www.flacsoandes.edu.

ec
 CHAPTER 8

THE ROLE OF
INDUSTRIAL
CLUSTERING AND
INCREASING RETURNS
TO SCALEIN ECONOMIC
DEVELOPMENT AND
URBAN GROWTH

PH ILIP M cC A N N

I n t r o d u c t io n

T h e size o f its cities is a cru cia l issue for the ab ility o f an e c o n o m y to p ro m o te

e co n o m ic g ro w th and d e ve lo p m e n t. T h e evid en ce in favor o f the role played b y
a g glo m era tio n e co n o m ie s in p ro m o tin g g ro w th is n o w so o v e rw h elm in g th at it
is really b e yo n d debate (V enables 2006). H ow ever, there are v a rio u s d istin ct
aspects to the roles played b y cities in p ro m o tin g gro w th . First, large cities are
associated w ith a h igh level o f m arket p o ten tial, in term s o f the size and accessi-
b ility o f m arkets a ffo rd ed to a p a rticu la r firm at a p a rticu la r lo c a tio n . Seco n d ,
such large h o m e m arket effects associated w ith cities are also im p o rta n t in
i68 UR BA N S T R U C T U R E , G ROW TH , AND T H E D E V E L O P M E N T P R O C E SS

p ro m o tin g eco n o m ie s o f scale th at facilitate the ca p tu rin g o f m arkets in o th e r

lo ca tio n s. T h ird , cities facilitate the v a rio u s e xtern alities and sp illo ve r m ech a -
n ism ch a racteristics o f a g g lo m e ra tio n e co n o m ie s, in c lu d in g , sh a rin g , so rtin g ,
m a tch in g (D u ra n to n and Puga 2004), c o m p e titio n v ia m u tu a l tra n sp a re n cy
(P o rter 1990), and c o o p e ra tio n via p ro x im ity (S co tt 1988). T h e litera tu re o n these
issues is n ow ad ays vast, so w e d o n o t n eed to dw ell o n th is fo r the p u rp o ses o f
this chapter. Suffice to say that the stra ig h tfo rw a rd o r th o d o x th e o re tica l a g g lo m ­
eratio n argu m en ts tend to im p ly th at cities sh o u ld sim p ly be a llo w ed to exp a n d
as the m arket requires, in o rd er to m a xim ise the e ffic ie n cy and w elfare gains
associated w ith a g g lo m era tio n effects.
If this is indeed the m ost appropriate interp retation o f the theory, then regional
plan n in g sim ply reduces to a fo rm o f cartography, in w h ich urban bo u n d aries are
adjusted as it becom es profitable to transfer rural land at the urban ff inge to urban
land uses. U nder this scenario the o n ly role o f regional p lan n in g therefore is to
ensure that sufficient quantities o f land are m ade available in the correct locations.
A s such, the o n ly problem here is to determ ine exactly w h at w e m ean b y “sufficient”
land and “correct” locations, b u t this is straightforw ard as the m arket signáis should
determ ine these issues.
In contrast, the argum ents p u t forw ard in this chapter are that the links b etw een
urban scale, econ o m ic grow th and d evelopm ent, and regional p lan n in g are far fforn
straightforw ard.
T he chapter is organized as follow s. In the next section I ad op t a transactions-
costs perspective in order to briefly review the different types o f industrial clusters
that exist. As w e shall see, they differ significantly in term s o f their location al logic,
the characteristics o f the interfirm know led ge exchanges, the firm size features, and
the objectives o f the firm s located in these variou s clusters. In the third section I
explore h o w cities have grow n d u rin g the tw entieth century. W e w ill see that at the
turn o f the tw entieth century, w h en A lfred M arshall w as w ritin g, the relationship
betw een cities and national econ o m ic developm en t was very sim ple. Large cities
m eant w ealthy econom ies. As such, urban izatio n and in d ustrialization were tw o
sides o f the sam e coin. H owever, this relationship changed dram atically d u rin g the
tw entieth century. A s w e shall see in the fo u rth section, b y the tu rn o f the tw enty-
first century, urban scale is n ot a sim ple ind icator o f u rb an productivity. In m an y
parts o f the w orld the m ost productive cities are not the largest. In these locations
city p ro d u ctivity appears to be driven far m ore by global co n n ectivity than b y scale.
As such, this fact poses m ajor challenges to regional plan n in g strategies based sim ­
ply on rural-urban land release strategies. In each case the appropriate regional
planning strategies can best be inferred w ith recourse to the industrial cluster types
discussed in the second section. T hese regional plan n in g issues are all discussed in
the fifth section. It w ill be argued that tw en ty-first-cen tu ry plan n in g strategies in
m any countries w ill increasingly involve the p ro m o tio n o f global connectivity, and
this tendency tow ard polycentric urban systems w ill com e to d o m ín ate land-use
planning strategies.
TH E ROLE OF I N D U S T R I A L C L U S T E R IN G AND IN C R E A S IN G R E T U R N S 169

D if f e r e n t F o r m s o f I n d u s t r ia l C l u s t e r i n g :
A g g l o m e r a t io n , C o m p l e x e s , a n d S o c ia l
N etw orks

All concentrations o f e co n o m ic activity and industrial clusters fall into three b roadly
defined stylized categories, each o f w hich exhibits a distinct set o f know led ge
exchanges and interfirm relational features (G o rd on and M cC an n 2000; M cC an n
and Sheppard 2003). T hese three distinct types o f industrial clusters are the
agglomeration, the industrial ,co
lex and the social
p
m T h e character
each o f the cluster types are Usted in table 8.1.
In the m odel o f puré ,a
ertininterfirm relations are inherently transient.
m
lo
g
Firm s are essentially atom istic, in the sense o f having alm ost no m arket power, and
they w ill co n tin u ou sly change their relations w ith other firm s and custom ers in
response to m arket arbitrage oppo rtu n ities, thereby leading to intense local com pe-
tition. As such, there is no loyalty betvveen firm s, ñor are any particular relations
long-term . T h e external benefits o f clustering accrue to all local firm s sim ply by
reason o f their local presence. T h e cost o f m em bership o f this cluster is sim ply the

Table 8.1 In d u strial C lusters: A T ran sactio n s-C o sts Perspective

C h a r a c te r is tic s P u ré A g g lo m e r a tio n In d u s tria l C o m p le x S o cia l N e tw o r k

F irm size atomistic some firms are large variable

C h a ra c te ris tic s o f nonidentifiable ¡dentifiable trust
re la tio n s fragmented stable trading loyalty
unstable joint lobbying
joint ventures
nonopportunistic
M em b ersh ip open closed partially open
A ccess to clu ste r rental payments ¡nternal investment history
location necessary location necessary experience
location necessary but
not sufficient
S p ace o u tco m e s rent appreciation no effect on rents partial rental
capitalization
N o tio n o f sp ace urban local but not urban local but not urban
E xa m p le o f clu ste r competitive urban Steel or Chemicals new industrial areas
economy production complex
A n a ly tic a l models of puré location-production social network theory
a p p ro a ch e s agglomeration theory
input-output analysis
170 U R BA N S T R U C T U R E , G ROW TH , AND T H E D E V E L O P M E N T P R O C E S S

local real estáte m arket rent. T here are no free riders, access to the cluster is open,
and consequently it is the grow th in the local real estáte rents that is the in d icator o f
the cluster’s perform ance. T h is idealized typ e is best represented b y the M arshall
(1890, 1920) m odel o f agglom eration and is the n o tio n o f clustering un d erlyin g
m odels o f new econ o m ic geograph y (K rugm an 1991a, 1991b; Fujita, K ru gm an n , and
Venables 1999). T he n o tion o f space in these m odels is essentially urban space in
that this typ e o f clustering exists o n ly w ith in in d ivid ual cities.
T he industrial complex is characterized p rim arily b y lo n g -term stable and pre­
d ic a b le relations betw een the firm s in the cluster. T h is typ e o f cluster is m ost co m -
m o n ly observed in industries such as Steel and Chemicals and is the typ e o f spatial
cluster typ ically discussed b y classical (W eber 1909) and neoclassical (M oses 1958)
lo catio n -p ro d u ctio n m odels, representing a fusión o f location al analysis w ith in put-
o u tp u t analysis (Isard and Kuenne 1953). C o m p o n en t firm s w ith in the spatial
grou pin g each undertake significant lo n g -term investm ents, particu larly in term s o f
physical capital and local real estáte, in order to becom e part o f the gro u p in g. Access
to the group is therefore severely restricted b y high e n try and exit costs, and the
rationale for spatial clustering in these types o f industries is that p ro xim ity is
required prim arily to m inim ize interfirm tran sp ort transactions costs. Rental appre-
ciation is not a feature o f the cluster because the land that has already been pur-
chased by the firm s is not for sale. T h e n o tio n o f space in the industrial co m p lex is
local, b u t not necessarily urban, in that these types o f com plexes can exist either
w ith in or outside o f an individual city. T his co m p lex m od el is actually the single
explicitly spatial elem ent in the transactions-costs approach o f W illiam son (1975),
where the focus is o n the types o f flow -process scale econ om ies that firm s can real-
ize by being part o f vertically integrated p ro d u ctio n com plexes.
T he third type o f spatial industrial cluster is the social network m odel. A lth o u gh
this has becom e extrem ely popular in research, it is not directly related to the previ-
ous research traditions o f location theory. T his cluster typ e is associated prim arily
w ith the w o rk o f G ranovetter (1973) and is a response to the hierarchies m odel o f
W illiam son (1975). The social netw ork m odel argües that m utual trust relations
between key decision-m aking agents in different organizations m ay be at least as
im portant as decision-m akin g hierarchies w ith in individual organizations. These
trust relations w ill be m anifested b y a variety o f features, such as jo in t lobbying, jo in t
ventures, inform al alliances, and reciprocal arrangem ents regarding trading relation-
ships. However, the key feature o f such trust relations is an absence o f o pportun ism ,
in that individual firm s w ill not fear reprisals after any reorganization o f interfirm
relations. Interfirm cooperative relations m ay therefore differ significantly ffo m the
organizational boundaries associated w ith individual firm s, and these relations m ay
be continually reconstituted. All o f these behavioral features rely on a co m m o n cu l­
ture o f m utual trust, the developm ent o f w hich depends largely on a shared history
and experience o f the decision-m akin g agents. T h is social netw ork m odel is essen­
tially aspatial, but ffo m the point o f view o f geography, it can be argued that spatial
proxim ity will tend to foster such trust relations, thereby leading to a local business
environm ent o f confidence, risk taking, and cooperation. Spatial proxim ity is neces-
THE ROLE OF IN D U S T R IA L C L U S T E R IN G AND IN C R EA SIN G R E T U R N S 171

sary but not sufficient to acquire access to the netw ork. As such, m em bership o f the
netw ork is on ly partially open, in that local rental paym ents w ill not guarantee access,
although they w ill im prove the chances o f access. T he geographic m anifestation o f
the social netw ork is the so-called new industrial areas m odel (Scott 1988), w hich has
been used to describe the characteristics and perform ance o f areas such as Silicon
Valley and the E m ilia-R om agna región o f Italy. In this m odel space is once again
local but not necessarily urban.
In reality, all spatial clusters w ill o b vio u sly contain characteristics o f one o r
m ore o f the ideal types described here, although one type w ill tend to be d o m in an t
in each observed cluster. T herefore, in order to understand the advantages to the
firm o f being located in any particular cluster, it is first necessary to determ ine w h ich
o f these ideal cluster types m ost accurately reflects the overall characteristics and
behavior o f the firm s in the cluster. However, sim ply o bservin g the clustering o f
econ o m ic activity is not sufficient evidence for the existence o f a cluster, if we con-
sider a cluster in term s o f local know ledge-related or transactions-related agglom -
eration effects, and there are tw o reasons for this. O n e reason is that the d istribution
o f activities across space is ran d o m (Ellison and Glaeser 1997), and the result o f this
random ness is that som e activities w ill appear clustered w hile others w ill appear
dispersed, even th ough there are no differences in the interactions betw een firm s.
O bservation s o f spatial industrial co n cen tration are not necessarily evidence o f
Porter-type clusters. Second, localized clustering m ay sim ply be a rational response
to the spatial organ ization o f industrial hierarchies, w ith ou t any recourse to local
interactions betw een the apparently clustered firm s (M cC an n 1995). Therefore,
understanding the m echanics o f a cluster requires a m ore detailed analysis.
T he co m m o n central elem ent in each o f these cluster types is that know ledge is
localized, while the differences betw een the cluster types relate to how that k n o w l­
edge is locally transm itted and appropriated and the spatial extent over w hich this
operates. In the agglom eration m odel know ledge flow s as tacit inform al spillovers
w ithin an explicitly urban setting; in the industrial com plex know ledge is an asset
that is transferred via form al long-term contracts w ithin a local regional setting; in
the social netw ork m odel, know ledge is transm itted via inform al social netw ork ties
over a local regional spatial scale. In the sim ple Porter (1990) m odel, the critical g eo ­
graphic dim ensión over w hich any such (inform ation) com petitive advantage is
assum ed to opérate is never specified. T his is problem atic because there is m uch
em pirical evidence to suggest that inform ation spillovers in som e dynam ic sectors
extend well beyond the dim ensions o f the individual m etropolitan areas, and m ay
well extend beyon d a State, regional, or even national level (Iam m arino and M cC an n
2006; M cC an n 2007). Therefore, how im portant it is for a particular type o f firm to
be located either w ith in or im m ediately adjacent to a particular type o f industrial
cluster or m etropolitan area is entirely unclear. It will depend on a careful evaluation
o f the know ledge and transactions-costs relations w ithin the cluster and how these
relate to the locational and organizational optim ization problem s faced by the firm.
W hile the vast m ajority o f papers o n the subject o f clustering have discussed the
advantages o f clustering, very few have discussed the disadvantages o f clustering.
172 UR BA N S T R U C T U R E , G ROW TH , A N D T H E D E V E L O P M E N T P R O C E S S

