-complex projects? A typical solution is to establish special project-
oriented modes of governance, with close ties between public and private
actors that frequently bypass, or at times even overrule, traditional forms
of deliberation associated with spatial planning in western democracies
(Moulaert et al. 2003). Necessary as it might look to handle the complex-
ity, these new forms are not without danger. The result is that these pro-
jects are vulnerable to criticism as neo-corporatist undertakings, set up and
managed business-style by small public-private elites informed by inter-
national consultancy agencies and trophy architects. Behind such a cur-
tain, public agencies desperate for private investment in office and housing
development, often take the greatest risks.
Therefore, it is important to find the right balance of governance innova-
tion. Decision-making processes have to be flexible enough to handle
fragmented goals and the numerous uncertainties related to these long-term
development projects, while at the same time must be vigorous enough to
arrive at concrete results. Appealing images of future urban qualities, as
sketched in the previous paragraph, can help pinpoint and align the frag-
mented preferences of actors towards a common discourse on place quality
(Gualini and Majoor 2007). However, if they are not accompanied by a
sound process-architecture, their value often stays on the drawing board.
Importantly, however, is that in cases like the Dutch New Key Projects,
there is not only a form of strategic complexity between actors (differences
in goals, for example), but there is also a large institutional complexity that
sometimes hampers concerted action in these projects (Klijn 2001). The
crux of the matter therefore seems to be not only to organise effective
forms of consensus-building within projects, but also to accompany them
with a growing awareness of the importance of connectivity between dif-
ferent domains of societal action where the projects have to be realised
(Salet et al. 2003). Only when projects are integrated in different domains,
including governmental: all the way from local authority to the EU (for
different forms of political, legal and financial support) private (for a
sustainable flow of private investments) and civic (for democratic sup-
port and societal identification with a place), can conditions be created for
an effective process of consensus-building at the project level (Majoor
2005).
The questions derived from this domain for our case study are therefore:
i.) to what extent have governance forms been capable of dealing with the
complexity of station area development, both in the relation between dif-
ferent branches and layers of government and in the cooperation between
public and private actors? and ii.) to what extent have innovative forms of
connecting governance been set up?
6.3 The New Key Projects in the Netherlands
6.3.1 Goals, funding, process
The future connection of the Netherlands to the new European HST net-
work and the new possibilities this connection is expected to create for ur-
ban development around stations are the main reasons for the second round
of Key Projects. Six projects were initiated by local governments for the
future (2010) seven HST stops. The seventh HST stop is Schiphol Airport
near Amsterdam; this station was already the centrepiece of an integral de-
velopment in the mid-1990s. The projects are situated in the four largest
cities (Amsterdam, Rotterdam, The Hague, and Utrecht) and in two me-
dium-sized cities near the border with Germany or Belgium (Arnhem and
Breda) (see Figure 6.1 above). According to the national government,
these six projects must contribute substantially to the national policy goal
to improve the physical and economic structure of the country by:
providing new employment;
developing better management of the growing mobility;
leading to intensive land use policy in urban areas;
improving the socio-economic vitality of the urban areas (Minis-
terie VROM, 2006).
Having a designated budget from an interdepartmental structure fund to
strengthen the economic base of the country makes this generation of Key
Projects different from previous ones. The budget comprises 244 million
(NSP-1) for improving the spatial quality of the surrounding public space
and 102 million (NSP-2) for the qualitative improvement of the transit
terminals, and is managed by the Ministry of Housing, Spatial Planning
and the Environment (Ministerie VROM). This special grant gives the
ministry a position at the table in all projects in order to negotiate the con-
ditions for the subsidy; the most important condition is that the process be
integral, where public (local government) and private parties work together
to realise goals in both infrastructure and urban revitalisation around the
future HST stations.
However, the involvement of the Ministry of Transport, Public Works
and Water Management (Ministerie van Verkeer en Waterstaat or V&W)
probably has higher importance. In the framework of the national Multi-
year Program for Infrastructure and Transport (MIT) nearly 1 billion is
set aside: 640 million for large stations (MIT-1) and 329 million for re-
lated infrastructure (MIT-2). Another 136 million is set aside for the
South Axis in Amsterdam. All projects involve major infrastructure in-
vestments in preparing the station and surrounding area for the future con-
nection of the HST (see Table 6.1).
Table 6.1. Overview of national government investments (in millions of ) in the

New Key Projects
NSP-1 NSP-2 MIT-1 MIT-2 addi- NoMo North Total
large sta- tional Wing Randstad
tions infrastructure infrastructure
Amsterdam 70 69 91 287 136 653
South Axis
Rotterdam 54 0 161 - 215
The Hague 34 15 81 - 130
Utrecht 54 - 253 - 307
Arnhem 16 9 24 35 84
Breda 16 9 30 7 62
Total 244 102 640 329 136 1451

Source: Ministerie VROM (2006)
The two ministerial sources of finance create a decision-making situa-

tion in which local authorities have to address both private investors and
the national government in these projects. Spatial planning policy in the
Netherlands is mainly a local affair. In order to regulate its role in these lo-
cal planning processes and structure its progress, national government set
up a framework for the planning and decision-making. Its four phases are:
Exploration: National government explores whether the locally-
initiated project receives the status of a New Key Project;
Fact finding: local authority and national government together
sketch the content and financial aspects of the project and reach an
intentional agreement;
Plan making: local authority makes a masterplan which will after-

wards be examined by the national government;
Financial agreements: implementation agreements about public-
private partnerships and the definitive national government grant.
We now briefly describe the state-of-affairs of the six projects (as of be-
ginning 2006).
6.3.2 Amsterdam, South Axis
The South Axis project in Amsterdam is one of the largest integrated urban
development projects currently underway in Europe. The project consists
of an area of 210 hectares around the southern ring road and railway lines
of Amsterdam, and is conveniently located close to the historic core of
Amsterdam and the Schiphol airport. The epicentre of the project will be a
public transportation node to facilitate rail and subway, and (possibly) a
high-speed train connection (Gemeente Amsterdam 2004). The initiation
of the project in the mid-1990s was strongly associated with private initia-
tives, especially when the financial and legal sectors changed their prefer-
ence to this location from the traditional office locations in the historic
core (Ploeger 2004; Salet and Majoor 2005; Majoor 2007).
The project emphasis has recently shifted towards a radical mix of uses
(Majoor 2006). While the initial plans had specifically focussed on the po-
tential for office development in the area (Gemeente Amsterdam 1998),
current plans indicate a ratio of 45-45-10 for offices, housing and facilities,
respectively. Claiming a new strategic regional centrality of the area, the
city of Amsterdam expects that South Axis will develop as a new second
urban centre of Amsterdam (Gemeente Amsterdam 2003).
While some parts of the project have already been realised, the plan to
cover the central infrastructure bundle (the so-called dock model) seen as
a necessary prerequisite to create a real integrated area, was delayed due to
a stalemate between Amsterdam and the national government (responsible
for infrastructure investments) (Majoor 2004). Both costs and benefits
especially the rentability of the huge office program are uncertain and a
negative outcome is most probable (CPB 2003). Without the dock, the am-
bitions on total program and integration of uses must be limited to a dike
model where the area remains divided into two parts. However, between
2004 and 2006, strong city-business lobbying behind the project was able
to mobilise more public and private support and investment in an attempt
to create a limited South Axis company to build the infrastructure in tun-
nels, together with the station, and develop the parcels on top of it. Final
decisions on this issue are expected in 2007.
6.3.3 Rotterdam, Central Station
Travelling from France and Belgium, Rotterdam, the second largest Dutch
city, will be the first real HST stop in the country. Rotterdam Central is
situated at the edge of the inner city area, which was rebuilt after heavy
bombing during World War II. Due to the arrival of the HST South and the
completion of a regional light rail network, the number of daily travellers
is expected to rise from 140 000 now to approximately 220 000 in 2010.
The goal of the project is to create a high quality public transport hub for
HST, train, RandstadRail (light rail), metro, trams, buses, and taxis (Kooij-
man and Wigmans 2003).
Transport goals are combined with plans for an extensive program of of-
fices and apartments and an improved public realm. The city of Rotterdam
has initiated the project and after an earlier ambitious master plan (made
by Alsop), new agreements were signed between the City and the national
government in 2003. The new team of designers (Team CS: Benthem
Crouwel, Meyer and van Schooten, West 8) delivered a new design for the
terminal and an urban design plan for the whole area, with plenty of atten-
tion given to public space. The existing network of streets, building fa-
ades and public ground floors will extend to the rail tracks, including a
connection between the south (front) and north (back) side of the station;
the present station square will be covered by a gigantic roof. The opening
of the new terminal is scheduled for 2009, with ground preparation work
and the construction of a new car underpass already underway.
6.3.4 Utrecht, Utrecht Station Project
Utrecht was present in both generations of Key Projects with this project
related to upgrading its central public transport station and direct environ-
ment. Built in the 1960s, the station was the most important node in the na-
tional train network and an important terminal for local and regional buses.
An indoor shopping centre crossing over the rail tracks, offices, and other
facilities were integrated. In Corbusian style different uses and modes of
transportation are separated, but nowadays this concept is perceived
widely as unpleasant and lacking in spatial quality. Instead of a connector,
the whole station complex acts as a massive barrier between the two sides
of the railway. Also due to the future arrival of the HST East (to Ger-
many), and thus an increase in ridership, the station has to be enlarged two
to three times its original size.
Different generations of proposals have tried to address these issues by
drawing up large integral masterplans for transforming the area. But con-
sensus building between the different actors was unable to endure beyond
preliminary planning phases. At the beginning of the 1990s the project
failed, largely as a result of the complexity of ownership in the area and
divergent wishes and demands of the involved actors (Verbart 2004).
Realisation of the project seemed even further away than ever by 2001
because public support for the project began to dwindle. The issue of pub-
lic support is like an exposed nerve, especially in Utrecht, due to the his-
tory of the current station area as an example of 1960s top-down decision-
making, which had brutalised Utrechts historic city structure. A local one-
issue party criticising the lack of progress and opposing the present plan as
megalomaniacal, won the local elections and successfully halted the cur-
rent plan (with the exception of the station itself).
Once in power they required a referendum to choose between two com-
parable new variants for the area: Plan 1 (Enlarged) and plan A (Compact).
The 2002 referendum resulted in the vast majority of the population prefer-
ring Plan A with fewer offices and shops, more housing, and two main
axes instead of one. Plan A was elaborated into a master plan as phase 4 in
2003, and the Key Project approach could be concluded in 2004 with an
implementation agreement between central government and the City of
Utrecht. Development Agreements are settled with the three other main ac-
tors: Dutch Rail Real Estate, Crio, and the Utrecht Trade Fair. The irony
is that, for the first time in the history of the project, and thanks to the op-
posing local party coming to power and holding its referendum, realisation
of the project (taking 10 years) now seems imminent.
6.3.5 The Hague, Nieuw Central
The Nieuw Central project in The Hague is part of a large inner city rede-
velopment strategy called New Centre conducted over the last decade. The
Hague Central Station is the only terminus station of the Dutch Key Pro-
jects. Built in the 1970s, it combines national train services with light rail,
trams, and local and regional buses. The purpose of the Nieuw Central pro-
ject is to expand the capacity of the station for the HST, provide necessary
infrastructures for a new regional light rail system (RandstadRail), and im-
prove its layout. The direct area of the station, now an unpleasant area
lacking urban design quality, will be revitalised. The Babylon shopping
centre will be upgraded with new offices and apartments. A new under-
ground parking garage will replace the old and open up the area with a
square in the middle. Research is in progress to identify how to build on
top of the railroads; this concept has already been elaborated on a much
larger scale (the very ambitious Hoog Hage plan, from 1997). However,
the vision was shelved for being too expensive and unrealistic in the short-
term.
Planning for the Nieuw Central project was delayed because of difficult
negotiations with the Ministry of Transport concerning the station revitali-
sation. Private investor commitment was also weak in a city with numer-
ous other interesting office locations. But in 2003 the master plan was ap-
proved by central government after a few necessary adaptations.
Construction started in 2004 with alterations of the platform for the trams
and the viaduct for the coming RandstadRail. The new terminal is ex-
pected to be ready in 2009 and the rest of the station area complete by
2011. The project will be the final piece of the renewal of the centre of The
Hague, but it is peculiar that The Hague Central Station is no real HST
stop. The direct HST from Amsterdam via Schiphol, Rotterdam to Bel-
gium and France will not stop there. Instead, the city will obtain a direct
connection with Brussels eight times per day, and a shuttle to the HST in
Rotterdam. The ultimate outcome for the international City of Peace and
Justice will turn out to be a somewhat disappointing second-class connec-
tion to the international HST network.
6.3.6 Arnhem, Arnhem Central
Arnhem is the first of three designated Dutch stops for the HST East arriv-
ing from Germany and is the capital of an internationally high-ranking
border region. The program focuses mainly on the terminal and nearby of-
fice development, but improvement of the public realm as a whole is also a
priority. The 24-hectare project aims for the integral development of the
train station and its direct surrounding area; a much larger (40 hectares)
revitalisation project in the inner city towards the river Rhine is also un-
derway: Rijnboog.
Planning of the station area (involving national government, local au-
thority, owners and investors, as well as regional government of the prov-
ince) began 10 years ago. Thus far the traffic tunnel, railway viaduct, un-
derground car park, bus station, and office towers are finished.
Construction of the new terminal will be undertaken from 2007-2009. The
terminal design by UN Studio and Arup, with its waving roof, walls and
floors, won an award for innovative, multiple and underground use of
space in 2005. The realisation agreement for the whole project was signed
by the city of Arnhem, the Ministry of VROM, the Ministry of V&W, and
Dutch Rail in 2005. The local morphology, with height differences of up to
20 metres, has been very well utilised.
6.3.7 Breda, Breda Stationskwartier
Breda is the smallest of the six New Key Projects, and like The Hague, not
a real HST station. The 12.6-hectare project comprises the Breda station,
which will serve as a future HST shuttle stop, and the immediate surround-
ing area. With the arrival of the HST shuttle in Breda, travel time towards
Amsterdam and Rotterdam and Antwerp, Belgium will be reduced spec-
tacularly by 50%, and a third train platform has already been built. Ac-
cording to the architect Van Velzen, the terminal is not a station, but rather
is one comprehensive building for the city, where all functions are com-
bined under one roof: train terminal, bus terminal, car park, commercial
uses, pedestrian underpass, 130 dwellings, and offices. Financial support
comes from national government, the province, and Dutch Rail. The con-
struction of the terminal is scheduled between 2007 and 2010. The station
area is part of a much larger development scheme, Via Breda, to create 1
million m2 floor space on 160 ha., which should be ready by 2025.
6.4 Analysis
Following the outline discussed in section 6.2, we now analyse the level of
integration of spaces (content) and the integration of decision-making
(process) of railway station development illustrated with the New Key Pro-
jects.
6.4.1 Creating integrated spaces: Urban planning assessment

criteria
Does the arrival of the HST really create a new premium layer of nodes
with a substantially higher level of connectedness, as its proponents ex-
pect? Are the opportunities for an HST connection, in the form of an ex-
pected higher node value, being capitalised on to create a new central
mixed high quality environment?
HST connection: A new premium layer of connected nodes?
Longer distances with few stops only in large cities are ideal for an HST
network. However, HST in the Netherlands shows different characteristics:
short high-speed tracks in a small country where the southern border with
Belgium and the eastern border with Germany are only 120 kilometres
away from the capital Amsterdam, the terminus of both lines. Although the
country is heavily urbanised, the largest four individual cities are modest
in size (between 260 000 and 750 000 inhabitants), all located in the so-
called Randstad (Rim City) Holland, relatively near to each other in the
west of the country. Selection of the number and location of the HST stops
is therefore difficult (see the case of The Hague bypass) and a complaint
about a possible Randstad bias is also probable (see the addition of Breda
by Parliament). Most existing large railway stations are centrally located in
the late 19th century zone adjacent to the inner city in an often mixed-use
but somewhat second-class environment. They either are or are increas-
ingly becoming the central node for all the modes of regional and local
public transport, combined with kiss-and-ride car traffic and bicycle park-
ing.
The main selection criterion for becoming a New Key Project in 1997
was to be a proposed HST station along the HST South or the HST East.
Since then, the number of real HST stations in The Netherlands has dimin-
ished year on year, starting with the degradation of The Hague and Breda
to only a stop for the HST shuttle (to Rotterdam or Antwerp), leaving Am-
sterdam, Schiphol Airport and Rotterdam as the only real HST stations
along the through-going HST South (Figures 6.1 and 6.2). The second deg-
radation was for Utrecht and Arnhem with the decision in 2001 to cancel
the HST East on dedicated tracks. The estimated extra ridership and travel
time gain between Utrecht and the German border were too low to justify
huge investment, especially after the cost overruns of the new freight rail-
way line Rotterdam-Germany. Instead, the choice was to use existing
tracks that permit a maximum speed of 140 kms (the same as the present
Intercity) as a kind of pseudo-HST.
At the beginning of this century it became clear that for Amsterdam, at
least until 2022, Amsterdam Central would be the HST terminus, not Am-
sterdam South Axis. The operator of the line, the High-Speed Alliance (NS
and KLM Royal Dutch Airlines), and not the subsidising government can
decide at which Amsterdam station this fast connection will call. So from
the six stations designated as HST stations in the New Key Project ap-
proach, only one (Rotterdam Central) deserves the special status. The oth-
ers have a second-class position in terms of HST connectivity, although
this is already a major improvement for Breda. The mismatch between this
lack of real HST connectivity for most of the six stations and the policy to
selectively spend 1.5 billion of central government money for HST sta-
tions and their environments with this premium connection is rather disap-
pointing. To smooth over this bleak result, one could say that apart from
the HST, additional investment in new regional and local transit connected
with the station (like RandstadRail, bus stations or a subway line) is made,
which will improve the hub status of the station and its overall connec-
tivity. Real high-speed railway lines are difficult to realise and very expen-
sive in the small but densely urbanised Netherlands while the time gain,
and therefore increased ridership, are small.
Source: www.hsa.nl
1 line = 1 train per hour (6 trains per hour between Amsterdam Central and Rot-
terdam Central)
Figure 6.2. Connections, stops and frequency of HST-South (in full operation in
2010)
HST stations: New mixed high quality environments?
Tables 6.2 and 6.3 give a quantitative overview of the program mix in the
six New Key Projects. In both tables uncertainty about Amsterdams dock
or dike model is incorporated, since it heavily influences the numbers.
One can first observe that Amsterdam South Axis (dock model) over-
shadows all other projects in size: 2.30 out of 3.76 million m2 or 61% of all
proposed NSP real estate. The overall volume, proposed density or floor
space index of these projects are much higher than at present, and quite
high for Dutch circumstances. In most of the station redevelopment pro-
jects, office use is dominant but the share of housing is remarkably high,
especially in Utrecht and Amsterdam South Axis compared to foreign pro-
jects. Social housing is often included, which mirrors the Dutch goal for
balanced neighbourhoods. Mixed uses are omnipresent, including facili-

ties. Together with the high densities this will help create new forms of ur-
banity. The attention paid to better use of railway stations and public
transport by concentrating activities and buildings near or on top of them,
is in line with national and local planning objectives. There is strong focus
during the planning stages on the public realm, and in some cases this is
even the integrator of different sub-areas. Such a focus is in marked con-
trast to the sometimes dilapidated situations originally present in those sta-
tion environments.
Table 6.2. Overview of the proposed building program (in square meters) of the
New Key Projects (status 2006)
Offices Housing Facilities Total
Amsterdam South Axis, dock 985 000 1.000 000 320 000 2.305 000
Amsterdam South Axis, dike 460 000 335 000 166 000 927 000
Rotterdam 183 000 84 000 19 000 286 000
The Hague 114 500 53 000 132 000 299 500
Utrecht 178 500 247 400 145 400 571 300
Arnhem 80 000 7 000 5 900 92 900
Breda 117 000 74 000 14 500 205 500
Total (dock) 1.658 000 1.465 400 636 800 3.760 200
Total (dike) 1.113 000 800 400 432 800 2.346 200
Source: Ministry VROM (2006) p. 37
Table 6.3. Proportions of different uses in the proposed building programs of the
New Key Projects (status 2006)
Offices Housing Facilities
Amsterdam South Axis, dock 43 43 14
Amsterdam South Axis, dike 46 36 18
Rotterdam 64 29 7
The Hague 38 18 44
Utrecht 31 43 25
Arnhem 86 8 6
Breda 57 36 7
Total (dock) 44 39 17
Total (dike) 42 36 22
Source: Ministry VROM (2006) p. 37
It is problematic that, within their institutional logic, NS and the Min-

istry of Transport do not consider the improvement of spatial quality and
the public realm as their business; they only support terminals at the basic
station level. Contributions from the Ministry of VROM, the municipality
and, wherever possible, the private sector, are therefore indispensable to
the creation of a higher level of public space. Already it is clearly visible

that some physical objectives of the NSP approach will be achieved, such
as creating mixed-use areas with intensive and multiple land use or im-
proving the vitality of the station areas. The quality level of the public
realm will be considerably higher, compared to other station areas in the
Netherlands, but it is still uncertain if they match European standards.
6.4.2 Creating integrated decision-making
We posed two questions in section 6.2.2 for the process-dimension of our

research: i.) to what extent have governance forms been able to deal with
the complexity of station area development, both in the relation between
different branches and layers of government and in the cooperation be-
tween public and private actors? and ii.) to what extent have innovative
forms of connecting governance been set up?
Towards new forms of connecting governance in the Dutch New Key

Projects?
As mentioned above, the Key Project policy is a new way to combine the
efforts and funding of different government branches and layers with pri-
vate development of real estate. On paper it is an attempt to coordinate the
roles of a multitude of actors around station redevelopment to create inte-
grated decision-making. However, results from nearly 10 years planning
and negotiation among the parties highlight the difficulty of the task. The
consultancy firm Berenschot was contracted by the national government in
2005 to critically evaluate the organisation and operational procedures for
the New Key Projects, especially the relationship between local and na-
tional government (Kort et al. 2005). We will next distil some observations
and conclusions from their evaluation report.
The special NSP status was given to the six projects in 1997-1998
(Phase 1: Exploration section 6.3.1) and the minister of VROM should
coordinate the interdepartmental project organisation. In 2001 the so-called
Audit-Luteijn already suggested organisational improvements after wide-
spread criticism of insufficient cooperation among the different govern-
ment parties, and that a focal address at the central level was missing. Al-
though the goals of the individual projects largely remained unchanged
since 1997, the context wherein the projects had to be realised, changed
quickly: we mentioned earlier that not all stations become real HST sta-
tions; another problematic issue was a serious and longstanding slump in
the office market, putting real pressure on the profitability of real estate
developments in the station areas. At that time all projects finished Phase
2: Fact-finding with an intentional agreement between the various layers
of government. Phase 3: Plan-making took much longer, and making a
masterplan to include all public and private parties has proven to be no
simple task.
The meaning of the masterplan itself was not clearly defined, thus re-
sulting in major differences among the six projects. Some already contain
a draft design for the transit terminal, while others deliver it in Phase 4
Financial agreements for example. At the end of Phase 3 masterplans are
examined by central government on five aspects: urban design, traffic and
transport, economic effects, external safety, and financial feasibility. That
some institutions cooperating in the masterplan were also involved in ex-
amining it, is noteworthy. The first masterplan for Rotterdam Central was
rejected as far too expensive, and for some other projects approval was
conditional. The implementation agreements at the end of Phase 4 are set-
tled in The Hague, Utrecht and Arnhem; in Utrecht and The Hague these
agreements also contain benefit-sharing arrangements.
The parties involved have different opinions on various organisational
aspects; all see the NSP goals as very ambitious. The local/regional gov-
ernment parties say that the attributed status, abstract goals and open
framework have created high expectations in terms of financial support by
central government. There were remarkable differences in ambition be-
tween the two ministries involved: V&W aimed at and paid for only sober
and efficient basic stations, while VROM strived for high-standing station
environments. There are also differences between the two ministries with
relation to phasing the projects: VROM works well with the four NSP
phases, while V&W prefers the MIT phasing (on investments of large-
scale infrastructure). The cities and private parties each have their own tra-
ditional phasing for large projects. Local governments and private parties
have also made critical remarks on the lack of clarity of the examination
criteria, and they regard central government financial support to be too lit-
tle too late (known). Central government replies that the total NSP budget
and the indicative sums of money per project were announced at an early
stage. This clash of expectations has caused serious delays and the down-
scaling of ambitions. The central parties are satisfied about the distribution
of responsibilities, but some local/regional government parties differ in
that opinion and are critical that (the different business units of) Dutch Rail
and ProRail (the provider of rail infrastructure of V&W) are involved in all
projects but are nonetheless detached from the interdepartmental project
organisation and decision-making at the central level.
The splitting of the formerly integral Dutch Rail (NS) creates a new
complexity in and around stations; in Key Projects NS Stations, NS Real
Estate, NS Commerce and NS Travellers are often separately involved in

the decision-making process, while ProRail is responsible for construction,
maintenance and management of the rail infrastructure. NS is still a pub-
licly-owned company based on private law, and ProRail is a separate or-
ganisation under the auspices of the Ministry of V&W. So the question of
who owns and manages what in and around the HST terminal is not so
easy to answer. NS Real Estate and NS Stations recently merged per Janu-
ary 2007 to overcome some of the organisational fragmentation.
Innovative forms of connecting governance
Officially, the municipalities are the drivers as well as responsible admin-

istrators of the comprehensive Key Project. They take the initiative, are re-
sponsible for masterplan making, receive subsidies, set up partnerships
with the private sector, and sign implementation agreements with central
government. In this proactive role they have the difficulty that, for the fo-
cal point of every project, the transit terminal, their responsibility and
competence is often minimal, compared to the surrounding areas. Never-
theless, there are different arrangements; in The Hague and Arnhem the
city is also (co)principal of the terminal. Financial means, formal responsi-
bilities and risk-taking become increasingly important in later phases of
each project, and those are not the strongest points of municipalities. They
complain at times that they have excess responsibility for matters they
cannot control or steer. The successful process design of earlier genera-
tions of Key Projects with the municipality as driver is continued, but we
ask if this is appropriate for HST terminal Key Projects. In any case, the
process is not designed together with the municipalities but is instead im-
posed upon them, which is no guarantee for success.
The Berenschot Report advises central government to build in stimuli
for cooperation and achievement of central government objectives by all
other parties. To do so, they could e.g., guarantee funding for a basic sta-
tion and withhold extra reserve money for projects that have progressed or
represent high quality. A proper definition of phases, products and players
is essential to manage expectations and prevent endless discussion on who
delivers what and when to whom.
In order to reply to the question if and what kind of innovative forms of
governance are set up, we observed some institutional innovation within
central government: e.g., special separate funding for the New Key Pro-
jects instead of using regular subsidies. At the same time we observed that
this separate funding remained a bit fragmented, as it was coming from
two ministries and working with five separate budgets (see Table 6.1).
Over time, the inter-ministerial collaboration and account management
(one address) improved. Between the different layers of government there

was much less innovation visible.
The public/private cooperation is mainly in the hands of local govern-
ment and differs strongly from project to project. A real innovation is that
(in the case of Amsterdam South Axis only) central government probably
is willing to participate directly as a risk-taking shareholder in the limited
company South Axis-Dock. A real innovation at local level was the refer-
endum in Utrecht, ending the stalemate between the population and the
proponents of the project. It is doubtful that these few innovations are ca-
pable of handling the complexity of projects such as these.
A few final words can be said about the speed of procedures and imple-
mentation of the projects. Between generation 0 and 2 of the Key Project
approach, the speed of selection and decision-making decreased markedly.
Having four phases now instead of one, and four ministries instead of one,
simply exacerbates the decision procedures. Separate funding also does not
facilitate quicker central government approval. The splitting up of NS
made things worse by slowing the decision-making process. Compared to
speedy French decision-making and implementation procedures relative to
dedicated HST tracks and stations, the Dutch lag far behind. This is nicely
illustrated by the title of the Progress Report 2006 of the second generation
(nine years after its start): New Key Projects on Steam (Ministerie
VROM 2006). The steam (train) is too slow in an era of high-speed and the
stations will not be ready when the HST arrives! Fortunately, stations and
their environments are meant to endure for many years and initial prob-
lems cannot predict that the journey itself will not be pleasant or worth-
while.
6.5 Conclusions
In this chapter we have analysed the complicated process of realising a

new generation of station projects in the Netherlands. If we step back from
our findings in the previous section, what lessons can we learn from our
study concerning decision-making processes on HST station redevelop-
ment for projects in both the Netherlands and internationally?
Taking our dual perspective on content and process as a point of depar-
ture, it is first important to indicate the massive potential of station areas as
focal urban points in a spatially-fragmented society. In the Netherlands the
contribution of the high-speed train itself is modest, since it will primarily
replace existing intercity trips. However, as the stations develop towards
real integrated public transport nodes, with connections to various other
transport modes, especially regional rail networks, the connectedness of

these places offers a natural condition and a strategic asset for the devel-
opment of high density mixed urban spaces.
The problem for station redevelopment, as we perceive it, is not the lack
of potential, good planning intentions, or imaginative ideas. For all the
projects handled in this chapter, extensive exhibitions could present the
full range of sketches, plans and scale models made over the previous dec-
ade. Instead, the major challenge lies in the decision-making, and perhaps
even more in the institutional fragmentation that needs to be resolved. To
begin with the first, a major task of redeveloping stations is to manage a
process that occurs in different domains of interaction, where different ac-
tors operate with different interests, goals and means, over different time
scales. We pointed out above that a decision on actual infrastructure be-
tween the stations is often made by national-level actors not connected to
local arenas of interaction on station (area) redevelopment. The Dutch Key
Project policy is an interesting attempt to establish connections between a
national level plan for infrastructure improvement and local station (area)
redevelopment projects. Nevertheless, the new policy could not prevent the
intriguing mismatch that only one of the six Key Projects will have a full
HST connection. On top of that, serious differences in the expectations of
the quality of the stations and station area redevelopments remained be-
tween key actors. While the Ministry of V&W reserved money to finance
basic stations, the municipalities made plans for expensive stations and
public spaces for which the Key Project subsidy of VROM was often too
low.
A sound process architecture, as proposed by Berenschot, to overcome
these difficulties is important, but not sufficient. The crux of the matter
seems to be in an improved connection between the relevant domains of
society that are important for station (area) redevelopment. We have ana-
lysed the problems between different spheres of government here as they
are central to the innovations the Dutch Key Project policy aimed to real-
ise. But this is only one relevant domain; connections to other domains are
also crucial to the success of station projects. Without private sector inter-
est, projects have no chance to be carried out, and no private capital gains
can contribute to public goals.
The extremely slow pace of Dutch Key Projects, especially due to prob-
lems in the government domain, often frustrates private investors. The
same is true for the domain of civic interests. Although stations and their
direct environments are arguably of the most public places of cities, trans-
lating that notion in order to attract input from the broader circles of soci-
ety has not been convincing. We are reminded of the arguments of Mou-
laert et al. (2003) on the closed, de-politicised nature of most large-scale
undertakings. Only in Utrecht, after a political turnaround, did the issue

become more politicised with a referendum. But most decision-making on
station projects takes place away from public scrutiny. Although it is
probably too simplistic to redraw all the issues in station redevelopment
towards two-choice referenda, seeking better ways to incorporate the
voice, the wishes and the demands of station users, is a challenge that has
yet to be met.
Acknowledgement
The authors would like to thank Marieke van Loon and Johan Olsthoorn
for gathering valuable data on the projects.
References

view 71:455 475
Areas. Spon Press, London
CPB [Centraal Plan Bureau] (2003) Kengetallen Kosten-batenanalyse Project
Zuidas Amsterdam. CPB Document 44, Den Haag
Gemeente Amsterdam (1998) Masterplan Zuidas. Dienst Ruimtelijke Ordening,
Gemeente Amsterdam, Amsterdam
Gemeente Amsterdam (2003) Kiezen voor Stedelijkheid, Structuurplan. Amster-
dam
Gemeente Amsterdam (2004) Visie Zuidas 2004. Amsterdam
Gualini E, Majoor SJH (2007) Innovative Practices in Large Urban Development
Projects: Conflicting Frames in the Quest for New Urbanity. Planning Theory
and Practice, forthcoming
Klijn EH (2001) Ruimtelijk Beleid voor Bestaand Stedelijk Gebied: De Sleutel-
projecten als Illustratie. Beleidswetenschap 15:444 460.
Kooijman D, Wigmans G (2003) Managing the City: Flows and Places at Rotter-
dam Centraal Station. City 7:301 326
Kort M, Mark R van der, Verhey T, Wal M de (2005) Evaluatie Organisatie en
Werkwijze Nieuwe Sleutelprojecten: Een Terugblik om van te Leren. Beren-
schot, Utrecht
Majoor SJH (2004) Amsterdam Zuidas: Verkeersbundel en Toplocatie. In: Bruijn
H de,Teisman GR, Edelenbos J, Veeneman W (eds) Meervoudig Ruimtege-
bruik en het Management van Meerstemmige Processen. Uitgeverij Lemma,
Utrecht, pp 77 98
Majoor SJH (2005) Paradox van Grote Projecten. Rooilijn 38:57 63
Majoor SJH (2006) Conditions for Multiple Land Use in Large-Scale Urban Pro-
jects. Journal of Housing and Built Environment 21:15 32
Majoor SJH (2007) Amsterdam Zuidas: The Dream of a New Urbanity. In: Salet
WGM, Gualini E (eds) Framing Strategic Urban Projects. Routledge, London,
pp 53 83
Ministerie VROM (2006) Nieuwe Sleutelprojecten op Stoom. Voortgangsrappor-
tage maart 2006, The Hague
Moulaert F, Rodrguez A, Swyngedouw E (2003) The Globalized City: Economic
Restructuring and Social Polarization in European Cities. Oxford University
Press, Oxford
Ploeger R (2004) Regulating Urban Office Provision: A Study of the Ebb and
Flow of Regimes of Urbanisation in Amsterdam and Frankfurt am Main,
1945-2000. University of Amsterdam, Amsterdam
Salet WGM, Thornley A, Kreukels AJ (eds) (2003) Metropolitan Governance and
Strategic Spatial Planning. Spon Press, London
Salet WGM, Majoor SJH (eds) (2005) Amsterdam Zuidas European Space. 010
Publishers, Rotterdam
Salet WGM, Gualini E (eds) (2007) Framing Strategic Urban Projects. Routledge,
London
Schuiling D (1996) Key Projects for Urban Regeneration: The Dutch Experience.
Planning Practice and Research 11:279 290
Trip JJ (2007) What Makes a City? Planning for Quality of Place. Delft Univer-
sity Press, Delft
Verbart J (2004) Management van Ruimtelijke Kwaliteit: De Ontwikkeling en
Verankering van Inrichtingsconcepten in het Utrechtse Stationsgebied. Ebu-
ron, Delft
7 A multidisciplinary approach of railway station
development: A case study of s-Hertogen-
bosch
Gert-Joost Peek
Strategy & Marketing, ING Real Estate Development International,

The Hague, the Netherlands
Erik Louw

7.1 Introduction
Railway stations have a dual character, as they are nodes in transportation

networks as well as places in the city (see chapter 3). A combination of ac-
cessibility and potential as a concentrated area of urban activity therefore
makes station areas favoured locations for the development of structuring
nodes in the network city (Bertolini and Dijst 2000 p. 41). With reference
to Castells (1989), Bertolini (1996 p. 332) subsequently pinpoints the ten-
sions this combination embodies: On one hand, stations offer a (potential)
connection to several material and immaterial flows that create value in the
current informational mode of development. Stations are (or may be-
come) important nodes in both transport and non-transport (e.g. business,
consumption) networks. The connection to ever denser, faster and further
reaching transportation systems, as well as the development there of office
complexes and shopping centres are materialisations of this network di-
mension of station areas. On the other hand, stations identify a place, a
both permanently and temporarily inhabited area of the city, a dense and
diverse conglomeration of uses and forms accumulated through time, that
may or may not share in the life of the node. The mixture of housing, small
126 Gert-Joost Peek and Erik Louw
business premises and informal public spaces of the stations neighbour-

hood are an expression of this local dimension.
These tensions may result in synergy when node and place reinforce
each other. For instance [] a high level of accessibility may provide the
critical mass of demand for the development of particular activities. In
turn, a high density of activities may induce the necessary support for the
development of transportation networks (Bertolini and Spit 1998 p. 9).
But these tensions are also accountable for the long and complex process
of (re)development in which both transportation and urban development
issues simultaneously need to be handled (Bertolini and Spit 1998 p. 17).
A multifarious array of both node and place-based actors crowd station
area redevelopment processes, of which local government and the railway
company are characteristic. Depending on the local context, other actors
will also play a decisive role. These include different levels of the public
administration, different transportation companies, and actors in the real
estate market, such as developers, investors, and end users. In addition,
particularly at station areas set in dense, historically-stratified urban dis-
tricts, local residents and businesses may also have a significant stake in
the process.
The objectives of this heterogeneous array of actors are diverse and their
expectations are often very high. Depending on their disciplinary back-
ground in education, their professional experience and their particular
business or policy objectives, all participants in the redevelopment process
see their own opportunities and have distinct views about the ways in
which others can assist in achieving their goals. The dual character of the
station area and its resulting possible synergies raise the expectations of
the participants. This makes them interdependent, so that capitalising on
the synergies requires collaboration; but such a condition is difficult to
achieve and maintain when the actors involved invariably have different
viewpoints.
The literature on urban redevelopment, including station location, can
roughly be divided into two areas of emphasis. On one side we see a focus
on the literature of the redevelopment, which describes more and less suc-
cessful cases, and provides us with content-based models as with the node-
place model of Bertolini. On the other side is the literature that deals with
the process management issue, in which most attention is given to the in-
teractions of the actors involved, and little attention is given to aspects
concerning the content of the redevelopment (i.i.g., Bruijn et al. 2002).
We believe the dual character of the station location to be such a mas-
sive force driving force for these redevelopment processes that it deserves
much more attention, especially from a process management perspective.
Actors should be seen in this context as individual representatives of the
A multidisciplinary approach of railway station development 127
public and members of private organisations who have to work together to

see a project reach fruition. Most are professionals with sound back-
grounds in one or several disciplines and, despite differences in expertise
or policy backgrounds, they all want to be part of a redevelopment project
that meets their professional standards.
In this chapter we analyse and explain the process of redevelopment us-
ing a content-based model, with the assumption that content-based driving
forces within the process are as important or influential as process-based
driving forces. We explore the explanatory capacity of a conceptual
framework to develop, in Bertolinis words, a balanced station area. Our
framework arises from an interventionist perspective, and is based on the
presupposition that synergies at station locations stem from the combina-
tion of four main approaches towards the redevelopment of a railway sta-
tion area (Peek 2006). The four approaches provided in section 7.2 refer to
the actors professional perspectives. We investigate the chances for syn-
ergies between approaches in section 7.3 and identify possible antago-
nisms among them. In section 7.4 we turn to the redevelopment process of
the station area in the Dutch city of s-Hertogenbosch, which is widely re-
garded as a success (Nijkamp et al. 2002; Hobma 2004); we see in this
case whether we can observe the process management decisions taken dur-
ing the redevelopment in terms of the four approaches. In so doing, we
hope to gain insights into the way actors in the process have utilised the re-
inforcing interrelations of the approaches and avoided possible threats.
Section 7.5 concludes.
7.2 The four approaches
Because stations are viewed as nodes and places, the approaches are cate-
gorised by these primary focusses. The focus of the station as a node is
found within the transportation discipline; whereas the station as place cor-
responds with urban planning and real estate discipline. A division can
then be made in accordance with a secondary focus. As station areas are
nodes within the network city, they can be viewed as separate locations,
but also as part of all connecting networks. With these primary and secon-
dary focusses in mind, four approaches are distinguished in Table 7.1.
Table 7.1. Four approaches to railway station development

Primary focus\Secondary focus Location Network
Node Connector Transportation node
Place Meeting place Urban centre
They are not theories as such, but rather are logical combinations of ar-
guments associated with similar theoretical or disciplinary backgrounds on
railway station redevelopment; the approaches may be interpreted as a type
of discourse (Hajer 2000 pp. 17-20) on the redevelopment of railway sta-
tions and their subsequent local effects.
The four approaches or discourses represent ideal types of station areas
in which certain features, valued within a specific discourse, are empha-
sised while others, which are regarded to be of less importance, are ig-
nored. Therefore, these ideal types are constructs and are not found in real-
ity (Zijderveld 1988 p. 45). Our approach has some similarities with the
node-place model of Bertolini described in chapter 3, but with one impor-
tant difference. Bertolini considers node and place as dimensions upon
which a station can be developed. Our approaches are conceptualised cate-
gories which represent ideal types less likely to appear in reality, but which
nevertheless represent coherent ways to think about station development.
The first approach typified as Connector combines the focusses of pri-
marily, the node and secondarily, the location. We find this approach with
engineers and architects who, when confronted with the design brief of a
station, view the station and surrounding area as a connector (Spek 2003);
that is, a build environment connecting the present transportation modes.
The aim is to make the connections as smooth as possible. Transferring
should be safe, reliable, fast, hassle-free, comfortable, and pleasant, thus
corresponding with travellers demands (Peek and Hagen 2002). Travellers
perceive transferring as up to three times more time consuming than it is in
reality (Waard 1989; Wardman 2004). The design of the connector should
lower this transfer resistance; the station area would be spatially defined by
all possible transfer routes connecting transport modes. The area could be
whimsically shaped, depending on the situation of the stops, and would ex-
tend to multiple levels to include indoor and outdoor space as well as pub-
lic and private space. Transferring consists of two main activities: moving
between stops and staying within the connector, which includes collecting
information and a ticket, shopping and waiting. Each transfer route would
feature physical, spatial and social aspects (Spek 2003 pp. 11-12).
The second approach typified as Transportation node combines the fo-
cusses of primarily, the node and secondarily, the network. We observe
this approach with transportation and traffic planners who view the station
as a node characterised by its hierarchical position within transportation
networks to which it is linked. They aim at optimising overall efficiency
and effectiveness of the networks by stimulating the use of transportation
modes in off-peak hours in the opposite direction of the mainstream.
Moreover, they would stimulate an even distribution of embarking and
disembarking travellers across all stops of a certain service. The position
of a station in this view is determined by number and type of services

stopping at the location which is subsequently determined by each sta-
tions individual contribution to overall network performance.
In this perspective the stations area of influence is spatially defined by
its accessibility. For every transportation network the distance or number
of stops in reach of a particular station within a certain period of time can
be determined. Considering rail networks the most important infrastruc-
ture-related variables are the number of directions and the spacing of sta-
tions along the lines. These, in combination with the service-related vari-
ables number of services and spacing of stops account for the travel
time. The spatial definition of the station area can be refined by taking ac-
count of frequencies and transfer options at other stations (Nes 2002 p.
49). The actual dimensions of the area largely depend on the period of time
chosen, which could result in a substantial area of influence.
The third approach typified as Meeting place combines the focusses of
primarily, the place and secondarily, the location. We find this approach
with sociologists who contemplate the role of the station area within the
life of city dwellers and view the station and its surroundings as heteroto-
pia (Hajer, 1996): a modern market place where people are still con-
fronted with urban life in all of its multiplicity. The combination of acces-
sibility and public space offers opportunities for planned and spontaneous
encounters of individuals. The contrasts in use of a station area are well-
depicted by the heterogeneity of the areas populations. The social com-
plexity of the metropolis is here in full display and includes a diverse array
of permanent and temporary residents, commuters, business people, and
area visitors. The absence of a dominant social-cultural group creates an
environment where many different individuals interrelate (Bertolini 2004;
Boelens et al. 1999 p. 28).
Whether people are open to meeting others in a positive sense depends
on their perception and cognition of their environment. Pocock (1973)
models this process in three phases. First, environmental stimuli are re-
ceived. The reception depends on physical characteristics of the environ-
ment for instance rain or sunshine and of the individual perspective
for instance running or sitting down. Second, processing the stimuli de-
pends on ones personal physical and mental state and social-cultural
schemata.
Third, this process leads to a certain perception and cognition of the en-
vironment upon which decisions are made regarding behaviour. Therefore,
creating meeting space in and around a station requires flexibility and pro-
grammatic complexity in order to allow for the dynamics of its use (Nio
2000 p. 124). In this approach it is not so much the presence of spaces
dedicated to or suitable for meetings that determines the location as meet-
ing place, but rather it is the use of space for individual encounters and col-
lective events that defines the station as a place that offers the experience
of urbanity (Boomkens 1999 pp. 70-71). Therefore, the approach of meet-
ing place should be interpreted in its metaphorical sense.
The fourth approach typified as Urban centre combines the focusses of
primarily the node, and secondarily, the network. We find this approach
with urban economists and city planners in search of locations to stimulate
the urban economy. The theory driving the redevelopment of station areas
as an economic development tool lies in its transit ability to provide higher
densities and create activity centres. From its position within the urban
network, station areas are focal points for urban regeneration: they offer
space for an array of urban functions, including offices, shops, dwellings
and all sorts of urban amenities. Furthermore, an attractive station area can
allow for more housing, business, and recreation activity; it also encour-
ages such dense development by providing a base of commuters as well as
a common destination. A transit centre essentially provides a scarce source
of land that accommodates dense, mixed-use development. Through this
kind of land use, economies of scale and diversity may be achieved, thus
leading to greater differentiation of urban activities and functions that, with
institutional support frameworks, leads to a more robust urban constella-
tion (Rodenburg et al. 2003). The diversity at the scale of a railway station
area greatly resembles the urban environment advocated by Jacobs (1961)
with its cultural variety and social interaction, and better known as Jacobs
externalities of synergy effects.
The American concept of transit-oriented development (TOD) is a good
example of the urban centre approach. TOD concentrates development in
nodes associated with transit stations, thus creating an urban regional
structure with clusters of uses arranged according to a density gradient
from transit stations. It stresses the need for a functional integration of
transit and the surrounding uses, which should be intense, mixed, compre-
hensive, and thus supportive of sustainable modes of transport and to eco-
nomic development (Belzer and Autler 2002). Although TOD is mainly
applied to Greenfield developments, the concept is applicable in an urban
context.
In Europe the Urban centre approach is often connected with the devel-
opment of the High-Speed Train (HST) network. When considering the ef-
fects of connecting cities to the HST network, Pol (2002) perceives three
investment categories next to the upgrading of the station itself. The first is
investments in space for new urban activities; this should result in a bal-
anced mix of functions and add value to the entire city by processes of dis-
tribution and generation. The second is investment in spatial quality,
meaning the upgrade of the station area and its integration in the surround-
ing urban fabric. Third is investment in the image of the urban region as a
whole and especially to improve the status of the urban area.
7.3 Possible synergies and antagonisms
The tensions between the station area as node and place and, simultane-
ously, between the perspective of a stand-alone location and of a location
within urban networks, results in possible synergies, but may also lead to
negative effects or antagonisms, as Bertolini and Spit indicated: [..] dense
patterns of use can make a locations transport infrastructure difficult to
expand and adapt. In the same vein, optimisation of a stations accessibil-
ity by all modes may negatively affect its liveability, and thus its attrac-
tiveness (Bertolini and Spit 1998 p. 9). Nevertheless, these complex node-
place and location-network interactions are the core of railway station re-
development. We explore these interactions through a combination of our
four ideal typical approaches. First, we consider the possible positive in-
teractions which might lead to synergy. Achieving synergies requires a
balanced attention to all approaches. Second, we address antagonistic ef-
fects that might occur when one approach dominates the others. This re-
sults in optimising only one approach without achieving potential synergy
among them.
Table 7.2 shows the possible synergies among the four approaches to
railway station development. The ideal types of connector and of transpor-
tation node reinforce each other with respect to transfer quality. While the
connector aims at lowering the transfer resistance, the transportation node
aims for an even distribution of travellers leading to an efficient and effec-
tive exploitation of transportation networks. Both improve the transfer
quality. By looking at the station, an even distribution of travellers helps to
match transfer facilities to travellers needs by reducing peak and off-peak
differences. At the network level a smooth transfer helps connect the
transportation networks of different scale levels and contributes to the per-
formance of the whole system.
Table 7.2. Possible synergies of approaches to railway station development

Focus Location Synergy Network
Node Connector Transfer quality Transportation node
Synergy Intensity Centrality
Place Meeting place Spatial quality Urban centre
The ideal types meeting place and urban centre have a mutually positive
effect on spatial quality. The concentration of urban activities leads to a
vibrant environment around the station and heightens the chance of sudden
encounters and urban events, while additional meeting places strengthen
the station areas position as a centre point in the urban network and cre-
ates a corresponding urban feeling. The ideal types connector and meeting
place are complementary concerning intensity. The intensity of use is,
among others, influenced by the smoothness of the transfer as well as a
pleasant atmosphere and dedicated meeting places. This presence might
lead to more travellers in off-peak hours helping to maintain a certain pres-
ence of people and so preserving a safe environment, especially at night. A
smooth transfer makes the station a preferred place to meet.
The ideal types transportation node and urban centre both influence the
centrality of the station area. The more connections a station has and the
faster these connections are, the more central the transportation system be-
comes. This centrality in turn would be capitalised by property developers,
the concentration of real estate subsequently would lead to a central loca-
tion within the urban network and create further travel demand. Private
railway companies in Japan combine these approaches by developing resi-
dential areas near new rail lines and thus are assured of an increase in
property value in the short-term and a steady stream of travellers in the
long-term (Mizutani 1999 p. 300).
A balanced station area, in Bertolinis words, should combine all four
ideal types (Uum and Boer 2003 p. 20) creating a quality of space that
matches the criteria set by Florida (2002 p. 215) in order to contribute to a
Creative City. Our standpoint is that the surroundings of the main inner
city stations provide us with the potential, both in terms of node-place and
location-network, to become balanced station areas. At the same time,
tensions between node and place could lead to a number of negative ef-
fects, and rather than transform the station area into a place of buzz
(Storper and Venables 2002) could instead predispose such a station area
as an urban manhole, where all perspectives of synergy have vanished.
Table 7.3. illustrates the possible antagonisms among the four ap-
proaches to railway station development. The ideal types connector and
transportation node could have a mutually-negative effect through a cou-
pling of linear and nodal infrastructures. The ways in which these infra-
structures are linked have great influence on the transfer quality. Hiddink
and Van Twist (2003) have provided us with an overview of possible
managerial arrangements of nodes. They first observe an uncoupled
model, in which the management of node and line infrastructure are inde-
pendent, as in air- or seaports. Second, they perceive a node-service
model, in which the transportation service provider manages the node.
This model is most common for the management of railway stations.

Third, is an integrated model, in which node and line infrastructure are
owned and managed by the same company. We find this model in metro
systems and private Japanese railway companies. The models differ in
their synergy-creation potential from the coupling of node and line infra-
structure. However, in most cases the choice of model is not motivated so
much by possible synergies but rather is affected most by national and in-
ternational policies with regard to privatisation and open markets.
Table 7.3. Possible antagonisms of approaches to railway station development

Focus Location Antagonism Network
Node Connector Coupling of linear and Transportation node
nodal infrastructures
Antagonism Accessibility & Embeddedness & Positioning & envi-
identity commerce ronmental impact
Place Meeting place Function mix Urban centre
The ideal types meeting place and urban centre could lead to synergy
threatening interrelations with regard to function mix. The pursuit of multi-
functionality in order to create a livable, attractive and secure environment
for meeting and public transport and long-term property exploitation pros-
pects could be strained if in a coincident relation with short-term profit-
ability. Offices and retail are, from a development perspective, the most
desired functions in and around stations. When a redevelopment pro-
gramme largely depends on property revenue it will be difficult to realise
multifunctional ambitions, certainly because non-residential functions also
require shielding the tracks. A large share of offices subsequently makes
the execution of the total programme largely dependent on thriving real es-
tate markets (Bertolini and Spit 1998 p. 215; Koppenjan and van Ham
2002 p. 447).
The ideal types connector and meeting place could conflict in the areas
of accessibility and identity. An exclusive orientation on the connector
could lead merely to stations designed as grey but efficient transfer ma-
chines. These non-places, as Aug (1995) calls them, have no identity
and almost entirely lack social and historical significance. They are pro-
grammed exclusively for the friction-free and disciplined transport of large
groups of people and are everything but meeting places. But an exclusive
prioritising of the meeting place could, however, lead to a station where
marginal groups of society would assemble.
Their presence could discourage travellers, especially the elderly,
women and single travellers, from using the station. How to combine the
character of the station as a place to encounter the fringe of society and yet
be a safe and secure transfer environment is an opportune question cur-

rently under discussion with regard to electronic ticketing, controlled ac-
cess, and local branding of stations (Peek 2001 and 2002; Pronk 2005).
Confrontation of the ideal types transportation node and urban centre
could lead to antagonistic effects regarding positioning and environmental
impacts. Seen from the perspective of the transportation node, stations
could best be positioned such that the ideal layout of the infrastructure
network is least affected. Stations are autonomous constructs situated out-
side the city and surrounded by open space for future infrastructure devel-
opment and parking. Some French TGV stations resemble this approach
(Valance TGV: Tiry 2001).
Such positioning has little to offer with respect to the station as an urban
centre. Conversely, the exclusive positioning of a station as an urban cen-
tre within the heart of the city and surrounded by buildings crammed with
multi-functional programmes, leads to a suboptimal infrastructure layout.
Consider for instance all termini in cities such as London and Paris. The
scope for future infrastructure expansion is also severely limited. Similar
conflicts arise when considering the environmental impact of the rail infra-
structure. While heart-to-heart connections between cities by rail compete
best with transport by car, environmental legislation limits the possibilities
of concentrating multi-functional programmes next to the tracks (Uum and
Boer 2003 pp. 13-14; Schutte-Postma 2004 p. 300).
The last series of antagonistic effects could arise from the confrontation
of the ideal types connector and urban centre and affect all approaches.
These deal with the embeddedness of the station in its surroundings and
commerce. From the perspective of the connector, a rather autonomous
and closed structure could very well guarantee the transfer quality. Graham
and Marvin (2001) discuss autonomous isles in the city and signal a des-
perate need, in particular, to imagine ways of weaving secessionary and
global network spaces into the finer-grained fabric of the urban spaces []
that surround them (Graham and Marvin 2001 p. 414). On the other hand,
such a closed structure leads from an interweaving of local pedestrian and
bike routes with the internal logistics (as observed in many Dutch railway
stations) to discussions about levels of restricted access and exact location
of barriers.
The embeddedness of stations in the city is also threatened by the space
around the station occupied by all access and egress modes. In line with
the goals of the connector bus station, parking and bicycle sheds are con-
veniently located nearby the stations entrance, forming a large field of as-
phalt full of buses and cars and hampering pedestrian connections to the
surrounding urban fabric. Potential solutions are dynamic bus stations and
underground parking.
Pedestrian flows attract commercial activities which could expand into

attractions by themselves in line with the ideal type of the urban centre.
The dependence of retail outlets on passenger flows nevertheless risks that
shoppers will disturb travellers who are intent on simply catching their
trains. A clear spatial division is advised in order to avoid these conflicts.
7.4 The station area in s-Hertogenbosch
Our case study station area is situated in the municipality s-

Hertogenbosch (132 000 inhabitants) a regional city in the southern half of
the Netherlands. The station is a major node in the Dutch railway network
because several intercity train services stop there. In the 1990s a new sta-
tion was built and the former adjoining industrial estate De Wolfsdonken
was transformed into a mixed use area called Paleiskwartier. This former
industrial estate is situated on the west side of the railway station, while
the city centre is on the east side. In the Netherlands this redevelopment is
seen as highly successful, both as an example of public-private partnership
and urban design (Nijkamp et al. 2002; Hobma 2004).
The industrial estate was developed in the 1950s. During the 1980s
signs of physical deterioration began to appear. Three driving forces even-
tually set the redevelopment in motion, the first being real estate agents
interest in redeveloping parts of De Wolfsdonken between 1985 and 1990.
In particular they wanted to develop commercial property near the station,
and made several proposals to the municipality. Although these initiatives
failed, they were encouraging signs of an emerging market for offices near
the station, and alerted the municipality to reconsider its policy to update
the estate. The second driving force was the municipal spatial policy for
strengthening the economy of the cramped city centre and redeveloping
the entire 200-hectare site west of the station; the resultant policy docu-
ment drafted in 1987 had the following goals:
Redevelopment of the station area (east and west of the station it-
self) from low to high quality land uses;
Provision of space for large offices that have to leave the city cen-
tre;
Improvement of accessibility to the city centre and the area itself.
The third driving force was the need to accommodate additional trains
passing through the station as part of a new ambitious timetable for the 21st
century. A new platform had to be built and the track layout altered, but
initially Dutch Railways did not plan to build a new station.
The redevelopment initiatives focussed in 1988 on both sides of the sta-

tion, and an agreement was signed between Dutch Railways and the mu-
nicipality to redevelop it. The municipality agreed to pay for a bridge over
the railway tracks to connect Station Square at the east side of the station
and the proposed redevelopment at the west side of the station (La Gare
area). Actual construction of the new station did not commence, however,
until 1995. In the meantime, redevelopment of De Wolfsdonken had accel-
erated; the first office building and an apartment block were built in La
Gare in 1991, and a regional court of justice was built soon after. Together
with increasing demand in the office market and the new emphasis of na-
tional spatial policy to use land in city centres intensively, this resulted in
the whole area of De Wolfsdonken becoming a redevelopment area from
1992.
The municipality published its first planning document in 1994 for the
whole area west of the station. In 1996 the urban design Masterplan for
Paleiskwartier was published with the goal of creating an area with a high
level of urban design, and a feasibility study was also carried out. Al-
though a Masterplan is not a statutory planning document, in practice it is
used as a kind of preliminary or draft land use plan. In 1998, the public-
private company Paleiskwartier BV was established, the participants being
the municipality, a developer and two investors. This company operated as
a land and property developer for the remaining two-thirds of the former
De Wolfsdonken (Paleiskwartier 2nd phase). In total, the station area en-
compasses 26 hectares and a property programme of 210 000 sq metres of-
fice space, 46 000 sq metres commercial space and 1500 apartments. By
2006 70% of this programme was completed.
From the above we observe that the prime concern of Dutch Railways
(DR) was to focus on the station as a transport node. Initially, they wanted
only to alter the track layout. Only in 1995 when DR split into various
companies did the unit concerned with the stations become interested in
the development of various station services (shops and cafs, etc.); this
situation provided a solid link to the approach of connector, thus providing
for a smooth and pleasant interchange. Spatially this interest was limited to
the station building itself.
The prime focus of the municipality was on development of the site in
the context of the urban centre approach. Its initiatives to redevelop the
Station Square and De Wolfsdonken were meant to improve competitive-
ness of the city centre and improve the citys economic performance.
However, this redevelopment was partly in response to initiatives from ac-
tors in the commercial real estate sector. Had the municipality agreed with
their initiatives, the outcome of the development would have been differ-
ent.
The redevelopment of the station area and Paleiskwartier is analysed

thoroughly by Bruil et al. (2004). She distinguishes decisive three events
for the successful development of the whole area; these events were recog-
nised after analysing the public decision processes, with the Kingdon
model (1995) of policy formation in mind. Kingdons model highlights
three processes or streams: i.) a problem stream in which problems are de-
fined and rise to a sufficient level of urgency that they find a place on poli-
cymakers agenda; ii.) a policy stream in which alternative policy solutions
are developed and selected through consensus within the political commu-
nity; and iii.) a political stream in which consensus on political issues is
developed in the broader political environment through coalition building.
The streams are influenced by their own institutional structures, but they
also interact.
When applied to our case study, the policy stream is the municipalitys
policy to strengthen the city centre. A policy stream is not sufficient by it-
self to implement policy; what is needed is an interaction between streams
in which policy entrepreneurs couple the three streams to produce desired
policy outcomes. Such an interaction is called a window of opportunity and
is created by a change in the problem or political stream. In the remainder
of this section we focus on how three policy windows or events, which oc-
curred in the s-Hertogenbosch case study and found by Bruil (2004), fit
into our four approaches.
The first policy window was created around the decision by the munici-
pality to subsidise the new railway station. As indicated above, DR had to
adapt the track layout along the station due to plans for a new timetable.
The municipality, however, now sees an opportunity to redevelop the sta-
tion and the Station Square. To achieve this, the municipality proposes to
invest 2.2 million into the station under three conditions:
DR has to build a new station building;
The bridge connecting the platforms has to cross all the railway
tracks to the area on the west side of the tracks;
The bridge has to be 15 metres wide and the old platform roof
construction will be preserved.
The board of DR accepts the offer and conditions and construction of
the bridge begins in 1995.
The municipal decision to invest (or subsidise) in the railway station
seems straightforward, but is in fact embedded in the urban centre ap-
proach of the municipality, while at the same time, a policy window which
had originated in the transportation node approaches of DR is used. Dutch
Railways were only interested in increasing the efficiency of the station as
a node in their intercity network; and to achieve this, only the platform and
track layout had to be altered.
The municipality, however, was interested in a new station building, a
new Station Square and the removal of the railway as a physical barrier be-
tween city centre and west side of tracks, which included the area De
Wolfsdonken. A direct link between De Wolfsdonken and the city centre
via a new railway station would remove the barrier and enable the munici-
pality to advance its urban centre approach. In Kingdons terms the trans-
portation node approaches of DR opened a policy window for the munici-
pality to implement its urban centre approach and redevelop De
Wolfsdonken as an extension of the city centre. By optimising the scope of
the station redevelopment for the DR as seen from the perspective of the
connector, the municipality was willing to invest in the station.
The second window of opportunity was the municipalitys decision to
not grant a building permit to a developer who wanted to build the first of-
fice building at the west side of the station in 1987. The alderman for spa-
tial planning was pleased with the plan of the developer, because it pro-
vided an incentive for the economic redevelopment of the area. However,
the highest civil servant for spatial development disagreed because it
would endanger the as yet undecided plans of the municipality: an invest-
ment now would probably increase land values, which would endanger
these plans. The proposed office building could also interfere with the
plans for the bridge over the railway tracks, which were under discussion
at the same time. The civil servant convinced the alderman not to adapt the
land use plan that was necessary to grant the building permit.
Although it seems that the refusal to grant the building permit closed a
window of opportunity, in fact it kept an emerging and even larger window
open. Granting the building permit seems logical and would create imme-
diate success for the urban centre approach in the form of a short-term
profitable real estate project. However, the changes necessary to create
sustainable synergies between the various approaches would have been
lost because it reduced the chances to level out the railway as a physical
barrier. In the long-run this would prevent centrality synergy between the
transportation and urban centre approaches in the form of capitalisation by
property development. Also synergies between meeting place and urban
centre were favoured, because it kept open opportunities for spatial quality
regarding mixed use development. Not granting the building permit there-
fore prohibited antagonism between meeting place and urban centre from
occurring.
This second window of opportunity also influenced the political stream.
After the municipality refused the building permit, the developer aban-
doned its initiatives for property development of the site. The municipality
informed another developer about this, who then bought the land; this de-
veloper was well-known to the municipality and became the developer of
three buildings in La Gare and a shareholder in the public-private company
Paleiskwartier. In a sense the refusal of the building permit paved the way
to public-private coalition building in the property development of the en-
tire site.
The third window is a classical example of the true sense of Kingdons
model: the coupling of streams leading to the decision to move the various
court offices in the city centre to one building in La Gare. The 40 000 sq
metre building is now the flagship project of the site. Due to reorganisation
and expansion, the courts needed more office space; staying in the city
centre was not possible and a new office in La Gare was not an attractive
option, because in the early 1990s La Gare was an unappealing building
site with a rundown industrial estate next door. However, the site was an
office location that suited the national governments policy on building
new offices near railway stations.
This policy, known as ABC-location policy, is an example of coordi-
nated transport and land use policies and aimed at settling employment
with high passenger transport profiles at public transport nodes (A-
locations) and employment with high freight profiles at sites highly acces-
sible for road transport (C-locations). The municipality used this national
policy, founded on the rationale behind the transportation node and urban
centre approach, as an argument to persuade the Ministry of Housing, Spa-
tial Planning and the Environment to build the new court building in La
Gare.
In this case the municipality used a national policy stream to its advan-
tage to handle the problem of relocating the court offices. However, the
large amount of floor space demand created its own problems. The exist-
ing urban design plan had to be altered because land assembly for the
building could not be completed on time. Therefore, a square that was
planned as urban entrance space for the Westside access of the station had
to be relocated. In its turn this diminished the possibilities for mixed-use
development around this square and the subsequent reduction in the func-
tional mix and urban quality. This antagonism effect weighed against the
securing of a flagship project on the site and gave way to a re-branding of
the area. The location there of the Palace of Justice put the area on the map
as the premium location in the local office market and even resulted in re-
naming the redevelopment area Paleiskwartier: Palace quarter.
For the redevelopment of La Gare, the designation of the area as an A-
location had yet another side effect. The ABC-location policy also contains
a regulation about the number of parking places allowed in an area. Ac-
cording to this policy, only 10 places for 100 office employees were al-
lowed in the s-Hertogenbosch station area. Although the municipality has

always mitigated this parking standard, some firms decided for this reason
to not establish their office in La Gare. So in this sense the policy to inte-
grate transport and spatial planning had a temporary antagonism effect on
the development of La Gare.
7.5 Conclusions
The redevelopment of station areas is a complex and time-consuming

process with many actors. Many studies on this subject are content-based
and give scant attention to the development process. In this chapter we
have tried to combine the content-based and the process-management as-
pects into a multidisciplinary approach. The four approaches we have dis-
cerned represent coherent ways of thinking about station development and
follow theoretical or disciplinary views on station development taken by
particular types of actors. To bring together these different actors, with
their diverse views and backgrounds, into the planning and development
process is one of the main challenges in station area redevelopment when
the ambition is to create synergistic effects.
We have applied our multidisciplinary approach to the redevelopment of
the station area in the Dutch town of s-Hertogenbosch. It came as no sur-
prise that Dutch Railways was clearly linked with the transportation node
approach, while the municipality was linked to the urban centre approach.
The real estate branch of DR only played an instrumental role in this last
approach. The success of the s-Hertogenbosch station area redevelopment
is mainly due to the fact that these to approaches collided at a particular
time and place where several policy windows could be opened. We also
noted that the entrepreneurial attitude of the municipality could prevent
some antagonistic effects.
Our case study also revealed that the beneficial effect of the transporta-
tion node approach was only short-lived. Its main influence was made
manifest during the initial phase of the redevelopment in the form of in-
vestment in the station and its immediate surroundings at La Gare.
This provided a threshold level to make the development of the entire
site viable to self sustain. The current development of Paleiskwartier 2nd
phase continues more or less independently from the station area itself. Its
success is, however, irrevocably linked to the policy decisions made in the
footsteps of the transportation node and urban centre approaches. In gen-
eral, it can be assumed that a combination of these approaches is essential
for a viable station area development, because this puts the station area
into a broader spatial context and scale than the connector and meeting
place approaches, which are location-specific.
To a certain extent our approach is a supplement to and a refinement of
Bertolinis node-place model. In our approach we extract the spatial di-
mension location vs. network from the original two functional dimen-
sions: node vs. place. By so doing we are able to give a broader perspec-
tive on the development potential in combination with the conditions that
determine if the potential will be realised. The potential and conditions for
the redevelopment of station areas are incorporated in the interaction of the
four approaches. We add to this a process perspective in order to analyse
and explain the interaction and thereby assess whether the potential is
eventually realised.
We interpret from our analysis that the potential for creating synergies
between the four approaches is, in our view, only possible when a multid-
isciplinary approach is applied. This implies a decision process in which
various and very different actors have to collaborate. When they do this ef-
fectively the four approaches are combined such that a maximum on syn-
ergies and a minimum of antagonism is realised, and a balanced situation
in Bertolinis words can arise. What such a balanced situation for a par-
ticular station is, however, depends on the characteristics (potential and
conditions) of the station area itself, and on the interaction among the ac-
tors involved. The multidisciplinary approach can help us understand why
in some redevelopment processes potential is realised and in others it is
not. Moreover, based on the approach interventions in redevelopment
processes, potential can be designed in order to better exploit these fa-
voured locations.
References
Aug M (1995) Non-Places: Introduction to an Anthropology of Supermodernity.

