policies for urbanism

Urban Studies, Vol. 41, No. 3, 579–603, March 2004

Policies for Urban Form and their Impact onTravel: The Netherlands Experience

T.SchwanenRegional Research Centre Utrecht (URU), Faculty of Geographical SciencesUtrecht University3508 TC UtrechtPO Box 80.115The Netherlands � 31 30 254 [email protected]

Tim Schwanen, Martin Dijst and Frans M. Dieleman

[Paper first received, November 2002; in final form, July 2003]

Summary. This paper documents an evaluation of the consequences of the Netherlands nationalphysical planning policy for an individual’s travel behaviour. Four components of this policy areconsidered: the concentrated decentralisation of the 1970s and 1980s; the strict compact-citypolicy of the 1980s and 1990s; the A-B-C location policy; and the spatial retailing policy. Usingdata from the 1998 Netherlands National Travel Survey, the article addresses the followingquestions. Did physical planning reduce the use of the private car and promote the use of publictransport together with cycling and walking? Did physical planning lead to shorter traveldistances and times? The analysis suggests that national spatial planning has been most effectivein retaining high shares of cycling and walking in the large and medium-sized cities, in particularfor shopping trips. In terms of travel time, however, spatial policy seems to have been lesssuccessful. The building of new towns and, more recently, the development of greenfieldneighbourhoods close to cities do not appear to have reduced commuting times. Alternativestrategies to promote the use of public transport, the bicycle and walking through the regulationof land use are discussed. Relaxing some of the present spatial planning controls is suggested toreduce car use and travel times.

1. Introduction

Fuelled by general concerns about environ-mental problems, over the past two decades,spatial planners have paid particular attentionto the impact of urban form on mode choiceand travel distance, especially for commut-ing. Compared with the US, where urbandevelopment is more often privately financedand less directed by national policies(Cervero, 1995), many European plannersand politicians have proposed or imple-mented measures to limit urban sprawl andpromote the development of compact urban

forms as a means of reducing energy con-sumption through transport (Hart, 1992;Williams et al. 2000). Other important rea-sons for influencing and controlling the de-velopment of urban forms have been thepreservation of agricultural land, open spaceand environmentally valuable areas, and thepromotion of investment in existing built-upareas to halt neighbourhood decline.

The Netherlands has a long history of theintervention by spatial planners in the devel-opment of urban forms. In the 1970s and

Tim Schwanen, Martin Dijst and Frans M. Dieleman are in the Urban and Regional Research Centre Utrecht (URU), Faculty ofGeosciences, Utrecht University, PO Box 80.115, 3508 TC Utrecht, The Netherlands. Fax: � 31 30 254 0604. E-mail:[email protected]; [email protected]; and [email protected]. The research reported here has been made possibleby grant 425–13–003 from the Netherlands National Science Foundation (NWO) to the Urban and Regional Research CentreUtrecht. Frans Dieleman was able to work on the theme of this article through a grant of the Royal Netherlands Academy of Artsand Sciences (KNAW) for the Netherlands Visiting Professorship at the University of Michigan, Ann Arbor.

0042-0980 Print/1360-063X On-line/04/030579–25 2004 The Editors of Urban StudiesDOI: 10.1080/0042098042000178690

TIM SCHWANEN ET AL.580

1980s, a policy of concentrated decentralis-ation of urban land use was implemented bydeveloping designated growth centres andprohibiting the growth of small rural settle-ments (Dieleman et al. 1999; Bontje, 2001).In the 1980s, a policy of compact urbangrowth was formulated; it has been imple-mented since the 1990s (Hayer and Zon-neveld, 2000). This policy was supplementedby policies of large investments in the re-newal of the urban housing stock and thestrict regulation of the location of retail fa-cilities with a ban on the development oflarge out-of-town shopping malls (Evers,2002). Additionally, the national governmenthas tried to channel new employment intonodes that were well served by public trans-port through the A-B-C location policy (Dijst,1997).

Because of planners’ intensive regulationof spatial development over a long period,the Netherlands provides an interesting casein which to evaluate the impact of nationalspatial planning policies on the travel behav-iour of individuals. This article documents anevaluation of the impact of some of thenational spatial policies on the developmentof urban form and through this on travelpatterns in the Netherlands. Which policieshave stimulated the use of public transportand of short trips? In what respect did thesepolicies fail to promote walking, cycling andthe use of public transport and to reducetravel times and distances, and what were thelikely causes of the failure? We support ourarguments with empirical data from the 1998Netherlands National Travel Survey, supple-mented by results from our own recent workon travel behaviour and urban form(Schwanen, Dieleman and Dijst, 2001, 2003a,2003b; Schwanen, Dijst and Dieleman, 2001,2002) and other studies on this theme.

The next section gives a summary of someof the main spatial policy measures of thepost-World-War II era in the Netherlands.Section 3 presents some empirical results onthe relationship between urban structure andtravel against the spatial policies imple-mented. In section 4, some conclusions aredrawn and some possible ways of making

spatial policies more effective in terms oftheir travel outcomes are discussed.

2. Physical Planning Controls on UrbanDynamics in the Netherlands

A whole range of policy measures has beendesigned and put into action to regulate ur-ban spatial dynamics in the Netherlands dur-ing the past three decades. We brieflysummarise spatial planning controls on urbanexpansion and the compact-city policy, in-vestments in urban renewal and attempts toinfluence the location choice of firms andretail planning.

In the Second Physical Planning Memo-randum (MVRO, 1966), the national govern-ment took a powerful stand against suburbansprawl that was at that time threatening toengulf the Green Heart of the Randstad Hol-land. This is the (relatively) open space sur-rounded by the major cities in the westernpart of the Netherlands (Figure 1; Faludi andvan der Valk, 1990; Dieleman et al., 1999).The report proposed channelling suburbani-sation into ‘concentrated decentralisation’.More specifically, new urban growth was tobe accommodated outside the existing citiesin a number of designated overspill centresreferred to as growth centres. The Nether-lands does not stand alone in this approach.Similar initiatives have been undertaken else-where in Europe—for example, in England,Sweden and the Paris region (Cervero,1995). This policy was put into practice inthe late 1970s and early 1980s. It was verysuccessful in terms of concentrating subur-ban residential growth into the designatedgrowth centres and prohibiting the growth ofrural villages in the Green Heart (Faludi andvan der Valk, 1990).

Nevertheless, the policy of concentrateddecentralisation of urban growth was broughtto an end during the 1980s (Dieleman et al.,1999; Hayer and Zonneveld, 2000). Inner-city decline became a major issue on thepolicy agenda. Many government officialsand academics blamed this decline on thepolicy of concentrated decentralisation(see—for example, MVROM, 1988). The

URBAN FORM AND TRAVEL 581

Figure 1. Growth centres and VINEX locations in the western part of the Netherlands.

policy did indeed foster an unprecedentedwave of suburbanisation in the Netherlands,leading to substantial income differences be-tween city centres and the suburbs (Dielemanand Wallet, 2003). A new policy of compacturban growth was developed and put intopractice. Under this new policy formulated inthe Fourth Physical Planning MemorandumExtra (MVROM, 1991) and now being imple-mented, urban growth is guided into(re)development locations within existing cit-ies (brownfield sites) and towards newgreenfield sites directly adjacent to the built-

up areas of the larger cities in the Netherlands(Figure 1). Leidsche Rijn—the greenfield de-velopment immediately to the west ofUtrecht—is the largest of these VINEX loca-tions. Here, some 30 000 dwellings and10 000 jobs are being created. The compact-city policy also promotes the construction ofrelatively dense new residential neighbour-hoods in suburban settlements that kept ex-panding and accommodating part of the urbangrowth outside the VINEX locations.

In the same period in which spatial plan-ning in the Netherlands changed track from


concentrating suburbanisation in growth cen-tres to compact urban redevelopment, thenational government embarked on an exten-sive programme of urban renewal. Largesubsidies were given to upgrade the qualityof the housing stock, particularly in the oldcores of the largest cities. Oskamp and col-leagues (2002) show how, in 1985, of all theDutch housing stock, 19 per cent was still inthe category of “in poor condition”, whereasthis share was reduced to 1 per cent in 2000.The major cities of Amsterdam, Rotterdam,The Hague and Utrecht have benefited mostfrom the urban renewal policy, which hashelped to upgrade the old private rental hous-ing stock in the urban cores. However, theurban renewal policy showed little concernfor retaining or attracting employment in thecore areas of the cities. This neglect hascertainly been one of the causes of the rapiddevelopment of the larger Dutch cities frommonocentric to polycentric functional units.The majority of jobs in many cities can nowbe found in new employment concentrationsoutside the old original city centres. Thepolicy of urban renewal has been reformu-lated in the Big Cities Policy that, notwith-standing its broader scope, has as one of itsmain aims the renewal and partial replace-ment of social housing estates built in the1950s and 1960s (van Kempen and Priemus,1999; van Kempen, 2000).

The Fourth Physical Planning Memoran-dum (MVROM, 1988) introduced a policyfor the location of firms, the A-B-C locationpolicy. This policy was explicitly formulatedto discourage the use of the private car and topromote the use of public transport togetherwith cycling and walking. A locations werecentrally located sites often close to mainrailway stations and hence readily accessibleby public transport. In contrast, B loca-tions—typically situated in developmentnodes outside the larger CBDs and the cen-tres of smaller urban settlements—were rea-sonably well connected to public transportand readily accessible by car. C locationsusually had very good motorway access.Typical examples were business zones in theurban fringe area or alongside motorways.