W here the disadvantages o f clustering are discussed it is alm ost always in the context
o f increasing local land costs, labor costs, or congestión. A s such, these are all pecuni-
ary effects. A lm ost now here is the issue o f know ledge costs discussed in relation to
clustering, except in a very few instances (Iam m arino and M cC an n 2006). T h e issues
here concern the problem s o f unintended outw ard know ledge spillovers, first dis­
cussed b y G rin d ley and Teece (1997) in a nonspatial context, although this argum ent
translates directly to problem s o f industrial clustering (Iam m arin o and M cC a n n
2006). T h e central problem is that geographic clustering increases the likelih ood o f
unintended outw ard know ledge spillovers occurring. A s such, the firm has to try to
balance the benefits o f unintended inward and outw ard know led ge spillovers associ-
ated w ith industrial clustering. T h e seriousness o f the effects o f unintended outw ard
know ledge spillovers on the firm ’s proprietary know ledge and profitability w ill deter­
m ine, and also be determ ined by, the organizational structure o f the firm and its
industry. Therefore, the particular m eans b y w hich the firm w ill attem pt to avoid
such spillovers w ill und erpin the extent to w h ich it engages in lo n g-term form al
contracts governing all aspects o f know ledge exchanges. T h is w ill also determ ine
w hich typ e o f industrial cluster is attractive to a particular typ e o f firm .
U nderstanding these three stylized types o f clusters is im p o rtan t because this
allow s us to begin to id en tify the role that d ifferent cities play in grow th . In general,
cities tend to fall into the puré agglom eration m odel, irrespective o f w h ether the
city is highly diversified or specialized in term s o f its industrial structure. O bviously,
large cities are likely to contain elem ents o f all three cluster types, b u t the d o m in an t
set o f transactions-costs characteristics exhibited b y large cities are those repre-
sented by the m odel o f puré agglom eration. In term s o f regional plann ing, this
initially appears to im p ly that cities sh ould be allow ed to g ro w large naturally in
order to generate the econom ies o f scale and know ledge spillovers that are essential
for national econ o m ic grow th. Indeed, such flexible spatial adjustm ent allow ing
urban grow th is one o f the m ajor tenets o f n ew econ o m ic geograph y m odels. As w e
shall see in the fo llo w in g tw o sections, how ever, the role played b y cities, and the
scale o f cities in particular, in national grow th and d evelopm en t is n o t nearly as
straightforw ard as it m ight first appear on the basis o f these argum ents, in that it
appears to have changed over tim e since M arshall (1890,1920) was w ritin g.

T h e G r o w t h o f C i t i e s in the
T w e n t ie t h C e n t u r y

Globally, the period that saw the m ost rapid grow th o f in d ustrialization w as that
spanning broadly the second h a lf o f the nineteenth cen tu ry and up to the eve o f
W orld W ar I. A t the start o f this period G reat Britain was the w o rld ’s leading eco ­
n om ic pow er (Findlay and O ’R ourke 2007), as evidenced b y the G reat E xhibition o f
T H E ROLE OF IN D U S T R IA L C L U S T E R IN G AND IN C R EA SIN G R E T U R N S 173

L ondon in 1851, follow ed by the new ly industrializing U nited States, w hose d ram atic
process o f m echan ization follow ed the technical developm ents m ade d u rin g the
A m erican C ivil W ar (Best 1990). In Europe, G erm án u n ification also spurred rapid
industrial grow th, w hile Japan also began to em erge as a m o d ern industrial econ-
om y fo llow in g the M eiji R estoration. D u rin g the fifty years leading up to W orld
War I, w hile the levels o f in d ustrialization in the U nited K in gd o m and France had
doubled, in the U nited States and G erm an y they had increased sixfold (Findlay and
O ’R ourke 2007). Betw een 1820 and 1913 these en o rm o u s increases in the levels o f
industrialization were also associated w ith rapidly increasing in equality betw een
the different parts o f the w orld, w ith the rich industrialized A tlantic econ om ies plus
British “o ffsh o ots” (M ad d ison 2006) such as Australia and N ew Zealand pullin g
away from the rest o f the w o rld (Findlay and O ’R ourke 2007). By the end o f the
nineteenth cen tu ry the U nited States o verto ok Great Britain to em erge as the w o rld ’s
largest and richest econom y.
From the start o f the era o f globalization that took place during the late sixteenth
and early seventeenth centuries it is possible to trace a fairly direct and clear link
between the grow th o f industrialization, trade grow th, and the levels o f urbanization.
However, this period o f enorm ous industrialization from the m id-nineteenth century
through to the eve o f W orld War I also coincides w ith the era o f the m ost rapid period
o f urbanization, in w hat are nowadays the richest countries. As industrialization pro-
ceeded rapidly during the nineteenth century, global urbanization rates had tripled
(Findlay and O ’Rourke 2007). M ost notable was the case o f Chicago, w hich was the
fastest-growing m ajor city in the w orld, increasing in population from 100,000 in 1858
to m ore than 1.7 m illion in 1900. T h e result was that the distribution o f the w o rld ’s
largest cities closely reflected the econom ies w ith both the highest levels o f per capita
productivity and also prod uctivity growth.
A s we see in table 8.2, by 1925 N ew Y ork had em erged as the w o rld ’s largest city,
as it was the d o m in an t city o f the w o rld ’s d o m in an t econom y. In addition, in 1925
fourteen o f the w o rld ’s largest fifteen cities were located in either Europe, the U nited
States, or Japan. O n ly Buenos Aires in A rgentin a, w h ich was the w o rld ’s fastest-
grow in g m ajor city in the early part o f the tw entieth century, was outside o f these
regions, and this too was located in w hat was then a very rich country. As such, in
1925 all o f the w o rld ’s largest cities were in the richest and largest econom ies. These
m ajor cities fu n ction ed as the engines and the m otors that pow ered the largest and
wealthiest econom ies, co n su m in g inputs sourced fforn all over the w orld, m ost o f
w hich originated w ith in the em pire systems o f these sam e m ajor nations.
Im portantly, for o u r purposes, the turn o f the last cen tu ry was the very period
in w hich A lffe d M arshall was w ritin g about the role and nature o f industrial dis-
tricts and agglom eration effects in the U nited K in gd om . As such, his analytical
observations w hich I have already referred to, and w h ich were based prim arily on
case studies o f the arm am ents in d ustry o f B irm in gh am and the cotton in d u stry o f
M anchester, reflected a fairly sim ple set o f relationships. In the late nineteenth and
early tw entieth centuries, big cities were associated w ith w ealthy econom ies. In this
sense, at the tim e that M arshall was w riting, although as he freely adm its the exact
Table 8.2 The World’s Largest Cities in 1925
1925 C ity P o p u la tio n o o o s C o u n t r y P o p u la tio n o o o s G D P $ ooos G D P p e r C a p ita $
(% ch a n g e 1900-1925) (% ch a n g e 1900-1925) ( % c h a n g e 1900-1925) (% c h a n g e 1900-1925)

New York 7 7 7 4
, (8 3 -2 ) 116,284(52.2) 7 3 0 ,5 4 5 (2 3 3 ) 6,282(53.5)
London 7,742(19.5) 45,059(9-48) 231,806(25.4) 5,144(14.5)
Tokyo 5 ,3 0 0 (354) 59,522 (86.0) 112,209 (216) 1,885 (5 9 -7 )
París 4,800(44-1) 40,610(11.7) 169,197 (4 4 -9 ) 4,166(44.8)
Berlín 4,013(48.2) 63,166 (87.2) 223,082(37.4) 3,532 (18.3)
Chicago 3,564(208) 116,284(52.2) 7 3 0 5 4 5
, (2 3 3 ) 6,282(53.5)
Ruhr 3,400 (443) 63,166 (87.2) 223,082(37.4) 3,532 (18.3)
Buenos Aires 2,410(299) 10,358 (221) 4 0 ,5 9 7 (2 3 3 ) 3 9 1 9
, (5 3 -5 )
Osaka 2,219 (228) 59,522 (86.0) 112,209 (314) 1,885 (18.3)
Philadelphía 2,085(47) 116,284(52.2) 7 3 0 ,5 4 5 (2 1 6 ) 6,282(53.5)
Vienna 1,865 (9 -8 ) 6582 (10.2) 22,161 (233) 3,367(204)
Boston 1,764(64.1) 116,284(52.2) 7 3 0 ,5 4 5 (2 8 .7 ) 6,282(53.5)
Moscow 1 7 6 4
, (5 7 .5 ) 158,983 (27.2XUSSR) 231,886 [1928] (50.5) 1,370 [1928] (10.)
Manchester 1,725 (20.2) 4 5 ,0 5 (9 .4 8 ) 231,806(25.4) 5,144(14.5)
Birmingham 1,700(36.2) 4 5 0 5 9
, (9 .4 8 ) 231,806(25.4) 5,144(14.5)

Sources: City population data (Chandler 1987); country population, GDP, and GDP per capita data (Maddison 2006). McCann and Acs (2011).
THE ROLE OF IN D U S T R IA L C L U S T E R IN G AND IN C R E A S IN G R E T U R N S 175

m echanism s o peratin g were beyond his analysis, in broad term s the sto ry was
actually very straightforw ard. C ity size and w ealth creation were in extricably linked
in that u rbanization and in d ustrialization were tw o sides o f the sam e coin. It is this
sim ple and direct lin k betw een c ity scale, local externalities, and p ro d u ctiv ity that
underpins the increasing returns to scale assum ptions o f new eco n o m ic geograph y
m odels, m od ern urban eco n o m ic theories, and the regional plan n in g insights
derived from these m odels, the m ajority o f w hich are N orth A m erican in origin .
Indeed, p ro d u ctivity and city size are highly correlated in the U nited States
(Rosenthal and Strange 2004).
D iscussions that em erged very m uch later than w hen M arshall w as w ritin g have
subsequently focused on the m ore subtle question o f h o w localization versus u rb a n ­
ization econom ies, or specialization (M arshall 1890,1920) versus diversity (Jacobs
1984) is related to urban size and productivity. M arshall h im self did not explicitly
consider these issues. His w ritings, w h ich also borrow ed heavily ffo m earlier U K
governm ent research (M cC an n 2005), were based on observations o f the features o f
two o f the w o rld ’s largest cities at the tim e, w hich in to d ay’s term in o lo g y w o u ld
have been view ed as m egacities, nam ely, B irm in gh am and M anchester. T h e special-
ization-versus-diversity debate had not arisen at this point, and for the purposes o f
this chapter, in the fo llo w in g four sections I therefore focus sim ply on the issue o f
scale and urban agglom eration in a m anner that is largely consistent w ith the
productivity-scale findings o f Rosenthal and Strange (2004). I co m m en t on this
m ore subtle, but rather later, specialization -versus-diversity issue o n ly in the final
section o f the chapter.
A n im portan t question for o u r purposes, however, is w hether the sim ple rela-
tionship betw een city size and p ro d u ctivity still holds globally in a w ay that it did
w hen M arshall was w riting. O n ce again I can lo o k at city sizes in different tim e
periods to provide som e clues, and here it becom es clear that the situation is now
far m ore com plex than was the case in M arshall’s tim e.
T h e tim e fo llo w in g the W all Street crash was a period d u rin g w h ich u rb an iza­
tion processes slow ed d ow n, and this was also associated w ith a massive global con-
traction o f both trade and foreign investm ent (C rafts 2004) in response to the Great
D epression o f the 1930S and also W orld W ar II. D u rin g this period, all m ajor e c o n o ­
mies increasingly reoriented their trade p rim arily to w ith in the sphere o f their ow n
colonial systems (Findlay and O ’ R ourke 2007), and given that urban ization had
always been closely associated w ith industrialization, as econ o m ic grow th and trade
fell, so therefore d id the grow th o f urban ization . It was o n ly after W orld W ar II, d u r­
ing the reconstruction era under the B retton-W oods system , that the grow th in
urbanization once again picked up. In 1950 there were globally 67 cities w ith m ore
than 1 m illion inhabitants (C h an d ler 1987), w hereas by 1975 there were 190 urban
agglom erations w ith m ore than 1 m illion inhabitants (C h and ler 1987). O f these 190
cities, the United States accounted for 25 o f these m illion -plus agglom erations,
w hile W estern Europe accounted for 34, and W arsaw Pact Europe accounted for 15
(C h andler 1987). T he p ro p o rtio n o f the U.S. p opu lation living in cities also increased
at a higher rate in the postw ar era than d u rin g the interw ar period.
176 UR BA N S T R U C T U R E , GROW TH, A N D T H E D E V E L O P M E N T P R O C E SS

