j.1467-9663.2011.00673.x Beyond Postuburbia_ Multifunctional Service Agglomeration in Vienna Urban...

8/11/2019 j.1467-9663.2011.00673.x Beyond Postuburbia_ Multifunctional Service Agglomeration in Vienna Urban Fringe

1/14

BEYOND POSTSUBURBIA? MULTIFUNCTIONALSERVICE AGGLOMERATION IN VIENNAS

URBAN FRINGEtesg_673 39..52

MARCO HELBICH

Institute of Geography, University of Heidelberg, Germany.E-mail: [email protected]

Received May 2010; accepted April 2011

ABSTRACT

The surroundings of European cities are becoming ever more complex and sophisticated land-scapes. In recent years there has been an ongoing debate about urban fringes that are developingover and beyond the influence of postsuburban processes into decentralised and non-centredurban landscapes. This paper addresses the urgent need to consider polycentric functional serviceclustering in the urban fringes based on empirical evidence gathered in Vienna, Austria, whichserved as a case study. A spatial scan statistic is applied to detect local spatial clusters of servicesand the bivariate K(d)-function is used to investigate the spatial interaction between each nucleus.Due to some significant functional clusters, it has been concluded that agglomeration economiesare still of prime interest. Furthermore, some significant attraction between the studied clusters isinvestigated. Overall, the results do not contradict the concept of postsuburbia yet, which is whya paradigm shift must be ruled out.

Key words: postsuburbanisation, cluster detection, spatial scan statistic, bivariateK(d)-function,Vienna (Austria)

BACKGROUND FROM SUBURBIA VIAPOSTSUBURBIA TO . . . ?

As the last decades have shown, urban regionsare a highly dynamic type of territory subject toconstant change due to endogenous and exog-enous influences. The impacts of globalisation,information economies, social and demo-graphic changes have given these areas a dis-tinct appearance (Castells 1989; Hall 1993).This realignment towards a postmodern societyand urban development (Dear & Flusty 2002),has resulted in a self-evident restructuring ofthe spatial organisation of metropolitan areasand their functional dependencies on the corecity. Nowadays the dichotomy between rural

and urban areas is becoming increasinglyblurred and cities have been turned inside out

(Muller 1997, p. 45). Suburban spatial struc-tures are now often a relic of the past. Froma theoretical perspective such urban formscannot adequately be conceptualised withinmonocentric urban models (Anas et al. 1998;Clark 2002), whereas some density gradients,like population density or land price, decreaseuniformly with distance to the core city(Mills 1970). This transformation process hasreceived considerable attention as an empiricalissue and from a theoretical point of view inurban geography, regional science, and territo-rial planning.

Several recent US (e.g. Giuliano & Small1991; McMillen & McDonald 1998; Soja 2000),and European studies (e.g. Blotevogel 1998;

Kagermeier et al. 2001; Kloosterman & Musterd2001; Helbich & Leitner 2010), confirm this

Tijdschrift voor Economische en Sociale Geografie 2012, DOI:10.1111/j.1467-9663.2011.00673.x, Vol. 103, No. 1, pp. 3952. 2011 The AuthorTijdschrift voor Economische en Sociale Geografie 2011 Royal Dutch Geographical Society KNAGPublished by Blackwell Publishing Ltd., 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA 02148, USA


2/14

ongoing (primarily economically aligned) evo-lution from a monocentric into a polycentricurban structure. This means that there arespatially delineated centres composed of large

populations and employment opportunitiesas well as services within the surroundings ofthe core city (Kloosterman & Musterd 2001).

Additionally, the concept of polycentrism,which will not be taken into account in thefollowing, serves as an integral planning doc-trine (Faludi 2002), not limited to a regionalscale as discussed below. It is also an essentialpart of the European spatial developmentperspective to get a balanced urban system(European Commission 1999).

Consequently, urban theory (Fishman 1987;Garreau 1992; Kling et al. 1995; Soja 2000;Brake 2005), has already incorporated this newurban fabric of metropolitan areas and hassought to deal with this process of maturation,including a functional enrichment of fringesand a spatial reorganisation that benefits sub-urban landscapes in various facets. Therefore, aterminologicaltabula rasaof different concepts(Vaughanet al. 2009), for example, postsubur-bia (Kling et al. 1995), edge cities (Garreau

1992), in-between-cities (Sieverts 1998), exopo-lis (Soja 2000), among others, is suggested fordealing with these novel US and Europeanstructures. One common feature is fragmentedand polynucleated functional patches of ser-

vices, located within the urban fringes.In the following, postsuburbia is examined

and discussed because it is the most compre-hensive concept, under which various currentdevelopment tendencies such as the agglom-eration of offices, retail space in the urban

fringes, fragmentation of the metropolitanarea, the diversity of the population and multi-nucleated structure in the surroundings ofthe core city can be subsumed. Because of thisdiversity, this study addresses only the latteraspect, namely the multinucleated urban land-scape. Lifestyle businesses and health servicesrelocating to the cities surroundings constitutea rather new and unique development com-pared to the suburbanisation era. These reloca-tions have resulted in a functional enrichment

of the urban fringes, which contrasts with thecore idea of Christallers central place theory(Brake 2005). Because of the increased im-portance of agglomeration economies and the

need to maximise profits, services aspire tospatial proximity and form spatial clusters (Soja2001).1 Bkemann (1999) sees a cluster as anagglomeration of services or institutions in a

particular location. The result is a polynucle-ated functional urban structure in the formersuburban fringes. This, in turn, has given wayto intra-regional interaction patterns, whichbecome more diffuse and multidirectional(Hesse 2001; Schwanen et al. 2001). Someauthors, Kunzmann (2001) and Brake (2005)among others, argue that the functionallyenriched polycentric pattern has alreadyresulted in an emancipation of the fringes fromthe core city. Thus, the traditional core city has

lost its primary status, becoming just one ofmany parts of the metropolitan area (Fishman1987; Brake 2005). Kunzmanns (2001) archi-pelago of the metropolitan area summarisesthis re-shaping process and backs his claim thattodays urban fringes consist of several more orless independent monofunctional clusters offirms.

