Professional Documents
Culture Documents
Railway Development
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
DOI 10.1007/978-3-7908-1972-4
Library of Congress Control Number: 2007936545
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
Frank Bruinsma
Eric Pels
Hugo Priemus
Piet Rietveld
Bert van Wee
Contents
Preface
Contributors 415
1 The impact of railway development on urban
dynamics
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
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
The impact of railway development on urban dynamics 5
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.
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.
6 Frank Bruinsma, Eric Pels, Hugo Priemus, Piet Rietveld and Bert van Wee
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
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
The impact of railway development on urban dynamics 7
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
The impact of railway development on urban dynamics 9
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
The impact of railway development on urban dynamics 11
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
Hugo Priemus
2.1 Introduction
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.
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.
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.
Urban dynamics and transport infrastructure: Towards greater synergy 21
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
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
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
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
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
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.
Urban dynamics and transport infrastructure: Towards greater synergy 31
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
Bertolini L, Spit T (1998) Cities on Rails: the Redevelopment of Railway Station
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
Urban dynamics and transport infrastructure: Towards greater synergy 33
Luca Bertolini
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.
Node
Unbalanced Stress
node
Balance
Dependence Unbalanced
place
Place
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
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.
Station areas as nodes and places in urban networks 41
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.
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
Station areas as nodes and places in urban networks 43
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
1
The author wishes to thank Angelique Klinkers for help in updating the origi-
nal analysis to 2005.
Station areas as nodes and places in urban networks 45
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
Station areas as nodes and places in urban networks 51
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.
Station areas as nodes and places in urban networks 53
3.4 Conclusions
References
Berg L van den, Pol P (1998) The European High-Speed Train and Urban Devel-
opment. Ashgate, Aldershot
Bertolini L (1996) Nodes and Places: Complexities of Railway Station Redevel-
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
Bertolini L, Spit T (1998) Cities on Rails: the Redevelopment of Railway Station
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.
Station areas as nodes and places in urban networks 57
Peter Pol
4.1 Introduction
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
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.
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
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.
Vision Leadership
Support Strategic
networks
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.
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
power over
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.
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
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.
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.
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
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).
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
HST stations and urban dynamics: experiences from four European cities 77
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
Pol PMJ (2002) A Renaissance of Stations, Railways and Cities: Economic Ef-
fects, Development Strategies and Organisational Issues of European High-
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
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
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
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.
What makes a city: urban quality 83
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.
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).
What makes a city: urban quality 85
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).
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.
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-
What makes a city: urban quality 87
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.
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
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
definition of
municipal plan-
functional
ning department
programme
architecture of
private
commercial
developers
real estate
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.
What makes a city: urban quality 91
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).
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.
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
What makes a city: urban quality 93
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.
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.
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.
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.
5.5 Conclusions
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
6.1 Introduction
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).
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
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-
New Key Projects for station redevelopment in the Netherlands 107
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?
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-
108 Stan Majoor and Dick Schuiling
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).
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.
110 Stan Majoor and Dick Schuiling
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.
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-
New Key Projects for station redevelopment in the Netherlands 111
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.
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,
112 Stan Majoor and Dick Schuiling
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.
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.
New Key Projects for station redevelopment in the Netherlands 113
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.
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?
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
114 Stan Majoor and Dick Schuiling
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-
New Key Projects for station redevelopment in the Netherlands 115
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)
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
116 Stan Majoor and Dick Schuiling
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
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
118 Stan Majoor and Dick Schuiling
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
New Key Projects for station redevelopment in the Netherlands 119
6.5 Conclusions
Acknowledgement
The authors would like to thank Marieke van Loon and Johan Olsthoorn
for gathering valuable data on the projects.
References
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
Erik Louw
7.1 Introduction
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.
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
A multidisciplinary approach of railway station development 129
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-
A multidisciplinary approach of railway station development 131
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.
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.
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.
A multidisciplinary approach of railway station development 133
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
134 Gert-Joost Peek and Erik Louw
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
A multidisciplinary approach of railway station development 139
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-
140 Gert-Joost Peek and Erik Louw
7.5 Conclusions
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
Piet Rietveld
8.1 Introduction
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
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
Ex ante evaluation of railway station development projects 151
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
152 Piet Rietveld and Bert van Wee
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.