T h is p o st-W o rld W ar II period , how ever, w h ich was characterized glo b ally b y a
renew ed gro w th o f u rbanization, also involved tw o m ajor qualitative differences
from the p re-W W II urban ization processes. First, am o n g the w ealth y nations,
urban ization in this perio d w as also associated w ith spatial dispersión and geo-
graphic decentralization. Im proved tran sp ortatio n tech n o lo gies and infrastructure
allow ed the urban fringe to m ove farther aw ay from the urban cores. In addition,
this process w as m agnified by industrial restructurin g. W h ile, as w e have seen, m ost
cities co n tin u ed to grow d u rin g this postw ar p erio d o f increasing urb an izatio n ,
som e industrial cities actually declined in p o p u latio n d u rin g this era. T h is develop-
m ent was p rim arily associated w ith declines in cities d o m in ated b y m an u factu rin g
and engineering industries, as m an u factu rin g as a w h o le exhibited a strong urban-
rural drift in response to changing space-extensive p ro d u ctio n techniques and the
need for greater accessibility to nation al tran sp ort netw orks. In Europe this was
particularly noticeable in the trad itional m an u factu rin g heartlands o f the U nited
K in gd o m (Fothergill, K itson, and M o n k 1985). As such, in b o th the U nited States
and Europe (Sassen 2006), m an y o f the very largest cities saw actual p o p u latio n
declines in the core parts o f the agglom erations, as b o th househ old s and firm s
m oved o u t to m ore suburban locations. N o t surprisingly, this process o f spatial
decentralization tended to be associated w ith ever-increasing h o u seh o ld co m m u t-
ing distances into the urban centers.
Second, as w e see in table 8.3, since the end o f W orld W ar II and fo r the first tim e
in the m od ern era, cities in the d evelopin g w o rld first started to enter the highest
echelons o f the w o rld c ity -size ranking. O ver the fo llo w in g five decades span nin g
the second h a lf o f the tw entieth century, n ot o n ly d id urban izatio n rates increase
globally, but this period o f u rban izatio n carne to be increasingly d o m in ated b y the
rise o f u rbanization in d evelopin g countries. By 1975 m ore than 75 o f the 190 urban
agglom erations w ith m ore than 1 m illio n inhabitants were fforn the so-called T h ird
W orld countries (C h and ler 1987). As w e see fforn table 8.4, seven cities in d evelo p ­
ing countries, nam ely, M éxico City, Sao Paulo, B uenos Aires, R io de Janeiro, Cairo,
Shanghai, and Kolkata are all am on g the w o rld ’s top fifteen cities. A m o n g the w o rld ’s
largest fifteen cities, betw een 1950 and 1975, eleven had experienced faster grow th
than their equivalent-ranked city in 1950 had experienced betw een 1925 and 1950.
A fter this initial phase o f p o st-W o rld W ar II urban ization , the process o f u rb a n ­
ization actually accelerated in the last three decades o f the tw entieth century. B y
2008, the num ber o f people living in urban areas across the w o rld had risen to
3.3 billion, such that for the first tim e in h istory the global urban p o p u latio n passed
50 percent o f the global po p u latio n (O E C D 2006; U N F PA 2007). M oreover, this
process o f increasing urban ization w as co m m o n to both the industrialized and the
industrializing w orld. T he result o f this was that by the year 2000, there w ere m ore
than 140 cities globally w ith populations o f m ore than 2 m illio n inhabitants, and 19
cities w ith population s o f m ore than 10 m illion (Le Gales 2002).
T his en orm ou s urban grow th that to o k place d u rin g the five decades fo llo w in g
W orld W ar II was p rim arily associated w ith the rise o f cities in the d evelopin g w orld.
By 2005 the d evelopin g w o rld ’s urban po p u latio n o f 2.4 b illion accounted for
Table 8.3 The World’s Largest Cities in 1950
1950 C ity P o p u la tio n o o o s C o u n t r y P o p u la tio n o o o s G D P $ ooos G D P p e r C a p ita $
(% c h a n g e 1925-1950) (% c h a n g e 1925-1950) ( % c h a n g e 1925-1950) ( % c h a n g e 1925-1950)

New York 12,463 (60.3) 152,271 (30.9) 1 4 5 5 ,9 1 6

, (9 9 -3 ) 9,561 (52.2)
London 8,86o (14.4) 50,127(11.2) 3 4 7 8 5 0
, (5 0 .1 ) 6 ,9 3 9 (3 4 -8 )
Tolcyo 7,000 (32.1) 83,805 (40.8) 160,966 (43.4) 1,921 (1.9)
París 5,900 (22.9) 41,829 (3.0) 220,492 (30.3) 5,271 (26.5)
Shanghai 5,407 (360) 546,815 (13.8) 239,903(io.o) 439 (-21.9)
Moscow 5,100 (289) ' 179,571 [USSR] (12.9) 510,243 (220) 2,841 (207)
Buenos Aires 5,000 (207) 17,150 (65.6) 85,524(210) 4,987 (27.2)
Chicago 4,906 (37.6) 152,271 (30.9) 1 4 5 5 9 1 6
, , (9 9 -3 ) 9,561 (52.2)
Ruhr 4,900 (44.1) 68,375 (8.2) 265,354 (18.9) 3,881 (9.9)
Kolkata (Calcutta) 4,800 (345) 359,000(12.2) 222,222 (30.3) 619 (-11.4)
Los Angeles 3.986 (347) 152,271 (30.9) 1 4 5 5 9 1 6
, , (9 9 .3 ) 9,561 (52.2)
Berlín 3 .7 0 7 (-7 7 ) 68,375 (8.2) 2 6 5 ,3 5 4 (18.9) 3,881 (9.9)
Osaka 3,341 (50.6) 83,805 (40.8) 160,966 (43.4) 1,921 (1.9)
Philadelphia 2,900 (39.1) 152,271 (30.9) 1 4 5 5 ,9 1 6
, (9 9 -3 ) 9,561 (52.2)
México City 2,872 (372) 28,485 (5 3 -3 ) 67,368 (223) 2,365 (7 3 -i)
World (1950) 2,524,324 5 3 2 9 ,7 1 9
, 2 ,111 (40) (1913-1950)

Sources: City population data (Chandler 1987); country population, GDP, and GDP per capita data (Maddison 2006).
Table 8.4 The World’s Largest Cities in 1975
1975 C ity P o p u la tio n o o o s C o u n t r y P o p u la tio n o o o s G D P $ ooos G D P p e r C a p ita $
(% ch a n g e 1950-1975) (% c h a n g e 1950-1975) (% c h a n g e 1950-1975) (% c h a n g e 1950-1975)

Tokyo 23,000 (328) in , 5 7 3 (3 3 -i) 1,265,661 (786) 1 1 ,3 4 4 (5 9 0 )

New York 17.150 (3 7 -6 ) 2 1 5 ,9 7 3 (4 1 -8 ) 3,516,825 (241) 16,284 (70.3)
Osaka 15,500 (464) 1 1 1 ,5 7 3 (3 3 .i) 1,265,661 (786) n , 3 4 4 (5 9 0 )
México City 1 1 ,3 3 9 (3 9 5 ) 60,828 (213) 312,998 (465) 5,146 (216)
Moscow 10,700 (209) 2 5 4 ,5 1 9 [USSR] (41.7) 1 5 6 1 ,3 9 9
, (306) 6,135 (216)
London 10,500 (18.5) 56,215 (12.1) 665,984 (9 1 .4 ) 11,847 (225)
Sao Paulo 10,041 (4 5 i) 108,824(204) 4 5 5 ,9 1 8 (510) 4,190 (257)
Paris 9,400 (5 9 -3 ) 52,758(26.1) 699,106(317) 1 3 ,7 7 3 (261)
Los Angeles 8,960 (225) 2 1 5 ,9 7 3 (4 i-8 ) 3,516,825 (241) 16,284 (70.3)
Buenos Aires 8,498 (69.9) 26,082 (52.1) 211,850 (247) 8,122 (62.8)
Cairo 8,400 (305) 3 6 ,9 5 2 (7 4 .3 ) 52,501 (272) 1,421 (56.1)
Rio de Janeiro 8,328 (290) 108,824(204) 455,918 (510) 4,190 (251)
Shanghai 8,000 (47.9) 9 1 6 ,3 9 5 (67.6) 800,876 (339) 8 7 4 (9 9 -i)
Kolkata (Calcutta) 7,875 (64.0) 607,000 (69.1) 544,683 (245) 8 9 7 (4 4 .9 )
Seoul 7,500 (483) 35,281 (69.2) 111,548 (695) 3,162 (411)
World 4,065,408 16,644,898 4 ,0 9 4 (9 3 -9 )
Sources: City population data (Chandler 1987); country population, GDP, and GDP per capita data (Maddison 2006).
THE ROLE OF IN D U S T R IA L C L U S T E R IN G AND IN C R EA SIN G R E T U R N S 179

approxim ately three-quarters o f the global urban p o pu lation (W orld B ank 2008;
UN FPA 2007). T h e p ro p o rtio n o f the p opu lation in the lovv- and m id d le-in co m e
countries o f the d evelop in g vvorld that lived in urban areas had increased from 37
percent to 44 percent betvveen 1990 and 2005, w hile d u rin g the sam e p eriod the
pro p o rtion o f the po p u latio n in d evelopin g countries living in cities o f m ore than
1 m illion inhabitants had increased from 14 percent to 17 percent (W orld B ank
2008). As we see in table 8.5, the result o f these changes in the d evelopin g w orld was
that by 2000, ten o f the w o rld ’s largest fifteen cities were from the d evelop in g w orld.
M oreover, this tend en cy tow ard m egacities in the d evelopin g w orld was not specific
to one or tw o countries, in that these ten cities were located in eight different
countries.
T h e annual gro w th o f the u rb an p o p u la tio n in low - and m id d le-in co m e
co u n tries in the d evelo p in g w o rld betw een 1990 and 2005 has been h igh er at
2.6 percent than that o f the h ig h -in co m e co u n tries at o n ly 1.1 percen t (W orld
B ank 2008). M oreover, all the evid en ce suggests that the rate o f u rb a n iza tio n in
the d evelo p in g w o rld w ill co n tin u é to increase even faster relative to that o f the
d evelo p ed w o rld . U nited N atio n s p red ictio n s suggest that the global urban p o p u ­
lation w ill increase to 4.9 b illio n by 2030, o f w h ich 3.9 b illio n w ill be in the d e ve l­
o p in g w o rld (U N F P A 2007). As such, the level o f u rb a n iza tio n in the d evelo p in g
w o rld w ill increase by 60 percent betw een 2000 and 2030, w h ich is som e 3.75 tim es
greater than the u rb a n iza tio n rate in the d evelo p ed w o rld over the sam e p erio d .
M oreover, this un p reced en ted u rb a n iza tio n is not sim p ly a result o f p o p u latio n
gro w th , in that o ver the sam e p erio d , the glo b al rural p o p u latio n is actu ally
expected to decrease (U N F P A 2007). Betw een 1950 and 2030, the total u rb a n iza ­
tion ín d ex o f the d evelo p in g w o rld w ill increase fro m 18 percent to 56 percen t
(U N F P A 2007), w ith the m a jo rity o f this increasing u rb a n iza tio n takin g place in
Asia and A frica. By then, Asia and A frica w ill a cco u n t fo r 80 percen t o f the glo bal
urban p o p u latio n (U N F P A 2007).
If we focus on the large and rap id ly g ro w in g d evelo p in g nation s, we see that
six o f the ten largest cities in the d evelo p in g w o rld are located in w h at the O E C D
calis the B R IIC S co u n tries (M cC a n n 2009) o f Brazil, Russia, India, Indonesia,
C h in a, and South A frica. As C o llie r and V enables (2008) argüe, tw o co m m o n fea-
tures o f alm ost all the fastest-gro w in g co u n tries in the d evelo p in g w o rld are that
they are la rg e-p o p u la tio n co u n tries that are able to sustain large agglo m eratio n s.
In the d evelo p in g w o rld , co u n tries unable to sustain ag glo m eratio n effects are
largely static o r even d eclin in g in w ealth (C o llie r and V enables 2008). A s such, in
the case o f d evelo p in g co u n tries to d ay there still seem s to be a fairly direct lin k
betw een urban scale and e co n o m ic g ro w th , exactly as M arshall (1890,1920) had
observed was the case for G reat B ritain at the turn o f the last century. T h erefo re,
the analytical assu m p tio n s o f M arshall, w h ich u n d erp in m any urban e co n o m ic
and new eco n o m ic geo g rap h y fram ew orks, still appear to capture m any essential
aspects o f the urban grow th processes at w o rk in the d evelo p in g co u n tries and
also, as we w ill see shortly, the urban g ro w th processes e vid en t in the U n ited
States, Japan, and Korea.
Table 8.5 The World’s Largest Cities in 2000
2000 City Population1 ooos Country Population ooos GDP $ooos GDP per Capita $
(% change 1975-2000) (% change 1975-2000) (% change 1975-2000) (% change 1975-2000)