There is an ongoing debate, initiated byGordon and Richardson (1996) and genera-lised by Lang (2003), Lang and LeFurgy (2003)

as well as Langet al. (2009), on spatial patternsdeveloping beyond postsuburban polycentriclandscapes. According to this concept, postsub-urban landscapes are just a transitory phaseleading up to a scattered intrametropolitanurban structure (Shearmuret al. 2007), whichis conceptualised and refined in Langs (2003)edgeless cities. This diffuse concept describesa kind of scattered office development through-out the urban landscape which never assumesthe scale, density, and cohesiveness of edge

cities.While spatial changes from sub to postsubur-bia are theoretically and empirically well knownat present (e.g. Fishman 1987; Klinget al. 1995;Soja 2000), there has been less research seekingto find evidence for urban structures beyondthese urban patterns (e.g. Lang & LeFurgy2003; Phelpset al. 2006; Lang et al. 2009). Onthis subject, this paper makes an initial empiri-cal contribution and charts current urban pat-terns. Therefore, first, the paper aims to find

local geographic agglomerations of services inthe Viennese urban fringe, which would refutethe thesis of diffuse urban forms. Second, ifsome clusters are available, the focus lies on the

40 MARCO HELBICH

2011 The AuthorTijdschrift voor Economische en Sociale Geografie 2011 Royal Dutch Geographical Society KNAG


3/14

spatial relationship between them, to evaluatepossible urban structures beyond Kunzmanns(2001) reference model. More specifically, thefollowing specific research questions are to be

addressed. Are there local functional spatialclusters in Viennas urban fringe? If this can beverified: what are their spatial interrelations? Isthere some spatial interaction between theiroccurrences?

The paper is structured as follows. Thefollowing section provides a literature reviewon empirical manifestations of polycentric anddecentralised urban forms. The study area,the data sets, and some data preprocessingsteps are then introduced. The subsequent

section briefly presents the methodology usedto detect local clusters of services and anapproach to analysing the relationship betweendifferent point patterns. Drawing on the previ-ous section, the results of the empirical analysisare discussed. The paper concludes with asummary, relates the findings to urban theoryand discusses the results in comparison to otherempirical findings.

EMPIRICAL EVIDENCE OF ALTERED

URBAN FORMS

Urban analysis is a diverse and extensivemultidisciplinary field of research (Harris &Larkham 1999). Little academic considerationsdeal empirically with European postsuburbanstructures, the most notable exceptionsbeing Phelps and Parsons (2003), Bontje andBurdack (2005), Phelpset al. (2006) as well asHelbich and Leitner (2009, 2010).

The transformation of metropolitan areas

into polycentric urban patterns has been exam-ined primarily in the US (e.g. McMillen 1996),and selected European cities (e.g. Dieleman &Faludi 1998). At present this is still a novelresearch topic as there is no common topologyand taxonomy concerning the methods andoperationalisation of evaluating local spatial

variability. One reason is probably the fact thata grand theory of polycentricity is still missing(Kloosterman & Musterd 2001, p. 624), whichcomplicates the comparison of empirical

findings due to the application of differenttechniques.For example, Feser and Sweeney (2000)

applied a point pattern approach and showed

an association between economic linkages andgeographic clustering in the US state of NorthCarolina. Enterprises in knowledge-based ortechnology-intensive sectors in particular tend

to cluster in space. In comparison, Maoh andKanaroglou (2007) related economic cluster-ing to urban form, using the city of Hamilton,Canada, as a case study. They concluded that amultinucleated urban form is likely to emerge.Finally, Cuthbert and Anderson (2002) demon-strate the strength of point pattern analysis forthe Halifax-Dartmouth region in Canada andfound evidence for a changing urban form witha tendency towards multinucleation.

Two prime examples of polycentric urban

landscapes are the US metropolises of Chicagoand Los Angeles. McMillen (1996) stated thatChicagos monocentric pattern already ceasedto exist in the 1960s in favour of subcentres,classified as edge cities with multifunctionalconcentrations of offices, retail, leisure, andhousing areas (McMillen & McDonald 1998).Several other scholars, notably Fishman(1987), Giuliano and Small (1991), Klinget al.(1995), and Soja (2000), agree that the Los

Angeles metropolitan area is a prototype for a

postsuburban landscape. Beyond that, thereis literature on a set of numerous otherNorth American case studies for polycentricstructures (e.g. Bogart & Ferry 1999; Coffey &Shearmur 2002).

In Europe examples of polycentric urbanpatterns have been identified in the DutchRandstad (Dieleman & Faludi 1998), theRhine-Ruhr metropolitan region (Blotevogel1998), Munich (Kagermeier et al. 2001), and

Vienna (Helbich & Leitner 2010), among

others. Bontje and Burdack (2005) evaluatedthe possible occurrence of edge cities in Euro-pean metropolitan areas, such as Paris and theDutch Randstad. Because Garreaus (1992) cri-teria2 of an edge city could not be found, theyconcluded that city edges are the Europeanpendants to the North American edge cities.Nevertheless, city edges have some similarityin terms of specialisation (e.g. financial andinsurance institutions), with edge cities, but arenot spatially independent of the traditional

core city. Also Phelps et al. (2006) emphasisethat the postmodernity of US cities can berarely found to the same degree in the Euro-pean context; as such their case studies (e.g.

MULTIFUNCTIONAL SERVICE AGGLOMERATION IN VIENNAS URBAN FRINGE 41



4/14

Noisy-le-Grand next to Paris) are unlike USones. Moreover, there is heterogeneity withinEuropean countries. Helbich and Leitner(2010) undertook the first attempts to quantify

postsuburban spatial phenomena for Viennasurban fringe. Their findings refer to the spa-tially clustering effects of services. However,since they worked with global clustering tests,their study clearly has methodological limita-tions. It was not possible to detect local, andthus mappable, clusters. In addition, they didnot investigate spatial interaction betweenservices.

Few examples exist that refute the existenceof a polycentric urban region in Europe.

Baumont et al. (2004) have analysed thespatial distribution of population and employ-ment in the metropolitan area of Dijon(France) and concluded that the spatialpattern still exhibits a monocentric character.

Also, Riguelleet al. (2007) claim that for Brus-sels spatial patterns towards polycentrism arestill weak.