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.
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.
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.
156 Piet Rietveld and Bert van Wee
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.
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.
2 At the time of the introduction of the Euro, 1 was equal to 2,20 Dutch Guild-
ers.
160 Piet Rietveld and Bert van Wee
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.
162 Piet Rietveld and Bert van Wee
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.
164 Piet Rietveld and Bert van Wee
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.
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,
Ex ante evaluation of railway station development projects 165
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.
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-
166 Piet Rietveld and Bert van Wee
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.
8.7 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-
Ex ante evaluation of railway station development projects 167
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
168 Piet Rietveld and Bert van Wee
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
Ex ante evaluation of railway station development projects 169
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
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
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
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
District Offices Dwellings Amenities Total
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
Table 9.3. The additional real estate development in the Dock alternative
District Offices Dwellings Amenities Total
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
MDCA of a high speed railway station area development project 179
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
District Offices Dwellings Amenities Total
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
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
1 Feeder tracks will be placed at the western and eastern sides of the terminal.
MDCA of a high speed railway station area development project 181
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.
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.
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.
MDCA of a high speed railway station area development project 183
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
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
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).
MDCA of a high speed railway station area development project 189
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
References
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
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-
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.
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
194 Carel Eijgenraam and Ioulia Ossokina
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
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
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).
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
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.)
CBA of railway station area development 197
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.
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.
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.
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.
CBA of railway station area development 199
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-
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
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
Residential estate
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.
202 Carel Eijgenraam and Ioulia Ossokina
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-
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.
CBA of railway station area development 203
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).
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
204 Carel Eijgenraam and Ioulia Ossokina
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.
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
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
Table 10.7 gives an overview of the external effects of the Dock project.
tion area of South/WTC. The described pattern of rents can be observed in the
South Axis area.
206 Carel Eijgenraam and Ioulia Ossokina
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
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-
CBA of railway station area development 207
10.7 Discussion
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.
CBA of railway station area development 209
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.
210 Carel Eijgenraam and Ioulia Ossokina
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
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
Caroline Rodenburg
11.1 Introduction
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.
216 Thomas de Graaf and Caroline Rodenburg
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.
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:
Measuring the WTP for shopping facilities around railway stations 219
[ + 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.
220 Thomas de Graaf and Caroline Rodenburg
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.
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.
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
Measuring the WTP for shopping facilities around railway stations 223
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.
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.
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
226 Thomas de Graaf and Caroline Rodenburg
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
Measuring the WTP for shopping facilities around railway stations 227
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
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.
Measuring the WTP for shopping facilities around railway stations 229
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),
230 Thomas de Graaf and Caroline Rodenburg
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
Measuring the WTP for shopping facilities around railway stations 231
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
Jasper Willigers
12.1 Introduction
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.
240 Jasper Willigers
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.
The impact of high speed railway developments on office locations 241
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.
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
The impact of high speed railway developments on office locations 243
j f ij = 0
exp[V ] > I
m
ijm max
Figure 12.5. Change in accessibility score towards employees due to the domes-
tic HSL South services
246 Jasper Willigers
Figure 12.7. Change in accessibility score towards business partners due to the
domestic HSL South services
The impact of high speed railway developments on office locations 247
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.
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.
248 Jasper Willigers
Figure 12.8. Effect of fare supplements for high-speed train services on potential
accessibility (gravity-type indicator) increase
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.
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
250 Jasper Willigers
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.
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
252 Jasper Willigers
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.
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
The impact of high speed railway developments on office locations 253
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
Reference With HSL South
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
Reference With HSL South
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.
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.
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-
The impact of high speed railway developments on office locations 255
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
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.
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
Reference With HSL South
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
Reference With HSL South
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
Reference With HSL South
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%)
Percentages between brackets are the changes in choice probability relative to the
reference case.
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%)
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-
260 Jasper Willigers
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.
The impact of high speed railway developments on office locations 261
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.
262 Jasper Willigers
12.8 Conclusions
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
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
264 Jasper Willigers
Ghebreegziabiher Debrezion
Jasper Willigers
13.1 Introduction
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.