Verso, London/New York
Belzer D, Autler G (2002) Transit-Oriented Development: Moving from Rhetoric
to Reality. The Brookings Institution, Washington DC
opment. European Planning Studies 4:331 345
Bertolini L (2004) Fostering Urbanity in a Mobile Society: An Exploration of Is-
sues and Concepts. Paper presented at AESOP Conference: Grenoble
Bertolini L, Dijst M (2000) Mobiliteitsmilieus: Ankers voor het Vluchtende Stede-
lijke Leven. In: Boelens L (ed) Nederland Netwerkenland: Een Inventarisatie
van de Nieuwe Condities van Planologie en Stedebouw. NAi Uitgevers, Rot-
terdam, pp 35 45

Areas. E & FN Spon, London
Boelens L, Sanders W, Stroeken F (1999) Infrastructuur als Katalysator: Op Weg
naar een Nieuwe Benadering van Spoor. Holland Railconsult, Utrecht
Boomkens R (1999) Van de Grote Stad Ging een Onbestemde Dreiging Uit: Hoe
Grootstedelijk is Nederland? In: Wouden R van der (ed) De Stad op Straat.
Sociaal en Cultureel Planbureau, Den Haag, pp 63 80
Bruil AW (2004) Sturing en Procesmanagement. In: Bruil AW, Hobma FAM,
Peek GJ, Wigmans G (eds) Integrale Gebiedsontwikkeling: Het Stationsge-
bied s-Hertogenbosch. Uitgeverij SUN, Amsterdam, pp 259 79
Bruil I, Hobma F, Peek GJ, Wigmans G (eds) (2004) Integrale Gebiedsontwikke-
ling: Het Stationsgebied s-Hertogenbosch. Uitgeverij SUN, Amsterdam
Bruijn JA de, Heuvelhof EF ten, Veld RJ in t (2002) Procesmanagement: Over
Procesmanagement en Besluitvorming. Academic Service, Schoonhoven
Castells M (1989) The Informational City: Information Technology, Economic
Restructuring and the Urban-Regional Process. Blackwell, Oxford
Florida R (2002) The Rise of the Creative Class: How it is Transforming Work,
Leisure, Community and Everyday Life. Basic Books, New York
Technological Mobilities and the Urban Condition. Routledge, London/New
York
Hajer M (1996) Heterotopia Nederland: Of wat Bunnik mist. Stedebouw & Ruim-
telijke Ordening 77/6:8 9
Hajer M (2000) Politiek als Vormgeving. Vossius AUP, Amsterdam
Hobma FAM (2004) Het Succes van s-Hertogenbosch. In: Bruil AW, Hobma
FAM, Peek GJ, Wigmans G (eds) Integrale Gebiedsontwikkeling: Het Stati-
onsgebied s-Hertogenbosch. Uitgeverij SUN, Amsterdam, pp 430 436
Hiddink CJW, Twist MJW van (2003) Niet de Grote Lijnen maar de Vele Kno-
pen. Economisch Statistische Berichten 88/4392:15 17
Jacobs J (1961) The Death and Life of Great American Cities. Random House,
New York
Kingdon JW (1995) Agendas, Alternatives and Public Policies. Harper Collins
College Publishers, New York
Koppenjan JFM, Ham JC van (2002) Patronen in de Totstandkoming van Publiek-
Private Samenwerking bij Transportinfrastructuur: Een Vergelijkende Analy-
se. In: Ham JC van, Koppenjan JFM (eds) Publiek-Private Samenwerking bij
Transportinfrastructuur: Wenkend of Wijkend Perspectief? Uitgeverij
LEMMA BV, Utrecht, pp 415 456
Mizutani F (1999) Japan. In: Velde DM van der (ed) Changing Trains: Railway
Reform and the Role of Competition; The Experience of Six Countries. Ash-
gate, Aldershot/Burlington, pp 255 306
Nes R van, (2002) Design of Multimodal Transport Networks: A Hierarchical Ap-
proach. DUP Science, Delft
Nijkamp P, Burch M van der, Vindigni G (2002) A Comparative Institutional
Evaluation of Public-Private Partnerships in Dutch Urban Land Use and Revi-
talisation Projects. Urban Studies 39:1865 1880
Nio I (2000) Paradoxen van de Netwerkstad. OASE Tijdschrift voor Architectuur

Vol. 53 April, pp 113 124
Peek GJ (2001) Verknopen: Knooppuntontwikkeling als Kernactiviteit voor NS.
Afdeling Bouwmanagement & Vastgoedbeheer, Technische Universiteit Delft
(not pubished), Delft
Peek GJ (2002) Locatiesynergie op Stationslocaties. Spitscollegereeks Ontwikke-
len van stationslocaties., ARKO Conferences, Rotterdam
Peek GJ (2006) Locatiesynergie: Een Participatieve Start van de Herontwikkeling
van Binnenstedelijke Stationslocaties. Eburon, Delft
Peek GJ, Hagen M van (2002) Creating Synergy in and Around Stations: Three
Strategies in and Around Stations. Transportation Research Record, Journal of
the Transportation Research Board 1793:1 6
Pocock DCD (1973) Environmental Perception: Process and Product. Tijdschrift
voor Economische en Sociale Geografie 69:251257
Speed Train Stations. Delft University Press, Delft
Pronk C (2005) Verslag Werkgroep Marktontwikkeling Project Stadsgeweste-
lijk vervoer. 20 September, Bureau Spoorbouwmeester/Railforum Nederland,
Utrecht/Hoofddorp
Rodenburg CA, Vreeker R, Nijkamp P (2003) Multifunctional Land Use: An Eco-
nomic Perspective. In: Nijkamp P, Rodenburg CA, Vreeker R (eds) The Eco-
nomics of Multifunctional Land Use. Shaker Publishers, Maastricht, pp 3 15
Schutte-Postma ET (2004) Milieurecht: Randvoorwaarden en Risicos. In: Bruil
AW, Hobma FAM, Peek GJ, Wigmans G (eds) Integrale Gebiedsontwikke-
ling: Het Stationsgebied s-Hertogenbosch. Uitgeverij SUN, Amsterdam, pp
294 315
Spek SC van der (2003) Connectors: The Way beyond Transferring. Delft Univer-
sity Press, Delft
Storper M, Venables AJ (2002) Buzz: The Economic Force of the City. DRUID
Summer Conference on Industrial Dynamics of the New and Old Economy:
Who is Embracing Whom?, Copenhagen
Tiry C (2001) Between City and Countryside: TGV Mditerrane Stations Paying
Tribute to Nature. Japan Railway & Transport Review 28 (September):29 31
Uum E, Boer J van der (2003) Knopen en Centra in Soorten en Maten. Ministerie
VROM, Den Haag
Waard J van der (1989) Onderzoek Weging Tijdelementen (deelrapport 5). Facul-
teit Civiele Techniek, Technische Universiteit Delft, Delft
Wardman M (2004) Public Transport Values of Time. Transport Policy 11:363
377
Zijderveld AC (1988) Sociologie als Cultuurwetenschap: Een Beknopte Methodo-
logie van de Cultuursociologie. Uitgeverij Lemma BV, Culemborg
Part B: Evaluation studies
8 Ex ante evaluation of railway station
development projects: Issues still to be solved
Piet Rietveld

the Netherlands
Bert van Wee

8.1 Introduction
Governments have traditionally had a major impact on railway systems, at

least in so far as the infrastructure is concerned, classical reasons being
network externalities and the large fixed costs, including its impact on tar-
iffs that would strongly exceed marginal costs. The latter means that wel-
fare maximising tariffs would lead to losses for the railway operators.
Moreover, in several countries rail services have been subsidised for over
40 years in order to increase their competitive status in comparison to the
car. Environmental externalities are an additional reason for government
involvement: e.g. national, regional and local authorities decide on the
best route for a new line to be constructed in order to mitigate impacts on
the local environment.
The high involvement of government in the rail system means that deci-
sions on possible new rail projects are frequently based partly on ex ante
evaluations of them, Cost-Benefit Analysis (CBA) being the method used
most often and, to a lesser extent, Multi Criteria Analysis (MCA) (Hayashi
and Morisugi 2000). In CBAs on the benefit side, travel time savings and
the benefits of induced demand are dominant. Environmental aspects also
play an important role in the policy debate. However, apart from these
148 Piet Rietveld and Bert van Wee
benefits, railway stations as nodes of multimodal transport have an impor-

tant impact on cities and towns (see also Bertolini and Peek and Louw,
chapters 3 and 7 respectively, in this book). Bertolini and Spit (1998) dis-
tinguish between the value of a railway station as a place and as a node.
The place value relates to the links between the station and its surround-
ings, the node value to its position in (multimodal) networks. In our opin-
ion benefits of stations related to the value as a node are more or less dealt
with satisfactorily, assuming that a state-of-the-art transport model is used.
Such a model will cover changes in travel times at the multi-modal level
and possibly also address the changes in the discomfort of having to
change trains or modes. In this chapter we explore the extent to which
there may be further benefits related to developments in railway networks
or in railway stations under consideration. In this context we pay special
attention to indirect effects of network improvements, effects on land
prices, and experiential values of railway stations and their environments.
Section 8.2 first gives an overview of rail project benefits as they are
usually included in CBAs. Indirect welfare effects of network improve-
ments are analysed in section 8.3, and accessibility benefits and their im-
pacts on land prices are then covered in section 8.4. Experiential benefits
of railway stations are addressed in section 8.5 and in section 8.6 we dis-
cuss various other place-related benefits. Conclusions follow in section
8.7.
8.2 Benefits of rail projects in CBA: State of the art1
In most western countries cost-benefit analysis (CBA) is the method used

for ex ante evaluations of transport infrastructure projects (Hayashi and
Morisugi 2000). Basically a CBA is an overview of all positives (benefits)
and negatives (costs) of a project. These costs and benefits are quantified
as much as possible and expressed in monetary terms. Costs and benefits
occur in different years within the time horizon of the CBA. To deal with
this, they are presented as so-called net present values, implying that
even after a correction for inflation it is better to have 1 euro or $1 dol-
lar nowadays than in, for example, 2020. The discount rate is used to ex-
press this valuation and final results are often presented in summary tables.
The main indicators used concern the difference between costs and bene-
fits, the return on investment, and the benefit-cost ratio. Almost every
1 This section as well as section 8.5 is partly based on Van Wee (2007 in press).
Ex ante evaluation of railway station development projects 149
handbook on transport economics pays attention to CBA in transport (But-

ton 1993).
There are several reasons for the popularity of CBA for the ex ante
evaluation of infrastructure projects and its role in decision-making. Most
of the costs and benefits are relatively well-known, at least theoretically.
Investment, maintenance and operation costs can be derived from data of
projects constructed in the past or from the tenders. The most important
benefits are travel time savings, both of travellers and for freight transport.
Models are generally used to estimate demand of passengers or volumes of
goods transport that benefit from a new project. In the case of passenger
travel time, savings per trip can be estimated easily by comparing travel
times with and without the proposed infrastructure project using changes
in network characteristics. Next, the so-called Value of Time (VOT) is
used to express the value of shorter travel times into monetary terms. VOT
is higher for business travel and goods transport than for commuting; lei-
sure travel has in fact the lowest value of time. VOT differs between
modes, income classes and other characteristics of travel and travellers
(e.g. Gunn 2001). The value of both time and other transport aspects, such
as reliability, damage risks, and the availability of information about the
transport process altogether often lead to a comparison of the valuation of
rail compared to other modes, mainly road, and sometimes barge or short
sea shipping. However, what seems to be more under debate nowadays are
the indirect effects (effects in addition to the direct effects resultant from
reduced generalised transport costs) as well as environmental effects.
A second reason for the popularity of CBA is its presumed neutral
character as opposed to its main competitor: Multi Criteria Analysis
(MCA). In MCA effects are presented and weighed using weights per ef-
fect. Setting the weights is not at all value free; it is much easier to ma-
nipulate final outcomes of an MCA compared to a CBA. Note that CBA is
not completely value free either, for example, because of the use of the
utilitarian concept, the assumption that price tags should be based on con-
sumer preferences; and because several methodologies exist to obtain these
price tags, the choice taken often has a major impact on the outcomes. In
addition, the models used to estimate transport effects can be manipulated
(which, of course, is also true for MCA). Nevertheless, there is a broad
consensus that CBA is much more value free than MCA. The essential dif-
ference between the two is that, in MCA, weights express the priorities of
policymakers, whereas the weights used in CBA are expressions of collec-
tive consumer preferences. The aim of CBA is hence to ex ante evaluate
projects based on market preferences, of course after correcting for market
imperfections where applicable.
Note that travel time savings, often the most important benefits of infra-
structure projects, are not fully expressed in GDP. Travel time savings for
business trips and goods transport lead to higher productivity and lower
costs and impact on GDP, but if a commuter can leave home later because
commuting times are reduced, or because it takes less time to travel to visit
a relative, GDP is not affected. A broad approach to welfare is common in
CBA, which implies that all benefits for consumers are included, even if
they are not expressed in GDP. The point is that time savings are important
from an economic perspective even when they do not affect GDP, the rea-
son being that they do affect welfare since welfare depends, among others,
on leisure.
Next to travel time savings, additional travel (induced demand) is an
important category of benefits of transport infrastructure projects; these
benefits are generally estimated using the so-called rule of half (Button
1993) in order to arrive at an appropriate estimate of the total change in
consumer surplus. Hence the change in consumer surplus consists of two
components: first, the change in generalised transport costs of those who
did not change their behaviour after the project, and second, the change in
welfare position of those who did adjust their behaviour after the project.
The rule of half means that, on average, the benefits of those who did ad-
just behaviour is equal to half of the benefits of those who did not change
behaviour.
Other benefits of rail projects can be changes in profits of the company
operating the service, environmental, and safety benefits. Note that
changes in safety and the environment might be both positive and nega-
tive, thus implying that safety and environmental changes can be listed un-
der costs as well as benefits.
For CBA (as well as for other evaluation methods) spatial and temporal
system boundaries are crucial. If the spatial boundaries are too tight, rele-
vant effects might occur outside the area under consideration. All non-
marginal effects at the network level should be included. Temporal
boundaries meanwhile are also relevant. First, the (assumed) year of the
opening of a new infrastructure project is very important, because only af-
ter the opening can benefits be realized. Due to the general use of discount
rates, longer construction periods can significantly reduce a projects bene-
fits. Next, the time horizon included in a CBA is very important; this
should be at least two or three decades. Note that benefits in the very long-
term hardly affect the net present value due to the generally used discount
rates.
Importantly, underlying an appropriate analysis of the benefits of in-
vestments in any type of infrastructure is that a reliable estimate can be
given for the long-run response of the demand side with respect to the new
services supplied. This means, for example, that the traffic predictions on
which CBA is based should account for long-run adjustments of employ-
ees in terms of work choice location and/or residence. Long-run adjust-
ments in choice of services such as shopping or education also have to be
taken on board. Note that such adjustments take place within the con-
straints that the total number of persons living in zones, and the total vol-
ume of activities, both remain unaffected. In technical terms, this boils
down to a doubly constrained transport model (Ortuzar and Willumsen
2001). However, in the long-run, ongoing adjustments may take place,
leading to differentiated growth and decline of zones, for example, by ex-
pansion of industrial sites or office areas, and residential construction ac-
tivities. An appropriate CBA is based on models that take these adjustment
processes into account. Integrated land use transport models have been de-
veloped for this purpose (Wegener and Frst 1999).
Of special relevance is that land use in general as well as the interaction
between land use and transport depends on both market forces as well as
land use policies of governments (Rietveld and Wagtendonk 2004). When
governments interfere strongly with land markets this may prevent certain
adjustment mechanisms, implying that potential transport benefits of infra-
structure projects remain underutilised. For example, when a city centre is
provided with improved railway services, but zoning hampers the expan-
sion of office buildings or retail activities, this will inevitably have effects
on the long-run transport benefits of the project. Note that land use policies
including zoning not only have negative aspects; there may also be good
reasons for such interventions see section 8.4. It is beyond the scope of
this chapter to discuss the benefits and discrepancies inherent in land use
policies. However, important to this chapter is that obviously, incorporat-
ing such a subtle balance between market forces and government interven-
tion imposes high demands on the integrated land use transport models
used as a basis for CBA.
In short, at least at first sight, the main costs and benefits of possible
new infrastructure projects seem relatively easy to estimate, thereby mak-
ing CBA an attractive tool for the ex ante evaluation of such projects. But
is this actually true? Are all relevant benefits included in CBAs, especially
those related to the place value? The likely answer is no. To illustrate, the
Dutch transport advisory Council for Transport and Public Works [Raad
voor Verkeer en Waterstaat] (2004a) distinguishes several benefit catego-
ries of public transport:
Spatial and economic development
Possibilities to participate in activities
Liveability in central urban areas
Safety
Stimulate innovation by market parties
Most of these benefits categories are not, or at least not fully, included
in CBAs. We will elaborate current practice in CBAs of railways and rail-
way station areas in the next sections, but it is beyond the scope of this
chapter to discuss all the categories. We will pay special attention to three
issues; the first issue concerns the question of the extent to which travel
time savings, or travel cost reductions owing to infrastructure investments,
completely capture the welfare gains. This is known as the surplus equiva-
lence problem and is addressed in section 8.3. The problem is general in
nature, since it not only applies to railway projects, but rather may apply to
any infrastructure project. The second issue discussed in section 8.4. ad-
dresses the relationship between transport cost changes and land prices.
This theme is also general, although in the case of railway projects, the ef-
fects on land prices are particularly relevant, since given their nature
the changes in land prices tend to be more focussed in space compared
with road projects, where effects are more dispersed. The third issue dis-
cussed in section 8.5 concerns the experiential benefits of projects. We will
indicate that, although such benefits are to some extent incorporated in
standard cost benefit studies, in particular in the case of value of time stud-
ies, other dimensions of experiential benefits are often ignored in studies
of this type. In particular, investments that not only affect their functional
use in terms of travel time, but also affect the experiential values of staying
at the station or its neighbourhood, may have additional benefits that are
relevant for cost-benefit analysis. Along with these three issues, we briefly
discuss other limitations of current CBAs for rail infrastructure projects, in
particular those related to the place value of railway stations.
8.3 Indirect effects: the surplus equivalence issue
There is increasing interest in the question on the extent to which changes

in consumer surplus would underestimate (or overestimate) actual welfare
changes in the case of market imperfections. This is called the surplus
equivalence problem by Jara-Diaz (1986). The starting point of the discus-
sion is the result of Jara-Diaz (1986) that, with an improvement of the
transport system, the change in consumer surplus as measured on the
transport market is equal to the welfare changes of all actors in the market
under the condition that all markets are perfect. The latter means that there
should not be imperfections such as monopolistic or oligopolistic price set-
ting on particular markets such as transport, the markets of traded goods,
or other markets such as labour market or land market. When imperfec-

tions do exist, this surplus equivalence no longer holds. During the past
decade lively debate has addressed the question of the extent to which
market imperfections do matter for an appropriate estimation of the bene-
fits of transport system improvements.
For example, the Transport and the Economy, Standing Advisory Com-
mittee on Trunk Road Assessment SACTRA report (1999) compares total
welfare increases of infrastructure improvements with changes in con-
sumer surplus, and finds that the former is about 30% higher than the lat-
ter; this analysis is based on a monopolistic competition model dealing
with regional trade. Another result is that Newbery (1998) finds indirect
welfare effects of 3 to 8% in the context of oligopolistic competition in a
spatial price equilibrium model. Davies (1999) obtains an indirect welfare
effect of about 12% in a partial oligopolistic model. Brcker and Schneider
(2002) arrive at indirect welfare effects of about 20% via a Computable
General Equilibrium (CGE) model for European regions. Oosterhaven and
Elhorst (2003) obtain indirect welfare effects ranging from 20 to 80% in a
CGE model for Dutch regions. And in a monopolistic competition model
with localisation economies, Venables (2004) finds increases ranging be-
tween 60-150%. In a recent paper, Zhu et al. (2006) find that indirect wel-
fare effects due to labour market imperfections are about 10-20% above
the direct consumer surplus-related effects of infrastructure improvements.
These figures reveal the substantial differences between the various
studies, depending on the nature of the imperfection and the type of model
used. The SACTRA report indicates that, from a theoretical perspective,
these indirect welfare effects may be both negative and positive. However,
the actual cases addressed thus far arrive at positive values. The policy
relevance of these studies is obvious. When there would be a tendency
that, cost-benefit analysis based on consumer surplus in the transport mar-
ket, would systematically underestimate total welfare change, this might
lead to biases in decision-making on infrastructure projects. This not only
appears in the case of infrastructure projects but may also occur in case of
other transport policies, such as infrastructure pricing. More to the point, if
the gap between the two welfare measures were to remain smaller than
(say) 10%, there would not be much reason to worry, but for larger gaps
there is a real risk that promising infrastructure plans would not be exe-
cuted.
The present state of knowledge is not yet sufficiently developed to give
definite answers concerning the possible size of indirect effects. Remem-
ber that SACTRA (1999) warns that these effects may be negative in cer-
tain cases. The degree and nature of market imperfection is of course an
important determinant of the size of the indirect effects. For example, in
which market does the imperfection take place: in the market of the goods
that are transported, or in the labour market? And how large is the imper-
fection, for example in terms of market power? Most of the studies thus far
indicate that the indirect effects are limited in size, though not negligible.
For example, Elhorst et al. (2005) suggest that plausible values for indirect
effects are between 0 and 30%. We close this discussion by emphasising
that in many countries market imperfections might be relatively large in
railway station areas because they are sited in densely-built areas heavily
under the influence of government policies. In some cases market imper-
fection is contingent on one or a few powerful actors, e.g. those that have
large land positions. The surplus equivalence problem may therefore be
relatively important for the ex ante evaluation of railways and railway sta-
tion changes.
8.4 Land values and transport costs
An important research question asks how changes in transport costs relate

to changes in land values. To what extent will the benefits of transport in-
frastructure investments be reflected by changes in land values? This issue
has been addressed among others in urban economic theory (Mohring
1961; Arnott and Stiglitz 1981; Fujita 1989). The contest of these contribu-
tions is the monocentric city model where transport takes place by car in a
uniform, non-differentiated transport network. Let us first consider the re-
lationship between transport costs per unit distance and total expenditures
on transport. When transport costs per unit distance decrease, there will be
an increase in demand for transport. Three possibilities then arise: demand
may increase more than proportional to the cost decrease, the increase may
be exactly proportional, or it may be less than proportional. As a conse-
quence, total transport-related expenditures (the product of price and quan-
tity) may increase, remain constant or decrease. Which of the three possi-
bilities will occur depends only on consumer preferences, more in
particular on the substitution elasticity between transport demand and de-
mand for other consumption goods.
A second step is the relationship between total transport-related expen-
ditures and total land rents in a city. Arnott and Stiglitz (1981) show that
under rather general assumptions in a monocentric city there is a fixed ra-
tio between total transport costs and total land rents. The ratio depends on
the shape of the city. For example, in a linear city the ratio of the two
would be 50-50, whereas in a circular city the ratio would be 67-33.
When we combine these two elements one may conclude that the effect
of a uniform decrease of transport costs on total land rents is uncertain. In-
deed, Arnott and Stiglitz (1981) ascertain: Aggregate land rents may, in
general, rise, remain the same, or fall in response to a transport improve-
ment. An important implication is that it would be misleading when one
infers the benefits of transport improvements from the changes in aggre-
gate land rents. There is no simple way to relate changes in transport costs
to changes in land rents.
Significantly, the above result only holds under specific assumptions.
We will discuss some of these in more detail below. One assumption is the
spatial uniformity in the transport costs per unit distance. Another assump-
tion concerns the absence of market imperfections.
8.4.1 Non-uniform change in transport costs
The above results hold when there is a uniform decrease in transport costs,
an obvious example being a decrease in the cost of fuel. However, the re-
sult does not hold in the case of non-uniform changes in transport costs,
such as in the case of a specific transport infrastructure improvement, re-
lated to a particular highway connecting urban fringe with city centre. One
then expects that some zones clearly benefit and others do not, but instead
experience a decline in relative terms. This case is studied by Mohring
(1993), who considers the effects of an introduction of a radial expressway
on land prices in a monocentric city. The expressway will lead to an ex-
pansion of the city, at least in the area it serves. In the context of this
model, where population is assumed to be given and long-run adjustments
are considered, the rest of the city which is farther away from the express-
way will shrink. Mohring demonstrates that, when the reduction in trans-
port costs made possible by the expressway in the part of the city served
by it is modest, this will dominate the negative effect on land values in the
rest of the city.
The overall net effect on land rents in the city will be an increase. If the
reduction of transport costs were to be substantial, its impact area would of
course expand, but it appears that the fraction of total travel cost savings
appearing as a rent increase becomes smaller. Mohring further shows that
the effects of transport infrastructure improvements on land rents depend
considerably on behavioural reactions in terms of trip frequencies and de-
mand for space per dwelling. Therefore, there is no clear-cut relationship
between the change in land rents and the change in transport costs.
Mohring (1993) concludes that increases in land rents in zones directly af-
fected by the infrastructure project overestimate benefits.
8.4.2 Market imperfections
Limiting ourselves to commuting, a central assumption in the urban eco-

nomic models above is that workers trade-off the benefits of living close to
the centre where they have their job, with the price of land (and indirectly
house price). Living close to the centre implies lower transport costs, but
in a free market this would lead to higher bids to use the land for residen-
tial purposes. Thus, a worker is entirely compensated for the monetary and
non-monetary costs involved in his transport activities (Fujita 1989). This
reasoning holds in the case of monocentric cities, but also applies to poly-
centric structures as long as substantial spatial concentrations of jobs exist.
However, in an imperfect market, this conclusion on compensation for
travel costs is no longer true. For example, in a labour market with unem-
ployment where search for jobs takes place and where workers and em-
ployers negotiate the wage, workers are only partially compensated for
their commuting costs (van Ommeren and Rietveld 2006).
The reason is that there is always a probability in the future that they
will be unemployed for a certain period, and during that period such com-
pensation of course does not occur. Should the costs of changing residence
be zero, workers would immediately move to low rent locations (at the
fringe of the metropolitan area) to avoid paying high rents. This would
lead to a concentration of unemployed at the fringes of cities, a situation
observed, for example, in the Paris region. However, this pattern may only
be expected when unemployment durations are very long. Otherwise the
transition costs related to changing residences would not be worth the ef-
fort. Thus, next to the urban shape factor that influences the ratio between
total transport costs and total land rents, we find another factor: the degree
of imperfections on the labour market.
Another relevant market imperfection concerns the land market itself. In
the case of spatial externalities, zoning may be applied to reduce negative
spillovers from one activity (for example airports) to other activities (for
example residential use) in zones nearby. Zoning may also help to preserve
open space and landscape quality in view of the potential claims of devel-
opers that ignore the positive contribution to other actors such as residents
who live near the open space and who benefit from it (cf. the green belt
policy around London). The point is that such positive spillovers would be
ignored by the developers.
Clearly, zoning measures as these potentially contribute to welfare im-
provements. Conversely, zoning may of course, also cause welfare losses,
i.e. when it would unnecessarily restrict market operations. A potentially
negative effect of zoning to create green belts is, for example, that cities
become crowded and congested, and also that commuting distances for
workers living beyond the green belt would be unnecessarily long. A sur-
vey of zoning measures is given by Kruk (2005) He finds that zoning has a
differentiated upward pressure on house prices outside the areas affected
by the zoning measures. Of particular interest is the study of Sivitanidou
and Wheaton (1992) who explore the effects of restrictions on commercial
areas in central cities; it appears that this drives up the costs and intensity
of commercial land use in central areas, which reduces the competitive po-
sition of cities, and implies smaller rent receipts for residential areas.
8.4.3 The case of railway projects in cities
We have already discussed the theme of non-uniform transport cost

changes in cities in the case of highway developments. An even stronger
case of non-uniform cost changes would be that of railway development.
This case is addressed, for example, in Debrezion (2006) who explores the
effects of investments in railway infrastructure in a theoretical urban eco-
nomics model. He found that such investments enlarge cities (in line with
the reasoning of Clark (1958), who found that railway investments were a
key driver of urban growth in the 19th century). Another important finding
is that changes in land rents are most pronounced near to stations, but they
also do occur in the whole urban area. Next, Debrezion finds that restric-
tions on commercial activities in centres lead to higher office buildings and
corresponding rents. The imposition of the constraints on the total com-
mercial rents is positive, implying that the decrease of the total size of land
available for commercial activity is more than offset by an increase in rent
per area. Thus, from this perspective the land use regulation would lead to
higher rents. However, an opposite development is that the total land rents
for residential activities would be considerably less, implying that total
rents would be negatively affected by the regulatory measures.
Our conclusion is that there is no simple recipe to link benefits of trans-
port cost reductions to changes in land values. The aggregate land values
may even decrease as a consequence of transport cost changes. Relevant
factors playing a role are the shape of the city, market imperfections, the
structure of the transport network, and the specific place in the network
where transport costs are reduced. This means that there is a considerable
need for context-specific studies on the effects of transport cost changes
(for example as represented by changes in accessibility), on land and prop-
erty values.
Another conclusion we draw here is that changes in land values is not a
reliable indicator of welfare gains in urban systems. This does not mean to
say, however, that such changes in land values are irrelevant. In this dis-
cussion on the use of social cost-benefit analysis of investment projects, it

is useful to provide a link with issues related to an uneven distribution of
costs and benefits and other possible barriers to implementation. The point
is that there may be projects that pass the tests of social cost-benefit analy-
sis, but where nevertheless the project cannot be implemented because the
main beneficiaries have insufficient financial resources. A social cost-
benefit analysis is focussed on the net balance of costs and benefits. When
the balance of an investment project is positive, it should in principle be
carried out. However, from a commercial perspective, it may well be that
the benefits mainly accrue to actors who do not pay for the project. For ex-
ample, car drivers usually do not pay directly for the specific road they
use, and owners of real estate property at a location where a station is built
benefit from the project without paying for it. In the case of railway station
areas, we refer to the chapter of Peek and Louw in this book.
This is one reason for the gap between the evaluation of investment pro-
jects based on private sector considerations and public considerations. One
way to reduce this gap is to use increases in land values or property values
as a way to increase the interest of the private sector. Thus we arrive at the
theme of value capturing, defined as procedures to (partly) charge benefi-
ciaries of public sector investments with the aim to mobilise resources for
these investments. Note that this definition of value capturing is somewhat
broader than the case described above: it also includes cases where prop-
erty values would not increase. As Table 8.1 shows, there are several
routes available (Offermans and Van der Velde 2004; Council for Trans-
port and Public Works 2004b).
Table 8.1. Value capturing alternatives

Actor contributing Degree of voluntariness Examples
Developers: Voluntary Public private partnerships
Benefit sharing
Developers Involuntary Auctioning of development
concession
Development charge
Owners of land and prop- Involuntary Development impact fee
erty
Increase of property tax
Internal actors Involuntary Real estate owned by public
national railways
The specific instrument to be used depends strongly on the institutional

environment in a given country. An important factor is for example the
role of property taxes in the total fiscal system. A small share of property
taxes in total taxes obviously limits the possibilities of using the property
tax instrument.
The discussion above underlines how important it is to assess the poten-
tial effects of infrastructure projects on property values. Even when these
effects do not play a direct role in cost-benefit analysis, they are neverthe-
less relevant because they may give an indication where the main benefici-
aries are located and whether their gains can be mobilised to increase the
financial resources for good projects.
8.5 Experiential benefits
8.5.1 Experiential benefits during trips
Investments in infrastructure and railway stations are usually evaluated in

terms of travel time implications for travellers. As already mentioned in
section 8.2, there is substantial literature on valuing the time use involved.
External circumstances certainly play a large role in the valuation of travel
time. The valuation of time depends strongly on the trip purpose: from
high in the case of business trips, to medium in the case of commuting, to
low for other trip purposes. Furthermore, there are some indications that
the value of travel time for car drivers increased less than proportional to
incomes in the past 10 years. This may be due to the fact that travel by car
is becoming more convenient owing to, for example, the gradual introduc-
tion of ICT equipment in cars. Note that next to income, other factors im-
pact on the value of travel time.
In line with the less than proportional increase of the value of travel time
with income is the trend in the value of time for train travel. Contrary to
expectations based on income developments, Gunn (2001) found that be-
tween 1988 and 1997 in the Netherlands the value of travel time hardly
changed and in some cases even decreased, despite the increase in incomes
during that period. He found the largest decrease for people travelling by
train for business purposes: a decrease of almost 25% (from 40.16 to 30.87
Dutch Guilders2 per hour), explained by the decreasing disutility of travel.
In the case of business train travellers, this might relate to the use of lap-
tops, cell phones, and in recent years and probably in the near future, to
wireless internet use.
2 At the time of the introduction of the Euro, 1 was equal to 2,20 Dutch Guild-
ers.
Another dimension of variations in the value of travel time is that trans-

port time is not homogeneous. This holds true even for car transport. For
example, driving under non-congested conditions is different from driving
in a queue and would probably lead to a higher willingness to pay for a
travel time gain due to a reduction in congestion than in the case of free
flow traffic. And obviously walking to the final destination after having
parked the car is valued much higher than time spent in the car. Such a dif-
ferentiation is even more relevant in public transport trips, where the walk-
ing and waiting elements are more pronounced. As found by Wardman
(2001), travel time spent on platforms is experienced much more nega-
tively than the in-vehicle time.
A differentiation in the valuation of travel times is that reliability is an
important component. As we will show, reliability has considerable expe-
riential dimensions that have implications for the perception of travel time.
We discuss the underlying dimensions below. An important reason why
travellers prefer reliable services is because they dislike the uncertainty in-
volved (RAND 2005). Even when they do not have to arrive at a certain
time, they will probably still dislike the uncertainty, as it leads to feelings
of stress. Travellers appear to be risk averse. They want to be in control,
and unreliable transport service reduces this feeling of being in control;
this appears to be stronger in public transport than in private transport.
Another reason why people dislike unreliability is that they expect that,
apart from the strict uncertainty effect on travel time, it may lead to incon-
venience. For example, unreliability can imply that trains become very
crowded so that there is a high probability that no seats will be available
for them. (Rietveld et al. 2001) Moreover, it is well-known that most of the
increases in travel time due to unreliability take place as waiting times on
platforms. And it is also well-known that travellers dislike travel time in-
side vehicles more than outside vehicles (Wardman 2001).
Finally, unreliability may be bad for the traveller who may have sched-
uling costs (Small 1992). By scheduling costs we mean that the traveller
wants to arrive at a certain time to his or her destination. Examples include
arriving on time for work, education, a business appointment, a job inter-
view, a community activity, before a shops closing time, before final
check-in at the airport, etc. The usual way to analyse scheduling costs is
that passenger utility is assumed to be dependent on ticket costs, total
travel time, plus the time that she is early (compared to the preferred arri-
val time) or the time that she is late. In the case of an early arrival (known
as schedule delay early, SDE) the time lost by arriving early will usually
not be evaluated the same as the time spent for the trip, since it may be
used for other useful or relaxing purposes. On the other hand, the time lost
by arriving late (usually called schedule delay late, SDL) is expected to be
more strongly valued than the time spent in transport. The latter of course
depends on the type of activity after the trip; the penalty for being late is
high when one misses a job interview, and lower when one merely arrives
late for a social gathering.
This discussion makes clear that time costs of travelling are differenti-
ated and that the conditions under which travelling takes place are indeed
important. This has led to special attention to the experiential dimensions
of travelling (van Hagen 2004). When travel time and waiting time are un-
avoidable, and we must also accept a certain amount of unreliability, it
makes sense to enhance the quality of the experience during the trip. Part
of this strategy would be to increase trip convenience. Examples would
comprise guaranteeing that a seat is available even during rush hour, that
platforms are safe and clean, and that waiting accommodation is heated.
One further step is that trips become an experience and, in particular,
that railway stations change in character, away from functional transfer
points towards places that are enjoyable. This approach is probably most
promising for non-daily travellers. In terms of clientele this is a large
group. A tendency can be observed that most railway trips are made by
those who use the train on a regular basis, but above all there is a much
larger group that uses trains only occasionally (see Table 8.2). For these
travellers the trip, including the time spent at the station, may be an experi-
ence valued positively. An indication of this is already found in the general
result that the value of time applied to commuting trips is substantially
higher than for social visits or recreational trips. It may be interesting in
this respect to compare railway stations to airports; a recent tendency is
that airports carry out substantial investments in side activities such as
shopping which has led to the airport city concept. In doing so the com-
mercial basis of airports is gradually broadened from mere transfer activi-
ties towards a complete range of services. Thus, investments in airports as
places with positive experiential qualities have positive economic impacts.
For example, airports score very high in terms of prices for floor area in re-
tail activities.
Table 8.2 underlines that airports and railway stations are rather differ-
ent in terms of passengers served, but for the larger stations there may be
opportunities to follow a similar strategy. Indeed, additional activities such
as the exploitation of stations are becoming increasingly important in the
total revenues or profits of railway companies. A further improvement of
this could therefore be achieved by improving the experiential quality of
railway stations and the surrounding areas. This will be the subject of the
next subsection.
Table 8.2. Comparison between railway stations and airports

Railway Stations Airports (for example,
Amsterdam Airport
Schiphol)
Number of passengers per day Varying from 500 to Schiphol: circa 100 000
500000 passengers per passengers per day
day (local and regional (national/international
market area) market area)
Share of transfer passengers Approx. 10% Schiphol: approx. 45%
Average duration of transfer Approx. 5-10 minutes Approx. 120 minutes
Share of friends/relatives who No Yes
accompany travellers to airport
important?
Share of daily users among 60% 2%
total number of passengers
(more than 3 times per week)
Share of travellers with trip Approx. 20% Approx. 60%
purposes recreation and
tourism
8.5.2 Experiential benefits of railway station areas
Railway stations potentially offer experiential benefits. Especially in the

past, railway stations in central urban areas and the surrounding areas were
expressing high architectural qualities. Some of these stations, often cre-
ated a century ago, have maintained their high quality, and are still re-
garded as attractive places to visit (for example Antwerp Central Station).
On the other hand, there are also many railway stations that are thought to
be undesirable places. For example, Bowes and Ihlanfeldt (2001) find that
in the USA, railway stations and their immediate environments (say, up to
250 metres) have a reputation for crime, noise and lack of safety. One im-
plication is that real estate values in the immediate environment of the sta-
tions in the USA are relatively low.
One may observe several efforts to create attractive railway stations and
surrounding areas in many countries over the last decades. For example,
the development of the high-speed rail network in Europe has stimulated
the development of new railway stations. As described by Kido (2005),
this has led to a good number of railway stations with high aesthetic quali-
ties in countries such as France, Germany and the UK. A recent example is
the opening of a new central station in Berlin, which is intended to become
a landmark in the city, so that it also has a broader positive impact. The
Amsterdam South project has a similar ambition, in that the plans to make
the whole station underground would create space for a top office location
where the infrastructure guarantees a high level of accessibility, but with-
out the present barrier effects related to railways, metros and expressways
at the location.
Experiential benefits of this type are usually ignored in CBAs. They are
of potential relevance, however, for the evaluation of transport projects.
Constructing railway station areas with a particularly high quality not only
makes them more enjoyable for visitors, workers and residents, but may
also lead to an increase in their number. Both factors are relevant in CBAs.
The higher appreciation for a railway station area implies a higher willing-
ness to pay for making a trip to this destination, and this implies a higher
consumer surplus. The same holds true for an increase in the number of
visitors. An obvious difficulty is that the effect of investments in spatial
quality on willingness to pay is difficult to predict, so that a cost-benefit
analysis of investing in quality is not easy. But one should be aware that
this difficulty does not arise only when one wants to conduct a cost-benefit
analysis, but also appears when one makes predictions on number of visi-
tors to the area. These not only depend on travel times, but also on the per-
ceived attractiveness of the area.
First steps towards the ex ante estimation of the benefits of investing in
quality can be made by analysing the present travel behaviour of people.
The actual choices individuals make in terms of destinations of trips such
as station areas, reveals their willingness to pay in terms of time and
money. Destinations with distinct qualities will be more attractive places to
visit. This method is known as the travel cost method in environmental
economics (Perman 2003); it enables the estimation of the implicit value
attached to a certain destination. Ex post analysis of choice among destina-
tions with varying qualities can be used to assess the contribution of spe-
cific artistic features of destinations towards their attractiveness. It is obvi-
ous that such an ex post analysis of actual travel behaviour cannot be easily
used for an ex ante analysis of projects that have not yet been realised, but
it would provide a promising step towards better understanding the impact
of spatial quality on travel behaviour.
8.6 Other place benefits
8.6.1 The link with land use
In section 8.3 we emphasised the role of public transport for spatial devel-
opment. Transport and land use interact: changes in travel times and costs
(e.g. due to new infrastructure projects, changed prices or subsidies) for
distinguished modes may affect land use, and land use changes may affect
travel behaviour (Wegener and Frst 1999). Many models as used for
CBA do not include this two-way interaction. If available, using a Land
Use Transport Interaction model (LUTI model) would allow researchers to
give an indication of the land use changes resulting from transport
changes, and then determine the subsequent related economic, social and
environmental impacts. These changes are even more important for rail-
ways and railway station areas than for roads, since differences of accessi-
bility over space are much larger for rail than for road transport (Geurs and
Ritsema van Eck 2001). The spatial contrast within a region largely relates
to differences between railway station areas and other areas.
A reliable LUTI model is often not available. If this is the case, an alter-
native might be to use expert judgement to estimate the land use changes
due to the transport changes. And even if a LUTI model is available, the
question is whether the benefits of applying it exceed the costs. Our gen-
eral impression is that if an existing model (both the software and scenario
runs including the data needed) is available anyway, adapting it for CBA
purposes may be worth the effort. But if a new model needs to be devel-
oped, the expert judgement approach may be the preferred method. Note
that if a traditional transport model is used for a CBA instead of a LUTI
model, ignoring the land use changes is not a shortcoming of CBA as a
method, but rather of the transport model itself.
8.6.2 Other land use issues
Apart from the importance of land use and transport interactions for de-
mand forecasting, the link between land use and transport offers other
challenges, some of which are discussed below. First, it is important to re-
alise that better transport reduces transport costs, which accentuates the
advantages of agglomeration. In the case of railways the agglomeration ef-
fect mainly occurs within a few kilometres from the main railway stations.
Agglomeration suggests that there are advantages of economic density,
which may or may not yield additional benefits in CBA, and which are
even not covered by LUTI models (assuming perfect models). The con-
sequences of agglomeration are well-known, but its causes, its specific na-
ture and quantitative relations are not. Discovering more about the causes
is an important challenge for CBA, and even more so for spatial econom-
ics. This, alongside the effects in CBA of rail infrastructure projects, is the
main issue. Second, the intensive use of land reduces urban sprawl and al-
lows for the preservation of open space. In CBA the valuation of open
space remains a challenge, not only because the agricultural value of land
is important, but also because of its recreational use and external effects.
Third, rail is a land use efficient transport mode, compared to road trans-
port, not only because of the efficient use of infrastructure space, but also
because of a lower claim for car parking. Measuring and valuing these
benefits is an important issue.
8.6.3 The robustness of policies
Another issue is the robustness of infrastructure and land use concepts.

How vulnerable are we to, for example, an anticipated or unexpected limi-
tation in energy availability for transport? Such limitations may be the re-
sult of political instability in oil producing countries, much higher prices
for fuel, or stringent environmental policies. The preferences of consumers
and firms may also change in the future. What will happen if sustainably-
produced energy becomes available at reasonable prices? The environ-
mental benefits and costs of project alternatives might then change dra-
matically.
8.6.4 The option value
Current CBAs focus on user benefits only. There is the question, however,
of whether non-user benefits are relevant. The first category of non-user
benefits relates to the so-called option value. The option value can be de-
scribed as an individuals valuation of the opportunity to use a particular
transport mode or piece of infrastructure for unexpected future use (Roson
2000). For example, car owners may value their ability to use a public
transport service when, for whatever reason, they cannot use the car due to
unavailability, or breakdown, bad weather, increase in fuel prices or other
car costs, or the loss of the ability to operate a car. Other non-use benefits
relate to altruistic and existence values: a person may value a rail network
even if he never uses it, because he appreciates that others have the possi-
bility to use it (altruistic value) or even its very existence (a classical ex-
ample being the South Pole: people value that it exists even if no one,
themselves included, would ever visit). We refer to Geurs et al. (forthcom-
ing) for an extensive discussion of the option value and two regional case
studies showing that this value may be of the same order of magnitude as
the use value.
8.6.5 Rail as part of a comprehensive plan
Rail projects are often considered as more or less independent infrastruc-

ture plans, ignoring the fact that they may be part of a policy package that
cannot be realised unless the rail infrastructure is expanded. This is par-
ticularly true for local and regional (light) rail. Reducing the role of the car
in central urban areas without at least increasing the capacity of alterna-
tives (including rail) may result in an unrealistic scenario or a decrease in
the functioning of these central urban areas, or at the least, a decline in po-
litical support. If strong connections between rail projects and other poli-
cies, such as reducing road capacity and parking space for cars or road
pricing, exist, the complete plans should be subject to CBA (or some other
evaluation methodology, such as MCA), and not the rail infrastructure by
itself. The explicit link between the congestion charge in London and the
use of revenues for the improvement of the Underground illustrates the
links between policies as mentioned above. Conversely, to only evaluate
mixed policies is a potentially dangerous mix up of the different responsi-
bilities of researchers and policymakers. Researchers should analyse all the
possibilities, that is, only limiting car use, only rail investment, and
combination of both. If the combination alone is analysed and presented,
it might turn out to be viable only because the limiting of car use is very
viable, while the rail investment may be very unprofitable. This underlying
information is important and should not be hidden by researchers to obtain
public support. Obtaining public support is the job of policymakers.
8.7 Conclusions
This chapter has shown that CBA is a popular method to ex ante evaluate
transport infrastructure projects. Its popularity is partly explained by the
fact that most costs and benefits seem to be well-known and may be ex-
pressed relatively easily in monetary terms. However, current practice with
respect to CBA of rail projects ignores several benefit categories. These
are partly related to the function of rail infrastructure as transport net-
works, but also to the place values of railway station areas. Important sub-
jects not currently dealt with properly include the surplus equivalence is-
sue, land values and experiential benefits. Ignored benefit categories might
be substantial. Including them can therefore significantly influence the
benefits-to-costs ratio of rail projects. Current CBAs are therefore only of
limited use for policy decisions. The problem lies not with the CBA
method in itself, but rather the input for the model as derived from trans-
port and other models.
Some of the additional benefits are very difficult to express directly in
monetary terms, such as experiential benefits. An alternative method is to
rank these benefits, thus allowing the researcher to position them between
other benefit categories that can be monetised. Then at least the interval of
monetised benefits would be known. Another solution to the problem
might be the combination of CBA and MCA, the additional benefits being
included in the MCA part of the ex ante evaluation methodology.
A special kind of benefits deserves attention: those that might occur
over the very long-term, such as those related to the robustness of the land
use and transport system. Discounting reduces their net present value to a
marginal level unless real benefits are assumed to increase over time with
a percentage in the order of magnitude of (or higher than) the discount rate
used.
References
Arnott R, Stiglitz JR (1981) Aggregate Land Rents and Aggregate Travel Costs.
Economic Journal 91:331-347
Bertolini L, Spit T (1998) Cities on Rail: The Redevelopment of Railway Station
Areas. E & FN Spon, London/New York
Bowes DR, Ihlanfeldt KR (2001) Identifying the Impacts of Rail Transit Stations
on Residential Property Values. Journal of Urban Economics 50:1-25
Brcker J, Schneider M (2002) How Does Economic Development in Eastern
Europe Affect Austrias Regions? A Multiregional General Equilibrium
Framework. Journal of Regional Science 42:257 285
Button KJ (1993) Transport Economics. 2nd edition. Edward Elgar, UK
Clark C (1958) Transport Maker and Breaker of Cities. The Town Planning Re-
view 28:237-250
Council for Transport and Public Works [Raad voor Verkeer en Waterstaat]
(2004a) Ieder zijn deel. The Hague
Council for Transport and Public Works [Raad voor Verkeer en Waterstaat]
(2004b) De Waarde van Openbaar Vervoer. The Hague
Davies S (1999) Review of the Evidence of the Incidence of Imperfect Competi-
tion in the UK. Report to the UK Standing Advisory Committee on Trunk
Road Assessment (SACTRA), London
Debrezion G (2006) Railway Station Effects on Real Estate Prices. Tinbergen In-
stitute, Amsterdam
Elhorst J, Heijma A, Koopmans CC, Oosterhaven J (2005) Indirecte Effecten In-
frastructuur Projecten. SEO, Amsterdam
Fujita M (1989) Urban Economic Theory. Cambridge University Press, Cam-
bridge
Geurs KT, Ritsema van Eck JR (2001) Accessibility Measures: Review and Ap-
plications. RIVM report 408505006, National Institute of Public Health and
the Environment, Bilthoven
Geurs KT, Haaijer R, Wee B van (forthcoming) The Option Value of Public
Transport: Methodology for Measurement and Case Study for Regional Rail
Links in the Netherlands. Submitted to Transport Reviews
Gunn H (2001) Spatial and Temporal Transferability of Relationships between
Travel Demand, Trip Cost and Travel Time. Transportation Research E
37:163 189
Hagen M van (2004) Klantwensen bij Verblijven en Verplaatsen. NS, Utrecht
Hayashi Y, Morisugi H (2000) International Comparison of Background Concept
and Methodology of Transportation Project Appraisal. Transport Policy 7:73
88
Jara-Diaz SR (1986) On the Relation between Users Benefits and the Economic
Effects of Transportation Activities. Journal of Regional Science 26:379 391
Kido EM (2005) Aesthetic Aspects of Railway Stations in Japan and Europe.
Journal of the Eastern Asia Society for Transportation Studies 6:381 396
Kruk R van der (2005) Hedonic Valuation of Dutch Wetlands. Tinbergen Institute
PhD thesis, Thela Thesis, Amsterdam
Mohring H (1961) Land Values and the Measurement of Highway Benefits. Jour-
nal of Political Economy 69:236-249
Mohring H (1993) Land Rents and Transport Improvements: Some Urban Par-
ables. Transportation 20:267-283
Newbery D (1998) Efficiency Benefits of Transport Cost Reductions: Report Pre-
pared for SACTRA. Department of Applied Economics, Cambridge
Offermans RN, Velde DM van der (2004) Value Capturing, Potentieel Financie-
ringsinstrument voor Nederland? Achtergrondrapport voor Ieder Zijn Deel:
Locatiebereikbaarheid anders Aanpakken. Raad voor Verkeer en Waterstaat,
Erasmus University
Ommeren J van, Rietveld P (2006) Incomplete Compensation for Transport Costs
in Urban Models with Imperfect Markets. Papers in Regional Science (forth-
coming)
Oosterhaven J, Elhorst JP (2003) Modelling Interactions between the Economy,
the Environment and Transportation at the Local and Regional Level, with an
Application to Dutch Maglev Projects. Paper presented at the TRIP research
conference on the Economic and Environmental Consequences of Regulating
Traffic, Copenhagen
Ortuzar de Dios J, Willumsen LG (2001) Modelling Transport. Wiley, New York
Perman R (2003) Natural Resource and Environmental Economics. Pearson, Har-
low
RAND (2005) The Value of Reliability in Transport: Provisional Values for the
Netherlands Based on Expert Opinion. RAND Europe
Rietveld P, Vuuren D van, Bruinsma F (2001) Coping with Unreliability in Public
Transport Chains: A Case Study for the Netherlands. Transportation Research
A 35:539 551
Rietveld P, Wagtendonk AJ (2004) The Location of New Residential Areas and
the Preservation of Open Space: Experiences in the Netherlands. Environment
and Planning A 36:2047 2063
Roson R (2000) Social Cost Pricing when Public Transport is an Option Value.
Innovation 13:81 94
SACTRA (1999) Transport and the Economy. Standing Advisory Committee on
Trunk Road Assessment (SACTRA), London
Sivitanidou R, Wheaton WC (1992) Wage and Rent Capitalization in the Com-
mercial Real Estate Market. Journal of Urban Economics 31:206 229
Small KA (1992) Urban Transportation Economics, Fundamentals of Pure and
Applied Economics. series 51, Harwood Academic Publishers, Chur
Venables AJ (2004) Evaluating Urban Transport Improvements: Cost-Benefit
Analysis in the Presence of Agglomeration and Income Taxation. Working
paper, London School of Economics, London
Wardman M (2001) A Review of British Evidence on Time and Service Quality
Valuations. Transportation Research Part E 37:107 128
Wee B van (2007) Rail Infrastructure: Challenges for CBA and Other Ex Ante
Evaluations. Transportation Planning and Technology (in press)
Wegener M, Frst F (1999) Land Use Transport Interaction: State of the Art. De-
liverable D2a of the project TRANSLAND (Integration of Transport and
Land Use Planning). Universitt Dortmund, Insititut fr Raumplanung, Dort-
mund
Zhu X, Ommeren J van, Rietveld P (2006) The Surplus Equivalence Problem in
the Case of Infrastructure Improvement with an Imperfect Labour Market.
Vrije Universiteit, Amsterdam
9 Multicriteria analysis of a high-speed railway
station area development project
Ron Vreeker

the Netherlands
9.1 Introduction
The introduction of high-speed trains in the Netherlands has led to the de-
sign of the so-called New Key Projects (Nieuwe Sleutelprojecten, NSPs)
whose aim is to capitalise on the synergy effects that may exist between
railway network improvements and urban redevelopment by applying the
compact city concept. In this chapter the largest of such New Key Projects,
the South Axis project, takes a central position as a case study; it consti-
tutes a massive investment in housing, offices and infrastructure in the
southern district of Amsterdam. It is believed that a variety of effects are
associated with these investments. The Dutch government and the Munici-
pality of Amsterdam, for example, proposed these investments in railway
infrastructure in order to reduce car mobility and promote more environ-
mentally friendly transport modalities (see also Newman and Kenworthy
1989a; 1989b; Newman 1992). Furthermore, knowledge-based business
services may benefit from economies of scale or scope by clustering
around such clients as large financial institutions or their competitors. Fi-
nally, by having the goal of compact land use at high densities, open space
and farmland at this location may be protected against ongoing urbanisa-
tion (De Roo and Miller 2000).
Because the aforementioned New Key Projects are often large, consist-
ing of multiple sub-projects and their realisation depends on numerous ac-
tors, they can be characterised as rather complex planning problems. The
multi-actor characteristic in particular takes the central position in this
chapter.
172 Ron Vreeker
Various methods exist to assess and evaluate aforementioned effects. A

well-known method, a form of applied welfare economics, is Social Cost-
Benefit Analysis (SCBA). By applying SCBA, one attempts to identify all
the costs and benefits of a project and value them in monetary terms. The
method also judges projects or alternatives in terms of their contribution to
social welfare.
A Cost-Benefit Analysis (CBA) has been conducted for the Amsterdam
South Axis project in which the effects of two alternative development op-
tions are evaluated (CPB 2003). CBA is discussed in detail in chapter 10;
the study concludes that both alternatives result in a welfare loss for Dutch
society. However, the results obtained from the CBA do not prevent the
most important actors to proceed with the project. We may thus conclude
that next to the sheer size of effects, other factors, for example the distribu-
tion of effects over actors, may also influence the outcomes of the deci-
sion-making process. In our case study example we sketch the positions
and attitudes of stakeholders involved in the South Axis project by means
of a multicriteria decision analysis (MCDA). In particular, we use MCDA
to describe the decision-problem from the perspectives of these different
actors involved and analyse which aspects they consider important. Based
on these identified perspectives we hope to identify groups of actors for
which interests apparently cluster.
In section 9.2 we discuss the characteristics of MCDA in deeper detail.
The history of the South Axis is handled in section 9.3 and we examine
how the area has transformed into its current situation. We also attend to
particular problems and opportunities attached to this district. Our analysis
focusses on the autonomous developments and various plans designed for
the integral development of the area. In section 9.4 we discuss the various
stakeholders involved in the project and offer a list of actors to be con-
sulted in a multi-actor multicriteria procedure. In section 9.5 we discuss
the decision criteria and in section 9.6 we present the results of the MCDA.
Conclusions follow.
9.2 Multicriteria decision analysis (MCDA) and the

inclusion of multiple actor perspectives
In the literature on evaluation techniques authors often distinguish between

monetary and non-monetary evaluation (Janssen and Munda 1999). In
monetary evaluations of projects attempts are made to measure all effects
of a project and express them in monetary units, whereas non-monetary
evaluation uses a wide variety of measurement units to express the effects
MDCA of a high speed railway station area development project 173
of a project. CBA belongs to the monetary category of techniques and

MCDA belongs to the non-monetary category.
MCDA is a family of methods rather than a specific appraisal method; it
consists of over 100 members, with sometimes differing theoretical foun-
dations (Nijkamp et al. 1990; Vincke 1992). In general, MCDA methods
are applied to assist decision-makers in selecting from a given set of alter-
natives. To assess the performance of alternatives, MCA techniques com-
bine a set of criteria with a set of weights reflecting the preferences of the
decision-maker.
Various classifications of MCA techniques do exist. Methods can be
classified according to the type of alternatives they address, their data han-
dling capabilities, or the type of decision problem they address (Roy
1996). MCDA techniques are mostly classified according to how they syn-
thesise information, such as criteria scores and weights, into an overall per-
formance score (Nijkamp et al. 1990; Vincke 1992). Three types of aggre-
gation procedures can be distinguished, namely iterative, complete and
partial aggregation. Iterative methods aim to identify an efficient alterna-
tive from a set of infinite alternatives. The evaluation of our case study en-
tails the ranking of a restricted number of discrete project alternatives
therefore the application of iterative methods is not suitable here.
With the application of MCDA methods based on complete aggregation,
we assume that the preference structure of a decision-maker (DM) can be
modelled by using only indifference (I) and preferential (P) relationships.
Both relationships are assumed to be transitive. Furthermore, it is assumed
that criteria used are independent, which is not always a realistic notion
(Vincke 1992). The application of this preference model (P, I) results in a
complete pre-order of alternatives. One of the main techniques belonging
to this school is the Multiple Attribute Utility Theory (MAUT), which has
been widely applied for economic and financial problems. MAUT is based
on utility theory, more specifically, on the expected utility theory of von
Neumann and Morgenstern (1947). The analysts role is to build a utility
function that represents the decision-makers (henceforth DM) preference
system as accurately as possible. We assume that the DM acts rational and
wants to maximise his utility function. Application of MAUT leads to a
well-defined mathematical problem in which the scores on every criterion
are used to calculate the overall utility of an alternative. The technique al-
lows that criteria scores compensate each other; the weights used by
MAUT can therefore be interpreted as trade-offs (Keeney and Raiffa
1976).
An important advantage of MAUT methods is that the aggregation pro-
cedure is fairly transparent and can be explained easily to the DM. How-
174 Ron Vreeker
ever, MAUT-methods also require intensive interaction with, which is

rather time-consuming.
Criticism of MAUT has led to the development of the so-called Multiple
Criteria Decision Aid and associated partial aggregation methods and out-
ranking methods. Within this school of thought it is not aimed to optimise
a mathematical model that fully reflects DMs preference structure, but
rather to learn about the decision-making situation and lead DM to an ac-
ceptable solution. The methods utilised model preference structures by
means of Preference (P), Indifference (I), Weak preference (Q), and In-
comparability (R) relationships. These relationships are considered to be
possibly intransitive.
Outranking methods typically use pair-wise comparisons to assess the
performance of the alternatives on each criterion and concordance and dis-
cordance indices to rank alternatives. We explain here the basics of con-
cordance analysis and consider a choice problem with a set of alternatives
k and a set of criteria l. After comparing alternative i with alternative j in
relation to all criteria, we select all criteria for which the former performs
better than, or equal to, the latter. We call this set of criteria the concor-
dance set. Similarly, we define the class of criteria for which alternative i
performs worse than, or equal to, alternative j. This set of criteria is called
the discordance set. In order to rank the alternatives, the concordance in-
dex is used, which is the sum of the weights related to the criteria for
which alternative i performs better than j. We call this sum Cij and one can
see that the higher the value of Cij, the greater is the dominance of alterna-
tive i over j.
Outranking methods usually calculate discordance indices to use the
cardinal information contained in the impact matrix (Nijkamp et al. 1990).
The discordance index indicates the maximum difference of scores for al-
ternatives i and j, on the criteria for which the latter performs better. The
information provided by the concordance and discordance indices can then
be used to obtain a ranking or selection of alternatives, and one seeks the
alternative with high values on the concordance index and low values on
the discordance index.
In reality, some differences between criteria scores are irrelevant or are
only taken into consideration by DM when they are of a certain magnitude.
Some outranking methods therefore use indifference and preference
thresholds to account for this aspect (Roy 1968, 1972; Brans et al. 1986).
Since outranking methods are based on partial or non-compensatory ag-
gregation procedures, the weights used by these methods should not be in-
terpreted as (marginal) rates of substitution; they reflect only the relative
importance of the criteria as assigned by DM.
Regime analysis is an outranking method applied in the case study pre-

sented here. The method is able to cope with both qualitative and quantita-
tive effect information (Nijkamp et al. 1990). Since the effects in our case
study are expressed on a variety of measurement scales, we have selected
this method (Hinloopen et al. 1982).
Although it is claimed that the MCDA approach is highly capable of in-
volving multiple actors in a decision-making process, this feature had only
a limited impact on the actual design of MCDA procedures (Bogetoft and
Pruzan 1991; Keeney 1992). Only recently have approaches emerged that
incorporate characteristics of the so-called group decision support systems
(GDSS) in MCDA procedures (see, for example, the SMCE approach).
Such approaches are developed to ensure that multiple views are present in
the decision process. This is accommodated by giving actors an influence
in the design and selection of evaluation criteria. Furthermore, actors are
consulted to elicit their preferences with regard to the weight given to the
selected criteria. Before we can actually involve actors in the decision-
making process, however, we have to identify them. In the next section we
describe the South Axis project, its history and the relevant actors.
9.3 History and current situation of the South Axis area
In the early 20th century the South Axis was a green belt separating two
districts in the southern part of the city of Amsterdam. Nowadays the area
is crossed by the southern branch of orbital motorway A10 as well as
tracks used by trains and light rail systems. Since the infrastructure divides
the area district into two parts, it is regarded as a major barrier. The area
has railway station South/WTC as its centre; as a main hub in the public
transport system, it links local, regional, and (inter) national public trans-
port systems.
During the 1990s the municipality of Amsterdam decided to attract
more foreign firms that offer financial, legal and business services. The fo-
cus here was on the development of a location for high-end offices that
competes with other European office locations for international business
activities. Initially it was aimed to develop the southern embankments of
River IJ as a preferred location. However, due to the availability of space
and good accessibility, real estate companies and other firms turned their
attention towards the South Axis area rather than the IJ-banks. To prevent
further uncoordinated and undesired developments in the South Axis area,
the municipality of Amsterdam designed an integral development plan
(DRO 1998), which aimed to create an attractive new urban centre offering
176 Ron Vreeker
a wide variety of high quality amenities such as museums, hotels, hospi-

tals, and shopping facilities.
The project area is depicted in Figure 9.1. The solid line in the figure in-
dicates where adjustments to the infrastructure will take place. Adjust-
ments include the expansion of motorway A10 and the current rail sys-
tems. Real estate development will be undertaken within the areas
demarcated by the dotted line. The municipality now strives for an even
presence of housing and offices in the area.
Figure 9.1. Project area Amsterdam South Axis
In order to fulfil the stated objectives, three alternatives were outlined