The intention of the policy was to guide newemployment and public services as much aspossible towards A and B locations.

The A-B-C policy turned out to be veryhard to implement (MVandW, 1999). Localgovernment authorities often gave higher pri-ority to attracting new employment than tothe locational requirements of the new firms.Furthermore, B locations were often less ac-cessible by public transport than had beenpredicted. Most importantly, however, thegrowth in employment in the office sectorwas much larger than had been planned inthe 1990s and could not be accommodatedfully in A and B locations as originallyplanned. As a result, the largest employmentgrowth in the 1990s occurred at C locations.The location policy has been modified as aresult of its malfunction and changes in pol-icy aims. The main aims of this renewed andmore broadly defined policy are not only toencourage the optimal use of different trans-port systems, as in the ‘old’ A-B-C policy,but also to encourage economic develop-ment, spatial quality and the quality of theliving environment (MVROM, 2001).

Finally, it is pertinent to discuss the veryspecific position taken in the Netherlandswith regard to retail planning (Evers, 2002).One of the salient features of the deconcen-tration of urban land uses in many countriesis the growth of out-of-town hypermarketsand shopping malls. In the Netherlands na-tional legislation was enacted to prohibitsuch developments in 1973, because the es-tablishment of out-of-town shopping mallswas seen as a threat to the vitality of towncentres and likely to generate extensive caruse. Netherlands retail policy has been highlyeffective in reaching its objectives. Manyshops are still located within the built-up areaof cities and towns, and within walking andcycling distance for local residents (Evers,2002). Nowadays, however, traditional retailplanning is under heavy pressure. In the cur-rent era of deregulation and privatisation, thepolicy is being blamed for curtailing retailproductivity and competitiveness and cre-ating barriers for new firms to enter the retailmarket.


In short, despite the failure of the A-B-Clocation policy, spatial planning has beenfairly successful during the past few decadesin guiding residential and retailing develop-ments towards the locations that had beenplanned. This success can be considered asubstantial achievement, given thedifficulties often encountered in implement-ing policies of compact urban growth andcontaining urban sprawl (Dieleman et al.,1999). The question remains, however, towhat extent these spatial policies have beensuccessful in terms of their travel-related ob-jectives. Have the policies led to largershares for public transport and walking andcycling? Or to shorter travel distances andtimes?

3. Physical Planning and Travel Behav-iour: A Confrontation

Using the 1998 Netherlands National TravelSurvey (NTS), we have analysed theinfluence of urban form on travel behaviourin the Netherlands. One-day travel-diary datafrom male and female heads of householdshave been used. For these individuals, infor-mation is available about all their trips on asingle day together with their socioeconomicand demographic situation (Statistics Nether-lands, 1999).

The NTS also contains information abouta household’s residential location at the levelof the municipality (n � 550 in 1998). Twocategorisations of these municipalities havebeen used to represent urban form. The firstclassifies the degree of urbanisation of theresidential environment into six categories(Figure 2); the second distinguishes fourtypes of daily urban systems: one monocen-tric (centralised) and three polycentric forms(Figure 3). Further details are provided inAppendix 1. All descriptive results presentedin the tables are weighted so that they arerepresentative for the whole Dutch popu-lation. In our evaluation of Netherlands na-tional spatial planning policy of the past fewdecades, we focus on three main dimensionsof travel behaviour. Sub-section 3.1 reportsan assessment of how spatial policies and

urban forms may have influenced modalchoice, while 3.2 discusses travel distanceand time. The focus is not only on commut-ing, but also on shopping trips and to a lesserdegree on leisure travel.

3.1 Modal Choice

Attempts to influence modal choice throughspatial planning are usually based on thepremise that urban compactness induceshigher shares for public transport, cyclingand walking at the expense of the private car.The concept of compactness comprises sev-eral interrelated urban form dimensions: highdensity, a land-use mix and short distances tothe urban core and suburban concentrationsof employment or retailing.

In line with the academic literature (Ca-magni et al., 2002; Cervero, 1996a, 1996b,2002; Frank and Pivo, 1994), we find that caruse for commuting falls and the share forpublic transport rises with the degree of ur-banisation. What is peculiar to the Nether-lands is that so many commuters travel bybicycle or on foot. Ranking residential envi-ronments within and outside the RandstadHolland from high to low urbanisation showsthat the levels of cycling and walking as wellas bus/tram/metro use fall with decreasinglevels of urbanisation, whereas the share ofthe car-driver mode rises (Table 1). Thelarger share for mass transit reflects the factthat higher population densities and levels ofland-use mixing facilitate more extensivepublic transport systems. The greater use ofslow modes of transport—cycling and walk-ing—is a result of the fact that, in the moreurbanised environments, more activity loca-tions can be accessed within acceptablewalking and cycling (time) limits. The effectof the level of urbanisation on modal choiceremains strong after account is taken of per-sonal and household attributes; after carownership, the level of urbanisation is themost important determinant of modal choice(Dieleman et al., 2002). Based on thesestatistics, we may conclude that the policiesof redeveloping brownfield sites, urban re-

584 TIM SCHWANEN ET AL.

Tab

le1.

Mod

alsp

litby

trip

purp

ose

and

leve

lof

urba

nisa

tion

inth

eN

ethe

rlan

ds,

1998

.

Ran

dsta

dR

est

ofth

eN

ethe

rlan

ds

Thr

eela

rge

Med

ium

-G

row

thR

ands

tad,

Mor

eL

ess

citie

ssi

zed

citie

sSu

burb

sce

ntre

sto

tal

urba

nise

dur

bani

sed

Per

cent

age

ofco

mm

uter

sus

ing

Car

driv

er37

.444

.460

.856

.850

.652

.764

.9C

yclin

g31

.229

.522

.418

.724

.532

.122

.6W

alki

ng8.

88.

14.

75.

06.

65.

24.

0T

rain

11.0

14.5

4.9

9.2

9.1

4.5

1.9

Bus

/tram

/met

ro14

.94.

83.

77.

97.

92.

31.

6

Per

cent

age

ofsh

oppe

rsus

ing

Car

driv

er25

.628

.444

.043

.935

.640

.352

.4C

yclin

g26

.438

.830

.426

.930

.533

.830

.0W

alki

ng41

.034

.322

.926

.531

.023

.614

.7T

rain

1.5

2.1

1.5

1.6

1.6

0.9

0.3

Bus

/tram

/met

ro12

.22.

91.

94.

25.

51.

80.

9


Figure 2. Classification of municipalities in the Netherlands: degree of urbanisation.

newal and upgrading the inner-city housingstock have contributed to a modal split forcommuting in the large and medium cities inwhich walking, cycling and the use of publictransport predominate (Table 1).

In terms of influencing the modal split forshopping trips, the policies of creating com-pact cities and retail planning seem to havebeen even more successful. As Table 1 indi-cates, the importance of walking and cyclingis even more pronounced for shopping thanfor commuting. Of the people engaging in

shopping in the three large cities and themedium-sized cities of the Randstad Hol-land, no less than 41 per cent and 34 per centrespectively travel on foot. These percent-ages stand in sharp contrast with the mere 14per cent for the least urbanised environmentsoutside the Randstad Holland. Furthermore,in the Netherlands many people use a bicyclefor shopping. In all residential environments,over a quarter of the population cycle to theshops; this percentage is highest in the me-dium-sized cities and more urbanised munic-


Figure 3. Classification of municipalities in the Netherlands: urban structure. Source: Based on Van derLaan (1998).

ipalities outside the Randstad Holland. Asurvey of the relevant academic literatureindicates that these high shares of the slowmodes of transport are exceptional. Whileelsewhere the bicycle is rarely used for shop-ping, particularly in the US, there is someevidence that in some high-density, mixedneighbourhoods in the UK and the US sharesfor walking do tend to mirror those for thelargest cities in the Netherlands. In general,however, the level of car use for shoppingseems to be much higher in urban areas in

these countries (see—for example, Frank andPivo, 1994; Handy and Clifton, 2001; Vanand Senior, 2000). Credit for this lower levelof car use in the Netherlands can undoubt-edly be attributed to the policies of stimulat-ing the development of compact and diverseresidential neighbourhoods and limiting re-tail sprawl.

While the strict compact-city policy of thepast decade has had beneficial effects interms of modal split, the impact of its prede-cessor—the policy of concentrated decentra-


lisation—seems to have been more mixed.As explained in section 2, in the 1970s andearly 1980s, national spatial planning fo-cused on the development of growth centresto accommodate suburban growth. A com-parison of the modal split in the growthcentres and the suburbs of Randstad Holland,which represent a more dispersed decentra-lisation pattern, shows that the inhabitants ofthe growth centres are somewhat less depen-dent on the private car for commuting. Thelevel of commuting by train is higher,reflecting the fact that almost all growth cen-tres are directly connected by rail to the mainemployment centres. However, the level ofbicycle use is lower in the growth centres,which seems to be related to the qualitativemismatch between residents and employmentin many new towns (section 3.2). With re-gard to shopping, the results again indicatethat the relatively good public transport con-nections between the new towns and themain urban centres lead to greater use of thebus/tram/metro. The share of the private carfor shopping trips in the growth centres is aslarge as in the suburbs of Randstad Holland.Although conclusions about the effectivenessof spatial policy should be drawn from theseresults with caution, it seems that the impactof the policy of concentrated decentralisationon modal split is relatively limited.