Tokyo 29,896 (30.0) 126,737 (13.6) 2,589,320 (204) 20,431 (80.0)

New York 2 4 .7 1 9 (4 4 -i) 270,561 (25.2) 7 ,3 9 4 , 5 9 8 (210) 2 7 ,3 3 1 (67.8)
Seoul 20,674 (275) 46,898 (30.7) 624,582 (559) 1 3 ,3 1 7 (421)
México City 19,081 (68.3) 98,553 (62.0) 655,910 (209) 6,665 (29.5)
Sao Paulo 1 7 .3 9 6 (7 3 .2 ) 169.897(56.0) 926,918 (203) 5,459 (30.2)
Manila 16,740 (310) 7 9 .3 7 6 (78.5) 181,886 (201) 2,291 (12.9)
Los Angeles 15,807 (76.4) 270,561 (25.2) 7 3 9 4 ,5 9 8
, (210) 2 7 ,3 3 1 (67.8)
Mumbai 15,769 (223) 991,691 (63.3) 1,803,172 (3.31) 1,818 (202)
Djakarta 15,086 (284) 207,429 (58.9) 628,753 (3.2) 3,031 (201)
Osaka 1 5 .0 3 9 (-3.0) 126,737 (13.6) 2,589,320 (204) 20,431 (80.0)
Delhi 1 3 .5 9 2 (309) 991,691 (63.3) 1,803,172 (3.31) 1,818 (202)
Kolkata 12,619 (60.2) 991.691 (63.3) 1,803,172 (3.31) 1,818 (202)
Buenos Aires 12,297 (44-7) 36,235 (3 9 .2 ) 3 3 4 ,3 1 4 (5 7 .8 ) 9,219 (13.2)
Shanghai 11,960 (49.5) 1,252,704 (36.6) 4,082,513 (509) 3 ,2 5 9 (3 7 2 )
Cairo 11,633 (38.4) 66,050 (78.7) 1 4 0 ,5 4 6 (3 3 9 ) 2,128 (89.8)
World [1998] 5,907,680 (45.3) 3 3 7 2 5 ,6 3 1
, (202) 5 ,7 0 9 (3 9 -4 )
1 The figures reported here are for conurbations and come from GEOPOLIS database rather than just city administrative boundaries. Therefore, New York City (five
boroughs) has only 7.5111, the greater New York area has 18.7111 (OECD 2006), and the total New York conurbation (including Philadelphia) has more than 24 million
inhabitants. Similarly, depending on the source used and methods o f defining urban areas, London city has only 7.4111 inhabitants (OECD 2006), but the London
conurbation has a population o f 8.5 m according to the World Urbanization Prospects (2008) (http://esa.un.0rg/unup/p2kodata.asp), a population o f more than 9.66m
according to GEOPOLIS, or a population o f 11.221x1 according to the World Gazetteer (www.world-gazatteer.com).
Sources: City population data (Chandler 1987; Le Gales 2002); country population, GDP, and GDP per capita data (Maddison 2006). McCann and Acs (2011).
T H E ROLE OF I N D U S T R IA L C L U S T E R IN G A ND IN C R E A SIN G R E T U R N S 181

ClTIES IN THE TWENTY-FIRST C e NTURY

As we saw in the second section, o u r analytical understanding o f the role o f agglom -

eration and urban scale in eco n o m ic d evelopm ent has increased en o rm o u sly dur-
ing the second h a lf o f the tw entieth cen tu ry and the first decade o f the tw enty-first
century. As we have just seen, the fairly sim ple M arshallian story appears to capture
m uch o f w hat is happ en ing in the U nited States, Japan, Korea, and m any d evelopin g
countries. W h at is m uch less und erstood , however, is that the situation in the rest o f
the d eveloped w orld is novv far m ore com plex, varied, and nuanced than the sim ple
M arshallian story im plies. A sim ple M arshallian story is not even approxim ately
sufficient to capture the essence o f w hat is curren tly taking place in the m ajo rity o f
advanced econom ies. T h e reason is that although the M arshallian assum ptions are
attractive for a range o f urban eco n o m ic m odels and new econ o m ic geograph y
argum ents, in m any rich countries the link betw een urban scale and wealth that was
so evident in the early tw entieth cen tu ry n ow appears to have largely broken dow n.
T h is is im portant in term s o f regional plann ing theories because m any such theo-
ries derive from urban econom ies and new econ o m ic geography, w h ich them selves
are dom in ated by a N orth A m erican and also Japanese agenda. As we w ill see here,
in the U nited States and Japan the relationship betw een city size and p ro d u ctivity
still largely reflect the patterns observed by M arshall, and in this sense his argu­
m ents appear to be valid in the cases o f these tw o countries. For o ther countries,
however, the reality is very different.
In order to see this, we have to reconsider how table 8.2 com pares w ith tables 8.3,
8.4, and 8.5. As has already becom e clear, post-W orld War II urban grow th was d o m i­
nated by developing countries, whereas prior to this period, large cities were associated
almost exclusively w ith wealthy countries. Yet, at the turn o f the m illennium only three
advanced countries were represented in terms o f the w orld’s largest cities, and these are
the United States, Japan, and Korea. T he rest o f the w orld’s wealthy countries are absent
ffo m the top echelons o f the w orld city-scale rankings. By the turn o f the m illennium
¡t was clear that the m ajority o f the w orld ’s largest cities are nowadays no longer located
in the w orld ’s m ost productive and richest econom ies. Rather, they are in the newly
industrializing countries. As such, between 1925 and 2000 the simple M arshallian rela­
tionship between city size and wealthy countries appears to have fundam entally
changed in m ost rich countries. T he scale o f the dom inant cities in a coun try is there-
fore no longer the simple indicator o f the wealth o f a country or society, in the w ay it
was w hen A lffed Marshall (1890,1920) was first observing the im portance o f agglom -
eration efifeets during the latter years o f the nineteenth century and the early years o f
the twentieth century.
T hese striking observations therefore lead to tw o im m ediate questions, both o f
w hich have direct relevance for urban and regional planning. First, w h y has the
sim ple M arshallian relationship betw een city size and wealth creation changed, and
w hat is the current role o f cities in the econom y? Second, w hat are the im plications
o f these changes for tw en ty-first-cen tury urban and regional planning? In the next
182 UR BA N S T R U C T U R E , G ROW TH , A N D T H E D E V E L O P M E N T PR O C ESS

tw o sub-sections I w ill deal w ith the first o f these tw o questions. T he subsequent full
section w ill deal w ith the second o f these tw o questions.

ClTIES, SCALE, AND PRODUCTIVITY

A t the start o f the tw enty-first century, in the industrialized w o rld the total urban
po p u latio n in 2006 was estim ated to be m ore than 870 m illion (U N F PA 2007), in
other w ords, approxim ately o n e-qu arter o f the global urban p o p u latio n o f 3.3 bil-
lion (W orld B ank 2008). By 2007 som e 53 percent o f the O E C D p o p u latio n lived in
urban areas; this figure rises to alm ost 80 percent if less densely p opu lated in term e­
díate urban areas are included in the estim ates (O E C D 2006). A s w ith the d evelop-
ing countries, the p ro p o rtion o f the po p u latio n livin g in urban areas in h igh -in com e
countries has been increasing in keeping w ith the global p h en om en o n o f u rb a n iza ­
r o n . Betw een 1990 and 2005 the levels o f urban ization in the developed countries
increased from 71 percent to 73 percent (W orld B ank 2008), and this process is
expected to contin ué. T h e U nited N ations anticipates that betw een 2000 and 2030
the urban p o p u latio n o f the developed w o rld w ill increase b y som e 16 percent from
870 m illion in 2000 to 1.01 b illion in 2030 (U N F PA 2007).
N o t surprisingly, this increasing urban ization in the w ealthy countries is also
associated w ith large cities. Indeed, by 2002 the O E C D con tain ed seventy-eight
m etropolitan urban regions w ith m ore than 1.5 m illion inhabitants. However, the
size o f the m ajor cities in m any o f the w orld richest countries suggests that the rela-
tionship betw een urban scale and w ealth is far from straightforw ard, and certainly
rather different than sim ple M arshallian assum ptions m igh t im ply.1
T he O E C D (2006) classifies cities o f m ore than 7 m illion inhabitants as “m ega-
cities.” As w e see in table 8.6, however, m ost o f the w o rld ’s highest-p ro d u ctivity
cities are not m egacities, and m ost are now here near the sam e order o f m agnitude
o f w hat are curren tly the w o rld ’s largest cities, as listed in table 8.5. In fact, o f the
w o rld ’s seventy-five high est-p ro d u ctivity cities (in clu d in g Singapore, H on g Kong,
and cities in Taiwan and Israel), tw enty-nin e are w hat the O E C D (2006) classifies as
“sm all m etro areas” o f fewer than 3 m illion inhabitants; th irty-tw o are w h at the
O E C D (2006) classifies as “ m éd ium to large m etro areas” o f betw een 3 and 6.99 m il­
lion inhabitants; and o n ly fourteen are m egacities o f at least 7 m illion inhabitants.
M ost rich countries contain d o m in an t cities that are not m egacities in term s o f
population size. As we see in table 8.5, am on g the rich countries, the exceptional

1. Combining the OECD (2006) metropolitan productivity data with PPP national
productivity data at current prices (World Bank 2008) would also rank Singapore as the
30th highest productive city in the world, below Stuttgart and above Milán, and Hong
Kong as the 48th highest productivity city in the world, below Auckland and above
Hamburg.
Table 8.6 The WorlcTs Most Productive Cities in 2002-2004
U.S. C itie s C ity P o p 1 M illio n s C ity p e r C a p ita N o n -U .S . O E C D C itie s C it y P o p M illio n s C it y p e r C a p ita
P r o d u c tiv ity P r o d u c tiv ity
(U S $000 P P P ) (U S $ P P P )

San Francisco 4 -2 62.3 London 7 -4 46.2

Washington DC 5 -i 61.6 París 11.2 42.7
Boston 4 -4 58.0 Dublin 1.6 38.9
SeattJe 3 -2 5 4 -4 Vienna 2.2 37.6
Minneapolis 3 -t 53 -0 Stockholm 2.2 36.7
New York 18.7 52.8 Stuttgart 2.7 36.4
Denver 2.3 50.8 Milán 7 -4 35.6
Philadelphia 5.8 50.5 Lyon 1.6 3 5 -2