(DRO 2001) and are described here in subsequent sections. The considered
alternatives differ with regard to the positioning of the infrastructure bun-
dle and, as a consequence, in their urban development programs. Besides
the three alternatives mentioned above, we also include an Autonomous
development alternative. We have summarised the characteristics of the
considered alternatives in Table 9.1.
Table 9.1. Real estate development in the Autonomous development scenario

District Offices Dwellings Amenities Total
2 2 2
Mahler4 158 754 m 37 700 m 26 800 m 223 254 m2
Gershwin 45 000 m2 163 180 m2 34 100 m2 242 280 m2
Vivaldi 192 650 m2 88 000 m2 38 000 m2 318 650 m2
Kop South 50 000 m2 50 000 m2 50 000 m2 150 000 m2
Axis
Total 446 404 m2 338 880 m2 148 900 m2 934 189 m2
9.3.1 Autonomous development alternative
In the Autonomous development alternative the expansion of rail infra-

structure is limited to the extension of the existing platform for heavy rail
(2 platform tracks to a length of 340 metres), and the construction of an
additional platform with two platform tracks. Within this scenario a new
subway line is also being constructed, connecting Amsterdam Central Sta-
tion with South/WTC. In total, four platforms-tracks and four feeder-tracks
will be available for light rail in 2011. The capacity of the current orbital
motorway A10 (2x3 lanes) crossing the area will not be sufficient to ac-
commodate the foreseeable growth in road transport. To create additional
capacity, a new configuration of 2x4 lanes will be implemented; this will
be built at the expense of the current emergency lanes.
With regard to the development of real estate, approximately 930 000
2
m of floor space will be constructed on both sides of the current infra-
structure dike (DRO 1999). This urban development program will be con-
structed over the period 2004-2012. The South Axis area will remain
dominated by offices.
9.3.2 Dike alternative
Since further expansion of tracks and platforms at Amsterdam Central Sta-

tion is almost impossible, terminal South/WTC has been assigned to ac-
commodate the foreseen growth in rail transport. Furthermore, the devel-
opment of real estate at the South Axis is expected to lead to additional
demand for public transport. Together with the introduction of the high-
speed link, this calls for large investments in rail infrastructure at
South/WTC. In total, three platforms will become operational in 2020. Due
to the addition of rail tracks, the southern lanes of orbital motorway A10
will be repositioned and enlarged from a 2x3 to a 2x4 main lanes configu-
ration plus emergency lanes. In this alternative the infrastructure dike will
be widened to 170 metres and will accommodate the expanded infrastruc-
ture bundle.
The urban development program of approximately 1,5 million m2 of
floor space will be effectuated at both sides of the extended infrastructure
dike. This program includes the real estate developed in the Autonomous
development alternative. Table 9.2 depicts the urban development pro-
gram to be constructed in addition to the Autonomous development al-
ternative.
In the Dike alternative various small tunnels will be constructed to im-
prove the connection between the districts of Oud-Zuid and Buitenveldert.
178 Ron Vreeker
Railway station South/WTC will, however, form the main link between
both districts. All construction works are expected to be finished in the
year 2020.
Table 9.2. The additional real estate development in the Dike alternative
2 2 2
Noordzone 25 000 m 46 000 m 10 600 m 81 600 m2
Beethoven 30 000 m2 30 000 m2 30 000 m2 90 000 m2
VU-quarter 102 000 m2 93 000 m2 25 300 m2 220 300 m2
Ravel 120 000 m2 74 000 m2 7000 m2 201 000 m2
Total 298 000 m2 298 000 m2 69 900 m2 589 900 m2
9.3.3 Dock alternative
In the Dock alternative the infrastructure is to be expanded as in the Dike

alternative, but will be placed underground. The associated tunnels will be
1,2 kilometres long, and the design of the public transport terminal is
based on the integration of all modalities within one terminal building.
By building the infrastructure underground, an additional 930 000 m2
will become available for real estate development. The complete urban de-
velopment program constitutes approximately 2,4 million m2 of floor space
(see Table 9.3). This program will be realised by means of compact land
use at high densities, and the choice of the Dock model (expected to take
30 years to complete), is aimed to develop a new urban centre. The Dock
model is supported by a consortium of real estate developers, private banks
and the Dutch government.
Table 9.3. The additional real estate development in the Dock alternative
2 2 2
Noordzone 25 000 m 46 000 m 10 600 m 81 600 m2
Beethoven 30 000 m2 30 000 m2 30 000 m2 90 000 m2
VU-quarter 102 000 m2 93 000 m2 25 300 m2 220 300 m2
Ravel 120 000 m2 74 000 m2 7000 m2 201 000 m2
Composer 419 400 m2 419 400 m2 93 200 m2 932 000 m2
Total 641 400 m2 717 400 m2 163 100 m2 1.521 900 m2
9.3.4 Combination of Dike and Dock
The expansion plans for the infrastructure in this alternative do not differ
from the Dike and Dock alternatives. However, in the combination alterna-
tive, parts of the infrastructure will be placed at surface level (rail) and
parts will be subterranean (A10 and light rail), resulting in an infrastruc-
ture dike less wide (80 metres wide) than the Dike alternative. As in the
Dike alternative, various corridors will be established to improve the ac-
cessibility of both districts, and station South/WTC will form the main
connection. Real estate development will proceed on both sides of the
dike. Office development will take place on top of the tunnel for light rail.
The space above orbital motorway A10 will not be used for real estate de-
velopment. Some of these tunnels, especially on the eastside of the termi-
nal, will be transformed into recreational areas.
Due to the smaller infrastructure dike, more space will become available
for real estate development in the combination alternative compared to the
Dike alternative. However, since not every tunnel will be used to accom-
modate real estate, the urban development program is smaller than that of
the Dock alternative. The complete urban development program consti-
tutes approximately 2,21 million m2 of floor space. This program will be
realised by means of compact land use at high densities and includes the
real estate developed in the Autonomous development alternative. Table
9.4 depicts the urban development program to be constructed in addition to
the Autonomous development alternative.
Table 9.4. The additional real estate development in the Combination alternative
2 2 2
Noordzone 25 000 m 46 000 m 10 600 m 81 600 m2
Beethoven 30 000 m2 30 000 m2 30 000 m2 90 000 m2
VU-quarter 102 000 m2 93 000 m2 25 300 m2 220 300 m2
Ravel 120 000 m2 74 000 m2 7 000 m2 201 000 m2
Composer 293 580 m2 293 580 m2 65 240 m2 652 400 m2
Total 515 580 m2 591 580 m2 135 140 m2 1.242 300 m2
We have summarised the characteristics of the South Axis alternatives

identified in Table 9.5. Note that the effects resulting from these alterna-
tives are the subject of section 9.5 where the selected evaluation criteria
and associated scores are discussed.
180 Ron Vreeker
Table 9.5. Overview of South Axis alternatives

Business Dike Dock Combination
as usual alternative alternative alternative
Railway tracks
Placement Surface Surface Subterranean Surface
Feeder-tracks1 4 8 8 8
Platform-tracks 2 6 6 6
Terminal
Placement Surface Surface Subterranean Surface
Light rail
Placement Surface Surface Subterranean Subterranean
Platform-tracks 4 4 4 4
Feeder-tracks 4 4 4 4
Motorway A10
Placement Surface Surface Subterranean Subterranean
Main lanes 2x4 2x4 2x2 2x2
Parallel lanes - - 2x2 2x2
Emergency lanes - 2x1 2x1 2x1
Real estate
Offices 446 404 m2 668 404 m2 1.087 804 m2 961 987 m2
Amenities 148 900 m2 218 800 m2 312 000 m2 284 040 m2
Residential 338 880 m2 636 880 m2 1.056 280 m2 961 984 m2
Total m2 934 184 m2 1.524 084 m2 2.456 084 m2 2.176 484 m2
9.4 Stakeholders in the South Axis development
The proper identification of relevant actors is important to the successful

application of participative MCDA procedures. Various methods exist to
identify and classify the actors involved in a decision-making process.
These methods are often based on analysing the reasons why people get
involved in a decision-making process and the power they can exercise
(Mason and Mitroff 1981; Weiner and Brown 1986). The significant ques-
tion is whether the actor can influence the decision-making process or is
only affected by it (Martin 1985; Fottler et al. 1989; Savage et al. 1991;
Banville et al. 1998). Although the various methods are useful to sketch
the decision-making arena, the various actors, their interactions, and so on,
1 Feeder tracks will be placed at the western and eastern sides of the terminal.
one should note that these methods do not guarantee a complete list of
relevant stakeholders. In general, the identification, selection and inclusion
of stakeholders is based on common sense, experience and intuition of the
MCDA practitioner (Banville et al. 1998).
The South Axis project is divided into multiple subprojects and in order
to realise it, various private and public parties have been brought together.
Among these actors are developers, architects, investors, and (local) public
authorities. In this section we provide an overview of the actors in the
South Axis project. For an actor to be selected it needs to be influential in
the decision-making process. Actors will also be consulted in case they are
affected by the decision taken but they are not at all influential. In our de-
scription we distinguish between standard and silent actors (Banville et al.
1998). Standard stakeholders are affected by, and influential in, the deci-
sion-making process. Silent stakeholders are affected by the outcomes of
the process, but do not have the means to participate directly or indirectly;
their influence in the decision-making process is therefore limited.
9.4.1 Standard stakeholders in the South Axis project
Influential in the South Axis project are the prospective shareholders in the
public-private partnership South Axis ltd. Among these are the Dutch gov-
ernment, the municipality of Amsterdam, and various real estate property
developers/investors. South Axis ltd. will be established in 2007 and will
be responsible for the integral development and construction of the Dock
area which also includes provision of the infrastructure. Since the revenues
are insufficient to recover the costs of the dock, the municipality has de-
cided to partly subsidise the project by investing rent revenues accruing
from parcels outside the dock area (395 mln). In the event that another al-
ternative is selected, the municipality will not make this investment.
The municipality also has a coordinating role in the project which is as-
signed to Project Office South Axis. Furthermore, the planning department
(DRO), on the other hand, is responsible for the design of the complete ur-
ban structure and infrastructure. To fulfil these tasks, the planning depart-
ment (DRO) works in close cooperation with the Department of Infrastruc-
ture, Traffic and Transport of the municipality of Amsterdam (DIVV).
Various other governmental agencies besides the municipality of Am-
sterdam play a role in the South Axis project; some will subsidise the pro-
ject. the infrastructure components of the project are subsidised by the
Ministry of Transport, Public Works and Water Management. The size of
the subsidy is independent from the choice of an alternative. Two coordi-
nating agencies of this ministry, Directorate North-Holland and ProRail,
182 Ron Vreeker
will be responsible for the expansion of the road and the rail infrastructure.
Although these coordinating agencies have strong opinions about the
specifications of the infrastructure, representatives of these organisations
emphasise that no preference for a certain alternative exists. The subsidy
provided can be used by South Axis ltd. as long as the infrastructure ex-
pands and functions according to the specifications set by the coordination
organisations.
To ensure that the South Axis is of a high quality, the Ministry of Hous-
ing, Spatial Planning and the Environment (VROM), the Province of
North-Holland, and the City Region Amsterdam subsidise the Dock alter-
native.
9.4.2 Silent stakeholders
Although affected by the project, some actors are not involved in the deci-
sion process. Current residents and users (e.g. firms, employees and visi-
tors) for example, may experience noise nuisance and be inconvenienced
by the construction activities. The real estate developments could even
lead to the conversion of green spaces into office and residential areas. Not
all residents, users or visitors will necessarily appreciate this possibility.
Due to the developments, some parties may even have to relocate their ac-
tivities to other locations; such drastic consequences hit especially various
sport clubs that are now (or were previously) located in the area. Users of
the current infrastructure, such as the orbital motorway and the railway sta-
tion, may experience delays during the construction phase. The aforemen-
tioned actors may also benefit from the investments in the South Axis (e.g.
transport benefits, agglomeration economies, urban quality, etc.). For
practical reasons most of the silent stakeholders will be represented by
agents or lobby groups in a participative MCDA approach. In Table 9.6 we
give an overview of standard and silent stakeholders involved in the South
Axis project, as well as their representative agents.
In this section we have identified the various actors involved in the
South Axis project. In our research we address the perspectives of the mu-
nicipality of Amsterdam (planning and economic development depart-
ments), the possible shareholders of South Axis ltd, current residents, of-
fice users, and the Ministries of Finance and Housing, Spatial Planning,
and the Environment (VROM). Furthermore, we address the shareholders
of South Axis ltd as one group. In the MCDA procedure we evaluate the
considered alternatives from the perspective of those actors. In the next
section we analyse which actors deem which effects to be important.
Table 9.6. Standard and silent stakeholders involved in the South Axis project
Actor Description Representative
Standard stakeholders
Dutch government Shareholder South Axis ltd. Ministry of Finance
Dutch government Provides subsidy Ministry of Housing, Spa-
tial Planning, and the Envi-
ronment (VROM)
Dutch government Supervisor rail project Ministry of Transport,
Public Works and Water
Management (ProRail)
Dutch government Supervisor A10 project Ministry of Transport,
Public Works and Water
Management (Directorate
North Holland)
ABN AMRO Shareholder South Axis ltd.
ING Real Estate Shareholder South Axis ltd.
Fortis Bank Shareholder South Axis ltd.
BNG Shareholder South Axis ltd.
Municipality Amsterdam Shareholder South Axis ltd. Economic Development
Department
Municipality Amsterdam Responsible for spatial de- Planning Department
sign (DRO)
Silent stakeholders
Residents and environ- Resident associations and
mental protection groups environmental protection
groups
Current real estate users Selection of real estate us-
ers
Future real estate users Developers offices
Developers dwellings
Public transport users Association of public
transport users
Visitors
9.5 Decision criteria and operational attributes
Although a set of evaluation criteria has to comply with certain require-

ments (Keeney and Raiffa 1976; Bouyssou 1990), no formal procedure yet
exists for the selection of criteria. In a participative MCDA procedure the
selection of evaluation criteria takes place when all relevant actors have
been identified and consulted to ensure that the criteria reflect their points
of view on the decision-problem.
184 Ron Vreeker
In this research we have constructed a long-list of criteria from which

the actors can eventually select the relevant ones to reflect their concerns.
The long-list is constructed by analysing various policy documents in
which the main objectives of the project and the relevant legislation are
stated. Various effects studies have also been analysed to obtain additional
criteria and the scores (e.g. CBA and interim report Environmental Impact
Assessment). We have moreover conducted interviews with relevant
stakeholders and invited them to check that the list is comprehensive or to
provide additional criteria as they see fit. We have categorised the decision
criteria under various themes and enumerated the corresponding indicators
in Table 9.7.
Furthermore, in this table the values of each alternative score on the in-
dicators are also given, as well as the units in which these scores are meas-
ured. From this table it can also be read whether an indicator can be la-
beled as a benefit or cost indicator. For a benefit indicator it holds that the
higher the indicator score, the better; the opposite applies for a cost indica-
tor. When we look closely at Table 9.7 the following observations can be
made. We see that the economic and environmental indicators are meas-
ured on ratio-scales leading to quantitative indicator scores. This is due to
the fact that it is mandatory in the Netherlands to conduct formal economic
and environmental effect studies (e.g. CBA and Environmental Impact As-
sessment) before an investment decision can be made. Some effect themes
only contain indicators measured on qualitative measurement scales. These
indicators and associated scores stem from an explorative effect study, in
which the alternatives were assessed on the themes nuisance, urban qual-
ity, and terminal and rail operations.
It is noteworthy that the included indicators solely measuring effects oc-
curring in the project area are in the adjacent region (e.g. city-region Am-
sterdam). This means that, with respect to the assessment of effects, the na-
tional scale is overlooked. So-called network effects (e.g. transport and
environment) are therefore not included. As a consequence, the evaluation
of South Axis alternatives by means of MCDA and this indicator set will
not lead to the same type of evaluation as a CBA.
The indicators included in the long-list are not relevant for all actors.
We have therefore conducted interviews with stakeholders and asked each
respondent to rate the selected criteria. Respondents were asked to assign a
value ranging from 1 to 5 to a criterion, where 1 means not important in
the choice between the alternatives and 5 very important. In Table 9.8 the
results per actor are given. Residents, for example, will be confronted with
noise and traffic nuisance during the long construction period and there-
fore assign high value to these criteria. Not surprisingly, private investors
in South Axis ltd. focus on the total costs and revenues of the dock con-
186 Ron Vreeker
struction. Since full recovery of development costs is problematic, the sub-

sidies given to the South Axis ltd. form a decisive element for these actors
in the decision process. With regard to the subsidy indicators, we note that
this may constitute a benefit indicator for one actor (South Axis ltd.) and a
cost for another (e.g. municipality, Ministry of Transport, etc.). Govern-
ment authorities consider the broadest ranges of effects. The planning de-
partment, for example, not only takes safety and environmental effects into
consideration, but is also responsible for the design and functioning of
public transport systems in the area.
9.6 Results of the Regime analysis
After having consulted the various actors about their preferences regarding
the different indicators, we use the results to construct so-called weight
vectors to reflect the perspectives of respondents. From the analysis it be-
comes clear that some actors have to choose between limited construction
nuisance occurring over the short-term and benefits accruing over the
longer term (see Table 9.9). This especially applies to residents, employees
and current real estate users at the South Axis. If equal weight is placed on
the indicators, these actors put the business-as-usual alternative at the
highest position in the ranking, which results in a choice for limited nui-
sance. Where emphasis is put on benefits, the Dock alternative takes first
choice position. For residents, these long-term benefits of the Dock alter-
native take the form of increased housing prices, the provision of urban
green and a reduction of the barrier effect.
Table 9.9. Ranking from the perspectives of the residents and the current real es-
tate users
Residents Current Real Estate Users
Rank Alternative Alternative
1 Autonomous Dock
2 Dock Combi
3 Combi Autonomous
4 Dike Dike
The same reasoning also applies to employees currently working at the

South Axis. These employees have to trade-off between the nuisance they
will have to endure (e.g. congestion, noise, etc.) against the increased
number of jobs occurring over time. Current office users place the Dock
alternative at the first position in the event that more weight is put on profit
188 Ron Vreeker
or employment indicators. Reduced accessibility to the area during the

construction phase may deter individuals such that many will decide to
vote against this alternative.
The governmental authorities included here all prefer the development
of the Dock alternative (see Tables 9.10 and 9.11). By investing 345 mil-
lion in the Dock alternative, the Amsterdam municipality realises various
benefits which are partly financed by other government agencies. The city-
region Amsterdam especially benefits from an increased number of jobs
due to the construction and use of real estate at the South Axis. Various
economic impact studies have shown that these jobs imply the relocation
of employment from outside the Amsterdam region towards the South
Axis (Rienstra et al. 1996; van der Mei et al. 1999; CPB 2003). This means
that on a national scale no additional jobs will result from the project alter-
natives. Whether this investment (345 mln) would result in higher bene-
fits by investing in other projects is not under debate. Here a choice is
made between the selected alternatives.
Table 9.10. Ranking from the perspectives of investors and the municipality of
Amsterdam
Investors Municipality Economic Municipality Planning De-
South Axis ltd. Department partment
Rank Alternative Alternative Alternative
1 Dock Dock Dock
2 Combi Combi Autonomous
3 Dike Dike Combi
4 Autonomous Autonomous Dike
Table 9.11. Ranking from the perspectives of the Ministry of Finance, Ministry
of VROM
Ministry of Finance Ministry of VROM
Rank Alternative Alternative
1 Dock Dock
2 Combi Combi
3 Dike Autonomous
4 Autonomous Dike
The investors in South Axis ltd. also prefer the Dock alternative above
the other alternatives. Although the construction costs of the Dock alterna-
tive are significantly higher than that of the other alternatives, the investors
still set the Dock alternative at the highest position in the ranking. This is
because the associated subsidies (955 mln) are just sufficient to overcome
the differences in costs (2,5 bln) and revenues (1,6 bln).
In sum, the standard stakeholders of the South Axis project prefer the
development of the Dock alternative. For silent stakeholders the Autono-
mous and Dock alternatives are the only development options. Where em-
phasis is placed on long-term benefits, silent stakeholders prefer the devel-
opment of the South Axis according to the Dock alternative.
9.7 Conclusions
In this chapter we have applied the MCDA technique Regime analysis to

the South Axis case study. We identified the various actors present in the
project and evaluated the various development options according to their
perspectives. The analysis shows that none of the standard stakeholders
opposes the Dock alternative. In order to convince silent stakeholders to
choose the Dock alternative, the (long-lasting) nuisance effects of the
building process should be mitigated. The analysis also shows that MCDA
techniques are useful in sketching the decision-making arena, and their re-
sults provide insights into the possible coalitions and conflicts among ac-
tors.
References
Banville C, Landry M, Martel JM, Boulaire C (1998) A Stakeholder Approach to

MCDA. Systems Research and Behavioral Science 15:1532
Bogetoft P, Pruzan P (1991) Planning with Multiple Criteria. North-Holland, Am-
sterdam
Bouyssou D (1990) Building Criteria: A Prerequisite for MCDA. In: Bana Costa
CA (ed) Readings in Multiple Criteria Decision Aid. Springer-Verlag, Berlin
Brans JP, Vincke P, Mareschal B (1986) How to Select and How to Rank Projects:
The Promethee Method. European Journal of Operational Research 24:228
238
CPB (2003) Kengetallen Kosten-Baten Analyse Project Zuid-As Amsterdam.
Centraal Planbureau, Den Haag
De Roo G, Miller D (2000) Compact Cities and Sustainable Urban Development.
Ashgate Publishing, Aldershot
DRO (1998) Masterplan Zuidas. Gemeente Amsterdam: Dienst Ruimtelijke Orde-
ning, Amsterdam
DRO (1999) Concept Visie Amsterdam Zuidas. Gemeente Amsterdam: Dienst
Ruimtelijke Ordening, Amsterdam
DRO (2001) Integrale Trace/M.E.R. studie Zuidas; Stedelijke Kwaliteit van de
Drie Alternatieven: Dijk, Kunstwerk en Dok. Gemeente Amsterdam: Dienst
Ruimtelijke Ordening, Amsterdam
190 Ron Vreeker
Fottler MD, Blair JD, Whitehead CJ, Laus MD, Savage GT (1989) Assessing Key
Stakeholders: Who Matters to Hospitals and Why? Hospital and Health Ser-
vices Administration 34:525546
Hinloopen E, Nijkamp P, Rietveld P (1982) The Regime Method: A New Multic-
riteria Method. In: Hansen P (ed) Essays and Surveys on Multiple Criteria
Decision Making. Springer, Berlin
Janssen R, Munda G (1999) Multicriteria Methods for Quantitative, Qualitative
and Fuzzy Evaluation Problems. In: Bergh JCJM van den (ed) Handbook of
Environmental and Resource Economics. Edgar Elgar, Aldershot
Keeney RL, Raiffa H (1976) Decision with Multiple Objectives: Preferences and
Value Trade-Offs. John Wiley & Sons, New York
Keeney RL (1992) Value-Focussed Thinking: A Path to Creative Decision-
Making. Harvard University Press, Boston MA
Martin AP (1985) The First Order of Business: Testing the Validity of the Objec-
tives. Professional Development Institute, Ottawa
Mason RO, Mitroff II (1981) Challenging Strategic Planning Assumptions: The-
ory, Cases and Techniques. John Wiley & Sons, New York
Mei JP van der, Ter Beek HM, Hers JFP, Nagengast E, Lambooy J (1999) Eco-
nomische Effecten Zuidas Amsterdam. Ernst & Young and Regioplan,
Utrecht
Neumann J von, Morgenstern O (1947) Theory of Games and Economic Behav-
ior. Princeton University Press, NJ
Newman P (1992) The Compact City: An Australian Perspective. Built Environ-
ment 18:285300
Newman P, Kenworthy J (1989a) Cities and Automobile Dependence: A Source-
book. Aldershot and Brookfield, Victoria
Newman P, Kenworthy J (1989b) Gasoline Consumption and Cities: A Compari-
son of US Cities with a Global Survey. Journal of the American Planning As-
sociation 55:2437
Nijkamp P, Rietveld P, Voogd H (1990) Multicriteria Analysis for Physical Plan-
ning. Elsevier, Amsterdam
Rienstra SA, Rietveld P, Bruinsma FR, Gorter C (1996) Regionaal-Economische
Effecten van de Ontwikkeling van de Amsterdamse Zuidas als Kantoorloka-
tie. Vrije Universiteit, Amsterdam
Roy B (1968) Classement et Choix en Presence de Points de Vue Multiple (La
Methode Electre). R.I.R.O. 2:5775
Roy B (1972) Decision avec Criteres Multiple. Metra 11:121151
Roy B (1996) Multicriteria Methodology for Decision Aiding. Kluwer Academic
Publishers, Dordrecht
Savage GT, Nix TW, Whitehead CJ, Blair JD (1991) Strategies for Assessing and
Managing Organizational Stakeholders. Academy of Management Executive
5:6175
Vincke P (1992) Multicriteria Decision-Aid. John Wiley & Sons, New-York
Weiner E, Brown A (1986) Stakeholder Analysis for Effective Issues Manage-
ment. Planning Review 14/5:2731
10 Cost-benefit analysis of railway station area
development: The case of Amsterdam
South Axis
Carel Eijgenraam and Ioulia Ossokina
CPB Netherlands Bureau for Economic Policy Analysis, The Hague,

the Netherlands
10.1 Introduction
Development of railway station areas has for several years been an impor-
tant issue on the Dutch spatial policy agenda. National and local govern-
ments have been financially supporting the development of the areas
around the stations of the new High-Speed Rail, the railway connection
from Amsterdam to France and Germany. The projects in question are de-
signed to create a multifunctional land use environment in which synergy
effects of the combination of the transport nodes with business and resi-
dential land uses can arise.1 This chapter evaluates the welfare effects of
probably the most ambitious of these projects, that which involves urban
construction and infrastructural investment in the Amsterdam South Axis
area. In this project the transport infrastructure on both sides of the railway
station Amsterdam South/WTC is to be brought underground, thus creat-
ing extra space for high quality urban construction. The expectation is that
this change in the pattern of land use in the South Axis will have positive
effects on the utility of residents and productivity of companies in the area,
and subsequently on the value of real estate there.2
This chapter is related to the literature on the effects of combining dif-
ferent land use functions. The literature suggests, for example, that prox-
1 See for a discussion of the multifunctional land use concept e.g. Vreeker et al.
2004; Rodenburg and Nijkamp 2004.
2 This paper builds on a study performed by CPB Netherlands Bureau for Eco-
nomic Policy Analysis (Eijgenraam and Ossokina 2006).

192 Carel Eijgenraam and Ioulia Ossokina
imity of open space and accessibility of transport infrastructure both posi-

tively affect the prices of residential and commercial property, while prox-
imity of industrial land use has a negative effect (see inter alia Cheshire
and Sheppard 1995; Irwin and Bockstael 2001; and Debrezion and Wil-
ligers in chapter 13 of this book). Other studies examine the effect of a
change in the pattern of land use in the neighbourhood, such as increased
diversity and fragmentation of land use on the property price (e.g. Geog-
hegan et al. 1997; Song and Knaap 2004). However, relatively little re-
search has been performed on the costs of multifunctional land use, such as
investment costs involved in combining different land use functions in the
same restricted area.3 In this chapter we therefore compare for a specific
project the benefits of multifunctional land use with the costs that their
creation may involve. Such an exercise may provide important information
on the welfare effects of combining different land uses, which is a relevant
issue in the development of railway station areas.4
We evaluate the welfare effects of the South Axis multifunctional land
use project using the technique of (partial) cost-benefit analysis (CBA).
This technique is often applied in studies of the welfare effects of large in-
vestment projects on a national scale and has for the Netherlands been ex-
tensively discussed in a guideline by Eijgenraam et al. (2000). The guide-
line does not, however, pay explicit attention to the treatment of the
welfare benefits from land issue for urban construction purposes and the
(external) effects of a change in the pattern of land uses.5 These two as-
pects play an important role in the present study. We suggest an approach
for how they can be accounted for in the cost-benefit analysis.
This chapter is organised as follows. Section 10.2 describes the method-
ology of defining a multifunctional land use project for the purposes of the
cost-benefit analysis. Section 10.3 introduces the major direct effects of
the project: the financial land yields and the external effects of the change
3 Coupland (1997) discusses the possible costs of multifunctional land use. Bes-
seling et al. (2003) show for an earlier version of the South Axis project that the
costs of bringing the transport infrastructure underground can be quite high in
comparison with the benefits.
4 Several existing studies of the South Axis project base their evaluations of it
largely on the fact that considerable benefits can be expected from the unique lo-
cation of the South Axis area and the (external) effects of multifunctional land use
(see e.g. City of Amsterdam 2005). This paper presents a quantitative estimation
of the most important benefits and costs of the project, thus providing policymak-
ers with additional information which can be used in decision-making concerning
the South Axis project.
5 These two effects receive little attention in the published foreign studies that
use cost-benefit analysis (see, for example, Layard and Glaister 1994).
CBA of railway station area development 193
in the pattern of land uses in the area, and discusses some general meth-
odological issues. Section 10.4 is devoted to the estimation of the financial
net benefits of land issue. We find that combining underground infrastruc-
ture and urban construction above ground in the considered specification
of the South Axis project is too costly in comparison with the financial
benefits it yields. An important rationale to exist for the South Axis multi-
functional land use project should thus be sought in its external effects. Ex-
ternal effects of multifunctional land use in South Axis are discussed in
Section 10.5. We find these effects to be insufficient to compensate for the
financial deficit of the project, but stress at the same time the large uncer-
tainty margin surrounding the estimation. Section 10.6 presents the overall
balance of costs and benefits and performs a sensitivity analysis of the out-
comes to the assumptions used. Section 10.7 summarises some methodo-
logical lessons which can be learned from this cost-benefit analysis and
section 10.8 concludes.
10.2 Multifunctional land use in the South Axis
10.2.1 Development of the South Axis area
The development of the South Axis area is a complicated project. Its most
ambitious version (the Dock) involves: i.) extending and bringing the
transport infrastructure underground in the neighbourhood of the railway
station Amsterdam South/WTC, and ii.) developing the vacant land above
the tunnel (on the Dock) in the direct proximity of the tunnel. Tables 10.1
and 10.2 show the ambition of the project in comparison with the situation
during 2006-2007. The expectation is that the construction program will be
completed in 2028.
Table 10.1. Real estate in South Axis: in 2006 and after the realisation of the
Dock project
2006 Dock project: additional
construction volume in
comparison with 2006
Offices (m2 gross floor area) 594 000 +752 000
Residential property (m2 gross floor area) 52 000 +1.022 000
Services (m2 gross floor area) 654 000 +357.000
Total 1.300 000 +2.131 000
Parking lots (units) 11 000 +17 000
Table 10.2. Transport infrastructure in South Axis: in 2007 and after the realisa-
tion of the Dock
2007 Dock-project
Railway tracks at the South/WTC station 4 6
Railway tracks east and west of the station 2 4
Motorway A10 lanes 2x4 2x5
NS railway station Amsterdam South/WTC 2 narrow platforms 3 larger platforms
Metro no North-South line,
subway
10.2.2 Defining the multifunctional land use project at South

Axis
The ambition of Dock is to create a second city-centre of Amsterdam in

the South Axis, which is an area comparable to La Defense in Paris, Ca-
nary Wharf in London or das Bankenviertel in Frankfurt. Expected in-
crease in the quality of the South Axis area is an important motive behind
the tunnelling part of the project. Placing the transport infrastructure below
ground removes a physical barrier between the Amsterdam districts South
and Buitenveldert and diminishes the nuisance currently caused by trans-
port infrastructure in the area. At the same time new land is created where
commercial and residential property can be built.
For the purposes of this CBA, we define the multifunctional land use in
the South Axis as a project involving only the tunnelling of transport infra-
structure and urban construction above the tunnel. To evaluate the costs
and benefits of multifunctional land use we need to specify a reference al-
ternative, i.e. give a description of how the area would look without tun-
nelling. Research shows that the same extension of transport infrastructure
is also possible above ground. The vacant land available for construction is
in the reference alternative, however, less extensive. Table 10.3 describes
the characteristics of the real estate construction program in the reference
alternative and in the Dock project.
By defining the reference alternative in this way, we split the South
Axis project in two parts: the extension of the transport infrastructure on
the one hand, and the tunnelling and real estate development on the other.
In this chapter we concentrate on the cost-benefit analysis of the second
part only.
Table 10.3. Construction program of the Dock project in comparison to the refer-
ence alternative
Reference alternative: South Axis Dock: addi-
(real estate in 2006 + real tional construction vol-
estate that can be realised ume in comparison with
without tunnelling the the reference alternative
transport infrastructure)
Offices (m2 gross floor area) 594 000 + 280 000 +472 000
Residential property (m2 gross 52 000 + 313 000 +709 000
floor area)
Services (m2 gross floor area) 654 000 + 183 000 +174 000
Total 1.300 000 + 776 000 +1.355 000
Parking lots (units) 11 000 + 6 000 +11 000
10.3 Methodological issues
10.3.1 Direct and indirect effects
In a cost-benefit analysis (CBA) the effects of the project are usually sepa-
rated into direct effects and indirect effects. In this study we define direct,
or primary, effects as: i.) effects that accrue to the owner or users of the
project, and ii.) externalities that arise from the realisation, existence, or
use of the project. Direct effects are thus connected to the markets on
which intended impacts of the project take place. In the case of tunnelling
only the land market is impacted.6
Realisation of the project can, however, result in secondary or tertiary
effects on other markets as well. When relevant markets are characterised
by perfect competition, these so-called indirect effects only result in a re-
distribution of direct effects. The possibility of additional welfare changes
exists only when the project impacts imperfect markets. Recent research
based on spatial general equilibrium models suggests that the indirect ef-
fects can yield a net welfare benefit, which is however, small compared to
the direct effects (Brcker 2003).
6 As the transport infrastructure in the project alternative is the same as in the

reference alternative, the transport market is hardly affected.
Direct effects of the tunnelling
In terms of demand and supply on the market for urban construction land,
the South Axis project has two consequences; their combination deter-
mines the new market price of land at the location. First, the supply curve
of the land available for construction purposes moves to the right (the vol-
ume effect). This happens because tunnelling of transport infrastructure
makes extra space available for construction. Second, the demand curve
for land available for real estate development at the location moves up-
wards (quality effect). The reason is that the change in the pattern of land
use realised in the South Axis project can be expected to have positive ef-
fects on the productivity of companies and utility of residents in the South
Axis. As a consequence, the willingness to pay for land on-the-spot in-
creases.7
The factors discussed above determine the main direct effects of the
Dock project: the net internal benefits that accrue to the developer of the
Dock, and the external effects of the Dock on the value of real estate in the
neighbourhood of the project. These effects will be discussed in-depth in
sections 10.4 and 10.5 below.
Indirect effects
One of the frequently discussed positive indirect effects of large projects is

their influence on the labour market and employment. In the case of the
South Axis, this is a long-run effect, since the time horizon of the project
amounts to more than 20 years. In the long-run it is usual to assume that
the national economy converges to the general market equilibrium in
which the structural employment and unemployment are determined by
macroeconomic factors and labour market institutions. In this framework
land development projects have scarcely any influence on the employment
in the Netherlands. While the employment at South Axis will rise, in equi-
librium this increase will be compensated by a fall in employment else-
where in the country. This results in a net effect on national employment
that can be neglected.
Regional relocation is another often mentioned positive indirect project
effect. As a result of the intensive urban construction in the Dock, else-
7 We assume that the effect of the change in the pattern of land use in the South
Axis is completely reflected in the increased value of land in the area. We there-
fore consider it among the direct effects. (The effect can also manifest as a higher
remuneration of employees or higher profits of companies at the location, in
which case it should be seen as an indirect effect. The issue of the distribution of
the benefits of the project is, however, outside the scope of our analysis.)
where in the Netherlands land will stay available for other uses (such as
agriculture, green areas, and so on). For the purposes of this study we as-
sume that: i.) the business activity and inhabitants that move to the South
Axis if the project is implemented are in the reference alternative spread
across different existing locations, so that their influence on the local land
market can be seen as marginal, and ii.) the regional zoning policy at the
locations in question reflects social preferences. This implies that the
choice of companies and people to move to the South Axis does not lead to
any social costs or benefits at other locations.8
The project can also have negative indirect effects. One example is the
excess burden of extra taxation, which arises when a project yields a finan-
cial deficit (as is the case with the Dock) and public finances are used to
bridge this financial deficit. A possible way to account for this effect is to
include in the balance of costs and benefits a negative premium equal to a
certain fraction of the government subsidy. Another negative indirect ef-
fect that can occur in the Dock project is the leaking of some benefits
abroad. This can especially be the case if foreign companies are attracted
to the South Axis. In this (partial) cost-benefit analysis we consider only
direct effects of the project and do not address its indirect effects.
10.3.2 Other methodological issues
Risk valuation
Land benefits are prone to macroeconomic risks risks that cannot be di-
versified as they concern developments that influence the whole econ-
omy. In times of high economic growth the value of land and real estate
grows much faster than during periods of low economic growth. Investors
are only ready to invest in real estate when they receive compensation for
bearing this uncertainty in the form of a risk-premium. Thus they require a
higher return on real estate than on government bonds. In this cost-benefit
analysis we account for this effect by using a risk-premium on the discount
rate for the benefits.
8 This is an application of the general assumption always made in a partial

CBA, namely that the marginal alternative projects returns are exactly equal to
the discount rate used for the project under consideration.
Discount rates used
In this CBA we derive the general required return on investments in real

estate from the realisation of the returns on real estate as documented in
the Dutch ROZ/IPD index.9 For effects that are prone to macroeconomic
risks, we use a real (i.e. inflation-free) discount rate of 5%, which consists
of a 2% risk-free component and a 3% risk-premium.10
The risk-free discount rate of 2% is applied to all costs and benefits of
the project up to the start of construction of the Dock. For the years after
the construction has started, we apply the risk-free discount rate to the
costs, and the discount rate including the risk premium to the benefits.
Time horizon and present value
In the South Axis project both the costs and the benefits are expected to be
spread over a rather long time period. For this reason we use an infinite
time horizon in this cost-benefit analysis. The economic life of the real es-
tate is assumed to be 50 years; we assume the land to be used for construc-
tion again thereafter. The net present value of the cash flows of the project
is calculated in 2006 prices. Table 10.4 summarises the methodological as-
sumptions discussed thus far.
Table 10.4. Parameters used in the net present value calculations

Real risk-free discount rate (all costs and benefits) 2% per year
Risk-premium for macro-economic risks (benefits) 3% per year
Time horizon Infinite
Price level 2006
10.4 Balance of costs and benefits for the developer
10.4.1 Land benefits
In this section we estimate the benefits accruing to the owner and users of
the Dock project. A company yet to be founded, the so-called South Axis
enterprise, will be responsible for the realisation of the tunnel and the issue
of land. The expected income of the company will consist of the land rents.
9 www.rozindex.nl
10 In section 10.6 we provide a sensitivity analysis of the cost-benefit balance to
the assumption about the discount rate and the risk-premium.
These land rents can be estimated as a residual of the expected value of the
real estate to be constructed, and the costs of construction and mainte-
nance.11 This approach builds on the differential theory of rent by Ricardo
(1817) and is currently used by municipalities to determine the rents for
newly-issued land. We assume that the South Axis enterprise will be able
to cream off all the extra rents.
To apply this method we need to estimate the revenues of the real estate
development and the costs of preparing the land for building. This requires
assumptions to be made concerning the developments in the land market.
The most important parameters in this estimation are the pace with which
the land is issued, and the real estate prices that will be realised in the fu-
ture. Both factors are not known with certainty in advance. In this CBA we
adopt the assumptions made in the Business Case Study for the future
South Axis enterprise (Deloitte 2006).
Table 10.5 gives an overview of the assumptions used to calculate the
land benefits. The scope of the real estate program and the rent and sale
prices have been taken from Deloitte (2006). Land rents presented in the
table have been calculated using a model for constructing and operating
the real estate.
Table 10.5. Overview of the land rents used in the cost-benefit analysis (in 2006
prices)
Land rents per m2 Underlying rent/ Time period of land
gross floor area sale price per m2 issue
Office (rent) 1350 300 2010-2028
Dwelling (sale) 850 3145 2010-2028
Shop (rent) 3338 375 2017-2019
Commercial ser- 876 175 2009-2028
vices (rent)
Non-commercial 145 - 2009-2028
services (rent)*
Social housing* 69 - 2011-2016
Parking lot (sale)* 10 000 - 2020-2027
* For these types of real estate land rents were not calculated, but taken from
Deloitte (2006)
The rent/sale prices reported in Table 10.5 reflect the fact that the real
estate is situated in a prime location in the Dutch real estate market, char-
acterised by very good local, national and international transport connec-
11Normal return to the capital is incorporated in the costs of construction and

maintenance.
tions,12 proximity of social and cultural activities of Amsterdam, and the

additional location quality achieved as a result of removing the infrastruc-
tural construction from the ground level. A study of the market develop-
ments on the Dutch market for offices and residential property (Eijgen-
raam and Ossokina 2006 pp. 30-55) suggests, however, that use of the
aforementioned parameters may lead to an overestimation of the land
benefits. While the projected pace of the land issue and the expected price
level in the South Axis area fit with the market conditions and historical
developments during the last decade of the 20th century, this does not nec-
essarily guarantee that they can be seen as a realistic forecast for the fu-
ture. In section 10.6 we will discuss the uncertainties surrounding the
yields from the issue of land for construction purposes.
Table 10.6 reports the ex ante balance of costs and benefits of the Dock
for the developer. The cost figures have been provided by Arcadis (2006).
Table 10.6. Financial costs and benefits of multifunctional land use in South Axis
(in 2006 prices)
Category Period of time Total expendi- Present value in
tures ( mln) 2006 ( mln)
Costs
Tunnelling 2009-2012/17/27 1430 1220
Costs for preparing land for 2006-2019 260 240
building
Maintenance starting with 2012 40
Total costs 1690 1500
Benefits
Land issue dwellings on Dock 2010-2028 290
Land issue offices on Dock 2010-2028 320
Land issue other real estate 2009-2028 410
Total benefits land issue 1020
Dock
Balance of benefits and costs -480
Table 10.6 shows that the balance of financial costs and benefits of the
multifunctional land use Dock project has a deficit of almost 0.5 billion.
This implies that the land yields are insufficient to compensate for the
12For instance, Amsterdam Schiphol international airport can be reached from

South Axis with the train in less than 10 minutes.
quite substantial costs of tunnelling the transport infrastructure.13 The re-

alisation of the multifunctional land use Dock project therefore must
largely be justified by its external effects. These effects are discussed in
the next section.
10.5 External effects
10.5.1 Location quality effect
The realisation of the Dock leads to a considerable change in the pattern of

land use in the South Axis railway station neighbourhood. The territory
currently occupied with transport infrastructure construction will in the
Dock project become available for large-scale urban construction. The new
urban quarter to be built above the tunnel foresees housing for approxi-
mately 5400 households, office space for almost 16 000 employees, and a
large-scale development of commercial services. Realisation of such an
urban environment on the place where rail and roads were previously situ-
ated can be expected to have important external effects on the utility of
residents and the productivity of companies located in the proximity of the
tunnel. We next discuss the possible foundations of these external location
quality effects.
Residential estate
Ex post studies of the housing market14 suggest that the characteristics of

land use in a neighbourhood can have a non-negligible influence on the
utility of residents of this neighbourhood and thus on house prices there.
Debrezion et al. (2006) show for the Netherlands that houses located
within a 250 metre distance from the railway, cost ceteris paribus 5% less
than houses situated farther than 500 metres away.15 This utility effect of
13 Financially, this deficit will be bridged by a transfer of the City of Amster-

dam to the South Axis enterprise of the building lots in the direct proximity of the
Dock area. The rationale for this transfer should be the size of the external effects
from Dock that accrue to the City or its inhabitants.
14 Studies discussed in this and further paragraphs are revealed preference stud-
ies using the hedonic price method.

15 Similarly, Cheshire and Sheppard (1995) and Rouwendal and van der
Straaten (2006) find that proximity of industrial land use can on the margin have a
negative effect equal to some percentage of the house price.
railway proximity seems to an important degree to be caused by noise nui-

sance and air contamination: one extra decibel of noise has been shown to
lead to a decrease of up to 2% in the house price (see e.g. Lijesen et al.
2006 and references). Also relevant are studies on possible effects of
commercial land use and land use diversity on the housing prices; these
find, however, less pronounced positive and sometimes ambiguous effects
(Song and Knaap 2004; Geoghegan et al. 1997). Finally, negative utility
effects due to loss of open space and an increase in population density can
take place (e.g. Cheshire and Sheppard 1995; Irwin and Bockstael 2001).
On the basis of the discussed studies, one can expect a positive utility
effect of the Dock-induced decrease in noise nuisance and air contamina-
tion, as well as of the Dock-induced increase in the accessibility of ser-
vices and possibly the diversity of land use. At the same time, the increase
in population density or the deterioration of the view outside the window
(skyscrapers instead of green sport fields) may have a negative utility ef-
fect. In a properly working housing market these location quality effects
can be expected to be revealed in house prices. As for the possible size of
these effects, the above discussed studies offer a range of possible out-
comes, most of which lie below 5% of the house price. Different studies
stress at the same time that the size of the effects of neighbourhood charac-
teristics on house price is very location-specific.16
Offices
The pattern of land use in the neighbourhood can also have effects on the
productivity of companies located there. Ciccone and Hall (1996) show
that a higher density of employment in an area can lead to a higher level of
productivity. A considerable increase in employment density in South Axis
as a result of the realisation of the Dock can thus via agglomeration
economies such as knowledge spillovers lead to an increase in firm pro-
ductivity at South Axis; its increased attractiveness as a work location for
employees can cause another productivity effect of the Dock. De Graaff
and Rodenburg (chapter 11 of this book) suggest that employees in South
Axis have a positive willingness to pay for the increased accessibility to
shopping facilities in the proximity of their offices. A Dock-induced in-
16 Cheshire and Sheppard (1995) show that in places characterised by relative
scarcity of open space, the influence of open space proximity on house prices is
relatively high. Rouwendal and Van der Straaten (2006) find that, in cities with a
relatively high concentration of industrial activities, the valuation by the housing
market of a decrease of the fraction of industrial land use in the neighbourhood is
relatively high.
crease in the accessibility of these facilities may thus facilitate (and there-
fore to some degree make it less expensive) for firms to attract highly pro-
ductive employees. The above mentioned positive productivity effects of
the Dock can be counteracted somewhat by negative effects, such as, for
example, the negative utility of firms which were first located in sight of
the highway, and used their buildings to attract travellers attention to the
company located there.
In a properly functioning market the mentioned effects will ultimately
be revealed in higher office rents at a location. Debrezion and Willigers
(chapter 13 of this book) is one of the few noteworthy studies on the influ-
ence of location characteristics on office rents. Its results are generally in
line with the literature on location quality effects on house prices (see
above).
Methodology for valuing the location quality effect
For the purposes of this CBA we assume that the bulk of the effects of the
Dock on the utility of residents and productivity of companies in the
neighbourhood will be revealed in higher real estate prices. Given the
broad range of neighbourhood effects that the Dock will possibly effectu-
ate, we have taken the 5% price premium as a lower frontier for the com-
bined location quality effect, which can be expected from moving the rail-
way infrastructure construction to below ground level (Debrezion et al.
2006). Furthermore, we have relied on two ex ante studies performed spe-
cifically for the Dock project. Buck Consultants International (Buck and
Glaudemans 2006) calculated for the South Axis area and a number of for-
eign locations a so-called area quality index that accounts for transport
accessibility, access to and quality of facilities, economies of scale, the
quality of real estate, and image effects. Comparing the value of this index
and the rent levels at several locations, they suggest that the realisation of
the Dock, together with the extension of the transport infrastructure, can
lead to a 15% increase in the rents in South Axis. Another study based on
interviews with large Dutch real estate agents conducted by Fakton (2002)
estimated a 10% value increase of real estate prices in South Axis due to
the implementation of the Dock project.
Our assumption about the size of the area in which location quality ef-
fects can be expected is based on existing ex post studies. The analyses ar-
gue that the influence of neighbourhood characteristics diminishes quickly
with distance. Orford (1999) shows for example, that the influence of the
major park in Cardiff (Wales) on house prices in the neighbourhood is half
as large at a distance of some hundred metres from the park than in its di-
rect proximity. Debrezion et al. (2006) find that houses within a 250 metre
distance of the railway have a larger negative price premium than houses
situated between 250 and 500 metres from the railway.
In this CBA we assume the combined location quality effect to result in
a 10% increase in real estate prices in the direct proximity of the Dock. We
define the direct proximity of the Dock as the area with a radius of about
250 metres from the tunnel. To account for the fact that effects diminish
with distance, we therefore assume a 5% price increase for the real estate
at a distance between 250 and 500 metres from the tunnel.
By assuming that the external effects of the Dock do not diminish
gradually with distance, but rather in two large steps (from 10% to 5% to
0%), we overestimate these effects to some degree. This overestimation is
compensated in some sense by the fact that we do not quantify the part of
the location quality effect not revealed in property prices (e.g. a higher
utility of tourists or higher salaries of employees in the South Axis). In or-
der to stress the uncertain character of the above calculation, we provide a
sensitivity analysis of the results for two variants: i.) the value increase of
real estate is 5 percentage points lower; ii.) the value increase is 5 percent-
age points higher and manifests in a larger area.
10.5.2 Other external effects
Consumer surplus
Land rents that have been estimated in section 10.4 using the expected
market prices of real estate, are equal to the producer surplus the Dock de-
veloper receives. If companies and inhabitants utility from using the real
estate is higher than market rents, consumer surplus arises. The location
quality effect discussed in the previous section will, for example, at least
initially manifest as a rise of the consumer surplus. For the real estate on
the Dock, however, one can expect that consumer surplus will be so small
in comparison with land rents that it can be neglected for the purposes of
the CBA, the reason being that land owners at South Axis can be expected
to cream off most of the location benefits of users through price discrimi-
nation. In the centre of the South Axis area near the railway station South
WTC, land rents will be highest, reflecting the highest agglomeration
economies that can be enjoyed there. The land rents will diminish with dis-
tance to the station.17
17 This is in accordance with the monocentric model of urban development (see

e.g. Fujita and Thisse 2002), which seems to be rather well applicable to the sta-
There is, however, one type of real estate on the Dock for which con-
sumer surplus cannot be neglected: social housing. The expected land rents
for social housing are less than 10% of land rents for other housing on the
Dock. We assume in this CBA that the utility of the inhabitants of social
housing is equal to the utility of other inhabitants of the Dock area. The
difference between the land rents determined on the land market and rents
requested for social housing is therefore considered as the consumer sur-
plus that inhabitants of social housing on the Dock will enjoy.
Transport benefits
The realisation of the Dock implies an increase in the number of travellers

at South Axis and a decrease elsewhere in the country. With respect to the
railway, the expectation is that the net effect will be negligible because the
necessary capacity of the railway is the same both in the Dock project and
the reference alternative. The situation differs with respect to the highway.
There is a possibility that an increase in congestion in the busy South Axis
area may have effects that cannot be compensated by a decrease in conges-
tion elsewhere. At the time this CBA was performed there was insufficient
information available to value these effects. We therefore include them in
the balance of the costs and benefits as a pro memoria item.
Balance of external effects
Table 10.7 gives an overview of the external effects of the Dock project.
Table 10.7. External effects of the Dock project

Category Method Present value in 2006
( mln in 2006 prices)
Location quality effect 10% and 5% of the rent/sale 360
prices
Consumer surplus social Land rents are set equal to the 40
housing rents for houses built for sale
Transport benefits highway No information available PM
Transport benefits train, Change in the capacity
metro and station
Total of the external effects 400 PM
tion area of South/WTC. The described pattern of rents can be observed in the
South Axis area.
10.6 Total balance of costs and benefits
The results of the CBA are summarised in Table 10.8. The table illustrates
that the total benefits of the project fall short of its costs. The cost-benefit
balance is -80 mln euro. There is, however, a large uncertainty margin
around this figure, not least due to uncertainty surrounding the estimation
of the external effects of the project.
Table 10.8. Summary of the welfare effects of the Dock project (present value in
2006, mln in 2006 prices)
Category Costs Benefits Total
Investment costs 1460
Maintenance expenses 40
Total costs 1500 1500
Land rents 1020

Consumer surplus social housing 40
Location quality effect 360
Transport benefits - PM
Total benefits 1420 PM 1420 PM
Balance of costs and benefits 80 PM
Table 10.9. Net present value of the variants (2006, mln, 2006 prices)
Difference with
the base variant
Base variant (risk-free real discount rate 2%, risk-premium -
benefits 3%)
Risk-free real discount rate 4%, risk-premium benefits 3% 130
Risk-free real discount rate 2%, risk-premium benefits 5% 350
Costs tunnelling transport infrastructure 20% higher 260
Half of m2 gross floor area offices Dock 2020-2028 and half of 60
the gross floor area shops used for residential construction
More optimistic estimates of the land rents (equal to those used + 60
in the Business Case of the South Axis enterprise)
Location quality effect 5 percentage points smaller 180
Idem 5 percentage points larger and applied to a larger area + 180
Sensitivity analysis of the above results (see Table 10.9) suggests that
the assumptions concerning the size of the investment costs, the discount
rate, and the size of the location quality effect have a major impact on the
cost-benefit balance of the project. Results are less sensitive to the assump-
tions concerning the composition of the real estate construction program,

the speed with which this program is realised, and the land rents.
10.7 Discussion
In this study we have used the method of cost-benefit analysis (CBA) to

study, structure and quantify the main effects of a large multifunctional
land use project around the railway station Amsterdam South/WTC. Cost-
benefit analysis has been increasingly gaining favour as a tool to support
policy decision-making, especially with respect to complicated investment
projects in transport infrastructure and land development. A CBA aims to
contribute to the rational and transparent assessment of all project alterna-
tives, mostly by trying to value the totality of welfare effects of the project
in the same (monetary) units; this proceeds smoothly as long as the ex-
pected effects can be reasonably well-estimated and quantified, and the
prices of affected goods and services can directly or indirectly be derived
from real markets. When this is not the case, a CBA method may not be
able to express the balance of costs and benefits as a single figure, or may
yield a balance of costs and benefits surrounded by a large degree of un-
certainty. In both cases additional care is needed in presenting and inter-
preting the CBA results.
A well-known example of effects that are rather difficult to account for
in a CBA are the implications of a project for nature, landscape and envi-
ronment (see Stolwijk 2006 for an extensive discussion). The CBA of the
multifunctional land use in South Axis discussed here offers another ex-
ample: the whole range of effects induced by the change in the pattern of
land use in the neighbourhood of the station South/ WTC. The conse-
quences of a combination of tunnelling and urban construction involve in-
ter alia a decrease in air contamination and noise nuisance, an increase in
accessibility to shopping facilities, and higher employment density. These
factors should contribute to an urban environment better suited in which to
work and reside. The mentioned factors can thus be summarised by the
term location quality improvement and can, through their influence on
the utility of inhabitants and visitors and on the productivity of companies,
lead to higher real estate prices in the neighbourhood of the Dock.
Measurement and valuation of this cluster of effects nevertheless pre-
sents challenges. How should one ex ante measure the expected change in
the location quality? And what is the precise relation between the size of
this change and the real estate prices in the neighbourhood? Existing ex
post studies provide some information on the possible valuation of sepa-
rate neighbourhood characteristics. Debrezion et al. (2006) suggest, for ex-

ample, that residential property located within a distance of 250 metres of
the railway has a negative price premium of circa 5%, compared to the
property at a distance of 500 metres. In the case of the Dock, however, it is
the combined effect of removing the negative influence of the railway
proximity and a number of other changes in different neighbourhood char-
acteristics that interests us. This combined effect may very well turn out to
be an unknown non-linear combination of the separate effects.
Several possible approaches for tackling this problem of estimating the
complex effects of a change in the land use pattern may be considered.
Thus, one could compose an index to measure and summarise the quality
of a location and try to find out the influence of the change in this index on
real estate prices.18 The area quality index of BCI (Buck and Glaude-
mans 2006) discussed in section 10.5 is such an index that can measure the
change in location characteristics. However, the well-known problem with
an index is the certain subjectivity implied by the weights given to differ-
ent components.
Studying the ex post effects of similar projects may be seen to some ex-
tent as an easier option. But this approach has its caveats as well. The chal-
lenge here is to correctly separate the effects of the project on real estate
prices in the neighbourhood from other possible effects. Furthermore, for
locations like South Axis, finding a similar project may on its own be a
rather difficult exercise.
Finally, expert opinion can perhaps provide relevant information. Real
estate agents seem to be experts in projects having to do with land devel-
opment. The louder criticism of their forecasts is, however, that their in-
formation is based on the most recent trends and does not (fully) account
for the possibility of future economic trend changes. In the case of South
Axis, a relevant trend change is, for example, the expected decrease in
population and employment growth after 2020 (Huizinga and Smid 2004),
which may have important negative implications for the demand for office
space.
In our discussion we have illustrated some of the limitations of the CBA
methodology when evaluating the effects of a change in the pattern of land
use. When (external) effects of location quality improvement present an
important rationale for a construction project, a CBA of this project is
likely to yield a result surrounded by a rather large degree of uncertainty.
18 This type of study is, for example, widely used in research on the impact of
environmental policy on companies location decisions. While the concept of en-
vironmental policy is broad and difficult to define, an index weighing different
types of policy measures can offer a solution.
Future research should assist in developing better methods to measure and

estimate these effects. Until then, additional care may be needed when pre-
senting and interpreting CBA results for projects under consideration, for
instance by showing the margins around the chosen valuation.19 We again
stress that knowing the difficulties and inherent limitations when trying to
measure, quantify and make comparable often very diverse effects of a
project is crucial in order to correctly interpret and use the results of CBAs
for decision-making purposes.
10.8 Conclusions
This chapter has reported on a case study of the welfare effects of the de-
velopment of a railway station area: Amsterdam South Axis the area
around the station Amsterdam South/WTC, in which tunnelling the trans-
port infrastructure and urban construction above the tunnel are to be com-
bined to create a new high quality urban district. Using the technique of
cost-benefit analysis (CBA), we have explicitly compared the benefits and
costs of this multifunctional land development.
The analysis suggests a methodological approach for evaluating the wel-
fare effects of multifunctional land use projects with the help of CBA. We
have inquired into the definition of multifunctional land use for the pur-
poses of the analysis, and have investigated how internal land benefits and
external effects of the change in the pattern of land use can be incorporated
into a CBA. We argue that, given the state-of-the art knowledge on effects
of changes in the pattern of land use, the ex ante estimates of these will, for
the purposes of CBA, most certainly have a rather large uncertainty mar-
gin. The implication therefore is that, for projects in which these effects
expectedly play an important role, extra attention should be paid by deci-
sion-makers to correctly interpreting the information provided by cost-
benefit analyses.
19 To stress the uncertain nature of the results, one may also opt for presenting
the results in terms of cost-efficiency and to determine the net financial costs in-
volved in the realisation of the external effects of location quality improvement. In
the case of the Dock project, these costs have been estimated at 480 mln. Under
the assumption that the external effects apply to an area of 500 metres around the
Dock, one can then calculate how large the effect should be per single office or
dwelling in order for the project to be welfare-enhancing.
Acknowledgement
The authors thank Paul Besseling, Jos Ebregt, Ruud Okker, Herman Stol-
wijk, Rafael Saitua and the participants of the ERSA-2005 conference for
helpful comments on earlier versions of this study. The usual disclaimer
applies.
References
Arcadis (2006) Costs of the Dock. Arcadis, Arnhem

Besseling P, Ebregt J, Saitua R (2003) Kengetallen Kosten-Batenanalyse Project
Zuidas Amsterdam. CPB Document 44, CPB, The Hague
Brcker J (2003) Indirect Effects: Do They Matter for Cost-Benefit Analysis? In:
Centraal Planbureau (eds) Twee Jaar Ervaring met OEEI. De Discussie over
Indirecte Effecten. CPB, The Hague
Buck R, Glaudemans J (2006) De Zuidas in Internationaal Perspectief. Real Estate
44:2024
Ciccone A, Hall RE (1996) Productivity and the Density of Economic Activity.
American Economic Review 86:5470
Cheshire P, Sheppard S (1995) On the Price of Land and the Value of Amenities.
Economica 62:247267
City of Amsterdam (2005) Visie Zuidas: Stand van zaken 2004, Amsterdam
Coupland A (1997) Reclaiming the City: Mixed Use Development. Spon, London
Debrezion G, Pels E, Rietveld P (2006) The Impact of Rail Transport on Real Es-
tate Prices: An Empirical Analysis of the Dutch Housing Market. Tinbergen
Institute Discussion Paper 2006-031/3, Tinbergen Institute and Vrije Univer-
siteit Amsterdam, Amsterdam
Deloitte (2006) Financieel model Zuidas Dok, version 20 June
Eijgenraam CJJ, Ossokina IV (2006) Kosten-batenanalyse Zuidas Amsterdam,
CPB document 134. CPB, The Hague
Eijgenraam CJJ, Koopmans CC, Tang JG, Verster ACP (2000) Evaluatie van In-
frastructuurprojecten. Leidraad voor Kosten-Batenanalyse. Sdu Uitgevers,
The Hague
Fakton (2002) Toelichting Scenario Onderzoek Zuidas Amsterdam, mimeo
Fujita M, Thisse JF (2002) Economics of Agglomeration. Cambridge University
Press, Cambridge
Geoghegan J, Wainger LA, Bockstael NE (1997) Spatial Landscape Indices in a
Hedonic Framework: An Ecological Economics Analysis Using GIS. Ecolo-
gical Economics 23:251264
Huizinga F, Smid B (2004) Vier Vergezichten op Nederland. Productie, Arbeid en
Sectorstructuur in Vier Scenarios tot 2040. Bijzondere publicatie 55, CPB,
The Hague
Irwin EG, Bockstael NE (2001) The Problem of Identifying Land Use Spillovers:
Measuring the Effects of Open Space on Residential Property Values. Ameri-
can Journal of Agricultural Economics 83:698704
Layard R, Glaister S (1994) Cost-Benefit Analysis. 2nd edition, Cambridge Uni-
versity Press, Cambridge
Lijesen M, Straaten JW van der, Dekkers J, Elk R van (2006) Geluidsnormen
voor Schiphol: Een Welvaartseconomische Benadering. CPB Document 116,
CPB Netherlands Bureau for Economic Policy Analysis, The Hague
Orford S (1999) Valuing Location in an Urban Housing Market. Mimeo
Ricardo D (1817) The Principles of Political Economy and Taxation. Guernsey
Press, London
Rodenburg CA, Nijkamp P (2004) Multifunctional Land Use in the City: A Typo-
logical Overview. Built Environment 30:274288
Rouwendal J, Straaten JW van der (2006) The Costs and Benefits of Providing
Open Space in Cities. Mimeo
Song Y, Knaap GJ (2004) Measuring the Effects of Mixed Land Uses on Housing
Values. Regional Science and Urban Economics 34:663680
Stolwijk H (2006) Kanttekeningen bij het Gebruik van Contingent Valuations in
Maatschappelijke Kosten-Batenanalyse. Tijdschrift voor Politieke Economie,
27/3:3751
Vreeker R, Groot HLF de, Verhoef ET (2004) Urban Multifunctional Land Use:
Theoretical and Empirical Insights on Economies of Scale, Scope and Diversi-
ty. Built Environment 30:289307
11 Measuring the WTP for shopping facilities
around railway stations
Thomas de Graaff

the Netherlands
Caroline Rodenburg
Real Estate Advisory Services and International Location Advisory