There are also several aspects of spatialplanning policy that seem to have had lessfavourable consequences for modal choice.Focusing again on the more recent policy ofconcentrating residential development withinand adjacent to existing built-up areas, it canbe argued that there are second-order effectsof a compact-city policy (Priemus et al.,2001). The more compact the city, the morelikely are some urban functions to relocate tothe suburbs. It may well be that the ‘cram-ming’ (Counsell, 2001) of housing into thebuilt-up and redeveloped parts of Dutch cit-ies has accelerated the deconcentration ofemployment towards C locations situatednear the motorways (section 2). Such loca-tion behaviour will almost certainly haveresulted in greater car use, because theselocations are not well served by public trans-

port and neither are they suited to bicycletravel.

In addition, although in the more ur-banised environments the car is partly substi-tuted for by public transport, its higher sharesare sometimes reached at the expense ofcycling. Comparison of the medium-sizedand three large cities for shopping in Table 1provides some evidence for the substitutionof public transport for cycling. In otherwords, higher-quality public transport at thelocal level seems to compete not only withthe private car, but also with the bicycle(Dieleman et al., 2002). Similar evidencewas found for leisure travel by Dutch seniorcitizens (people aged 50 or older and livingindependently in a one- or two-person house-hold) (Schwanen, Dijst and Dieleman, 2001)and in a comparative study of commutingbehaviour in 11 European cities (Schwanen,2002).

Furthermore, while the development ofhousing on brownfield sites within the exist-ing urban fabric may lead to less car travel, itremains to be seen whether the developmentof compact greenfield sites adjacent to theexisting built-up areas of cities reduces theprobability of using a car. Some preliminaryevidence is available for people who haverecently moved to the newly built VINEXneighbourhood of Leidsche Rijn situated onthe fringe of the city of Utrecht (Figure 1).Dijst and colleagues (2000) found that, aftermoving to Leidsche Rijn, a shift from anyother means of transport to the private caroccurred more often than the reverse. Relo-cating to this greenfield development is thusassociated with higher levels of car depen-dence. This may be attributed to the fact that,in order to minimise operational deficits,most public transport services in these loca-tions are only made available after most resi-dential units have been completed. Inaddition, Leidsche Rijn has very good accessto motorways and the residents are to a largeextent young, two-earner couples and famil-ies with a relatively high income. Thewomen in such households in particular relyon the private car for commuting to ease thecombination of paid labour with household


maintenance activities (Hjorthol, 2000;Schwanen et al., 2003b). Furthermore, it ap-pears that policy-makers have incorrectly as-sumed that the new residents of LeidscheRijn would be employed in the city ofUtrecht and its surroundings. However, sinceLeidsche Rijn provides excellent access tojobs located in many parts of the RandstadHolland (van Ham et al., 2001), many resi-dents commute by private car to work loca-tions outside the Utrecht area.

3.2 Travel Distance and Time

Since the amount of energy used for trans-port depends on distance travelled as well asmode choice, many previous studies havefocused on the impact of urban form ontravel distance. We have therefore also as-sessed the impact of spatial planning policieson the distance people travel per day forcommuting, shopping and leisure activities.However, since “time use is … a criticalelement for a thorough understanding oftravel behaviour” (Kitamura et al., 1997,p. 171), we have also considered how timespent on travelling varies with the character-istics of the built environment.

With respect to distance travelled, manyprevious studies have indicated that a higherdegree of urbanisation is associated withshorter travel distances in general as well asdistance travelled by private car (Næss et al.,1995; Kenworthy and Laube, 1999; Kockel-man, 1997). Compact urban structures thusresult in travel patterns that are efficient inthe sense that the distances travelled areshorter than in lower-density or more dis-persed urban areas (Camagni et al., 2002).Evidence for shorter daily travel distanceswith a higher level of urbanisation is alsoborne out for shopping in Table 2. For com-muting, the average distance travelled as acar driver is somewhat lower in the threelarge cities of Amsterdam, Rotterdam andThe Hague and in the more urbanised munic-ipalities outside Randstad Holland than else-where.

The relationship between degree of urbani-sation and travel time appears to be more

complicated. For the US, Levinson and Ku-mar (1997) found that commuting times byprivate car first decrease with populationdensity, but then start to rise with the numberof persons per square mile above some den-sity threshold. In the Netherlands, traveltimes for car drivers tend to rise somewhatwith the level of urbanisation for shopping(Table 2) and leisure (not shown due to spacelimitations). For commuting, the results areagain a little more complicated. Neverthe-less, the average for the medium-sized citiesis higher than for the suburbs. In addition,the average for the Randstad Holland as awhole is higher than that for the rest of theNetherlands. Furthermore, the difference be-tween the more urbanised and less urbanisedenvironments outside the Randstad Hollandis much smaller than in the case of commut-ing distance.

Because the results in Table 2 may partlyreflect differences in population compositionand the variation around the mean values isconsiderable, we have estimated four multi-level regression models with sociodemo-graphic factors and the typology ofresidential environments as independent vari-ables. Multilevel regression modelling is anelaboration of standard ordinary least squaresregression. Compared with standard re-gression modelling, parameter estimates andt-statistics especially for variables at thehigher levels of analysis (the residential mu-nicipality) are improved, because residualvariance is not captured by one but by sev-eral variance terms. Technical details areprovided in Appendix 2. The sociodemo-graphic variables included in the analysisreflect socioeconomic status (household in-come, education), transport availability (caravailability), life-cycle (age, household type)and role within the household (gender, inter-actions of gender and household type). Theywere selected on the basis of extensivepreliminary analysis. Their estimatedcoefficients (Table 3) are consistent with ex-pectations and previous analyses. For in-stance, in line with numerous other studies(Johnston-Anumonwo, 1992; Turner andNiemeier, 1997), our models show that gen-

589URBAN FORM AND TRAVEL

Tab

le2.

Dai

lytr

avel

dist

ance

(km

)an

dtim

e(m

inut

es)

per

pers

onas

aca

rdr

iver

,by

trip

purp

ose

and

leve

lof

urba

nisa

tion

ofth

ere

side

ntia

lm

unic

ipal

ityin

the

Net

herl

ands

,19

98

Com

mut

ing

Shop

ping

Dis

tanc

eT

ime

Dis

tanc

eT

ime

Stan

dard

Stan

dard

Stan

dard

Stan

dard

Mea

nde

viat

ion

Mea

nde

viat

ion

Mea

nde

viat

ion

Mea

nde

viat

ion

Ran

dsta

dH

olla

ndT

hree

larg

eci

ties

39.6

44.2

53.2

40.8

14.9

21.0

31.5

32.6

Med

ium

-siz

edci

ties

43.9

50.6

56.6

45.4

15.5

21.4

29.6

23.5

Subu

rbs

41.9

45.9

52.5

40.8

15.7

21.9

28.1

25.0

Gro

wth

cent

res

43.9

45.2

56.1

43.5

16.9

23.8

29.5

25.7

Tot

al42

.046

.253

.842

.015

.721

.929

.326

.8

Res

tof

the

Net

herl

ands

Mor

eur

bani

sed

38.7

49.0

48.6

42.7

13.8

18.3

27.2

23.0

Les

sur

bani

sed

41.4

49.1

49.5

42.9

16.3

26.0

27.4

24.3

Not

e:D

ata

are

aver

ages

for

peop

lew

hodr

ive

aca

rto

wor

kor

for

shop

ping

.

590 TIM SCHWANEN ET AL.

Tab

le3.

Mul

tilev

elre

gres

sion

mod

els

for

tota

lda

ilytr

avel

dist

ance

and

time

asa

car

driv

erfo

rco

mm

utin

gan

dsh

oppi

ng

Com

mut

ing

Shop

ping

Dis

tanc

eT

ime

Dis

tanc

eT

ime

Coe

ffici

ent

T-s

tatis

ticC

oeffi

cien

tT

-sta

tistic

Coe

ffici

ent

T-s

tatis

ticC

oeffi

cien

tT

-sta

tistic

Fix

edpa

rtC

onst

ant

2.98

942

.78

3.27

161

.98

1.97

534

.93

2.76

181

.95

Soci

odem

ogra

phic

fact

ors

Age

(in

year

s)�

0.00

2�

1.69

0.00

44.

930.

004

4.38

0.00

711

.64

Fem

ale

�0.

535

�17

.54

�0.

293

�13

.42

�0.

266

�11

.23

�0.

141

�8.

84Si

ngle

fem

ale,

wor

king

�12

hpe

rw

eek

0.17

13.

030.

113

2.35

0.25

84.

010.

237

3.72

Sing

lefe

mal

e,w

orki

ng0–

12h

per

wee

k1.

180

3.73

0.81

93.

51Fe

mal

ein

two-

wor

ker

coup

le0.

239

6.32

0.13

34.

86Fe

mal

ein

one-

wor

ker

coup

le0.