Dallas 5 -7 50.1 Munich 6.1 3 5 -2

Atlanta 4 -7 47.8 Oslo 1 -7 35.0

Houston 5.2 4 7 -4 Sydney 4.2 3 5-0

San Diego 2.9 46.8 Brussels 3 -8 35.0

Chicago 9 -4 4 5 -6 Toronto 4 -7 3 4 -9

Los Angeles 12.9 4 5 -3 Helsinki 1.8 3 4 -0

Detroit 4 -5 44.0 Frankfurt 5.6 3 3-6

1 The city population figures here are firom the OECD metropolitan database (OECD 2006) and differ slightly from the urban definitions employed in table 8.5.
2 Combining the OECD (2006) metropolitan productivity data with PPP national productivity data at current prices (World Bank 2008) would also rank Singapore as
the thirtieth highest productivity city in the world, below Stuttgart and above Milán, and Hong Kong as the forty-eighth highest productivity city in the world, below
Auckland and above Hamburg.
Sources: OECD (2006,38-40); World Bank (2008).2McCann and Acs (2011).
184 UR BA N S T R U C T U R E , GROW TH, A N D T H E D E V E L O P M E N T P R O C E SS

nations in this regard are U nited States, Japan, and Korea. T hese are the o n ly three
w ealthy countries that contain cities large en o u gh to reach the top echelons o f the
w o rld c ity -size rankings.
T his o bservation suggests that in term s o f u rban gro w th and perform an ce, the
m echanism s that are operating in m any o f the o ther O E C D co u n tries m ay be rather
different than those that are operating in the U nited States, Japan, and Korea. It can
therefore be instructive to also focus specifically o n the rest o f the o ther advanced
O E C D countries, in that th ey can be regarded in som e sense as b ein g m ore repre-
sentative o f rich countries. O n this basis, exclud in g the top tw enty-three cities in the
U nited States and focusing o n ly on the rest o f the O E C D , w e see that o f the w o rld ’s
m ost prod uctive fifty-tw o non-U .S. cities, tw en ty-on e are sm all m etro areas o f few er
than 3 m illion inhabitants; tw en ty are w hat the O E C D (2006) classifies as “ m éd iu m
to large m etro areas” o f betw een 3 and 6.99 m illion inhabitants; and o n ly eleven are
m egacities o f at least 7 m illion inhabitants.
T he results o f these observation s is that for cross-sectional estim ations o f the
relationship betw een urban scale and urban p ro d u ctiv ity across cities o f m ore than
1.25 m illion but less than 10 m illion inhabitants, the positive relationship betw een
city per capita p ro d u ctivity and city p o p u latio n is both statistically significant and
strong (O E C D 2006). In contrast, for cities o f m ore than 6 m illion in h a b ita n ts ,,
there is a w eak but statistically significant negative cross-sectional relationship
betw een city per capita p ro d u ctiv ity and city p opulation . As such, am o n g O E C D
cities there appears to be som eth ing o f an inverted U -shaped relationship betw een
city per capita p ro d u ctivity and p o p u latio n scale, in w h ich urban p ro d u ctivity
appears to be m axim ized at around 5 to 7 m illion inhabitants.
Taken together, the overall result o f these different relationships is that for all
O E C D cities o f m ore than 1.25 m illion inhabitants, there is actually o n ly a v ery w eak
statistically significant cross-sectional link betw een city per capita p ro d u ctiv ity and
city p opulation, and if anything, the relationship is actually slightly negative (O E C D
2006). M oreover, if w e also were to also inelude in o u r cross-sectional sam ple the
rest o f the largest cities ffo m d evelopin g countries, then clearly the negative p ro d u c­
tivity effect o f p o pu lation scale w ill be very m uch m ore m arked am o n g the m egaci­
ties than is the case w ith just the O E C D cities. D oes this m ean that M arshall’s
agglom eration argum ents are w rong?
Part o f the p ro b lem here is that it is n o t clear fro m u rb an e c o n o m ic th e o ry
w h ich cities sh o u ld be in clu d ed in these types o f cro ss-sectio n al estim ation s. O n e
plausible argum en t is that d o m estic e c o n o m ic and in stitu tion al co n d itio n s d iffer
m arked ly betw een d ifferen t co u n tries, and given that c ity g ro w th is associated
w ith changes in land use, then in stitu tion al variatio n s p reelud e sim ple direct
in tern ation al com parison s. Second , sim ple u rb an p ro d u c tiv ity m easures d o not
a cco u n t for the n ation al im paets on urban p ro d u ctivity. In o rd er to circu m ven t
these problem s w e can con vert the urban p ro d u c tiv ity ran kin gs fro m absolute to
relative m etrics.
Table 8.7 reports the rankings o f the w o rld ’s h igh est-p ro d u ctivity cities, calcu-
lated on absolute levels o f urban G D P per capita as depicted in table 8.6, but then
Table 8.7 Relative Performance of the Highest-Productivity Cities in the O ECD
U.S. C itie s C ity P o p u la tio n 1 R elative N o n -U .S . O E C D C itie s C it y P o p u la tio n R ela tiv e
M illio n P r o d u c tiv ity M illio n P r o d u c tiv ity

San Francisco 4.2 1.72 London 7 -4 t- 5 9

Washington 5 -1 1.70 París 11.2 1 -5 3

Boston 4 -4 1.60 Dublin 1.6 1.18

Seattle 3 -2 1.50 Vienna 2.2 1.27
Minneapolis 3-1 1.46 Stockholm 2.2 1.29
New York 18.7 1-4 5 Stuttgart 2.7 1 -3 4

Denver 2 -3 1.40 Milán 7 -4 1.29

Philadelphia 5-8 1-3 9 Lyon 1.6 1.26
Dallas 5-7 1.38 Munich 6.1 1.30
Atlanta 4 -7 1.32 Oslo 1 -7 0.95
Houston 5-2 1.31 Sydney 4 -2 1.07
San Diego 2.9 1.29 Brussels 3 -8 1.19
Chicago 9 -4 1.26 Toronto 4 -7 1.08
Los Angeles 12.9 1.25 Helsinki 1.8 1.19
Detroit 4-5 1.21 Frankfurt 5 -6 1.24

1 The figures here are from the OECD metropolitan database (OECD 2006,38-40) and differ slightly from the urban definitions employed in table 8.5.
Sources: Calculations based on OECD (2006,38-40); World Bank (2008). McCann and Acs (2011).
i8 6 UR BA N S T R U C T U R E , GROW TH, AND T H E D E V E L O P M E N T P R O C E SS

converts these m etrics into relative valúes, calculated w ith respect to the nation al
G D P per capita o f the co u n try in w h ich the city is located. O n ce again as w ith table
8.6, in table 8.7 the U.S. cities are split from the o ther rich cities in o ther O E C D
countries. Table 8.7 therefore lists the top fifteen per capita p ro d u ctiv ity U.S. cities
and also the top fifteen non-U .S. O E C D cities and indicates their city per capita
p ro d u ctiv ity relative to the nation al average p ro d u ctiv ity for the countries in w h ich
th ey are located.
It is o b v io u s that the U.S. ra n kin gs rem ain u n ch a n g ed , b u t the n o n -U .S .
ran kin gs on the rig h t-h a n d side o f table 8.7 d iffer sig n ifica n tly w h en m easured
relative to th eir n a tio n al p ro d u c tiv ity levels in co m p a riso n w ith the rig h t-h a n d
ran kin gs in table 8.6. A s such, the relative p erfo rm a n ce o f the in d iv id u a l cities
w ith respect to th eir o w n co u n tries is rath er d ifferen t fro m w h a t m ig h t be in ferred
sim p ly fro m absolute u rb an p ro d u c tiv ity ran kin gs. T h e m o st n o tab le ch an ge is
the p erfo rm a n ce o f E u rop e’s tw o largest cities, nam ely, L o n d o n and París, relative
to the p erfo rm a n ce o f th eir n a tio n al eco n o m ies. In relative term s th e y are even
m ore im p o rta n t d o m e stica lly than th ey are in abso lu te term s, yet th eir relative
p erfo rm a n ce is still o f the sam e o rd er o f m a g n itu d e as fo r m an y U.S. cities. A t
the sam e tim e, apart fro m L o n d o n and París, the relative p erfo rm a n ce o f m o st o f
the o th er n o n -U .S . h ig h -p ro d u c tiv ity cities is a ctu ally n o tice a b ly less th an is the
case for the U.S. cities. T h erefo re, apart fro m L o n d o n and París, m o st h ig h -p ro -
d u ctiv ity cities in o th er rich O E C D co u n trie s o u tsid e o f the U n ited States are
b o th sm aller and also less p ro d u ctiv e in b o th absolute term s and relative term s,
than the U.S. cities.
To further investígate these issues, in table 8.8, w e repeat exactly the sam e exer-
cise, b u t this tim e after exclud ing all U.S. cities. T he first three co lu m n s on the left-
hand side o f table 8.8 inelude h ig h -p ro d u ctiv ity cities o n ly in the w ealthy O E C D
countries, w hereas the three colum ns on the righ t-han d side inelude cities located
in the form er transition O E C D econom ies, as w ell as in M éxico and Turkey.
If we consider the first three left-hand colu m n s o f table 8.8 we see that tw elve o f
these cities are the sam e as those listed in the three righ t-h an d colu m n s o f table 8.7.
In contrast, i f we n o w consider the second three righ t-h an d colu m n s o f table 8.8, a
dram atic change in both relative p ro d u ctiv ity scorings and also city com position
now occurs. T he highest relative prod u ctivity-ran k ed non-U .S. O E C D cities n o w
have relative pro d u ctivity scores that are actually even higher in general than those
o f the U.S. cities. M ost notably, however, w e see that the rankings are n o w d o m i-
nated b y cities in the poorer nations o f the O E C D . Ten o u t o f the top fifteen relative-
p ro d u ctivity cities are located in either fo rm er tran sition -eco n o m y countries o r
developing co u n try m em bers o f the O E C D . Two o f these cities, nam ely, M éxico C ity
and Istanbul, are very large indeed and can easily be placed in the O E C D category
o f m egacities.
These observations suggest that the processes d rivin g agglom eration in the U.S.
cities and also the cities in the po o rer parts o f the O E C D m ay have m uch in com -
m on. In particular, cities in the U nited States and also the p oorer O E C D countries
appear on average to exhibit very high relative urban p ro d u ctiv ity levels. In contrast,
Table 8.8 The Highest Non-U.S. Relative Productivity Cities in the O ECD
N on -U .S. O E C D C itie s C ity P o p u latio n R elative N on -U .S. O E C D C ity P o p u la tio n R elative
E x clu d in g F o rm e r M illio n s P ro d u c tiv ity C itie s M illio n s P ro d u c tiv ity
T ra n sitio n E c o n o m ie s, (A ll O E C D
M éxico an d T urk ey c o u n trie s)

London 7 -4 1 -5 9 Warsaw 3.0 1.99

París 11.2 1-5 3 Monterrey 3-2 1.98
Lisbon 2 .7 1.39 Istanbul 11.4 1.60
Auckland 1.2 1 -3 4 London 7 -4 1 -5 9

Stuttgart 2 .7 1 -3 4 Budapest 2.8 1 -5 9

Milán 7.4 1.31 París 11.2 i- 5 3

Munich 6.1 1.30 Prague 2.3 1.51
Stockholm 2.2 1.29 México City 18.4 1.49
Vienna 2.2 1.27 Izmir 3 -4 1.46
Lyon 1.6 1.26 Ankara 4.0 1.41
Frankfurt 5.6 1.24 Guadalajara 3 -5 1-3 9

Madrid 5.6 1.24 Lisbon 2-7 1-3 9

Rome 3.7 1.21 Puebla 2.1 1.36

Brussels 3.8 1.19 Auckland 1.2 1 -3 4

Helsinki 1.8 1.19 Stuttgart 2.7 1-3 4

Sou rces: Calculations based on OECD (2006,38-40); World Bank (2008). McCann and Acs (2011).
188 UR BA N S T R U C T U R E , G ROW TH, AND T H E D E V E L O P M E N T P R O C E SS

m an y h ig h -p ro d u ctiv ity cities in w ealthy O E C D co u n tries tend to exhibit relative

p ro d u ctiv ity scores that are som ew hat low er than those in either the U nited States
o r the po o rer O E C D countries. In ad dition, as w e have already seen, m any o f the
cities in these w ealthy coun tries are also m uch sm aller than those in the U nited
States.
T hese various observation s suggest that the relationship betw een city size and
p ro d u ctiv ity is far from clear-cut in m any countries. W h ile there is still clearly an
im p o rtan t role for very large cities in the d evelopin g w o rld in term s o f d rivin g
grow th, for w ealthy countries the relationship betw een city size and p ro d u ctiv ity is
now adays not nearly as straightforw ard as it w as w h en M arshall (1890,1920) was
w riting. T h e effect o f urban scale o n p ro d u ctiv ity is very clear in the case o f the
U nited States (Rosenthal and Strange 2004), b u t b eyo n d the cases o f the U nited
States, Japan, and Korea, the situation appears to be rather m ore co m p lex for the
rest o f the O E C D . T h e po o rer O E C D coun tries appear to m im ic the U nited States
in term s o f the relative im p ortan ce o f cities, w hereas the other w ealthy O E C D c o u n ­
tries are clearly v ery different. In contrast, in the other w ealthy O E C D countries the
m ost productive cities are generally m u ch sm aller than in the U nited States, and
their relative p ro d u ctiv ity scores are also m uch lower.
In term s o f the ch an ges in these pattern s, o n ce again the p ictu re is v e ry c o m ­
plex. W h ile A u stralia, C a ñ a d a , and N e w Z e a la n d are e x p e rie n cin g ra p id ly in creas-
ing levels o f u rb a n iza tio n (O E C D 2006), the e xp erien ce o f the E u rop ean
e co n o m ies is v e ry d ifferen t. T h e share o f G D P in E u rop e a cco u n te d fo r b y u rb an
areas o f m ore than 250,000 in h a b itan ts has rem ain ed a lm o st co n sta n t o ver the
last d ecad e (E C 2009). In the E U co u n trie s, w h ile som e cities have b een g ro w in g ,
others have been d e clin in g . In W estern E urop e in particu lar, little u rb a n g ro w th
arises fro m ru ra l-to -u rb a n m ig ra tio n , w ith rath er m o re associated w ith u rb a n -
to -u rb a n la b o r flow s. Such la b o r m ig ra tio n flow s are p rim a rily fro m cities spe-
cia lized in d e clin in g in d u stries to o th e r cities sp ecialized in g ro w in g sectors. In
the case o f the fo rm e r tra n sitio n eco n o m ie s, u rb a n g ro w th is associated w ith
in creasin g ru ra l-to -u rb a n la b o r flow s as w ell as u rb a n -to -u rb a n flow s. In fact,
o ver recent decades, m o re cities in th e d e ve lo p ed w o rld have a ctu a lly sh ru n k in
size than have g ro w n (U N -H A B IT A T 2008). A s such, the fl-s h a p e d rela tio n sh ip
betw een c ity per cap ita p ro d u c tiv ity and p o p u la tio n scale m ay have a ctu ally
shifted sligh tly to the left.
T h e various observation s reported here suggest that the sim ple M arshallian
relationship betw een urban scale and p ro d u ctivity no longer obtains in m any c o u n ­
tries. W h ile this relationship was fairly clear-cut at the turn o f tw entieth century, b y
the turn o f the m illennium the relationship betw een city size and perform ance had
becom e far m ore com plex than m any urban econ o m ic and new econ o m ic geogra-
phy m odels allow for. W h ile the sim ple M arshallian scale-p ro d u ctivity relationship
m ay be a useful approxim ation tod ay in the case o f the d evelopin g countries plus
the U nited States, Japan, and Korea, for m any advanced coun tries this appears to be
no longer the case.
T H E ROLE OF I N D U S T R IA L C L U S T E R IN G AND IN C R EA SIN G R u 189