Few investigations deal with intrametro-politan scatteration and edgeless cities, res-pectively. Yet, the results from Gordon and

Richardson (1996) suggest that Los Angeleshad already matured from a polycentric to adispersed metropolitan area. Also, Lang(2003) agreed with the notion of Los Angelesbeing a dispersed metropolitan area. Lang(2003) and Langet al. (2009) found evidencefor such a pattern in Miami and Philadelphiatoo. In addition, Lee (2007) empirically sup-ports the hypothesis of continuing decen-tralisation for Portland and Philadelphia.Recently, based on case studies of the Cana-

dian cities of Toronto, Montreal, and Vancou-ver, Shearmur et al. (2007) found evidencethat both concepts, polycentrism and scattera-tion, occurred simultaneously on differentspatial and temporal scales. However, it is stillunclear what theoretical implications this willhave.

In general, since the results of US cities areonly partially valid for European metropolitanareas (Phelpset al. 2006), deeper insights intospatial processes will certainly require more

research. This will enable more general con-clusions and knowledge as to what extentpostsuburban processes shape European urbanfringes.

STUDY AREA AND DATA

Study area The Vienna urban fringe servesas the study area, where the emergence of

postsuburban patterns is, most likely to haveoccurred within the Austrian context. Since the1980s this area has been altered by migrationfrom urban to rural areas. Thereby, the po-pulation is not homogenously distributedthroughout the fringe. Hot spots that are sig-nificant in terms of population are located tothe north of Vienna and along the main trafficaxis in the south (Helbich & Leitner 2009).This corresponds to the spatial distribution ofservices that followed after a short time lag.

These services completed the facilities in thesurroundings (Fassmann et al. 2009). But thedesignation of protective areas, like the bio-sphere reserve Wienerwald, in the southwest-ern parts of the fringe, as well as tight normativeplanning laws and local regulatory powerprevent undesirable developments and uncon-trolled settlement growth. Nevertheless, Musiland Pindur (2008) predict stronger urbandevelopment pressure in far more peripheralareas for the coming years.

For delimitation purposes a slightly modifiedouter boundary compared to the one of Fuchs(1997) is used. In the applied definition of theurban fringe, the core city is excluded becausethe focus lies on polycentric patterns in thesurroundings and, in addition, the urbanstructure outside the traditional core city hasalready matured, becoming independent, andhas (theoretically) broken away from its core(Phelps 2004; Brake 2005). Consequently, thecore city has become marginalised, since it is

only one of many nuclei of the urban structure(Fishman 1987; Brake 2005). Further, currentspatial processes manifest themselves in urbanfringes (e.g. Soja 2000). This also holds truefor the structure of postsuburban landscapes

where service clusters are exclusively located inthe surroundings of the core city. However, thespecific inner city subcentres are in this case notof interest but still certainly present.

Data and data preprocessing For this study

geo-referenced locations of service sector firmsfrom April 2006 are used to evaluate polynucle-ated urban structures in the Viennese fringes.The micro-geographic dataset was collected

42 MARCO HELBICH



5/14

and sold as Marketing CD Business Geo

from the commercial data provider HeroldBusiness Data. It contains highly accurate geo-referenced firm locations and associatedattributes, as, for instance, the industrial classi-fication code and business volumes.

As mentioned above, it is characteristic ofa postsuburban structure, namely, atypical ofsuburban conditions, that firms are increas-ingly located in the urban fringes. A largernumber of highly specialised firms in the urbanfringes results in their functional enrichment

and in a growing demographic and socio-economic diversity of their population. This, inturn, has changed the proportion betweensupply and demand of goods and services andhas attracted new services to the urban fringes(Brake 2005). While these service locationsallow a cross-linked production, large-scaleindustries have lost their importance (Soja2001; Fassmann 2004). Therefore, this paperfocuses on the service sector and its locationpatterns. From now on these services are

referred to as postsuburban services. Due todiverse lineage, development trajectories, andcomplexity Phelps and Parsons (2003) arguethat it is not advisable to apply North Americanconcepts to European settings. Thus, Kunz-manns (2001) less common archipelago of themetropolitan area serves as the base model. Inthis case it seems appropriate because it is theo-retically based on the European metro-politan area of Munich, having almost similareconomic, social, and demographic conditions

and spatial planning policies compared withVienna. He argues that todays metropolitanareas have a multinucleated urban pattern andconsist of several monofunctional clusters of

services. As a next step, based on Kunzmann

(2001), the postsuburban services are aggre-gated to supergroups of postsuburban services.Therefore, the Herold classification code, aclassification of economic activities, was used toestablish the four postsuburban supergroups(Figure 1).

The resulting spatial distribution, used forfurther analysis, is visualised in Figure 2.Overall, there are approximately 25,700 ser-

vices located in the Viennese fringe, about7 per cent corresponding to one of the four

postsuburban services.METHODOLOGY

Spatial statistics offers numerous techniquesfor quantifying spatial point patterns. Globalstatistics evaluates a point pattern throughoutthe whole study site. Thus, it is not possible topinpoint local spatial clusters. Jacquez (2008)noted that local cluster detection methods(e.g. Openshaw et al. 1987; Besag & Newell

1991; Kulldorff & Nagarwalla 1995) overcomethis shortcoming and they are essential forquantifying geographic patterns, without beingaffected by the well-known modifiable arealunit problem in spatial analysis (Openshaw1984). From a statistical point of view a spatialcluster is a spatially bounded group of cases ofsufficient size and concentration to be unlikelyby chance alone (Knox 1989). In this paper aspatial scan statistic approach is used to detect apolycentric structure of postsuburban services

and bivariate K(d)-functions are examinedto analyse the spatial relationships betweenthe point patterns of postsuburban services in

Viennas urban fringe.

software (114),

software devel. & service (56),internet (51),

data processing (28),computer repair & service (17),

computer network (16),computer peripherals (12),

internet service provider (11),...

advertising agency (132),advertising art & design (58),

graphic design (34),web design (31),multimedia (16),

computer graphic (1)

IT services(N= 305)

Creative services(N= 272)

Lifestyle and health services(N= 545)

Economic and legal services(N= 557)

psychotherapist (78),

psychologists (71),nutritional advice (27),

biological products (22),alternative therapy (21),

coaching & mediation (21),kinesiology (17),feng-shui (14),

...

management consultancy (183),

chartered accountant (163),investment adviser (46),

public relations (41),financial advice &

business manag. consulting (31)marketing (25),

management (17),management consulting (15), ...

Figure 1. Aggregation of individual economic activities to supergroups.