347

4.60

�0.

099

�1.

32Fe

mal

ein

one-

wor

ker

fam

ily�

0.10

7�

2.01

Sing

lepe

rson

,w

orki

ng�

12h

per

wee

k�

0.05

9�

1.90

�0.

122

�2.

71Si

ngle

pers

on,

wor

king

0–12

hpe

rw

eek

�0.

370

�1.

80�

0.32

7�

2.13

Tw

o-w

orke

rco

uple

0.16

75.

02O

ne-w

orke

rco

uple

0.19

53.

60Z

ero-

wor

ker

coup

le0.

111

2.58

Tw

o-w

orke

rfa

mily

�0.

081

�3.

71O

ne-w

orke

rfa

mily

0.07

52.

890.

041

2.10

0.10

42.

38Fa

mily

with

adul

t(s)

wor

king

0–12

hpe

rw

eek

0.29

01.

710.

273

2.17

Oth

erho

useh

old

type

s0.

059

2.52

0.04

12.

420.

070

2.33

Car

avai

labi

lity

inde

x0.

229

8.94

0.13

26.

910.

114

3.74

Em

ploy

edfo

r�

30h

per

wee

k0.

237

8.40

0.17

38.

31L

owed

ucat

iona

lat

tain

men

t�

0.22

8�

10.0

9�

0.14

2�

8.49

�0.

028

�1.

34M

ediu

med

ucat

iona

lat

tain

men

t�

0.14

1�

6.78

�0.

107

�6.

95H

ouse

hold

inco

me

0.00

35.

460.

002

4.07

Hou

seho

ldin

com

eis

unkn

own

�0.

022

�0.

850.

027

0.02


Res

iden

tial

envi

ronm

ent

Ran

dsta

dH

olla

nd,

thre

ebi

gci

ties

�0.

136

�2.

550.

069

2.05

�0.

122

2.37

0.12

13.

71R

ands

tad

Hol

land

,m

ediu

m-s

ized

citie

s�

0.01

7�

0.29

0.12

63.

51�

0.07

4�

1.37

0.11

23.

45R

ands

tad

Hol

land

,su

burb

s�

0.00

8�

0.27

0.05

72.

91�

0.05

0�

1.64

0.01

40.

69R

ands

tad

Hol

land

,gr

owth

cent

res

0.07

21.

300.

139

4.11

�0.

033

�0.

640.

062

1.98

Res

tN

ethe

rlan

ds,

mor

eur

bani

sed

�0.

155

�5.

01�

0.03

5�

1.77

�0.

123

�4.

150.

018

0.96

Res

tN

ethe

rlan

ds,

less

urba

nise

d0

00

0

Ran

dom

part

Var

ianc

ete

rmfo

rin

divi

dual

leve

l(

2 e0)

1.07

385

.90

0.58

986

.05

1.04

124

.07

0.57

024

.07

Var

ianc

ete

rmfo

rho

useh

old

leve

l(

2 u0)

0.11

52.

760.

674

29.5

4V

aria

nce

term

for

mun

icip

alle

vel

(2 v0

)0.

024

6.09

0.00

63.

700.

011

3.00

0.00

00.

00

Num

ber

ofob

serv

atio

ns15

215

1521

512

034

1203

4�

2lo

glik

elih

ood

atco

nsta

nt45

819.

836

277.

136

347.

629

117.

7�

2lo

glik

elih

ood

atco

nver

genc

e44

492.

935

260.

035

985.

328

713.

7C

hi-s

quar

edm

odel

impr

ovem

ent

132

7.0

101

7.2

362.

240

3.9

Dep

ende

ntva

riab

leL

N(d

aily

trav

elL

N(d

aily

trav

eltim

eL

N(d

aily

trav

eldi

stan

ceL

N(d

aily

trav

eltim

edi

stan

ceas

aca

rdr

iver

asa

car

driv

erfo

ras

aca

rdr

iver

for

asa

car

driv

erfo

rfo

rco

mm

utin

g)co

mm

utin

g)sh

oppi

ng)

shop

ping

)


der is an important determinant of commut-ing distance and time, but its effect dependson household structure. Table 3 also showsthat travel patterns for shopping are to alesser degree dependent on sociodemo-graphic situation than commuting travel.

Regarding the residential environment, theestimated coefficients largely corroborate theobservations on the basis of Table 2. Both forcommuting and shopping travel, distance as acar driver tends to fall with the degree ofurbanisation within and outside RandstadHolland. For travel time, the conclusions arealmost opposite: travel time rises with urban-isation level for shopping, while for commut-ing time the highest coefficients areassociated with the growth centres and themedium-sized cities of the Randstad Hol-land. Only small differences exist in traveltime between more and less urbanised mu-nicipalities outside Randstad Holland.

For travel modes other than the privatecar, the association between the degree ofurbanisation and travel time is similar to thatbetween distance and urbanisation level.Whereas there is no discernible relationshipbetween time spent walking to shops andresidential environment, the average distanceand time travelled by bicycle are greater inmore urbanised environments (Table 4). Thesame is true for commuting and leisure travel(Dieleman et al., 2002; Schwanen, Dijst andDieleman, 2002). In contrast, travel distanceand time for public transport modes tend tobe lower in more urbanised environments.This finding is related to the fact that, in thethree large cities and medium-sized cities,more people make use of public transportsystems at the city or daily urban systemlevel. Such systems operate at higher fre-quencies here than elsewhere in the Nether-lands.

What conclusions about the effectivenessof national spatial planning policies can bedrawn from Tables 2–4? A first conclusion isthat, despite the fact that the commute dis-tance and time for car drivers vary with thetype of residential environment, the import-ance of urban form in influencing commutingbehaviour should not be overestimated. The

variation within residential environmenttypes is large, as the standard deviations inTable 2 show. Furthermore, the varianceterms estimated in the multilevel models inTable 3 can be used to determine the share ofthe municipal level in the total variation in adependent variable. These shares can be cal-culated by dividing the estimate for the resi-dential level in a model with only a constantincluded as an explanatory variable by thesum of all variance estimates in such a model(Snijders and Bosker, 1999). For the fourvariables analysed here, these shares are: 2.6per cent (commuting distance as a cardriver); 1.7 per cent (commuting time as acar driver); 1.0 per cent (distance as a cardriver for shopping); 0.4 per cent (travel timeas a car driver for shopping). These resultsclearly illustrate that variations between trav-ellers within residential municipalities aremuch larger than differences between suchspatial units. Similar conclusions can befound in Herz (1983) and Weber and Kwan(2003).

The fact that travel times tend to be longerin more urbanised environments is both apositive and a negative outcome of nationalphysical planning policy. On the one hand,the slightly greater travel times in more ur-banised environments depress car use. Thus,for shopping travel we might infer that thecombination of strict retail planning and theconcentration of housing development withinor adjacent to existing built-up areas hascontributed to travel patterns that are lessdetrimental from an environmental perspec-tive than would have been the case had manylarge-scale shopping malls been developed inurban fringe areas. On the other hand, thelonger travel times in the more urbanisedenvironments may have important second-order effects (Priemus et al., 2001). It hasrepeatedly been shown that households tendto favour residential locations providinggood access, or limited travel time, to em-ployment locations (Brun and Fagnani, 1994;McDowell, 1997) as well as shopping oppor-tunities (Lerman, 1976) and recreational ac-tivities (Guo and Bhat, 2002). Manyhouseholds might thus prefer residential


Tab

le4.

Ave

rage

daily

trav

eldi

stan

ce(k

m)

and

time

(min

utes

)pe

rpe

rson

for

shop

ping

activ

ities

,by

tran

spor

tm

ode

othe

rth

anth

epr

ivat

eca

ran

dle

vel

ofur

bani

satio

nin

the

Net

herl

ands

,19

98

Cyc

ling

Wal

king

Tra

inB

us/tr

am/m

etro

Dis

tanc

eT

ime

Dis

tanc

eT

ime

Dis

tanc

eT

ime

Dis

tanc

eT

ime

Stan

dard

Stan

dard

Stan

dard

Stan

dard

Stan

dard

Stan

dard

Stan

dard

Stan

dard

Mea

nde

viat

ion

Mea

nde

viat

ion

Mea

nde

viat

ion

Mea

nde

viat

ion

Mea

nde

viat

ion

Mea

nde

viat

ion

Mea

nde

viat

ion

Mea

nde

viat

ion

Ran

dsta

dH

olla

ndT

hree

larg

eci

ties

4.6

4.6

25.2

23.5

1.7

1.7

20.7

20.3

32.9

22.7

73.9

43.8

10.0

7.9

46.6

42.3

Med

ium

-siz

edci

ties

4.4

3.6

23.3

18.6

1.7

1.7

21.1

21.2

52.5

35.9

80.1

38.7

12.9

12.4

51.4

31.1

Subu

rbs

4.2

4.3

21.5

19.1

1.8

1.7

21.6

21.5

60.6

60.4

89.6

59.3

18.4

13.3

50.7

32.5

Gro

wth

cent

res

4.4

4.0

22.2

16.6

1.7

1.7

20.3

25.4

38.6

27.5

61.6

26.5

20.6

17.4

54.4

36.8

Tot

al4.

44.

223

.020

.01.

71.