Services, Ernst & Young, Utrecht, the Netherlands
11.1 Introduction
Railway stations often function as a nexus of various activities, such as

transport, shopping and working. Larger stations especially act as nodes
for several transport modes, including heavy rail, light rail and city bus
transport. Therefore, it is precisely due to their strategic and accessible lo-
cations that specific railway stations increasingly become more attractive
for the location of firms. Because station areas potentially act as magnets
for service sector firms (particularly) and in combination with increased
traffic density, these areas also attract many smaller facilitative firms, such
as shops, childcare centres, and restaurants. This situation leads to a vari-
ety of (Marshallian) localised external economies of scale; examples of
railway stations that have induced such economies of scale are the high-
speed railway station in Lille, Gare Montparnasse in Paris, Broadgate in
London, Lehrter Bahnhof in Berlin, and the train-metro-tram station Zuid-
as in Amsterdam. At present, the area around the latter is witnessing a
rapid transformation towards a completely new central business district
(CBD) (for more details, see Rodenburg 2005).
Due to the dependence on accessibility and transport opportunities,
these agglomeration forces may not always arise spontaneously, but may
instead require the coordinated planning of transport networks, railway sta-
214 Thomas de Graaf and Caroline Rodenburg
tion areas and real estate investments in the vicinity. When (local) gov-
ernments invest in such areas they will usually require insights into the so-
cial costs and alternatives of such station areas. Unfortunately, such costs
and benefits cannot always be observed directly in actual markets, espe-
cially not when they concern external effects such as economies of ag-
glomeration, or when the markets concerned simply do not yet exist. A vi-
able alternative is then to measure such costs or benefits by research
questionnaire on stated preferences. This chapter presents such a study
by focussing on employees evaluations of various shop types in a so-
called multifunctional development currently under consideration at the
Amsterdam South Axis.
City planners introduced the term multifunctional land use as a spatial
planning concept aimed at the combination of different land use functions
in order to save scarce land and exploit synergies between land use func-
tions. Because of the high intensity of land use, such areas are normally
planned near large public transport nodes like railway stations. Gathering
information on the costs and benefits for different groups of stakeholders
in multifunctional land use projects is essential in order to provide a better
foundation for investment decisions in multifunctional land use projects
(see for an introduction into multifunctional land use, inter alia, Jacobs
1961; Coupland 1997; Priemus et al. 2000; Nijkamp et al. 2003).
Previous research claims that people attach positive values to multifunc-
tional land use projects (Geoghegan et al. 1997; Song and Knaap 2004; Ir-
win and Bockstael 2001; Irwin 2002). Possible reasons for this may be that
multifunctional land use has a positive impact on how people perceive and
appreciate the built environment (Van Wee 2003). This perception value
of areas usually around railway stations is a major element of city market-
ing, which is important for competition between cities in attracting firms
and residents, and influencing the economic and social liveability of a city
(Florida 2002). Other possible reasons are the existence of an option
value, which expresses the value people derive from having access to
goods and services, independent of whether they actually use them (Weis-
brod 1964; Johansson 1987, 1991; Geurs and Van Wee 2004), and the be-
lief that it generates less traffic than mono-functional land use patterns
(Bannister 1994). In railway station areas, moreover, the option value is
strengthened by the possibilities for people to use various transport modes
(Geurs and Ritsema van Eck 2001). There could also be a conventional
use-value, when some of the functions at railway station areas involve
goods and services that the sites users want to consume. In this chapter we
concentrate on such tangible use-values for the presence of different
shops at the Amsterdam South Axis area by a particular group of stake-
Measuring the WTP for shopping facilities around railway stations 215
holders, viz. the employees (for a detailed description of the South Axis
we refer to chapter 9 in this volume).
Our approach in this chapter is to circumvent the absence of markets for
multifunctional land use characteristics by using individuals stated behav-
iour on hypothetical markets for these characteristics. By means of a stated
preference survey we identify employees willingness-to-pay (henceforth
denoted as WTP) for the presence of various types of shopping facilities
(see for an overview of stated preference surveys, e.g., Louviere et al.
2000). WTP surveys allow us to infer price-related changes in behaviour
by drawing from hypothetical price changes. WTP surveys basically
measure potential demand for products or services by asking consumers,
for example: Would you purchase this product or make use of this service
if it were offered at this price? We use a similar questionnaire approach to
quantify the benefits of shopping facilities measured by the use-value
as valued by office employees around the railway station in the South Axis
area. Office employees will form a substantial share of the users of the
South Axis, and is thus an important group to take into consideration, since
in current development plans, 45% of the area consists of office space.
In the questionnaire respondents were requested only to give detailed
preferences over the subset of shops they frequented most often. Our inten-
tion has been to minimise the burden of plodding through a long list of
similar stated preference questions. Unfortunately, this set-up induces a
potential entanglement of the use-value of shopping facilities at the Am-
sterdam South Axis and the general use-value unconditional on the loca-
tion. The latter may arise if employees do not have general access to the
shopping facilities in question because they are not present at their residen-
tial location. This does not lead to inconsistent results, but might seriously
obscure the estimation results of the use-values of shopping facilities at the
South Axis, and thus might ultimately lead to wrong conclusions concern-
ing the benefits of shopping facilities at the Amsterdam South Axis instead
of the benefits of shopping facilities in general.
The remainder of the chapter is organised such that the next section lays
out the research framework, first dealing with the questionnaire, and then
showing how WTP values may be derived from questionnaire answers.
The following section deals with the data and provides evidence for the
importance of accounting for the number of shops at the location of resi-
dence. Thereafter, we continue with the empirical specification and the es-
timation results. To analyse the WTP for shopping facilities at the Amster-
dam South Axis, a questionnaire was sent to employees at the Amsterdam
South Axis. The next subsection outlines the particulars of the question-
naire, followed by a discussion of the methodology used to find WTP val-
ues for shopping facilities.
11.2 The questionnaire
The questionnaire was developed in electronic format (in Dutch and Eng-
lish) for respondents to complete on the internet. An invitation to partici-
pate was sent to about 6600 employees via an internal mailing by compa-
nies located within the South Axis area. For our current purpose, the two
most important parts of the questionnaire concern questions about prefer-
ences and the expected use frequencies for several types of shopping facili-
ties nearby their work location. The answers to these questions were used
to create scenarios for which respondents were subsequently asked to ex-
press changes in expenditure shares. The second part explicitly requests
employees personal characteristics such as age, gender, education level,
income, family situation, commuting mode, average travel time between
home and work location, and zip code of residence. In the end, the resul-
tant database contains 1952 respondents (the response rate was therefore
32%), of which 1271 individuals completed all relevant questions (includ-
ing location of residence). For this type of research the response rate is
normal (Sheehan 2001).
Sample selection biases might result from several sources. Because we
conducted an internet survey, a selection bias might arise when not every
employee has access to the internet. In business companies this may be the
case for those employees who do not need internet access to carry out their
jobs. Furthermore, the survey sample might not be fully representative for
employees at the South Axis, since not all companies in the South Axis
area participated. This is due to the fact that i.) some companies were un-
willing to participate, and ii.) at that time there was no complete overview
available of all companies located at the South Axis. Most participating
companies are located close to the Amsterdam Zuid World Trade Centre
railway station (approximately a two-minute walk), which could lead to a
slightly distorted picture in terms of, for example, commuting mode used,
assessment of accessibility of the South Axis, and preferences for facili-
ties. Most companies situated near to the Amsterdam Zuid-WTC railway
station are financial institutions and law offices, whereas other employ-
ment (e.g. the public sector) is located somewhat farther away from the
Amsterdam Zuid-WTC railway station.
11.3 WTP for shopping facilities at the South Axis
As explained above, the questionnaire is designed to identify how many

(more) people are willing to pay for the use of shopping facilities plus the
existence of these shops close to work at a multifunctionally-designed site

above having access to the same facilities elsewhere. Because people have
different preferences regarding the most desired type of shops, various
types of shops were considered in the questionnaire. To avoid excessive
repetition and fatigue for the respondents, we decided to ask that they
value only their two most frequently visited shops.
Because individuals are not accustomed to paying for access to shops,
we used the following approach for the valuation of shopping facilities.
We asked respondents for their two most frequently visited shops at pre-
sent and, thereafter, asked them about their changes in expenditure shares
in both shops for four different hypothetical price levels when those two
shops were also present at Amsterdam South Axis. We asked for most fre-
quently visited shops and not most preferred shops because in the former
case we were certain that respondents (shopping experience) would have a
reasonably accurate idea of the normal price level in these shops else-
where.
For our purposes the expenditure share is defined as the share of the
total expenditure in a certain type of shop as spent on the South Axis if that
type of shop were present there. The reason we asked respondents to indi-
cate changes in expenditure share, and not changes in purchased quantities
of a specific product, is that questions in terms of expenses are more natu-
ral to formulate and easier to answer. This is important, since people are
expected to be particularly reluctant towards paying an entrance fee for
permission to use shopping facilities. The use of budget shares enables us
to assess individuals responsiveness towards price changes in shopping
facilities in a multifunctionally-designed area, compared to shopping fa-
cilities elsewhere.
Thus, the type of shop for which respondents had to fill in their change
in expenditure shares depended on the frequency of visits to that particular
shop, which they had answered in a previous survey question. In the open-
ended question that followed about respondents expenditure shares in
specific shops, four scenarios were presented. In the first we hypothesised
0
an equal price level at the South Axis p ZA compared to alternative options
(e.g. shops that respondents currently use). In subsequent scenarios the
(relative) price level in the South Axis area (pZA) was presented as 10%,
25% and 50% higher, respectively. For each scenario respondents indi-
cated which percentage of their total expenditures (E) on a specific type of
shop (e.g. supermarket, depending on earlier answers) they expect to spend
in an outlet in the South Axis area.
By asking respondents about their budget share in a specific shop at dif-
ferent price levels, we obtain indirect information about the excess WTP
of employees for the use of shopping facilities at the South Axis, relative
to using their current outlet. The extent to which demand decreases when
relative prices exceed unity can namely be seen as a measure for respon-
dents WTP. Responses to this question have allowed us to estimate a de-
mand function for their purchases at the South Axis, rather than their buy-
ing these goods elsewhere. Figure 11.1 shows this demand function, for
which the non-South Axis price and quantity (and hence expenditures) are
normalised to one.
Figure 11.1. Hypothesised relation between price level (pZA) and expenditure
level (qZA) at the Amsterdam South Axis (ZA)
Figure 11.1 shows the budget shares qZA at different relative price levels
0
pZA. The area between the demand function and the line p ZA can thus be
seen as a measure for respondents WTP to use shopping facilities at the
South Axis. Now, because we have four (aggregate) observation points, we
might e.g. fit a quadratic function (in the form:
p(q ZA

) = + 1 q ZA + 2 (q ZA ) ) through those four points by means of or-
2
dinary regression, as indicated in Figure 11.1. The area, and thus the WTP,
would then reveal itself by the following integral:
[ + q ]
0
q ZA
WTP =
+ 2 (q ZA

) dq ZA q ZA0 .
2
1 ZA (1)
0
Note that this WTP is a measure for the change in budget shares. Divid-
ing by total expenditures leads to the desired WTP in monetary terms. See
too, that self-selection is an issue since respondents were asked to fill in
the open-ended question for their two most frequently visited shopping fa-
cilities. Because we know the relative importance employees attach to the
various shops as well, we are able to correct for self-selection bias. In this
case self-selection is caused by the fact that employees sort themselves out
when answering questions regarding their WTP. Note that this happens
only when frequency of shop visits coincides with employees preference
structure for shop facilities, which is rather likely. For further methodo-
logical issues, we refer to De Graaff et al. (2005).
11.4 Data
Our data originates from two sources. The first is the questionnaire de-
scribed in section 11.2; the second source is information gathered from the
Yellow Pages on the internet. The latter dataset concerns the number of
shops present in the area of residence of each respondent at a 4-digit post-
code. Subsequently, we match this area-specific data with individual spe-
cific data using the 4-digit postcode of the residential location as the indi-
cator variable. We will next deal with the characteristics of the
respondents, followed by an explicit discussion on the characteristics of
visited and preferred shops (as reported), and the number of shops in the
employees residential location.
11.4.1 Respondents
The second column in Table 11.1 presents the descriptive statistics of the
distribution of respondents in the stated preference survey. To give a char-
acterisation of the workforce at the Amsterdam South Axis, we present
data for the active workforce in the Netherlands from 2003 (CBS 2004) in
the third column of Table 11.1 for comparison.
Table 11.1. Characteristics of the respondents (percentages of population)

Characteristics Value South Axis Value Netherlands
Gender
Male 60 59
Age
<30 26 23
31-40 35 29
>41 39 48
Highest education degree
Bachelor/Master 82 27
High school 16 64
Other 2 8
Working days per week
3 or less 10 17
4 31 20
5 59 63
Main mode of commuting
Car 29 60
Public transport 38 8
Bicycle 30 25
Other transport modes 3 7
Net personal monthly income
< 2000 35 90
2000 3000 30 3
> 3000 27 1
Unknown 5 6
Family situation
Living alone without children 26 14
Living alone with children 2 3
Living with partner and with children 38 54
Living with partner and without children 34 29
Partner (when living together)
Works full time 51 14
Work Location
< 2 min. walk from railway station 50
A comparison of the two datasets reveals characteristics in which the

average South Axis employee differs from the average employee in the
Netherlands. We see quite clearly that the average employee at the South
Axis is higher educated than his counterpart in the Netherlands, which is
also reflected in the income distribution. This difference is probably
caused by the types of companies present at the South Axis, which are pre-
dominantly law offices and head offices of international companies. Con-
sequently, they employ more high-skilled workers. The active workforce at
the South Axis is moreover relatively young compared with the active
workforce in the Netherlands, and employees there seem more often to live
alone or with a partner without children. Finally, we recognise a very high
share of employees at the South Axis who use public transport. This might
be because the South Axis has attracted companies that emphasise accessi-
bility by public transport due to the national government targeting it as a
place designed to be easily accessible by public transport. Meanwhile, its
proximity to the Amsterdam Zuid-WTC railway station shows that the
South Axis, as well as the dataset, indicates that large numbers of employ-
ees work close to the large (and expanding) station.
11.4.2 Shops and the location of residence
As indicated above, this large group of high-skilled workers causes the la-
bour market for the common employee at the South Axis to be relatively
thin, and the commuting distance for the average employee to be relatively
large. To illustrate this high average commuting distance, Figure 11.2
shows the spatial distribution of the residential location of employees at
the South Axis at the 2-digit (a) and 3-digit (b) postcode level.
Figure 11.2a. Residential location of employees at the Amsterdam South Axis at

the 2-digit postcode level
Figure 11.2b. Residential location of employees at the Amsterdam South Axis at

the 3-digit postcode level
As Figure 11.2-a clearly shows, employees at the South Axis are widely
distributed across (the Western part of) the Netherlands. Moreover, at the
3-digit postcode level (11.2-b), we notice that many employees do not live
in the surrounding cities of Haarlem, Utrecht, or The Hague, with the ex-
ception of course of Amsterdam. A distinct sample of employees lives
in more rural areas surrounding Amsterdam. Note that among them the
most favourable residential locations are those 3-digit postcode areas pos-
sessing high accessibility characteristics (by train and/or highway connec-
tions). Probably one reason why employees are able to commute this far is
the accessibility of the South Axis itself, with its good transport connec-
tions in all directions.
As explained above, we asked respondents to indicate the percentage
change in their budget shares at shops at the South Axis if prices there
were equal or 10%, 25%, and 50% higher. The shopping facilities re-
quested were supermarkets, drugstores, dry cleaners, bookshops, flower
shops, and hairdressers. These facilities were chosen because they repre-
sent a wide range of shop types, where some are more uniformly distrib-
uted over space (e.g. supermarkets) and others are clustered (e.g. book-
shops). A negative relation is likely between the number of a particular
type of shop in the residential location and the WTP for that type of shop
at the South Axis. For example, if there are no bookshops near employee
residences, the employee has a high WTP for a bookshop at the South
Axis. But this WTP has no connection to the use-value of bookshops at the
South Axis; the employees simply value better access to bookshops (or the
general use-value, regardless of location).
Since a significant percentage of our sample lives outside the larger cit-
ies, to not account for the number of shops present at the residential loca-
tion might obscure the results, in that merely overall accessibility to shop-
ping facilities is estimated but the multifunctionality of the Amsterdam
South Axis is not. Insight into the precise nature of the WTP for shopping
facilities is important, because accessibility to shops close to the place of
residence might easily change through e-shopping or new shopping malls,
which consequently lowers the WTP for shopping facilities at the South
Axis. If the multifunctionality of the Amsterdam South Axis is truly esti-
mated, then WTP values will remain fairly stable, thus instigating conse-
quences for error-margins in the cost-benefit analysis of this area. To illus-
trate the spatial distribution of shopping facilities, Figure 11.3 shows the
national distribution of the number of supermarkets and bookshops per
km2 in the Netherlands.
Figure 11.3. Spatial distribution of supermarkets and bookshops across Dutch 4-

digit postcode areas
Clearly there are differences in scope and clustering across types of

shops. Where some shops benefit more from a wide distribution (because
e.g. of the homogeneity of products), other shops thrive when situated in
clustered patterns (e.g. in cities) by using a variety of scale economies. Ta-
ble 11.2 offers descriptive statistics of the shopping facilities respondents
have valued at the South Axis (including percentage of those chosen as
most frequently visited, average value attached to the shops, average value
attached when chosen), and the characteristics of the same type of shops in
the location of residence, including the number of shops per km,2 maxi-
mum number of that type of shop in a 4-digit postcode area, and the per-
centage of 4-digit postcode areas that do not contain such particular shops.
Table 11.2. Characteristics of the chosen shopping facilities

Shops at the South Axis Shops in the 4-digit postcode of
residence
Chosen Value Value when Number of Max. no. No. of no
(%) (1 5) chosen (1 5) shops p/km2 of shops shops (%)
Supermarkets 35 3.51 3.93 1.82 9 6

Drugstore 28 3.37 3.73 1.42 14 21
Dry cleaners 5 2.41 3.71 0.77 4 37
Bookshop 18 3.37 3.68 1.74 20 32
Flower shop 12 3.30 3.59 1.80 13 12
Hairdressers 1 1.98 3.32 6.26 40 3
As Table 11.2 shows, respondents report supermarkets as most fre-

quently visited shops, closely followed by drugstores. If we look at the av-
erage preference value respondents attach to shops earmarked for the
South Axis, then supermarkets, drugstores, bookshops, and flower shops
are considered to be most valuable. Dry cleaners and especially hairdress-
ers are valued much less. This value, however, does not say anything about
employees WTP for these facilities. Preference values report only the per-
ception of the employee in terms of relative importance. The fourth col-
umn, value when chosen, offers an impression of the size of the selection
bias when we do not control for it. Even the perceived value for hairdress-
ers is then more or less on par with perceived values for supermarkets and
drugstores. Obviously, reported expenditure shares are then biased
(mostly) upwards as well. The fifth column of Table 11.2 indicates the av-
erage number of that particular shopping facility across 4-digit postcode
areas.
Hairdressers in particular are abundant, which explain the low value
people attach to additional hairdressers at the South Axis. If we also take
the maximum number of shops into account, we see that some postcode
areas have a relatively large number of bookshops, although the average
number per postcode area is rather low. This indicates again that facilities
such as bookshops are more densely clustered and benefit more from each
others presence. The opposite is apparently true for dry cleaners, which
seem to be spread evenly across postcode areas (if present at all). The last
column presents the percentage of residential postcode areas as reported by
the respondent that contains no shops of that type. We notice that super-
markets and hairdressers function almost as necessary shops, whereas
dry cleaners and bookshops are less often near the place of residence.
Thus, supermarkets and drugstores are on average considered to be the
most valuable. However, whether this truly reflects the amount of the WTP
for shops at the South Axis, or if it is correlated with personal characteris-
tics remains to be seen. The impact of the number of shops in a particular
residential location on the WTP for that type of shop is as yet also unclear.
The next section will therefore deal in a more rigourous manner with the
determinants of WTPs for the various shops.
11.5 Empirical application
As mentioned above, six shop types were distinguished in the question-

naire: namely, supermarkets, drugstores, hairdressers, dry cleaners, flower
shops, and bookshops. Each individual was first asked to indicate the two
most frequently visited shops and subsequently indicate any change in ex-
penditure level for both shops when prices at the South Axis (pZA) were
0%, 10%, 25%, and 50% higher. Although the two variables most fre-
quently visited shop and most preferred shop are not identical, Table
11.2 indicates that self-selection is most likely an issue. Because each re-
spondent also had to indicate a preference ranking of shops separately
from most frequently visited shops, we are able to construct a ranking
structure s for each shop s. Normalising the prices ( p ZA
0
) at one enables us

to specify the relation between quantities ( q ZA ) and price levels ( p ZA ) for
shopping facility s as follows

q ZA | s = D | s + (D rs ) | s + Z | s + s , (2)

where D is a matrix of dummies for price levels at the South Axis ( p ZA )
with 24 columns four relative price levels ( {1.0, 1.1, 1.25, 1.5})
for each of the six shopping facilities D rs are cross-effects between the
24 dummies for the four price levels for the six shops, and the number of
shops present in the residential location rs, Z is a set of individual charac-
teristics; , , and are parameters to be estimated, and is a normally
distributed error term. Note that s indicates that this specification corrects
for self-selection as well (see De Graaff et al. 2005 for specification esti-
mation issues). Table 11.3 offers the estimation results of specification 2.
Table 11.3. Estimation results for shopping facilities (at 5% significant in bold)
Standard Standard
Variable Coefficient Variable Coefficient
error error
Price dummies for supermarkets () pZA = 1.0 0.077 0.020
pZA = 1.0 0.348 0.008 pZA = 1.1 0.042 0.023
pZA = 1.1 0.229 0.010 pZA = 1.25 -0.006 0.019
pZA = 1.25 0.150 0.009 pZA = 1.5 -0.042 0.012
pZA = 1.5 0.114 0.008 Cross-effects for drugstores ()
Price dummies for drugstores () pZA = 1.0 0.115 0.036
pZA = 1.0 0.470 0.011 pZA = 1.1 0.097 0.020
pZA = 1.1 0.318 0.011 pZA = 1.25 0.043 0.014
pZA = 1.25 0.178 0.011 pZA = 1.5 -0.029 0.014
pZA = 1.5 0.106 0.010 Cross-effects for dry cleaners ()
Price dummies for dry cleaners () pZA = 1.0 -0.028 0.019
pZA = 1.0 0.628 0.019 pZA = 1.1 -0.178 0.032
pZA = 1.1 0.444 0.021 pZA = 1.25 0.049 0.025
pZA = 1.25 0.095 0.017 pZA = 1.5 0.014 0.023
pZA = 1.5 -0.092 0.018 Cross-effects for bookshops ()
Price dummies for bookshops () pZA = 1.0 -0.181 0.017
pZA = 1.0 0.498 0.012 pZA = 1.1 0.188 0.020
pZA = 1.1 0.314 0.013 pZA = 1.25 0.140 0.032
pZA = 1.25 0.134 0.013 pZA = 1.5 0.118 0.028
pZA = 1.5 0.036 0.015 Cross-effects for flower shops ()
Price dummies for flower shops () pZA = 1.0 0.036 0.021
pZA = 1.0 0.479 0.014 pZA = 1.1 0.060 0.051
pZA = 1.1 0.341 0.016 pZA = 1.25 0.081 0.028
pZA = 1.25 0.145 0.014 pZA = 1.5 0.091 0.022
pZA = 1.5 0.025 0.014 Cross-effects for hairdressers ()
Price dummies for hairdressers () pZA = 1.0 -0.265 0.079
pZA = 1.0 0.393 0.026 pZA = 1.1 0.171 0.111
pZA = 1.1 0.219 0.035 pZA = 1.25 0.051 0.075
pZA = 1.25 -0.092 0.026 pZA = 1.5 0.016 0.019
pZA = 1.5 -0.252 0.028 Age (base 30)
Cross-effects for supermarkets () 31 40 -0.045 0.005
To be continued
Table 11.3. Estimation results for shopping facilities (at 5% significant in bold,
continued)
Standard Standard
Variable Coefficient Variable Coefficient
error error
> 41 -0.076 0.006 Transport mode (base = car)
Missing -0.140 0.013 Train -0.018 0.006
Travel days ( base = 1 3) Tram/bus 0.042 0.006
4 0.030 0.007 Walk/bike -0.036 0.006
5 0.028 0.006 Other modes -0.016 0.012
Net personal monthly income (base
Gender (base = male)
1500)
Unknown (5.1%) 0.039 0.019 Female 0.026 0.004
Urbanisation of place of residence (base
1500 2000 0.013 0.006
= < 500 addresses per km2
2000 3000 0.029 0.006 500 1000 -0.020 0.005
> 3000 0.019 0.007 1000 1500 -0.015 0.007
Travel Time (base = 0 10 minutes) 1500 2500 -0.027 0.008
10 20 minutes -0.048 0.007 > 2500 0.039 0.010
20 30 minutes -0.072 0.007 Missing -0.031 0.010
30 45 minutes -0.086 0.007
45 60 minutes -0.100 0.007 Number of observations 10168
> 60 minutes -0.102 0.008 Mean Log likelihood 0.22
The estimation results in Table 11.3 focus on the impact of individual

characteristics on the WTP for shopping facilities at the Amsterdam South
Axis and on the impact of the number of shops in the residential location
(). Although the individual characteristics directly influence expenditure
shares in shopping facilities at the South Axis, direct inference on the level
of WTP values is not feasible, because information on individual-specific
variation in monthly household expenditures is unavailable. However, we
are able to interpret the sign of the coefficients as the direction WTP val-
ues change due to individual characteristics. We first deal with the impact
of individual characteristics; afterwards we look specifically at mean
WTPs for shopping facilities at the South Axis without individual charac-
teristics.
Almost all coefficients of the individual characteristics are statistically
significant. To start, we observe that young employees have higher prefer-
ences for shopping facilities than older employees, which may indicate that
they are less loyal to local shops around their place of residence. Workers
who are more often at the South Axis are willing to pay more for the pres-
ence of shops, as the probability increases that they actually use these fa-
cilities. The income parameters point to the theoretically justifiable
pattern that workers with higher incomes also have higher WTPs for shop-
ping facilities. Rather significantly, workers with lowest travel time also
have the highest WTP for shopping facilities. These workers probably not
only expect to frequent these shops during working hours, but also during
non-working hours, e.g. at weekends, since shops are located close to their
place of residence. This indicates that workers value an overall increase in
accessibility to shops higher than merely an increase in accessibility to
shops at the work location.
We observe that, contrary to intuition, workers who walk or cycle to
work have the lowest WTP for shopping facilities whereas workers who
use the car, and especially bus or tram, have the highest WTP. Most likely
this is correlated with the ease of carrying home the purchases made at the
shops. Overall, women seem to have a higher WTP for shopping facilities.
Finally, workers who live in either very densely populated or very sparsely
populated areas have a high preference for shops at the South Axis. This is
probably due to two effects, first, because workers who live in rural areas
better value access to shops anyway, irrespective of whether shops are pre-
sent in the proximity of residence or near the work place. Secondly, work-
ers who live in very densely populated areas might live there because they
have an altogether higher preference for shopping facilities.
When we examine the various cross-effects () that measure the impact
of the number of shops in the residential location, we can then observe as
expected mostly negative effects. However, an exception is drugstores,
which indicates that a larger number of them can be found at the residen-
tial location, thus leading to a higher WTP for drugstores at the South
Axis. However, the other shopping facilities show an (overall) negative re-
lation for cross-effects. Figure 11.4 depicts the estimated change in budget
shares in the various types of shops when the number of shops in the resi-
dential location increases.1 Clearly, the individual budget shares depend
largely on residential situation. If employees live in (rural) areas with low
accessibility to shopping facilities, then their WTP for shops at the South
Axis is high (even for hairdressers) because they nevertheless value acces-
sibility. As argued above, such a high WTP is not a correct measure for
multifunctional land use. Thus, when assessing benefits of the South Axis
with respect to multifunctional land use, accessibility to a vast array of fa-
1 Note that the change in budget shares in Figure 11.3 depends on the baseline
case, which means in this case young car driving males, who work less than four
days per week at the Zuidas, who are low-earners and live in very sparsely popu-
lated municipalities. However, we need to include individual characteristics, be-
cause these are highly correlated with the number of shops in the residential loca-
tion; leaving them out induces unobserved heterogeneity bias.
cilities in the residential location has to be taken explicitly into account in

order to address the benefits of multifunctional land use correctly.
Figure 11.4. Relation between number of shops in the residential location and the
change in budget share
Note that the price dummies in Table 11.3 directly give the effect of
prices on quantities for a sample of young car driving males who work less
than four days per week at the South Axis, who are low-earners, and live
in very sparsely populated municipalities. Before examining the total sam-
ple, a few general comments are appropriate here. For all shop types, the
analyses result in vertical intercepts within the range 0.63 to 0.35, imply-
ing that the employee belonging to the sample as described above antici-
pates spending 35% to 63% of his total expenditure in these kinds of shops
at the South Axis when prices remain the same. For hairdressers and dry
cleaners, workers will actually be spending less at the South Axis, com-
pared to their normal expenditure pattern when prices will be raised too
high at the South Axis.
To derive average WTPs for the total sample we use the approach as ex-
plained in subsection 11.2.1 for an estimation without covariates. Esti-
mates are provided by (CBS, 2004) for the average monthly household ex-
penditures at supermarkets (260), drugstores (35), hairdressers (13),
dry cleaners (1.40), flower shops (11.40), and bookshops (26.70). Un-
fortunately, budget shares for sub-samples of the Dutch population are un-
available. This might induce a bias in the sense that budget shares from our
sample may as a result deviate from those of the Dutch population. The
second column in Table 11.4 shows the share of an employees average
monthly expenses in a specific type of shop he or she is willing to spend
extra for the use of that type of shop in the South Axis area. The third col-
umn shows the final conversion to the corresponding WTP values.
Table 11.4. Expenditure shares and sample average WTP values for shopping fa-
cilities at the South Axis
Shopping facility Expenditure share Sample average WTP
Supermarket 5.95% 15.46
Drugstore 5.50% 1.92
Dry cleaners 3.96% 0.06
Bookshop 3.69% 0.98
Flower shop 3.40% 0.39
Hairdressers 22.51% 2.87
Total WTP 21.68
Table 11.4 shows that the WTP values for supermarkets are highest,
which is primarily due to the fact that average monthly household ex-
penses for non-supermarket shops are much lower than for supermarkets.
If we look at the WTP for the use of different types of shops in a multi-
functionally-designed area as a share of average monthly expenditures,
however, we see that the WTP in terms of expenditure shares are for all
shops more or less equal, with the notable exception being hairdressers.
Presumably, frequency of visits and value attached to hairdressers do no
coincide, but instead have opposite signs. This means that workers who do
not often visit hairdressers value them higher than workers who visit them
frequently. The other expenditure shares all drop when accounting for self-
selection, with the exception of the expenditure share for supermarkets,
which more or less remains the same.
11.6 Conclusions
The larger railway station often serves as a significant attraction for vari-
ous activities, such as living, shopping and working. The broad range of
activities on limited zones of land is also referred to as multifunctional
land use. To create such external economies of scale, careful and extensive
cooperation is required from all involved stakeholders. One of the largest
stakeholders is often the (local) government, whose main function is to in-

vest heavily, especially in infrastructure. Therefore, to assess whether
these multifunctional land use projects are beneficial (in terms of social
welfare), insights into direct and indirect benefits for various stakeholders
is essential. This chapter has examined one particular benefit for a specific
group of stakeholders, namely, the additional use-value that shops create
for employees at the Amsterdam South Axis. The reasoning behind the in-
vestigation is that shops situated close to the work location create addi-
tional use-value because they might save valuable shopping time. To ac-
count for idiosyncratic preferences we have controlled for a wide range of
individual characteristics.
The results show that employees attach additional use-value to (future)
shops at the South Axis. In terms of expenditure shares, they expect to
spend on average about 5% at the South Axis. When we take budget shares
into account, the WTP for the presence of this set of shops amounts to
about 21 a month, which is mainly driven by high budget shares for su-
permarkets. Considering the individual characteristics, it seems that in par-
ticular the employees who live near the South Axis have a high WTP for
future shops, mainly because their total use-value is enlarged (they can
visit those shops during weekends as well). Females, youngsters and high
wage earners moreover, all seem to have higher than average WTPs for
shopping facilities, as do employees who work full-time.
In this chapter we have taken a specific type of bias into account,
namely, that if we correct for the number of shops in the residential loca-
tion, then expenditure shares (and hence WTPs) decrease with the number
of shops (except drugstores). This entails that what we exactly measure is
partly a general use-value, or a general accessibility to shopping facilities,
instead of primarily a characteristic of the railway station area concerned.
Controlling for this enables us to specify the exact use-value at the South
Axis for a specific group of employees (in specific residential locations).
This might also be important if employees at the South Axis do not yet live
in fully developed residential areas. If these residential areas begin to ex-
hibit more variety in shops in the future, individual WTP for shopping fa-
cilities at the South Axis may decrease, and thus the perceived benefits the
area surrounding the railway station creates for its stakeholders will de-
crease as well. This scenario also has consequences for cost-benefit analy-
ses, because (indirect) benefits from shopping facilities might be less cer-
tain when created by a general use-value (an increase in accessibility)
instead of a specific characteristic of the railway station area (the multi-
functionality).
However, it seems that shopping facilities at the Amsterdam South Axis

will indeed create additional value (up to 21 a month), which might be
incorporated into a future cost-benefit analysis of the South Axis project.
Acknowledgement
The authors would like to thank H.L.F. de Groot, E.T. Verhoef, F. Bruin-
sma, and an anonymous referee for useful remarks. This research has been
made possible by a BSIK Habiforum research grant. The usual disclaimer
applies.
References
Bannister D (1994) Viewpoint: Reducing the Need to Travel Through Planning.

Town Planning Review 65:349 354
CBS (2004) Statistische Berichten Centraal Bureau voor de Statistiek, Voor-
burg/Heerlen
Coupland A (1997) An Introduction to Mixed Use Development. Spon, London
De Graaff T, Groot HFL de, Rodenburg CA, Verhoef ET (2005) The WTP for Fa-
cilities at the Amsterdam Zuidas. Tinbergen Institute discussion paper TI-
2005-090/3, Amsterdam
Florida R (2002) The Rise of the Creative Class and How it is Transforming
Work, Leisure, Community and Everyday Life. Basic Books, New York
Geoghegan J, Wainger LA, Bockstael NE (1997) Spatial Landscape Indices in a
Hedonic Framework: An Ecological Economics Analysis Using GIS. Eco-
logical Economics 23:251 264
Geurs KT, Ritsema van Eck JR (2001) Accessibility Measure: Review and Appli-
cations: Evaluation of Accessibility Impacts on Land Use Transport Scenarios
and Related Social and Economic Impacts. Rijksinstituut voor Volksgezond-
heid en Milieu (RIVM), Bilthoven
Geurs KT, Wee B van (2004) Accessibility Evaluation of Land Use and Transport
Strategies: Review and Research Directions. Journal of Transport Geography
12:127 140
Irwin EG (2002) The Effects of Open Space on Residential Property Values. Land
Economics 78465 480
Irwin EG, Bockstael NE (2001) The Problem of Identifying Land Use Spillovers:
Measuring the Effects of Open Space on Residential Property Values. Ameri-
can Journal of Agricultural Economics 83:698 704
Jacobs J (1961) The Life and Death of Great American Cities. Penguin, Har-
mondsworth
Johansson PO (1991) Valuing Environmental Damage: Economic Policy Towards
the Environment. Blackwell Publishers, Oxford
Johansson PO (1987) The Economic Theory and Measurement of Environmental

Benefits. Cambridge University Press, Cambridge
Louviere JJ, Hensher DA, Swait JD (2000) Stated Choice Methods: Analysis and
Applications. Cambridge University Press, Cambridge
Nijkamp P, Rodenburg CA, Vreeker R (2003) The Economics of Multiple Land
Use. Shaker Publishing, Maastricht
Priemus H, Nijkamp P, Dieleman FM (2000) Meervoudig Ruimtegebruik: Stimu-
lansen en Belemmeringen. Delft University Press, Delft
Rodenburg CA (2005) Measuring Benefits of Multifunctional Land Use. Thesis,
Vrije Universiteit, Amsterdam
Sheehan K (2001) E-mail Survey Response Rates: A Review. Journal of Com-
puter Mediated Communication 6/2 Online: http://jcmc.indiana.edu/vol6/ is-
sue2 /sheehan.html
Song Y, Knaap GJ (2004) Measuring the Effects of Mixed Land Uses on Housing
Values. Regional Science and Urban Economics 34:663 680
Wee B van (2003) Assessment of Benefits of Multifunctional Land Use. In: Nij-
kamp P, Rodenburg CA, Vreeker R (eds) The Economics of Multifunctional
Land Use. Shaker Publishing, Maastricht
Weisbrod B (1964) Collective Consumption Services of Individual Consumption
Goods. Quarterly Journal of Economic 78:471 477
Part C: High-speed rail and urban dynamics
12 The impact of high-speed railway
developments on office locations: A
scenario study approach
Jasper Willigers
Significance, Leiden, the Netherlands
12.1 Introduction
The introduction of high-speed rail in the Netherlands is accompanied by

large redevelopment projects for several station areas; one such redevel-
opment site is the Amsterdam South axis.
Although high-speed rail is often used as an incentive for large-scale ur-
ban regeneration projects, the manner and extent to which high-speed rail
contributes to the attractiveness of station areas as locations for offices is
as yet unknown. We observe too, that even without high speed connec-
tions, a station area can become more attractive if the quality of the site
improves. For example, the cost-benefit analysis for a new infrastructure
link from Amsterdam and Arnhem to Frankfurt showed that travel time
savings within the Netherlands has a small impact on the attractiveness of
the station areas (Dijkman et al. 2000). Most benefits result from the rede-
velopment activities themselves, e.g. improved local/regional accessibility
and from an image effect. For the High-Speed Line (HSL) South from
Amsterdam and Brussels (and farther on to Paris), however, the travel time
savings are considerably larger, and the cities it connects are seen as inter-
nationally competing cities for which high-speed rail accessibility can be
especially relevant.
Furthermore, uncertainty exists on several aspects of the high-speed rail
network, especially concerning the HSL South. The HSL South will be-
come operational by the end of 2007, but several alternatives are still pos-
sible with regard to how the train services are scheduled. Issues that re-
main susceptible to change are predominantly the height of the fare
supplement for the high-speed train services, and whether the HSL South
238 Jasper Willigers
services have their end stop at Amsterdam Central station or at Amsterdam

South station. In addition, the introduction of other new high-speed train
services (for instance to the north of the country) can be relevant for loca-
tion choice and accessibility.
The analyses presented here intend to clarify how the HSL South influ-
ences the attractiveness of station areas for offices and how attractiveness
depends on the choices concerning the train services on this new line, the
Amsterdam South axis. Although the aim of the South axis is to attract in-
ternational head offices, given the experience in France (Sands 1993;
Mannone 1997), it is likely that most employment effects will be intrare-
gional distributional by attracting offices previously located at sites within
Amsterdam, or elsewhere in the Randstad. Our analyses are limited to
these intraregional effects.
Because of aforementioned alternatives existing for high-speed train
services, the study described here takes the form of a scenario study. Each
scenario is composed of a different combination of the alternatives and the
analyses are restricted to HSL South alternatives. The scenario study
should be regarded as an explorative analysis used to reveal changes in the
general image of the attractiveness of locations for offices and does not
give exact forecasts of transport and land use effects for a specific year. In
each scenario we distinguish among three aspects of location attractive-
ness:
The potential accessibility of locations in order to convey the
size of the labour and product markets, and thus the potential for
spatial interaction;
The attractiveness of cities (defined as micro-economic utility
for location decision-makers) in order to evaluate the role of
high-speed rail in the competition between cities and;
The attractiveness of different locations (again defined as util-
ity) within or near Amsterdam, to analyse the relative attrac-
tiveness of different office sites in the same city.
Attention is given to how location attractiveness and the impact of high-
speed rail differ among types of offices. These aspects together allow us to
draw conclusions for the expected role of high-speed rail in the attractive-
ness of locations for offices. The next section gives more detail on the sce-
narios that are evaluated. Thereafter the results of the scenarios are dis-
cussed for potential accessibility, the choice of an urban region for an
office establishment, and attractiveness of locations within the Amsterdam
urban region. The final sections integrate and discuss the findings reported
here.
The impact of high speed railway developments on office locations 239
12.2 Description of scenarios
12.2.1 Scenarios for the future HSL South high-speed railway
The HSL South, a high-speed railway line under construction at the time of
this writing, will offer services in 2007 or 2008 (Het Financile Dagblad
2005; Min. V&W 2005), and two types of high-speed train service will use
this rail infrastructure: i.) international Thalys services from Amsterdam
to Belgium and France with stops at Amsterdam Central station, Schiphol
airport, and Rotterdam Central station (the current stop in The Hague will
discontinue when the HSL South is put into service), and ii.) high-speed
train services for shorter distances on two trajectories: Amster-
dam/Rotterdam/Breda and The Hague/Rotterdam/Breda/Belgium (Min.
V&W 2005). In this scenario study for the international services we take
only connectivity into account. Domestic services are expected to be a vi-
able option for more trips, so its effect on potential accessibility may be
larger; these train services are accounted for via both a connectivity effect
and a potential accessibility effect.
The frequencies of domestic rail services are relatively high for the Am-
sterdam Rotterdam connection, where four trains per direction per hour
are planned (Rijkswaterstaat 2006). At Rotterdam half of these services
will go through to Breda, but The Hague/Breda services are far less fre-
quent, with one train per direction per hour. These trains travel farther to
Brussels, Belgium, but we do not include this in the analysis. Figure 12.1
gives an overview of the assumed in-vehicle travel times and service fre-
quencies.
The high-speed train services on the new infrastructure will serve Am-
sterdam Central station. In future, an alternative will be to offer services
running from the station at the Amsterdam South axis. From this station
travel time along the HSL South is seven minutes shorter than from Am-
sterdam Central station, as indicated in Figure 12.1. We observe that an
Amsterdam South axis connection would provide a higher gain in potential
accessibility; however, by having fewer conventional train services than
Amsterdam Central, its accessibility effect on the wider region cannot be
identified straightforwardly. In our case study the two possible locations of
the Amsterdam high-speed train station are analysed as different scenarios.
Figure 12.1. In-train travel times and frequencies (trains per direction per hour)
between stations as assumed for the future domestic services on the HSL South
(based on Rijkswaterstaat 2006)
Another relevant issue for the HSL South scenarios is the fare supple-
ment charged to travellers who use the high-speed train services. This sup-
plement can have a large impact on the accessibility effect, and several
supplement schemes are possible; but in this case we will discuss two fare
structures: a percentage addition to a non-high-speed fare on the same
connection, and a fixed supplement above the non-high-speed fare. The
fixed supplement is currently in force on the Amsterdam
/Arnhem/Germany service at 2. We assume the same supplement for the
HSL South services. Two supplement percentages, 10% and 25%, are
evaluated here.
The alternatives discussed above have resulted in a total of six evaluated
scenarios (see Figure 12.2). In order to reduce the number of possible sce-
narios, the HSL South station choice and fare supplement alternatives are
modelled for only one scenario each.
Railway network Station for HSL South services Fare supplement
Current situation (reference) Not applicable Not applicable
HSL South Amsterdam Central station No

10%
25%
2
Amsterdam South station 10%

Figure 12.2. Scenarios evaluated as a combination of different railway develop-
ments
12.2.2 The Amsterdam South axis area
As mentioned above, our focus is predominantly on Amsterdam and the

Amsterdam South axis area in particular. At this location a high-status of-
fice park is created to attract international head offices (Rienstra and Riet-
veld 1999). However, the location has to compete with several other office
locations in the Randstad for which redevelopment projects are also un-
derway. These locations include the central station areas of Rotterdam and
Utrecht. But the South axis also faces competition from other locations
within the Amsterdam urban region, most notably Schiphol airport and the
Amsterdam Central station area (Figure 12.3).
12.3 Methodology
The scenario study in this chapter uses models based on empirical data of
travel behaviour and location choices. In this section we describe the gen-
eral outline of the models in concise terms. Two distinct models: one for
determining potential accessibility and another for location choices of of-
fice decision-makers, are described below.
12.3.1 Potential accessibility
Potential accessibility represents the total number of possible spatial inter-

actions and can be interpreted as the size of a market, e.g. a labour market
or product market. In this chapter we use indicators that combine the travel
impedance (or travel resistance due to travel time, cost, etc.) with the mode
Figure 12.3. Amsterdam with railway lines and stations
categories car, train and other public transport. Impedance functions are
derived from the travel disutilities of a multinomial logit model (e.g.
McFadden 1974; Ben-Akiva and Lerman 1985) of combined mode and
destination choice. The observed utility for train travel is hereby a function
of station-to-station generalised cost (the train fare plus the monetary
equivalent of the travel time), station access distance, and station egress
distance. Waiting time, as part of the station-to-station travel time, is half
the service headway and is weighted as twice the in-train travel time. Pa-
rameters are estimated based on the Dutch national travel survey for the
year 2000.
For commuting, the potential accessibility indicator represents the size
of the labour market. A gravity-type of potential accessibility indicator is
used and formulated as:
Ai = D j exp Vijm [ ]
j m
where Ai is potential accessibility of zone i; Dj is the employment popu-

lation in zone j, and Vijm is the observed utility for travelling from zone i to
zone j by mode category m.
For business travel a contour indicator is found to be more relevant in
the location choice model. This type of indicator represents long-distance
travel impedance well, whereas gravity-type indicators put most emphasis
on intra-urban travel. The contour indicator sums all destinations within a
certain travel impedance value:

f ij = 1

exp[V ] I
ijm max
Ai = D j f ij
m
j f ij = 0
exp[V ] > I
m
ijm max
where Imax is a critical travel impedance, determined empirically in the

location choice model, and Dj is the number of jobs in zone j; other terms
are as above.
12.3.2 Office location choices
A location choice model for office decision-makers is used to determine

the attractiveness of locations. Locations are hereby represented by grid
cells in a geographical information system. A nested logit structure
(McFadden 1978; Ben-Akiva and Lerman 1985) is used to make a distinc-
tion between the intraregional and interregional level. Data on stated
choices (to study high-speed train connectivity) and revealed choices (for
giving the model a realistic base) are combined in a single location choice
model by using the full-information maximum likelihood technique
(Louviere et al. 2000).
Location attractiveness is represented in the model by a number of ac-
cessibility and non-accessibility characteristics. Three types of location at-
tributes are distinguished here:
Potential accessibility indicators as above;
Connectivity: distance to a station, level-of-service for this sta-
tion, travel time to a motorway ramp;
A typology of land uses (following Maat et al. 2005), which
represents both the type of urban environment (e.g. city centre,
business site) and the density of land use.
Utility functions are thereby assumed to be different for distinct seg-
ments of offices. Three segmentation criteria are found to be most relevant
for the current research:
Being in the financial or business services branch of industry;

Having employees who make international business trips regu-
larly and;
Having a national or international spatial orientation to custom-
ers.
The second aspect is essential to our topic, because it largely determines
the relevance of international high-speed train services at a station. The
third aspect takes into account where the majority of an offices customers
are located, and thus the time taken for business trips. The contour acces-
sibility indicator for business travel is assumed to be relevant only for na-
tionally and internationally-oriented offices.
This model is used first to study the attractiveness of cities in the Rand-
stad and then to compare different office sites in the Amsterdam urban re-
gion: Amsterdam city centre, Amsterdam South axis, and Schiphol airport.
Since the modelled probability of choosing individual grid cells is accom-
panied by a relatively large uncertainty, the intraregional differences in lo-
cation attractiveness are presented aggregated to larger areas (see Fig-
ure12.3).
12.4 Potential accessibility
Accessibility is directly affected by new high-speed railways or other

transport infrastructure developments. Accessibility is therefore the highest
priority when evaluating location attractiveness. The first subsection below
discusses the results for the HSL South with a 10% fare supplement as a
basic HSL South scenario. Thereafter we attend to the potential accessibil-
ity effects of three aspects of high-speed rail developments in the Nether-
lands in relation to the first high-speed rail scenario: the influence of the
fare supplement, the effect of which station in Amsterdam receives HSL
South services, and the impact of a second high-speed rail link from Schip-
hol to Groningen.
12.4.1 Accessibility effect of the HSL South
In gravity-type accessibility indicators short-distance travel typically has a

larger impact on the accessibility score than long-distance travel, because
most impedance functions are less steep for higher impedance values. The
highest accessibility scores can therefore be found in large cities, as can be
seen in Figure 12.4.
Figure 12.4. Reference accessibility score towards potential employees
Figure 12.5. Change in accessibility score towards employees due to the domes-
tic HSL South services
Figure 12.6. Reference accessibility score towards potential business partners
Figure 12.7. Change in accessibility score towards business partners due to the
domestic HSL South services
Figure 12.5 shows the influence of the HSL South on potential accessi-
bility for employees. Here we assume a 10% fare and a stop at Amsterdam
Central station; this serves as a reference for the other HSL South scenar-
ios. Compared to overall accessibility the accessibility increase is small;
relative to accessibility for train alone in the reference scenario, the in-
crease is significant but not very large.
The Amsterdam and Rotterdam regions are the main gainers of the HSL
South. Remarkably, within Amsterdam, the South axis area has the highest
absolute increase in potential accessibility, even in this scenario where
there is not yet direct high-speed train services from its station. Other
places within the Randstad region scarcely or do not at all benefit from the
high-speed railway for potential accessibility to employees. Cities and
towns between Amsterdam and Rotterdam are bypassed by the high-speed
trains (the so-called tunnel effect of high-speed railway infrastructure,
Plassard 1991; Vickerman 1997), whereas for places farther north and
south of the line, the effort to get to the high-speed trains is often too large
to have an advantage over conventional rail. For The Hague the impact on
the accessibility score is also negligible due to the very low frequency of
services that stop there.
The contour-based indicators show a more dispersed image of accessi-
bility, as can be seen in Figure 12.6 for the reference scenario. Figure 12.7
shows the impact of the HSL South domestic services on the contour po-
tential accessibility indicator. We see from the figure that the accessibility
effect for this indicator is much larger than for the gravity-type indicator.
Accessibility increases are calculated up to 6.7% of the maximum accessi-
bility in the reference situation. Furthermore, high increases in accessibil-
ity are less restricted to the cities with high-speed train stations; several cit-
ies and towns with good train connections to the high-speed stations are
also seen to gain accessibility.
12.4.2 Influence of the fare supplement
The travel time effect of a new high-speed train service seems normally
quite robust, but for the train fare much more variability is possible. Train
supplements can nevertheless reduce the attractiveness of using a high-
speed rail link, so it seems relevant to consider the impact of this supple-
ment on the potential accessibility effect. Figure 12.8 illustrates the effect
of different fare supplements on potential accessibility for locations along
the HSL South, both for commuting and business travel. Gravity-type in-
dicators are used here for both trip purposes.
Figure 12.8. Effect of fare supplements for high-speed train services on potential
accessibility (gravity-type indicator) increase
Commuters on average have a relatively low value of time, which

means that, for many connections, the fare supplement is not fully com-
pensated by a decrease in train travel time. The fare supplement therefore
has a high impact on the accessibility effect of the high-speed train ser-
vices. For the gravity indicator for business travellers (who value time
higher) the fare has much less influence.
Remarkably, it can be seen from Figure 12.8 that the impact of the sup-
plement is not equal in size for different locations. For example, the in-
crease of commuting accessibility without a fare supplement is for Am-
sterdam city centre about as high as for Schiphol airport; but if a
supplement is introduced then the accessibility effect for Amsterdam is
lower than for Schiphol. The reason is because for a trip from Amsterdam
Central station to Rotterdam, the fare has a higher share in the generalised
cost than for a trip from Amsterdam South station to Rotterdam. The most
efficient trip (for an average traveller) from Amsterdam South station has a
transfer at Schiphol airport station, while for trips from Amsterdam Central
station such a transfer is avoided, but a fare supplement is accounted for
the trip section to Schiphol. Note that transfers cost time (valued as twice
the in-vehicle-time), but not money.
Another noteworthy component from Figure 12.8 is that the type of sup-
plement, either as a percentage addition or as a fixed supplement, does not
lead to significantly different accessibility effects. For all locations the ac-
cessibility score for the 2 fixed fare is just above the score for a 25% ad-
dition. Thus, for trips on which the HSL South offers an advantage over
conventional trains, the 2 distance-independent supplement is on average
below 25% of the basic fare.
In the remainder of this study we assume a fare addition of 10% of the
basic fare. This is not a high fare that precludes noteworthy accessibility
effects, but it is nevertheless high enough to consider its network effects.
For the analyses in the next sections it should be borne in mind that the re-
sults can be sensitive to both the structure and the height of the fare.
12.4.3 Influence of HSL South services from the Amsterdam

South axis rather than Amsterdam Central station
The potential accessibility results showed that even in the situation with a
stop at Amsterdam Central station, the South axis already has a higher in-
crease in accessibility than the city centre of Amsterdam. For the same rea-
son the possible shift of high-speed train services to the Amsterdam South
station offers little extra advantage for the South axis. The high frequency
of trains between Amsterdam South station and Schiphol makes a transfer
less problematic, while the fare supplement for high-speed train services
and the absence of a travel time gain until Schiphol airport decrease the
advantage of a direct connection from the South axis. Direct high-speed
train services from the South axis therefore result in only a small increase
in the gravity-type accessibility indicators for the South and Southeast of
Amsterdam. The decrease in travel impedance for the 2 contour indicator
is not enough to result in any effect for the South axis area.
In addition to this weak positive effect for the South axis area, accessi-
bility decreases for most other places in the Randstad. In the Amsterdam
region the city centre is directly influenced by the loss of the direct con-
nection. As in the Netherlands train fares per kilometre decrease with the
distance travelled by train, the fare supplement is lower relative to the
travel time on the Schiphol/Amsterdam Central station connection than on
the Schiphol/Amsterdam South station line. But the same applies for most
other places in the Amsterdam region: a stop at Amsterdam Central station
is most advantageous, because it is better connected to the regional train
network than Amsterdam South station. For Rotterdam and its surround-
ings all indicators also give a higher accessibility for the stop at Amster-
dam Central station. The direct connection to Amsterdam South offers al-
most no advantage for a trip from Rotterdam to Amsterdam South axis
(and/or beyond), and the increased travel impedance to Amsterdam city
centre results in a lower accessibility score for Rotterdam.
We can conclude that several network effects play a role in how the
choice between Amsterdam Central station and Amsterdam South station
influences the attractiveness of locations within and outside the Amster-
dam region. The most important role is for the trade-off between an extra
transfer at Schiphol and the fare supplement of high-speed train services
(assumed here to be a percentage of the basic train fare), which has differ-
ent outcomes per origin-destination pair. For the Amsterdam South axis a
stop at the Amsterdam South station yields almost no accessibility benefit,
but does result in a lower accessibility increase for other locations in the
Randstad. Overall this has a negative influence. However, for the competi-
tiveness of the South axis it is an advantage. On an interregional scale the
urban region of Amsterdam as a whole benefits less from the HSL South,
as does the Rotterdam region. But on an intraregional scale the South axis
area is more competitive towards other locations in Amsterdam. For the
remainder of this chapter we assume that high-speed train services on the
HSL South serve Amsterdam Central station unless explicitly stated oth-
erwise.
12.5 Probability of region choice
The potential accessibility indicators give a first indication of how high-

speed rail influences location attractiveness, but a more complete image
with both centrality and connectivity aspects can be achieved by applying
a location choice model. This section focusses on the choice of an urban
region. The next subsection discusses the resulting average region choice
probabilities for different scenarios, thus compromising the main part of
the scenario study at the regional level and, in particular, the evaluation of
the interregional distributive effect of different high-speed train network
alternatives. Thereafter, the region choice of relevant sub-populations of
offices is studied on the basis of segmentation criteria in the model. Our
aim is to clarify the variation in the location choices between different

types of offices.
12.5.1 Average region choice probabilities
The four main cities in the Randstad area are most frequently chosen, ac-
cording to the location choice model (see Figure 12.9), as might be ex-
pected. In absolute figures Utrecht is most often chosen partly because this
region is the largest of all urban regions in the study area. A better indica-
tor might be the choice probability proportional to the total number of
choice options in the region; according to this, a location in The Hague is
on average chosen most often, with Utrecht second. Amsterdam and Rot-
terdam have a much lower relative probability, because both regions also
include peripheral, less attractive areas. Apart from the four main cities,
the regions in the peripheral north of the study area show a very low prob-
ability of being chosen by businesses for office space.
Figure 12.9. Probability of region choice (in percentages) for the reference case.
Grayscale indicates probabilities relative to the regional number of opportunities
Figure 12.10. Change in probability of region choice in percentage points by the

HSL South. Grayscale indicates the change relative to the reference case
Figure 12.10 illustrates how the region choice probabilities change when
the HSL South is implemented. The Amsterdam region has the highest
gain in accessibility, while Rotterdam and the southeast of the province of
Zuid-Holland also benefit in attractiveness. All other regions lose their
probability of being chosen; this loss is extensive, particularly for Utrecht.
Due to the HSL South the increase in region choice is 2.7% of the refer-
ence scenario for the Amsterdam region and 1.3% for Rotterdam. These
are not high figures, especially given that the research population of firms
and institutions has been pre-selected to be potentially sensitive to train ac-
cessibility in general and high-speed train accessibility in particular.
Within this population, however, further differences exist between types of
offices. We next examine region choices of a number of relevant sub-
groups of offices.
12.5.2 Heterogeneity in region choice probabilities
The location choice model takes into account several segmentation criteria
that are a cause of heterogeneity among the location preferences of offices.
This taste heterogeneity causes different types of offices to be unevenly
distributed over the urban regions. Another consequence is that the out-
come of the high-speed rail scenarios is different per office. The segmenta-
tion is meant to study this consequence of heterogeneity.
The results of the reference case and HSL South scenario for different
segments of offices are shown in Table 12.1. A first segmentation criterion
is the offices branch of industry, which mainly influences sensitivity to
the type of urban environment, but also depends on the importance of a
stations total train frequency. Offices in the business and financial ser-
vices are more susceptible to a stations train frequency than other offices.
However, total train frequency is only marginally increased at stations in-
tended to receive new high-speed train services; this explains why the per-
centage changes in region choice for offices in the financial and business
services are broadly comparable to the average values.
Table 12.1. Differences between office segments: forecasted probabilities for re-
gion choices per segment and the change due to the HSL South scenario
Branch of industry is financial or business services
Reference With HSL South
Amsterdam 17.5% 18.0% (+2.9%)
Rotterdam 13.1% 13.4% (+1.8%)
The Hague 13.3% 13.3% (+0.1%)
Utrecht 24.5% 24.0% (1.9%)
Other 31.6% 31.3% (0.9%)
Has employees who make regular international business trips
Amsterdam 16.8% 17.6% (+4.7%)
Rotterdam 13.9% 14.2% (+2.2%)
The Hague 11.9% 11.8% (0.8%)
Utrecht 24.9% 24.3% (2.4%)
Other 32.5% 32.1% (1.3%)
Spatial orientation to customers is national or international
Amsterdam 16.3% 16.9% (+3.8%)
Rotterdam 12.7% 13.1% (+2.6%)
The Hague 12.0% 11.9% (0.8%)
Utrecht 28.4% 27.6% (2.5%)
Other 30.7% 30.5% (0.5%)
Percentages between brackets are the changes in choice probability relative to the
reference case.
A second segmentation criterion is whether or not an office has employ-

ees who make regular international business trips. The HSL South services
are on average a larger influence for this segment than for all offices. The
Amsterdam region especially benefits from the high-speed train, because it

receives two stations with international high-speed train services.
Finally, the third heterogeneity aspect considered in Table 12.1 is the
spatial orientation to customers. Offices with customers or clients located
across the country and abroad are assumed to be sensitive to long distance
accessibility. The Utrecht region dominates for this type of office because
of its central location, both geographically and within the Dutch railway
network. With the arrival of the HSL South, Rotterdam and the region
around Dordrecht (southeast of Rotterdam) predominantly benefit most.
If a distinction is made between different segments, then the effect of
high-speed train services can be seen to be larger for some sub-populations
of offices than for others. Spatial orientation and travel behaviour are rele-
vant in this respect. The changes in region choice probabilities reported in
Table 12.1 are modest nonetheless. We can expect that changes in choice
probabilities on the intraregional level will be larger. This subject is ad-
dressed in the next section.
12.6 Attractiveness of locations within the Amsterdam

urban region
We are now interested in examining the effect of the high-speed train sce-
narios for competing locations within the Amsterdam region. Intraregional
location choices are analysed by simulating location choices between all
GIS grid cells available in the Amsterdam region. Grid cell probabilities
and changes in these probabilities are aggregated to larger areas: Amster-
dam city centre, South axis and Schiphol airport, (shown as locations) in
Figure 12.3 above.
We first analyse the modelled average choice probabilities of some rele-
vant locations in the Amsterdam region for different high-speed rail sce-
narios. Thereafter, we focus on the variation in taste heterogeneity for in-
traregional location choices. Finally, we explore an aspect that influences
the location choices at this spatial scale: the choice of station in Amster-
dam to be used for HSL South services, and the impact of a land use
change at the Amsterdam South axis.
12.6.1 Average probabilities for intraregional location choices
Within the Amsterdam urban region the South axis area competes with
many other locations including the city centre and Schiphol airport. Each
location benefits from the HSL South. Figure 12.11 shows how the aver-
age utility at these three sites increases with the implementation of the
HSL South. Utility is used here as a measure of attractiveness. Since utility
is a relative quantity and its absolute increase does not have a readily in-
terpretable meaning, the utility in the figure is normalised to the change for
the Amsterdam South axis.
Figure 12.11. Increase in average location utility due to HSL South services
Amsterdam city centre gains most in attractiveness. It benefits from both

a large connectivity effect and a centrality effect; whereas the South axis
has only a centrality effect. Schiphol has an approximately equal connec-
tivity increase as Amsterdam city centre but a lower gain in potential ac-
cessibility. In the direct vicinity of the Schiphol station the connectivity ef-
fect is larger than for Amsterdam city centre. Since high-speed trains stop
at Schiphol in two directions, the increase in train frequency there is twice
that of Amsterdam Central station, the end point of the service. In addition,
the initial frequency is higher for Amsterdam Central station than for
Schiphol; higher initial frequencies will lower the effect of additional train
services. However, at Schiphol airport itself little space is available for of-
fices. Therefore in this section a somewhat larger area is used that does not
fully fall within Schiphol stations catchment area.
The utility per grid cell results in a choice probability. Table 12.2 shows
the calculated probabilities for location choices conditional on the region
choice. In the reference case the city centre of Amsterdam is the most of-
ten chosen location. The high level-of-service of Amsterdam Central sta-
tion is an important factor for this, as is the high potential accessibility,
which compensates for poor motorway connectivity. The Schiphol zone
has a low probability, because in this area relatively little space is available
for offices, and because the self-potential for commuting accessibility is
low. Finally, because the Amsterdam region is a large region, there is also
a considerable probability for choosing a site other than one of the three
examined here.
Table 12.2. Forecasted changes in average probabilities for location choices

within Amsterdam due to different high-speed rail scenarios
Location Reference With HSL South
South axis 7.3% 7.3% (0.9%)
City centre 8.8% 10.1% (+15.2%)
Schiphol 1.8% 2.0% (+14.6%)
Other 82.1% 80.6% (1.9%)
reference case.
A stop at Amsterdam Central station benefits the city centre of Amster-

dam and Schiphol the most in the HSL South scenario. Other locations, in-
cluding the South axis, are less often chosen conditional on the region
choice, even though these also have an increase in potential accessibility.
The decrease in intraregional choice probability is lower than the increase
in region choice probability (which was 2.7%). So in total, the choice
probabilities for these locations increase on average. From these results we
can conclude that the intraregional distributive effect of high-speed rail is
likely to be substantial for the Amsterdam region. The next subsection ad-
dresses the extent to which these location choices differ between the seg-
ments in the model.
12.6.2 Heterogeneity in intraregional location choice

probabilities
Similar to the choice of a region for an intraregional location, choice taste

heterogeneity can be of influence. Table 12.3 shows the intraregional
choice probabilities subdivided to the same segments as in Table 12.1. The
South axis and city centre are considerably more attractive to offices in the
financial and business services than for other offices. This is mainly
caused by the sensitivity of the different types of offices to the type of ur-
ban environment. For accessibility attributes less diversity exists between
these branches of industry, although Schiphol gains relatively more for the
business and financial services than for other branches of industry.
International business trips make an office much more responsive to
connectivity to international high-speed rail services. For this type of of-
fice the HSL South results in a large increase in the probability that the
city centre and Schiphol airport will be chosen, mainly at the expense of
the South axis, which in combination with the region choice probability,
suffers a net loss in location choice probability for this type of office.
Table 12.3. Differences between office segments: forecasted probabilities for lo-
cation choices within Amsterdam per segment and the change due to the HSL
South scenario
Branch of industry is financial or business services
South axis 8.6% 8.4% (2.3%)
City centre 12.2% 14.1% (+15.1%)
Schiphol 1.5% 1.8% (+19.0%)
Other 77.6% 75.7% (2.5%)
Has employees who make regular international business trips
South axis 7.2% 6.8% (5.3%)
City centre 8.6% 11.0% (+28.1%)
Schiphol 1.8% 2.3% (+27.0%)
Other 82.4% 79.9% (3.0%)
Spatial orientation to customers is national or international
South axis 7.5% 7.4% (1.4%)
City centre 9.4% 10.8% (+15.9%)
Schiphol 1.8% 2.1% (+16.8%)
Other 81.4% 79.7% (2.1%)
reference case.
The results for offices with national or international spatial orientations

are quite comparable with the average probabilities. The spatial orientation
interacts with the availability of intercity services and the contour indicator
for business travel potential accessibility. Intercity services do not change
in this scenario, while the potential accessibility indicator impacts most at
an interregional level rather than for the intraregional location choices. For
the HSL South scenario, Schiphol gains more offices with a national or in-
ternational orientation to customers, but the difference is not large.
The only office characteristic in this model that really leads to a substan-
tially larger impact of the HSL South at the intraregional level is the rou-
tine international business trip. The South axis becomes considerably less
attractive for this type of office relative to the city centre. However, this
outcome is conditional on the choice of Amsterdam Central station as the
end stop for HSL South services. The next subsection explores how these
results change if Amsterdam South station is chosen.
12.6.3 Choice between Amsterdam Central station and