6/14

Spatial scan statistic Spatial scan statistic (SSS;Kulldorff & Nagarwalla 1995; Kulldorff 1997);is a cutting-edge and reliable local geographicalcluster detection methodology. Numerousapplications in diverse disciplines have shownthe utility and potential of this approach (e.g.Pearlet al. 2006; Leitner & Helbich forthcom-ing). Compared to earlier approaches to deter-

mine subcentres (e.g. Giuliano & Small 1991;Bogart & Ferry 1999), there is no need to setthreshold values. On the other hand, limitedexperience exists in the application on urbangeographical analysis (e.g. Montrone et al.2009; Tuiaet al. 2009).

Here, the Bernoulli-based SSS (Kulldorff1997), is applied to detect polycentric urbanforms, meaning that dichotomised data (in themanner of cases and controls) are compared.This approach enables one to take into account

heterogeneous distributed background popu-lation across space. Thus, the aim is to deter-mine significant local clusters (hot spots) of thecases (services aggregated above) compared to

the heterogeneous distributed control popula-tion. The algorithm can be briefly describedas follows: A two-dimensional circular scan

window moves from caseito casejthroughoutthe study site R and compares the observedcases inside each window according to the nullhypothesis with the expected amount of cases.To detect different sizes of clusters the scan

window continuously changes its size from zeroup to a recommended value of 50 per cent ofall cases (Kulldorff 2010), which allows thedetection of varying cluster sizes without anypre-selection bias. The reason for scanningup to such a large cluster is the lack of priorknowledge about the true cluster size. For each

window the following likelihood function iscalculated (Kulldorff 1997):

c

n

n c

n

C c

N nN n C c

N n I

c n c C c

N n C c ( )

( )

( )( ) ( )

( )

( ) ( )

O

(1)

(a) (b)

(c) (d)

IT services

District capitals

Highways

Settlement areas

Fringe (municipalities)

Districts

Lifestyle andhealth servicesDistrict capitals

Highways

Settlement areas


Districts

Economic andlegal servicesDistrict capitals

Highways

Settlement areasFringe (municipalities)

Districts

Data: Herold, Corine,

Lower AustrianGovernment, N= 305





Data: Herold, Corine,Lower Austrian

Government, N= 557

Creative services

District capitals

Highways

Settlement areas


Districts

Figure 2. Spatial distribution of the postsuburban services: (a) IT services; (b) creative services; (c) lifestyle and healthservices; and (d) economic and legal services.

44 MARCO HELBICH



7/14

where care the observed cases inside a scanwindow,Cis the total amount of cases,nis thetotal amount of cases and controls inside a scan

window and N are all cases and controls of

the data set. The most likely cluster also calledprimary cluster, maximises the maximum like-lihood ratio statistic. In comparison, secondaryclusters are clusters with reduced likelihood

values. Monte Carlo hypothesis testing is usedto determine significance.

Compared to other methods (e.g. Openshawet al. 1987; Besag & Newell 1991), SSS considersmultiple hypotheses testing and allows forcovariates adjustment. Performance tests havealreadyshownthesuitabilityofSSS,inparticular

if the cluster size and morphology is not knowna priori (Song & Kulldorff 2003). Kulldorffet al. (2004) have shown that SSS has the abilitytodetectnon-circularclusters(e.g.alongatrans-port axis) as well. Nevertheless, this issue hasdrawn some criticism. For instance, Tango andTakahashi(2005)argued that a circularwindowis not appropriate because most geographicalclusters do not follow such a circular morphol-ogy and promoted their flexibly shaped spatialscanstatistic. Finally,Kulldorffet al. (2006) refute

this by means of an empirical study and empha-sise the additional virtue of this approach.

BivariateK(d)-function A common approachused to quantify the strength and type of thespatial association between different point pat-terns is to apply the bivariate K(d)-function(Ripley 1976). Thus, one is interested in therelative relationship between two point pat-ternsiandj. Therefore, it is not necessary toaccount for spatial heterogeneity as discussed

above. Following Dixon (2002), the bivariateK(d)-function is defined as:

K E jd

ij j= [

1 number of type case within

a distance of a randomly choseentype casei ]

where l are the points per unit area (Ripley1976; Dixon 2002). One possible edge cor-rected estimator commonly used is:

, ,

K d R w i j I d d ij i j k l ik jl lk

( ) =

( ) ( ) 0 will presentsome attraction between the two point patternsat distance d, tantamount to both patternsoccurring in spatial proximity. Values ofL dij( )


8/14

fringe, namely campus21 (Brunn am Gebirge)and Concorde (Schwechat). Both advertisetheiroutstandingconnectiontoViennaandtheinternational airport, unique infrastructure

(e.g. ready-for-occupancy offices, shops, fitnesscentre), etc. The former, for instance, alsoaccommodates some global acting companies,like the Austrian corporate headquarters ofMcDonalds. Without doubt these issues mustbe interpreted as selective postsuburban devel-opments. Second, normative planning lawsdetermine where certain facilities are allowedto be located and prevent an homogenous dis-tribution. Nevertheless, these subjective inter-pretations of service clustering require some

spatial statistical analysis to formally test for thepresence of clusters.To verify the stated hypotheses of local geo-

graphic clusters of postsuburban services in theViennese urban fringe, resulting in a polycen-tric urban landscape, SSS was used to scan forareas with a high amount of cases. In addition

to cluster detection this method assigns eachcluster some likelihood value and allows forinference. Thus, a control point pattern wasneeded for each point pattern of postsuburban

services to adjust for the a priori heterogeneousdistributed background population acrossspace. For this purpose, random samples of 20per cent of all 23,983 service locations in thestudy area were determined, excluding thefour supergroups that were previously selected.Sampling was necessary to keep the computa-tion feasible. For Monte Carlo inference, 9,999replications were performed. In this studythe null hypothesis of no spatial cluster wasrejected when the simulated p-value was less

than 0.05.Figure 3 shows the locations of detected clus-ters for every supergroup, whereas each clusteris labelled with its cluster category on the basisof its significance values. Every pattern has onehighly significant primary cluster (C1), mainlylocated in the south of Vienna, meaning that

(a)

Data: Herold, Corine,Lower Austrian Government


Lower Austrian Government





Primary clusterp< 0.001Secondary cluster

p= 0.068Secondary clusternot signif.District capitals

Highways

Settlement areas


Districts

Primary clusterp< 0.001Secondary clusternot signif.