G lo ba l C it ie s and C o n n e c t iv it y

T he reasons the urban scale-p ro d u ctivity relationship has becom e so co m p lex

appear to be related to issues o f knovvledge. For advanced econ om ies today, kn ow l-
edge (C aniels 2000), creativity, and in n ovation and entrepreneurship (Acs 2002)
appear to be far m ore im portan t for p ro d u ctivity than sim ply urban scale. Indeed,
am on g the rich countries, tvvelve o u t o f the fifteen m ost entrepren eurial cities are
sm all to m éd ium sized (M cC an n and Acs 2011), w hile eleven o u t o f the w o rld ’s
fifteen m ost com petitive cities in G D P per capita term s are sm all to m éd iu m sized
(C o rp o ra tio n o f L on d on 2008). T h e result o f this is that w hile very large cities such
as L ondon , N ew York, and París are high ly productive, across the O E C D m any o f
the richest cities are not very large (O E C D 2006).
Yet there appears at first to be so m eth in g o f a puzzle here. W h y sh ould know l-
edge issues today be associated w ith sm aller cities? It has lon g been u n d erstood that
the advantages o f cities are prim arily associated w ith know ledge spillovers, exactly
as b oth M arshall (1890,1920) and Jacobs (1970) observed, yet their argum ents tended
to point to urban scale as being a crucial facilitator o f such know ledge spillovers, for
exactly the puré agglom eration reasons discussed in section 2. In contrast, however,
the evidence in m any w ealthy countries presented earlier suggests that d u rin g the
tw entieth cen tury there have been qualitative changes that favor the com petitive
advantages o f sm aller or m ed ium -sized cities rather than very large cities. W h y
should such a change take place? W h ile the answ er is not im m ediately obviou s,
clues for the solu tion com e from Porter (1990), w h o points o u t that it is n ot regions
or cities that com pete but the firm s located in regions. As such, the reasons that
particular cities are highly prod uctive m ust lie in the types o f firm s that are located
there.
For firm s that invest heavily in know led ge assets in order to generate the
required returns to their know led ge investm ents, m an y o f the know ledge-based
firm s located in such cities m ust capture m arkets that extend w ell beyo n d the bor-
ders o f their o w n country. Traditionally, these returns were generated b y exports.
H owever, one o f the key features o f the current phase o f globalization is that there
is now an increasing prem ium associated w ith face-to-face contact (M cC an n 2008),
and this im plies that the global engagem ent facilitated b y direct international invest-
m ent is becom in g relatively far m ore im p o rtan t than exportin g (M cC an n 2009). As
such, the relationship betw een cities, countries, and globalization is increasingly
dependent o n the role o f m ultination al firm s as cond uits and facilitators o f such
global engagem ent (M cC an n and Acs 2011). N ow adays, the cities w ith the strongest
m ultination al connections w ill be those that m ost benefit from the global m arkets.
T his is, therefore, a rather different em phasis to the traditional agglom eration argu-
m ent based on urban scale as facilitating local know ledge spillovers. In the global
engagem ent argum ent, it is not the scale o f the city that is crucial but the nature o f
the firm s located in it.
190 U R BA N S T R U C T U R E , G ROW TH , A N D T H E D E V E L O P M E N T P R O C E SS

T h e m o st im p o rta n t p o in t here is th e p rin c ip ie o f w h a t so c io lo g ists cali

“co n n e ctiv ity .” T h e co n c e p t o f c o n n e c tiv ity is n o t c o n n e cted n e ss b u t ra th er the
c o m b in a tio n o f in te rn a tio n a l co n n e cted n e ss an d g e o g ra p h ic a ccessib ility, allied
w ith the fin a n cia l and d e c isio n -m a k in g in d e p e n d e n c e a n d d is c re tio n to act on
th o se d ecisio n s. C ities c o n ta in in g m a n y m u ltin a tio n a l firm s w ill e x h ib it c o n ­
n ectiv ity, and there is in crea sin g e v id e n c e th a t e c o n o m ic g ro w th at the in te rn a ­
tio n a l scale is in d eed b e in g in c re a sin g ly d o m in a te d b y n e tw o rk s o f p a rtic u la r
m a jo r u rb a n centers, w h ich are o fte n referred to as th e “g lo b a l” cities (Sassen
2001). T h ese g lo b a l cities are th e p a rticu la r u rb a n cen ters w h o se lo c a tio n s n o t
o n ly e x h ib it sig n ific a n t a g g lo m e ra tio n ad van tages b u t also in tera ct p rim a rily
w ith o th e r sim ila r g lo b a lly o rie n te d cities in o th e r c o u n trie s , ra th er th a n w ith
o th e r sm aller u rb a n cen ters w ith in th e ir o w n c o u n trie s. T h e reaso n is th a t the
a ctiv ities u n d e rta k en b y the m u ltin a tio n a l firm s lo ca ted in these g lo b a l cities
are in crea sin g ly o rie n te d to in te rn a tio n a l m arkets and as su ch are in c re a sin g ly
d ifferen t fro m th e d o m e stica lly o rie n te d roles u n d e rta k e n b y the firm s lo ca ted
in sm aller cities.
T h ere is already a v a riety o f e vid en ce p o in tin g to these glo b a l c ity and glo b al
engagem ent argum en ts. In m an y sectors such as fin an cial Services (M cC a n n and
Acs 2010; M asterCard 2008), there is alread y m u ch evid en ce that glo b a l m arkets
are in creasin gly d o m in ate d b y netw o rks o f glo b a l cities (Sassen 2002) such as
L o n d on , París, T okyo, Sydney, and N e w York. T h e in creasin g relative d o m in an ce
o f these glo b al cities appears to be associated w ith the d en sity o f k n o w led ge
(Sim m ie 2004) and in fo rm a tio n te c h n o lo g y assets in the c ity (Sassen 2002).
E m pirical evid en ce also suggests that the im p o rta n ce o f m a jo r u rb an n od es w ith in
such n etw orks is also rein forced b y the existen ce o f hubs w ith in the glo b al air, rail,
and m arin e tra n sp o rta tio n system s (Bel and Fageda 2008). In a d d itio n , there is
evidence that the p erfo rm an ce o f these d o m in a n t glo b al cities is p layin g an
in creasin gly im p o rta n t role even w ith respect to th eir o w n h in terlan d n ation al
and co n tin en tal eco n o m ie s.2
It is possible to intégrate the various types o f evidence o f global city engage­
m ent and the co n n ectivity o f the firm s located there b y d evelop in g extensive algo-
rithm s that intégrate a w ide range o f data o n trade flow s, airline passenger
m ovem ents, curren cy and financial flows, headquarters and corporate decision-
m aking centers, telecom m unication s traffic, and the q u an tity o f international assets
under m anagem ent from that city, and the presence o f m ajor m ultin ation al com pa-
nies (Taylor 2004; Taylor et al. 2008). B oth the g lo b a l-city centers o f com m erce
rankings (MasterCard 2008) and also the rankings o f global financial centers

2. Combining the OECD (2006) metropolitan productivity data with PPP national
productivity data at current prices (World Bank (2008) would also rank Singapore as the
30th highest productive city in the world, below Stuttgart and above Milán, and Hong
Kong as the 48th highest productivity city in the world, below Auckland and above
Hamburg.
THE ROLE OF IN D U S T R IA L C L U S T E R IN G AND IN C R EA SIN G R E T U R N S 191

(C o rp o ra tio n o f L ond on 2008) are calculated in this way, and these com posite
indicators provide the m ost com prehensive rankings o f the level o f global connec-
tivity o f different cities.
Table 8.9 provides the recent international rankings o f these tw o types o f global
engagem ent rankings for different cities. T h e three left-hand colu m n s report the
overall g lo b al-city centers o f com m erce rankings (MasterCard 2008). T hese ran k­
ings are the global engagem ent rankings o f cities calculated w ith respect to the vari-
ous types o f international engagem ent across all industries and all sectors, in clud in g
the valúes o f trade, in fo rm atio n and Com m unications tech n o lo gy traffic, airline
passenger m ovem ents, international funds under m anagem ent, and the n um ber o f
headquarters and d ecisio n -m akin g linkages. T he three righ t-han d co lu m n s report
the global financial centers (C o rp o ra tio n o f L ond on 2008) rankings that are ca lcu ­
lated specifically w ith respect to international financial flow s and financial m an age­
m ent indicators.
N ot surp risingly, as w e see in table 8.9, there is a very cióse co rresp o n d en ce
betw een the level o f glo b a l c o n n e ctiv ity o f the cities and its G D P per capita. O f
the top fifty m ost p ro d u ctive cities in the w o rld , th irty-five o f them are in the top
fifty o f the g lo b a l-city centers o f co m m erce rankings (M asterCard 2008). I f w e
were also to inelude the h ig h e st-p ro d u c tiv ity n o n -O E C D cities o f Tel A viv.T a ip ei,
Singapore, and H o n g K o n g ,3 then the n u m b er w o u ld rise to th irty-n in e o u t o f the
top fifty m ost p ro d u ctive cities in the w o rld that are also in the top fifty g lo b a lly
co n n ected cities. In term s o f the glo b al financial centers, ig n o rin g sm all tax
havens,4 but also in clu d in g H o n g K on g and Singapore, th irty o u t o f the top th irty-
seven global financial centers are a m o n g the w o rld ’s fifty m ost p ro d u ctive cities.
O f the ten largest cities fro m the new ly in d u strializin g w o rld , w h ich w e see fro m
table 8.4 are all am o n g the fifteen largest cities in the w o rld , o n ly nine are a m o n g
the w o rld ’s top seven ty w o rld w id e centers o f co m m erce (M asterCard 2008), and
o n ly tw o are also a m o n g the w o rld ’s top th irty-seven g lo b al centers o f finance
(M asterCard 2008).
T his striking corresponden ce betw een the w orld w id e centers o f com m erce
rankings (MasterCard 2008) as reflected in the three left-hand colu m n s o f table 8.8;
the global financial center rankings (C o rp o ra tio n o f L ond on 2008) as reflected in
the three righ t-hand colum ns o f table 8.9; and the w orld c ity -p ro d u ctiv ity rankings
(O E C D 2006) as reflected in table 8.6 is consistent w ith the argum ent that the cities
w ith the highest levels o f global co n n ectivity are also largely the w o rld ’s m ost
productive cities.
F ollow ing the argum ents o f Porter (1990), the reason there is such a high co r­
respondence betw een the global co n n ectivity o f a city and its p ro d u ctivity is that
cities that have a m ajor presence o f m ultination al com panies that are consistently
engaging globally are those cities that provide the best environm ent to achieve

3. For our purposes we can regará Singapore and Hong Kong as city-states
4. O f which there are thirteen in the top fifty world financial centers.
Table 8.9 Global City Rankings
G lo b a l C it y 2004 P o p o o o s' 2008 G lo b a l C it y In d ex G lo b a l F in a n cia l C e n tr e Pop ooos 2008 G lo b a l F in a n c ia l
C e n tr e In d ex

London 7,400 79 -1 7 London 7,400 7 9 5

New York 18,700 7 2 -7 7 New York 18,700 786

Tokyo 34,200 66.60 Hong Kong 7,000 695
Singapore 4,000 66.16 Singapore 4,000 6 7 5

Chicago 9 > 4 0 0 65.24 Zurich 2,500 665

Hong Kong 7,000 63.94 Frankfurt 5,600 642
Paris 11,200 63.87 Geneva 4 5 0 640
Frankfurt 5,600 ► 62.34 Tokyo 34,200 628
Seoul 23,500 61.83 Sydney 4,200 621
Amsterdam 7,500 60.06 Boston 4 ,4 0 0 618
Madrid 5,600 5 8 .3 4 San Francisco 4,200 614
Sydney 4,200 5 8 .3 3 Dublin 1,600 613
Toronto 4 ,7 0 0 58.16 Paris 11,200 612
Copenhagen 2,400 5 7 -9 9 Toronto 4 ,7 0 0 610
Zurich 2,500 56.86 Washington 5,ioo 5 9 7

1 Population data come from the OECD metropolitan database (OECD 2006,38-40). The data for Amsterdam are that o f the Randstad, and the data for London are only the area
within the London city boundaries and do not inelude the whole London commuting area (OECD 2006).
2 Combining the OECD (2006) metropolitan productivity data with PPP national productivity data at current prices (World Bank (2008) would also rank Singapore as the
thirtieth highest productivity city in the world, below Stuttgart and above Milán, and Hong Kong as the forty-eighth highest productivity city in the world, below Auckland and
above Hamburg.