Amsterdam South station for HSL South services
In subsection 12.6.1 we concluded that, with the alignment of the railway

network as in the year 2000, a stop of the HSL South services at the Am-
sterdam South station leads to a marginally higher potential accessibility
for the South axis. But it also results in a substantially lower potential ac-
cessibility for the city centre and some other locations in the Amsterdam
region. Although this has a negative influence on the Amsterdam region as
a whole, it makes the South axis more competitive towards other locations
in Amsterdam. This is even more the case if connectivity is also taken into
account. Table 12.4 shows how the intraregional location choice probabili-
ties change according to which station is chosen. If HSL South services
stop at the Amsterdam South station then this stations environment is
about as equally chosen as the city centre. Schiphol is chosen more often
than in the case with a stop at Amsterdam Central station, because it now
has less severe competition from other locations in the Amsterdam region;
overall we conclude from these results that a stop at the South axis would
be beneficial for its development impact.
Table 12.4. Impact for location choices within Amsterdam of the chosen station
for HSL South services
Location Reference Stop at Central station Stop at South station
South axis 7.3% 7.3% (0.9%) 8.8% (+20.7%)
City centre 8.8% 10.1% (+15.2%) 8.6% (2.0%)
Schiphol 1.8% 2.0% (+14.6%) 2.1% (+16.4%)
Other 82.1% 80.6% (1.9%) 80.5% (2.0%)
Another factor of uncertainty in addition to the choice of the end station

for HSL South services are development plans under consideration or al-
ready under construction. The discussion below will synthesise accessibil-
ity and location results and draw conclusions relating to the literature. We
first give an integral overview of the results and illustrate the differences
between scenarios. Thereafter we discuss the role of high-speed rail in the
competition between cities, and finally we examine how high-speed rail
has impacted on the attractiveness of different sites within the same city.
12.7 Synthesis of scenario results
As noted earlier, the purpose of the models used in this scenario study is to
investigate the structure of changes in the attractiveness of locations within
the Randstad area rather than to forecast precisely for a future situation.
Results are therefore discussed in a qualitative way. Table 12.5 gives an
overview of the high-speed rail scenario results. The impact of high-speed
rail on potential accessibility is moderate, since long-distance trips account
for only a small proportion of all trips. A specific indicator for long-
distance accessibility shows a larger increase, but this indicator is assumed
to be valid only for a particular segment of offices having a national or in-
ternational orientation. The accessibility benefit is limited to Amsterdam,
Rotterdam and nearby cities with good train connections to Amsterdam or
Rotterdam.
Table 12.5. Overview of the effects of HSL South services for different scenarios
Region/location HSL South only
Potential accessibility Location choice
Interregional Amsterdam + +
Rotterdam + +
The Hague 0 0
Utrecht 0
Other 0
Intraregional Amsterdam South axis +
Amsterdam city centre + ++
Schiphol + ++
Other Amsterdam +
Location choices are based not only on potential accessibility but also
on aspects of connectivity. The connectivity effects further enhance the at-
tractiveness of the Amsterdam and Rotterdam regions. We can also ob-
serve for all high-speed rail scenarios a decrease in the choice probability
for other locations within the Randstad. This is especially true for Utrecht,
because many of the offices there are sensitive to long-distance accessibil-
ity. Compared to the choice of region, the intraregional location choice re-
acts stronger to a change in the accessibility attributes; this is a conse-
quence of the spatial hierarchy in location choices that was found to be
relevant. Locations within Amsterdam lacking an accessibility increase in
the high-speed railway scenarios suffer a larger loss in choice probability
than similar locations in other regions.
The scenario outcomes are susceptible to the types of offices considered.
In the location choice model three office characteristics are found that ex-
plain taste heterogeneity within the data set. These segmentation criteria
have an influence on different aspects of the location choice. Table 12.6
gives an overview of this. The branch of industry is a commonly used
segmentation criterion in location choice studies and models (e.g. Abra-
ham and Hunt 1999; Waddell and Ulfarsson 2003). In the current applica-
tion it indeed has a profound influence on the distribution of economic ac-
tivities, both on an interregional and intraregional level. However, it is a
less distinctive factor for reactions in location preferences due to a change
in accessibility. On the other hand, the two accessibility-related segmenta-
tion criteria are of larger influence on the responsiveness of location
choices to accessibility, but these have less impact on the spatial distribu-
tion of activities in the reference case. The spatial orientation of an office
therefore has greater influence on the interregional scale than on the in-
traregional scale.
Table 12.6. Influence of office characteristics on location choices and on the im-
pact of high-speed rail
Characteristic Reference re- Region choice Reference lo- Location
gion choice responsiveness cation choice choice respon-
siveness
Branch of industry is ++ + ++ +
financial or business
services
Has employees who 0 ++ 0 ++
make regular interna-
tional business trips
Spatial orientation to + ++ 0 +
customers is national
or international
The scenario study in this chapter has also expressed the sensitivity of
the model results for several aspects relating to the high-speed rail imple-
mentation. Firstly, if Amsterdam South station is used for HSL South ser-
vices, then the effect on location choice is inclined towards the South axis
but is less positive for other locations in Amsterdam as well as for Rotter-
dam. However, results for this option can improve if Amsterdam South
station receives more and better connections to other stations within the
region. A second aspect that influences the model outcomes is the height
of the fare supplement for travelling by high-speed train. This fare is found
to have a considerable restrictive effect on the potential accessibility in-
crease of sites for which the new high-speed rail services are an option.
High-speed rail is found to have an impact on office location choices at

different spatial scales. Different issues are relevant at the regional scale
than at the urban scale. The next two sections further elaborate on the re-
sults of the scenario study for these two scales respectively, and relate
them to concepts in the literature.
12.7.1 High-speed rail and competition between cities
An important motive for building the HSL South is the presumed impor-
tance for the Randstad to be connected to the Western European high-
speed railway network in order to gain and/or preserve the international
business service offices (Min. V&W 1994). Therefore, Amsterdam, Rot-
terdam and The Hague are seen as internationally competing cities. The
possible connection of The Hague to the high-speed railway network was
consequently an important point of debate for policymakers when the HSL
South was given the go-ahead (see Min. V&W 1994). However, competi-
tion also takes place between cities in the Netherlands. This makes the ef-
fect of the HSL South very relevant for office location choices within the
Randstad.
Previous accessibility studies on the Trans-European Networks program
(e.g. Gutirrez et al. 1996; Spiekermann and Wegener 1996) have already
shown that, on a European scale, the main metropolises gain most from
new high-speed railway links. The scenario study elaborated here shows
similar results. Within the Randstad region Amsterdam and Rotterdam, the
two largest cities in the study area, benefit from their mutual increased po-
tential. More high-speed train stops in the Netherlands would lead to a less
concentrated impact from the high-speed rail, but would at the same time
cause a crowding-out effect because of the limited number of offices af-
fected by high-speed rail.
For The Hague an absence of international high-speed train (Thalys)
services results in a small loss in attractiveness for internationally-oriented
offices. However, other cities such as Utrecht suffer much more from be-
ing bypassed.
High-speed rail in the Netherlands leads to an increased competitiveness
of the main urban centres (Amsterdam and Rotterdam) compared to other
cities in the Randstad, and similar to the studies at the European level. As
shown in Table 12.6 this consequence is especially true for nationally and
internationally-oriented offices. For other types of offices high-speed rail
has less influence.
12.7.2 Development at high-speed train station sites
In the literature it is suggested that the employment location effect of high-

speed rail is largely intraregional distributive (Sands 1993; Mannone
1997). Furthermore, e.g. Bonnafous (1987) and Mannone (1997) have
concluded from entrepreneurial surveys that high-speed rail itself has
played only an inferior role in the development of the high-speed train sta-
tion areas.
In order to answer the question to what extent high-speed rail contrib-
utes to the attractiveness of a location, the results of the scenario study can
be evaluated in a site-oriented way. As shown by the results in section
12.5, a site with a high-speed train connection will be chosen (conditional
on region choice) around 20% more often than the reference situation.
Compared to the case where a competing location has a high-speed train
connection, the impact of that connection is of course larger. It can thus be
seen that the influence of high-speed rail on location choices is much more
apparent at the intraregional level than on the interregional level.
The question of how high-speed rail influences location attractiveness
seems more appropriately answered with an orientation towards the office
decision-makers. From the results we see that the number of offices (for
which high-speed rail is of decisive importance) is not very large. The ma-
jority of offices that choose a high-speed train station site would also have
chosen this location in a situation without high-speed rail, although some
of these offices would choose a competing site if it too had a high-speed
train connection. Variation in the relevance of high-speed rail for offices is
also visible. With high-speed train services at Amsterdam Central station
the city centre is most attractive for offices whose employees regularly
make international business trips, due to the international high-speed train
connectivity. Nationally-oriented offices on the other hand are more influ-
enced by the potential accessibility effect, which is higher for the South
axis.
12.8 Conclusions
Overall we can conclude that high-speed rail has a remarkable effect on

the attractiveness of locations, especially at the intraregional level. The
connectivity effect is dominant; this is a combination of the option value
and image effect characteristics of high-speed rail. Travel time advantages
are expressed as increases of potential accessibility, which is most influen-
tial at an intraregional level (e.g. the competition between cities), but less
so for the development of specific sites. However, this potential accessibil-
ity effect can suffer markedly if a high fare supplement is charged. Finally,
given the South axis connection to the high-speed train network, analysis
of the data indicates that it is not essential in order to attract sufficient em-
ployment in general; but if the aim is to specifically attract internationally-
oriented offices, a high-speed train connection is of considerable signifi-
cance.
Acknowledgement
This chapter is based on research funded by Connekt-NWO as part of the

stimulation program Gebruik en waardering van vervoersnetwerken (Use
and valuation of transport networks).
References
Abraham JE, Hunt JD (1999) Firm Location in the MEPLAN Model of Sacra-
mento. Transportation Research Record 1685:187198
Ben-Akiva M, Lerman SR (1985) Discrete Choice Analysis: Theory and Applica-
tion to Travel Demand. The MIT Press, Cambridge
Bonnafous A (1987) The Regional Impact of the TGV. Transportation 14:127
137
Dijkman H, Koopmans CC, Vromans M (2000) Cost-Benefit Analysis of High
Speed Rail. CPB report 2000/2, CPB, s-Gravenhage
Gutirrez J, Gonzlez R, Gmez G (1996) The European High-Speed Train Net-
work: Predicted Effects on Accessibility Patterns. Journal of Transport Geo-
graphy 4:227238
Het Financile Dagblad (2005) Treinen op Hsl-Zuid Gaan Maanden Later Rij-
den.12 October
Louviere JJ, Hensher DA, Swait JD (2000) Stated Choice Methods: Analysis and
Applications. Cambridge University Press, Cambridge
Maat K, Harts JJ, Zeijlmans van Emmichoven M, Goetgeluk R (2005) Dynamiek
van Stedelijke Milieus: 1996-2002. Ministerie van Volkshuisvesting, Ruimte-
lijke Ordening en Milieu, Den Haag
Mannone V (1997) Gares TGV et Nouvelles Dynamiques Urbaines en Centre Vil-
le: Le Cas des Villes Desservies par le TGV Sud-Est, Les Cahiers Scientifi-
ques du Transport 31:7197
McFadden D (1974) The Measurement of Urban Travel Demand. Journal of Pub-
lic Economics 3:303328
McFadden D (1978) Modelling the Choice of Residential Location. In: Karlqvist
A, Lundqvist L, Snickars F, Weibull JW (eds) Spatial Interaction Theory and
Planning Models. North-Holland, Amsterdam, pp 7596
Min. V&W (1994) Nieuwe HSL-Nota: Nederlands Deel Hogesnelheidsspoorver-

binding Amsterdam-Brussels-Parijs/Londen. Ministerie van Verkeer en Wa-
terstaat (Ministery of Transport and Water Works), Sdu Uitgeverij, Den Haag
Min. V&W (2005) Voortgangsrapport 17 Hogesnelheidslijn-Zuid. Rijkswater-
staat, Ministerie van Verkeer en Waterstaat, Den Haag
Plassard F (1991) Le Train Grande Vitesse et le Rseau des Villes. Transports
345:1423
Rienstra SA, Rietveld P (1999) Spatial Economic Impacts of International Head
Office Locations: A Case Study of Amsterdam South. European Urban and
Regional Studies 6:685689
Rijkswaterstaat (2006) HSL Zuid. http://www.hslzuid.nl, 8 February
Sands B (1993) The Development Effects of High-Speed Rail Stations and Impli-
cations for California. Built Environment 19:257284
Spiekermann K, Wegener M (1996) Trans-European Networks and Unequal Ac-
cessibility in Europe. European Journal of Regional Development 4:3542
Vickerman R (1997) High-Speed Rail in Europe: Experience and Issues for Future
Development. The Annals of Regional Science 31:2138
Waddell P, Ulfarsson G (2003) Accessibility and Agglomeration: Discrete-Choice
Models of Employment Location by Industry Sector. 82nd Annual meeting of
the Transportation Research Board, Washington DC
13 The effect of railway stations on office
space rent levels: The implication of HSL
South in station Amsterdam South Axis
Ghebreegziabiher Debrezion

the Netherlands
Jasper Willigers
Significance, Leiden, the Netherlands
13.1 Introduction
Accessibility is an important determinant of property values. Property is

broadly defined in urban economics as an estate ranging from a vacant
piece of land to an area occupied by all sorts of buildings: residential,
commercial, industrial, etc. (Brigham 1965). Several studies have been
conducted on the theme of accessibility and property values. Different
modes of transport can contribute to the accessibility level a property en-
joys. In this regard, accessibility related to highways (freeways) and rail-
ways is frequently studied, in which accessibility is explained primarily in
reference to the nodes or outlets across the line of the transportation net-
work. Different methodological approaches have been employed to ac-
count for accessibility to highway entry/exit points and railway stations.
However, the general understanding about accessibility in the literature re-
lates to the fact that the value of accessibility is capitalised on the value of
property.
But the capitalisation process can take various forms across different
property types (Debrezion et al. 2007). In the Dutch context several em-
pirical analyses on the impact of railway accessibility on property values
are given in Debrezion (2006). In this chapter we focus in particular on the
capitalisation pattern of railway accessibility on the rent levels of office
266 Ghebreegziabiher Debrezion and Jasper Willigers
spaces in the Netherlands. Our study is generally empirical in nature and

includes three important steps. First, we outline the methodological ap-
proach employed in measuring railway accessibility. Next, based on the
rent contract data recorded by the DTZ Zadelhoff, we estimate spatial he-
donic price models. Third, the result of the output is used to forecast the
effect of the planned railway accessibility for the Amsterdam South Axis
with regard to the High-Speed Line (HSL South) on rent levels of office
spaces around the station.
In discussing railway station effects it is essential to distinguish between
railway stations at the origin and at the destination side of trips, because
departure and destination features of a railway station each have different
implications for residential and commercial property values. In the deci-
sions undertaken for the location of their enterprise, businesses mostly as-
sess the value of a station by its trip-attraction quality as a destination sta-
tion. This represents our point of departure for assessing the value of a
station. In the egress part of a trip, visitors or employees rely mostly on
walking to reach the location of the commercial property. On the activity
end of a railway trip, walking accounts for about 46% of the share of ac-
cess to and from railway stations (Rietveld 2000). Thus, the distance range
at which the influence of station proximity on commercial land rent is
measurable is expected to be quite limited.
The discussion here is organised as follows. In section 13.2 we review
the literature on the effect of railway stations on office space rent levels. In
section 13.3 we discuss the data and methodological approach. After giv-
ing the autocorrelation diagnosis in section 13.4, we discuss the estimation
results of the spatial hedonic price model in 13.5. Section 13.6 provides a
discussion of the implementation implication of HSL South in South Axis
station on the office space rental market. Conclusions follow.
13.2 Railway accessibility in hedonic pricing studies
Several determinants of office space rental are noteworthy and can be

grouped into three general categories: structural, accessibility and envi-
ronmental. The effect of railway accessibility on property values has at-
tracted attention in the literature. Thus we observe studies approaching the
effect of railway accessibility on real estate from two angles: effects on
land use and effects on property values. In one of the earliest studies,
Quackenbush et al. (1987) analysed the impact of the Red Line in Boston
on land use. They found that the largest effect was on commercial proper-
ties and observed only a slight effect on residential properties. Weinstein
The effect of railway stations on office space rents 267
and Clower (1999) moreover has showed that, on the announcement of

Dallas Area Rapid Transit (DART), the occupancy rate of commercial
properties within mile of the station(s) increased on average by 5%.
A number of different findings on the effect of railway stations on
commercial property values are found in the literature. The study on the ef-
fect of proximity to a metro station on commercial property values in
Washington D.C. (Damm et al. 1980) found that the values of commercial
properties decline with distance. Proximity to a metro station results in a
steeper effect on commercial property values as compared with the effect
on residential property values. The elasticity of proximity to the railway
station on property values was around four times higher for commercial
properties than for residential properties. This shows that in the immediate
neighbourhood the premium of closeness to a station is greater on com-
mercial properties. Fejarang (1994) found that commercial space in Los
Angeles within mile of a rail transit station had an additional $31 in-
crease in mean sales price per square foot over comparable parcels outside
the corridor.
In addition to the land use changes as a result of the announcement of
the opening of DART, Weinstein and Clower (1999) observed an increase
in the rent of three classes of offices within mile of a station ranging
from 20.9% to 47.4%, compared with the same kind of offices outside that
range. Similarly, a study done by the US Federal Transit Administration
(FTA) indicates the price per square foot of commercial space decreases
by about $2.3 for every 1000 feet further away from a railway station. This
value accounts for approximately 2% of the value (FTA 2002). Further-
more, Nelson (1998) found that the price per square metre in Atlanta de-
creases by $75 for every metre further away from a transit station.
Weinberger (2001) tested several hedonic price models on the rental rates
of commercial property in Santa Clara County. The finding reveals that the
results of commercial properties within mile of a light rail station are
10% higher than rents of commercial properties beyond mile of a light
rail station. When controlling for highway access, the rail proximity bene-
fit was maintained. A similar study was also done by Cervero and Duncan
(2001) in the same county. They found that commercial properties within
mile of a light rail station were sold at prices 23% higher than commer-
cial properties outside this range.
The capitalisation is even further pronounced in the case of proximity to
a commuter railway station. Commercial properties within mile of the
commuter railway station sell at prices higher than 120% above commer-
cial properties outside this range. But, contrary to the above positive ef-
fects of proximity to a railway station, Cervero and Duncan (2001, 2002a,
2002b) found a mixed effect of proximity to a railway station on commer-
cial property values. The study in Los Angeles County shows that for
commercial properties located within mile of a station, the impact of dif-
ferent stations ranges from a negative effect of as much as 30%, to a posi-
tive effect of 16%, compared with the values of properties outside the
mile range. Similarly, the study in San Diego County reveals that the im-
pact of proximity to a railway station within mile on commercial prop-
erty values ranges from a negative effect of 10% to a positive effect as
large as 90%. Landis et al. (1995) found no premium for commercial land.
However, the inability to find a positive impact is attributed to a data and
methodological problem rather than to a lack of a real value premium.
13.3 Data and methodology
13.3.1 Data source
The main sources of data for the estimation in this study are the recoded
rent contracts of office spaces from Zadelhoff DTZ, which extend over a
period of 23 years (1983 to 2005). Geographically it covers all provinces
of the Netherlands. The dataset includes information on the rent per square
metre of office floor space, building status, type of rent contract, and cate-
gory of business. The five types of building status are identified as: first-
user buildings, second-user buildings, buildings under construction, build-
ings under renovation, and land yet to be developed. The three types of
rent contracts are, namely, first-hand new rent contracts, rent extension
contracts, and subleases. In addition, the dataset identifies the type of busi-
ness undertaken by the user. One could assert that the data actually relate
to the user and not to the building itself. However, these data can be used
as a proxy for the nature of the building, because the qualities of the re-
quired building can differ according to the business orientation of the user.
For instance, the type of building needed by a banking or insurance firm is
generally different from that required by a transportation and storage busi-
ness. A number of other variables are also brought into the analysis. To ac-
count for the environmental features we include the share of different land
use types in the postcode area. Because the data includes rent contracts for
a long period, we use year dummies to capture the temporal change in the
rent levels related to inflation and real value changes.
Two types of accessibility variables are included. Railway accessibility
is measured by the proximity of the office location to the nearest railway
station. Accessibility by road is measured by the distance to the nearest
highway entry/exit point from the location of the office. In order to com-
pute these distance measures, the stations, railway line, highway entry/exit
points, and office location had to be geo-coded; coding was carried out at a
detailed address level, because the office rent is generally expected to be
sensitive to distance and, according to the literature, the range of distance
at which the rent of commercial properties responds to proximity is rather
limited. To account for the effect of business locations opportunities for
interaction with customers and employees on the rent level, we introduce a
derived opportunity index for business locations. The statistical tests over
the different opportunity (accessibility) indicators made by Song (1996)
indicate that gravity-type opportunity measures generally perform better
than other measures. We define the opportunity index as the cumulative
population of all postcode areas in the country weighted by the inverse of
distance from the office location to the centroid of the postcode area:
R Population j
Opportunity indexi = (1)
j =1 d ij
where d ij is the Euclidean distance between the location of office i and
the centroid of postcode area j . The opportunity index used in the empiri-
cal estimation of this study is base on the 4004 post code areas comprising
the Netherlands. Distance is measured in metres.
Railway accessibility is explained by two variables: a measure of the
rail service at the railway station, and the distance to the station. We use a
derived index as the measure of rail services at a station. As mentioned
earlier, for explaining the commercial property value it is relevant to con-
sider the station at the destination side of trips. Thus the explanation of the
index assumes that the station under consideration is a destination station.
In the following subsection we elaborate on the derivation of this index.
The descriptive statistics of the data used in the estimation is given in Ta-
ble 13.A1 in the Appendix.
13.3.2 Railway accessibility: Rail Service Quality Index (RSQI)
Railway stations differ from each other in the quality of the rail services
offered to passengers. In many empirical applications it has been noted
that there is a need to distinguish between stations on the basis of service
levels. The real estate analysis indicates that the intensity of the effect ex-
erted by a station differs from station to station. Our dataset includes some
indicators of rail service at the station level, such as daily frequency of
trains leaving the station, number of destinations having a direct connec-
tion with the station under consideration, and whether the station has inter-
city services. By including these factors in the choice analysis we account
for the rail service quality of a station. However, the usefulness of these
factors is limited because they do not take into account the location of the
station in relation to important origin stations. So, in addition to the afore-
mentioned factors, the stations position in the network towards origin sta-
tions and the importance of these origin stations are significant in the de-
termination of a stations service quality. The need for a comprehensive
rail service quality indicator for each station leads us to a modelling exer-
cise on underlying railway trip data. We refer to this index as the Rail Ser-
vice Quality Index (RSQI ) of a railway station. We next discuss the ser-
vice quality index of a station as a destination point. The level of pure rail
service quality of a station as a departure point is a function of the impor-
tance level of the origin stations, the generalised journey time it takes to
travel from these stations, and the ratio of generalised journey time to dis-
tance. The importance of an origin station can be explained by the size of
the station as an origin point.
RSQIdestination j = i
f (Oi , GJTij , GJTij / d ij ) (2)
where, Oi is the total number of trips generated by an origin station i .

This is taken as an indicator of importance level of origin station i for the
concerned destination station j . GJTij is the generalised journey time be-
tween an origin station i and a destination station j ; generalised journey
time is a measure of the time needed to travel between the stations. It in-
cludes the average waiting time at the origin station, in-vehicle time, trans-
fer time, and some penalty for the number of transfers. The generalised
journey time measure encompasses several distinct rail service features at
a station. For instance, the frequency of trains leaving the station per pe-
riod of time is reflected in the average waiting time component of the gen-
eralised journey time. The distinction between the types of stations is ex-
pected to be reflected by in-vehicle time, transfer time, and number of
transfer penalty components. For example, an intercity railway station
mostly provides more direct services than stop train stations. This leads to
less in-vehicle and transfer time, and thus less generalised travel time than
stop train stations. In addition, the connection time and penalty for connec-
tion show the amount of direct connections a station has with other sta-
tions. d ij is the (Euclidean) distance measure between origin station i and
destination station j .
Doubly constrained model
We use the doubly constrained spatial interaction model to model and es-
timate the parameters in determining the RSQI of a railway station. As the
name implies, the model is constrained both from the origin and destina-
tion stations. The constraint pertains to the production capacity of an origin
station and attraction capacity of a destination station. These are con-
strained to be equal to the sum of all trips origination at the departure sta-
tion and ending at the destination station, respectively. These are among
the many other factors that determine the number of trips between two sta-
tions. In addition, we assume that the number of trips between any two sta-
tions is a function of the generalised journey time and generalised journey
time to distance ratio. The general form of the doubly constrained gravity
model used to depict spatial interaction (Fortheringham and OKelly 1989)
for trips between stations is given as follows:
Tij = Ai Oi B j D j f (GJTij ) f (GJTij / d ij ) exp( ij ) (3)
Oi = j Tij (4)
D j = Tij (5)
i
where Tij is the number of trips between origin station i and destination
station j ; Ai and B j are the balancing factors which ensure that the con-
straints on origins and destinations (given by Eqs. 4 and 5) are met; Oi is
the total number of trips originated in station i ; D j is the total number of
trips attracted by a destination station j ; ij is the error component of the
model which follows an independently and identically normal distribution.
The functions of generalised journey time ( f (GJTij ) ) and the ratio of
generalised journey time and distance ( f (GJTij / d ij ) ) are specified in
Eqs. 6 and 7, respectively.
C
f (GJTij ) = exp( c DGJTc )
ij
(6)
c =1
ij
This is a stepwise function of the generalised journey time. DGJTc is
a dummy variable equal to 1 if GJTij falls in the generalised journey time
category c , and zero otherwise. c is the coefficient for generalised jour-
ney time category c .
f (GJTij / d ij ) = (GJTij / d ij )

(7)
is the power coefficient for the ratio of generalised journey time and
distance. Thus our estimation of the doubly constrained gravity model is
given by:
C ij
Tij = Ai Oi B j D j exp c DGJTc (GJTij / d ij ) exp(ij )

(8)
c=1
This equation can be linearised by taking the natural logarithm of both
sides:
C ij
ln (Tij /(Oi D j ) ) = ln Ai + ln B j + c DGJTc
c=1 (9)
+ ln (GJTij / d ij ) + ij
The coefficient of the generalised journey time categories, the ratio of
generalised journey time, and the balancing factors will be estimated from
the above equation. The logs of the balancing factors in the equation there-
fore represent the coefficients to be estimated. This requires that the logs
of the balancing factors be multiplied by the dummy variable for the corre-
sponding station; the equation to be estimated is given as:
T N N 1
C ij
ln ij = ln A~i S ~i + ln B~j S ~j + c DGJTc
O D ~i =1 c=1
i j
~
j =1
(10)
GJTij
+ ln + .
d ij
ij
where N is the number of railway stations in the railway network; and
~
S ~i and S ~j are dummy variables for departure station i and destination
~ ~ ~
station j . They assume the value 1 when i = i and j = j respectively,
and 0 otherwise. Given the assumption on the error components above, Eq.
10 can be estimated using ordinary least squares (OLS). The estimated co-
efficients are then used in determining the RSQIs for each station. The
RSQI for any destination station j is determined by summing the function
over all origin stations as follows:
C ij
RSQIdestination j = Ai Oi exp c DGJTc (GJTij / d ij )

(11)
i c=1
13.3.3 Methodology: Econometric models
Standard hedonic price models
The analysis in this chapter is based on hedonic pricing model estimation.

The variables of interest relate to accessibility in general and railway ac-
cessibility in particular. There are two types of variables related to rail-
ways: distance to the nearest railway station, and the rail service quality
index (RSQI) at the nearest station, and both are expected to capture the
effect of railway accessibility. In addition, the model includes accessibility
to the highway, measured by the distance to the nearest highway entry/exit
point. Furthermore, the model includes access to potential employees and
business customers by assuming radial access to the office location. A
semi-logarithmic hedonic model is specified. The dependent variable is
given in the natural logarithmic form; the values of the coefficients thus
represent percentage change. The specifications of the standard models
used in the estimations are given by Eqs. 12 and 13. Distances from the of-
fices to the nearest railway station are divided into six (6) distance catego-
ries, where distances above 4 kilometres are taken as the reference group.
Distance and RSQI are included separately in the first model. In the second
model, however, a cross product of distance and RSQI is included with the
aim of observing the effect of service quality on office rent levels at differ-
ent distance classes. The respective base models have the following form:
ln (rent i ) = + ' BS DBuildStat us i + 'CT DContrType i
+ ' BT DBusType i + ' dc Distcategr ail i
+ RSQIdest + ln hwdist i (12)
RSQIdest hw
+ opport Opportunit yIndex i

+ ' Neighb DNeighb i + 'time Dtime i + i
ln (renti ) = + ' BS DBuildStatusi + 'CT DContrTypei

+ ' BT DBusTypei + ' dc Distcategraili RSQIdest
(13)
+ hw ln hwdisti + opport OpportunityIndexi
+ ' Neighb Neighbi + 'time Dtimei + i
Where renti is the rent per square metre of space for office i ,
DBuildStatusi is the building status of office i , DContrTypei is the
rent contract type of office i , DBusTypei is the business type for office
i , Distcategrail i is the category for the distance at which office i is lo-

cated from the railway line. A positive sign is expected for coefficients for
the distance categories, with the highest effect in the immediate categories
to the nearest station, and decreasing outwards. RSQIdesti is the service
quality index of the nearest station for office i . A positive effect is ex-
pected, showing that an increase in the service quality index of the nearest
railway station to the office location leads to higher rents. hwdist i is the
distance between office i and the nearest highway entry/exit point. We
expect the office space rents to decline as the distance to the nearest high-
way entry/exit point increases. OpportunityIndexi is the opportunity in-
dex for office i (defined by Eq. 1). The larger the opportunity for interac-
tion with potential customers and employees, the higher the office rent
level. Neighbi is the land use feature for the post code areas in which of-
fice i is located, Dtimei is a time dummy variable representing the year
when the rent contract was signed and i is the iid error term. The expla-
nation and descriptive statistics of the variables are given in Table 13.A1
in the Appendix.
Spatial hedonic price models
The standard hedonic price models discussed above assume that rent levels
of offices in the sample are independent from each other. The law gener-
ally referred to as Toblers first law of geography states everything is re-
lated to everything else, but near things are more related than distant
things (Tobler, 1970); however, there is the possibility that the assump-
tion could be violated. Offices in the same area tend to share similar physi-
cal, environmental and accessibility features, which results in spatially-
correlated rent levels. At the same time, location-related characteristics are
generally difficult to observe and quantify, and the omission of variables
from the hedonic price model results in spatially-correlated error terms.
The violation in the assumption of independence of the error term leads to
inefficiency in the parameters estimate by ordinary least squares (OLS)
methods. There are two well-known ways of dealing with the spatial de-
pendence in the literature. The first approach includes the weighted aver-
age of neighbouring office rents, and the second involves modelling the
structure of the error term of the standard model. The general cases of spa-
tial hedonic price models corresponding to the earlier discussed models are
given below:
ln (rent i ) = wij . ln( rent j ) + + ' BS DBuildStat us i

j
+ 'CT DContrType i + ' BT DBusType i

+ ' dc Distcategr ail i + RSQIdest (14)
RSQIdest
+ hw ln hwdist i + opport Opportunit yIndex i

+ ' Neighb DNeighb i + 'time Dtime i + wij . i + u i
j
ln (renti ) = wij . ln(rent j ) + + ' BS DBuildStatusi

j
+ 'CT DContrTypei + ' BT DBusTypei

+ ' dc Distcategraili RSQIdest + ln hwdist i (15)
hw
+ opport OpportunityIndexi + ' Neighb Neighbi

+ 'time Dtimei + wij . i + ui
j
where and are the weighted lag and error coefficients; wij is an
indicator of the proximity of office j to office i in the row standardised
weights matrix; i is the residual of the OLS estimate for office i ; and u
is white noise error term (u ~ N (0,1)) . If = 0 , the model reduces to the
spatial lag model. In this case the office rent level is dependent on the
weighted average rent of the neighbouring offices. But if = 0 , the
model reduces to the spatial error model. If both coefficients are different
from 0, we get a higher-order spatial specification that involves both spa-
tial lag and spatial error models. In our analysis here, the estimation con-
siderations will be limited to the case where either one of the two coeffi-
cients is 0.
13.4 Diagnosis for spatial autocorrelation
To assess the spatial dependency in the office space rents we use Morans I
test. A row standardised weights matrix of proximity based on a 3 kilome-
tre cut-off distance is used to model the spatial structure of the depend-
ency. By showing the level of spatial dependency on the data, Morans I
test indicates whether or not the standard (non spatial) model is wrongly
specified. However, the test does not provide any information about which
spatial model is appropriate for the data. Identifying the suitable spatial
model is based on Lagrange Multiplier tests. Table 13.1 provides five La-
grange Multiplier test results. The first two (LM lag and Robust LM lag)
are tests on the appropriateness of the spatial lag model. The next two (LM
error and Robust LM error) relate to the spatial error model as an alterna-
tive model. The last, Lagrange Multiplier, tested for a higher order alterna-
tive specification involving both spatial lag and spatial error terms. The
specifications of the entire test statistic are listed in the appendix.
Table 13.1. Diagnostics for spatial dependence

Test MI/DF Value Prob
1 Separate effect of distance and station quality index
Moans I (error) 0.1088 56.75 0.000
Lagrange Multiplier 1 78.89 0.000
Robust LM (lag) 1 0.48 0.488
Lagrange Multiplier (error) 1 2911.72 0.000
Robust LM (error) 1 2833.31 0.000
2. Cross effect of distance and station quality index

Moans I (error) 0.1088 57.11 0.000
Lagrange Multiplier 1 84.47 0.000
Robust LM (lag) 1 0.15 0.698
Lagrange Multiplier (error) 1 2969.62 0.000
Robust LM (error) 1 2885.30 0.000
The Morans values are positive and significant, indicating that error
component(s) of the standard model for neighbouring offices are positively
correlated a violation in the independence assumption of the error term.
Thus the ordinary (non spatial) model estimations have resulted in biased
estimates. This called for the use of a spatial autocorrelation model for the
rent data. The choice of the proper approach for modelling the spatial
autocorrelation on the data is based on Lagrange multiplier tests, of which
two categories, a standard and a robust form, are proposed separately for
each of the modelling approaches. The specifications of the test statistic
are given in the appendix. Both the standard forms of the Lagrange Multi-
plier tests (LM lag and LM error) are significant, indicating that both spa-
tial lag and spatial error models can be used to model our data. However,
from the robust forms only the robust Lagrange Multiplier test is signifi-
cant. This indicated that the spatial error model is the preferred model for
the data. Based on these test results we apply the spatial error model to
model our data.
13.5 Estimation and discussion
The resulting spatial autocorrelation diagnosis discussed above shows that

the spatial error model is preferred. We therefore estimate both the ordi-
nary least squares (OLS) and spatial error hedonic price model (SEM) for
each of the base models. The spatial autocorrelation models are estimated
using Geoda 9.5-i5 software. The estimation results related to the variables
of interest are given in Table 13.2. The coefficients of the remaining vari-
ables of the models are listed in Table 13.A2 in the Appendix.
Table 13.2. Estimation results

Separate effect of distance Cross-effect of distance
and rail service quality in- and rail service quality
dex index
OLS SEM OLS SEM
CONSTANT 2.499*** 3.132*** 2.487*** 3.138***
(23.271) (14.584) (23.452) (14.615)
raildist0_250 0.097*** 0.139*** 0.169*** 0.176***
(6.147) (5.934) (9.826) (8.886)
raildist250_500 0.123*** 0.127*** 0.221*** 0.176***
(8.923) (5.704) (14.712) (9.800)
raildist500_1000 0.072*** 0.074*** 0.141*** 0.087***
(5.758) (3.423) (11.458) (5.445)
raildist1000_2000 0.051*** 0.038* 0.115*** 0.054***
(4.211) (1.836) (10.265) (3.585)
raildist2000_4000 0.035*** 0.034* 0.091*** 0.050***
(2.844) (1.689) (7.409) (3.187)
Rail service quality index 0.144*** 0.080***
(RSQI) (12.749) (5.246)
0.376*** 0.267*** 0.387*** 0.274***
Ln(opportunity index) (22.320) (7.437) (23.211) (7.710)
-0.043*** -0.044*** -0.042*** -0.044***
Ln(hwdist) (-11.581) (-9.634) (-11.245) (-9.652)
0.709*** 0.707***
Lambda (36.555) (36.252)
Number of observations (N) 9,357 9,357 9,357 9,357
R-squared 0.3602 0.4255 0.3603 0.4257
***
stands for a significance level of less than 1%
**
*
The first set of estimates is based on the model, which separately pre-
sents the effect of proximity to railway stations and service quality at the
station. The second set of estimates is based on the model, which treats
railway accessibility as a cross-product of distance and the RSQI. A

piecewise approach is used for the distance to the nearest railway station.
The effect of proximity to the nearest railway station on office rent levels
is inferred by reference to the rent levels of offices located beyond 4 kilo-
metres from the nearest railway station. Our discussion will be based on
the spatial hedonic error models (SEMs). The spatial error parameter in
both spatial error models ( ) is equal to 0.71, and is highly significant be-
cause it shows that the unobservable components of the model for
neighbouring offices are positively correlated.
13.5.1 Effect of accessibility on office space rent levels
Railway accessibility
From Table 13.2 we can see that the proximity to a railway station has a
positive effect on office rent levels. The spatial error model estimation on
the separate effect of proximity and rail service quality shows that offices
within 250 metres of a railway station have a rent of about 14% above that
for offices beyond 4 kilometres of a railway station. A downward-sloping
effect is found: the effect of proximity to a railway station on office rent
levels decreases as the distance away from the railway station increases. A
statistically weaker positive effect of station proximity is found for offices
between 1 km and 4 km compared with offices located beyond 4 km from
the nearest railway station. This confirms the assertion that the effect of
proximity to the railway station on commercial property is limited to the
range that is within walking distance (Debrezion et al. 2007). A graphical
description of the effect of proximity to the railway station on office rent
levels is given in Figure 13.1.
On the other hand, keeping all other things constant, a unit increase in
the RSQI of a station leads to an average increase of the rent level by 8%.
The refinement of this effect is achieved by observing the effect of a
change in rail service quality on the rent level at different distance ranges
from the station. The estimation of the cross-effect of rail service quality
with station proximity shows the effect of service quality at different dis-
tance categories. A unit increase of the rail service quality at the nearest
station leads to about an 18% increase in the rent level of offices within
500 metres of a station, compared with the rent levels of offices beyond 4
kilometres of a railway station. The effect is halved in areas between 500
metres and 1 kilometre. The effect of an increase in rail service quality on
rent levels declines with distance from the station. A graphical illustration
16
14
12
10
Effect (%)
raildist1000_2000
raildist2000_4000
raildist500_1000
raildist250_500
raildist0_250
OLS
SEM Distance categories (metres)
Figure 13.1. Effect of distance to nearest railway station on office space rent
25
20
15
Effect (%)
10
0
raildist0_250
raildist250_500
raildist500_1000
raildist1000_2000
raildist2000_4000
OLS
SEM Distance categories (me tres)
Figure 13.2. Cross-effect of rail service quality and distance to the nearest station
on office space rent
of the effect of rail service quality at different distance categories is given

in Figure 13.2. A doubling of the frequency of services on the existing
network setting (which halves the average waiting time) increases the av-
erage rail service quality indices of the stations by 0.2. This increase in rail
service level leads to a 3.6% increase in rent level for offices within 500
metres of a station, compared with offices beyond the 4 kilometre range. In
the last distance category (between 2 and 4 kilometres), the change is
translated into a 1% increase in rent levels.
Similarly, a decrease in the in-vehicle time component of the general-
ised journey time, by increasing the speed of the vehicles, leads to an in-
crease in the RSQI. For example, a 50% increase in the speed of the trains
directed to the railway stations used in the analysis results on average in an
increase of the RSQI of the stations by half a unit (0.5). This in turn leads
to a 9% rent increase for offices located within 500 metres of the stations,
compared with offices located beyond the 4 kilometre range. For offices
located in the range of 500 metres to 1 kilometre, the effect on office rents
of increasing the speed of trains by 50% is about 4.5%, compared with the
effect on office rents beyond 4 kilometres from a railway station. Because
the RSQI of a station integrates all components of the generalised journey
time, it is possible to compare the effect of changes in the time compo-
nents on office space rent. Given the current setting of the railway net-
work, doubling the frequency of train service and increasing the speed of
the trains by 20% results in an equivalent increase of the RSQI, and thus of
office rent.
Effect of highway accessibility
Road accessibility, which includes distance to the nearest highway en-

try/exit point and opportunity index defined by Eq. 1, have significant ef-
fects with the expected signs. The elasticities of distance to the nearest
highway entry/exit and opportunity index on office space rent levels are -
0.044 and -0.270, respectively, on both models. This means that a 1% in-
crease in both factors leads to a decrease of 0.044% and an increase of
0.270% on office space rent levels, respectively.
13.5.2 Effect of building status and nature of business
The analysis found significant rent differences among different building

statuses. The reference building status is taken to be first user office
spaces. This means that the occupier rents the office space directly after
the completion of the office space. Compared to this reference building
status, second user office spaces rent for approximately 11% lower. On the
other hand, offices occupied after renovation rent 5% lower than new of-
fices (first user offices). No significant difference is found on the other
building statuses.
Moreover, the estimation result shows that a significant rent difference
is observed on some natures of business of the occupants. Our analysis
takes industrial companies and public utilities as a reference group. Higher
rent levels are observed for offices occupied by Credit and Insurance com-
panies. Such office spaces rent for around 10% higher than the rent levels
of the reference group. Similarly, Financial business companies rent for
5% more, compared to the reference group. Conversely, offices occupied
by a Trade and Repair company, and Education and Health Care rent for
about 7% and 10% lower than the reference group, respectively. The
analysis shows no significant difference on the office space rent levels of
other occupant types.
Among the different types of rent contracts, we found only significant
differences between direct rent from the owner and sublease contracts. In
the case of offices rented by sublease, contracts are found to rent about 5%
higher than direct rent contracts from the owner.
13.5.3 Temporal effect
Figure 13.3 below shows the temporal development of the rent prices. The
rent prices can be seen to follow the development of demand and supply of
office space. The demand and supply of real estate is surveyed by Dynamis
(2006). Between about 1995 and 2001 there was a relatively tight office
market, reflected by a sharp increase in the real estate price. The shortage
of office space stimulated new office builds, which were completed with a
several year time lag. After a peak of office space take-up in 2002, the de-
mand for office space declined, but supply of new offices soared by the
political initiatives taken during the tight market period. The large over-
supply of office real estate after 2002 led to a decline in the real estate
price; during this time the demand for office space also witnessed a down-
turn. After year 2000, there was a general demand slow down in the office
market, which is believed to be a result of general slow down of the Dutch
economy.
50
40
effect (%)
30
20
10
0
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
Year
Figure 13.3. Increase in rent levels by year compared to rent levels before 1990
13.6 Implication of South Axis investment on office

space rent levels
The model described in the previous sections is applied to the case of the
South Axis in Amsterdam. Near the South Axis several new railway infra-
structures have recently been completed. A new direct intercity connection
to Utrecht and Eindhoven has also been accomplished. These infrastruc-
tures allow for more train services to and from the Amsterdam South Axis
Station and can therefore be expected to influence real estate prices in this
area. The most notable railway development is the HSL South high-speed
railway from Schiphol airport to Rotterdam and farther, to Belgium.
The question of whether the Amsterdam South Axis Station will ac-
commodate high-speed train service is still indefinite. In the long-term
both Amsterdam South Axis Station and Amsterdam Central Station are
options for HSL. In our analysis we assume that all high-speed train ser-
vices use Amsterdam South Axis Station. The domestic connections ac-
count for the largest part of the train services that will use the new high-
speed railway. According to the projected schedule (High-Speed Alliance
2006) of all 96 trains leaving Amsterdam per day, only one-third goes to
Belgium, half of which continues to Paris. With this schedule travel times
between Amsterdam and Rotterdam decrease from 53 to 30 minutes, and
between Amsterdam and Breda in the south of the Netherlands from 1

hour and 42 minutes to 54 minutes. This reduction of travel times can have
a large impact on rail service quality at the South Axis Station.
13.6.1 Effect of high-speed rail on rail service quality index of

South Axis Station
The implementation of high-speed rail reduces the in-vehicle travel time

and thereby the generalised journey time from which the rail service qual-
ity index used in this study is derived. Figure 13.4 shows the RSQI of the
station to increase on average at a rate of 0.1 per 10% increase in the speed
of vehicles. Combined with Table 12.3, we can observe that the rent levels
for offices within 500 metres of the station increase at a rate of 1.8% for
every 10% increase in the speed of trains directed to the station.
2
Rail Service Quality Indexh
1.5
0.5
0
1.00 1.20 1.40 1.60 1.80 2.00 2.20 2.40
Factor of current level of trains speed
Figure 13.4. The effect of speed level (as a factor of current level) of train di-
rected to South Axis Station on the RSQI of the station
The HSL South from Amsterdam has three national destinations: Schip-
hol airport, Rotterdam and Breda. On average, the travel time is halved,
implying an operational speed of factor 2. Based on the current setting, the
operation of HSL South is expected to increase the RSQI of the South Axis
Station by 0.30 from 0.74 to 1.04. According to the model prediction, this
increase in RSQI at the South Axis Station leads to an increase of rent lev-
els office within the 500 metres range of the station by about 5.4%, com-
pared to the rent of offices located beyond 4 kilometres from the station.
13.7 Conclusions
We have in this chapter discussed methodological issues in addressing

railway accessibility from the point of view of commercial entities. The
relevance of a railway station as a destination point outweighs the rele-
vance of a railway station as a departure station. The reverse case is true
for households. Different from the existing simple methods of measuring
railway accessibility by proximity, or at most differentiating railway sta-
tion by frequency of rail service, this chapter has applied spatial interaction
models using a wider range of railway service elements to measure the
railway service provided in a station. This method constitutes a compre-
hensive approach and is expected to be superior to existing approaches.
From the spatial autocorrelation diagnosis, we can draw several conclu-
sions. First, the data on office space rent used in the analysis exhibit spatial
autocorrelation. Second, the appropriateness of the spatial error model for
the model estimation indicates that neighbouring office locations share
common features unobserved by the model. These unobserved features can
range from unaccounted structural features to environmental features to la-
tent location factors, such as the image of a site caused by neighbouring
buildings. Spatial autocorrelation improves model estimates by reducing
the bias that can result due to correlation in the error component of the
model.
By controlling for a wide range of features, the spatial autocorrelation
model estimated in this chapter found a significant effect with expected
signs for accessibility features on office space rent levels. Both railway
and highway accessibilities are included. Our main focus here has been to
analyse the effect of railway accessibility on rent levels of office space.
The study shows the relevance of railway accessibility as accounted for by
the proximity and rail service quality measure for office space rent in the
Netherlands. Rent levels decline as the distance from the nearest railway
station increases. On the other hand, rail service quality of a railway sta-
tion has a positive effect on office space rent levels.
Furthermore, the cross-effect of distance and service quality on rent
shows a declining effect of rail service quality of a station with distance. A
stronger effect is observed on offices located in the immediate vicinity of a
railway station, indicating that the range at which railway accessibility will
have a meaningful effect on the rent levels of office is quite limited. As has
been pointed out in several earlier empirical studies, this range represents a
reasonable walking distance. Due to the fact that office locations lay on the
egress part of rail trips, access to them is expected to depend mostly on
walking.
The flexibility of service quality measure of a railway station allows us
to make model predictions based on expected changes in the railway net-
work setting. The study accessed the implication of HSL South implemen-
tation on office space rent levels. The operation of the high-speed line is
expected to upgrade the rail service quality measure of South Axis Station
which in turn is expected to increase the rent level of office spaces around
the station. Based on the foreseen change, the study found that, on average,
rent levels are expected to rise by 5.4% for offices located within 500 me-
tres of the station. In reality the effect can be a bit higher for two main rea-
sons. First, only changes in high-speed line setting are considered. Im-
provements in terms of the ordinary rail operation have not been
addressed. Second, the high-speed line changes mainly consider changes in
the national railway network. In the case of high-speed operations, interna-
tional origins can play a major role in upgrading the rail service quality
status of the South Axis Station.
References
Anselin L (ed) (1995) SpaceStat Version 1.80 Users Guide. University of Illinois,
Urbana-Champaign
Brigham E (1965) The Determinants of Residential Land Values. Land Economics
41:325334
Cervero R, Duncan M (2001) Rail Transits Value Added: Effect of Proximity to
Light and Commuter Rail Transit on Commercial Land Values in Santa Clara
County California. Paper prepared for National Association of Realtors Urban
Land Institute
Cervero R, Duncan M (2002a) Land Value Impact of Rail Transit Services in Los
Angeles County. Report prepared for National Association of Realtors Urban
Land Institute
Cervero R, Duncan M (2002b) Land Value Impact of Rail Transit Services in San
Diego County. Report prepared for National Association of Realtors Urban
Land Institute
Cliff A, Ord JK (1971) Evaluating the Percentage Points of a Spatial Autocorrela-
tion Coefficient. Geographical Analysis 3:51-61
Damm D, Lerman SR, Lerner-Lam E, Young J (1980) Response of Urban Real
Estate Values in Anticipation of the Washington Metro. Journal of Transport
Economics and Policy 14:315336
Debrezion G (2006) Railway Impacts on Real Estate Prices. Ph.D. dissertation,

Tinbergen Institute Research Series 389, Amsterdam
Debrezion G, Pels E, Rietveld P (2007) The Impact of Railway Stations on Resi-
dential and Commercial Property Values: A Meta-Analysis. Journal of Real
Estate Finance and Economics, forthcoming
Dynamis (2006) Sprekende Cijfers Kantorenmarkten 2006. Dynamis, Amersfoort
Fejarang RA (1994) Impact on Property Values: A Study of the Los Angeles
Metro Rail. Preprint, Transportation Research Board, 73rd Annual Meeting,
January 9-13, Washington, D.C.
Fortheringham AS, OKelly ME (1989) Spatial Interaction Models: Formulations
and Applications. Kluwer Academic Publishers, Dordrecht
FTA (2002) Commercial Property Benefit of Transits. Hickling Lewis Brod INC
High Speed Alliance (2006) http://www.hslzuid.nl
Landis J, Cervero R, Guhathukurta S, Loutzenheiser D, Zhang M (1995) Rail
Transit Investments, Real Estate Values, and Land Use Change: A Compara-
tive Analysis of Five California Rail Transit Systems. Institute of Urban and
Regional Studies, Monograph 48, University of California, Berkeley
Moran PAP (1948) The Interpretation of Statistical Maps. Journal of the Royal
Statistical Society B 10:243251
Nelson AC (1998) Transit Stations and Commercial Property Values: Case Study
with Policy and Land Use Implications. Presented at Transportation Research
Board 77th Annual Meeting
Quackenbush KH, Holst-Roness E, Humphrey TJ, Slater G (1987) Red Line Ex-
tension to Alewife: Before/After Study Appendices. Central Transportation
Planning Staff for the Massachusetts Bay Transportation Authority, Boston
Rietveld P (2000) The Accessibility of Railway Stations: The Role of the Bicycle
in the Netherlands. Transportation Research Part D 5:7175
Sen A (1976) Large Sample-Size Distribution of Statistics Used in Testing for
Spatial Correlation. Geographical Analysis 8:174-184
Song S (1996) Some Tests of Alternative Accessibility Measures: A Population
Density Approach. Land Economics 72:474482
Tobler W (1970) A Computer Movie Simulating Urban Growth in the Detroit Re-
gion. Economic Geography 46:234-240
Weinberger R (2001) Light Rail Proximity: Benefit or Detriment, the Case of
Santa Clara County, California. Presented at Transportation Research Board
80th Annual Meeting, January 7-11, Washington, D.C.
Weinstein BL, Clower TL (1999) The Initial Economic Impact of the DART LRT
System. Unpublished manuscript
Appendix
Spatial autocorrelation test statistics: Morans I
Morans I test statistics is the most commonly used test for checking spa-
tial autocorrelation in the data. The test is developed by Moran (1948). The
test statistics is specified as follows:
I = ( N / S 0 )(e' We / e' e) (A.1)
where N is the number of observations, e is a vector of the OLS re-

siduals, S 0 is the standardisation factor, which is the sum of the elements
of the weights matrix W. For a row standardised W A.1 I is reduced to
( I = e' We / e' e ). The value of the statistic ranges between -1 and 1. A
value of -1 indicates perfect negative correlation, where offices with high
rent are generally neighboured by office with lower rent and vice versa. On
the other hand, a value of 1 indicates perfect positive correlation, where of-
fices with high rent are neighboured by offices with high rent and vice
versa. A value of 0 shows no spatial autocorrelation. The statistics is as-
ymptotic to a normal distribution approximation (Cliff and Ord 1971; Sen
1976).
Lagrange multiplier tests
1. LM error = [e' We /(e' e / N )] 2 /[ tr( W 2 + W ' W )] (A.2)
2. LM lag = [e' Wy /(e' e / N )] 2 / D (A.3)

where,
D = [ WX )' (I X( X' X) 1 X' )( WX ) / 2 ] + tr( W 2 + W' W)
The diagnosis on the spatial autocorrelation is based on a series of tests.
~
Robust LM error = [e' We /(e' e / N ) T ( RJ . ) 1
3. (e' Wy /(e' e / N ))]2 (A.4)
~
/[T T 2 ( RJ . ) 1 ]
~
( RJ . ) 1 = [T ( WX )' (I X( X' X) 1 X' )( WX ) /
With
(e' e / N )]1
and T = tr( W 2 + W' W)
Robust LM lag = [e' Wy /(e' e / N )

4. ~ (A.5)
(e' We /(e' e / N ))]2 /[ RJ . T ]
All tests are distributed as 2 , with one degree of freedom (Anselin
1995).
Table 13.A1. Descriptive statistics of the variables included in the estimation of

office rent levels
Std. De-
N Minimum MaximumMean viation
Rent (Euro/square metre) 8.80 12992.22 125.11 145.06
ACCESSIBILITY
Rail service index (destina-
tion station) 9340 0.034 1.464 0.686 0.300
Distance to nearest railway
station (m) 11298 30 20,139 1,686 1,751
Distance to highway en-
try/exit points (m) 11298 21 35,372 1,878 1,826
Opportunity index 11298 133.64 980.35 387.12 103.29
BUILDING STATUS
First user 1508 0 1 0.133
Second user 9114 0 1 0.807
Renovation 463 0 1 0.041
Under construction 61 0 1 0.005
Yet to be built 26 0 1 0.002
TYPE OF CONTRACT
Direct rent 10967 0 1 0.971
Rent extension 109 0 1 0.010
Sub lease 222 0 1 0.020
TYPE OF BUSINESS
Industrial companies and
public utilities 661 0 1 0.059
Building and civil engineer-
ing 260 0 1 0.023
Trade and repairing compa-
nies 627 0 1 0.055
Transportation and storage 281 0 1 0.025
Communication companies 418 0 1 0.037
Credit and insurance ser-
vices 508 0 1 0.045
Financial business services 685 0 1 0.061
Other business services 2566 0 1 0.227
Computer companies 1104 0 1 0.098
Public administration, de-
fence or social security 728 0 1 0.064
Education and health care 707 0 1 0.063
Other institutions and com-
panies 1406 0 1 0.124
Missing category 1347 0 1 0.119
To be continued
Table 13.A1. Descriptive statistics of the variables included in the estimation of

office rent levels (continued)
Std. De-
N Minimum Maximum Mean viation
LAND USE
cultivation under glass 9357 0 0.509 0.002 0.015
other agricultural use 9357 0 0.963 0.141 0.220
forest 9357 0 0.708 0.028 0.083
residential area 9357 0 0.967 0.322 0.239
extraction of minerals 9357 0 0.119 0.001 0.005
industrial ground 9357 0 0.941 0.104 0.169
service facilities 9357 0 0.733 0.098 0.156
other public facilities 9357 0 0.192 0.010 0.024
socio-cultural facilities 9357 0 0.488 0.037 0.054
railway 9357 0 0.443 0.024 0.037
asphalt road 9357 0 0.277 0.054 0.039
airport 9357 0 0.722 0.002 0.034
park or public garden 9357 0 0.491 0.043 0.078
sports park 9357 0 0.405 0.023 0.038
dry natural ground 9357 0 0.550 0.002 0.023
wet natural ground 9357 0 0.369 0.002 0.015
waters broader than 6 m 9357 0 0.582 0.047 0.056
Table 13.A2. Estimation results (t, z scores in brackets)

Separate effect of dis- Cross-effect of dis-
tance and rail service tance and rail service
quality index quality index
OLS SEM OLS SEM
Second user -0.115*** -0.109*** -0.115*** -0.109***
(-13.093) (-12.991) (-13.117) (-13.009)
Renovation -0.039** -0.050*** -0.039** -0.050***
(-2.393) (-3.276) (-2.408) (-3.289)
Under construction -0.006 0.022 -0.010 0.021
(-0.132) (0.548) (-0.243) (0.521)
Yet to be built 0.026 0.049 0.024 0.048
(0.420) (0.841) (0.399) (0.825)
Rent extension 0.032 0.024 0.034 0.026
(1.091) (0.891) (1.159) (0.950)
Sub lease 0.066*** 0.048** 0.067*** 0.048**
(3.260) (2.507) (3.294) (2.524)
Building and civil engineering -0.036 -0.030 -0.037* -0.030
(-1.637) (-1.432) (-1.657) (-1.431)
Trade and repair companies -0.069*** -0.067*** -0.070*** -0.068***
(-4.099) (-4.228) (-4.157) (-4.273)
Transportation and storage -0.020 -0.025 -0.023 -0.028
(-0.956) (-1.258) (-1.093) (-1.407)
Communication companies -0.002 -0.013 -0.002 -0.011
(-0.101) (-0.705) (-0.105) (-0.630)
Credit and insurance services 0.117*** 0.107*** 0.116*** 0.108***
(6.480) (6.260) (6.410) (6.313)
Financial business services 0.074*** 0.052*** 0.073*** 0.052***
(4.441) (3.304) (4.407) (3.342)
Other business services 0.018 0.008 0.018 0.009
(1.345) (0.668) (1.331) (0.738)
Computer companies -0.001 -0.006 -0.002 -0.007
(-0.039) (-0.438) (-0.117) (-0.520)
Public administration, defence or -0.014 -0.004 -0.015 -0.003
social security (-0.845) (-0.271) (-0.877) (-0.197)
Education and health care -0.108*** -0.096*** -0.109*** -0.095***
(-6.391) (-6.063) (-6.459) (-6.002)
Other institutions and companies -0.076*** -0.071*** -0.076*** -0.070***
(-5.255) (-5.208) (-5.292) (-5.124)
Missing category 0.023 0.020 0.021 0.019
(1.505) (1.410) (1.406) (1.358)
cultivation under glass -1.115*** -0.764*** -1.061*** -0.743***
(-5.536) (-3.232) (-5.277) (-3.147)
other agricultural use -0.054 -0.014 -0.046 -0.015
(-1.519) (-0.270) (-1.321) (-0.294)
Table 13.A2. Estimation results (t, z scores in brackets, continued)

tance and rail service in- tance and rail service
dex quality index
OLS SEM OLS SEM
Forest 0.109** 0.165** 0.117** 0.171**
(2.227) (2.483) (2.393) (2.572)
residential area -0.094*** -0.047 -0.078** -0.034
(-2.705) (-0.958) (-2.252) (-0.697)
extraction of minerals -1.665*** -1.471** -1.650*** -1.524**
(-2.715) (-2.300) (-2.692) (-2.384)
industrial ground -0.286*** -0.175*** -0.273*** -0.169***
(-7.395) (-3.336) (-7.076) (-3.218)
service facilities -0.062 0.144 -0.056 0.144***
(-1.550) (2.660) (-1.403) (2.664)
other public facilities -0.002 0.192 -0.050 0.132
(-0.013) (1.270) (-0.367) (0.869)
socio-cultural facilities -0.097 -0.007 -0.085 0.012
(-1.489) (-0.088) (-1.304) (0.159)
Railway -0.297*** -0.132 -0.300*** -0.127
(-3.170) (-1.306) (-3.204) (-1.251)
asphalt road -0.108 -0.073 -0.106 -0.068
(-1.246) (-0.756) (-1.219) (-0.699)
Airport 0.194** 0.017 0.210** 0.013
(2.089) (0.163) (2.257) (0.128)
park or public garden 0.296*** 0.435*** 0.296*** 0.426***
(5.657) (6.999) (5.683) (6.885)
sports park -0.167* -0.052 -0.139 -0.055
(-1.852) (-0.508) (-1.556) (-0.531)
dry natural ground 0.369*** 0.673*** 0.436*** 0.694***
(2.763) (4.372) (3.277) (4.538)
wet natural ground -0.198 0.156 -0.177 0.146
(-0.987) (0.724) (-0.882) (0.679)
waters broader than 6 m 0.063 -0.156* 0.059 -0.154*
(0.946) (-1.814) (0.890) (-1.793)
Year 1986 -0.011 -0.053 -0.009 -0.053
(-0.264) (-1.375) (-0.232) (-1.371)
Year 1987 0.019 0.005 0.021 0.006
(0.493) (0.143) (0.536) (0.170)
Year 1988 0.026 -0.001 0.029 -0.002
(0.710) (-0.021) (0.804) (-0.052)
Year 1989 0.026 0.001 0.028 0.001
(0.744) (0.034) (0.800) (0.026)
Year 1990 0.097*** 0.076** 0.097*** 0.075**
(2.803) (2.346) (2.811) (2.306)
Table 13.A2. Estimation results (t, z scores in brackets, continued)

tance and rail service in- tance and rail service
dex quality index
OLS SEM OLS SEM
Year 1991 0.132*** 0.112*** 0.133*** 0.111***
(3.995) (3.587) (4.021) (3.555)
Year 1992 0.137*** 0.107*** 0.137*** 0.107***
(4.123) (3.395) (4.115) (3.399)
Year 1993 0.141*** 0.115*** 0.140*** 0.114***
(4.272) (3.694) (4.257) (3.678)
Year 1994 0.140*** 0.120*** 0.140*** 0.118***
(4.344) (3.940) (4.321) (3.879)
Year 1995 0.178*** 0.167*** 0.178*** 0.166***
(5.663) (5.597) (5.653) (5.561)
Year 1996 0.217*** 0.192*** 0.216*** 0.190***
(6.955) (6.528) (6.932) (6.451)
Year 1997 0.237*** 0.220*** 0.237*** 0.219***
(7.630) (7.497) (7.621) (7.449)
Year 1998 0.278*** 0.261*** 0.279*** 0.261***
(8.910) (8.873) (8.939) (8.859)
Year 1999 0.348*** 0.342*** 0.348*** 0.342***
(11.058) (11.504) (11.050) (11.506)
Year 2000 0.426*** 0.424*** 0.426*** 0.424***
(13.826) (14.573) (13.811) (14.558)
Year 2001 0.485*** 0.493*** 0.484*** 0.493***
(15.513) (16.723) (15.498) (16.706)
Year 2002 0.520*** 0.508*** 0.520*** 0.507***
(16.742) (17.301) (16.725) (17.265)
Year 2003 0.489 0.488*** 0.489*** 0.487***
(16.128) (17.057) (16.103) (17.021)
Year 2004 0.471*** 0.468*** 0.471*** 0.468***
(15.529) (16.336) (15.511) (16.313)
Year 2005 0.468*** 0.474*** 0.468*** 0.475***
(15.052) (16.174) (15.055) (16.182)
Number of observations (N) 9357 9357 9357 9357
R-squared 0.3602 0.4255 0.3603 0.4257
***
**
*
14 Regional high-speed trains on the Svealand
line: Evaluation of effects
Oskar Fridh and Bo-Lennart Nelldal
Division of Transport and Logistics, KTH, Stockholm, Sweden
14.1 Introduction
Several new or upgraded railway lines, primarily for regional or medium

distance travel at high speed, have been opened in Sweden since the 1990s.
The investment decisions were based on the expected societal benefits of
increased accessibility. It was thought that it might be possible to turn re-
gional imbalances, for example in the Stockholm-Mlaren region, into re-
gional development through high-speed train commuting, and unemploy-
ment might thus decrease. However, some economists and organisations,
for various reasons, have questioned the whole idea of constructing new
railways and their possible effects on the travel market and regional devel-
opment. Therefore, in order to evaluate the effects of the radical change in
train service supply, a before and after study was conducted on the new
Svealand line.
14.1.1 Background
A transport policy reform to establish a separate state authority for rail in-
frastructure (Banverket) was adopted in Sweden in 1988; rail companies in
future would merely have to operate the network. Combined with a socio-
economic view of investment in tracks in the same way as for roads, this
has meant that considerable sums have been invested in improvements to
the Swedish rail network for over 15 years.
In the Stockholm-Mlaren region in particular, several new lines have
been built or older lines modernised to cope with speeds of up to 200
km/h. The main aims were to facilitate daily commuting so that the people
living in the region can live and work in different places, thus smoothing
296 Oskar Fridh and Bo-Lennart Nelldal
imbalances in the housing and labour markets, and spreading the economic
growth from Greater Stockholm to the surrounding area through improved
accessibility. Such investments in regional high-speed trains to achieve
these ends are rare in an international perspective.
The Svealand line was opened in 1997; the purpose was primarily to in-
crease accessibility in the region, not to replace the old line, which never-
theless was threatened by abandonment. It consists of a newly built, and
for the most part, single-track line from Sdertlje to Eskilstuna (79 km),
and a modernised stretch from Eskilstuna west to Valskog (35 km; see
map in Figure 14.1). Trains can thus continue towards rebro and Halls-
berg. Together with other infrastructure investments in the Stockholm area,
this means that regional high-speed trains can cover the distance between
Eskilstuna and Stockholm (115 km) in just an hour, making five stops
along the way.
Sala Heby Morgon-