District capitals

Highways

Settlement areas


Districts

Primary clusterp< 0.001Secondary clusternot signif.

District capitals

Highways

Settlement areas


Districts

Primary clusterp= 0.002Secondary cluster

not signif.District capitals

Highways

Settlement areas


Districts

(b)

(c) (d)

Figure 3. Spatial clusters of the postsuburban services: (a) IT services; (b) creative services; (c) lifestyle and health services;and (d) economic and legal services.

46 MARCO HELBICH



9/14

there are more postsuburban services thanother services from the control population. Allsecondary clusters are not significant atp< 0.05(IT services have ap-value approximately equal

to the critical value, Fig. 3a).Additionally, Table 1 presents some clusterdetails and summarises their properties. Theconsiderable differences in the cluster sizes arestriking. For instance, the small primary clusterof IT services consists of 25 objects and its coun-terpart the primary cluster of the economicand legal service supergroup has nearly 15times more objects. Moreover, Figure 3 refersto some geographical overlap of the postsubur-ban clusters in the southern area of the urban

fringe, which could be a first note on somespatial relationships.To quantify this possible spatial association

between the four point patterns bivariateK(d)-functions are applied. To evaluate the spatialrelationship, the L dij( )-functions are plottedagainst a distanced(m) (Figure 4). Significance

was again obtained by 499 Monte Carlo simu-lations.3 These simulation runs create criticalenvelops. Values around L dij( ) =0 (nullhypothesis) provide a reference for indepen-

dence of the two spatial processes.Overall, all four plots show positive devia-tions from zero, tantamount to spatial attrac-tion. For most distances up to five kilometresthe estimated functions are above the confi-dence envelopes. Thus, the null hypothesismust be rejected. This means that the numberof IT services available in the neighbourhood ofcreative services (or equivalently the number ofcreative services in the neighbourhood of ITservices), is significantly higher than expected.

This interpretation is true for all other plots inFigure 4. Beyond a distance of five kilometresthe patterns are still attracted to each otherbut no longer significant up to the analysed

distance of 20 kilometres. The comparisonbetween these six bivariate L dij( )-functionssuggests differences in the strength of the rela-tionships. The strongest attraction is observable

between the lifestyle and health services versusthe economic and legal services. The finalsection summarises findings from this researchand draws conclusions and further implicationsfor the theoretical discourse.

DISCUSSION

Today metropolitan areas are facing pervasivechanges of their urban spatial structure and arereshaped by spatial processes. This paper exam-

ined the spatial structure of the Viennese urbanfringe as a case study based on the distributionof specific services. On the one hand, localspatial clusters are investigated and on theother hand, their spatial interrelations arequantified applying spatial statistics.

First it analysed the current spatial formationof services and addressed the question, whetherthere are multinucleated monofunctionalclusters of higher-order services, which canbe interpreted as a polycentric pattern. This

state would be a characteristic of postsuburbandevelopment processes (Kunzmann 2001;Brake 2005). On this basis, the second partexamined these postsuburban clusters to see

whether they have already become some newurban landscape and evaluated the spatial rela-tionship between the detected monofunctionalservice clusters. Hence, this study follows Langet al. (2009) who investigated US metropolises,to see if the observable spatial patterns ofoffices are already beyond edge cities and thus

consequently beyond postsuburbanisation pro-cesses. The first section argues that the citedterms such as edge city, postsuburbia or exopo-lis highlight a maturity process (e.g. functional

Table 1. Statistics of the mapped primary clusters (C1).

Numberof cases

Expectedcases

Observedexpected ratio

Log likeli-hood ratio

p-value

IT services 25 5.160 4.840 22.801


10/14

enrichment of services, spatial reorganisation)of the urban landscapes referring to a postmod-ern society (Dear & Flusty 2002). The method-ological framework uses micro-geographic dataof service locations for the year 2006 and spatialstatistics, namely the spatial scan statistic andthe bivariate K(d)-function, to answer theseresearch questions.

The findings show that the urban fringe ofVienna consists of a polycentricurbanstructure.

Allresultsindicatesignificantfunctionalclustersbecause analysed services are clearly more clus-tered than other services in the surroundings.In detail, every point pattern has one highly

significantprimarycluster,mostlysituatedinthesouthernpartofthecorecity,whereastheclustersizes vary. Most of the primary clusters (e.g. cre-ative services), have a spatial range extendingover several administrative units, thus allowingus to conclude that there are no specific loca-tional policies. Compared with another study,the detected clusters aremainly located in tradi-tional suburban areas and reshaping them, ascomprehensively analysed in Fassmann et al.

(2009). Thus, the cluster locations offer highaccessibilitytoVienna.Theresultsareconsistentwith those of Eisenreich and Schenk (2002) forthe German Filderregion next to Stuttgart and

0 5.000 10.000 15.000 20.000

0

2.

000

6.

000

IT services vs. Creative services

d(m)

L

(d)

^12

L (d)-functionConfidence envelopes

^12

0 5.000 10.000 15.000 20.000

0

2.

000

6.

000

IT services vs. Lifestyle services

d(m)

L

(d)

^12

0 5.000 10.000 15.000 20.000

0

2.

000

6

.000

IT services vs. Economic services

d(m)

L

(

d)

^12

0 5.000 10.000 15.000 20.000

0

2.

000

6.

000

Creative services vs. Lifestyle services

d(m)

L

(

d)

^12

0 5.000 10.000 15.000 20.000

0

2.

000

6

.000

Creative services vs. Economic services

d(m)

L

(d

)

^12

0 5.000 10.000 15.000 20.000

0

2.

000

6.

000

Lifestyle services vs. Economic services

d(m)

L

(d

)

^12





L (d)-functionConfidence envelopes12

^12

^12

^12

^ ^12

Figure 4. Estimated bivariate L dij( )-functions of the postsuburban services. Dashed lines represent the simulatedconfidence envelopes and solid lines represent the empirical functions (headings are abbreviated).