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
The effect of railway stations on office space rents 269
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.
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-
270 Ghebreegziabiher Debrezion and Jasper Willigers
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)
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 .
272 Ghebreegziabiher Debrezion and Jasper Willigers
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
The effect of railway stations on office space rents 273
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
274 Ghebreegziabiher Debrezion and Jasper Willigers
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:
The effect of railway stations on office space rents 275
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.
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
276 Ghebreegziabiher Debrezion and Jasper Willigers
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.
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.
The effect of railway stations on office space rents 277
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
278 Ghebreegziabiher Debrezion and Jasper Willigers
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
The effect of railway stations on office space rents 279
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
280 Ghebreegziabiher Debrezion and Jasper Willigers
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.
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.
282 Ghebreegziabiher Debrezion and Jasper Willigers
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
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
The effect of railway stations on office space rents 283
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-
284 Ghebreegziabiher Debrezion and Jasper Willigers
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
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
286 Ghebreegziabiher Debrezion and Jasper Willigers
Appendix
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:
~
( RJ . ) 1 = [T ( WX )' (I X( X' X) 1 X' )( WX ) /
With
(e' e / N )]1
and T = tr( W 2 + W' W)
14.1 Introduction
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.
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
14.2 Methods
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.
Route choice
The Svealand line
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).
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%.
Regional high-speed trains on the Svealand line 299
Q
Change in supply
- new infrastructure Supply
Demand
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.
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
300 Oskar Fridh and Bo-Lennart Nelldal
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
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
Regional high-speed trains on the Svealand line 301
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%).
302 Oskar Fridh and Bo-Lennart Nelldal
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
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
304 Oskar Fridh and Bo-Lennart Nelldal
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
Regional high-speed trains on the Svealand line 305
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
306 Oskar Fridh and Bo-Lennart Nelldal
20% Proportion of
commuters
15% Eskilstuna out
Strngns out
10% Nykping out
5%
0%
1993 1994 1995 1996 1997 1998 1999 2000
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.
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
310 Oskar Fridh and Bo-Lennart Nelldal
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
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.
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.
312 Oskar Fridh and Bo-Lennart Nelldal
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
Regional high-speed trains on the Svealand line 313
References
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
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
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.
318 Eric Pels, Erik Louw and Jan Jacob Trip
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.
320 Eric Pels, Erik Louw and Jan Jacob Trip
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.
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).
Rail pricing and the supply of complementary commercial goods 321
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.
In this section we model the interaction between railway services and eco-
nomic activities as described above. We discuss four different organisa-
322 Eric Pels, Erik Louw and Jan Jacob Trip
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
15.3.1 Households
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.
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
2
pi = u (..) qi = + qi + q i + il y l (6)
l =1
324 Eric Pels, Erik Louw and Jan Jacob Trip
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 )
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
Rail pricing and the supply of complementary commercial goods 325
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
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:
In this case the railway company no longer takes into account the inter-
action effect between rail services and commercial activities. In terms of
326 Eric Pels, Erik Louw and Jan Jacob Trip
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
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.
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
Rail pricing and the supply of complementary commercial goods 327
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.
328 Eric Pels, Erik Louw and Jan Jacob Trip
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.
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 15.2 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
330 Eric Pels, Erik Louw and Jan Jacob Trip
-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 15.4 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).
Rail pricing and the supply of complementary commercial goods 331
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).
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
332 Eric Pels, Erik Louw and Jan Jacob Trip
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
Bertolini L (2000) Planning in the Borderless City: A Conceptualisation and an
Application to the Case of Station Area Redevelopment. Town Planning Re-
view 71:455475
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
Rail pricing and the supply of complementary commercial goods 333
Enrica Papa
Francesca Pagliara
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.
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.
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
340 Enrica Papa, Francesca Pagliara and Luca Bertolini
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
342 Enrica Papa, Francesca Pagliara and Luca Bertolini
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.
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.
344 Enrica Papa, Francesca Pagliara and Luca Bertolini
p
value2004 kt value1991kt
%valueit = 100 (6)
k =1 value1991k
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
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.
