You are on page 1of 28

Urban Geography

ISSN: 0272-3638 (Print) 1938-2847 (Online) Journal homepage: http://www.tandfonline.com/loi/rurb20

Are major canadian city-regions monocentric,


polycentric, or dispersed?

Matthias N. Sweet, Bronson Bullivant & Pavlos S. Kanaroglou

To cite this article: Matthias N. Sweet, Bronson Bullivant & Pavlos S. Kanaroglou (2016): Are
major canadian city-regions monocentric, polycentric, or dispersed?, Urban Geography, DOI:
10.1080/02723638.2016.1200279

To link to this article: http://dx.doi.org/10.1080/02723638.2016.1200279

Published online: 29 Jun 2016.

Submit your article to this journal

Article views: 10

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=rurb20

Download by: [Library Services City University London] Date: 01 July 2016, At: 20:50
URBAN GEOGRAPHY, 2016
http://dx.doi.org/10.1080/02723638.2016.1200279

Are major canadian city-regions monocentric, polycentric, or


dispersed?
Matthias N. Sweeta, Bronson Bullivantb and Pavlos S. Kanaroglouc†
a
School of Urban & Regional Planning, Ryerson University, Toronto, ON, Canada; bMalatest & Associates,
Ltd., Victoria, BC, Canada; cMcMaster Institute for Transportation and Logistics, School of Geography and
Earth Sciences, McMaster University, Hamilton, ON, Canada
Downloaded by [Library Services City University London] at 20:50 01 July 2016

ABSTRACT ARTICLE HISTORY


Relatively little research has explored spatial structure in modern Received 19 November 2014
major Canadian regions. Three common models are monocentricity, Accepted 6 May 2016
polycentricity, and dispersion, but these are not always mutually KEYWORDS
exclusive in the complex spatial structures of contemporary city- Canada; dispersal;
regions. Shifts between these models are discussed in the context monocentricity;
of three explanations of economic growth and restructuring: acces- polycentricity; spatial
sibility, municipal competition, and globalization. All three explana- structure; urban form
tions suggest a trend away from monocentricity. While this appears
clearly in US cities, disagreement surrounds whether Canadian cities
are following the same path. This study uses cross-sectional data
from InfoGroup in 2011 to estimate the relative strengths of mono-
centricity, polycentricity, and dispersion for characterizing eight
major regions. Results indicate that elements of each model are
evident in all eight study regions, but each tends to dominate in
different contexts. When focusing on Montreal, Toronto, and
Vancouver, results imply that all three forces which guide spatial
structure play a role and that job centers appear to play a particu-
larly important structural role in larger regions.

Introduction
While urban spatial structure and land use patterns are central determinants of city
function, relatively little previous research has explored spatial structure in modern major
Canadian cities, perhaps most critically due to insufficient data. Urban economic literature
from the United States has documented the extent to which cities have become more
polycentric and dispersed over time, finding that the monocentric model of city structure,
while still important, is less important than in the past (Arribas-Bel & Sanz-Gracia, 2014;
Gordon & Richardson, 1996). Much disagreement surrounds whether Canadian cities have
followed the same path toward dispersed and polycentric cities (Filion, Buntin G,
McSpurren, & Tse, 2004). This study focuses on these questions.
This study begins by reviewing the monocentric, polycentric, and dispersed models
of spatial structure at the scale of the city-region. The extent to which each model
reflects the spatial organization of cities has implications for how policy shapes land

CONTACT Matthias N. Sweet matthiassweet@ryerson.ca


†Deceased
© 2016 Informa UK Limited, trading as Taylor & Francis Group
2 M. N. SWEET ET AL.

uses and redistribute the benefits from intervention. This study uses 2011 data from
InfoGroup to estimate the relative strength of monocentric, polycentric, and dispersed
models of spatial structures in characterizing eight major Canadian Census
Metropolitan Areas (CMAs)—henceforth referred to as city-regions. The study regions
contain 50.3% of the Canadian population and include Toronto, Hamilton, Ottawa,
Vancouver, Calgary, Winnipeg, Edmonton, and Montreal (Statistics Canada, 2011).
Three key drivers of city growth and restructuring, accessibility, governance, and
globalization, are discussed to better interpret the contexts under which different
types of spatial structure might be expected.