48 MARCO HELBICH



11/14

those of Kagermeieret al. (2001) for the metro-politanareaofMunich.Thesenewclustershavedeveloped out of existing clusters and have not

yet created new economic centres. This leads to

the conclusion that there are specific locationalqualities in these cluster locations. From aneconomic geography point of view this meansthat space still matters and, followingMalmbergand Maskell (2002), density can boost localisedlearning and the innovation process. In particu-lar, agglomeration economies like face-to-facecontacts and the creation, exchange, and thediffusionofideasandknowledgeareessentialinsuch economies (Storper & Venables 2004). Bycontrast, telecommunication technologies still

lack the ability to soften traditional agglome-ration economics (Coffey & Shearmur 2002).Using above point pattern methodology, thisstudygenerallydoesnotallowseriousstatementsabouttherelationshipbetweenthecorecityandthe fringe and, specifically, no conclusions canbe drawn about the importance of the core city.Therefore, spatiotemporal interaction studiesare still required.

Summing up, the hypothesis of monofunc-tional service agglomerations can be verified by

reference to a postsuburban spatial organisa-tion of the Viennese urban fringe. Further,these findings converge with Burdack andHerfert (1998) and Phelpset al. (2006) in theirclaim that future European metropolitan areas

will be characterised by a polycentric urbanstructure and disconfirm some evolutionarydevelopment towards a diffuse urban fabric asproposed by Gordon and Richardson (1996)and Lang (2003).

The second part addressed the spatial

relationship between these monofunctionalclusters and evaluated the hypothesis of urbanstructures beyond postsuburbia. Applyingseveral bivariateK(d)-functions, the study pro-

vided evidence for spatial interaction betweentheir occurrences and are not independent ofeach other. Thus, clusters aspire spatial prox-imity next to clusters of another category. Withregard to postsuburban processes this is hith-erto not known for European fringes. This factis not accounted for in Kunzmanns (2001)

archipelago, which categorically excludes inter-action effects, using crisp cluster delimitations.More appropriate would be conglomerates ofclusters of postsuburban services. Those are

more suitable for the Viennese case as well.Finally, it should be mentioned that the clusterdetection algorithm used generates per se crispinstead of fuzzy cluster demarcations, which

does not correspond to most real-world phe-nomena as stated in Leung (1987). A constraintworth mentioning in this study are the sam-pling procedures, inducing some selectionbias. Nevertheless, a further development isassessed, which deviates from our base model,the archipelago of the metropolitan area. Com-pared to Kunzmann (2001), this partial resultrejects the thesis of specific European postsub-urban landscape styles and resembles theirNorth American counterparts of mixed service

clusters. Proposing a paradigm shift for theViennese surroundings from a postsuburbandevelopment to something new beyond post-suburbia would currently be premature.

Future research is needed to see whetherthe findings reported here are robust across a

wider range of metropolitan areas. Such studiescould address the spatial and spatiotemporaldevelopment of service clusters, their morpho-logical morphing, and how the centres andtheir specialisation evolve over time. In addi-

tion, comparative studies with other Europeanmetropolitan areas are of great theoretical andpractical importance to find out whether theresults from Vienna are unique or can be gen-eralised to other metropolitan areas. It is hopedthat this study will stimulate further investiga-tion in this field. Possibly further work alongthis line will be fruitful.

Acknowledgements

I acknowledge the valuable feedback I have received

from Alois Humer and Peter Grgl on successivedrafts of this paper. In addition, I want to thankAlexander Zipf for helping me obtain the post-doctoral position at the University of Heidelberg,making it possible to prepare this publication.Further, I would also like to acknowledge the con-structive comments and feedback from the reviewers.Finally, I want to dedicate this paper to Daniela forher never-ending patience.

Notes

1. Current research suggests that Marshallianstheory of agglomeration may be insufficient forexplaining polycentric urban forms and that




12/14

polycentricity is not simply a matter of agglo-meration economies producing agglomeration(Krugman 1996, p. 88 cited in Phelps 2004, p.978). For a detailed discussion see Phelps and

Ozawa (2003) and Phelps (2004).2. After Garreau (1992) an edge city consists of450,000 m2 office space, 55,000 m2 retail space,and 24,000 employees.

3. The 499 runs used are a moderate number ofsimulations, but reduce computation time consid-erably. I therefore advise the reader to interpretthe results, especially the significant ones, withsome caution.

REFERENCES

Anas, A., R. Arnott & K. Small (1998), UrbanSpatial Structure.Journal of Economic Literature36,pp. 14261464.

Baumont, C., C. Ertur & J . Le Gallo (2004),Spatial Analysis of Employment and PopulationDensity: The Case of the Agglomeration of Dijon.Geographical Analysis36, pp. 146176.

Besag, J. (1977), Contribution to the Discussion ofDr. Ripleys Paper. Journal of the Royal StatisticalSociety B39, pp. 193195.

Besag, J . & J. Newell (1991), The Detection ofClusters in Rare Diseases.Journal of the Royal Statis-tical Society A154, pp. 143155.

Blotevogel, H. (1998), The Rhine-Ruhr Metro-politan Region: Reality and Discourse. EuropeanPlanning Studies6, pp. 395410.

Bkemann, D. (1999), Theorie der Raumplanung:Regionalwissenschaftliche Grundlagen fr die Stadt-,Regional- und Landesplanung. Mnchen: Olden-bourg Verlag.

Bogart, W . & W . Ferry (1999), Employment

Centres in Greater Cleveland: Evidence of Evolu-tion in a Formerly Monocentric City.Urban Studies36, pp. 20992110.

Bontje, M . & J . Burdack (2005), Edge Cities,European-style: Examples from Paris and theRandstad.Cities22, pp. 317330.

Brake, K. (2005), Der Suburbane Raum als Standort-typ.In: K. Brake, I. Einacker& H. Mding, eds.,Krfte, Prozesse, Akteure zur Empirie der Zwischen-stadt. Zwischenstadt Band 3, pp. 965. Wuppertal:Mller +Busmann.

Burdack, J. & G. Herfert(1998), Neue Entwick-lungen an der Peripherie europischerGrostdte. Ein berblick. Europa Regional 6,pp. 2644.

Castells,M.(1989), Information Technology, EconomicRestructuring, and the Urban-Regional Process.Oxford: Blackwell.

Clark, W. (2002), Monocentric to Polycentric: New

Urban Forms and Old Paradigms. In: G. Bridge&S. Watson, eds., Companion to the City, pp. 141154.Oxford: Blackwell.

Coffey, W. & R. Shearmur(2002), Agglomerationand Dispersion of High-order Service Employmentin the Montreal Metropolitan Region, 198196.Urban Studies39, pp. 359378.