721

.021

.448

.144

.279

.347

.912

.411

.248

.439

.6

Res

tof

the

Net

herl

ands

Mor

eur

bani

sed

4.4

3.9

22.4

19.5

1.7

1.7

20.8

23.3

93.3

98.0

107.

282

.916

.026

.944

.445

.9L

ess

urba

nise

d3.

93.

919

.917

.31.

51.

520

.827

.949

.436

.269

.935

.322

.016

.447

.928

.5

Not

e:D

ata

are

aver

ages

for

peop

lew

hogo

shop

ping

and

use

agi

ven

mod

e.


locations where travel to relevant activitiestakes less time and therefore choose to relo-cate to a lower-density environment if theywere given the opportunity to do so. In otherwords, a further suburbanisation of house-holds may be one of the longer-term conse-quences of the compact-city policy and thismay offset some of the short-term benefits,such as fewer car trips and car kilometrestravelled.

In short, we may conclude that spatialpolicies have probably only made a modestcontribution to an increase in travelefficiency in terms of travel distance andtime in certain parts of the urban system or intravel behaviour in the country as a whole.Some additional empirical evidence can begiven to support this conjecture.

First, while comparisons of travel patternsin urban areas in different countries aredifficult because of large differences in typeof analysis and level of detail, some empiri-cal evidence about travel distance and time isavailable to put the above results in an inter-national perspective. This material suggeststhat, notwithstanding its strong planning tra-dition, commuting patterns in the Nether-lands are not exceptionally efficient. A studythat compared average commuting distancesand times across metropolitan regions re-vealed that Amsterdam takes an intermediateposition in Europe (Schwanen, 2002). Forthe country as a whole, the evidence points ata less rather than a more efficient commute.The Dutch commute on average 46 minutesper day, substantially more than in manyother countries in Europe (NRC Handels-blad, 2001). This comparatively long timeseems to result from a combination of longercommute distances (Jansen, 1993) and moresevere road congestion (Jansen et al., 2002;Bovy and Salomon, 2002). However, theNetherlands do differ from other west Eu-ropean countries in terms of the share of totaldistance that is covered by slow transportmodes. Whereas the share of walking andbicycling in the total distance was 4 per centin 1990 for western Europe as a whole (Hart,2001), this share was 12 per cent for theNetherlands (Dieleman et al., 1999). Expla-

nations for this difference are, for instance,the pro-public transport and bicycle planningand less positive attitudes towards the car inthe Netherlands.

Secondly, the policy of concentrated de-centralisation does not seem to have resultedin more efficient commuting patterns. Tables2 and 3 indicate that, for car drivers, totalcommute distances and times are larger ingrowth centres than in other suburban partsof the Randstad Holland. These high aver-ages seem to be primarily related to thequantitative and qualitative mismatch be-tween workers and employment in the newtowns. In quantitative terms, the number ofjobs has long lagged behind the number ofhouses in these growth centres, turning theminto dormitory towns (van der Laan, 1998).Over time, however, the number of jobs inthese settlements has increased. Neverthe-less, many workers in growth centres resideelsewhere and many residents of these newtowns work elsewhere, suggesting that theskills and preferences of growth-centre resi-dents correspond less with the available jobsthan in other suburban communities (qualita-tive mismatch). For shopping, the differencesbetween growth centres and other suburbanlocalities are less pronounced, indicating thatover time retailing supply in the new townshas reached a mature state, comparable withthat in other communities.

Thirdly, a comparison of commuting be-haviour in monocentric and polycentric areasof the daily urban systems (DUS) in theNetherlands provides little evidence of spa-tial planning efforts having created efficienttravel patterns. Rather, the reverse seems tobe the case. It has often been argued thatpolycentric urban areas have developed be-cause of households’ and employers’ prefer-ence for locations where congestion islimited (Anas et al., 1998; Carlino and Chat-terjee, 2002). To escape congestion, house-holds and their members periodically changejobs and/or residential location, so that theycan travel shorter distances, or use less con-gested routes. Employers move to lower-density environments for similar reasons.Nevertheless, there is a tendency among


Table 5. Average daily travel distance (km) and time (minutes) for commuting asa car driver, by type of daily urban system in the Netherlands, 1998

Commuting distance Commuting time

Standard StandardAverage deviation Average deviation

Centralised 37.6 46.1 49.5 41.1Decentralised 40.9 45.3 53.6 41.9Cross commuting 35.5 44.4 47.4 38.2Exchange commuting 42.9 54.2 53.3 50.7

Note: Data are averages for car-driving commuters.

firms to concentrate in suburban nodes ofdevelopment, or edge cities (Garreau, 1991),where they can benefit from agglomerationeconomies resulting from the presence andproximity of other firms. This line of reason-ing is known as the co-location hypothesis;Gordon and colleagues (1989a, 1989b and1991) and Levinson and Kumar (1994) pre-sent some empirical support with data from arange of US urban areas. However, the con-tention that polycentrism leads to moreefficient commuting patterns has been chal-lenged. Clark and Kuijpers-Linde (1994) andCervero and Wu (1998), for example, foundthat car travel times and distances are higherin polycentric than in monocentric areas.1 Bycomparing commuting behaviour in 26 urbanareas in the Netherlands, we have investi-gated how polycentrism is associated withlower commute times and distances for cardrivers in the Netherlands. The classificationof daily urban systems (DUSs) developed byvan der Laan (1998) and shown in Figure 3has been used for this.

By and large, commuting is less efficientfor car drivers in polycentric than in mono-centric regions (Table 5). Only in archetypalpolycentric regions consisting of several rela-tively self-contained concentrations of em-ployment and residences outside the old citycentre (cross-commuting DUSs) are averagecommute distances and times lower than intraditional, centralised DUSs. However, thenumber of cross-commuting systems in theNetherlands is very small (Figure 3); mostpolycentric regions belong to the decen-

tralised type, where average commute timesand distances are much higher. Schwanen etal. (2003b) show that the differences be-tween monocentric and polycentric urban ar-eas persist after accounting for differences inthe sociodemographic variables, residentialdensity, urban size and other metropolitanstructure indicators.

In our view, spatial planning policies canbe held to account at least in part for therelatively large commute distances and timesin most polycentric-oriented urban regions inthe Netherlands. While residential growthhas been concentrated in central-city loca-tions and VINEX neighbourhoods, the A-B-Clocation policy has failed and much employ-ment was situated on car-accessible locationsthat were poorly served by public transportand could not be reached readily on foot orbicycle. As a consequence, a limited numberof new housing units have been built nearnew or strongly growing employment con-centrations, many of which can be found at aconsiderable distance from existing residen-tial areas. Several reasons for this mismatchcan be given, including high land prices andthe relatively scarcity of land. This scarcityfor development is at least partly the result ofthe national government’s sustained effortsto keep the Green Heart open. A chain ofevents has led this scarcity of land to becomeone of the causes of the severe distortions ofthe Dutch housing market, which is currentlycharacterised by insufficient new construc-tion, extremely high prices in the owner-occupier stock and long waiting-lists in the


extensive social-rented sector.2 For manyhouseholds, it has become very difficult torent or buy a dwelling near the main bread-winner’s workplace location, or to find em-ployment at a short distance from thehousehold’s residential location.

4. Conclusions and Discussion

The aim of this study was to assess theimpact of Dutch spatial policies on modalsplit and efficiency in travel distance andtime. Table 6 summarises the implicationsfor travel patterns of the main spatial policiesdiscussed in section 2: the policy of concen-trated decentralisation of the 1970s and1980s; the strict compact-city policy of the1980s and 1990s; the A-B-C policy in the1990s; retail planning policy in the past threedecades. It should be stressed that Table 6does not present the results of formal tests ofthe impact of each of the policies mentionedon travel. Instead, the table reflects ourevaluation of how national spatial planninghas affected travel behaviour in the Nether-lands through its regulation of the develop-ment of urban form. In addition, it is difficultto separate the impact of physical planningon travel patterns from other influences, suchas pricing measures impacting travel costs.However, we do feel that these tentativeresults illuminate some of the strengths andweaknesses of spatial planning policy andhence may have some relevance for policy-makers seeking to influence travel behaviourthrough spatial-planning measures.

As Table 6 indicates, the policy of concen-trated decentralisation—the development ofgrowth centres—has probably stimulated theuse of public transport, but has not led tomore cycling or walking. As a result, car useis only slightly lower in growth centres thanin suburban environments. Furthermore,travel distance and time are large in growthcentres. The strict compact-city policy, thesuccessor to the policy of creating growthcentres, seems to have had some positiveoutcomes in terms of travel patterns. Theempirical evidence presented indicates lowertravel distances as well as higher shares for

public transport and cycling and walking. Ithas been suggested, however, that the strictcompact-city policy might have some lessbeneficial effects in the longer term. It mayencourage the suburbanisation of householdsto lower-density environments because oftravel time considerations, which would re-sult in higher shares for and larger distancesby the private car. The A-B-C policy shouldhave led to more public transport use, butfew of the aims of (re)directing employmentgrowth have been met. The policy’s impacton travel behaviour seems to have beenslight; the only effect seems to have beenthat, as a result of developments in the vicin-ity of large railway stations, the shares ofpublic transport for commuting at these loca-tions have increased slightly. Perhaps thegreatest success of Netherlands national spa-tial planning so far is that, thanks to spatialretail planning, walking and cycling forshopping still prevail and that, comparedwith other European countries, the use of thecar for shopping is relatively limited.