Literature review
Downloaded by [Library Services City University London] at 20:50 01 July 2016

Urban spatial structure and land use patterns are both causes and consequences of city
function (Anas, Arnott, & Small, 1998). Knowledge and resource sharing at city and
global scales have strengthened big-city agglomeration economies (Shearmur, 2012).
Spatial structure and land use patterns in Canadian urban regions shape public policy,
how social outcomes can be achieved, and how growth might occur (or should occur)
(Filion, 2000). As such, while early geographic theories of spatial structure were largely
descriptive, more recent research have built normative theory partly in an effort to
improve policymaking (Beauregard, 2007).
The scale at which spatial relationships in land use patterns matter is important for
policymaking (Anas et al., 1998). National, local, and regionally scaled housing policy,
public service improvements, and infrastructure projects all impact spatial structure
(Banister & Berechman, 2000). Spatially fixed interventions shape regional outcomes
and impact society and the reallocation of resources among residents and businesses.
With localized or dispersed spatial structure, spatially fixed policy interventions may
diffuse through the regional land and labor markets—implying a weaker role for
policymaking (Gordon & Richardson, 1996). Urban spatial structure is one way to
understand cities, but it is also important for exploring how benefits and costs of public
policy disperse within or across regions.

The monocentric city


The monocentric city is an important normative theory of city form, which has
emerged from the field of urban economics. It establishes how early transportation
technologies created concentrated central cities and how automobility has enabled
dispersed spatial structures (Freestone & Murphy, 2007; Polese & Champagne, 1999).
Alonso (1964) proposed the model, which holds that multiple agents act according to
economic motivations, and it was updated in subsequent work by Muth (1969), Mills
(1967), and many others (Anas et al., 1998). The model establishes the expectation that
land values decline with distance from the central business district (CBD) and it has
been further expanded to include explanations for different housing types, commercial
and industrial uses, and prices (Brueckner, 2007; Wheaton, 2004). Accordingly, one
would expect one central job center to which workers commute and the cost of
accessibility enters as the chief determinant of land rent and use. The model charac-
terizes many elements of city organization and function through economic agents
URBAN GEOGRAPHY 3

(Alonso, 1964; Mills, 1967; Muth, 1969). Accessibility, travel costs, externalities (both
positive, from urban access, and negative, from urban inconvenience), and the char-
acteristics of residents and firms feature as key determinants of city structure (Anas &
Xu, 1999). Most subsequent urban economic models of urban spatial structure are
based on mathematical revisions or special cases of the monocentric city.

Polycentric cities
Polycentric models of cities hold that multiple centers develop to establish agglomera-
tion economies which together function as a region (Fujita & Ogawa, 1982) or system
of regions, often referred to as mega regions (Fujita, Krugman, & Mori, 1999; Gordon &
Downloaded by [Library Services City University London] at 20:50 01 July 2016