Cuthbert, A. & W. Anderson(2002), Using SpatialStatistics to Examine the Pattern of Urban LandDevelopment in Halifax-Dartmouth. The Profes-sional Geographer54, pp. 521532.

Dear, M. & S. Flusty (2002), The Resistible Riseof the L.A. School. In: M. Dear, eds.,From Chicagoto L.A. Making Sense of Urban Theory, pp. 616.Thousand Oaks, CA: Sage Publications.

Dieleman, F. & A. Faludi (1998), Poly-nucleated Urban Regions in Europe: Theme ofthe Special Issue. European Planning Studies 6,pp. 365378.

Dixon, P. (2002), Ripleys K Function. In: A.El-Shaarawi& W. Piegorsch, eds.,Encyclopediaof Environmetrics, 3, pp. 17961803. Chichester:

Wiley.Eisenreich, D. & W. Schenk(2002), Abkopplung-stendenzen des suburbanen Raums von der Kern-stadt das Beispiel der Filderregion sdlich vonStuttgart. Standort Zeitschrift fr Angewandte Geo-graphie26, pp. 173179.

European Commission (1999), ESDP EuropeanSpatial Development Perspective. Towards Balancedand Sustainable Development of the Territoryof the European Union. Luxembourg: Officefor Official Publications of the European

Communities.Faludi, A. (2002), European spatial planning. Cam-bridge: Lincoln Institute of Land Policy.

Fassmann, H. (2004), Stadtgeographie 1. Braunsch-weig: Westermann.

Fassmann, H., P. Grgl & M. Helbich (2009),Atlasder Wachsenden Stadtregion. PGO: Wien.

Feser, E . & S . Sweeney (2000), A Test for theCoincident Economic and Spatial Clustering ofBusiness Enterprises.Journal of Geographical Systems2, pp. 349373.

Fishman, R. (1987), Bourgeois Utopias. The Rise andFall of Suburbia. New York: Basic Books.Fuchs, I. (1997), Stadtregionen 1991 das Konzept.

Statistische Nachrichten2, pp. 7683.

50 MARCO HELBICH



13/14

Garreau, J. (1992),Edge City: Life on the New Frontier.New York: Anchor Books.

Giuliano, G. & K. Small(1991), Subcenters in theLos Angeles Region. Regional Science and Urban

Economics21, pp. 163182.Gordon, P. & H. Richardson(1996), Beyond Poly-centricity: The Dispersed Metropolis, Los Angeles,19701990.Journal of the American Planning Associa-tion62, pp. 289295.

Hall, P. (1993), Forces Shaping Urban Europe.Urban Studies30, pp. 883898.

Harris, R. & P. Larkham(1999),Changing Suburbs:Foundation, Form and Function. London: E & FNSpon.

Helbich, M. & M. Leitner(2009), Spatial Analysis

of the Urban-to-rural Migration Determinants inthe Viennese Metropolitan Area. A Transitionfrom Suburbia to Postsuburbia? Applied SpatialAnalysis and Policy2, pp. 237261.

Helbich, M. & M. Leiter (2010), PostsuburbanSpatial Evolution of Viennas Urban Fringe:Evidence from Point Process Modeling. UrbanGeography31, pp. 11001117.

Hesse, M. (2001), Mobilitt und Verkehr in (Post-)Suburbia ein Ausblick. RaumPlanung 95, pp.6569.

Jacquez, G.M. (2008), Spatial Cluster Analysis. In:S. Fotheringham & J. Wilson, eds.,The Handbookof Geographic Information Science, pp. 395416.Malden, MA: Blackwell Publishing.

Kagermeier, A. , M. Miosga & K . Schussmann(2001), Die Region Mnchen Auf dem Weg zuregionalen Patchworkstrukturen.In: K. Brake, J.Dangschat& G. Herfert, eds.,Suburbanisierungin Deutschland. Aktuelle Tendenzen, pp. 163173.Opladen: Leske +Budrich.

Kling, R . , S . Olin & M . Poster (1995), The

Emergence of Postsuburbia: An Introduction.In:R. Kling, S. Olin& M. Poster, eds.,PostsuburbanCalifornia. The Transformation of Orange County sinceWorld War II, pp. 130. Berkley, CA: University ofCalifornia Press.

Kloosterman, R . & S . Musterd (2001), ThePolycentric Urban Region: Towards a ResearchAgenda.Urban Studies38, pp. 623633.

Knox, E.G. (1989), Detection of Clusters. In: P.Elliott, ed., Methodology of Enquiries Into DiseaseClustering, pp. 1720. London: Small Area Health

Statistics Unit.Kulldorff, M. (1997), A Spatial Scan Statistic.Communications in Statistics: Theory and Methods26, pp. 14811496.

Kulldorff, M. (2010), SaTScanTM User Guidefor version 9.0. Available at.

Kulldorff, M. & N. Nagarwalla (1995), Spatial

Disease Clusters: Detection and Inference. Statis-tics in Medicine14, pp. 799810.Kulldorff, M., Z. Zhang, J. Hartman, R.

Heffernan, L. Huang & F. Mostashari(2004), Evaluating Disease Outbreak DetectionMethods: Benchmark Data and Power Calcu-lations. Morbidity and Mortality Weekly Report 53,pp. 144151.

Kulldorff, M., L. Huang, L. Pickle& L. Duczmal(2006), An Elliptic Spatial Scan Statistic.Statisticsin Medicine25, 39293943.

Kunzmann, K. (2001), Welche Zuknfte fr Subur-bia? Acht Inseln im Archipel der Stadtregion.In: K. Brake, J. Dangschat& G. Herfert, eds.,Suburbanisierung in Deutschland. Aktuelle Tendenzen,pp. 213221. Opladen: Leske +Budrich.

Lang, R. (2003), Edgeless Cities: Exploring the ElusiveMetropolis. Washington, DC: Brookings InstitutionPress.

Lang, R . & J . LeFurgy (2003), Edgeless Cities:Examining the Non-centred Metropolis. HousingPolicy Debate14, pp 427460.

Lang, R., T.W. Sanchez & A.C. Oner (2009),Beyond Edge City: Office Geography in the NewMetropolis.Urban Geography30, pp. 726755.

Lee, B. (2007), Edge or Edgeless Cities? UrbanSpatial Structure in U.S. Metropolitan Areas, 1980to 2000.Journal of Regional Science47, pp. 479515.