Having said that spatial planning policyinfluenced travel behaviour in the Nether-lands, but only to a limited extent and notalways as intended, we may ask ourselveswhether further efforts should be made to tryto influence travel behaviour through spatialplanning. The discussants in the debate onthe role of planning can be broadly dividedinto two camps. There are the ‘liberals’, likeGordon and Richardson (1997), who arguethat market mechanisms result in efficientsettlement patterns in terms of travel andenergy use. In their view, land-use regula-tions and subsidies destabilise the market;direct investment in central cities and publictransport lead to inefficient land-use patterns.Then there are the ‘regulators’, like Ewing(1997) and Cervero (1998), who advocate theregulation of the development of urban formby planners to curtail the costs of urbansprawl, such as the growing use of the pri-vate car.

The discussion in the previous sections onphysical planning and travel patterns in theNetherlands does not provide a convincinganswer to the question whether spatial plan-


Tab

le6.

Perf

orm

ance

ofna

tiona

lsp

atia

lpl

anni

ngpo

licie

sin

the

Net

herl

ands

inte

rms

oftr

avel

effic

ienc

y

Mod

alch

oice

Stim

ulat

ion

Stim

ulat

ion

ofSh

orte

rtr

avel

Shor

ter

Red

uctio

nof

publ

iccy

clin

gan

ddi

stan

cetr

avel

time

ofdr

ivin

gtr

ansp

ort

wal

king

(pri

vate

car)

(pri

vate

car)

Polic

yof

conc

entr

ated

��

��

�de

cent

ralis

atio

n(1

970s

and

1980

s)St

rict

com

pact

-city

polic

y�

��

��

(198

0san

d19

90s)

A-B

-Clo

catio

n�

��

��

Polic

yR

etai

lpl

anni

ng�

/��

��

��

�

‘�’

polic

ym

ade

posi

tive

cont

ribu

tion;

‘�’

polic

yha

dlit

tle(n

eutr

al)

effe

ct;

‘-’

polic

ym

ade

nega

tive

cont

ribu

tion.


ning helps to create travel patterns that aremore efficient in terms of travel distance andtime. The message from the preceding sec-tions is that the promotion of relatively com-pact urban structures can have positiveeffects on walking, cycling and the use ofpublic transport. However, strict designationof a small number of greenfield sites wheredeconcentration and urban growth are to beaccommodated is counter-productive, be-cause short travel times and distances wouldnot be the result. On the contrary, these latteraims might have been reached more readilyunder a regime of fewer land-use controls. Inthat case, the co-location of residences andemployment in close proximity to each othermight have occurred to a greater extent (Gor-don and Richardson, 1997; Levinson andKumar, 1994).

The argument the liberals use—that it isless rather than more land-use control thatmakes travel patterns efficient—is very im-portant, given the fact that we are currentlyheading towards a network society, also ingeographical terms (Castells, 1996; Hayerand Zonneveld, 2000). This type of society ischaracterised by highly diversified and spa-tially expanding activity and travel patternswhich have to be realised within tight timebudgets. The growth of the network societyseems to undercut the ideas of proximity andland-use control as ways to influence socio-spatial development. New spatialconfigurations emerge that can probably becharacterised better by nodes and flows thanin terms of land-use patterns. As a conse-quence, travel will increasingly be perceivedin time spent rather than distance covered(Dijst, 1999; Hayer and Zonneveld, 2000). Inour opinion, spatial planners have underesti-mated this temporal dimension of travel be-haviour in their policies. For instance, theplanning concept of the compact city usesgeographical proximity as the organisingprinciple and may become less relevant inthe light of the emerging network society.

Although a situation of more freedom inlocation choice behaviour—that is fewerrestrictions imposed by physical planningpolicies—might result in time-efficient

settlement patterns, it will probably result ina dispersion of urban functions to lower-density environments and areas that are nowbeing used for agriculture, recreation andnature conservation. In relation to this disper-sion, the potential for public transport will bevastly reduced. This reduction will have anegative impact on the daily life of thosewho cannot afford a car because of financialconstraints or physical impairments; manypeople in this category belong to low-incomehouseholds, are female and/or have reachedretirement age.

Planners are thus facing a dilemma. Thebest way to resolve it, we believe, is todevelop a spatial planning approach combin-ing ‘the best of both worlds’. In the remain-der of this paper, we discuss an outline ofsuch an approach. It aims to optimise thecompetitiveness of public transport and theslow transport modes in terms of patronageand time efficiency. Although the approach istailored to the Dutch situation, we believethat the underlying principles have a widerapplicability, in particular to other countriesin north-west Europe.

First, the use of public transport and walk-ing and cycling can best be promoted bystimulating relatively compact urban areas.With respect to mode choice, Dutch spatialplanning has been most effective in theRandstad Holland (section 3). In terms oftravel time and travel distance, however, themore urbanised settlements outside the Rand-stad (at least 25 000–50 000 inhabitants) per-form quite well. These results lead us tothink that spatial policies aiming at the con-centration of land uses in smaller citieswould be highly beneficial from a travel-behaviour perspective. For the UK, Banister(1992) reached a comparable conclusion.

This national spatial policy is a necessary,but not sufficient, condition for improvingthe competitiveness of public transport. Thepolicy has to be supplemented by invest-ments in relatively high-speed public trans-port systems, such as intercity trains at thenational level and light-rail and metro sys-tems at the conurbation level (Newman andKenworthy, 2000). These systems can


guarantee travel times which are competitivewith the private car (Bovy et al., 1990; vanden Heuvel, 1997). On the other hand, atthe local level, competition between thebicycle—a cheap and flexible transportmode—and public transport should be pre-vented as far as possible. Rather, the comple-mentarity of the environmentally friendlytransport systems of the bus/tram/metro andthe bicycle should be promoted. The inte-gration of both systems could be improvedby allowing the transport of bicycles on pub-lic transport vehicles such as trams and met-ros. Furthermore, demand-responsive publictransport could replace line-haul local sys-tems to guarantee public transport to all po-tential users.

The competitiveness of public transportmay also be improved by local land-use poli-cies focusing on nodes in the public transportsystem. In general, the attractiveness of pub-lic transport decreases rapidly beyond 500–800 metres from a public transport stop. It istherefore necessary to encourage high-density, mixed-use development within 500–800 metres of a public transport stop(Cervero, 1996b). In our opinion, a system offinancial incentives would contribute to thedevelopment of such public transport nodes.Two types of incentive might be offered. Onthe demand side, firms and facilities wouldreceive an incentive when locating in publictransport nodes. The magnitude of this incen-tive would depend on the distance to the stopand the amount of land occupied. Shortwalking distances and efficient use of space(high-rise buildings for example) would berewarded. In addition, incentives would beoffered to real-estate developers on the sup-ply side of the market. The more varied interms of land use a (re)developed publictransport node became, the greater the incen-tive the developers could receive. Becauseshort walking and cycling distances would bepromoted, these locational incentives wouldfavour not only the use of public transportbut also walking and cycling.

Opponents of the regulations describedhere may argue that this approach would notimprove the efficiency of travel times or

distances. A strict designation of new build-ing sites for commercial use at a small num-ber of public transport nodes would not becapable of providing sufficient opportunitiesfor households, firms and services to(re)locate to sites in close proximity to eachother. Hence, ‘the best of both worlds’ plan-ning approach should also include a partialderegulation of the land and housing mar-kets.

According to the World Bank (1993), oneway of making the housing market workmore efficiently would be to select a largenumber of competitive new greenfield sitesfor development and equip these with thenecessary infrastructure provisions (such asroad and rail networks) far in advance of theconstruction of new housing and other landuses. Such a partial deregulation could yieldthree kinds of travel-related benefit. First, alarge supply of new building sites in publictransport nodes would improve the competi-tiveness of public transport, cycling andwalking as outlined above. Secondly, thesupply of high-quality public transport innew residential areas before the arrival of thefirst new residents would offer better oppor-tunities for discouraging the development ofcar-based travel patterns. Finally, a largesupply of housing would encourage residen-tial mobility, which could result in moreefficient travel times and distances for bothpublic transport and private car users.

The combination of spatial policies at na-tional and local levels, investments in publictransport systems and a system of locationalincentives would give ample opportunities tohouseholds, firms and services to makeefficient locational choices. At the sametime, this spatial planning approach wouldprovide public-transport captives withsufficient opportunities for travelling in to-day’s society and preserve valuable culturaland natural landscapes.

Notes

1. A detailed review of these and other studiesis presented in Schwanen et al. (2002 and2003a).


2. The problematic situation of the Dutch hous-ing market can be illustrated by means of thenumber of housing units available per 1000inhabitants. For the Netherlands, this figureis 406, whereas the weighted mean for 9other European countries is 443 (Feddes andDieleman, 2000).

References

ANAS, A., ARNOTT, R. and SMALL, K. A. (1998)Urban spatial structure, Journal of EconomicLiterature, 26, pp. 1426–1464.

BANISTER, D. (1992) Energy use, transport andsettlement patterns, in: M. J. BREHENY (Ed.)Sustainable Development and Urban Form,pp. 160–182. London: Pion Ltd.