Richardson, 1996; Ross, 2009). These systems of centers include both primary urban
centers and secondary employment nodes (referred to in this paper as subcenters),
which are usually auto-centric (Cervero, 1986). While the monocentric model of cities
continues to reflect the spatial structure of most North American city-regions, large
regions have become more polycentric (Arribas-Bel & Sanz-Gracia, 2014). The poly-
centric city model represents a different interpretation of how accessibility explains
spatial structure, even while subcenters appear to often be imperfect substitutes for a
strong core (Meijers, 2008). Instead, the polycentric model highlights new conditions
under which localized physical proximity is important in supporting regionally and
globally scaled processes.
Polycentric regions have been popularized using the construct of Edge Cities
(Garreau, 1992) but critiqued on the basis of being overly descriptive and without
strong normative theory (Parr, 2004). There appear to be many routes to the poly-
centric city. Research indicates that polycentricity evolves from the codependencies of
formerly independent smaller cities into larger regional conurbations, especially in the
European context (Lopez, 2005; Romero, Solis, & De Urena, 2014). But polycentricity
more frequently stems from the creation of new suburban economic clusters outside of
the urban core in the context of North American city-regions (Cervero, 1986).
Regionally scaled spatial dependencies manifest through clusters of sub-nodes
(Giuliano, Redfearn, Agarwal, Li, & Zhuang, 2007; Giuliano & Small, 1991; McMillen
& Smith, 2003; Song, 1992).
One might expect the polycentric region to emerge from a revision to the Alonso–
Mills–Muth monocentric model of the city (Fujita & Ogawa, 1982). As land prices,
regulation, and congestion constrain the benefits and drive up the prices of central
locations, polycentric nodes develop (Fujita & Ogawa, 1982; Henderson & Mitra, 1996)
partly to enable better labor market access for firms (Anas & Xu, 1999). Thus, using the
monocentric city as a base model which assumes travel costs as the key determinant,
one would expect cities to become polycentric in response to congestion and localized
agglomeration economies (Fujita & Ogawa, 1982; Maoh & Kanaroglou, 2007; McMillen
& Smith, 2003; Wheaton, 2004).

Dispersed cities
Others argue that cities are now dispersed and edgeless cities, and that the transition
toward dispersion emerges from the declining role of physical proximity as a means of
4 M. N. SWEET ET AL.

fostering accessibility in the production process (Gordon & Richardson, 1996; Lang,
2003; Shearmur, 2012; Soja, 2011; Wheaton, 2004). Targeted regionally-scaled (Filion,
2000) and nationally-scaled (Baum-Snow, 2007) public policies have encouraged the
dispersion of jobs, people, and land uses into suburbs and exurbs and have reduced the
role of proximity-related city benefits, resulting in time–space compression (Harvey,
1990).
Technology has featured prominently in the trend toward dispersion. For example,
automobiles have increased mobility, making proximity relatively less important and
leading one to expect dispersed land use patterns around central cities when agglom-
eration economies are weak and transportation costs are low (Wheaton, 1998, 1974).
Glaeser and Kahn (2003) conclude that automobility, the core precondition for a
Downloaded by [Library Services City University London] at 20:50 01 July 2016

dispersed spatial structure, is so pervasive in North American cities that it shapes all
attributes of society and leaves the relative spatial advantages within regions at local
scales less important than in the past. Beyond automobility, others focus on how
telecommunications have enabled production and communication without physical
proximity (Castells, 1996; Golob & Regan, 2001) and can reduce the role of urban
accessibility in shaping growth (Gaspar & Glaeser, 1998). But empirical suggests that
telecommunications and proximity-related city benefits are complements, not substi-
tutes (Gaspar & Glaeser, 1998; Mokhtarian & Meenakshisundaram, 1999).

Forces shaping urban growth


Scholars have focused on many reasons for urban restructuring, some of which are
internal characteristics of regions, including accessibility to important economic inputs
(e.g. labor, capital, or natural endowments) (Storper, 2010) and governance (Storper,
2010), and some of which are external to city-regions, including the forces of globaliza-
tion (Friedmann, 1986; Sassen, 2001). Much debate surrounds the conditions under
which combinations of these causal factors lead to both net regional growth and
localized growth and restructuring among parts of a given city-region.

Accessibility
Many elements of accessibility and spatial proximity contribute to city growth between
and within city-regions. Agglomeration economies refer to the conditions under which
spatial co-location of firms and people enable higher production and growth through
positive spillover effects (Glaeser, Kallal, Scheinkman, & Shleifer, 1992). The forces of
accessibility and agglomeration encourage the centering of jobs.
There are three primary types of agglomeration economies (Glaeser et al., 1992).
First, Marshall–Arrow–Romer agglomeration externalities describe spillover effects
which result from scale economies through which shared inputs in specific industries
are internalized by vertical production chains, frequently leading to specialization
(Henderson, Kuncoro, & Turner, 1995). Second, Porter (1990) contends that returns
to scale are achieved through competition between firms induced by the agglomeration
of firms and people (Porter, 1990). Third, Jacobian agglomeration externalities focus
less on returns to scale from capital inputs and more on knowledge transfers and
returns to scale from large and diverse labor pools. Jacobs (2000) posits that a diversi-
fied economy is both a necessary precondition for, and an outcome of, agglomeration
URBAN GEOGRAPHY 5