Sources: Global City Index (MasterCard 2008); Global Financial Centre Index (Corporation of London 2008); city populations (OECD 2006); World Bank 2008).12McCann and Acs (2011).
TH E ROLE OF IN D U S T R I A L C L U S T E R IN G AND IN C R E A S IN G R E T U R N S 193

m arket scale effects. As such, it is the m ultination al firm s that facilítate and provide
for regional connectivity. M oreover, the know ledge spillover argum en ts typ ically
em ployed in urban e co n o m ic explanations o f agglom eration suggest that the
im portance o f urban global co n n ectivity m ust be just as im p o rtan t for sm all or
m edium -sized firm s as for large firm s. T his is because it is the local m u ltin ation al
firm s that vvill be the p rim ary cond uits via w hich global know led ge flow s opérate
and the natural channels via w h ich d om estic firm s can distribute their goods
(M cC an n and Acs 2011). As such, the export potential o f non exporters and n o n m u l-
tinational firm s sh ould also be higher in city-regions that are characterized b y lots
o f m ultination al firm s. Indeed, this is exactly w hat was fo u n d b y A itken, H anson,
and H arrison (1997), w h o d em onstrated that it is pro xim ity to m ultinationals, and
not pro xim ity to other exporters, that increases export propensities. M oreover, their
finding was robust to the inclusión o f other m easures such as the overall industrial
activity o f the región, p ro xim ity to the capital city and border regions, and price and
costs variables. In addition, m ore recent evidence from Sweden shows that the
export propensities o f n on m u ltin atio n al dom estic firm s are not o n ly higher for the
dom estically o w n ed firm s located in larger and denser cities b u t also correlated w ith
the degree o f m ultination al activity in that city (A ndersson 2009; Johansson and
L o o f 2009).
T hese findings do not im p ly that sm all and m ed ium -sized n o n m u ltin ation al
com panies are not im portan t for urban grow th. O n the contrary, w e k n o w that
sm all entrepreneurial start-ups are critical for local in n ovation and city grow th (Acs
2002). Ñ or do these observation s im ply that agglom eration and scale are not im p o r­
tant for city perform ance. Rather, the global city argum ents o u tlin ed earlier im p ly
that cities w ith greater levels o f global co n n ectivity are those that provid e the great-
est global m arket potential. In this case, now adays the size o f the city itself o r even
the size o f the dom estic e co n o m y is relatively m uch less im p o rtan t than it used to
be, whereas international accessibility and co n n ectivity o f the city are far m ore
im portan t than they used to be. Indeed, som e recent evidence suggests that the c o n ­
nectivity issue now adays dom inates, both city size and d om estic m arket size (Bel
and Fageda 2008).
AU the evid e n ce presented in section s 4.1 and 4.2 suggests that u rb a n scale is
o n ly part o f the sto ry o f w ealth creatio n . In the n od e and n e tw o rk stru ctu re o f
in tern atio n a l trad e and fo reig n d irect in vestm en t (M cC a n n 2008) in to d a y ’s
glo b a l m arketplace, n ot o n ly is g lo b a l c o n n e c tiv ity an altern ative w ay o f ach iev-
ing the u rb an scale and p ro d u c tiv ity effects first o b served b y M arsh all (1890) and
Jacobs (1970), b u t in m an y situ a tio n s such c o n n e ctiv ity appears to be a m o re
efficien t w ay o f a ch iev in g such scale effects. U rban scale still appears to o p érate
largely in the m an n er d escrib ed b y M arsh all (1890, 1920) and Jacobs (1970) in
the ases o f the U n ited States, Japan, K orea, and the em erg in g eco n o m ie s o f b o th
the d evelo p in g w o rld and the fo rm e r tra n sitio n eco n o m ies. In the rest o f the
ad van ced co u n tries, how ever, glo b a l c o n n e c tiv ity and the presence o f m u ltin a ­
tion al firm s appear to be a m u ch m ore im p o rta n t ex p la n a tio n o f u rb a n p e rfo r­
m ance th an u rb a n scale.
1 94 UR BA N S T R U C T U R E , G ROW TH, AND T H E D E V E L O P M E N T P R O C E SS

U r b a n S iz e , P e r f o r m a n c e , a n d L a n d -U s e
P l a n n in g I s s u e s

The global co n n ectivity argum ents o u tlin ed earlier are largely overlooked b y u rban
econom ists, yet th ey have p ro fo u n d im plication s for h o w w e consid er the role o f
urban plan n in g in the tw enty-first century. T h e reason is that th ey im m ed iately cali
into question land-use p lan n in g strategies that focus p rim a rily on allow in g urban
bound aries to expand m ore o r less co n tin u o u sly in response to urban land-use
dem ands. If sm aller cities are often m ore p rod uctive than large cides, then m ore
com plex approaches to regional p lan n in g m ay be required, such as the p ro m o tio n
o f polycentric city-regions or targeted urban land-use zon in g. H ow ever, w h at is not
yet clear is w h ether or w h en it is preferable to plan for urban polycentricity, to plan
for urban scale via land release, or plan via specific industrial and spatial zonin g.
F ollow ing the argum ents o u tlin ed in section 2, it is clear that urban expansión
facilitated by plan n in g policies that p ro m o te urban land availability are consistent
p rim arily w ith the puré agglom eration m od el described in table 8.1. As we have seen
in section 2, agglom eration argum ents based on M arshall (1890,1920) and Jacobs
(1970) point to urban expansión as a key feature o f eco n o m ic grow th, and these
argum ents therefore im p ly that land-use p lan n in g strategies sh ould p rim arily be
about releasing land for urban expansión. In su p p o rt o f these argum ents, there is a
large literature that argües that urban ffin g e bou n d aries o r greenbelts w h ich lim it
urban expansión necessarily im pose land-use inefficiency and w elfare costs on soci-
ety. O ne m anifestation o f this is that land prices w ill be m uch higher in restrictive
planning areas than in the localities that allow land to be m ore ffeely released in
response to dem and. In the U nited States, m any cities have adopted these m ore flex­
ible and less restrictive land-use policies over m any years, and those that have done
so have provid ed for m ore affordable living than those w ith high ly restrictive
policies.
In static term s, the o u tco m e o f such restrictive land-use policies in the presence
o f the puré agglom eration m odel w ill be that at the m argin land is not bein g allo-
cated efficiently b y the m arket. T his inefficiency w ill be revealed in term s o f signifi-
cant rural-urban discontinuities in the price o f land on the urban fringe. M oreover,
to the extent that higher-incom e groups are able to pay land price prem ium s, these
h igher-incom e groups w ill also be able to better internalize the rents associated w ith
being adjacent to such boundaries, to the d etrim ent o f low er-in com e groups
(M cC an n 2001). As such, the negative welfare consequences o f restrictive land-use
policies m ay adversely im pact on low er-incom e groups m ore than h igher-in com e
groups. M eanw hile, in dynam ic term s, these sam e agglom eration argum ents also
suggest that the level o f local know ledge spillovers and other such positive externali-
ties w ill be artificially lim ited b y such restrictive land-use plan n in g strategies. In the
long run this w ill reduce econ o m ic grow th and con sequen tly lo n g -ru n social
welfare.
THE ROLE OF IN D U S T R IA L C L U S T E R IN G AND IN C R E A S IN G R E T U R N S 195

Land release at the urban fringe sh ould therefore be encouraged in situations

where urban scale is the m ost im p o rtan t feature for p ro m o tin g know ledge spill-
overs and other M arshall- and Jacobs-type externalities. In social environm ents
w ith a great deal o f labor ch u rn and lovv lon g-term interpersonal loyalty or trust
betvveen individuáis, allo w in g a puré agglom eration to grow via land research on
the urban fringe vvill therefore be the best regional plan n in g strategy. To the extent
that transport policies are also involved in these urban expansión policies, a m o n o -
centric and radial pattern is also likely to be the m ost appropriate spatial structure.
T he argum ents in the previous sections suggest that plan n in g policies based on
land release as a m eans o f facilitating urban grovvth m ay be appropriate in parts o f
the d evelopin g vvorld, the form er transition econom ies, and also the U nited States,
Korea, and Japan, H owever, these conclusions m ust be treated w ith a degree o f
caution. T h e reason is that if the urban grow th is a result o f interregional m igration,
one problem o f puré agglom eration from a social welfare perspective is that m ig ra­
tion flow s to the grow in g urban center are always likely to be too large. T h is is
because m igration is driven by urban wages, w hich them selves reflect average urban
p ro d u ctivity rather than m arginal urban productivity. As such, the wage signáis that
trigger m igration m ay im pose social welfare costs due to excess congestión associ-
ated w ith the cities b eing to o large. O n these argum ents, the overcon gestion o f m any
o f the m egacities o f the d evelop in g w o rld poses m ajor challenges in term s o f pov-
erty reduction, environm ental d egradation, health care, and housing.
These argum ents suggest that in som e cases city sizes should actually be con-
strained by restrictive land-use plann ing systems in order to prom ote efficien cy and
welfare. Indeed, as we have already seen, the evidence provid ed in the preceding sec­
tions indicates that in m any coun tries the straightforw ard M arshall- or Jacobs-type
links betw een city size and p ro d u ctivity d o not appear to opérate. T hese observa-
tions therefore suggest that in som e situations, other approaches to plan n in g for
h igh-perform an ce urban environm ents m ay be possible or even preferable to p o li­
cies that em phasize increasing urban land availability.
If we m ove away from discussions o f the relationship between urban scale and
wealth and consider h o w the specialization -versus-diversity debate relates to urban
scale, the picture becom es even m ore com plex. A sim ple interpretation o f the analy-
sis in the second section, set in a new econ o m ic geography or central place type o f
analysis, w o u ld im ply that larger cities w ill tend to be m ore diverse than sm aller
cities, w ith larger cities relying rather m ore o n classic urbanization agglom eration
effects w hile sm aller cities w ill rely m ore on industrial com plex and social n e tw o rk -
type local econ o m ic structures em bedded in localization econom ies. A t the same
tim e, we w ou ld expect social netw orks to link these various nodes in the urban
hierarchy as w ell as transport systems. H owever, in the case o f Europe, the picture
does not even closely correspond to these argum ents. T he legacy effects o f different
institutional and cultural contexts on national and regional land-use plan n in g sys­
tems (N adin and Stead 2008) m ean that larger cities in Europe are not necessarily
m ore diverse than sm aller cities. T h e characteristics o f the m ajor globally connected
cities in Europe have developed a range o f typ ologies (B agnasco and Le Gales 2000;
196 UR BA N S T R U C T U R E , G ROW TH, AND T H E D E V E L O P M E N T P R O C E S S