BONTJE, M. (2001) Idealism, realism and theDutch compact city, Town and Country Plan-ning, December, pp. 336–337.

BOVY, P. H. L. and SALOMON, I. (2002) Conges-tion in Europe: measurements, patterns andpolicies, in: E. STERN, I. SALOMON and P. H. L.BOVY (Eds) Travel Behaviour: Spatial Pat-terns, Congestion and Modelling, pp. 143–179.Cheltenham: Edward Elgar.

BOVY, P. H. L., BAANDERS, A. and WAARD, J. VAN

DER (1990) Hoe kan dat nou? De discussie overde substitutiemogelijkheden tussen auto enopenbaar vervoer [How is that possible? Thediscussion about substitution between the pri-vate car and public transport], in: J. M. JAGER

(Ed.) Colloquium Vervoersplanologisch Speur-werk: Measuring, Modelling, Monitoring,Recent Developments in Research Methods,pp. 121–142. Delft: CVS.

BRUN, J. and FAGNANI, J. (1994) Lifestyles andlocational choices—trade-offs and compro-mises: a case-study of middle-class couplesliving in the Ile-de-France region, UrbanStudies, 31(6), pp. 921–934.

CAMAGNI, R., GIBELLI, M. C. and RIGAMONTI, P.(2002) Urban mobility and urban form: thesocial and environmental costs of different pat-terns of urban expansion, Ecological Econom-ics, 40, pp. 199–216.

CARLINO, G. and CHATTERJEE, S. (2002) Employ-ment deconcentration: a new perspective onAmerica’s postwar urban evolution, Journal ofRegional Science, 42(3), pp. 445–475.

CASTELLS, M. (1996) The Rise of the NetworkSociety: Economy, Society, Culture. Oxford:Blackwell.

CERVERO, R. (1995) Planned communities, self-containment and commuting: a cross-nationalperspective, Urban Studies, 32(7), pp. 1135–1161.

CERVERO, R. (1996a) Traditional neighborhoods

and commuting in the San Francisco Bay Area,Transportation, 23, pp. 373–394.

CERVERO, R. (1996b) Mixed land uses and com-muting: evidence from the American HousingSurvey, Transportation Research A, 30,pp. 361–377.

CERVERO, R. (1998) The Transit Metropolis: AGlobal Inquiry. Washington, DC: Island Press.

CERVERO, R. (2002) Built environments and modechoice: toward a normative framework, Trans-portation Research D, 7, pp. 265–284.

CERVERO, R. and WU, K.-L. (1998) Sub-centringand commuting: evidence from the San Fran-cisco Bay Area, Urban Studies, 35, pp. 1059–1076.

CLARK, W. A. V. and KUIJPERS-LINDE, M. (1994)Commuting in restructuring urban regions.Urban Studies, 31, pp. 465–483.

COUNSELL, D. (2001) A regional approach to sus-tainable urban form?, Town and Country Plan-ning, December, pp. 332–335.

DIELEMAN, F. M. and WALLET, C. (2003) Incomedifferences between central cities and suburbsin the Dutch urban regions, Tijdschrift voorEconomische en Sociale Geografie, 94(2),pp. 265–275.

DIELEMAN, F. M., DIJST, M. and BURGHOUWT, G.(2002) Urban form and travel behaviour: mi-cro-level household attributes and residentialcontext, Urban Studies, 39, pp. 507–527.

DIELEMAN, F. M., DIJST, M. J. and SPIT, T. (1999)Planning the compact city: the Randstad Hol-land experience, European Planning Studies,7(5), pp. 605–621.

DIJST, M. J. (1997) Spatial policy and passengertransportation, Netherlands Journal of Housingand the Built Environment, 12(1), pp. 91–111.

DIJST, M. J. (1999) Action space as a planningconcept in spatial planning, Netherlands Jour-nal of Housing and the Built Environment,14(2), pp. 163–182.

DIJST, M., TIMMERMANS, P. and STEENBRINK, P. A.(2000) Parkeren voor de deur? Mobiliteit inLeidsche Rijn [Parking in front of the home?Travel behaviour in Leidsche Rijn], in: Leid-sche Rijn Monitor 2000. Viva Vinex?, pp. 31–44. Utrecht: Berenschot, Universiteit Utrecht,Utrechts Nieuwsblad.

EVERS, D. (2002) The rise (and fall?) of nationalretail planning, Tijdschrift voor Economischeen Sociale Geografie, 93(1), pp. 107–113.

EWING, R. (1997) Is Los Angeles-style sprawldesirable?, Journal of the American PlanningAssociation, 63(1), pp. 107–126.

FALUDI, A. and VALK, A. VAN DER (1990) DeGroeikernen als Hoekstenen van het Neder-lands Ruimtelijke Ordeningsdoctrine [TheGrowth Centres as Cornerstones of the DutchSpatial Planning Doctrine]. Assen/Maastricht:van Gorcum.


FEDDES, A. and DIELEMAN, F. M. (2000) Receptvoor een functionerende woningmarkt [Pre-scription for a functioning housing market],Vastgoedmarkt, April, pp. 38–41.

FRANK, L. D. and PIVO, G. (1994) Impacts ofmixed use and density on utilization of threemodes of travel: single-occupant vehicle, tran-sit, and walking, Transportation ResearchRecord, 1466, pp. 44–52.

GARREAU, J. (1991) Edge City: Life on the NewFrontier. New York: Doubleday.

GOLDSTEIN, H. (1995) Multilevel Statistical Mod-els. London: Edward Arnold.

GORDON, P. and RICHARDSON, H. W. (1997) Arecompact cities a desirable planning goal?, Jour-nal of the American Planning Association, 63,pp. 95–106.

GORDON, P., KUMAR, A. and RICHARDSON, H. W.(1989a) The influence of metropolitan spatialstructure on commuting time, Journal of UrbanEconomics, 26, pp. 138–151.

GORDON, P., KUMAR, A. and RICHARDSON, H. W.(1989b) Congestion, changing metropolitanstructure and city size in the United States,International Regional Science Review, 12,pp. 45–56.

GORDON, P., RICHARDSON, H. W. and JUN, M.-J.(1991) The commuting paradox: evidence fromthe top twenty, Journal of the American Plan-ning Association, 57, pp. 416–420.

GUO, J. and BHAT, C. (2002) Residential LocationModeling: Accommodating Sociodemographic,School Quality and Accessibility Effects.Austin, TX: Department of Civil Engineering,University of Texas, Austin.

HAM, M. VAN, HOOIMEIJER, P. and MULDER, C. H.(2001) Urban form and job access: disparaterealities in the Randstad, Tijdschrift voorEconomische en Sociale Geografie, 92(2),pp. 231–246.

HANDY, S. L. and CLIFTON, K. J. (2001) Localshopping as a strategy for reducing automobiletravel, Transportation, 28, pp. 317–346.

HART, T. (1992) Transport, the urban pattern andregional change, Urban Studies, 29(3/4),pp. 483–503.

HART, T. (2001) Transport and the city, in: R.PADDISON (Ed.) Handbook of Urban Studies,pp. 102–123. London: Sage Publications.

HAYER, M. and ZONNEVELD, W. (2000) Spatialplanning in the network society: rethinking theprinciples of planning in the Netherlands,European Planning Studies, 8, pp. 337–355.

HERZ, R. (1983) Stability, variability andflexibility in everyday behaviour, in: S. CAR-

PENTER and P. JONES (Eds) Recent Advances inTravel Demand Analysis, pp. 385–400. Alder-shot: Gower.

HEUVEL, M. G. VAN DEN (1997) Openbaar Vervoerin the Randstad: Een Systematische Benadering

[Public Transport in the Randstad Holland: ASystematic Approach]. Amsterdam: ThesisPublishers.

HJORTHOL, R. J. (2000) Same city—different op-tions: an analysis of the work trips of marriedcouples in the metropolitan area of Oslo, Jour-nal of Transport Geography, 8, pp. 213–220.

JANSEN, G. R. M. (1993) Commuting: homesprawl, job sprawl, traffic jams, in: I. Salomon,P. BOVY and J.-P. ORFEUIL (Eds) A BillionTrips a Day: Tradition and Transition in Eu-ropean Travel Patterns, pp. 101–127. Dor-drecht: Kluwer Academic Publishers.

JANSEN, G. R. M., HILBERS, H. and WILMINK, I.(2002) Transport networks and mobility: acomparison of the Randstad, the Rhine–Ruhrand the Antwerp–Brussels–Ghent region, in: E.STERN, I. SALOMON and P. H. L. BOVY (Eds)Travel Behaviour: Spatial Patterns, Congestionand Modelling, pp. 20–36. Cheltenham:Edward Elgar.

JOHNSTON-ANUMONWO, I. (1992) The influence ofhousehold type on gender differences in worktrip distance, The Professional Geographer, 44,pp. 161–169.

KEMPEN, R. VAN (2000) Big cities policies in theNetherlands, Tijdschrift voor Economische enSociale Geografie, 91, pp. 197–203.

KEMPEN, R. VAN and PRIEMUS, H. (1999) Undiv-ided cities in the Netherlands: present situationand political rhetoric, Housing Studies, 14,pp. 641–657.