348 Enrica Papa, Francesca Pagliara and Luca Bertolini
Figure 16.4. Property value change of low quality housing in the station areas
1991-2004
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-
Rail system development and urban transformations 351
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
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
Rail system development and urban transformations 353
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
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
Rail system development and urban transformations 355
References
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
17.1 Introduction
17.2 Motivation
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
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)
S ( x, y ) R ( x , y ) = [ y T ( x ) ]
(2)
Z ( x, y ) = (1 ) [ y T ( x) ]
u * ( S , Z ) = (1 )1
[ y T ( x) ] (3)
R ( x, y )
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.
Rail-transit and real estate values in a polycentric city 363
y
S (0, y ) = S (0, y0 ) ; 0 < < 1 (5)
y0
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
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
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)
max = FV ( L + M ) = AL M 1 V ( L + M ) (10)
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
M (1 )
= (13)
L
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)
R ( x, y )
V ( x, y ) = (16)
k
1
1
A
( x, y ) = (1 )
R ( x , y ) (17)
k
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
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
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
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)
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
dist ( x, i )
t ( x, i ) = a (Ci ) 1 e dist ( x ,i ) + (22)
w0
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.
Rail-transit and real estate values in a polycentric city 371
$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
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
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
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
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)] 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-
Rail-transit and real estate values in a polycentric city 375
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.
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
Rail-transit and real estate values in a polycentric city 377
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:
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
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:
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
Rail-transit and real estate values in a polycentric city 379
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
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
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.
Rail-transit and real estate values in a polycentric city 381
Figure 17.11. The changing of the San Francisco skyline, looking toward Ember-
codero station
17.10 Conclusions
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
Rail-transit and real estate values in a polycentric city 383
Shlomo Bekhor
18.1 Introduction
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.
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.
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).
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).
Land use variables in trip generation models 389
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
390 Avigail Ferdman, Daniel Shefer and Shlomo Bekhor
18.3 Methodology
Based on the theoretical and empirical literature regarding land use ef-
fects on travel demand, three hypotheses are tested. The first states that
Land use variables in trip generation models 391
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.
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
392 Avigail Ferdman, Daniel Shefer and Shlomo Bekhor
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.
Land use variables in trip generation models 393
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.
394 Avigail Ferdman, Daniel Shefer and Shlomo Bekhor
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).
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
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
Figure 18.3. Comparison of the five scenarios, based on population density, job
density and dependency ratio
18.4 Results
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
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.
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
Land use variables in trip generation models 399
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.
400 Avigail Ferdman, Daniel Shefer and Shlomo Bekhor
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.
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.
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.
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.
Land use variables in trip generation models 401
Three models are presented in the case of HBO trips: the original NTA
model, a model with general zonal variables (service employees and total
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.
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
Land use variables in trip generation models 403
(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.
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.
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
406 Avigail Ferdman, Daniel Shefer and Shlomo Bekhor
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-
408 Avigail Ferdman, Daniel Shefer and Shlomo Bekhor
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.
Land use variables in trip generation models 409
Acknowledgement
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
412 Avigail Ferdman, Daniel Shefer and Shlomo Bekhor
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
Department of Spatial Economics
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
Technion Israel Institute of Technology
Haifa 32000
Israel
Oskar Fridh
Division of Transport and Logistics
KTH
S-100 44 Stockholm
Sweden
Daniel Gat
Center for Urban and Regional Studies
Technion Israel Institute of Technology
Haifa 32000
Israel
Thomas de Graaff
Department of Spatial Economics
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
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
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
Department of Spatial Economics
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
Delft University of Technology
P.O. Box 5015
2600 GA Delft
the Netherlands
Piet Rietveld
Department of Spatial Economics
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
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
Contributors 419
Daniel Shefer
Center for Urban and Regional Studies
Transport Research Institute
Technion Israel Institute of Technology
Haifa 32000
Israel
Ron Vreeker
Department of Spatial Economics
Free University
De Boelelaan 1105
1081 HV Amsterdam
the Netherlands
Jasper Willigers
Significance
Schipholweg 13-15
2316 XB Leiden
the Netherlands