N
The Mlar valley Virsbo
gva
Uppsala
Ransta
and surroundings
Ramns
Knivsta
Sura- Tillberga
Railways in 2006 hammar
Arlanda
Mrsta
0 10 20 30 40 km Hallsta- Vsters
hammar Enkping
Kolbck Blsta
Dingtuna
Upplands
Kping Bro Vsby
Frvi
Valskog Kvicksund
Torshlla
Kungsr Strngns STOCK-
Arboga Eskilstuna Sundbyberg HOLM
Hovsta central
lvsj
Mariefred
kers Lggesta
rebro styckebruk Sdertlje Tumba Flemings-
berg
Nykvarn
Hlleforsns Vster-
Kumla haninge
Mellsa
Jrna
Flen
Hallsberg
Mlnbo
Vingker Gnesta
Katrineholm
Vagnhrad
Nyns-
Trosa hamn
Figure 14.1. The Svealand line (shown in semi-bold) opened in 1997
A new high-speed line for regional travel in a corridor with a modest

proportion of public transport was a radical change in the supply of train
services. Questions were raised as to what effects on the travel market,
travel behaviour, and accessibility the Svealand line might be expected to
have. The market effects of the introduction of the regional high-speed
trains on the Svealand line has (already) been studied in an ex ante, ex post
case study in a research project run at the Royal Institute of Technology,
Regional high-speed trains on the Svealand line 297
Kungliga Tekniska Hgskolan (KTH), in Stockholm from 1996 to 2002. A

doctoral thesis was published (Fridh 2003), as well as a paper (Fridh
2005). Our aim in this chapter is to briefly present the effects of the
Svealand line. In addition, we will discuss valuation of the supply, and the
supplydemandaccessibility interaction of transportation.
14.2 Methods
14.2.1 The research project
The aim of the research project was to find general market effects arising
from the improved market supply, i.e. the connection between a radically
improved supply of train services on the one hand, and travel demand and
accessibility on the other. The general market effects expected to occur can
be used as indicators that changes in the social structure will become ap-
parent in the long-term. The research project was carried out as a before
and after-study (ex ante, ex post) of the supply and its influence on demand
and accessibility.
14.2.2 The four-step model
In conventional forecasting models changes in services are normally calcu-

lated according to given external prerequisites in four steps, the so-called
four-step model (see Figure 14.2).
Location choices Trip decision one day three years

Some years before
some decades after
Destination choice one day one year
Mode choice one day
Route choice
The Svealand line
Figure 14.2. The four-step model

The four-step model states that changes in traffic supply affect demand
momentarily, and that in reality the effects materialise at different times.
The four steps can also be found in the land use transport feedback cycle,
i.e. trip generation, trip distribution, mode choice, and route choice. By es-
timating demand with the future requisites (the supply) at equilibrium, a
forecasting model based on the four-step model can calculate the amount
of future travel, i.e. a demand forecast. The research project studied only
the Svealand line corridor, so only route choice is given. Instead it is of
greater interest to include how location affects travelling. A new first step,
location choice, has therefore been added and route choice need not be
considered; this then creates a theoretical linkage to the long-term trip-
generating effects of the social structure.
The changes in demand begin immediately as supply is changed. Some
typical periods for the changes to materialise are from one day to the next
for mode choice, up to a year for trip distribution, over a year for trip gen-
eration, and a change in location can affect demand for several decades af-
ter the change in supply. In general, it takes at least two to three years to
attain a new theoretical equilibrium in the traffic system, and several dec-
ades before equilibrium is attained in the location system. Detailed de-
scriptions of the four-step model can be found, for example, in Ortzar and
Willumsen (2001) or McNally (2000).
14.2.3 Research methods
In the research project the before and after study survey times needed to be
synchronised to the change in supply on the Svealand line (see Figure
14.3).
The study included quantitative field surveys with questionnaires sent
by post to people living along the line, and interviews with public transport
users with the aid of portable computers on buses and trains along the
stretch in question. Both the Stated Preferences (SP) and Revealed Prefer-
ences (RP) methods were used in the interviews to study individuals
knowledge and valuation of the supply and travel behaviour.
The postal questionnaires were sent each year the surveys were made to
a sample consisting of 2400-3000 people between ages 16 and 74 who re-
side along the line. The surveys were carried out three times; in 1997, for
bus traffic before the Svealand line was opened, and in 1998 and 2000 for
high-speed trains. The frequency of response was in the order of 70%. At
the same time, each year surveys were made, some 500 interviews were
held using portable computers among bus and rail passengers, with a re-
sponse frequency in the order of 60%.
Q
Change in supply
- new infrastructure Supply
Demand
1997 1998 2000
Figure 14.3. Outline description of the survey times linkage to the change in
supply on the Svealand line
The same surveys have also been carried out of people living in the ref-
erence population centre, Nykping, and on the Nykping line. Nykping
(27 000 inhabitants, 105 km south west of Stockholm) has not seen the
radical change in public transport supply over the period, but otherwise has
comparable characteristics.
14.2.4 Methodology issues
None of the methods employed in the research project can alone give a
true picture of the changes between before and after the opening of the
Svealand line with acceptable accuracy. The combination of several meth-
ods and different approaches, on the other hand, can identify the changes
and their magnitude. The time aspect is important; there is no way to ob-
tain reliable answers to questions about how a person would have travelled
in the past with different prerequisites, or in the future with unknown, or
possibly only partly known, prerequisites.
Regarding the interviews, the importance of having a verified and repre-
sentative sample of respondents cannot be emphasised enough. The main
objection in the field study involving passengers on buses and trains is that
the selection in the sample is not truly representative. Another method, for
example, a total survey over one or more days with questionnaires, would
probably have given more useful results as regards representative public
transport travellers, in the same way that car number recording on the E20
would have given representative car travellers. It was estimated from the
very beginning that a field survey of residents along the line should be able
to compensate for these shortcomings. Aggregated statistics also need to
be included to obtain a better picture.
Regional effects are difficult to identify since the effects are often slow
to appear and are sometimes very diffuse. In the Svealand line research
project this has largely had to be deferred. A longer survey period and
more resources for analysis would improve the prerequisites for describing
the regional effects. The regional effects are nonetheless a result of
changes in accessibility, manifested in changes in the demand for travel.
There are therefore very good prospects that it will be possible to pinpoint
the regional effects of the train service on the Svealand line if the current
research should continue.
All in all, the conclusions regarding the effects of the supply on the
Svealand line on demand and accessibility can nonetheless be assumed to
stand on fairly solid ground. This is due to the fact that many of the effects
are relatively large and distinct, and that several methods have been used
to analyse the effects. In most cases the results point in the same direction.
14.3 Results
14.3.1 Supply and demand
There had been a railway on this route for a long time but supply was far
from good with long travel times and infrequent trains. While the Svealand
line was being built, the old train service was discontinued and replaced by
buses with, in principle, the same frequency as the new trains on the
Svealand line. All the buses went via Strngns, unlike the old rail service,
which required a connection from kers styckebruk. Table 14.1 shows the
main supply and demand data for SJs trains, buses, and high-speed (HS)
trains between Eskilstuna and Stockholm.
When the new trains began operating, bus services were discontinued
and travel times were halved. Frequency is one train an hour, with some
extra trains at peak periods. During the first few years the service was op-
erated using comfortable high-speed trains of the X 2000 type, and travel
time was one hour between Eskilstuna and Stockholm. This resulted in a
marked increase in demand with regard to regional travel by public trans-
port. Travel by train between Eskilstuna and Stockholm increased to 1.6
million trips across the county border in 2001, seven times as many as with
the train service on the old line in 1993. To this increase should be added
the inter-regional travel that takes place without crossing the county border
and the inter-regional travel between rebro and Stockholm that had pre-
viously mainly gone by way of Hallsberg.
Table 14.1. Supply and demand, regional travel on SJ between Eskilstuna and
Stockholm
Period No. of services Travel time Fare, single, No. of jour- Incr.
Mon-Fri, each (hrs:mins) 2nd class neys factor
direction (Skr)1 (000s/yr)2
Up to spring 1993 8 trains 1:40 115 230 1
Autumn 1993-spring 18 buses 1:55-2:20 105-120 440 2
1997
Summer 1997 17 HS trains 1:00 55 1400 6
Autumn 1997 17 HS trains 1:00 110 1200 5
2001 18 HS trains 1:02 113-135 1600 7
1
Fares are shown in current prices. 1 is approx. 9.25 Swedish crowns (Skr; Sep-
tember 2006)
2
Regional travel over the county border (LggestaNykvarn section)
Regional car traffic fell when the high-speed train service began on the
Svealand line, but has subsequently increased, among other things, through
the conversion of the parallel-running 2-lane E20 road to a 4-lane motor-
way. The rail services market share has increased from 6% to about 30%
for regional trips (between Eskilstuna and Stockholm or shorter) in the
E20/Svealand line corridor (see Figure 14.4). For trips between
Eskilstuna/Strngns and Stockholm the market share is even greater. For
travel that involves a connection to the most important destinations, prin-
cipally via the Lggesta (change from Mariefred and kers styckebruk)
and Sdertlje South stations, the market share for train travel is noticeably
lower. However, for trips to and from Stockholm especially, the Svealand
line has meant increased travelling. Approximately half of the high-speed
train passengers on the Svealand line in 1998 had previously been bus pas-
sengers. The other (approximate) half were newcomers to public transport:
car travellers (15%) and new travellers (30%).
Regional trips
(millions per year)
5
Car
3 Estimated regional travel
Long-distance
express bus
2
Sdermanland county
PTA buses
1
Train
The Svealand line
Train SJ bus
0
1993 1994 1995 1996 1997 1998 1999 2000 2001
Figure 14.4. Estimated total regional cross-border travelling between the counties
of Sdermanland and Stockholm (LggestaNykvarn section) 1993-2001
A number of background factors have been especially advantageous

over the Svealand lines introduction period. Employment and private con-
sumption have risen at the same time as the price of petrol increased be-
tween 1998 and 2000. The communities along the Svealand line, their
population and the influx of new residents, have both increased over the
same period, while car ownership has stagnated and even declined, relative
to the country as a whole, in the areas closest to the railway stations. The
three last effects, and also the increase in employment, can partly be ex-
plained by the improved accessibility resultant from the train services on
the Svealand line. These factors are summarised in Table 14.2 (Fridh
2003).
Table 14.2. Factors determining demand for train services

Factor 1990-1997 1998-2000
Population
Influx
Employment
Car ownership 0 0( )
Private consumption
Price of petrol 0
14.3.2 Valuation and socio-economic factors
The surveys show that residents knowledge of the supply was good and
that their valuation of it was high. Motorists especially are attracted by
high-speed trains, both with regard to travel times and comfort, and proba-
bly also image. A number of other surveys also point in the same direction;
design, a clean and fresh impression and modern trains, and a high degree
of comfort are key factors in attracting travellers (see for example Kotten-
hoff 1999). This is otherwise often one of the weaknesses of the railways;
the trains endure for longer, they can still operate even though they are old,
outmoded and uncomfortable. But, as can be seen, this results in fewer
passengers. Consequently it would seem that the Svealand line has got off
to a good start with more passengers, thanks to the high-speed trains on the
line during its first years of operation.
The low, or insignificant, valuation of trains and high-speed trains as
compared to buses evident among the residents of Mariefred, kers sty-
ckebruk and Nykvarn, can probably be attributed partly to the change to
and from bus and train in Lggesta, and partly to the fact that these travel-
lers often find that the Stockholm-bound trains are full at peak periods.
The advantages of high-speed trains cannot compensate for the need to
change and perhaps be required to stand on the train. Commuters care less,
however, about the mode of transport, and more about travel time, fre-
quency of service, and fares. Residents of Eskilstuna and Strngns on av-
erage value the high-speed train mode of transport at 40 Skr and 20 Skr re-
spectively, and the ordinary train at 20 Skr and 10 Skr respectively, more
than the bus. This value includes the greater comfort and better service of
the train compared to the bus (see Figure 14.5) .
The travel time for regional journeys by the Svealand line (50-70 Skr/h)
is valued up to the same level as for interregional journeys (70 Skr/h),
twice the figure for regional journeys given in Banverkets calculation
guide (35 Skr/h; Berkningshandledning 2001). For the Nykping line,
however, the valuation of the regional journeys agrees with the values in
the calculation guide.
Motorists in particular highly value the high-speed trains, both because
they are fast and because they are very comfortable, while normal trains
are less attractive, and buses are hardly attractive at all to motorists. Both
before and after the opening of the Svealand line, travel by public transport
was greatest among those with access to a car at times, i.e. the infrequent
motorists. This group is comprised of many people with a relatively high
degree of mobility who live in households with a car, but who do not have
continuous access to it. The group therefore consists predominantly of
married and cohabiting women, but also a number of young people living
with parents.
Proportion
80%
70%
High speed train
60%
1998+2000
50%
1997
40% Train
30% 1998+2000
1997
20% 1998+2000
Bus
10%
1997
0%
70 Skr 120 Skr Fare
Figure 14.5. The proportion of motorists resident in the centre of Eskilstuna who
rated both a journey by car and a journey by public transport between home and
the centre of Stockholm, but gave the bus, train or high-speed train a higher score
than the car. The slope of the line between the 70 Skr and 120 Skr points gives an
indication of the price elasticity, but since there are only two points, it is impossi-
ble to get any idea of the shape of the curve. (1 = 9.25 Swedish Crowns/Skr, Sep-
tember 2006).
The survey shows that it is people who always have access to a car, the
habitual motorists, who have proportionally increased their travelling by
public transport the most. This group consists predominantly of married or
cohabiting middle-aged men. The non-motorists include many single peo-
ple who for financial or health reasons often lack the prerequisites to ac-
quire a car. For all groups, trips made less often (in the past month) have
increased since the regional high-speed trains began operating; this indi-
cates an increase in social and cultural contact outside peoples home areas
through the use of public transport. The socio-economic factors that give
significant increases in travel by public transport include men, age group
25-44, self-employed, full-time employed, or university educated people.
Significant reductions in car travel among full-time employed or secondary
school educated people can also be seen.
The proportion of households with two or more cars has not increased
along the Svealand line since the train service began, while the proportion
increased by three percentage points for the country as a whole between
1997 and 2000. For the latter year, consequently, there are significantly
fewer households with more than one car in the centres of population
along the Svealand line, and this trend can be seen most clearly in the cen-
tral parts of Eskilstuna and Strngns.
The improvement in supply and greater accessibility to public transport
as a consequence of the Svealand line train service has given rise to differ-
ences in the generation of journeys, in car ownership, in distribution of
mode of transport, and to a certain extent, in the choice of destination. The
effects are most apparent among residents close to the railway stations,
within walking distance of the stations in Eskilstuna and Strngns. For
example, they have changed the distribution of transport mode for journeys
on the E20/Svealand line from 20-25% public transport in 1997 to 45-50%
public transport in 2000. Farther away from the stations, the proportions
using public transport are substantially smaller.
14.3.3 Accessibility
Travel times with public transport along the Svealand line were signifi-
cantly shorter once the line had opened, just as the new, comfortable trains
had raised the level of comfort. For a trip from the central parts of
Eskilstuna, Strngns and Nykvarn to the centre of Stockholm, the train is
always faster than the car. The potential accessibility of, first and foremost,
Stockholms large labour markets, has been influenced by the faster public
transport connections. Figure 14.6 explains the potential accessibility to
workplaces from central Eskilstuna.
1 000 000
900 000
800 000
700 000
600 000
No. of jobs
Train
500 000
Bus
400 000
300 000
200 000
100 000
0
0 20 40 60 80 100 120
Travelling time (minutes)
Figure 14.6. The number of places of work within 120 minutes travel time door-
to-door from the station in Eskilstuna
High-speed trains enable individuals to reach more places of work than

was possible with the bus service within 60 minutes door-to-door travel
time from the centre of Eskilstuna. The increase within an hours travel
time, which is most frequent for travelling to and from work, is therefore
comparatively modest. Half a million work places can be reached within
an hour and 25 minutes by high-speed train. Longer journeys to work than
this are unusual, and a large proportion of Stockholms labour markets are
too far away, despite the regional high-speed trains on the Svealand line.
Generally speaking, commuting has increased over the period studied.
There are, however, differences between the various locations, as can be
seen by a comparison with Nykping. The increase is greatest in
Eskilstuna, but the increase is also substantial in Strngns, and both in-
creases coincide with the introduction of the new train service on the
Svealand line. The number of people commuting to the municipality of
Stockholm from Eskilstuna increased by 125% between 1996 and 2000.
Commuting to Eskilstuna from Stockholm also increased, but from a lower
level. Commuting to and from Strngns was greater from the outset and
has increased by 40-60% in four years (Figure 14.7).
25% The Svealand line opens
20% Proportion of
commuters
15% Eskilstuna out
Strngns out
10% Nykping out
5%
0%
1993 1994 1995 1996 1997 1998 1999 2000
Source: Statistics Sweden (SCB)

Figure 14.7. Proportion of commuters travelling to the municipality of Stock-
holm from the municipalities of Eskilstuna and Strngns along the Svealand line,
and from the reference population centre, Nykping.
14.4 Discussion and conclusions
14.4.1 Market behaviour accessibility
Conditions in the Stockholm region are not unique from an international

perspective, but they probably contribute to the comparatively large effects
that the train services on the Svealand line have had. What is important is
that Stockholm is the centre of the region, but at the same time it is also a
national centre, which means it has a large supply of qualified places of
work, and household and company services. It also means a heavy traffic
load, along with high generalised costs for car driving in the form of con-
gestion and high parking charges in the city centre. Compared to the car al-
ternative, regional high-speed trains therefore offer an attractive way of
providing this supply.
One of the most noticeable results of the increased accessibility is that
commuting in both Eskilstuna and Strngns shows a marked increase.
The train service on the Svealand line has put Eskilstuna within commut-
ing distance of Stockholm. Travel times are too long to make the car an at-
tractive alternative here. Consequently, the improved supply has resulted
in a leap in accessibility.
Commuting by train on the EskilstunaStockholm section is a distance
of 115 km and a one hour journey. This is somewhat longer, both in time
and distance, than what is considered to be the general norm for journeys
to work. In an accessibility context, the usual length of acceptable journeys
to work sustainable in the long-run is 45 minutes by car or 60 minutes by
train door-to-door, based on surveys of travel habits. The real travel times
by high-speed train are consequently acceptable to more people than ear-
lier surveys of travel habits had indicated. Under these circumstances fast
regional trains contribute to longer commuting distances through the re-
ductions in generalised travel costs.
The limitations as to which sections are acceptable for commuting are,
however, not absolute for the individual, and they can be assumed to be a
function of the generalised cost of the journey (travel time, fare, comfort,
etc.) and the individuals benefit of commuting a certain distance. The rea-
sons for long journeys to work might therefore be that the income differ-
ence is so great that regional commuting is economically profitable, that
people take other members of the family into consideration when choosing
where to live, or that the journey is comfortable and that the travel time
may also be used for work. Figure 14.8 illustrates principal changes in
travel frequency and generalised travel costs.
Travel frequency Generalised

travelling cost
Regional Inter- Regional

Local regional Inter-
Local regional
15 min 60 min Travelling

time 15 min 60 min T. time
Figure 14.8. The effects of regional high-speed trains can be described by the
broken line; increased frequency of travel in a time interval with longer regional,
and also to a certain extent interregional journeys, resulting from lower general-
ised travel costs. Travel times are approximate.
Regional high-speed trains attract new travellers, primarily because of

their speed, but also for other attributes such as high comfort and good
service. The new supply means that the generalised travel costs are lower.
The consequences of this are that frequency of travel increases in the re-
gional high-speed trains market niche, and that daily journeys on average
are longer. At the same time as the total transportation increases, a greater
proportion of the journeys will be made by high-speed train instead of by
other means of transport. Some of the new journeys will be substitutions
for slower journeys by other means of transport.
The proportion of new journeys is especially large from Eskilstuna to
Stockholm. Strngns already had considerable commuting to and from
the county of Stockholm before the Svealand line opened. Strngns is
within the range of the car from Stockholm, i.e. the area that can be
reached by car for daily commuting, even if daily car journeys take a rela-
tively long time. The transferral of trips from the car to the train is large
here. Outside this area, the high-speed trains have made daily commuting
by train possible, which was previously hardly feasible by train, by bus or
even by car, and this has resulted in a large number of new journeys being
made by train. The range of the car does not stretch to the area between
Eskilstuna and Stockholm for daily commuting, whereas the regional high-
speed train services do.
One conclusion that can be drawn is that clear goals for travel time,
based on accessibility, should be set when planning new infrastructure
where leap effects are identified. Will commuting to work, business trips,
etc. increase with a certain travel time, or are more (or less) comprehensive
measures needed? With these prerequisites it is important to realise that ef-
fects are non-linear, and that there are threshold values.
Limited time, and to a certain extent economic restrictions, determine
how long daily journeys are allowed to take. Considering the fact that most
people have to work, sleep, do housework, and spend time with other
adults and children each day, the time available for a daily journey is about
1 hour one way (Hgerstrand 1970). This figure naturally varies among
different individuals and societies, and is also dependent on what the la-
bour and housing markets look like, what the journeys cost, and comfort of
the journey. Based on the conditions that normally exist in Sweden and
many other industrialised countries, many people would apparently be
prepared to accept a one way journey to work of up to 1.5 hours in total,
which with current modes of transport, limits the maximum commuting
range to about 100 km.
Initially, the regional train has an average speed of 75 km/h and its ter-
minal time is 30 minutes. The bus has the same terminal time but a slightly
slower speed of 70 km/h. The car has a terminal time of 5 minutes and a
speed of 50 km/h for a period of 10 minutes due to urban traffic, and then
80 km/h. With an initial travel time of 1.5 hours, a commuter can travel
about 80 km by car, about 70 km by train, and about 60 km by bus; in 3
hours, 160 km by bus, 170 km by train, and 180 km by car. It is also evi-
dent that the train can never catch up with the car because its initial aver-
age speed is not higher.
Expanded rail services with a maximum speed of 200 km/h and fast re-
gional trains increase the trains average speed to 120 km/h. The train is
then faster than the car over distances greater than 30 km and the 1.5 hour
commuting range increases to 130 km, i.e. almost double, and to 300 km in
3 hours. This means that the train can then not only compete with the car
(and bus) but can also create completely new travel possibilities. Daily
trips up to 130 km one way become possible. With high-speed trains trav-
elling at speeds of up to 300 km/h, the length of a daily journey can be ex-
tended to nearly 200 km, but the cost of daily commuting might then be
very high.
Shorter travel times than today on the Svealand line would have sub-
stantial effects.
The diagram of Figure 14.9 shows the number of jobs that can be
reached by train within two hours from the centre of Eskilstuna for two
different travel speeds. Travel time by train today from Eskilstuna Central
station to Stockholm Central station is 1 hour, and the example shows an
alternative with 20 minutes less travel time. For a travel time of 40 min-
utes, a double track line is needed to avoid trains needing to stop in order
to pass each other, and a speed increase to 250 km/h with new trains. A
differentiated supply is also required with more trains, of which some are
non-stop trains to enable shorter travel times on longer routes. The curve in
the cumulative diagram is steep for the area studied, meaning that each
minute cut from the travel time gives a relatively large increase in potential
accessibility. From the centre of Eskilstuna, 20 minutes shorter travel
times mean that it would be possible to reach 350 000 jobs within one hour
instead of 100 000, and almost 800 000 instead of 350 000 within 1 hr 20
mins (few people would be able to manage longer commute times than this
in the long-run). The situation is similar for Strngns. Travel time by train
in 2000 (48 minutes) is almost what is necessary to give substantial poten-
tial accessibility to the whole of the Stockholm labour market. Ten minutes
less travel time between Strngns and Stockholm would mean that
250 000 more jobs would be within one hours reach of people in
Strngns.
1 100 000
1 000 000
900 000
800 000
700 000
600 000 Faster trains
No. of jobs
500 000 Todays trains

400 000
300 000
200 000
100 000
0
0 20 40 60 80 100 120
Travelling time (minutes)
Figure 14.9. With 20 minutes shorter travel time to Stockholm the number of
jobs that can be reached within reasonable commuting times by train from the cen-
tre of Eskilstuna is greatly increased.
In practice, faster trains would put more of Stockholms labour market

within reach. Another aspect is that shorter travel times would also im-
prove accessibility for people living a little farther away from the railway
stations on the Svealand line than the central areas of the conurbations.
This would then widen the areas around the railway stations in both home
and activity localities.
14.4.2 Valuation of the supply
A good supply seems to increase the valuation of the travel time. This
might be explained by the fact that the short travel times make it possible
for new groups to take the train, and for more trip purposes, given that the
frequency of service is sufficiently high. The new groups value their time
more highly, due in part to higher incomes. Consequently, the value of the
shorter travel time is higher.
Some research findings also suggest that the value is not a linear utility
function; small gains or losses of time (3-5 minutes) are practically unim-
portant, while the value increases dramatically according to a non-linear
function for gains larger than 10-15 minutes. The value also proved to be
lower for long journeys than for short ones, which runs contrary to previ-
ous results obtained using different methods (Hultkrantz and Mortazavi
2001). Transferring these results to the Svealand line means that a gain in
time of the magnitude given by the new train service compared to the old
supply would increase the valuation of the travel time compared with the
mean value that is normally applied, partly because the travel times are
radically shortened, and partly because the distances travelled have be-
come shorter, thus causing shifts in travel frequency and the reasons for
travelling. A greater number of business trips and more commuting mean
high valuations of time compared to leisure journeys.
There are also differences between the valuation of a supply before and
after a change. Generally speaking, the supply is valued more highly after
the change than the same supply described before the change (Kottenhoff
and Lindh 1996; Brthen and Hervik 1997). This would explain part of the
difference between the before and after studies of the Svealand line, and
between the Svealand line and the Nykping line.
Using standardised factors for travel time without differentiating be-
tween commuting and leisure travel, and not differentiating between a
good and a bad supply, and with a linear valuation of improvements, will
consequently disadvantage new investment in a good supply. The valua-
tion of the saving in travel time on fast regional journeys is thus far appar-
ently too low, which means that the calculated economic benefit to society
is too small for this type of project objective. The result is that fewer rail-
ways for regional high-speed services are constructed; instead either road
projects are favoured or there remains lower accessibility without the suit-
able infrastructure.
14.4.3 General effects of the Svealand line
The high-speed train service on the new line meant a considerable im-
provement in supply along the Svealand line. This prompted an immediate
market response in the form of considerably greater demand. The trains
replaced buses, which meant that a group existed who had already trav-
elled by public transport. But these passengers constituted only about half
of all the train passengers on the Svealand line. Consequently, there were
passengers on the trains who had previously not travelled at all, and some
who had previously preferred to travel by car (transferring travellers). The
new passengers are a kind of quantitative measure of the attractiveness of
the supply and the effects on accessibility.
The shortcomings in the old train service supply were so great that it
was largely people who had no alternative who used it. The new train ser-
vice halved travel times, which meant that it was faster to go by train than
by car in certain tours, and this paved the way for the threefold increase in
travelling. The new supply also means that motorists choose to travel by
train, and that train service attracts passengers on its own merits.
One general conclusion that can be drawn is that there is a clear linkage
between supply and demand in public transport. When supply is suffi-
ciently good, demand increases substantially. The greatest effects are
achieved if passengers can simultaneously be offered short travel times,
high frequency of service, low ticket prices, and a high degree of comfort.
The system must also be reliable. Short travel times are the most important
factor in attracting people to public transport, given a reasonable price, but
the high-speed train mode of transport in itself is also important.
The large increase in demand can therefore be explained partly by a
switch from car to train and partly by a switch from bus to train, but also
by a large amount of new travel. The new travel is interesting. While the
substituted travel can be presumed to have positive effects for the envi-
ronment and a gain in travel time that can be used for productive work or
improved quality of life, the new travel should to a greater degree contrib-
ute to the regional effects of the investment. The new journeys can be as-
sumed to be for business purposes, commuting and leisure trips, but with
different degrees of sensitivity to changes in the supply, depending on the
purpose of the journey. Consequently, the greater economic and social ex-
change between the various parts of the region can be assumed partly to
increase growth in the region and partly to contribute to strengthening the
functional region, and thus contribute to equalise local variations in, for
example, the labour markets.
Accessibility by car is already good, which reduces the overall im-
provement in accessibility resulting from the new train service. Despite the
comparatively small general effects, there is reason to conclude that the

train service really has brought about considerable improvements in certain
areas and for certain categories of people. Many people of a productive age
with access to a car have begun to travel by train. In particular, accessibil-
ity to the central parts of Stockholm has increased. Journeys by car to the
city centre of Stockholm mean coping with traffic congestion, as well as
likely parking problems, which limits accessibility in the road system.
Accessibility did improve along the line when the train services began
operating on the Svealand line. However, it appears to be difficult to gain
accessibility over wider areas generally speaking, since it entails a need to
change and travel by public transport feeder connections. Improvements in
accessibility between before and after the change in supply are therefore
greatest nearby the railway stations. The stations catchment areas appear
to be rather limited in size. On the other hand, there is a certain demand for
car parking at the stations. Experience from AVE in Spain (Menendez et
al. 2002), among others, indicates that demand can be assumed to increase
if the train service becomes even faster, compared to travelling by car for
the whole journey.
The Svealand line is mostly meant to contribute to a regional integration
for daily commuting over medium distances. The speed is therefore impor-
tant for opening new markets. As one such project, the Svealand line has
successfully met its expectations.
References
Berkningshandledning (2001) Calculation Guide 2001: BVH 706.00. Banverket

[the National Rail Administration], Borlnge
Brthen S, Hervik A (1997) Strait Crossings and Economic Development: Devel-
oping Economic Impact Assessment by Means of Ex Post Analyses. Trans-
port Policy 4:193200
Fridh O (2003) Introduction of Regional High-Speed Trains: A Study of the Ef-
fects of the Svealand Line on the Travel Market, Travel Behaviour and Acces-
sibility. (Ph.D. thesis) KTH, Division of Transportation and Logistics, Stock-
holm
Fridh O (2005) Market Effects of Regional High-Speed Train on the Svealand
Line. Journal of Transport Geography 13:352361
Hultkrantz L, Mortazavi R (2001) Anomalies in the Value of Travel-Time
Changes. Journal of Transport Economics and Policy 35:285300
Hgerstrand T (1970) What about People in Regional Science? Papers of the Re-
gional Science Association 14:721.
Kottenhoff K, Lindh C (1996) The Value and Effects of Introducing High Stan-
dard Train and Bus Concepts in Blekinge, Sweden. Transport Policy 2:235
241
Kottenhoff K (1999) Evaluation of Passenger Train Concepts: Methods and Re-
sults of Measuring Travellers Preferences in Relation to Costs (Ph.D. thesis).
KTH, Division of Traffic and Transport Planning, Stockholm
McNally MG (2000) The Four-Step Model. In: Hensher DA, Button KJ (eds)
Handbook of Transport Modelling. Pergamon, Oxford, pp 3552
Menndez JM, Coronado JM, Rivas A (2002) El AVE en Ciudad Real y Puertol-
lano. E.T.S.I Caminos, Canales y Puertos, University of Castilla-La Mancha,
Ciudad Real
Ortzar J de D, Willumsen LG (2001) Modelling Transport. 33rd edn Wiley, New
York
15 Rail pricing and the supply of
complementary commercial goods
Eric Pels

the Netherlands
Erik Louw and Jan Jacob Trip

15.1 Introduction
During the second half of the 19th century, railways were the dominant
transport mode, but in the 20th century the dynamism of the railway began
to lose its momentum. Presently, car and truck have become the leading
modes for passenger and freight transport.
However, a new turn in the development of the railway now seems to be
taking place. In many countries large investments are made to improve the
quality of railway systems, in particular the construction of high-speed rail
connections (Vickerman 1996; Haynes 1997). Congestion on the road
network makes rail more competitive, and capacity shortages in the avia-
tion network have led to an interest in applying high-speed rail links as an
alternative to feeder flights, which also reinforces the position of railways
for international trips. These railway developments offer renewed opportu-
nities for areas of larger cities that had previously been experiencing peri-
ods of decline. Railway related issues, be it the development of terminals
for high-speed rail, the construction of high quality office areas near rail-
way stations, or the introduction of light rail have become important
themes in policies to revitalise these cities (cf. Bertolini and Spit 1998; van
den Berg and Pol 1999). The essence of these plans is that railway stations
are not merely nodes where people change from one vehicle to the other,
316 Eric Pels, Erik Louw and Jan Jacob Trip
but where, furthermore, spatial concentrations of high value activities are

realised, which have positive impacts on the local economy. The mobilisa-
tion of resources for these plans is, however, an important problem. Al-
though the dynamics of the railway sector has improved its role in most
transport markets, its role is still limited, so that investments may be con-
sidered as risky. An open question is to what extent it is possible to use the
transport function of railway nodes in such a way that positive spillovers
can be realised for development opportunities in the cities concerned.
In order to address this problem, one first needs to analyse the behaviour
of public sector controlled, corporatised or privatised suppliers of railway
services. These are the suppliers of services that ultimately have a decisive
impact on the attractiveness of cities. The local authority can offer the rail
company a good location (station with sufficient capacity) with good ac-
cessibility and high potential demand, hoping that the rail company will
offer high quality services (high frequency and good [inter]national con-
nectivity). These high quality services may be an important factor in the
location decision of (international) businesses, and may also attract pas-
sengers (e.g. because of higher frequencies) that may boost the local econ-
omy. A profit maximising rail company could, however, offer services that
do not necessarily make the city as attractive as envisaged. Moreover,
railway services provided by an operator are typically a network product,
connecting many cities. Multiple cities may be interested in rail develop-
ment to enhance the local economy, but the number of stops in for exam-
ple, an intercity network, is limited. Competition between cities for sta-
tions then becomes a focus of attention.
The network product may face competition from, but can also be com-
plementary to, services of other rail operators as well as other modes
(Economides and Salop 1992; Roson and van den Bergh 2000). A local au-
thority can try to improve the attractiveness of the node (by investing in
accessibility, improving bus services, and so on) to attract more (compet-
ing) rail services and thus passengers, but then the railway company also
must find it optimal to transport the additional passengers. As rail compa-
nies develop a stronger commercial orientation, their pricing strategies, if
unregulated, may become more like a strategy one would expect from a
monopolist.
The scientific challenge is to analyse the behaviour of railway compa-
nies and the implications for network formation as well as railway station
development. The question is how a privatised rail company arranges its
schedule (and prices) in the face of competition (of other modes). The
questions immediately following are then whether the interaction between
the private rail operators in a competitive environment leads to optimal in-
vestment in new lines and stations, and where new lines and stations are to
Rail pricing and the supply of complementary commercial goods 317
be located (network density and spacing of train stations). The feedback of

railway development on urban and regional development also has to be
considered in this context.
The aim of this chapter is to analyse the economics of rail networks. Op-
timal outputs for a railway company are determined in relation to con-
sumer preferences concerning shopping goods. Various organisation struc-
tures for the railway sector are modelled.
In rail network design (including the spacing of stations), various strate-
gies have been analysed, e.g. maximisation of the number of direct travel-
lers or cost minimisation (Claessens et al. 1998). Since the rail company
and the passenger each has different objectives, the optimal network (from
the operators point of view) may be sub-optimal from the passengers
point of view (see e.g. van Nes 2002). Whether this will occur in practice
depends on the ownership structure of the rail sector (do new entrants have
access to infrastructure or does the private rail company also control the
infrastructure?) and the cost structure of a rail company. In this chapter we
assume that the railway company has a simple profit maximising strategy.
The chapter is organised as follows. We begin in section 15.2 with a
discussion of the synergy between railway operations and area develop-
ment; this section sets the scene for the theoretical model introduced in
15.3. All scenarios introduced in 15.3 are hypothetical. Section 15.4 dis-
cusses the simulation results and 15.5 concludes.
15.2 Synergy between railway operations and station

area redevelopment
Railway stations historically located at the urban fringes were once the true
gateways to cities (Richards and MacKenzie 1986). Because both goods
and passengers entered by way of the city by the railway station, the sta-
tion area became a nodal point in the local economy. Many offices and
factories located there and central business districts became inextricably
linked with the railway stations (Turnock 1998; Juchelka 2002 p. 12).
Even early suburbs developed around railway stations (Hall 1988 p. 274).
However, the car superseded the railway during the 20th century and
gradually became the main transport mode for travel to and from the city.
The share of the number of railway trips in the total number of trips is, at
present, only 2% in the Netherlands. This has made accessibility by road a
much more important location factor than railway accessibility (numerical
examples are given in Bruinsma and Rietveld 1998), both at local and re-
gional levels.
Accordingly, it also became a much more important factor in urban de-

velopment (Graham and Marvin 2001 p. 120). In particular, locations at
motorway nodes became attractive for business activities. The interna-
tional orientation of the railway network also decreased strongly after the
breakthrough of aviation. Railway stations lost their attractiveness for
business activities, and within a few decades station areas largely lost their
leading position in the local economy. Manufacturing moved onto indus-
trial estates at fringe locations, and was soon followed by offices situating
themselves in business parks. As a result, many once lively station quarters
with their shops, hotels, pubs, and restaurants eventually fell into decay.
With the increasing need to encourage the use of public transport, to-
gether with a renewed interest in the revitalisation of inner cities, planners
attention shifted towards railway stations once again. In the 1980s the first
initiatives were taken to redevelop these areas, now located more centrally
in the expanded cities, and to revitalise these still potentially important
nodes in the urban transport system. The main idea is that the availability
at one place, of a wide variety of transport flows and services, enables
planners to utilise existing local development potential, rather than have to
generate economic growth all by themselves. Hence, a variety of transport
flows and services could have a counterbalancing effect in the develop-
ment of a larger diversity of functions. However, because stations have lost
their natural dominant position in this system, redevelopment now has to
be integrated with transport services and local urban redevelopment poli-
cies. With regard to this, a distinction is often made between the station it-
self and the area directly around it, the fast area in which transport is the
primary function, and the slow area, the larger station area where focus is
on residential and commercial functions (Van Hagen 2004).
Thus railway stations are not just considered as nodes where people
change between transport modes (in the fast area), but where spatial con-
centrations of high value activities are also realised (mostly in the slow
area), which have positive impacts on the local economy. Studies on the
local economic effects of these initiatives are rare, however; most focus on
new infrastructure or new, or renewed, high-speed railway stations such as
Euralille, Antwerp Central Station, Brussels Midi, Lige Guillemins,
Dortmund, Rive Gauche and Montparnasse in Paris, and Kings Cross/St
Pancras in London (see for instance, Bertolini 2000; Spaans 2002; Pol
2002). In contrast, little is known about the redevelopment of ordinary sta-
tions in ordinary towns, although Bertolini (1998) mentions a large number
of such projects in France, Switzerland, Sweden, the Netherlands, and the
UK. Moreover, most studies focus on the spatial, architectural and institu-
tional aspects of these projects, which are indeed related to the economic
effects of the projects, but relatively little attention is being paid to the ur-
ban-economic effects as such.
One of the studies that did take into account the redevelopment of exist-
ing stations has been the analysis, carried out by Dutch Railways (NS), on
the effect of current policy initiatives to achieve synergy effects between
the attributes of the station area and the transport services (Vaessens
2004). In this research synergy is interpreted as a kind of agglomeration ef-
fects on the scale of a particular area, more specifically as the added value
between both identities of a station: as a node in a transport network and as
an urban place or location. An important reason for Dutch Railways to
start this research was its division into various companies: NS Passengers,
NS Stations and NS Real Estate. Although each of these companies should
be profitable on its own, they should also create synergy effects for each
other: location synergy should bring about NS group synergy. Instead of
being instrumental to transport services, land and property are now re-
garded as potential sources of profit and cross-subsidisation of less profit-
able transport activities (Bertolini and Spit 1998).
This striving for synergy between the different branches of Dutch Rail-
ways is strongly motivated by the way rail transport has been organised af-
ter the former state company was split up and partly privatised. Dutch
Railways are responsible for paying an infrastructure fee to use the rail
network; in 2005 this amounted to 146 million for the use of the main rail
network, and another 150 million for the exploitation of the Thalys high-
speed service to Brussels and Paris, in which Dutch Railways has a 90%
share (NS 2006). This can hardly be recouped merely with the income
from rail services. Dutch Railways therefore obtained the exploitation
rights of its stations; in 2005 almost 40% of its profit came from the devel-
opment and exploitation of stations, while this comprised only 15% of
turnover (NS 2006). The opposite approach, separating station and railway
exploitation, could lead to a considerable loss of direct income for the
railway system (Harman 1993 p. 22).
In the study by Vaessens, synergy should manifest as an increase in the
performance of the station areas; it was measured by four indicators be-
tween 1996/1997 and 2004:
number of passengers;
retail turnover (in station shops);
real estate prices in the station area (rent level);
customer satisfaction on station artefacts.
The research was carried out in 10 station areas of the so-called type 2
and 3 stations; these stations offer intercity, fast and local train services.
Type 2 stations (n = 7) are situated in the centre of middle-sized towns.
Type 3 (n = 3) stations are suburban stations in large towns and are repre-
sentative as nodes in the local transport system. In two of the type 2 station
areas (including the station itself) Amersfoort and s-Hertogenbosch, vari-
ous measures were taken to create synergy by improving transfer facilities,
increasing the densities in the vicinity of the station, and increasing area at-
tractiveness. Such measures involved enhancing the quality of the urban
design and adding shops and leisure facilities to the area and station. In
these synergy stations a new railway station was constructed, but several
offices and apartment buildings were also built.
Although the study by Vaessens (2004) was limited in scope, as it in-
cluded only a small number of stations and focussed mainly on the effects
within (and between) the various NS companies, it nevertheless reveals the
remarkable differences between the performance of station areas and indi-
cates that synergies are likely to occur. Figure 15.1 summarises the results
by showing that in general the synergy stations have increased their per-
formance more than the other type 2 and type 3 stations. The synergy sta-
tions are only outperformed in passenger number by type 3 stations, but
this is mainly due to the increase of the number of inhabitants in the station
areas. Particularly striking is that commercial activities (retail turnover and
real estate prices) increased most at the synergy stations; this shows that
synergy between various NS companies does exist.
200 synergy stations

type 2 stations
180
type 3 stations
160
index
140
120
100
80
number of customer retail real estate
passengers satisfaction turnover prices
Source: Vaessen 2004
Figure 15.1. Performance of three types of stations for number of passengers, cus-
tomer satisfaction, retail turnover and real estate prices between 1996/1997 and
2004 (1996/1997 = 100).
But what about the wider implications for the station area, the slow
area? Some conclusions can be drawn from the two synergy stations s-
Hertogenbosch and Amersfoort. Although in both cases a new station
building was built and the surrounding area was transformed, in s-
Hertogenbosch the transformation of the station area was carried out more
thoroughly (see chapter 7 of this book, and Peek and Van Hagen 2001;
Bruil et al 2004). Here, in contrast to Amersfoort, a joint vision existed be-
tween Dutch Railways and the municipality. Furthermore, the transforma-
tion covered a larger area than in Amersfoort. In s-Hertogenbosch the sta-
tion passageway to the platforms also functions as a connection between
the city centre and a former industrial estate that is being transformed into
a mixed-use area, including residential and office functions, but also cater-
ing, and the Court of Justice. The station as a node is therefore integrated
within the urban context.
The mobilisation of resources for these plans is an important challenge.
Many station area redevelopment projects are motivated primarily by the
local spatial and economic context, and are induced by the transport ser-
vices as a type of catalyst; this is the case for many of the projects dis-
cussed in this book, such as Euralille and the South Axis. Based on the de-
velopment potential of the railway station the assumed synergy between
station and spatial-economic development these plans potentially serve
as a platform to unite participants and investors towards an already exist-
ing goal (Trip 2005). However, whereas the railway sector is more dy-
namic than previously, its role in transport markets remains limited. In-
vestments based on railway station development may therefore be
considered as risky. Private developers especially that are involved in vir-
tually all station redevelopment projects, still tend to prefer motorway lo-
cations, although their appreciation of rail transport has increased, particu-
larly where prestigious high-speed trains are involved. It therefore remains
an open question as to what extent the transport function of railway nodes
could be used such that positive spillovers can be realised for development
opportunities in the cities concerned. In the next section we model the in-
teraction between railway activities (travel) and commercial activities
(shopping). We determine under what conditions (organisational struc-
tures) the spillovers mentioned above will be the highest.
15.3 The model
In this section we model the interaction between railway services and eco-
nomic activities as described above. We discuss four different organisa-
tional structures for the railway sector (public, private, private operators of
complementary goods, and competition between nodes). Following our
discussion of organisational structures we use simulation analysis to de-
termine which structure yields the highest welfare and profit levels. The
network analysed here is depicted in Figure 15.2, in which there are three
stations, two links and two markets. Passengers travel from an origin (O)
to a destination (D1 or D2) for shopping, and then return to the origin. For
simplicity, there is no market between D1 and D2.
At locations D1 or D2 (and only at these locations) a good is consumed.
As will be explained below, the willingness-to-pay (WTP) for travel de-
pends directly on the consumption of the other good (i.e. shopping). In
other words, if a consumer plans to spend a lot while shopping, the WTP
for travel is relatively low (and vice versa).
O D D
Figure 15.2. Network configuration
15.3.1 Households
Households residing in location O travel to location Di, where the subscript

i denotes the exact location, and consume the shopping good yi at that spe-
cific location (the good is not available at location O). The people living in
O derive utility from the shopping good, but also from travelling itself:
travelling to the shopping location and walking around also contributes to
utility (i.e. the higher the quality of the place, the higher the utility). We
denote the number of trips from O to destination Di as qi, and the amount
of the shopping good consumed at location i as yi. Note that we assume
that the traffic flows between the two end-points in a market are symmet-
ric, so that we need not specify a return-market. Finally, qo is the consump-
tion of all other goods (qo is used as the numeraire good). We assume that
the utility function is quasi-linear (linear in qo and possibly non-linear in
the other goods):
U (q1 , q 2 , y1 , y 2 , q o ) = u (q1 , q 2 , y1 , y 2 ) + q o (1)
Note that this formulation explicitly considers the possible interaction

between the number of consumer (shopping) goods bought at both loca-
tions, and the number of trips to these locations. By substituting the

budget-constraint, we can rewrite the utility maximisation problem as fol-
lows:
max U (..) = u (q1 , q2 , y1 , y 2 ) +

q1 ,q2 , y1 , y2
(2)
z p1 q1 p2 q2 t1 y1 t 2 y 2
where z is total income, pi is the fare for a return trip to Di,, and ti is the
price of one return trip (yi). Using standard microeconomic theory, we can
determine the inverse demand functions for trips and consumer products
from the first-order necessary conditions for utility maximisation:
pi = u (..) qi (3)
t i = u (..) y i (4)
At each level of qi and yi, and given income, the inverse demand func-
tions give us the prices that make qi and yi the desirable (i.e. utility maxi-
mising) consumption levels. In other words, for each level of consumption,
the inverse demand functions give us the maximum WTP for these goods.
We assume a quadratic structure for u(..):
2 2

u (q1 , q2 , y1 , y 2 ) = qi + qi + qi q i +
2
i =1 i =1 2
2 2

y + 2 y +
2
i i (5)
i =1 i =1
2 2
yi y i + il qi yl
i =1 l =1
This structure is far less restrictive than a linear specification, and it

guarantees a linear relation between pi on the one hand, and qi and yi on the
other:
2
pi = u (..) qi = + qi + q i + il y l (6)
l =1
In other words, consumption in the market for rail trips (shopping

goods) depends on the consumption of the shopping good (rail trips). The
inverse demand function is used in the following sub-sections to formulate
the profit and welfare functions. We next model different organisation
structures for the railway systems. The welfare economic impacts of these
different structures are described using the inverse demand function as de-
scribed above.
15.3.2 The railway company: Joint profit maximisation
The first organisational structure analysed is that which a single (private)

railway company offers both the transport services and the shopping facili-
ties. The railway company in this scenario resembles a company aiming to
exploit the synergies as described in the previous section. In this model the
synergies are limited to demand effects as discussed below.
In this case the railway company maximises profits from operations be-
tween O and Di, i = 1,2 and the profits from commercial activities at or
near the stations. The revenues obtained from the transport activities are
simply the ticket sales to passengers (n(p1q1 + p2q2), where n is the
number of consumers). The railway company incurs a cost cj (including an
infrastructure charge) for each passenger movement on link j. Total (vari-
able) costs are therefore n[q1(c1)-q2(c1+c2)]. Moreover, the railway
company receives revenues from the commercial activities
(n(t1y1+t2y2)), and incurs a cost ki per unit sold at location i. The inverse
demand functions for travel between O and Di and the commercial good
are given by Eqs. (3) and (4).The objective function therefore is:
max = n ( p1 q1 + p 2 q 2 + t1 y1 + t 2 y 2 )
q1 , q 2 , y1 , y 2
(7)
n (q1 c1 + q 2 [c1 + c 2 ] + y1 k1 + y 2 k 2 )
The first-order necessary conditions for profit maximisation are (for

simplicity and without loss of generality, we impose n = 1):
2 2
q i = + 2 q i +2 q i + 2 il y l ij c ji = 0 (8)
l =1 j =1
2
y l = + 2 y l +2 y l + 2 il qi k l = 0 (9)
i =1
where i,j takes on the value 1 if link j is used in market i. Rewriting Eq.
(8), making use of Eq. (6), we see that the price per train trip equals mar-
ginal cost plus a mark up over marginal cost due to market power.
2 2
p i = ij c ji q i q i il y l (10)
j =1 l =1
Note that this mark up also includes a component representing commer-

cial activities (the last term on the right-hand side). If there is a strong de-
pendency between rail trips and shopping at location Di, and the combina-
tion of (or interaction between) shopping and travelling results in an
increase in utility (i.e. i,l > 0), the profit maximising railway company sets
the ticket price relatively low; this increases demand for rail trips and
shopping goods increases. A potential loss in profits from the rail activity
is compensated by an increase in profits from the shopping activity if the
rail fare meets the condition in Eq. (10). Note that, if the interaction effect

( il y l ) is larger than the market power effect (- qi- q-i), the rail
fare may in fact become negative. Whether this is a feasible solution de-
pends on the parameterisation of the model.
Equations (8) and (9) form a system of 4 equations in 4 unknowns,
which can be solved analytically. Because the analytical solution is quite
difficult to interpret, we present a numerical solution in the next section.
We have shown here that a profit maximising railway company accounts
for the interaction effects between the various products. Strong consumer
preferences for commercial activities and rail trips lead to relatively low
rail fares.
15.3.3 Independent profit maximisation
Next we look at the case where the (private) operations of the transport and
commercial activities are separated: each activity is operated by an inde-
pendent (private) company. We thus have two objective functions:
max r = n ( p1 q1 + p 2 q 2 ) n (q1 c1 + q 2 [c1 + c 2 ]) (11)

q1 , q 2
max c = n (t1 y1 + t 2 y 2 ) n ( y1 k1 y 2 k 2 ) (12)

y1 , y 2
In this case the railway company no longer takes into account the inter-
action effect between rail services and commercial activities. In terms of
the discussion in section 15.2, this means that there is no effort to cash in
on synergy effects. We observe that this is reflected in the first-order con-
ditions (for simplicity and without loss of generality, we impose n = 1):
2 2
r qi = + 2 q i +2 q i + il y l ij c ji = 0 (13)
l =1 j =1
2
c y l = + 2 y l +2 y l + il qi k l = 0 (14)
i =1
Rewriting Eq. (13), again using Eq. (6), we obtain:
2
pi = ij c ji q i q i (15)
j =1
The monopolistic mark up over marginal costs does not include a com-
ponent representing commercial activities: the dependency between travel
and shopping is ignored by the independent monopolists. As a result, ticket
prices will be relatively high compared to the previous case.
Equations (13) and (14) again form a solvable system of 4 equations in 4
unknowns. Again, the analytical solution is difficult to interpret. We there-
fore present a numerical solution in the next section.
15.3.4 Local competition
In the previous scenarios passengers had to choose between two stations

(shopping locations), but the shopping activities at the two locations were
offered by a single agent. In this scenario we consider the case where two
independent firms offer commercial activities at or close to stations. We
have three players: the railway company, and two (spatially-differentiated)
suppliers of the commercial good. In this case the profit function for the
commercial agent at location i is:
max c ,i = n y i (t i k i ) (16)
yi
where
2
t i = u (..) t i = + y i + y i + il ql (17)
l =1
The first-order necessary condition for profit maximisation is:
2
c ,i y l = + 2 y l + y l + il qi k l = 0 (18)
i =1
By rewriting Eq. (18) we see that the fare is equal to the marginal cost
plus a monopolistic mark up (ti = ki- yi). Note that this monopolistic mark
up is smaller than the mark up in scenario 2 where the price for the shop-
ping good can be written as ti = ki- yi- y-i. 1 The price in scenario 2 is
therefore higher than in the current scenario as long as < 0. From Eq.
(17) we expect to be negative (since yi and y-i are substitutes, an increase
in y-i reduces demand and revenues at location i). Firm is pricing (output)
decisions have a direct impact on the output and profitability of commer-
cial activity at the other location (-i): a relatively low price at i reduces
demand and profitability at -i. But because we now consider two separate
firms, only firm i considers the impact of its decisions on its own profits. A
firm that maximises profits from all commercial activities (scenario 2) in-
ternalises the effect of pricing decisions in market i on outputs in market -i,
and will balance an increase in demand at i against a decrease in demand at
-i. A firm in the current scenario need not account for the fact that a price
that is too low at location i will decrease demand at location -i. The prices
in the two scenarios are therefore different.2
Because the rail fares directly depend on the commercial output, (Eq. 6),
the ownership structure in the shopping sector has a direct impact on the
rail fare. We expect the commercial output to be higher in the case of local
competition (because prices are expected to be lower). If i,l > 0, this
means that the rail fare in scenario 4 is higher compared to the fare in sce-
nario 2. In scenario 4 the prices for the shopping goods are relatively low
so that the output is relatively high. The railway company now cashes in
on the cross-effect with rail: consumers value the interaction between
shopping and rail. When commercial output is relatively high, utility is
relatively high, and demand for rail trips is relatively high. The railway
1 This price is determined by taking the first-order condition for profit maximi-
sation for the commercial activities and substituting for the inverse demand func-
tion.
2 The analysis in the two previous sections rests on the fact that q and y are
i i
complements; the results were first obtained by Economides and Salop (1992).
The shopping goods yi and y-i are assumed to be substitutes: competition means
lower prices.
company attempts to capture part of the benefits by setting relatively high

fares.
15.3.5 Global welfare maximisation
Finally, we scrutinise the welfare optimum. In this case a public party

maximises overall welfare (total consumer surplus plus profits). This sce-
nario provides the welfare level (and output and price levels) against which
the outcomes of the other scenarios are evaluated in the next section. The
welfare function (the public partys objective function) is:
max = n [u (..) + y p1 q1 p 2 q 2 t1 y1 t 2 y 2 ] + (19)

q1 , q 2 , y1 , y 2
The first RHS term (in brackets) is the total consumer surplus (total
benefits minus expenditures on travel and shopping (see e.g. Arnott and
Yan 2000). The second RHS term is the total profit of railway operations,
including commercial transactions. Note that this expression can be simpli-
fied further because consumer expenditures and railway company profits
cancel each other out. The first-order necessary conditions for welfare
maximisation are (again assuming that n = 1 for simplification):
2
qi = p i ij c ji = 0 (20)
j =1
y l = t l k l = 0 (21)
As expected, both prices are set at the marginal cost level. Eqs. (20) and
(21) form a system of 4 equations in 4 unknowns, and can be solved ana-
lytically. Since the analytical solution is quite involved, we present a nu-
merical solution in the next section.
15.4 Numerical solutions
In this section we present numerical solutions to models described above.