KENWORTHY, J. R. and LAUBE, F. B. (1999) Pat-terns of automobile dependence in cities: aninternational overview of key physical andeconomic dimensions with some implicationsfor urban policy, Transportation Research A,33, pp. 691–723.

KITAMURA, R., HOORN, T. VAN DEN and WIJK, F.VAN (1997) A comparative analysis of dailytime use and the development of an activity-based traveller benefit measure, in: D. ETTEMA

and H. TIMMERMANS (Eds) Activity-based Ap-proaches to Travel Analysis, pp. 171–188. Ox-ford: Pergamon.

KOCKELMAN, K. M. (1997) Travel behavior as afunction of accessibility, land use mixing, andland use balance, Transportation ResearchRecord, 1607, pp. 116–125.

LAAN, L. VAN DER (1998) Changing urbansystems: an empirical analysis at two spatiallevels, Regional Studies, 32, pp. 235–247.

LERMAN, S. R. (1976) Location, housing, automo-bile ownership, and mode to work: a jointchoice model, Transportation ResearchRecord, 610, pp. 5–11.

LEVINSON, D. M. and KUMAR, A. (1994) The ratio-nal locator: why travel times have remainedstable, Journal of the American PlanningAssociation, 60, pp. 319–332.


LEVINSON, D. M. and KUMAR, A. (1997) Densityand the journey to work, Growth and Change,28(2), pp. 147–172.

MCDOWELL, L. (1997) The new service class:housing, consumption and lifestyle among Lon-don bankers in the 1990s, Environment andPlanning A, 29, pp. 2061–2078.

MVRO (MINISTRY OF HOUSING AND PHYSICAL

PLANNING) (1966) Tweede Nota RuimtelijkeOrdening [Second Physical Planning Memo-randum]. The Hague: Ministry of Housing andPhysical Planning.

MVROM (MINISTRY OF HOUSING, PHYSICAL PLAN-

NING AND THE ENVIRONMENT) (1988) VierdeNota Ruimtelijke Ordening [Fourth PhysicalPlanning Memorandum]. The Hague: Ministryof Housing, Physical Planning and the Environ-ment.


NING AND THE ENVIRONMENT) (1991) VierdeNota Ruimtelijke Ordening Extra [Fourth Spa-tial Planning Memorandum Extra]. The Hague:Ministry of Housing, Physical Planning and theEnvironment.


NING AND THE ENVIRONMENT) (2001) VijfdeNota Ruimtelijke Ordening 2000/2020: RuimteMaken, Ruimte Delen [Fifth Spatial PlanningMemorandum: Making Space, Sharing Space].The Hague: Ministry of Housing, PhysicalPlanning and the Environment

MV&W (MINISTRY OF TRANSPORT AND PUBLIC

WORKS) (1999) Perspectievennota Verkeer enVervoer [Report of Perspectives on Traffic andTransport]. The Hague: Ministry of Transportand Public Works.

NÆSS, P., RØE, P. G. and LARSEN, S. (1995) Trav-elling distances, modal split and transportationin thirty residential areas in Oslo, Journal ofEnvironmental Planning and Management, 38,pp. 349–370.

NEWMAN, P. and KENWORTHY, J. (2000) Sustain-able urban form: the big picture, in: K.WILLIAMS, E. BURTON and M. JENKS (Eds)Achieving Sustainable Urban Form, pp. 139–148. London: E & FN Spon.

NRC HANDELSBLAD (2001) The Dutch are work-ing and commuting extensively, 12 April.

OSKAMP, A., POULUS, C. and GORDIJN, H. (2002)Stadsvernieuwing gemeten [Measuring urbanrenewal], Kwartaalbericht ABF Research, 1,pp. 3–6.

PRIEMUS, H., NIJKAMP, P. and BANISTER, D. (2001)Mobility and spatial dynamics: an uneasy rela-tionship, Journal of Transport Geography,9(3), pp. 167–171.

SCHWANEN, T. (2002) Urban form and commutingbehaviour: a cross-European perspective, Tijd-schrift voor Economische en Sociale Geografie,93(3), pp. 336–343.

SCHWANEN, T., DIELEMAN, F. M. and DIJST, M.(2001) Travel behaviour in Dutch monocentricand polycentric urban systems, Journal ofTransport Geography, 9(3), pp. 173–186.

SCHWANEN, T., DIELEMAN, F. M. and DIJST, M.(2003a) Car use in Netherlands daily urbansystems: does polycentrism result in lowercommute times?, Urban Geography, 24(5),pp. 340–360.

SCHWANEN, T., DIELEMAN, F. M. and DIJST, M.(2003b) A multilevel analysis of the impact ofmetropolitan structure on commute behaviourof urban residents in the Netherlands. Paperpresented at the 82nd Annual Meeting of theTransportation Research Board, January,Washington, DC.

SCHWANEN, T., DIJST, M. and DIELEMAN, F. M.(2001) Leisure trips of senior citizens: determi-nants of modal choice, Tijdschrift voorEconomische en Sociale Geografie, 92(3),pp. 347–360.

SCHWANEN, T., DIJST, M. and DIELEMAN, F. M.(2002) A microlevel analysis of residential con-text and travel time, Environment and PlanningA, 34(8), pp. 1487–1507.

SNIJDERS, T. A. B. and BOSKER, R. J. (1999) Multi-level Analysis: An Introduction to Basic andAdvanced Multilevel Modeling. London: SagePublications.

STATISTICS NETHERLANDS (1999) Onderzoek Ver-plaatsingsgedrag 1998: Documentatie voorTape-gebruikers [1998 National Travel Survey:Documentation for Tape-users]. Voorburg/Heerlen: Statistics Netherlands.

TURNER, T. and NIEMEIER, D. (1997) Travel towork and household responsibility: new evi-dence, Transportation, 24, pp. 397–419.

VAN, U.-P. and SENIOR, M. (2000) The contribu-tion of mixed land uses to sustainable travel incities, in: K. WILLIAMS, E. BURTON and M.JENKS (Eds) Achieving Sustainable UrbanForm, pp. 139–148. London: E & FN Spon.

WEBER, J. and KWAN, M.-P. (2003) Evaluating theeffect of geographic contexts on individual ac-cessibility: a multilevel level approach, UrbanGeography, 24 (Forthcoming)

WILLIAMS, K., BURTON, E. and JENKS, M. (2000)Achieving Sustainable Urban Form. London: E& FN Spon.

WORLD BANK (1993) Housing: Enabling Marketsto Work. Washington, DC: The World Bank.

Appendix 1

Degree of urbanisation is a composite variablewhich incorporates various dimensions of urbanform: urban size, residential density, distance tothe urban core, the degree of land-use mixing andprovision of infrastructure. The variable distin-


guishes first between the Randstad Holland andthe Rest of the Netherlands, because residentialdensities in the former are much greater and morepublic transport infrastructure is available. Withinthe Randstad Holland, we discern from most toleast urbanised: three large cities (Amsterdam,Rotterdam, The Hague); medium-sized cities; andsuburbs. The growth centres are considered to bea separate category. The national governmentconsciously planned these settlements during the1970s and 1980s to become self-contained com-munities, in order to curb the decentralisation ofhouseholds and firms from the larger cities. How-ever, they turned into dormitory towns whoseinhabitants had to travel extensively to reach thedestinations they preferred to visit. Outside theRandstad Holland, a distinction has been drawnbetween more urbanised and less urbanised areas.

With regard to urban structure, we use aclassification of daily urban systems in four cate-gories based on van der Laan (1998), rather thandrawing a distinction between monocentric orpolycentric urban areas

(1) Centralised: these resemble monocentric sys-tems in which morning peak-hour commutingis primarily directed towards the core city ofthe DUS.

(2) Decentralised: many morning commuters areattracted to the suburban parts of the system,where much employment is located.

(3) Cross commuting: these structures resemblethe classic polycentric region consisting ofrelatively independent, self-contained devel-opment nodes. Suburban commuters tend towork in the suburbs; core-city residents oftenwork in the core city.

(4) Exchange commuting: these systems havemany reciprocal relationships between thesuburbs and the core city. Many suburbancommuters work in the city, while many cen-tral-city residents work in the suburbs.

Appendix 2

Multilevel models expand the random part of amodel; rather than summarising the residual vari-ance in a single random parameter as in ordinaryleast squares regression, more than one randomterm is estimated to accommodate the nestedstructure of the data. A random-intercept modelwith three levels of analysis—for instance, indi-viduals within households within municipalities—can be written as (Goldstein, 1995)

yijk � �0 � �iX1ijk � (e0ijk � u0ijk � 0k)

where, yijk represents the dependent variable at thelowest level of analysis; �0 indicates the intercept;X1ijk is an independent variable at the level of theindividual; and �1 a fixed regression; coefficientto be estimated. The parentheses in the equationindicate the random part of the model. The ran-dom term e0ijk refers to the individual, u0jk to thehousehold; and 0k to the municipality of resi-dence. These random terms are assumed to bemutually independent and to be normally dis-tributed—that is, they have a mean of zero andcan be summarised by their variances 2

e0, 2u0

and 2v0, respectively. The coefficients of multi-

level models are estimated with maximum likeli-hood estimation procedures (more details inGoldstein, 1995).

policies for urbanism

Documents