Le Gales 2002) in w h ich scale and global co n n ectivity are n ot necessarily related to
urban diversity. T h e m ost notable case here is that o f L o n d on , w h ich is actually less
sectorally diverse than m any m u ch sm aller cities even w ith in the U nited K in gd o m
(D ew hurst and M cC an n 2002,2007). M u ch m ore research is required regarding the
specialization -versus-diversity debate in the con text o f large cities, and h o w these
patterns relate to polycentric urban system s, because in m an y advanced countries,
larger cities are n o t necessarily m ore diverse than sm aller cities.
For o u r purposes, the im p ortan ce o f these observation s lies in the fact that the
urban p ro d u ctiv ity evidence discussed earlier suggests that in m an y advanced econ-
om ies, p ublic p o licy and land-use p lan n in g strategies that are designed to encour-
age and p rom ote co n n ectivity betw een localities m ay be m ore im p o rta n t than
urban expansión policies. T hese alternative types o f p lan n in g policies fo cu s o n the
developm ent o f high-sp eed and h igh -q u ality tran sp ort infrastructure, w h ich allow s
for rapid flows o f large num bers o f people and go od s betw een groups o f cities. Such
policies are often described as p ro m o tin g p olycen tric system s o f cities, akin to those
that exist in the R andstad regions o f the w est o f the N etherlands, the R uhr in the
N ordrhein-W estphalia región o f G erm any, or in L o m b ard y and P ied m on te in
northern Italy. A s such, this is m u ch m ore o f a n etw o rk approach to spatial p lan ­
ning, rather than an urban-scale approach to land-use plann ing. Indeed, this n et­
w o rk and polycentric approach is the d o m in an t th in kin g beh in d spatial plan n in g
principies adopted in m any European countries such as the N etherlands and
Germ any. In addition, p ro m o tin g co n n ectivity betw een key city-regio n s b y devel-
oping transport infrastructure along particular strategic corrid ors is a m ainstay o f
E U regional plan n in g and cohesión policy.
Yet a polycentric system o f cities, in w h ich the spatial expansión possibilities o f
an individual urban area are restricted in favor o f im proved interurban tran sp orta­
r o n netw orks, w ill reduce spillover possibilities, precisely because o f the artificially
im posed spatial separation o f m an y o f the individuáis. As such, in order for a
polycentric system to be successful, the social relations betw een the people in the
different urban centers m ust be strong and long-lasting. T h e reason is that a p o lycen ­
tric netw ork can be efficient o n ly i f sufficient num bers o f people in each constrained
urban area k n o w sufficient num bers o f people in the o ther urban nodes, such that
th ey are w illin g to cooperate and trust each other in order to overeóm e the barriers
to interpersonal contact im posed by the spatial separation. O n this argum ent, the
social environm ents in the N etherlands, N ordrhein-W estphalia, and n o rth ern Italy
m ay indeed be ideal locations for such polycentric plan n in g approaches precisely
because they are localities w ith high levels o f social capital and lo w labor turnover.
In these localities, the integration o f land-use plan n in g w ith a polycentric node-
netw ork structure o f transport linkages is the m ost appropriate p o licy setting.
Regional planning strategies based on policies o f land-use zo n in g at specific
locations for particular types o f activities fforn particular industries are consistent
w ith the industrial com plex m odel. In this case, having firm s in the sam e o r closely
related industries allows the clustered group o f firm s to becom e increasingly effi­
cient b y reducing interfirm transactions costs. Such zo n in g on the urban firinge is
TH E ROLE OF I N D U S T R I A L C L U S T E R IN G AND IN C R EA SIN G R E T U R N S 197

often undertaken because the industries involved generate significant p o llu tion .
However, even in the absence o f significant p ollution , such plan n in g policies w ill be
appropriate w here industries co m m it significant capital investm ents in order to
particípate in a clustered industrial co m p lex m odel.
For each o f these three possible broad planning strategies, m ore generally as we
have already seen, global con n ectivity is b ecom in g increasingly im portan t in term s
o f urban perform ance. Prom otin g such global connectivity depends on the develop-
m ent o f appropriate transport infrastructure and, in particular, air transport infra-
structure. O bviously, not all cities in a co u n try can becom e global cities, and not all
cities will becom e nodes w ith in the global netw orks o f intercontinental transporta-
tion (Bel and Fageda 2008). T herefore, the argum ents regarding the know ledge
exchanges and access to global know ledge flows im ply that co n n ectivity betw een a
co u n try’s various urban centers and the co u n try’s global city or cities is essential, as
it is the key global city that w ill act as the co u n ty’s nodal access point to the global
m arket. As such, urban plann ing w ill increasingly becom e interrelated w ith trans-
portation and m obility planning. For m any countries inter-urban co n n ectivity w ill
becom e increasingly im portant relative to urban scale, and polycentric plann ing
policies w ill becom e increasingly appropriate. This is particularly so in European
econom ies, w here the history o f spatial p opu lation change, aging dem ographics, and
land-use planning ffam ew orks largely preeludes the grow th o f m egacities.

REFERENCES

Acs, Z. J. 2002. I r m o v a t i o n a n d t h e G r o w t h o f C i t i e s . Cheltenham: Edward Elga

Aitken, B., G. Hanson, and A. E. Harrison. 1997. “Spillovers, Foreign Investment, and
Export Behaviour.” J o u r n a l o f I n t e r n a t i o n a l E c o n o m i e s 43:103-132.
Andersson, M. 2009. “External Trade and Internal Geography.” CESIS Working Paper.
Stockholm: Jonkoping University, Jonkoping and Royal Institute of Technology.
Bagnasco, A., and P. Le Gales. 2000. C i t i e s i n C o n t e m p o r a r y Cambridge: Cambridge
University Press.
Bel, G., and X. Fageda. 2008. “Getting There Fast: Globalization, Intercontinental Flights
and Location of Headquarters.” J o u r n a l o f E c o n o m i c G e o g r a p h y 8:471-495.
Best, M. H. 1990. T h e N e w C o m p e t i t i o n : I n s t i t u t i o n s o f I n d u s t r i a l R e s t r u c t u r i n g . Cambridge:
Polity Press.
Caniels, M. C. J. 2000. K n o w l e d g e S p i l l o v e r s a n d E c o n o m i c G r o w t h : R e g i o n a l G r o w t h
D i f f e r e n c e s a c r o s s E u r o p e . Cheltenham: Edward Elgar.

Chandler, T. 1987. F o u r T h o u s a n d Y e a r s o f U r b a n G r o w t h : A n H i s t o r i c a l C e n s u s . Lampeter,

UK: Edwin Mellen Press.
Collier, P., and A. J. Venables. 2008. "Trade and Economic Performance: Does Africa’s
Fragmentation Matter?” Working paper, Department of Economies,
Oxford University
Corporation of London. 2008. T h e G l o b a l F i n a n c i a l C e n t r e s I n d e x 3 . London: Corporation
of London.
198 U R BA N S T R U C T U R E , G ROW TH, AND T H E D E V E L O P M E N T P R O C E S S

Crafts, N. 2004. “Globalization and Economic Growth: A Historical P erspective World

Economy 27 (i ):45~58.
Dewhurst, J. H. Ll., and P. McCann. 2002. “A Comparison of Measures of Industrial
Specialisation for Travel-to-Work Areas in Great Britain 1981-97.” Regional Studies
36:541-551.
---------. 2007. “Specialisation and Regional Size.” In New Directions in Economic ,
ed. B. Fingleton. Cheltenham: Edward Elgar.
Duranton, G., and D. Puga. 2004. “Microfoundations of Urban Agglomeration Economies.”
In Handbook of Regional and Urban Economies. Vol. 4, Ciñes and Economic Geography,
ed. J.-F. Thisse and J. V. Henderson, 2063-2117. North Holland: Elsevier.
EC. 2009. “Metropolitan Regions in the EU.” Regional Focus Paper 01/2009. Brussels:
European Commission.
Ellison, G., and E. L. Glaeser. 1997. “Geographic Concentration in US Manufacturing
Industries: A Dartboard Approach.” Journal ofPolitical Economy 105:889-927.
Findlay, R., and O ’Rourke, K. 2007. Power and Plenty: Trade, War, and the World Economy
in the Second Millennium. Princeton, NJ: Princeton University Press.
Fothergill, S., M. Kitson, and S. Monk. 1985. Urban Industrial Decline: The Causes o f Urban/
Rural Contrasts in Manufacturing Employment Change. London: HMSO.
Fujita, M., P. Krugman, and A. J. Venables. 1999. The Spatial Economy: Ciñes Regions and
International Trade. Cambridge, MA: MIT Press.
Gordon, I. R., and P. McCann. 2000. “Industrial Clusters: Complexes, Agglomeration and/
or Social Networks?” Urban Studies 37:513-532.
Granovetter, M. 1973. “The Strength of Weak Ties.” Journal of Sociology
78:1360-1380.
Grindley, P. C., and D. J. Teece. 1997. “Managing Intellectual Capital: Licensing and
Cross-Licensing in Semiconductors and Electronics.” California Management Review
39 (2):8-4i.
Iammarino, S., and P. McCann. 2006. “The Structure and Evolution of Industrial Clusters:
Transactions, Technology and Knowledge Spillovers.” Research 35:1018-1036.
Isard, W., and R. E. Kuenne. 1953. “The Impact of Steel upon the Greater New York-
Philadelphia Industrial Región.” Review o f Economies and Statístícs 35:289-301.
Jacobs, J. 1970. The Economy of Ciñes. New York: Vintage Books.
-------- . 1984. Ciñes and the Wealth ofNatíons. New York: Vintage Books.
Johansson, B., and H. Loof. 2009. “How Are Non-MNEs Affected by Globalization?” CESIS
Working Paper. Stockholm: Jonkoping University, Jonkoping and Royal Institute of
Technology.
Krugman, P. 1991a. Geography and Trade. Cambridge, MA: MIT Press.
-------- . 1991b. “Increasing Returns and Economic Geography.” of Political Economy
99:483-499.
Le Gales, P. 2002. European Ciñes. Oxford: Oxford University Press.
Maddison, A. 2006. The World Economy. Vol. 1, A Millennial Perspective. Vol. 2, Historical
Statistics. París: OECD.
Marshall, A. 1890. Principies of Economies. London: Macmillan.
-------- . 1920. Principies of Economies. 8th ed. London: Macmillan.
MasterCard. 2008. Worldwide Centers of Commerce Index, www.mastercardworldwide.com.
McCann, P. 1995. “Rethinking the Economies of Location and Agglomeration.” Urban
Studies 32:563-577.
---------. 2001. Urban and Regional Economies. Oxford: Oxford University Press.
-------- . 2005. “Transport Costs and New Economic Geography.” of Economic
Geography 5:305-318.
THE ROLE OF IN D U S T R IA L C L U S T E R IN G AND IN C R E A SIN G R E T U R N S 199

-------- . 2007. “Sketching Out a Model of Innovation, Face-to-Face Interaction and

Economic Geography.” Spíifúi/ E c o n o tn ic 2
si :117-134.
ly
a
n
A

-------- . 2008. “Globalization and Economic Geography: The World Is Curved, Not Fiat.”
Cambridge Journal of Regions, Econotny and Society 1:351-370.
-------- . 2009. “Globalisation, Multinationals and the BRIICS Countries.” In Globalisation
and Emerging ,E
iesed. R. Lattimore and R. Safadi, 71-116. Paris: OECD.
m
n
co
McCann, P., and Z. J. Acs. 2011. “Globalisation: Countries, Cities and Multinationals.”
Regional Studies. 45:17-32.
McCann, P., and S. C. Sheppard. 2003. “The Rise, Fall and Rise Again of Industrial Location
Theory.” Regional Studies 37:649-663.
Moses, L. N. 1958. “Location and the Theory of Production.” Quarterly Journal of
Economies. 73:259-272.
Nadin, V., and D. Stead. 2008. “European Spatial Planning Systems: Social Models and
Learning.” disP 172 (i):25—47.
OECD. 2006. Competitive Cities in the Global Econotny. Paris: OECD.
Porter, M. E. 1990. The Competitive Advantage ofNations. New York: Free Press.
Rosenthal, S. S., and W. C. Strange. 2004. “Evidence on the Nature and Sources of
Agglomeration Economies.” In Hatulbook of Urban and Regional Economies, Vol. 4, ed.
V. Henderson and J.-F. Thisse, 2119-2171. Amsterdam: North-Holland.
Sassen, S. 2001. The Global City. Princeton, NJ: Princeton University Press.
---------, ed. 2002. Global Networks: Linked Cities. London: Routledge.
---------. 2006. Cities in a World Econotny. Thousand Oaks, CA: Pine Forge Press.
Scott, A. J. 1988. New Industrial Spaces. London: Pión.
Simmie, J. 2004. “Innovation and Clustering in the Globalized International Economy.”
IJrban Studies 41:1095-1112.
Taylor, P. J. 2004. World City NetWork: A Global Urban Analysis. London: Routledge.
Taylor, P. J., D. M. Evans, and K. Pain. 2008. “Application of the Interlocking NetWork
Model to Mega-City Regions: Measuring Polycentricity within and beyond City-
Regions.” Regional Studies 42:1079-1093.
UNFPA. 2007. State of the World’sPopulation: Unleashing the Potential of Urban G
New York: United Nations Population Fund.
UN-HABITAT. 2008. State of the World's Cities Report 2008/09. Nairobi: United
Nations-HABITAT Programme.
Venables, A. J. 2006. “Shifts in EconomicGeography and Their Causes.” Federal Reserve
Bank of Katisas City Economic Review 91 (4):6i-85.
Weber, A. 1909. Theory of the Location of Industries. Trans. C. J. Friedrich, 1929. Chicago:
University of Chicago Press.
Williamson, O. E. 1975. Markets and Hierarchies. New York: Free Press.
World Bank. 2007. 2007 World Development Indicators. Washington, DC: World Bank.
---------. 2008.2008 World Development Indicators. Washington, DC: World Bank.