Table 15.1 contains the parameter values used in the analysis. Since the
objective is not, at this stage, to model real life networks, the parameter
values may not correspond to those of real life. However, the parameter
values are chosen such that the demand elasticities in equilibrium corre-
spond to real life values. With these parameters we are able to derive out-
put and price levels, and compare the welfare and profit levels in the dif-
ferent scenarios. The units in which prices and outputs are measured are
not essential for the interpretation of the results. When this model is ap-
plied to a more realistic case, this is of course no longer true.
Table 15.1. Parameter values
1,1= 2,2 1,2= 2,1 c1 c2 k1 k2

50 -0.75 -0.01 10 -0.2 0.01 0 0.1 1 1 0.5 0.5
Using these parameter values, we can derive the equilibrium values

given in Table 15.2, in which we consider the four cases described above.
In scenario I a private railway company maximises joint profits (i.e. profits
derived from railway operations and commercial operations). In scenario II
the profits of the railway and commercial activities are maximised sepa-
rately by two separate (private) operators. In scenario III two local compa-
nies offer the shopping good to consumers arriving at the nearby station,
while a single private company operates the rail network. In scenario IV
welfare is maximised by a single public authority operating both the rail-
way and commercial activities.
Table 15.2. Optimal prices and outputs and profits

p1 p2 q1 q2 t1 t2 y1 y2
I 25.5 26 37.58 36.85 5.25 5.25 40.53 40.15
II 27.04 27.53 34.27 33.58 6.96 6.93 30.79 30.61
III 27.08 27.57 34.32 33.66 6.81 6.77 31.54 31.67
IV 1 2 75.16 73.70 0.5 0.5 81.06 80.30
Table 15.2 contains the equilibrium prices and outputs. As could be ex-
pected, prices are set at the marginal cost level in scenario IV, and are
therefore welfare maximising. The profit maximising prices are higher
when two separate firms operate railway and commercial activities (sce-
nario II), compared to the scenario in which one firm maximises profits of
both activities (scenario I). This is because these separate firms do not ac-
count for the fact that a high price for one activity (rail travel) has a nega-
tive effect on the demand for, and profitability of, the other activity (shop-
ping). The private operator of both activities internalises this externality
and thus charges lower prices. In scenario III (competition between loca-
tions) the prices are lower compared to scenario 2, as expected. The rea-
soning was explained above: the local producer at location i does not in-
ternalise the effect its pricing decision has on the demand level at location
-i. Note that the total commercial output is highest in the scenario account-
ing for synergies (scenario I), as we also found in section 15.2.
The point price elasticities of demand are reported in Table 15.3. The
rail elasticities in scenarios I, II and III correspond to values encountered
in the literature, so that we analyse the effects of pricing and output deci-
sions on a relevant segment of the demand curve.3
Table 15.3. Point price elasticities of demand

rail OD1 rail OD2 commercial D1 commercial D2
I -0.94 -0.94 -0.65 -0.65
II -1.05 -1.09 -1.13 -1.13
III -1.05 -1.09 -1.08 -1.08
IV -0.02 -0.04 -0.03 -0.03
Table 15.4 contains the profit and (local and total) welfare levels for the
4 different scenarios. Total profits are highest in scenario I, as was ex-
pected (the single operator of rail and commercial activities internalises all
pricing effects). Generally speaking, one would expect the prices of trans-
portation and commercial activities in scenarios II and III to be higher
compared to the prices in scenario I, because the cross-effect of the price
of one good on the demand for the other good (and vice versa) is ignored.
As a result, prices are relatively high and demand is relatively low in sce-
narios II and III, so that aggregate profits are relatively low (compared to
I). Commercial profits are, however, higher in scenarios II and III, com-
pared to I. Because the demand for the commercial good is, by assumption,
relatively sensitive to price changes (as indicated by the slope of the in-
verse demand function), demand is relatively low (compared to the de-
mand for rail trips) in scenarios II and III, and the equilibrium price will be
relatively high.
As a result, revenues are relatively low, but because demand is low, the
corresponding low cost level causes profits to be relatively high in scenar-
ios II and III. Because demand for rail services is less elastic, the reduction
in costs does not outweigh the reduction in revenues, so that rail profits
decrease compared to scenario I. Note that commercial profits are highest
in scenario III (395.84). In this scenario the (local) suppliers of the shop-
ping good set the prices relatively high compared to the other scenarios.
Profits are relatively high as a result. The relatively high commercial
prices in scenario III cause a slight decrease in rail demand and a slight in-
crease in rail fares. This results in a decrease in rail costs in scenario III, so
3
Oum and Waters II (2000) present elasticities for the (intracity) rail segment
of -0.15 (peak), -1.00 (off-peak) and -0.12 -1.80 (all day).
that rail profits are higher compared to scenario II. Although the differ-
ences are small, the results indicate that local (private) developers may
prefer scenario III (profits of the local shopping activity are maximised).
Table 15.4. Profits and welfare levels
rail commercial total D1 D2

I 1805.15 383.24 2188.39 3282.59 1676.04 1606.55
II 1749.80 395.77 2145.57 3010.06 1537.36 1472.70
III 1754.97 199.03 (D1) 2150.81 3022.34 1543.56 1478.78
196.81 (D2)
IV 0 0 0 4376.78 2247.19 2129.59
Welfare is (of course) maximised in scenario IV. Scenario I has the

highest overall welfare level of the other three scenarios because its de-
mand externalities are internalised. Because the shopping goods are of-
fered by a monopolist in scenario II, and by two competing local suppliers
in scenario III, prices are lower and outputs are higher in scenario III. As a
result, aggregate profits are higher, and due to the fact that rail trips and
shopping goods are complementary goods, rail profits are also higher. The
higher rail fares reduce consumer surplus in the rail markets, but the con-
sumer surplus in the markets for the commercial goods increases due to the
lower prices. Overall, welfare is higher in scenario III compared to sce-
nario II.
15.5 Conclusions
In this chapter the interaction between rail services and demand for com-
mercial goods at or near stations has been modelled. Section 15.2 dis-
cussed synergies between rail and commercial activities, and these were
modelled in section 15.3. In the consumers utility function both goods ap-
pear, so that the demand for trips depends on the consumption of the
commercial good. Although the model is, of course, a simplification of re-
ality, the results offer relevant policy insights. The fact that joint opera-
tions of the rail and commercial activities results in higher profits is not
surprising, because the pricing effect (the impact of the rail price on com-
mercial profits, and vice versa) is internalised. But the simulation results
show that it is possible that profits of the commercial activity are highest
when it is operated separate from the other activity and is offered by spa-
tially-differentiated suppliers. This leads to some interesting policy consid-
erations. From a national perspective, scenario I (joint operations of rail
and commercial activities) is preferable, because welfare (and profits) is

(are) higher compared to separate operations. Joint exploitation of rail and
commercial activities is also preferable from a local perspective when
looking at the local welfare levels, although the (potential) operator of the
commercial activities would prefer a separate exploitation. In the decision
on separate or joint operations, the government must therefore weigh the
benefits of many anonymous consumers (increase in consumer surplus)
against the benefit of the (single) operator or developer of commercial ac-
tivities (profits). It remains to be seen whether private developers can be
attracted in scenario I. If not, they may prefer motorway projects as men-
tioned in section 15.2.
In a more realistic model the railway company would operate under in-
creasing returns to scale, which would reinforce the findings described
above. Furthermore, the rail network would be more extensive, while the
commercial activities might be exploited by different agents at different
stations. The main conclusions, however, will be similar. Given the com-
plementarity of the products, joint exploitation increases welfare. Further-
more, competition between location, for instance the independent devel-
opment of stations and surrounding areas, will result in welfare losses.
References
Arnott R, Yan A (2000) The Two Mode Problem: Second Best Pricing and Capac-
ity. Review of Urban and Regional Development Studies 12:170-199
Berg L van den, Pol P (1999) The High-Speed Train Station and Urban Revitaliza-
tion. European Institute for Comparative Urban Research, Erasmus Univer-
sity, Rotterdam
Bertolini L (1998) Station Area Redevelopment in Five European Countries: An
International Perspective on a Complex Planning Challenge. International
Planning Studies 3:163184
view 71:455475
Bertolini L, Spit T (1998) Cities on Rails: The Development of Railway Station
Areas. E&FN Spon, London/New York
Bruil I, Hobma F, Peek GJ, Wigmans G (eds) (2004) Integrale Gebiedsontwikke-
ling: Het Stationsgebied s-Hertogenbosch. SUN, Amsterdam
Bruinsma F, Rietveld P (1998) The Accessibility of European Cities: Theoretical
Framework and Comparison of Approaches. Environment and Planning A
30:499521
Claessens MT, Dijk NM van, Zwaneveld PJ (1998) Cost Optimal Allocation of
Rail Passenger Lines. European Journal of Operational Research 110:474-489
Economides N, Salop SC (1992) Competition and Integration among Comple-

ments and Network Market Structure. Journal of Industrial Economics
15:105123
Hagen M van (2002) Belevingswaarde Station. NS Productmanagement, Marke-
ting, Onderzoek en Advies, Utrecht
Hall P (1988) Cities of Tomorrow: An Intellectual History of Urban Planning and
Design in the Twentieth Century. Blackwell, Oxford
Harman R (1993) Railway Privatization: Does It Bring New Opportunities? Public
Money & Management 13:1925
Haynes KE (1997) Labor Markets and Regional Transportation Improvements:
The Case of High-Speed Trains. The Annals of Regional Science 31:5776
Technological Mobilities and the Urban Condition. Routledge, London/New
York
Juchelka R (2002) Bahnhof und Bahnhofsumfeld: Ein Standortkomplex im Wan-
del. Standort Zeitschrift fr Angewandte Geographie 26:1216
NS [Nederlandse Spoorwegen] (2006) In Perspectief: NS Jaarverslag 2005.
Utrecht
Nes R van (2002) Design of Multimodal Transport Networks: A Hierarchical Ap-
proach. Ph.D. Thesis, Delft University Press
Oum TH, Waters II WG (2000) Transport Demand Elasticities. In: Hensher DA,
Button KJ (eds) Handbook of Transport Modelling. Elsevier Science, Kidling-
ton
Peek GJ, Hagen M van (2001) Synergie op Stationslocaties: Investeren in Kwali-
teit Geeft Meerwaarde. Stedebouw en Ruimtelijke Ordening 4:4853
Speed Train Stations. Delft University Press, Delft.
Richards J, MacKenzie JM (1986) The Railway Station: A Social History. Oxford
University Press, Oxford
Roson R, Bergh JCJM van den (2000) Network Markets and Structure of Net-
works. Annals of Regional Science 34:197212
Spaans M (2002) The Implementation of Urban Revitalization Projects. Delft Uni-
versity Press, Delft
Trip JJ (2005) Railway Station Development in Post-Industrial Rotterdam: Path
Dependency and Shifting Priorities. Paper presented at the 45th Congress of
the European Regional Science Association: Land Use and Water Manage-
ment in a Sustainable Network Society, 23-27 August, Amsterdam
Vaessens B (2004) Synergy op stationslocaties. Colloquium Vervoersplanlogisch
Speurwerk, Antwerpen, 24-25 November
Vickerman R (1996) High-Speed Rail in Europe: Experience and Issues for Future
Development. The Annals of Regional Science 31:2138
Part D: Light rail and urban dynamics
16 Rail system development and urban
transformations: Towards a spatial
decision support system
Enrica Papa
DiPiST, Department of Urban and Regional Planning, University of

Naples Federico II, Italy
Francesca Pagliara
Department of Transportation Engineering, University of Naples

Federico II, Italy
Luca Bertolini

16.1 Introduction
System analysis techniques have long been used to interpret the transport
system behaviour and the performance of its components. More recently,
increasing attention has been given to the relationships between the trans-
port system and the global system to which it belongs (Meyer and Miller
2001). The study of these relationships is central to the transportation sys-
tem analysis and to the definition of integrated transport/land use planning
strategies.
This chapter is based on the theoretical framework of transport/land use
interaction and provides responses to the following questions: what are
possible urban structure changes in response to a transit system develop-
ment? How can information on the interaction between transit network
evolution and station area transformations support the search for adequate
policies?
338 Enrica Papa, Francesca Pagliara and Luca Bertolini
The chapter is organised in two main parts. The first section defines the
theoretical literature framework on the transit and urban system links and
the second proposes an application in the Naples urban area, in which a
new rail transit network is being developed. In particular, spatial and eco-
nomic transit impacts on the urban system are measured with the support
of a GIS. Moreover, some applications of the node-place model (Bertolini
1999) are proposed with a view to a more detailed investigation of the
status of the different transit/land uses in 1991 and 2004, and of the way in
which each station area changes in response to improved network connec-
tivity conditions. Our application of a 2011 node-place model defines the
first steps towards a Spatial Decision Support System (SDSS) which is in-
tended to support management of the integrated transit/urban system trans-
formation process in accordance with Transit Oriented Development
(TOD) principles, and to maximise the return on transit investments.
16.2 Transit networks and urban transformations: An

overview
16.2.1 Transit and urban system interaction: Theories and

planning practices
Interaction phenomena between rail infrastructure and urban system have

been studied extensively using different approaches in both transport and
urban studies disciplines, with the aim of defining theories and analysis
methods applicable to transport/land use system behaviour (Nijkamp and
Blaas 1994; Burmeister and Joignaux1998; Waddell 2001).
However, it is not until recently that researchers have applied a plan-
ning perspective to this topic and defined common and integrated strate-
gies for the management of transport/land use transformations (Hall and
Marshall 2000; Banister 1995; 2002).
In fact, despite evident interrelations between the two systems, land use
and transportation planning have tended to be separate operations in prac-
tice. From a strictly transport definition, the transport system is an urban
sub-system whose components generate transport demand between origins
and destinations, and a supply system is conceived to satisfy this demand
(Cascetta 1998, 2001). Transportation planning has tended to be based on
standard future land use patterns that are usually derived from market pro-
jections rather than a land use plan. Transport systems therefore generally
Rail system development and urban transformations 339
tend to reinforce past development trends rather than respond to urban plan
directions (Giuliano 1999).
Meanwhile, land use planners have typically taken account of the trans-
port plan outside of the decision process, with only mere acceptance of the
proposed transport interventions rather than endeavour to coordinate the
transport plan with future land use (Kaiser et al. 1995). In fact, the urban
plan defines the densities and the location of the activities without consid-
ering the future impacts of these choices on the transport system.
Both transport and urban planning disciplines uncover the need for a
new holistic approach based on complex theory and supportive of a trans-
port/land use system integration. In practice, approaches and methods have
been developed to define strategies for a more coordinated and cooperative
planning process between urban and transport policies (Cervero 1998).
This form of coordination is intended to generate synergy between urban
and transport system interventions, and such an approach aims to balance
the use of land use policy as an input to transportation planning with the
notion that transportation is a determinant in land use projection and land
use planning (Kaiser et al. 1995). In particular, the combined and inte-
grated set of strategies involving transit infrastructure investments, urban
development along transit lines, and integrated policies have been analysed
by many Transit Oriented Development (TOD) and Transit Joint Devel-
opment (TJD) experts (Dunphy et al. 2005; Cervero 2004; Dittmar and Oh-
land 2004). The practices have two main goals: the short-term goal is to
maximise the return on transport investments by increasing residential and
job density, urban quality, and a functional mix along transit corridors; and
the long-term goal is to increase the use of the transit system and control
urban sprawl.
16.2.2 Spatial and economic rail transit impacts on urban

systems: Results from empirical studies
Empirical studies on transit and land use system interactions have focussed
mainly on measuring and interpreting transit impacts on urban features, in-
cluding an analysis of different impact types and the use of various inter-
pretative methods (Rietveld 1994; Debrezion et al. 2004a, 2004b; RICS
2002; Van de Walle et al. 2004; Vessalli 1996). These impacts can be clas-
sified into spatial, economic, social, and environmental, but most of the
empirical studies focus on spatial and economic effects. Most studies have
been carried out in the US, where data quality and availability allows for
the application of more sophisticated analysis methods, such as correlation
analysis or hedonic price models (Haider and Miller 2000). In Europe
quantitative methods have only been applied relatively recently (Transplus

2002; Transecon 2003; Sesame 1998). Some of the empirical study results
are contradictory; this may be due to the different analytical techniques,
data quality and regional differences.
Rail transit system spatial impacts involve urban physical transforma-
tions such as land development or renewal interventions in the urban sys-
tem (Newmann and Kenworthy 1999; Cervero 1997) and activity pattern
changes, such as resident and jobs reallocation (Landis et al. 1995; Kim et
al. 2004; Transecon 2003). Most of the studies reviewed found some level
of land use change resulting from transit improvements and an activity
clustering effect close to urban rail stations. In the case of Madrid (Transe-
con 2003) a new transit line construction was accompanied by the shift of
residents from the city centre to suburban areas.
On the other hand, property value studies show higher impact intensities
than the land use impact studies, but the results are even more variable.
The economic transformations related to rail system evolution consist of
microeconomic impacts, such as property and rent value changes for dif-
ferent land uses, and macroeconomic effects such as urban economic com-
petition variation (Banister and Berechman 2000), potential development
increase, or economic viability of the central business districts (Arrington
1995; Parson 2001; Berechman and Paswell 1983; APTA 2002). A number
of empirical studies show an increase in property values in the new station
areas which is higher than the municipality average change value, as dem-
onstrated in the Vienna and Athens study cases (Transecon 2003; Golias
2002; Roider and Klementschitz 2002).
Many factors contribute to these urban transformations, such as the re-
gional economic trend or physical characteristics of the properties (Cam-
bridge Systematics 1999). In the available literature different methods
have been developed to analyse these factors and, in particular, hedonic
pricing is widely applied to define the property value change as a function
of the physical or functional characteristics of different properties (Cervero
and Duncan 2002; Bowes and Ihlanfeldt 2000). The extent of the impacts
varies from study to study and results often have to be interpreted cau-
tiously; they are generally small and indirect and depend on the presence
of several complementary factors, such as vacant land near new transit sta-
tions, a positive regional economic trend, good physical quality of the ur-
ban texture, and a central location of the new stations. However, only a
small number of empirical studies include an interpretation of the results
aimed to define planning strategies for the specific urban context where
they are applied, despite the common assumption that impact intensity also
depends on the presence of proactive urban planning policies in the transit
corridors and transit station areas.
Accordingly, one of our main objectives in this study is to describe im-

pacts so that they can support the identification of strategies and integrated
interventions. This goal is achieved in two different steps, both of which
are applied to the Naples case. The first more traditional step consists of
the quantitative measurement and interpretation of the impacts of a new
metro line on the urban system (population and property value changes). In
the second less orthodox step, the study proposes a node-place model
(Bertolini 1999) application for analysing the correlation between the con-
nectivity increase and the impacts intensity in the urban system. By so do-
ing it becomes possible to highlight the potential for transformation in each
station area. This information could be used to identify intervention strate-
gies and define priorities for a transit land use transformation management,
as discussed in the conclusions.
16.3 An application example: The Naples case
Naples is the largest city in southern Italy and comprises about one million
residents in an area of 117 sq km; it is part of a conurbation (Province of
Naples) of about three million residents (Istat 2001). The rail transit sys-
tem has been expanding since 1993, with the addition of new lines and
new stations, as shown in Figures 16.1 and 16.2. The rail network had con-
sisted in 1991 of five transit lines and 45 stations, but by 2004, the rail sys-
tem consisted of six lines, 69 stations and seven interchange nodes. The
main change involved the construction of a new transit line (Line 1) wind-
ing from the city centre to the northern periphery. The first six stations on
the new line were opened in July 1993, with a further three stations com-
pleted in 1995. In one year, from July 2001 to July 2002, four new stations
began operating: the line now has 14 stations and is connected with the ex-
isting rail network with two interchange nodes (Vanvitelli and Museo sta-
tions).
The station areas transformation assessment was carried out over a 13-
year period (from 1991 to 2004) using longitudinal data that allows for be-
fore and after comparisons of property values and residents in the rail tran-
sit station areas. In addition, GIS analysis techniques allow us to represent,
organise and query the time series spatial database. Geo-referred time se-
ries data has been related to Naples census tracts as well as to each transit
node in the network. This has enabled us to make geographic correlations
between the urban transformation indicators (residents change and prop-
erty values change) and the accessibility indicators change on the entire
rail network. Furthermore, with the GIS support, station influence areas
Figure 16.1. The Naples rail transit network in 1991
Figure 16.2. The Naples rail transit network in 2004

have been defined as the union of the census tracts within 500 metres from
the station exits. This measure is used widely as the average walking dis-
tance to reach a station (Landis et al. 1995).
The next section reports the main results and focusses on the Line 1 sta-
tion area transformations. Figures 16.3 to 16.6 illustrate the transformation
intensity in the station areas, in accordance with their location in the urban
structure and their connectivity in the transit network.
16.3.1 Spatial and economic changes in the station areas from

1991 to 2004
The main data sources used for the measurement of spatial and economic
transformations in the Naples urban area are the ISTAT National Institute
of Statistic data and the Agenzia del Territorio data concerning property
values for different property types (high quality house, low quality house,
single family house, parking spaces, retail, and offices). Table 16.1 shows
resident and property value changes within Line 1 station areas, and other
features that could have influenced the spatial and economic transforma-
tion as the station area location (central, semi central, suburban). The sta-
tion opening year is also shown. Average values for the whole of the mu-
nicipality are included for comparison. Table 16.1 illustrates the values of
the following indicators measured in Line 1 station areas:
1. Network index: is the normalised value of the measure of the net-
work connectivity of the station area i in 2004, defined as an adap-
tation of the Shimbel connectivity index (Pumain and Saint-Julien
2004), as defined in formula (1), where t ij is the network access
time1 from station area i to station area j, and n is the total number
of station areas:
n
t
j =1
ij
indconni = (1)
t
ij ij
2. built area index: is a measure of vacant land availability nearby
the station area and quantifies the concentration of built area, as
defined in formula (2), where Sbuilt k is the building covered area
1 The network times have been calculated with Distance/Travel time Calculator
software written by Dr Evert Verkuijlen of the GIS-Centre of the Department of
Geography and Planning at the University of Amsterdam.
in the census tract k; S k is the extension of census tract k; p is the

number of census tracts belonging to station area i:
p Sbuilt k
built i = (2)
k =1 Sk
3. Station use index: is a measure of the use of each station in 2004

as defined in formula (3), where U i is the number of users in sta-
tion i, and U tot is the total number of users of Line 1:
Ui
Stusei = (3)
U tot
4. Functional mix index: is the normalised value of the indicator
which measures the presence of different economic activities (re-
tail, offices, etc.) in station area i in 2004, as defined in formula
(4), where jobsia is the number of jobs for the economic activity
a in station area i; max jobsa is the maximum number of jobs for
the economic activity a; min jobsa is the minimum number of jobs
for the economic activity a; c a is the number of economic activi-
ties in the station area i; m is the total number of economic activity
categories:
m
jobsia
mixfunz i = ca (4)
a =1 max jobsa min jobsa
5. Resident change: is measured as average number of residents

change in census tracts k that belong to station area i, where
res 2004 k are residents in census tract k in 2004 and res1991k are
residents in census tract k in 1991:
p
res 2004 k res1991k
%resi = 100 (5)
k =1 res1991k
6. Property value change (for different types of property t): is com-
puted as the average change of the values in census tracts k that
belong to station area i and is measured in /sqm as defined in
formula (6), where value2004 kt is the value of the property in 2004
in census tract k for property type t:
p
value2004 kt value1991kt
%valueit = 100 (6)
k =1 value1991k
Some interesting conclusions can be drawn. First, a general resident

density decrease is shown along the metro line (as in the entire Naples mu-
nicipality), but there is also a population decentralisation trend. As shown
in Table 16.1, each area has evolved over time in relation to the specific
urban, historical and functional context. The new transit system seems to
have supported the central area transformation by decreasing residential
density in the central districts. In fact, the correlation between the network
index and the population decrease is strong. The analysis shows a higher
decrease in residents in new central station areas (e.g. Cilea 10%,
Medaglie dOro 11%, Vanvitelli 9,6%). This phenomenon is also con-
nected to the considerable above average increase in property values in
these areas. However, in the suburban station areas the reverse is evident; a
population increase has accompanied the new transit line construction (e.g.
Piscinola + 3,9%, Chiaiano + 1%, Frullone + 5,3% ). These trends corre-
spond to those reported in other studies (Transecon 2003) as in the Madrid
study case, where a new transit line construction contributed to a shift of
residents from city centre to peripheral areas.
Figure 16.3. Resident change in the rail station areas 1991-2004

Table 16.1. Spatial and economic transformations in the Naples L1 station areas:
1991-2004
Built Station
Station Network Functional
area use Residents
Station areas* opening Location index mix index
index index 2004
year 2004 2004
2004 2004
Piscinola July 1995 suburban 46,52 0,20 8,30 0,69 9759
Chiaiano July 1995 suburban 54,64 0,22 14,40 2,25 5166
Frullone July 1995 suburban 61,72 0,15 3,40 0,00 2876
semi-
Colli Aminei July 1993 71,46 0,18 5,80 4,73 4058
central
semi-
Policlinico July 1993 77,54 0,21 5,80 4,13 742
central
semi-
Rione Alto July 1993 81,05 0,36 9,70 38,88 19365
central
semi-
Montedonzelli July 1993 91,41 0,30 3,70 14,54 11281
central
Medaglie dOro July 1993 central 104,24 0,35 17,00 64,87 19592
Vanvitelli July 1993 central 116,43 0,49 26,00 65,95 9577
Cilea April 2001central 100,03 0,40 1,60 65,64 15594
Salvator Rosa April 2001 central 94,91 0,35 6,00 14,73 8768
Materdei July 2003 central 94,29 0,48 - 22,34 14471
Museo April 2001 central 106,10 0,58 3,80 46,45 8279
Dante April 2002central 96,46 0,62 - 100,00 7789
Average L1 85,49 0,35 8,30 31,80 9808
Municipality
- - - - -
average
Bold figures in the column represent higher L1 average in each group
Source: ISTAT and Agenzia del Territorio
The analysis shows that, with regard to economic transformation, prop-

erty values for different property types in station areas increased faster
than the urban municipality average value, particularly where offices are
concerned. In new and central subway stations, the prices also increase
faster and with a higher intensity due to other urban renewal interventions
in these areas: examples are house property value changes in such stations
as Dante (+41%), Materdei (+45%), and Salvator Rosa (+47%). In the
suburban and semi-central station areas, property values increase more
slowly; this may explain the resident decentralisation phenomenon in the
Naples area. Retail property value changes have a lower, both absolute and
relative, increase when compared to other uses. This can be explained by
the competition for retail spaces on the property market. As shown in GIS
layouts (Figures 16.4, 16.5 and 16.6), spatial and economic transit impacts
Table 16.1. Spatial and economic transformations in the Naples L1 station areas:
1991-2004 (continued)
%
% % sin-
good % % %
low qual- gle fam-
quality parking retail office
% ity house ily house
house space propertyproperty
residents property property
Station areas* property value value value
1991- value value
value 1991- 1991- 1991-
2004 1991- 1991-
1991- 2004 2004 2004
2004 2004
2004 (/sqm) (/sqm) (/sqm)
(/sqm) (/sqm)
(/sqm)
Piscinola 3,90 34,40 11,52 20,28 36,83 6,27 22,27
Chiaiano 0,90 33,60 16,12 28,32 32,36 6,42 20,55
Frullone 5,30 34,53 16,45 29,29 33,61 6,40 22,56
Colli Aminei -8.04 30,84 15,26 40,32 26,49 7,26 22,84
Policlinico -7,00 32,32 17,86 47,49 55,76 9,49 34,16
Rione Alto -11,70 30,84 35,26 40,32 26,49 12,26 23,84
Montedonzelli -9,30 35,00 33,43 21,32 32,69 13,85 38,02
Medaglie
-11,10 44,14 42,63 21,06 33,67 24,47 41,64
dOro
Vanvitelli -9,60 43,13 46,62 36,33 35,54 28,79 41,20
Cilea -10,00 46,23 42,08 22,46 33,84 24,87 41,60
Salvator Rosa -12,20 47,67 45,44 29,16 34,66 26,76 41,40
Materdei -6,70 45,41 40,94 21,90 32,03 25,99 43.0
Museo -8,30 45,27 49,40 61,34 50,09 24,59 55,18
Dante -4,10 41,62 42,84 38,50 27,14 27,68 54,08
Average L1 -6,10 38,93 32,56 32,72 36,51 17,50 37,88
Municipality
-8,25 33,68 18,01 29,28 30,43 14,82 23,42
average
Bold figures in the column represent higher L1 average in each group
Source: ISTAT and Agenzia del Territorio
are not uniform and occur with stronger intensity only where other eco-
nomic conditions are already beneficial to these increases. For example,
the GIS layout clearly shows the impact on new urban transformation in
the increasing property values in the Bagnoli area in the west periphery,
where a very large unused industrial area is being transformed into an ur-
ban park.
Finally, the combined results of resident and property value changes in
the new station areas draw attention to the structuring effect of the transit
system construction on urban transformation. In fact, the average increase
of the new Line 1 property values for different uses (both residential and
non-residential) is higher compared to the municipality average change.
Figure 16.4. Property value change of low quality housing in the station areas
1991-2004
Figure 16.5. Retail property value change 1991-2004

Figure 16.6. Offices property value change 1991-2004
16.3.2 Node-place model applications for the Naples area
The time-series and longitudinal data in the previous section, supported by

GIS and descriptive statistical methods, serves to underline particular fea-
tures of the urban transformations connected with the transit network de-
velopment. However, they represent the interaction of transit and urban
development too indirectly to be used in a discussion of the potential need
and possible scope of policy intervention. In order to improve on this, the
following section proposes an elaboration of the node-place model (Berto-
lini 1999). This model provides an analytical framework for understanding
and measuring the relationships between two aspects of the railway station.
Each subway station can be considered both as a node of the rail transport
network and as a location in the urban system (Bertolini and Spit 1998). In
order to study these different features of the station, the node-place model
allows each station to be represented in an xy diagram. One axis value cor-
responds to a node index, which is a measure of the network connectivity
of the station. The other axis value corresponds to a place index, which is a
measure of the station area urban features. The relative position of each
station area in the diagram provides information on the quality of the rela-
tionship between the node and the place aspect in the station. Furthermore,
the distribution of the scatterplot in the diagram can provide insights into
how the transit network structures the urban system into the station area
hierarchy in the network, and information on the extent to which the urban
system is connected with the rail system.
The model in this study is applied to analyse the station area transforma-
tions between 1991 and 2004. The use of the node-place model illustrates
how each station has evolved over time and how the urban system has
transformed following connectivity improvements. Existing GIS layouts
show how the station areas have changed spatially in the time period, but
the correlation between the urban feature change and the transit network
connectivity improvements has not yet been made evident. The node-place
model instead shows the transport provision change and the related trans-
formation intensity in each station area. In this way it is possible to under-
stand how the improved connectivity condition has or has not created con-
ditions favourable to urban transformation, and the direction each station
area has taken after realising its potential for transformation.
Figure 16.7 illustrates the scatterplot of the station areas in the xy dia-
grams. The y value is the normalised value of the network index change as
defined in (1). The x value of the diagram on the left is the normalised
value of the resident change in each station area, and the x value on the
right of the diagram is the normalised value of the house property value
change. The comparison of the two diagrams provides interesting informa-
tion. First, all central station areas of the new Line 1 (points inside the con-
tinued eclipses) that are characterised by a high increase of the network in-
dex, were subject to a simultaneous negative change in the number of
residents (from 11% to 4%) and a positive change in house property
values (from +41% to +44%), which is higher than the municipality aver-
age value (+34%). Second, the three new suburban stations (points inside
the dotted circle) that experienced a medium increase in the network index,
are characterised by a small increase in the resident number (from +1% to
+5%), and an increase in the property value change (from +30% to +35%),
which is very close to the municipality average. From a systems perspec-
tive, the analysis shows a resident decentralisation process associated to
the transit network evolution that may coincidentally imply the transforma-
tion of central urban areas with higher concentrations of services and
commercial activity. The strong above average increase in retail and office
property values in the high connectivity stations documented in Table 16.1
also points in this direction.
These diagrams, which can also be implemented using other indicators,
provide a communicative and powerful management support system for
the definition of land use/transport intervention in new and existing sta-
tions areas. The diagrams might show that some station area development
paths did not correspond to land use transport planning goals, and thus in-
dicate the need for intervention. For instance, in the CBD stations areas
where the level of connection has been increased, some finalised policies
might be required to avoid reductions in the number of residents in order to
preserve the functional mix, and thus the urban quality of the places.
Figure 16.7. Population, property value and network index change in the station
areas from 1991 to 2004
Other applications of the node-place model are proposed in the dia-

grams in Figures 16.8, 16.9 and 16.10. Figures 16.8 and 16.9 show the re-
lationship between network connectivity, number of residents and house
property values in 1991 and 2004. The comparison of the diagrams is use-
ful for studying the transit land use transformations at system level, since
the station areas scatterplot, the position and inclination of correlation
lines and the correlation indices in the two charts represent the urban sys-
tem state in 1991 and 2004.
In Figure 16.8, where the x axis represents the number of residents in
each station area, the 1991 scatterplot shows a diffused lack of transit sup-
ply and the absence of a structuring transit network for the urban system.
The increase in transport supply improvements and the decrease in number
of residents cause the correlation line in the 2004 scatterplot to rotate anti-
clockwise, thereby indicating an increase in the hierarchy level in the net-
work. Moreover, the higher value of the correlation index in 2004 is a sign
of an increasing structuring effect of the transit network on the urban struc-
ture.
Figure 16.8. Network index and population in 1991 and 2004
Figure 16.9 shows the evolution of the relationship between the network
index and house property values. On the basis of these indexes, the corre-
lation line is more steeply inclined in the 2004 diagram. This is an indica-
tion that the transit network has influenced the property values market and
caused an increase in property value differences among different areas in
the urban system. The same evolution has also been confirmed by node-
place model applications with property values for other uses (shops, of-
fices, etc.)
Figure 16.9. Network index and house property values in 1991 and 2004
In the future transit/land use system scenario, Figure 16.10 shows the
diagrams where the y value is the network index in the 2011 scenario
(when the transit network will have 113 stations, according to the Naples
Transport Plan) and the x value is the number of residents in 2004. The
diagram shows the high network index increase in the 2011 scenario that,
according to the past development path of the transit land use system,
could lead to a new state characterised by the position of the station areas
in the upper left-hand section of the diagram. This would mean a large
number of unbalanced station areas, since the points inside the dotted
eclipses will experience a high increase in network connectivity, but are
not used intensively in the present scenario. In these areas the increase of
the network index creates a transformation potential that can be set against
planning goals. This implies that some proactive polices according to TOD
principles (for instance, promoting residential use for central station areas)
need to be developed for these station areas.
Figure 16.10. Network index in the 2011 scenario and resident numbers in 2004
We have to underline that the node-place diagrams are not finalised to

the extent that they can predict the system evolution. This is because net-
work index increase from 2004 to 2011 is not comparable to the 1991 to
2004 network index change. Furthermore, the transport/land use tendency
towards a more balanced and transit-oriented dynamic equilibrium will
depend on the general economic trend and the presence of transit-oriented
planning strategies. However, the diagrams do provide key information for
urban system transformation governance, the definition of transformation
opportunities, and the risks associated with the connectivity increase for
each station area. The network index value for each station area is one
transformation condition that can be specifically tailored to achieve plan-
ning objectives.
16.4 Conclusions
This study has proposed a framework for the analysis of the transport/land
use interaction, with particular attention given to the transit role in the ur-
ban transformation process. Using a GIS impact assessment application,
we have presented the Naples study case; evidence pertaining to the tran-
sit/land use system has been verified in a quantitative manner with particu-
lar attention given to the new Line 1 station areas. Finally, the study has
proposed a number of applications of the node-place model (Bertolini
1999) in 1991, 2004 and 2011 to highlight impact analysis elements and to
provide a management support system for the integrated transit/land use
intervention and strategies development.
The impact assessment indicates that central station areas of the new
Line 1 have experienced both a lower population decrease and a concur-
rent higher increase in property values for all land uses compared to the
municipality value. This is particularly the case for central, high connec-
tivity stations. Suburban stations are differently characterised by an in-
crease in number of residents and an increase in property values which are
close to the urban system average value (for houses and offices) and lower
than the municipality value for retail. Results show an overall resident de-
centralisation with a shift of residents from city centre to suburban areas.
The average decrease in number of residents in new station areas is in fact
lower than the municipality average value. Second, all sorts of property
along the line have experienced an average increase in value, which is
markedly higher than the city-wide average. The two effects together sug-
gest an increasing structuring effect of the transit line on urban transforma-
tion and (especially) on property values.
Applications of the node-place model have also shown the transport
provision change and the related transformation intensity in each station
area. In this way, we were able to understand how the improved connec-
tivity condition has or has not created conditions favourable to urban trans-
formation, and also determine how each station area has developed after
the realisation of its potential for transformation. The application of the
node-place model has illustrated the different states the urban system has
achieved in 1991 and 2004, and how each station area has responded to the
improved network connectivity conditions. Certain conclusions can be

drawn from the comparison of the 1991 and 2004 diagrams. First, the cor-
relation line shifting upwards in the 2004 scatterplot indicates a connec-
tivity increase in the urban system. Second, the anti-clockwise rotation of
the correlation line shows an increase in network hierarchy among the sta-
tion areas. Finally, the correlation index value increase in the 1991 to 2004
scatterplot indicates that, during the observation period, the transit network
had a structuring effect on the urban system. The analysis verifies and
demonstrates these phenomena using population and residential property
values. With regard to the 2011 node-place model application, the diagram
provides basic information for urban system transformation governance,
defining the transformation potential related to the connectivity increase
for each station area. This information could support a transit land use
planning process in existing and future stations in order to achieve a
greater transit-oriented urban system.
Given these findings, the research agenda aims on the one hand, to fur-
ther investigate land use changes in station areas, particularly by extending
the analysis to the number of job changes for different economic activities;
this is essential in order to interpret which factors might influence a devel-
opment path. Findings here may also assist researchers in understanding
land use pattern changes in each station area, and comparing general pat-
terns with focussed analyses of different station area case studies. On the
other hand, future research can focus on exploring how the information
provided could be used effectively to support a debate on the required pol-
icy interventions in station areas and transit corridors.
References
APTA (2002) Rail Transit and Property Values. http://www.apta.com, Information

Centre Briefing 1
Arrington GB (1995) Beyond the Field of Dreams: Light Rail and Growth in Port-
land. Tri-Met, Portland, Oregon
Banister D (1995) Transport and Urban Development. Chapman & Hall, London
Banister D (2002) Transport Planning 2nd edn. SPON Press, London
Banister D, Berechman J (2000) Transport Investment and Economic Develop-
ment. UCL Press, London
Berechman J, Paswell RE (1983) Rail Rapid Transit Investment and CBD Revi-
talization: Methodology and Results. Urban Studies 20:471486
Bertolini L (1999) Spatial Development Patterns and Public Transport: The Appli-
cation of an Analytical Model in the Netherlands. Planning Practice and Re-
search 14:199210
Bertolini L, Spit T (1998) Cities on Rail: The Redevelopment of Railway Station

Areas. E&FN Spon, London
Bowes DR, Ihlanfeldt KR (2000) Identifying the Impacts of Rail Transit Stations
on Residential Property Values. Journal of Urban Economics 50:125
Burmeister A, Joignaux G (eds) (1998) Infrastructures de Transport at Territoires
Approches de Quelques Grand Projets. LHarmattan, Paris
Cambridge Systematics Inc. (1999) Quantifying the Impacts of a Light Rail Tran-
sit Station on Adjacent Property Values. Unpublished manuscript
Cascetta E (1998) Teoria e Metodi dellIngegneria dei Sistemi di Trasporto.
UTET, Torino
Cascetta E (2001) Transportation Systems Engineering: Theory and Methods.
Kluwer Academic Publishers, Boston
Cervero R (1997) Light Rail Transit and Urban Development. Journal of Ameri-
can Planning Association 50: 133147
Cervero R (1998) The Transit Metropolis: A Global Inquiry. Island Press, Wash-
ington D.C., California
Cervero R (2004) Transit-Oriented Development in the United States: Experi-
ences, Challenges and Prospects. TCRP Report 102, Washington
Cervero R, Duncan M (2002) Benefits of Proximity to Rail on Housing Markets.
Journal of Public Transportation 5:118
Debrezion G, Pels E, Rietveld P (2004a) The Impact of Railway Stations on Resi-
dential and Commercial Property Value. Tinbergen Institute Discussion Paper
TI2004-023/3, Amsterdam
Debrezion G, Pels E, Rietveld P (2004b) The Effects of Railway Investments in a
Polycentric City. Tinbergen Institute Discussion Paper TI2004-089/3, Am-
sterdam
Dittmar H, Ohland G (eds) (2004) The New Transit Town: Best Practices in Tran-
sit-Oriented Development. Island Press, Washington
Dunphy R, Cervero R, Dock F, McAvey M, Porter D (2005) Developing Around
Transit: Strategies and Solutions That Work. Urban Land Institute, Washing-
ton
Giuliano G (1999) Land Use Impacts of Transportation Investments: Highway and
Transit. In: Hanson, S. (ed.) The Geography of Urban Transportation, 2nd
edn. Guilford Press, New York
Golias JC (2002) Analysis of Traffic Corridor Impacts from the Introduction of
the New Athens Metro System. Journal of Transport Geography 10:9197
Haider M, Miller EJ (2000) Effects of Transportation Infrastructure and Location
on Residential Land Values: Application of Spatial-Autoregressive Tech-
nique. Transportation Research Record 1722:18
Hall P, Marshall S (2000) Transport and Land Use Development. Independent
Transport Commission London, University College London, London Press
Istat (2001) 14 Censimento della Popolazione e della Abitazione. Instituto Nazi-
onale di Statistics, Rome
Kaiser EK, Godschalk DR, Chapin Jr S (1995) Urban Land Use Planning. Univer-
sity of Illinois Press, Urbana IL
Kim JH, Pagliara F, Preston J (2004) Transport Policy Impact on Residential Lo-
cation. International Review of Public Administration 9:7187
Landis J, Cervero R, Guhathukurta S, Loutzenheiser D, Zhang M (1995) Rail
Transit Investments, Real Estate Values, and Land Use Change: A Compara-
tive Analysis of Five California Rail Transit System. Monograph 48, Institute
of Urban and Regional Studies, University of California, Berkeley
Meyer MD, Miller EJ (2001) Urban Transportation Planning: A Decision-
Oriented Approach. 2nd edn. McGraw Hill, New York
Newman P, Kenworthy J (1999) Sustainability and Cities: Overcoming Automo-
bile Dependence. Island Press, Washington DC
Nijkamp P, Blaas E (1994) Impacts Assessment and Evaluation in Transportation
Planning. Kluwer Academic Publishers, Boston
Parson B (2001) The Effect of Rail Transit on Property Values: A Summary of
Studies. Research carried out for Project 21439S, Task 7 (Draft). NEORail II,
Ohio, Cleveland
Pumain D, Saint-Julien T (2004) Lanalyse spatiale. Armand Colin, Paris
RICS (2002) Transport Development Areas: A Study into Achieving Higher Den-
sity Development around Public Transport Nodes. The Royal Institution of
Chartered Surveyors, London
Rietveld P (1994) Spatial Economic Impacts of Transport Infrastructure Supply.
Transportation Research A 28:329341
Roider O, Klementschitz R (2002) The Socio-Economic Effects of the Metro Line
U3 in Vienna. Institute for Transport Studies, University of Bodenkultur
Sesame Project (1998) Urban Form and Mobility D4: Report on Analysis of Rela-
tionships. Mimeo
Transecon Consortium (2003) Urban Transport and Local Socio-Economic De-
velopment. http://www.transecon.org, Deliverables 4 and 5
Transplus Project (2002) Land Use and Transportation Planning: Experiences in
European Cities. TNO Report 2002-67, Delft
Vessalli KV (1996) Land Use Impacts of Rapid Transit: A Review of Empirical
Studies. Berkeley Planning Journal 11:71105
Waddel P (2001) Review of the Literature and Operational Models. Systematics
Inc., Cambridge
Walle S van de, Steenberghen T, Paulley N, Pedler A, Martens M (2004) The Role
of Indicators in the Assessment of Integrated Land Use and Transport Policies
in European Cities. International Planning Studies 9:173196
17 Rail-transit and real estate values in a
polycentric city: A theoretic simulation
approach
Daniel Gat
Center for Urban and Regional Studies, Technion Israel Institute of

Technology, Haifa, Israel
17.1 Introduction
The purpose of this chapter is to provide a microeconomic framework for

tracing the impact of urban rail transit on the residential real estate market.
It presents an urban polycentric model that is quite simple, yet it is com-
plex enough to handle automobile traffic congestion as well as transit and
park-and-ride. The model is derived from a long line of microeconomic
monocentric urban models. The model is estimable, but not empirically es-
timated here; instead, it is run as a spreadsheet simulation with artificial
parameter values. Results are computed and shown graphically and geo-
graphically; they support the claim that transit introduction is likely to
raise real estate values, but are conditional with regard to instigating inner
city redevelopment.
17.2 Motivation
The debate on whether or not to introduce rail transit into transit-less or

transit-poor cities has become a hot topic. Those in favour of rail praise the
civilising effect of the transit-friendly city and its accompanying pedestrian
friendliness, and they point out the high amenity values of European cities
that rely highly on rail transit (e.g. Bernick and Cervero 1997; Cervero
1998; Newman and Kenworthy 1999). More recently, arguments in favour
of transit state that, in computing costs and benefits thus far, important en-
360 Daniel Gat
vironmental and health effects have been ignored. (Frank et al. 2003;
Frumkin et al. 2004).
Furthermore, on the benefit side, city-shaping influences and real estate
added values have been claimed by many and revealed by some, according
to Ryan (1999). Ryan concludes that property values are expected to rise
when and where transit endows commuters with significant time savings
over the car. A year after Ryans paper, Scheurer et al. (2001) claim that
property added-values caused by transit are significant and should be cap-
tured to help pay for transit.
In contrast, those who oppose new transit investments praise the advan-
tages of the spread city and of cheap automobile transportation (e.g. Web-
ber 1963, 1995, 1998); they stress the high costs of transit investment and
of its up-keep. They also claim that recent examples have shown that rail
transit adds insignificantly to existing accessibility in, for example, US cit-
ies. (Webber 1995; Mills 1999). The debate carries on and for good reason.
Essentially it poses the question: will urban life be better if we let the car
completely dominate our cities, or should we switch paradigms and insist
on multi-modal cities at great economic and political cost? Research that
addresses this dilemma has two alternative directions. One is empiric: to
conduct field studies of cities with transit and isolate its benefits and costs.
Bowes and Ihlanfeldt (2001) have done so with regard to the MARTA sys-
tem in Atlanta, applying a hedonic analysis of housing values in the vicin-
ity of stations. As a bonus they review a large collection of other studies
that take an empiric approach.
This chapter takes the second approach. Its purpose is to enhance theo-
retic discussion regarding the real estate market response to the introduc-
tion of transit facilities. Does urban-economic theory support the claim of
significant real estate added-value? Does it predict that transit would pre-
cipitate new construction? To this end the chapter extends a well-known
version of the standard monocentric (single-modal) residential model,
turning it polycentric and duo-modal.
Accordingly, the task is to produce a model able to simulate the real-
estate market impact of introducing un-congested transit into a previously
single mode vehicle-congested polycentric city. The proposed model is
built up gradually as a sequence of mutations starting with the well-known
standard monocentric model. Descending from Alonso (1964), the stan-
dard monocentric model is an appropriate point of departure, since it de-
rives the structure of a city as a relationship between the centre, a place of
work and shopping, and the home, mediated by the transportation system.
The result is a set of gradient curves, downward-sloping functions of dis-
tance, and/or of travel time, which describe housing rent, land value, floor-
area ratio (FAR), and population density. A change in the money or time
Rail-transit and real estate values in a polycentric city 361
cost of transport alters these gradients and thus, potentially, the physical
structure of the city.
In order to be sensitive to the introduction of a transit system, the stan-
dard monocentric urban model needs to be extended in more than one way.
The following is a list of the special features the new model must possess.
Items 1 and 2 rely heavily on the prior work of others. Items 3, 4 and 5 are
my own contributions, although they cope with well-known urban issues
and use familiar economic tools.
Floor space production function. Locations that are more desirable elicit
higher rents and as a result lead real estate entrepreneurs to build at higher
FAR. Although some early pioneering studies have incorporated this char-
acteristic (e.g. Muth 1969; Mills 1972; Richardson 1983; Bertaud and
Bruekner 2003; and Hatta and Okawara 1993), other even very recent pa-
pers ignore this trade-off between land and non-land inputs within a pro-
duction function. But this feature is a must for debating transportation and
real-estate relationships. I am therefore using the Bertaud and Bruekner
2003 model described in their appendix as the cornerstone for the proposed
model.
Road congestion is treated by some urban economic model builders (e.g.
Wheaton 1998; Anas et al. 2000). But congestion is ignored by many other
authors since it introduces serious non-linearity and does not lead to a
closed form solution. Anas et al. have proposed a simplified version which
I further simplify below.
Polycentric City. By definition, a monocentric model is not designed to
represent a polycentric city. But polycentricity is essential to the modelling
of transit, since sub-centres are the most appropriate locations for transit
stops. In such a model there is a need to include a mechanism for selecting
between destinations an element that is absent from the single centre
model.
Multi-Modality. The standard monocentric model is mono-modal. A
model of a city with more than one travel mode needs to include a choice
mechanism for selecting between modes.
Demolish and Redevelop. The standard model depicts the city, at least
implicitly, as a soft and malleable entity, changing its form constantly in
response to changing market and demographic conditions. A real city is
hard and brittle. One has to destroy a building if, in its place, a newer, lar-
ger one is preferred. The extended model needs to have a mechanism for
making the keep or demolish decision.
The balance of this chapter follows the sequential construction of a
model that incorporates all the aforementioned essential properties; it then
reports the results of numeric runs that simulate the introduction of urban
rail transit.
362 Daniel Gat
17.3 The monocentric single mode standard model
The city is open and inhabited by identical households that earn the same
income y and have identical preferences u. The household maximises its
welfare function of home size S and income Z net of rent R and transport
costs T:
u ( S , Z ) = S Z 1 (1)
Optimal allocation yields housing expenses that are a proportion of

income net of the transportation cost T(x), which is imposed once x the
home location is selected.
S ( x, y ) R ( x , y ) = [ y T ( x ) ]
(2)
Z ( x, y ) = (1 ) [ y T ( x) ]
Optimal S and Z are inserted into (1) to yield:
u * ( S , Z ) = (1 )1
[ y T ( x) ] (3)
R ( x, y )
City-wide equilibrium is reached when all households, regardless of lo-

cation have the same level of welfare. This enables the display of bid rent
R(x, y) at any location x as a function of rent at the CBD where no travel is
needed.
1
T ( x)
R ( x, y ) = R(0, y ) 1
y
(4)

Assume from here on (without serious loss of generality) that the only
form of transportation cost is the loss of income because time travelled is
not spent at work while earning. Therefore an income change would leave
the square bracketed term unchanged. But we know through observation
that households with higher incomes live in more spacious homes.
Added household space is modelled in the following way: we stay with
the result that total housing expenditures are a fixed proportion of income.
However, when income grows, part of the added expenditure is offered as
a higher bid rent and the rest is accounted for as a demand for more space.
Let S(0,yo) be a benchmark household demand for housing space at the

CBD when income is yo. For that, a household is willing to pay at the rate
of Ro = R(0,yo) $ /sqm-month. The changed demand for space at the CBD,
given a new income y, is a power function of the ratio between the new
and old income:

y
S (0, y ) = S (0, y0 ) ; 0 < < 1 (5)
y0
The special case of equation (2) at the CBD yields:
S (0, y ) R(0, y ) = y
S (0, y0 ) R(0, y0 ) = y0 (6)
S (0, y0 ) y
R(0, y ) = R (0, y0 )
S (0, y ) y0
Now (5) is plugged into the last line of (6) and the result is:
1 1
y y y y
R (0, y ) = R (0, y0 ) 0 = R(0, y0 ) = R0 (7)
y y0 y0 y0
This result is inserted into (4) to give:
1
1
y T ( x)
R ( x, y ) = R0 1
y
(8)
y0
Equation (8) is the foundation upon which the rest of the modelling
structure is built. It will be seen below that much effort is devoted to
evaluate the value of travel time under several conditions which are: (a) a
polycentric city; (b) a city with vehicular mode congestion; and (c) a dual
mode city where the second mode represents park-and-ride. Walking, cy-
cling and buses are not included so as to not clutter the exposition. How-
ever, leaving them out implies no loss of generality.
364 Daniel Gat
17.3.1 Producing housing structures
I continue now to add the supply side of the housing real estate, still within
the monocentric model. The production function for built floor-space is a
Cobb-Douglas function of land L and money M inputs.
F = AL M 1 (9)
In (9) A is the productivity and is the weight of land in the produc-

tion process. Measured empiric values of are between 1/4 and 1/3. To
keep things simple the interim financing cost is left out. However, insert-
ing back the construction duration is easy; it entails maximising present
value rather than profit, the latter being the case below.
Let 1$ be the unit cost of M, the cost of 1 square-meter (sqm) of land,
and V the value of 1 sqm of finished floor-space. Then the developer,
modelled as a price taker, seeks to maximise profit:
max = FV ( L + M ) = AL M 1 V ( L + M ) (10)
In a competitive market the maximum profit is zero, and the result of an

efficient project is to maximise the optimal value of land, assuming that
the optimal floor-area ratio (FAR) is selected. At this optimum the alloca-
tion of land and non-land (money) inputs is:
L = FV
(11)
M = (1 ) FV
Line one of (11) is now used for computing in terms of the optimal
floor-area-ratio (FAR) and V.
F
= V = FAR V (12)
L
From (11) we also obtain:
M (1 )
= (13)
L
Equation (9) is now used for computing the floor-area-ratio (FAR).

1
M
FAR = A (14)
L
Combining (12), (13) and (14) yields the value of the optimal unit
value of land in terms of V, the value of one sqm of floor-space.
1
= (1 ) ( A V )
1
(15)
Note that is a power function of the productivity times the value of

floor-space, and that the power 1 is much bigger than unity. In other
words, optimal land value is an increasing convex function of floor-space
value due to the fact that, as the latter increases in value, builders charge
more and pack more floor-space onto a given site. There remains the ques-
tion of floor-space value V. It is computed in the usual way, by dividing
bid rent by the (monthly) cost of capital k appropriate for similar risk pro-
jects.
R ( x, y )
V ( x, y ) = (16)
k
Inserting (16) into (15) we obtain:
1

1
A
( x, y ) = (1 )
R ( x , y ) (17)
k
And inserting (8) into (17) yields ( x, y ) as a function of T(x), the

cost of transportation. Since T(x) is a measure of the value of travel time,
we should expect that ( x, y ) will be highly sensitive to even small time
savings stemming from a more efficient transportation system.
1
1
1

1
A R0 y T ( x)

( x, y ) = (1 ) 1
k y0 y
(18)

Expression (18) shows that the steepest shift upwards in land value will
ensue from a combination of several occurrences: a rise in construction
productivity, a decline in the cost of capital, a rise in income, an increase
366 Daniel Gat
in the travel velocity and, of course an expected rise in Ro, core CBD rents
that tend to rise during a speculative run. The monocentric single-mode
model needs to be completed by computing the value of optimal FAR in
terms of location x, and by computing the travel time and travel expendi-
tures T(x). Combine (12) and (15) to obtain the optimal FAR given the bid
rent R(x, y):
1

1
(1 ) R( x, y )
FAR ( x, y ) = A [ (1 )V ]

1 1

= A

(19)
k
17.3.2 Travel time, congestion modelling and travel

expenditures
In a congestion-free monocentric mono-modal city, a single trip travel time

is simply the distance x between the origin (the home) and the single desti-
nation at the CBD divided by wo the free-flow velocity. However, conges-
tion plagues most medium to large size cities and is a major consideration
when contemplating congestion-free transit.
Vehicular Congestion is difficult to model correctly, and is even harder
to do within the framework of a model involving not just travel, but as
here, the interaction between travel and the real-estate market. Therefore, a
simple modelling strategy is followed, one that distills the essential charac-
teristics of congestions impact on travel velocity, without attempting to
follow the classic theory (e.g. Sheffi 1985). Especially since roads are not
an ingredient of the model, congestion is made a characteristic of vehicle
position on the home to centre ray.
The main assumption is that congestion is highest in the immediate vi-
cinity of the centre, and tapers off as a car drives away from that centre. To
be more specific, car speed at the centre is very slow and speed picks up as
distance from the centre grows, tending asymptotically towards free-flow
speed at the edge of the built area. The same idea is expressed in (Anas et
al. 2000) with somewhat more sophisticated math which ensures that speed
exactly matches free-flow speed at the circular and computable city edge.
This matching is achieved analytically, but is made possible because the
Anas et al. model remains monocentric and therefore symmetrical. The
present congestion sub-model has to serve, as the chapter unfolds, a poly-
centric city model that has no symmetric structure and whose boundary is
highly non-symmetric and thus incapable of being solved analytically. A
somewhat simpler congestion model is therefore formulated:
We begin by presenting a formal model of the duration of a single trip
under congestion as a (concave) function of distance from the centre to the
home (see Figure 17.1). The home to centre trip is assumed to be the con-
vex mirror image and therefore of equal duration.
x
t ( x) = a(C ) 1 e x +
w0
0
t (0) = a (C ) 1 e0 + =0 (20)
w0
x x
t ( x >> 0) = a(C ) [1 o( x) ] + a(C ) +
w0 w0
Figure 17.1. Travel time under congestion
The congestion sub-model computes the single trip travel time t(x). This
is converted into money terms in two steps: It is multiplied by the mobility
m, which is the number of trips a single household generates during each
month to give the total number of hours spent travelling and thus not spent
earning. That product is then multiplied by the value of travel time to
yield T(x), the value that the household loses each month due to travel.
Given the computed values of T(x), the four related gradients: value of
floor-space and of land, rent, and the optimal FAR can then be computed
(Figures 17.2 and 17.3).
368 Daniel Gat
Figure 17.2. Values of floor-space and of land
Figure 17.3. Rent and FAR
This concludes the presentation of the monocentric car-only urban

model. The ground is now prepared for us to analyse the polycentric
model.
17.4 The polycentric single-mode model
The monocentric models task and achievement is to derive the rent and
floor area gradients in space within a city that has exactly one centre: that
centre being the destination of all household trips. But it side-steps an im-
portant problem, that is, what if there are several centres? (see Figure 17.4)
Figure 17.4. Geography for a city with 10 centres
Urban scholars have long been grappling with the many-centred city
problem, and they have usually focussed on two important questions: why
do secondary centres get created? And where would they locate? Both of
these questions are not relevant to our topic here. Instead, we focus on a
third major question: How will residential land use align itself, given a
spatial distribution of several centres? This latter problem is a direct exten-
sion of the monocentric models problem; it is solved here by following
the strategy of that model, namely, by asking how much travel will an x
located household generate when there is more than one centre and loca-
tion is no longer a single number x; it is a pair of coordinates x = (x1, x2)?
This idea is the brainchild of Romanos (1976) who did not follow through
in detail. In that new situation households need to select a destination for
each intended trip.
The mechanism for selecting destinations in an urban model has a long
tradition in urban studies, beginning with Reillys Rule, moving on to the
gravity models many versions, and settling more recently into the general
framework of discrete choice modelling. Here, the following stochastic
version is used:
370 Daniel Gat
Bi e t ( x ,i )
(i x) = N
(21)
B
k =1
k
t ( x , j )
e
Equation (21) displays the probability of destination i being selected

when a trip from location x is contemplated. That probability is an increas-
ing function of the size Ci of the given centre, and it declines geometrically
with t(x, i), the travel-time from x to i. Thus, a large centre will attract
more trips, but since its size contributes to congestion, that attraction is at-
tenuated. Trip time is no longer a single number as computed in (20); it is
a vector of elements t(x, i) computed for each distance dist(x, i):
dist ( x, i )
t ( x, i ) = a (Ci ) 1 e dist ( x ,i ) + (22)
w0
It is now possible to compute the expected amount of monthly travel

generated by a household living at location x, as the mean trip duration
E t poly ( x) = ( x k ) t ( x, k ) multiplied by the monthly number of
k
trips and by the wage rate:
Tpoly ( x) = w m ( x k ) t ( x, k ) (23)
k
The final step in moving from mono to poly is to insert the new Tpoly(x)
instead of the old T(x) into all of the above equations that are dependent on
it. The result is a set of 3-d gradient graphs with peaks at the locations of
each of the centres. The height of the peak is an increasing function of the
centre size. Since all of the 3-d gradient maps in our analysis in this chap-
ter are related to the rent gradient, that one only is shown in Figure 17.5.
This completes the derivation of the polycentric car-only model. The
following section develops the next step: proceeding from a single travel-
mode to a duo-mode that includes cars and a transit system.
$6.00
$5.50
$5.00
$4.50
$4.00
$3.50 10
$3.00 8
$2.50
$2.00 6
$1.50 4
$1.00
$0.50 2
$0.00
0
-10
-8 -2
-6
-4 -4
-2
0 -6
2
4 -8
6
8 -10
10
Figure 17.5. An example of the 3-d polycentric rent gradient
17.5 A city with two (or more) travel modes
In a city with more than one mode, decisions about travel need two deci-
sion mechanisms; one, as described in the previous section, is to select a
destination, and the other is the decision on mode of travel. This section
treats the two decisions as a sequential process. Since travel time is an in-
put to the choice of destination, it needs to be known in advance to the
would-be traveller. This is best shown as a two-step backward folding de-
cision tree. (see Figure 17.6).
When transit is introduced it cannot be used on its own since transit
stops need to be reached. Thus we can envision walk and ride (transit),
bike and ride, park-and-ride or even bus and ride. However, to keep the
presentation uncluttered without loss of generality, only two modes are
shown: door-to-door car and park-and-ride (P&R). To further simplify,
transit stations have been assigned to each of the centres and only to these
centres.
The computation begins at extreme tips of the tree wherein t(x, i: car)
and t(x, i: P&R), the times to reach centre i by car or by park-and-ride are
placed. Then, the probability of taking either mode conditional on the ori-
372 Daniel Gat
gin and destination is computed and placed on the arms of the mode deci-
sion fork.
car t ( x ,i car )
e
(car x, i) = t ( x ,i car ) t ( x ,i P & R )
e car + e tran (24)
( park & ride x, i) = 1 (car x, i)
t (x, 2 : car)
Destination 1 Prob(car : x, 2
)
Destination 2
Prob(P&R : x, 2 )
Destination 3
t (x, 2 : P&R)
Destination 4
Destination 5
Figure 17.6. A two-step backward folding decision tree
Following that, the computed expected-travel-time is located at the tip

of each destination choice arm. From there we progress as in section 4 to
compute T(x) the monthly travel expenses associated with each location x
within the territory of the model. But before that we need to pause in order
to explain how the model computes travel time for park-and-ride.
0.90 - 1.00
0.80 - 0.90
0.70 - 0.80
0.60 - 0.70
0.50 - 0.60
0.40 - 0.50
0.30 - 0.40
0.20 - 0.30
0.10 - 0.20
- - 0.10
Figure 17.7a. Probability of taking transit when the destination is the CBD
Figure 17.7b. Probability of taking transit when destination is centre no. 7
17.6 Transit and park-and-ride travel time
The most glaring fact about a rail transit system is the existence of a fixed
rail network that constitutes an exclusive right-of-way for its vehicles. Let
us also assume that all road crossings are at separate levels. These two
374 Daniel Gat
characteristics combined give the system a high carrying capacity, which

enables transit to avoid slowdowns caused by congestion.
However, transit carries an added cost in comparison to the car mode:
stops need to be reached and following that, there is an expected waiting
time for vehicle arrival. Thus, park-and-ride time is made up of at least
three components: home to station by car, waiting, and transit riding. In a
very large and complex transit network, time for changing trains is also a
factor, but that is ignored here. Travel Time on the Transit Network is
computed by applying the method of min-addition, a highly efficient
shortest route algorithm (Pandit 1962). We begin with an origin-
destination matrix of travel time on direct links, the one-step ttran1(i,j)
where unlinked pairs are given a very high value. Elements of the two-step
matrix are computed by the following minimisation:
ttran 2 (i, j ) = Min ttran1 (i, k ) + ttran1 (k , j ) (25)

k
This procedure can be repeated and at each repetition the number of

steps is raised to the power of 2. The min-addition is stopped when there is
no further reduction in the value of transit travel time. A similar procedure
is used in the model to select the best park-and-ride choice. For each loca-
tion x, the (congested) travel time by car to each of the stations is min-
added to the best transit time matrix. Only one pass is needed here.
t ( x, i P & R ) = Min t ( x, k car ) + ttranbest (k , j ) (26)

k
The next step is to compute the expected travel time of a single trip in
the polycentric dual-mode city, given the home location x. The congested
car-only travel time t(x,i:car) is computed just as in (20). The park-and-
ride travel time t(x,i: P&R) is available in (26). Their probabilities,
(car x, i) and ( P & R x, i) are available from (24). So the expected
trip duration generated by a household living at x is:
E t ( x, i) = (car x, i) t ( x, i car ) + ( P & R x, i) t ( P & R car )

(27)
E[t(x, i)] from (27) can now replace t(x,i) in (21) and (22) so as to com-
pute TP&D(x) the monthly travel expenditures of a household living at loca-
tion x in a polycentric dual-mode city. TP&D(x) is then inserted into the rent
and all related gradient functions so as to yield the real estate functional
surfaces of the polycentric dual-mode city (see Figures 17.8a-c).
27.50 - 30.00
25.00 - 27.50
22.50 - 25.00
20.00 - 22.50 30.00
27.50
17.50 - 20.00 25.00
22.50
15.00 - 17.50 20.00
17.50
12.50 - 15.00 15.00
10.00
8.00
12.50
10.00 - 12.50 10.00 6.00
7.50
7.50 - 10.00 5.00 4.00
2.50
2.00
5.00 - 7.50 -
-10 0.00
2.50 - 5.00 -8
-2.00
-6
- - 2.50 -4
-4.00
-2
0
-6.00
2
4 -8.00
8 -10.00
10
Figure 17.8a. E[t(x, i)] the expected trip duration without regard to mode for the
No-Transit City
30.00
27.50
25.00
22.50
20.00
17.50
10.00
15.00
12.50 8.00
10.00 6.00
7.50
5.00 4.00
2.50
2.00
-
-10 0.00
-8
-6 -2.00
-4
-4.00
-2
0
-6.00
2
4 -8.00
6
8 -10.00
10
Figure 17.8b. E[t(x, i)] the expected trip duration without regard to mode for
the Light Rail City (Transit speed is 45 km/hr)
376 Daniel Gat
30.00
27.50
25.00
22.50
20.00
17.50
10.00
15.00
12.50 8.00
10.00 6.00
7.50
5.00 4.00
2.50
2.00
-
-10 0.00
-8
-6 -2.00
-4
-4.00
-2
0
-6.00
2
4 -8.00
6
8 -10.00
10
Figure 17.8c. E[t(x, i)] the expected trip duration without regard to mode for the
Heavy Rail City (Transit speed is 75 km/hr)
As transit is introduced and its speed increases, overall travel time de-
creases due to two mutually reinforcing reasons: i.) faster travel by transit
and ii.) higher probabilities assigned to the faster routes. This is especially
true for residents living near the stations (as Figure 17.8a-c clearly shows).
Transits impact on FAR. There is a serious catch when we attempt to
compute the impact on floor area ratio. While the rent as well as the poten-
tial optimal FAR and optimal land value functions can react immediately
following the installation of transit, realised FAR must consider that which
is already in existence. Functional buildings that exist have staying power.
They are practically immune to enlargement, and are not torn down readily
since they continue to supply real estate services.
Open land at the edge of the city can begin to react quickly to an upward
rent shock, whether it is caused by income, demography, an improved
transportation system, or a combination thereof. But within the built up
space of the inner city, especially favourable conditions must prevail to
cause redevelopment by demolition. This is the topic of the next section.
17.7 Redevelopment by demolition
When an existing building is torn down to make room for a new one, the
value of that old building, plus its demolition costs are an investment as a
sunk cost that must be made to acquire a recycled site. This investment has
to be justified. The common investment rule says that the return on in-
vestment must be at least as great as a hurdle value usually the cost of
capital of a similarly risked project, say kR. Formally:
value of recycled site

(1 + k R )T (28)
val of old bldg + demolition cst
Let F be the size of the existing building, L the size of the site, * the
optimal value of the retrieved site conditional on the new optimal FAR.
Let be the demolition cost of 1 square metre of the old building, and let
T be the time it takes to demolish and prepare the new site. Then (28) is
equivalent to:
*L V F * F* FAR*
= = = (1 + k R )T
(V + ) F (V + ) F (1 + V ) F (1 + V ) FAR
(29)
FAR*
(1 + V ) (1 + k R )T /
FAR
So the optimal strategy is to demolish and build if the ratio of the Opti-
mal FAR to the Existing FAR is at least as great
as (1 + V ) (1 + k R )T / . Let us put some numbers on that. If both
V and k R are 10% per year, and if T is one year and is 1/3, then the
hurdle ratio is roughly 3.6. This ratio is congruent with the situation in
many cities: single floor buildings of an early development stage are fre-
quently torn down to make room for 3 to 4 floor walk-ups. These are
sometimes demolished and replaced by 10 to 15-story tower blocks.
The picture could change drastically if the old building is in a deterio-
rated state and is valued at only a small proportion < 1 of a new building
of the same size. That would reduce the hurdle ratio to
(1 + V ) ( + k R ) / . This concludes the final step of formulating the
polycentric dual-mode urban model. The rest of the chapter will discuss
numeric simulation runs that compare the single mode city with a city with
rail transit.
378 Daniel Gat
17.8 Simulation runs and their results
This section reports the outcome of nine simulation runs, a Cartesian prod-
uct of three transit states (including no-transit, light rail and heavy rail) and
three income states, as the following table shows:
Table 17.1. The simulation table