Leitner, M. & M. Helbich (forthcoming), TheImpact of Hurricanes on Crime: A Spatio-temporalAnalysis in the City of Houston, TX. Cartographyand Geographic Information Science37.

Leung, Y. (1987), On the Imprecision of Boundaries.

Geographical Analysis19, pp. 125151.Malmberg, A. & P. Maskell (2002), The ElusiveConcept of Localization Economies: Towards aKnowledge-based Theory of Spatial Clustering.Environment and Planning A34, pp. 429449.

Maoh, H. & P. Kanaroglou (2007), GeographicClustering of Firms and Urban Form: A Multi-variate Analysis. Journal of Geographical Systems9,pp. 2952.

McMillen, D. (1996), One Hundred Fifty Yearsof Land Values in Chicago: A Nonparametric

Approach.Journal of Urban Economics40, pp. 100124.McMillen, D. & J. McDonald (1998), Suburban

Subcenters and Employment Density in Metro-




14/14

politan Chicago. Journal of Urban Economics 43,pp. 157180.

Mills, E.S. (1970), Urban density functions. UrbanStudies7, pp. 520.

Montrone, S., P. Perchinunno, A. Di Giuro, F.Rotondo& C.M. Torre(2009), Identification ofHot Spots of Social and Housing Difficulty inUrban Areas: Scan Statistics for Housing Marketand Urban Planning Policies. In: B. Murgante,G. Borruso, A. Lapucci, eds.,Geocomputation &Urban Planning, Studies in Computational Intelligence,Volume 176, pp. 5778. Berlin: Springer.

Muller, P. (1997), The Suburban Transformationof the Globalizing American City. Annals of theAmerican Academy of Political and Social Science551,

pp. 4458.Musil, R. & P. Pindur (2008), Nachhaltig trotzsuburbanen Nutzungsdrucks? Siedlungsentwick-lung im Biosphrenpark Wienerwald.Mitteilungender sterreichischen Geographischen Gesellschaft 150,pp. 121145.

Openshaw, S. (1984),Modifiable Areal Unit Problem.Norwich: GeoBooks.

Openshaw, S., M. Charlton, C. Wymer & A.W.Craft(1987), A Mark 1 Analysis Machine for theAutomated Analysis of Point Data Sets. Interna-

tional Journal of Geographical Information Systems1,pp. 335358.Pearl,D.L.,M.Louie, L . Chui, K . Dor,K.M.Grim-

srud, D. Leedell, S.W. Martin, P. Michel, L.W.Svenson& S.A. McEwen(2006), The Use of Out-break Information in the Interpretation of Cluster-ing of Reported Cases of Escherichia Coli O157 inSpace and Time in Alberta, Canada, 20002002.Epidemiology and Infection134, pp. 699711.

Phelps, N. (2004), Clusters, Dispersion and theSpaces in Between: For an Economic Geography

of the Banal.Urban Studies41, pp. 971989.Phelps, N . & T . Ozawa (2003), Contrasts inAgglomeration: Proto-industrial, Industrial andPost-industrial Forms Compared. Progress inHuman Geography27, pp. 583604.

Phelps, N . & N . Parsons (2003), Edge UrbanGeographies: Notes from the Margins of EuropesCapital Cities.Urban Studies40, pp. 17251749.

Phelps, N., N. Parsons, D. Ballas, A. Dowling(2006),Post-suburban Europe. Planning and Politicsat the Margins of Europes Capital Cities. New York:

Palgrave Macmillan.

Riguelle, F., I. Thomas & A. Verhetsel (2007),Measuring Urban Polycentrism: A European CaseStudy and its Implications. Journal of EconomicGeography7, pp. 193215.

Ripley, B. (1976), The Second-order Analysis ofStationary Point Processes.Journal of Applied Prob-ability13, pp. 255266.

Rowlingson, B. & P. Diggle (1993), SPLANCS:Spatial Point Pattern Analysis Code in S-Plus.Com-puters and Geosciences19, pp. 627655.

Schwanen, T., M. Frans, F. Dieleman& M. Dijst(2001), Travel Behaviour in Dutch Monocentricand Policentric Urban Systems.Journal of TransportGeography9, pp. 173186.

Shearmur, R., W. Coffey, C. Dub& R. Barbonne

(2007), Intrametropolitan Employment Structure:Polycentricity, Scatteration, Dispersal and Chaosin Toronto, Montreal and Vancouver, 19962001.Urban Studies44, pp. 17131738.

Sieverts, T. (1998),Zwischenstadt. Zwischen Ort undWelt, Raum und Zeit, Stadt und Land. Braunschweig:Vieweg.

Soja, E. (2000),Postmetropolis. Critical Studies of Citiesand Regions. Oxford: Blackwell.

Soja, E. (2001), Exploring the Postmetropolis. In: C.Minca,ed., Postmodern Geography: Theory and Praxis,

pp. 3756. Oxford: Blackwell.Song, C. & M. Kulldorff(2003), Power Evaluationof Disease Clustering Tests.International Journal ofHealth Geographics2, 9.

Storper, M. & A. Venables (2004), Buzz: Face-to-face Contact and the Urban Economy. Journal ofEconomic Geography4, pp. 351370.

Tango, T . & K . Takahashi (2005), A FlexiblyShaped Spatial Scan Statistic for Detecting Clus-ters. International Journal of Health Geographics 4,411.

Tuia, D., C. Kaiser, A. DaCunha& M. Kanevski(2009), Detection of Urban Socio-economicPatterns Using Clustering Techniques. In: B.Murgante, G . Borruso, A . Lapucci, eds., Geocom-putation & Urban Planning, Studies in ComputationalIntelligence, Volume 176, pp. 1936. Berlin:Springer.

Vaughan, L., S. Griffiths, M. Haklay& C. Jones(2009), Do the suburbs exist? Discovering Com-plexity and Specificity in Suburban Built Form.Transactions of the Institute of British Geographers34,

pp. 475488.

52 MARCO HELBICH


j.1467-9663.2011.00673.x Beyond Postuburbia_ Multifunctional Service Agglomeration in Vienna Urban...

Documents

Transcript of j.1467-9663.2011.00673.x Beyond Postuburbia_ Multifunctional Service Agglomeration in Vienna Urban...