Income No Transit Transit at 45 km/hr Transit at 75 km/hr
$2,000
$2,500
$3,000
350%
350% 350%
300%
300% 300%
250%
250% 250%
200% 10
200% 10 200%
150% 8 10
150% 8 150% 8
100% 6
100% 6 100% 6
50% 4
50% 4 50% 4
0% 2
0% 2 0% 2
0
0 0
-10 -10 -2
-2 -8 -10 -2
-8 -6 -4 -8
-6 -4 -4 -6 -4
-4 -2 -6 -4
-2 -6 0 -2 -6
0 2 -8 0
2 -8
4 2 -8
4 6 -10 4
6 -10 -10
No Transit / Income $2000 8 6
8 Transit at 45 km/hr -- Income $2000 10 Transit at 75 km/hr -- Income $2000 8
10 10
350% 350%
350%
300% 300%
300%
250% 250%
250%
200% 10
200% 10
200% 10
150% 8 150% 8
150% 8
100% 6 100% 6
4
100% 6
4
50% 4 50%
2 50% 2
0% 2 0%
0 0% 0
0
-10 -2 -10 -2
-8 -10 -2 -8
-6 -4 -8 -6 -4
-4 -6 -4 -4
-2 -6 -4 -2 -6
0 -2 -6 0
2 -8 0 2 -8
4 2 -8 4
6 -10 4 6 -10
No Transit / Income $2500 8 6 -10
Transit at 75 km/hr -- Income $2500 8
10 Transit at 45 km/hr -- Income $2500 8 10
10
350% 350%
350%
300% 300%
300%
250% 250%
250%
200% 10 200% 10
200% 10
150% 8 150% 8
150% 8
100% 6 100% 6
100% 6
50% 4 50% 4
50% 4
0% 2 0% 2
0% 2
0 0
0
-10 -2 -10 -2
-8 -10 -2 -8
-6 -4 -8 -6 -4
-4 -6 -4 -4
-2 -6 -4 -2 -6
0 -2 -6 0
2 -8 0 2 -8
4 2 -8 4
6 -10 4 6 -10
6 -10
No Transit -- Income $3000 8 Transit a t75 km/hr -- Income $3000 8
10 Transit at 45 km/hr -- Income $3000 8 10
10
Figure 17.9. The simulations of realised FAR under the nine scenarios
There are many types of results associated with each scenario along
with nine scenarios. Therefore, a real risk exists of swamping the reader
with too many figures and graphs. To avoid that, we show what happens to
only two variables under each of the nine scenarios. The Realised Floor
Area Ratio and the CBD Land Values is selected as the representative
summary measures, since they reflect a likely eventual outcome in real es-
tate terms. Figure 17.9 shows the spatial distribution of accumulated hous-
ing capital geo-cells belonging to the initial city (income $2000 and no
transit) are colored grey and kept grey, even as some such cells redevelop.
For new buildings to be built, new land values must be created in suffi-
cient intensity in order to attract builders. This happens rather easily on
green-fields at the fringes of cities, and it is illustrated in all the simula-
tions. Higher income and/or faster transit generate a secondary wave of re-
development by demolition just outside of the old city boundary.
The story is of course different in the inner city. As long as buildings are
well maintained and keep their values near the CBD, they are immune
from demolition and redevelopment. This shows up as a development la-
goon or crater at the old city; it takes a combination of very fast transit and
a very steep income hike to make redevelopment through demolition eco-
nomically feasible. This is manifest in the very last simulation diagram
(the lower right-hand side of Figure 17.10).
CBD Land Values given Urban Scenarios
3,500
3,000
2,500
2,000
1,500
Land Value ($ per sqm)
1,000
500
$4,000
-
$3,000
no transit household income
transportation status light rail $2,000
heavy rail
Figure 17.10. The simulations of CBD land values under the nine scenarios
The Time Dimension. The simulation runs refer to the immediate impact
after an income shock or a transit line opening. Prices change in response
to these shocks (or even in anticipation of them). Shortly after that, devel-
opers respond by initiating project plans and building starts. Completions
vary in time, depending on size of project and local building technology.
380 Daniel Gat
17.9 A brief note on urban growth and non-housing real

estate
Successful cities continue to grow by attracting and holding onto new

households, since they can offer a high quality of life. Their higher welfare
is comprised of increased income and more of the urban amenities called
city lights. Here we concentrate on the growth of urban income, namely,
its causes and consequences in terms of the non-housing real estate market.
The argument is easier to follow if we reverse the order: first conse-
quences, then causes.
Consequences: The housing model developed above relies on the as-
sumption of inhabitant workers of equal skills and therefore of equal
wages y. Furthermore, let us assume that such income is the wage paid by
a typical firm located at the very centre. It hires E employees, leases F sqm
of business floor-space, and applies $K of capital. It produces a staple
product Q, which is consumed at home and is also exported at price P, de-
termined by the larger exogenous market. Besides wages, it pays a rent Rbiz
and its cost of capital is also exogenously dictated.
The product Q is produced according to the following production func-
tion with inputs allocated optimally:
Q = B E F K , + + = 1, , , > 0
The added value Q.P is distributed to labour, real estate and capital
owners in the proportions : : to yield:
QP
y= $ / employee month
E
QP
Rbiz = $ / sqm month
F
QP
= % / month
K
We can easily see that if either Q or P (or both) increase, then input
owners demand their fair share of the new product. The wage y has to in-
crease, but the commercial rent Rbiz must also increase. (The cost of capital
is exogenous, so it cannot change. Instead, K the value of capital,
changes to reflect the inflated product.) If office rent rises sufficiently, it
could trigger redevelopment by demolition, as in the case of residential,
and new and taller office structures may emerge.
This is what happened to the centre of San Francisco, as the following

photos (see figure 17.11) show, probably with some help from BART.
Source: Access Magazine, University of California Transportation Center
Figure 17.11. The changing of the San Francisco skyline, looking toward Ember-
codero station
Causes. Having seen the consequences of growth in P and/or Q, we

need to ask ourselves: how can that growth arise? From observation of
economic history we know that economically successful cities are the cru-
cible of invention and creativity (Jacobs 1969, 1984). They attract what
Florida (2005) calls the creative class people who contribute new
ideas, products and processes. An improved product often stimulates a
surge in demand outside the home market that subsequently lifts its price.
An improved process leads to higher productivity.
382 Daniel Gat
17.10 Conclusions
This chapter has developed a simulation model of a single travel mode

polycentric city that takes on a transit network, becoming duo-modal. That
model was applied to tracing the impact on the residential real estate
market of turning on and accelerating the transit system.
Basically, there were three impacts: i.) Transit that connects urban cen-
tres causes a general decline in travel time and a corresponding flattening
of the rent gradients; ii.) as land at the fringe becomes more valuable,
green-field development takes place; iii.) a combination of sufficiently fast
transit and a strong upward income shock may also trigger development by
demolition at the inner city and the CBD.
However, the reader must bear in mind that, although the urban model
was designed to emulate the behaviour of a real city, some important in-
gredients are missing and should be included in future studies. Namely: i.)
a more detailed treatment of non-residential real estate; ii.) a mechanism
for incorporating quality and the aging of quality; iii.) more than one type
of consumer; and iv.) additional travel modes, especially walking, biking
and busing.
References
Alonso W (1964) Location and Land Use: Toward a General Theory of Land
Rent. Harvard University Press, Cambridge, Massachusetts
Anas A, Arnott RJ, Small KA (2000) The Pan-Exponential Monocentric Model.
Journal of Urban Economics 47:165179
Bernick M, Cervero R (1997) Transit Villages in the 21st Century. McGraw-Hill,
New York
Bertaud A, Brueckner JK (2003) Analysing Building Height Restrictions: Pre-
dicted Impacts, Welfare Costs and a Case Study of Bangalore India. In: Alain
Bertauds Academic Homepage
Bowes D, Ihlanfeldt KR (2001) Identifying the Impact of Rail Transit on Residen-
tial Property Values. Journal of Urban Economics 50:125
Cervero R (1998) The Transit Metropolis. Island Press, Washington
Florida R (2005) Cities and the Creative Class. Routledge, Oxford
Frank LD, Engelke PO, Schmid TL (2003) Health and Community Design. Island
Press, Washington
Frumkin H, Frank L, Jackson R (2004) Urban Sprawl and Public Health: Design-
ing, Planning and Building for Healthy Communities. Island Press, Washing-
ton
Hatta T, Okawara T (2001) Population, Employment, and Land Price Distribution

in the Tokyo Metropolitan Area. The Journal of Real Estate Finance and Eco-
nomics 6:103-128
Jacobs J (1969) The Economy of Cities. Random House, New York
Jacobs J (1984) Cities and the Wealth of Nations. Random House, New York
Mills ES (1972) Urban Economics. Scott Foresman, London
Mills ES (1999) Truly Smart Smart Growth. Illinois Real Estate Letter, Office of
Real Estate Research, University of Illinois, Urbana
Muth RF (1969) Cities and Housing: The Spatial Pattern of Urban Residential
Land Use. University of Chicago Press, Chicago
Newman P, Kenworthy J (1999) Sustainability and Cities. Island Press, Washing-
ton
Pandit SNN (1961) A New Matrix Calculus. Journal of the Society for Industrial
and Applied Mathematics 9:632639
Richardson HW (1983) The New Urban Economics. Routeledge, Oxford
Romanos MC (1976) Residential Spatial Structure. Lexington Books, Lanham
Ryan S (1999) Property Values and Transportation Facilities: Finding the Trans-
portation-Land Use Connection. Journal of Planning Literature 13
Scheurer J, Newman P, Kenworthy J, Gallagher T (2001) Can Rail Pay? Institute
for Sustainability and Technology Policy, Murdoch University, Perth
Sheffi Y (1985) Urban Transportation Networks. Prentice-Hall, New Jersey
Wheaton W (1998) Land Use and Density in Cities with Congestion. Journal of
Urban Economics 43:258272
Webber M (1963) Order in Diversity: Community without Propinquity. In: Wingo
Jr L (ed) Cities in Space: The Future Use of Urban Land. Johns Hopkins
Press, Baltimore
Webber M (1995) Bart: The Lesson. Discussion paper, University of California
Transportation Center, Berkeley
Webber M (1998) The Joys of Spread City. Urban Design International 3:201-
206
18 Land use variables in trip generation
models: The case of the light rail transit in
Tel Aviv
Avigail Ferdman and Daniel Shefer
Center for Urban and Regional Studies, Technion Israel Institute of

Technology, Haifa, Israel
Shlomo Bekhor
Transportation Research Institute, Technion Israel Institute of Technology,

Haifa, Israel
18.1 Introduction
Transportation and urban land use maintain a complex, symbiotic relation-

ship. Transportation affects land use by improving accessibility to urban
functions, and the built environment affects travel through its distribution
and density. It is important to explore these dynamics in order to under-
stand the effects that determine urban and metropolitan daily travel.
The main claim of this chapter is that land use characteristics account
for at least some of the daily travel habits in the metropolis. Current prac-
tice in transportation demand modelling emphasises socioeconomic factors
as explanatory variables for daily travel rates. Land use characteristics,
however, are less considered, if at all, in such models.
This chapter presents findings from a study conducted on Light Rail
Transit (LRT) stations in Tel Aviv Metropolitan Area (TAMA), in which
land use characteristics were explicitly included in travel generation mod-
els and forecasts. These land use variables were then tested in several land
use development policies in terms of population and employment density
and distribution.
The analysis is organised as follows: The first part reviews the literature
dealing with land use effects on urban travel trends. The general dynamics
386 Avigail Ferdman, Daniel Shefer and Shlomo Bekhor
of land use and travel trends are discussed, followed by a review of land
use variables such as density and diversity, land use mix, and their im-
pact on travel behaviour. Thereafter follows a brief overview of land use
parameters involved in transportation modelling.
The methodological section presents the trip-generation models and
briefly describes the simulations of land use development policy scenarios
for the LRT stations. The results of the estimation process and the com-
parison of land use policies is then presented, followed by a discussion of
some key findings regarding several land use variables which seem to be
especially influential in determining trip patterns. Finally, conclusions are
drawn regarding land use variables ability to represent intrinsic travel de-
mand.
18.2 Review of the literature
The literature review has been divided according to the different subjects
covered here: land use and trip trends, land use density, land use mix, and
land use variables in trip-generation modelling.
18.2.1 Land use and trip patterns
Numerous theoretical and empirical studies strongly suggest that certain

land use characteristics may promote urban sustainable growth by reducing
the use of private cars. The theoretical studies contend that transportation
improvements enhance the number of interactions between different urban
functions, and allow activities to exploit these improvements by distancing
themselves from the centre (Giuliano 1995). Transit cannot provide an ef-
ficient service in a sprawled environment, since it relies on scale econo-
mies. To counter and contain uncontrolled urban sprawl and promote the
use of transit, residential and employment densities alongside good public
transport infrastructure and agglomeration economies will increase tran-
sits modal share (Wegener and Frst 1999). Regarding trip frequencies,
however, Wegener and Frst, based on Zahavi (1979), assert that these
will not be affected by land use characteristics, since they are subject to
fixed time or budget constraints. People maximise their activities or oppor-
tunities in response to their time or budget constraints, and will not neces-
sarily increase their trip-production rates, but will rather select a different
trip mode or trip destination.
Pushkarev and Zupan (1977) were the first to report on urban residential
density and its capability to positively influence transit demand and thus
Land use variables in trip generation models 387
reduce car ownership and use. Other researchers have demonstrated simi-
lar trends (Kenworthy and Laube 1999a, 1999b; Cervero 2001; Parsons
Brinckerhoff Quade and Douglas 1996a). At the micro, mostly neighbour-
hood level, results are somewhat more varied. Cervero and Kockelman
(1997) concluded that density and diversity play a significant, albeit mar-
ginal role in determining trip rates. Other studies found that residential
proximity to a transit station will attract more riders to the station (Parsons
Brinckerhoff Quade and Douglas 1996b; Stringham 1982).
Critics of the capability of land use attributes to affect travel demand
and travel rates in particular, state that land use does not generate signifi-
cant results in demand forecasts, and that the most influential factors still,
are travellers socio-economic-demographic characteristics (Asensio 2002)
or policy variables, such as parking supply and fees (Hess 2001; Parsons
Brinckerhoff Quade and Douglas 1996a). High residential density, for ex-
ample, will have very little effect on car travel distances if cost measures
are not included (Wegener and Frst 1999). Others state that, although the
connection between residential density and travel habits is visible, it is
vague (Gordon and Richardson 1997; Breheny 1992; Schimek 1996).
18.2.2 Land use density
Since employment and residential densities seem to affect travel patterns

differently, it is useful to distinguish between them.
Employment density: Ewing and Cervero (2001) argue that employment
density should draw more attention in both the literature and the actual
planning process, due to its ability to determine some aspects of travel pat-
terns. Practically speaking, employment density could be much easier to
pursue in non-residential areas (Badoe and Miller 2000), mainly since de-
veloping high densities among residential areas might encounter opposi-
tion from nearby residents (Chatman 2003). Miller and Ibrahim (1998) be-
lieve that the employment density effects on higher trip rates to the CBD
might be the result of the CBD central location and public transportation
service quality. In contrast, Ewing and Cervero (2001) state that higher trip
rates might be the result of drawbacks of car ownership in dense areas,
such as congestion and limited parking supply. In any case, CBDs have the
highest transit modal split among all metropolitan areas, and the propensity
to walk to/from a CBD station is five times higher than elsewhere (Parsons
Brinckerhoff Quade and Douglas 1996a). Job concentrations around com-
muter rail stations in the suburbs also enhance transit use, walking and
ridesharing (Frank and Pivo 1994; Parsons Brinkerhoff Quade and Doug-
las 1996a).
Residential density: proponents of Transit Oriented Development (TOD)

often argue for higher residential density around transit stations. Empirical
studies show that people living in denser areas use transit more frequently
(Cervero 1993). People are more inclined to walk in denser neighbour-
hoods (Ewing and Cervero 2001; Pushkarev and Zupan 1977) and generate
shorter trips all in all (Dunphy and Fisher 1996; Ewing and Cervero 2001).
These studies also found that in less populated areas car ownership levels
are higher, ceteris paribus, increasing the probability of car use. Light rail
is considered to be particularly affected by residential density near its sta-
tions, more so than commuter rail transit. Residential densitys biggest im-
pact on LRT stations occurs when a low density, such as a four dwelling
unit per acre, is increased to medium density, a 10-16 dwelling unit per
acre. Densities higher than 30 persons per acre will attract more riders, but
at a diminishing rate (Bernick and Cervero 1997).
Nonetheless, the ability to actually divert automobile riders to transit
will depend on other complementary factors, such as the size of the metro-
politan area, station spacing, CBD job densities, parking supply, and the
physical and functional surroundings of the station, which determine the
areas level of land use density and mix.
18.2.3 Land use mix
The main advantage of a mixed use area, in transportation terms, lies in the
proximity of the activities that shorten travelling distances between them.
Even in suburban commercial centres, the proximity of restaurants to
shops and other services may discourage car use (Cervero 1996b).
Empirical studies suggest that mixed uses decrease single occupant ve-
hicle use, but the results are far from conclusive. It is not yet clear whether
land use mix has the same effect on shopping trips, work trips, and other
trips (education, errands, etc.) (Cervero 1996b). Kumar and Levinson
(1995) found, however, that proximity to activities does not necessarily
lead to the more efficient travel pattern, trip chaining.
The effects of land use mix on travel have been receiving less attention
than residential or employment density (Cervero and Kockelman 1997);
this may be due to difficulties in estimating these effects. Since dense areas
tend to be mixed, separating the impact of mixed use from the density im-
pacts is problematical. Furthermore, these two land use attributes are not
measured on the same scale density is numerical, while mixed uses are
frequently based on nominal or ordinal scales (Rodenburg 2005). To date
there is insufficient knowledge as to how to quantify and measure multi-
functional-mixed land uses (Shefer 2003).
18.2.4 Land use variables in trip-generation modelling
One of the most prevalent methods for evaluating land use impacts on
travel trends is the transportation forecasting model, which combines sup-
ply with demand. The supply side is generally represented by the highway
and the transit networks. The demand side is a series of mathematical
models, the most common being the classical four-step model. In the
four-step model, the first step or model is trip-generation (production and
attraction); second is trip-distribution, third is modal split-mode choice,
and fourth is trip-assignment. The trip-generation model usually treats de-
mand as an inelastic process, that is, it is unaffected by the results of the
modal split or the assignment stage (Ortuzar and Willumsen 1990). The
trip-generation model moreover does not take into account that, essen-
tially, demand for travel is derived, i.e. it is subject to individual needs and
preferences. Trip frequencies are thought to be influenced primarily by
socio-economic-demographic variables, and only secondarily by the built
environment; therefore, studies focussing on trip-production and trip-
attraction estimations the subject of this study are less prevalent.
Most of the studies that do investigate trip frequencies do so in a
neighbourhood-type comparative way (i.e. traditional vs. contemporary,
car vs. pedestrian, urban vs. suburban). According to Ewing and Cervero
(2001), only three out of 28 studies on land use patterns and their influence
on travel patterns have used trip frequencies as dependant variables. Dun-
phy and Fisher (1996) found that vehicle trips are less frequent at higher
densities. Ewing et al. (1996), however, found no significant relationship
between total trip frequencies and land use. Sun et al. (1998) also found
that total trip frequency is less sensitive to land use variables than total
VMT (vehicle miles travelled). Ewing and Cervero (2001) thus conclude
that travel demand is inelastic with respect to accessibility, and overall trip
frequencies differ little, if at all, in different built environments. Wegener
and Frst (1999) also maintain that trip frequencies are subject to fixed
time and budget constraints. Trip lengths and mode choice, however, are
expected to be influenced by land use patterns.
According to Wegener and Frst (1999), trip-production rates, measured
at the household level, are not expected to be affected by land use changes.
But Aviram (2001) found that income elasticity in respect to transportation
in 1995 in Israel was higher than one, meaning that expenditure on trans-
portation increases at a rate higher than increase in income. A transporta-
tion improvement that reduces transportation costs increases the household
disposable income, thereby raising the expenditure on transportation.
Some of this expenditure could take the form of a rise in total number of
trips produced. A major land use change could bring about, via improved
accessibility, a reduction in transportation costs, thus potentially, through

the income effect, increase the total number of trips generated.
18.3 Methodology
The methodology employed in the present study is twofold: first, in order

to verify whether the inclusion of land use variables affects trip demand in
the TAMA, new trip-production/trip-attraction demand models are esti-
mated. Next, in order to explore whether different land use scenarios also
affect the trip frequency rates, the results of the demand models are applied
to five different land use development scenarios representing the surround-
ings of planned LRT stations (Ferdman 2005). Figure 18.1 presents the
LRT alignment and stations.
Figure 18.1. LRT Red Line in TAMA
Based on the theoretical and empirical literature regarding land use ef-
fects on travel demand, three hypotheses are tested. The first states that
mixed land use reduces motorised trip production and attraction. The sec-
ond states that dense land use attracts fewer motorised trips, and the third
states that dense land use produces fewer motorised trips. The trip rates in
this chapter refer exclusively to motorised trips, irrespective of the internal
modal split (no distinction between car or transit travel). The reason for in-
vestigating motorised trips is twofold: first, we want to analyse the overall
sensitivity of the demand for all motorised trips to land use. Second, we
aim to take advantage of a rich dataset as elaborated below, even though it
did not include records of non-motorised trips.
The first hypothesis follows other studies by assuming that the prox-
imity of activities in a multifunctional land use pattern shortens distances,
thus reducing the need for daily motorised trips while increasing the use of
non motorised modes, mainly walking and cycling.
The second hypothesis is based on agglomeration effects: the denser the
environment, the better it can be served by transit and especially by non-
motorised modes, and thus may lead to fewer daily motorised trips at-
tracted to it.
The third hypothesis states that a densification of land uses entails new
prospects (or disadvantages) that would change a households preferences,
and consequently its budget and time constraints. Since dense land use
may be also mixed, it will produce fewer daily motorised trips.
18.3.1 Model estimation
The first part of the research methodology is an analysis of land use vari-
ables in trip-generation models in the TAMA. A previous transportation
study was conducted in the TAMA to forecast LRT ridership (NTA 2001).
The trip-generation models in that study were estimated based solely on
travellers socio-economic-demographic characteristics.
These models formed the first step of a conventional four-step model (with
feedback) used to forecast LRT ridership. The database used for model es-
timation was constructed from the 1996/97 National Travel Habits Survey
(NTHS) and the 1995 Census (CBS 1998, 1999).
Our study further investigates the trip generation models, using the same
datasets, augmented by land use variables. The additional land use attrib-
utes include job density, population density, job-population balance, ser-
vice employment density (density of employees in the service sector), de-
pendency ratio (the ratio between the employed population to the total
population in a zone), intra-zonal workforce ratio, and a dummy variable
for mixed land use. These variables were extracted from the 1995 Census
and added to the NTHS database. The dataset preparation is briefly de-
scribed as follows.
The national census includes 1.113 420 records at the person level (20%
of the population in 1995). This data set was aggregated to statistical zones
in order to arrive at zonal population and employment attributes. The sta-
tistical zones of the TAMA were then identified, creating a dataset for 702
statistical zones, corresponding to a population of 2.282 465, and a work-
force of 1.007 570.
The 1996/97 national trip survey file originally contained 333 532 re-
cords of daily trips. Out of these, 155 390 correspond to persons living in
the TAMA. These records were then aggregated by household, yielding 18
823 records. The land use variables from the census were merged with the
household dataset using the same statistical zone.
Table 18.1 summarises the land use variables included in the dataset;
not all estimations were found to be significant in trip production and at-
traction models.
Table 18.1. Land use variables tested for inclusion in the production and attraction
models
land use variable units
density variables
population density zonal population/acre
job density jobs per acre
workforce density registered workers in a zone/acre
total household density households/acre
education employment density education jobs/acre
service employment density service jobs/acre
service employment ratio service jobs/acre
retail employment density retail jobs/acre
total service and commercial em- service and commercial jobs*/acre
ployment density
ratio variables
total registered workers in a zone/total zonal
dependency ratio
population
job-population balance jobs/population
employees residing & working in same
intra-zonal workforce
zone/total workforce of the zone
education employment ratio zonal education jobs /zonal jobs
service employment ratio zonal service jobs /zonal jobs
retail employment ratio zonal retail jobs /zonal jobs
mixed land use dummy variable 0 or 1
* Including retail, catering, communication, banking, services, education, health,
personal and organisation.
The land use variables were designed to measure either density aspects
or diversity of use. For lack of explicit spatial data in GIS format, the di-
versity variables (job population balance, mixed land use dummy variable)
had to be computed from the census data.
Trip rates are measured as daily trips produced by a household or at-
tracted by the workplace. A total of nine models were estimated: four
models for the trip-production process, for estimating home-based trips1
(home-based-work: HBW, home-based-education: HBE, home-based-
other: HBO) and non-home-based NHB trips. Five models were cali-
brated for trip-attraction, for estimating the home-based trips mentioned
above and also home-based-shopping (HBS) trips.
The techniques used for parameter calibration are purposely chosen so
that they resemble the techniques used in the original NTA model: Multi-
ple Classification Analysis (MCA) (Andrews et al. 1973) for trip produc-
tion models, and multiple regression equations for trip attraction models.
The reason for using different models for trip production and attraction
stems from the level of aggregation of the observations. The production
models are estimated at the more refined household level. The attraction
models are estimated at the zonal level, since there is no disaggregated
data related to jobs and other land use variables.
MCA uses the household characteristics to categorise them into groups,
in order to estimate the response of different kinds of households as a func-
tion of different household attributes (Ortuzar and Willumsen 1990). The
coefficients are presented in a table and expressed as deviations from the
overall mean; the constant term in the predicting equation is the overall
mean.
The MCA model specifies that a coefficient be assigned to each cate-
gory of each predictor, and that each individuals score on the dependent
variable be treated as the sum of the coefficients assigned to categories
characterising that individual, plus the average for all cases, plus an error
term.
Yij...n = + ai + bj + . . . .+ eij..n
where:
Yij...n = The score on the dependent variable of individual n who falls in
category i of predictor A, category j of predictor B, etc.
= Grand mean of the dependent variable.
ai = The effect of the membership in the i th category of predictor A.
1 Home-based trips are trips which either originate or terminate at home, in con-
trast to non-home-based trips, where neither end is at home.
bj = The effect of the membership in the j th category of predictor B.

eij..n = Error term for this individual n.
18.3.2 Land use scenarios
The second part of the research methodology was to compare various land
use development policies applied to the station areas. Four different land
use scenarios, in terms of density and diversity, were constructed and
compared to previous land use forecasts. The scenarios modified popula-
tion, households, jobs, and workforce density features. By intending to
solely examine the influence that land use has on LRT station areas, only
Traffic Analysis Zones (TAZs) surrounding the proposed LRT line were
chosen.2 The proposed LRT line will pass through the heart of the CBD of
Tel Aviv and will consist of 33 stations. The TAZs around the line were
divided into three groups: CBD stations, inner-areas station and outer-area
stations; in so doing we aim to explore whether the geographical distance
from the CBD, combined with the land use changes will have a significant
effect on trip generation.
Two types of scenarios were created: the first includes three high den-
sity scenarios and the second includes one mixed land use scenario. All
four scenarios were compared to a base scenario (relative to the NTA fore-
cast 2001).
The high density scenarios represent different rates of population and
job densities, varying in the location of the crowded TAZs. The dense sce-
narios all increase the commercial densities in the CBD (the areas sur-
rounding the most central stations of the proposed LRT line) to 80
jobs/acre, and the service and commercial employment ratio to 0.25 and
0.2 of the CBD employees, respectively. As for the non-CBD stations, the
population densities were increased in the inner-area and outer-area sta-
tions of the line: Scenario 1 increases population densities to medium (60
people/acre) in the inner-area stations, and high (100 people/acre) in the
outer-area stations. Scenario 2 increases population densities to high in the
inner-area stations and medium in the outer-area stations, and Scenario 3
increases both the inner and outer area stations to high density. The high
density values represent the high category values of the entire TAMA, so
as to conform to regional characteristics as much as possible (see Figure
18.2a-d).
2 These TAZs are included in a 500m radius buffer, a distance recommended in

the literature as an average comfortable walking distance to a station (Untermann
1984; OSullivan and Morrall 1996).
A: Scenario 1 B: Scenario2
C: Scenario 3 D: Scenario 4
Figure 18.2 a-d. Comparison of the five scenarios, based on population density,
job density and dependency ratio
Since household characteristics are regarded as the most influential at-

tributes of trip-production, the dependency ratio (the ratio between the em-
ployed population to total population in a zone) in these scenarios was in-
creased to 0.55 in the CBD and 0.5 in the non CBD areas (compared to a
regional 0.46 in the original [or base] forecast of NTA 2001), in anticipa-
tion that these areas host such households or individuals whose participa-
tion in the workforce will be high, especially in the CBD area.
In the mixed land use scenario approximately half of the TAZs were
designated as having mixed land uses, and were chosen as alternating areas
along the line (see Figure 18.2d). Four measures were used as attributes of
mixed land uses: i.) a high rate of intra-zonal workforce (employees who
reside and work at the same TAZ), ii.) a high rate of dependency ratio, iii.)
high ratios of service and commercial employment as in the dense scenar-
ios, and iv.) a balanced job-population ratio, as much as possible. In addi-
tion, each zone was assigned a dummy variable representing mixed and
non-mixed land use patterns. Figure 18.3 presents a comparison of the five
scenarios, based on the modified variables: population density, job density
and dependency ratio.
90.0 0.52
80.0
70.0 0.50
60.0
0.48
50.0
40.0
0.46
30.0
20.0 0.44
10.0
0.0 0.42
NTA scenario scenario 1 scenario 2 scenario 3 scenario 4
population density (pop/acre) job density (job/acre) dependency ratio
Figure 18.3. Comparison of the five scenarios, based on population density, job
density and dependency ratio
18.4 Results
18.4.1 Trip production models
Two models were estimated for each trip purpose. The first model contains
only household characteristics variables, similar to NTA models. This
model was purposely estimated for comparison with the documented re-
sults (NTA 2001). The second model contains the same household vari-
ables with one or more additional land use variable. Several land use vari-
ables were tested for each trip purpose.3 Some were found to be
3 The trip purposes investigated were: home-based-work, home-based-

education, home-based-shopping, home-based-other and non-home-based.
insignificant, and some simply did not improve the models explanatory
power. The chosen models for later application were those with the best
statistical fit, as shown in Table 18.2.
Table 18.2. Trip production models estimation

trip purpose home-base-work home-base-education
sub model NTA NTA
variable model our model model our model
Grand mean 1,67 1,66 0,48 0,48
0 -0,25 -0,21 -0,11 -0,11
number of Cars in 1 0,05 0,03 0,002 0,003
HH 2 0,37 0,33 0,17 0,16
3+ 0,74 0,74 0,45 0,45
0 -1,22 -1,24
1 -0,34 -0,31
employees in HH
2 0,92 0,95
3+ 2,25 2,22
1
2
3
household size
4
5
6+
0 -0,31 -0,31
children aged 8-18 1 0,22 0,22
in HH 2 0,76 0,75
3+ 1,32 1,32
intra-zonal workforce 1.5 (F=117,922) 0.52 (F=28,597)
job density -0.01 (F=5,288)
workforce density -0.004 (F=688)
total household density -0.2 (F=126,114)

dependency ratio
job-population balance
observations 18.153 16.302 18.528 18.486
R square 0,46 0,47 0,20 0,20
F test (model) 354.752 167.014 63.189 59.100
F test (main effect
(combined)) 523.035 481.423 153.421 148.548
F test (covariates
(combined) 62.503 28.597
Table 18.2. Trip production models estimation (continued)

trip purpose home-base-other non-home-base
sub model NTA NTA
variable model our model model our model
Grand mean 1.64 1.64 0.91 0.91
0 -0.8 -0.78 -0.36 -0.36
number of Cars in 1 0.2 0.19 -0.02 -0.02
HH 2 1.12 1.08 0.72 0.71
3+ 1.82 1.76 1.07 1.07
0 0.48 0.49 -0.19 -0.19
1 0.04 0.05 -0.05 -0.05
employees in HH
2 -0.39 -0.4 0.15 0.15
3+ -0.3 -0.31 0.31 0.31
1 -0.7 -0.76 -0.19 -0.19
2 -0.24 -0.26 -0.1 -0.1
3 -0.05 -0.05 -0.02 -0.02
household size
4 0.18 0.2 0.14 0.14
5 0.71 0.75 0.14 0.14
6+ 0.7 0.8 0.19 0.2
0 -0.09 -0.09
children aged 8-18 1 0.2 0.2
in HH 2 0.13 0.11
3+ 0.18 0.19
intra-zonal workforce 0.6 (F=8,571) 0.55 (F=14,036)
job density
workforce density
total household density
dependency ratio 1.48 (F=10,062)
job-population balance 0.04 (F=1,420)
observations 18,153 18,114 18,153 18,111
R square 0.15 0.15 0.13 0.13
F test (model) 45,074 40,015 48,077 40,615
F test (main effect
(combined)) 45,074 44,400 48,077 46,594
F test (covariates
(combined) 9,316 7,728
Home-based-work trip production model
The HBW model, which improved the original NTA model most signifi-
cantly, was that which added intra-zonal workforce, job density, workforce
density, and household density, together with two categorical variables
cars in household (grouped as 0,1,2 or 3+) and number of employees in

household (also grouped as 0,1,2,3+). The coefficients of the categorical
variables are quite similar between the models, because the additional
variables do not interact with the categorical ones. The goodness of fit
measure by R was never expected to be very high because of the high
level of data disaggregation.
The land use variables coefficients reported here demonstrate very low
zonal values. For example, raising the workforce density of the travel zone
from 24 workers/acre (the mean of the samples) by one standard deviation
to 44.5 workers/acre, reduces zonal trip production by merely 0.02 trips
per household.
The intra-zonal variable was the only land use variable with a positive co-
efficient. A possible explanation is that people who work in the proximity
of their residence are likely to generate more HBW trips due to the possi-
bility, for example, to return home on the lunch break and travel again to
work in the afternoon.4
Other trip production models
The addition of land use variables to the other trip purpose models (home-
base-education, home-base-other, non-home-base) did not much improve
the explanatory power of the original NTA sub-models, as can be seen in
Table 18.2.
In short, the trip production models obtained concur with the hypothe-
ses made by researchers like Wegener and Frst (1999) and Ewing and
Cervero (2001), that household trip production is mainly influenced by its
characteristics: size, income (in the current study we used number of cars
per household as a surrogate for income a non-available variable), and
number of employees. The results indicate that household generation of
trips is influenced most by the socio-economic variables rather than the
built environment and household location. It should be noted that the trip
production sub-models above measured only the total number of trips
made by the household, whereas other travel patterns, such as trip length
and mode choice, were not in the scope of the current study.
4Since the models only test for total number of trips, we can say nothing about
whether these trips are actually shorter, or what mode they utilise.
18.4.2 Trip attraction models
Contrary to the case of the trip-production models, the land use variables
included in the trip-attraction models showed a significant impact on four
out of five model outcomes. The next section presents the results for each
trip purpose.
Home-based-work trip attraction model
The only attraction sub-model which did not benefit from the explicit in-
clusion of land use variables is the model explaining commuting trips the
HBW sub-model. The addition of land use variables to the NTA explana-
tory variable total zonal employees, such as job density, service sector
density, commercial sector density etc. produced insignificant results. The
only HBW sub-model chosen for later application on the land use scenar-
ios was the NTA sub model presented in Table 18.3.
Table 18.3. Daily HBW trip attraction model*

Variable Original NTA model
Total employees 1.46 (37.39)
Observations 158
R square 0.9
* t values in parentheses
The finding that the HBW sub-model did not benefit from the density or
diversity variables suggests that these factors alone do not have the power
to alter motorised commuting trip frequencies.
Home-based-education trip attraction model
The only variable found significant in the original NTA model was the
number of employees in education. Table 18.4 shows that the additional
land use variable included in the model (service employment density) is
significant at the 0.05 level; a possible explanation for this is that a zone
containing both service and non-formal education facilities may attract
trips to higher education facilities, informal education activities, adult edu-
cation, and so on. Indeed, a correlation was found between zonal education
and service employees (Pearson Correlation equal to 0.407, significant at
0.01). Hence, this finding suggests that the land use variable employed
might also affect education trips attraction power.
Table 18.4. Daily HBE trip attraction models*

variable Original NTA model Our model
Employees in education 4.12 (28.88) 4.05 (28.04)
Service employment density - 19.61 (2.27)
Observations 159 159
R square 0.84 0.84
Home-based-shopping trip attraction model
Each commerce employee in the NTA model is predicted to attract 1.47

home-based-shopping trips (Table 18.5). However, when the land use
variable is introduced the effect of an additional employee is smaller. We
keep in mind that, although one might expect only one explanatory vari-
able to be sufficient, a low correlation5 between the two variables was
found, thus allowing the use of both. The log of the job density coefficient
shows that higher job densities increase the zones attraction power. The
variable, log of job density per km, was chosen over the job density itself,
since the distribution of job density is not normal (Ferdman 2005).
The sub-model results clearly indicate that higher job densities attract
more home-based-shopping trips. This finding, coupled with Cerveros
(1996b) suggestion that employment centres should include more retail ac-
tivities, strengthens the claim that a mixture of land use in employment
centres could boost trip chaining and decrease private car use for shopping
trips.
Table 18.5. Daily HBS trip attraction models*

variable Original NTA model Our model
Commerce employees 1.47 (12.13) 0.94 (5.92)
Log of job density per km2 - 332.97 (4.87)
Observations 158 155
R square 0.48 0.55
Home-based-other trip attraction model
Three models are presented in the case of HBO trips: the original NTA
model, a model with general zonal variables (service employees and total
5 The DurbinWatson value for these two variables is 2.15.

households) only, and a similar model with explicit land use variables. The
reason for presenting the intermediate model is to allow easy comparison
between the models. Table 18.6 presents the results. Two land use vari-
ables were found significant, with positive coefficients. The mixed land
use dummy variable, a category attributed to only 12 zones out of the 158,
indicates that areas with diverse activities draw a large number of trips
with a purpose other than work, such as shopping or education, which
originate or end at home.
Zones with a high dependency ratio also attract home-based-other trips,
thus indicating the existence of other activities (apart from the residential
activities) which attract these home-based-other trips.
Table 18.6. Daily HBO trip attraction models*

variable Original NTA Our model (with no Our model (with
model land use variables) land use variables)
Commerce employees 0.90 (1.87)
Service employees 1.07 (7.47) 1.15 (12.40) 0.92 (9.27)
Total households 0.92 (10.66) 0.67 (5.25)
Mixed use (dummy) 2558.29 (2.72)
Dependency ratio 3985.92 (2.44)
Observations 158 158 156
R square 0.8 0.82 0.85
Table 18.7. Daily NHB trip attraction models*

variabe Original NTA model Our model
Employees in services 0.63 (8.74) 0.55 (7.33)
Employees in commerce 1.48 (6.03) 1.27 (5.06)
Square of total households 0.001 (7.36)
Dependency ratio 4329.75 (7.48
Mixed use (dummy) 1222.60 (2.20)
Obsevations 158 156
R square 0.87 0.87
Non-home-based trip attraction model
The NHB NTA model tested service employment rates, commerce em-
ployment rates, and the square value of total households. The latter was
used as a proxy for zone size, which assists in estimating the rates of non-
home-based trips. Once again in the new model, the variables signifying
the diversity of a zone (the mixed land use dummy variable) and its zonal
workforce (dependency ratio) are significant with positive coefficients
(Table 18.7). These trends are rather similar to those discussed in the
home-based-other sub-model analysis, namely, diverse areas can attract
substantial numbers of motorised chained trips that do not end at the trav-
ellers residence.
18.4.3 Land use scenarios
Comparison of the various land use development policy scenarios, which

compose the second part of this study, allows us to estimate alternative
policy development effects on trip rates. To recapitulate, the scenarios are
as follows: three scenarios represent high population and job densities rela-
tive to the base scenario. Scenario 1 has an average population density of
75 per acre, Scenario 2 of 70 per acre, and Scenario 3 of approximately 80
persons per acre. Their job density is 55 jobs per acre, and their depend-
ency ratio 0.5-0.51. The fourth scenario represents a high degree of land
use diversity; each alternating TAZ was assigned the mixed land use
dummy variable and a high rate of intra-zonal workforce (employees who
reside and work at the same TAZ), a high rate of dependency ratio, high
ratios of service and commercial employment, and a balanced job-
population ratio, to the greatest possible extent. The fourth scenario is
denser in terms of population and jobs that the base scenario, but less
dense than the three dense scenarios (Figure 18.3).
The model application results show that an apparent distinction between
the land use scenarios in terms of trip-production per household, in Figure
18.4, is the result of the higher dependency ratio. It is evident that prior to
the increase in the dependency ratio, the variation between the scenarios
was smaller. The scenarios most sensitive to the increase are the dense
scenarios. The difference in total trips produced is brought about by the
home-based-work trip sensitivity to the dependency ratio. The dynamics
are obvious: households having more workers produce more trips to work
than households with fewer workers. The home-based-work trips have the
greatest effect on the total household trip-production rates. Therefore, the
total trip-production rates are also higher in the high dependency ratio sce-
narios.
Differences between the scenarios are also apparent when changes in
population density are introduced. Even when the dependency ratio is low
for all the scenarios, population density does not affect trip rates uni-
formly. There is a clear trend towards a reduction in trip-production when
accompanied by an increase in population density (Scenarios 1-2-3). Sce-
nario 2, having the lowest average density (69 people/acre) in the high
density group scenarios, produces the fewest daily trips, compared with all
other scenarios, including the base. The denser scenarios, 1 and 3 (77 and
81 people/acre, respectively), produce more motorised trips per household
than Scenario 2.
Figure 18.4. Total trip production per household
Figure 18.5. Total trip attraction per workplace

Regarding trip-attraction trends, as found in the different land use sce-

narios, it seems that attraction rates in models with land use variables are
significantly different than models with no land use variables (Figure
18.5). The forecasts are as much as 18% different from the non-land use
model estimates, implying that land use variables, when explicitly used in
the trip-attraction forecasting model, could have a substantial effect on
model outputs.
Furthermore, commuting trips proportion was found to be lower for the
dense scenarios, and especially for the mixed land use scenario, even when
the total number of the trip-attraction per workplace is the same across all
scenarios.
18.5 Discussion
This study has focussed on trip frequency, that is, the estimation of the to-
tal number of trips made by a household. Furthermore, due to data limita-
tions, only motorised trips were included in the model estimation. Given
the limitations, care should be taken with respect to the conclusions. The
following paragraphs discuss some of the issues related to the present re-
sults in light of previous studies.
Based on our hypothesis regarding trip-production rates, dense land use
patterns produce fewer motorised trips, due to the proximity of land use
functions, which reduces the need for motorised trips. The results of the
model estimations, however, do not support this hypothesis, since in our
models all land use variables added little, if anything, to the models ex-
planatory power.
Generally, the results of the trip-production models concur with the hy-
pothesis made by researchers Wegener and Frst (1999) and Ewing and
Cervero (2001), that household trip production is mainly influenced by
household size, income (number of cars per household, a surrogate for in-
come used in our study), and number of employees. A household facing
the need to reach its various daily activities will produce trips according to
its socio-economic-demographic characteristics, rather than the character-
istics of the built environment in which it is located.
There are, however, some land use variables which were prominent in
the modelling process estimations and could be easily integrated in other
trip generation models. The positive sign of the intra-zonal workforce co-
efficient indicates that residing in proximity to the workplace could in-
crease the motivation for making more daily trips, and reduce the motiva-
tion for chaining trips. This conclusion is supported by the results of the
non-home-based trip-production model of the mixed use scenario, in

which non-home-based trip rates are lower than other scenarios, even
though its intra-zonal workforce ratio is the highest. This finding indicates
that the proximity of workplace and residence does not encourage trip
chaining. Furthermore, it concurs with Kumar and Levinsons (1995) as-
sertion that living in locations that provide good accessibility reduces the
motivation to combine (chain) trips.
Some authors favour a zonal balance between jobs and housing
(Cervero 1996a; Shefer and Degani 1998), which enhances walking and is
supposed to reduce overall motorised trips. The intra-zonal coefficient
used here might imply to the contrary, i.e. the short distance actually moti-
vates individuals to generate more (maybe shorter) motorised daily trips.
However, since the data used in this study involve only motorised trips, it
is impossible to determine whether these trips replace walking and cycling
or whether they add to the total number of the trips produced.
In any case, if this market segment (people working near their place of
residence) were to be inclined to use transit, or could potentially be di-
verted to transit, it is a worthwhile investigation. The challenge, according
to Cervero (1996a) and Shefer and Degani (1998), is to supply areas with
suitable workplaces in order to realise the potential of short distances be-
tween workplace and residence. A possible solution, for example, is add-
ing an assortment of housing types in CBDs, where there is a large and di-
verse concentration of jobs.
Land use variables in trip-attraction models have a greater effect on trip
attraction rates than on trip production rates. Home-based-education trips,
for instance, are positively affected by the incorporation of the land use
variable service sector employment density, suggesting that the commer-
cial activity of a zone might affect its education trip-attraction power, con-
sidering that education consists of various types of education facilities.
This result may be interpreted in two ways: first, the linkage between edu-
cation trips and locations catering for high levels of services might point to
a mixed land use environment. Second, the motivation of a trip maker to
link an education trip with service facilities might indicate a tendency to-
wards chain activities and thereby chain trips, which forms a more effi-
cient travel pattern.
The home-based-shopping trips model also benefits from the inclusion
of the land use variable. The log of the job density coefficient shows that
when job density is higher, it tends to attract more shopping trips. Our
finding concurs with Cerveros (1996b) suggestion to promote diverse land
uses; it reinforces the claim that a mixture of land uses in employment cen-
tres could boost trip chaining and decrease car use for shopping purposes.
Home-based-other and non-home-base trips attraction models both em-

ployed a mixed land use dummy variable and the dependency ratio vari-
able. This result may contradict Kumar and Levinsons (1995) assertion
that areas accessible to a multitude of activities do not encourage trip
chaining. In this case, from the planning perspective, there is a strong ar-
gument for intensifying and diversifying transit hubs, first for diverting
some of the home-based-other trips into chained trips, and second for en-
hancing the transit/non motorised modes among the non-home-based trips:
the chained trips.
The dependency ratio variable has also proven to be a potent explana-
tory variable, especially when applied to the land use scenarios. It can be
argued that, more than a land use feature it is a socio-demographic feature.
However, although it is not strictly a land use attribute, it could well be af-
fected by land use regulations. For example, a supply of residential units to
suit small households may attract precisely the populations that produce
the high rates of home-based-work trips. High dependency ratio zones
demonstrated moreover a greater tendency towards chaining trips (at the
expense of home-based-other trips). Were this market segment found to be
inclined towards transit and non motorised trips (this assumption would
have to be tested with a mode choice analysis), LRT station surroundings
could include such residential land use and improve their transit ridership
share, by diverting a substantial share of the produced home-based-work
trips to transit.
It is noteworthy that high dependency ratio scenarios have generated trip
forecasts implying that the scenarios could incorporate other uses, and po-
tentially serve as a proxy for a zones degree of mixed land use. High val-
ues of dependency ratio may indicate more than merely the ratio between
employed and total population of an area; it is the expression of the diverse
demand for services by the employed population. This demand influences
the allocation of land use, which in turn might affect travel demand pat-
terns. This effect may especially manifest itself in CBDs and other mixed
land use areas where households belong to a variety of employment sec-
tors.
18.5.1 Land use scenarios: Density and mixed land use
Our study has found that, contrary to Wegener and Frsts (1999) state-
ment that trip rates would not be influenced by land use features, house-
holds do seem to respond differently in different land use patterns. Despite
the preliminary calibration results showing that land use variables can have
little or no effect on trip-production trends, there are actually some differ-
ences between the scenarios. High population densities and high depend-
ency ratio (the ratio between employed and total population in the zone)
can influence the number of trips made by the household.
The enhanced opportunities (or obstacles) materialising in the process of
population densification, could change the household preferences and in-
fluence its trip-production choice. It seems that dense areas, as demon-
strated by dense scenarios, indeed affect regional opportunities, which in
turn reduce the original trip-production rates.
It is interesting to note that, among the high density scenarios, however,
this trend is not linear. The higher density scenarios produce more trips per
household than the lower density scenario, a finding which may indicate
that above a certain density threshold, trip-production rates will increase
and density effects will gradually diminish. This might be due to the fact
that the denser an area, the larger the number of activities it contains. A
larger number of activities may be the explanation for the increase in num-
ber of trips produced. Previous work has identified that, above a density
threshold of 30 persons/acre LRT transit ridership rises, but at diminishing
rates (Bernick and Cervero 1997). A combination of these trends suggests
that, not only will motorised trip rates rise in high density areas, but transit
share will also actually decrease. Therefore, those planning high densities
around transit stations should consider this effect.
Another finding illuminates possible shifts in trip purposes in different
land use patterns. For example, there seems to be a counterbalance be-
tween home-based-work trips and home-based-other trips when the de-
pendency ratio is higher; households with more workers will produce more
home-based-work daily trips and fewer home-based-other trips.
This counterbalance also manifests in trip attraction trends; the dense
and especially mixed land use scenarios reduce the home-based-work trips,
and increase the home-based-other and non-home-base trips attracted,
compared to the NTA base scenario. According to previous studies, in-
creasing employment densities and mixed land uses enhance the use of
transit and walking (Cervero 1996b; Ewing and Cervero 2001; Frank and
Pivo 1994). The trip attraction rates discussed here represent only motor-
ised trips, with no reference to modal split, so it is difficult to either con-
firm or refute the findings of the aforementioned studies. On the one hand,
both the dense and mixed scenarios yielded higher trip-attraction results
than the NTA base scenario, meaning that they actually do not reduce the
demand, but rather increase it. Conversely, since only the motorised de-
mand is accounted for, the data has no information regarding the actual
share of non-motorised modes, and especially about the demand for transit
modes.
Surprisingly, the mixed land use scenario predicts higher trip-attraction

rates than the NTA base scenario, which contradicts the finding and con-
clusions of the literature reviewed above, stating that mixed land use in-
creases non-motorised modes use, thus reducing motorised demand. The
current results indicate that, contrary to the assumption, mixed land uses
encourage motorised trips, but these results are a consequence of the cali-
brated coefficients that measured trip-attraction in the mixed land uses. A
more complete dataset, with explicit data on land use diversity, could yield
different coefficients to explain the relationship between mixed land uses
and trips. The study also does not yield information as to the trip by mode
or length. We therefore cannot conclude whether mixed land use, in the
case of the TAMA, decreases or increases everyday travel.
18.6 Summary and conclusions
This chapter has estimated trip generation (production/attraction) models

to verify whether the inclusion of land use variables affects the demand for
trips in the TAMA. This analysis offers empirical evidence to the debate
regarding the need to include land use variables in trip-generation models.
In order to explore whether different land use scenarios also affect trip
rates, the results of the trip generation models were applied to five differ-
ent land use scenarios representing alternative land use development poli-
cies for the area surrounding a proposed LRT line. The outcomes give us a
clearer picture of dense and diverse land use impacts on daily travel rates.
The findings strongly suggest that the zone density and mixed land uses
can play a major role in determining trip generation rates. The first conclu-
sion we can draw from the modelling process is that excluding land use
variables from the set of explanatory variables in a forecasting model
could have major implications on its forecasting accuracy. Feeding partial
or biased trip-production and trip-attraction levels for the sequential model
methodology could result in a miscalculation of the demand for car travel,
transit and non motorised modes, and a consequent miscalculation of the
network assignment.
The land use variable that stands out among the production model vari-
ables is the intra-zonal workforce ratio. The higher trip rates found in high
intra-zonal workforce zones could suggest more frequent cold ignitions (a
modal split model test is needed in order to establish car trips), which re-
sult in increased harmful emissions and compromise air quality. But short
distances between home and work could actually encourage transit use, es-
pecially in dense areas, where parking is costly or insufficient, and public

transportation is highly accessible.
In any case, the main contribution of the intra-zonal workforce variable
is that the modelling methodology is universal; it can serve as an explana-
tory variable in trip-generation models based on datasets including motor-
ised and non-motorised trips, and thus help determine travel patterns of
this particular market segment.
Another variable affecting trip demand is the dependency ratio variable,
a measure of socio-demographic features with the capability to represent
intrinsic demand for travel. The simulation of land use development poli-
cies has also proved instrumental in distinguishing land use affects on trip
generation trends. High density zones, for example, attract fewer home-
based-work trips per worker, relative to the other home-base trips. Ag-
glomeration effects may be in play in this case; dense and diverse settings
include a variety of uses which draw more trips and possibly of varied
types. This finding is consistent with previous studies, maintaining that
dense job concentrations reduce the need for car travel, either due to transit
increased efficiency, or to the effects that density produces (insufficient or
costly parking and traffic congestion).
Earlier studies reveal that trip frequencies are subject to household char-
acteristics (Ewing and Cervero 2001) and fixed time and money budget
constraints (Wegener and Frst 1999). Our study shows that trip frequen-
cies are also subject to land use patterns. Households in denser areas
probably respond to increased opportunities (or disadvantages) in the met-
ropolitan area and alter their trip production rates. In Wegener and Frsts
terms, land use patterns might affect household disposable income, and
hence affect its trip-making pattern. In urban agglomeration terms, densely
populated areas consume more, and possibly different services, than dis-
persed population areas, therefore increasing their daily trip rates pro-
duced.
Joint development of LRT stations and land use could stimulate and
promote potential development incentives around the stations, thus raising
both activity and trip rates. This chapter has presented a methodology to
assess the full potential of LRT in terms of travel demand forecasting. Fur-
ther research will investigate the effect of land use variables to other com-
ponents of transportation models such as mode choice, in order to assess
the LRT market share and its benefits for the population in terms of allevi-
ating traffic congestion.
Acknowledgement
This research was partially supported by the Henry Ford II Transportation

Research Fund at the Technion Israel Institute of Technology.
References
Andrews FM, Morgan JN, Sonquist JA, Klem L (1973) Multiple Classification
Analysis. 2nd edn. University of Michigan, Ann Arbor
Asensio J (2002) Transport Mode Choice by Commuters to Barcelonas CBD. Ur-
ban Studies 39:18811895
Aviram H (2001) Evaluating the Impact of Urban Transportation Projects on Land
Values. Haifa: Technion Israel Institute of Technology (in Hebrew)
Badoe DA, Miller EJ (2000) Transportation Land Use Interaction: Empirical
Findings in North America and Their Implications for Modelling. Transporta-
tion Research Part D 5:235263
Bernick M, Cervero R (1997) Transit Villages in the 21st Century. McGraw-Hill,
New York
Breheny MJ (1982) The Contradictions of the Compact City; A Review. In: Bre-
heny MJ (ed) Sustainable Development and Urban Form. Pion, London, pp
138-159
CBS Central Bureau of Statistics (1998) Travel Habits Survey 1996/7. Jerusalem
CBS Central Bureau of Statistics (1999) Transportation File; 1995 Census. Jeru-
salem
Cervero R (1993) Ridership Impacts of Transit-Focused Development in Califor-
nia. Working Paper, Institute of Urban and Regional Development, University
of California, Berkeley
Cervero R (1996a) Jobs-Housing Balance Revisited: Trends and Impacts in the
San Francisco Bay Area. Journal of the American Planning Association
62:492511
Cervero R (1996b) Mixed Land Uses and Commuting: Evidence from the Ameri-
can Housing Survey. Transportation Research Part A 30:361377
Cervero R (2001) Efficient Urbanisation: Economic Performance and the Shape of
the Metropolis. Urban Studies 38:16511671
Cervero R, Kockelman K (1997) Travel Demand and the 3Ds: Density, Diversity
and Design. Transportation Research Part D 2:199219
Chatman DG (2003) How Density and Mixed Uses at the Workplace Affect Per-
sonal Commercial Travel and Commute Mode Choice. Transportation Re-
search Record 1831:193201
Dunphy RT, Fisher K (1996) Transportation, Congestion and Density: New In-
sights. Transportation Research Record, 1552:8996
Ewing R, Cervero R (2001) Travel and the Built Environment: A Synthesis.
Transportation Research Record 1780:87113
Ewing R, DeAnna M, Li SC (1996) Land Use Impacts on Trip Generation Rates.

Transportation Research Record 1518:16
Ferdman A (2005) Modelling Joint Development of Land Use and Light Rail
Transit Stations: The Case of Tel Aviv. MSc Thesis, Haifa: Technion Israel
Institute of Technology
Frank LD, Pivo G (1994) Impacts of Mixed Use and Density on Utilization of
Three Modes of Travel: Single-Occupant Vehicle, Transit, Walking. Trans-
portation Research Record 1466:4452
Giuliano G (1995) Land Use Impacts of Transportation Investments: Highway and
Transit. In: Hanson S (ed) The Geography of Urban Transportation,Guilford
Press, New York, pp 305341
Gordon P, Richardson HW (1997) Are Compact Cities a Desirable Planning
Goal? Journal of the American Planning Association 63:95106
Hess DB (2001) Effect of Free Parking on Commuter Mode Choice. Transporta-
tion Research 1735:3542
Kenworthy JR, Laube FB (1999a) An International Sourcebook of Automobile
Dependence in Cities 1969-1990. University Press of Colorado, Boulder
Kenworthy JR, Laube FB (1999b) Patterns of Automobile Dependence in Cities:
An International Overview of Key Physical and Economic Dimensions with
Some Implications for Urban Policy. Transportation Research Part A 33:691
723
Kumar A, Levinson D (1995) Chained Trips in Montgomery County, Maryland.
ITE Journal 65/5:2732
Miller EJ, Ibrahim A (1998) Urban Form and Vehicular Travel, Some Empirical
Findings. Transportation Research Record 1617:1827
NTA (2001) Technical Memo No. 8, Trip Generation Models. Tel Aviv Metro-
politan Mass Transit System: Travel Demand Forecasting Model (MIP), Tel
Aviv
Ortuzar J de Dios, Willumsen LG (1990) Modelling Transport. Wiley, Chichester
Parsons Brinckerhoff Quade and Douglas, Inc. (1996a) Transit and Urban Form,
Transit, Urban Form and the Building Environment: A Summary of Knowl-
edge. TCRP Report 16, Vol. 1, Part I, Transportation Research Board, Wash-
ington, DC
OSullivan S, Morall J (1996) Walking Distances to and from Light Rail Transit
Stations. Transportation Research Record 1538:19-26
Parsons Brinckerhoff Quade and Douglas, Inc. (1996b) Commuter and Light Rail
Transit Corridors: The Land Use Connection. TCRP Report 16, Vol. 1, Part
II, Transportation Research Board, Washington, DC
Pushkarev B, Zupan JM (1977) Public Transportation and Land Use Policy. Indi-
ana University Press, Bloomington
Rodenburg CA (2005) Measuring Benefits of Multifunctional Land Use. Ph.D.
Thesis in the Department of Spatial Economics, The Free University, Amster-
dam
Schimek P (1996) Household Motor Vehicle Ownership and Use: How Much
Does Residential Density Matter? Transportation Research Record 1552:120
125
Shefer D (2003( Economic Aspects of Multifunctional Land Use and Urban Re-
generation. In: Nijkamp P, Rodenburg CA, Vreeker R (eds) The Economics of
Multifunctional Land Use. Shaker Publishing, Maastricht
Shefer D, Degani R (1998) Gender Differences in Travel-to-Work Patterns after a
Move to the Suburbs: An Israeli Case Study. European Spatial Research and
Policy 5:6780
Stringham M (1982) Travel Behaviour Associated with Land Use Adjacent to
Rapid Transit Stations. ITE Journal 52/4:18-22
Sun X, Wilmot CG, Kasturi T (1998) Household Travel, Household Characteris-
tics and Land Use: An Empirical Study from the 1994 Portland Activity-
Based Travel Survey. Transportation Research Record 1617:1017
Untermann RK (1984) Characteristics of Walking. In: Accommodating the Pedes-
trian, Adapting Towns and Neighbourhoods for Walking and Bicycling, Van
Nostrand Reinhold, New York, pp 23-59
Wegener M, Frst F (1999) Land Use Transport Interaction: State of the Art.
Berichte aus dem Institut fr Raumplanung 46. Institut fr Raumplanung,
Universitt Dortmund, Dortmund
Zahavi Y (1979) The UMOT Project. Report DOT-RSPA-DPB-20-79-3, US De-
partment of Transportation/Ministry of Transport, Federal Republic of Ger-
many, Washington, DC/Bonn
Contributors
Shlomo Bekhor
Transport Research Institute
Technion Israel Institute of Technology
Haifa 32000
Israel
Luca Bertolini
Department of Geography, Planning and
International Development Studies
Faculty of Social and Behavioral Sciences
Universiteit van Amsterdam
Nieuwe Prinsengracht 130
1018 VZ Amsterdam
the Netherlands
Frank Bruinsma
Department of Spatial Economics
Free University
De Boelelaan 1105
1081 HV Amsterdam
the Netherlands
Ghebreegziabiher Debrezion
Free University
De Boelelaan 1105
1081 HV Amsterdam
the Netherlands
Carel Eijgenraam
CPB Netherlands Bureau for Economic Policy Analysis
P.O. Box 80510
2508 GM The Hague
the Netherlands
416 Contributors
Avigail Ferdman
Center for Urban and Regional Studies
Haifa 32000
Israel
Oskar Fridh
Division of Transport and Logistics
KTH
S-100 44 Stockholm
Sweden
Daniel Gat
Haifa 32000
Israel
Thomas de Graaff
Free University
De Boelelaan 1105
1081 HV Amsterdam
the Netherlands
Erik Louw
OTB Research Institute for Housing, Urban and
Mobility Studies
Delft University of Technology
Jaffalaan 9
2628 BX Delft
the Netherlands
Stan Majoor
1018 VZ Amsterdam
the Netherlands
Contributors 417
Bo-Lennart Nelldal
Division of Transport and Logistics
KTH
S-100 44 Stockholm
Sweden
Ioulia Ossokina
CPB Netherlands Bureau for Economic Policy Analysis
P.O. Box 80510
2508 GM The Hague
the Netherlands
Francesca Pagliara
Department of Transport Engineering
University of Naples Federico II
Via Claudio 21
80125 Naples
Italy
Enrica Papa
DiPiST
Department of Urban and Regional Planning
University of Naples Federico II
Piazzale Tecchio 80
80125 Naples
Italy
Gert-Joost Peek
Strategy & Marketing
ING Real Estate Development International
P.O. Box 90463
2509 LL The Hague
the Netherlands
Eric Pels
Free University
De Boelelaan 1105
1081 HV Amsterdam
the Netherlands
418 Contributors
Peter Pol
Department of Applied Economics
Erasmus University Rotterdam
P.O. Box 1738
3000 DR Rotterdam
the Netherlands
Hugo Priemus
Faculty of Technology, Policy and Management
P.O. Box 5015
2600 GA Delft
the Netherlands
Piet Rietveld
Free University
De Boelelaan 1105
1081 HV Amsterdam
the Netherlands
Caroline Rodenburg
Real Estate Advisory Services and
International Location Advisory Services
Ernst & Young
Euclideslaan 1
3584 BL Utrecht
the Netherlands
Dick Schuiling
1018 VZ Amsterdam
the Netherlands
Contributors 419
Daniel Shefer
Transport Research Institute
Haifa 32000
Israel
Jan Jacob Trip

OTB Research Institute for Housing, Urban and
Mobility Studies
Jaffalaan 9
2628 BX Delft
the Netherlands
Ron Vreeker
Free University
De Boelelaan 1105
1081 HV Amsterdam
the Netherlands
Bert van Wee

Faculty of Technology, Policy and Management
P.O. Box 5015
2600 GA Delft
the Netherlands
Jasper Willigers
Significance
Schipholweg 13-15
2316 XB Leiden
the Netherlands

Frank Bruinsma, Eric Pels, Piet Rietveld, Hugo Priemus, Bert Van Wee (Auth.), Dr. Frank Bruinsma, Dr. Eric Pels, Prof. Dr. Piet Rietveld, Prof. Dr. Hugo Priemus, Prof. Dr. Bert Van Wee (Eds.)-Railway

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Frank Bruinsma, Eric Pels, Piet Rietveld, Hugo Priemus, Bert Van Wee (Auth.), Dr. Frank Bruinsma, Dr. Eric Pels, Prof. Dr. Piet Rietveld, Prof. Dr. Hugo Priemus, Prof. Dr. Bert Van Wee (Eds.)-Railway

Uploaded by

Copyright:

Available Formats

Railway Development

Impacts on Urban Dynamics

Impacts on Urban Dynamics

ISBN 978-3-7908-1971-7 e-ISBN 978-3-7908-1972-4

c 2008 Physica-Verlag Heidelberg

programs. By this the scope of the research extended from a Dutch to a

Amsterdam July 2007

1 The impact of railway development on urban 1

Part A: Setting the scene

2 Urban dynamics and transport infrastructure: 15

3 Station areas as nodes and places in urban 35

4 HST stations and urban dynamics: Experiences 59

5 What makes a city: Urban quality in Euralille, 79

6 New Key Projects for station redevelopment in 101

7 A multidisciplinary approach of railway station 125

Part B: Evaluation studies

8 Ex ante evaluation of railway station 147

9 Multicriteria analysis of a high-speed railway 171

10 Cost-benefit analysis of railway station area 191

11 Measuring the WTP for shopping facilities 213

Part C: High-speed rail and urban dynamics

12 The impact of high-speed railway developments 237

13 The effect of railway stations on office space 265

14 Regional high-speed trains on the Svealand line: 295

15 Rail pricing and the supply of complementary 315

Part D: Light rail and urban dynamics

16 Rail system development and urban 337

17 Rail-transit and real estate values in a polycentric 359

18 Land use variables in trip generation models: 385

Frank Bruinsma, Eric Pels and Piet Rietveld

Department of Spatial Economics, Free University, Amsterdam,

Hugo Priemus and Bert van Wee

Faculty of Technology, Policy and Management, Delft University of

A new turn in the development of the railway system is observed at the

1.2 Historical review

According to Sjoberg (1960), pre-industrial cities in various areas of the

larities in location, structure and architecture. They were relatively small

came increasingly aware of the various harmful external effects of road

1.3 Current issues

1.4 Future perspectives

1.5 Outline of the book

1.5.1 Part A: Setting the scene

opportunity and on strengthening weak location factors of the station areas.

development potential in combination with the conditions determining

1.5.2 Part B: Evaluation studies

to pay techniques for the valuation of various shopping facilities at the

We have already mentioned that, instead of a new transport modality or

1.5.3 Part C: High-speed rail and urban dynamics

High-speed rail networks add a strong international dimension to railway

Apart from impacts on individual high-speed railway stations, the net-

1.5.4 Part D: Light rail and urban dynamics

Banister D (2002) Transport Planning. Spon, London

Faculty of Technology, Policy and Management, Delft University of

Both urban planning and infrastructure development are highly path-

ing the development of poly-nuclear urban regions (urban networks and

2.2 Some historical notes on the development of the

If we want to know what impact modern high-speed train (HST) stations

2.3 Towards a new dynamic in railway networks and

In most Western European countries nowadays railway companies are

2.4 Changes in modal split: The growing share of high-

Table 2.2. Development of passenger transport in the Netherlands, 1990-2002, se-

Although we devote our attention in this chapter to the redevelopment

Table 2.4 shows how a number of countries participated in high-speed

Table 2.6. High-speed lines under construction in the EU, (2004)