which generates value through knowledge transfer (Glaeser et al., 1992). More recently,
the role of spatially fixed opportunities for consumption has been highlighted as a
source of agglomeration benefit which attracts labor, leading to Jacobian agglomeration
externalities (Carlino, 2005).
These three general theories of agglomeration differ, but each posits that spatial
accessibility and proximity are critical and that they can be realized at multiple scales
(Henderson et al., 1995), even in the Canadian context (Partridge, Olfert, & Alasia,
2007)—leading to centered (monocentric or polycentric) spatial structures.

Governance
Second, governance and institutions play key roles in shaping growth (Storper, 2010),
Downloaded by [Library Services City University London] at 20:50 01 July 2016

primarily through the value of public services which are reflected in resident and firm
location choices (Boarnet, 1994; Tiebout, 1956). Residents (Tiebout, 1956), and by
extension firms (Boarnet, 1994), seek municipalities which deliver preferred services
at a competitive cost which are regulated and priced through zoning (Hamilton, 1975).
Thus, prices are reflected in land values and firms and residents self-sort according to
their own willingness to pay.
Municipal competition can lead to the relative decline in the role of the accessible
city center in North American cities. Many governmental decisions feature into the
efficiency of public service provision (Bel & Fageda, 2006), so a market in municipal
services does not guarantee efficient public services. But whether job subcentering or
dispersion occurs depends both on the centripetal forces of local proximity and
accessibility (Boarnet, 1994) and the role of land use regulation (Levine, 2006).
Agglomeration benefits often accrue at local, spatially constrained scales (Alcacer &
Chung, 2013; Combes, Duranton, & Gobillon, 2011; Henderson et al., 1995). As a
result, when localized agglomeration benefits can be produced in both inner and outer
municipalities, one would expect jobs to cluster in both the urban core and suburban
subcenters.
But municipal competition and governance can generate dispersed land use patterns
even when localized agglomeration benefits could materialize outside the urban core
(Gordon, Richardson, & Yu, 1998), primarily due to restrictive zoning (Levine, 2006).
Agglomeration benefits may accrue at the regional rather than local scale, perhaps
through labor pool sharing (Gordon et al., 1998). In regions with a large market in
municipalities, one would expect public service levels to be preserved through zoning
policies (Hamilton, 1975). Restrictive development policies would then constrain local
agglomeration benefits and may encourage sprawl (Levine, 2006). In fact, the regiona-
lization movement has been justified largely on the basis of slowing the sprawl-inducing
effects of local municipalities’ restrictive zoning policies (Orfield, 2008).
But both regionally scaled and local municipal governance occur at a different scales
from locally produced urban agglomeration benefits, which lead to regional economic
spillover benefits (Henderson et al., 1995; McMillen & Smith, 2003). Thus, although
regionally scaled growth management legislation (e.g. Ontario’s Places to Grow Act or
Metro Vancouver’s Regional Growth Strategy) may shape where development cannot
occur, it has less leverage in producing localized agglomeration. With strong localized
agglomeration externalities, economic activity may not cluster due to building height
restrictions, land use controls, prohibitive parking standards, or zoning regulations. As
6 M. N. SWEET ET AL.

such, the extent and type of clustering depends both on regulatory conditions and on
the market benefits of clustering.

Globalization
Globalization is an external force with varied and context-specific consequences for
urban spatial structure. Firms and globally relevant economic activities locate in specific
regions to drive global production processes; globalization both influences city growth
and vice versa (Storper, 2011; Taylor, 2004). Hierarchies of cities support globalization,
whereby world or global cities (such as New York, London, and Tokyo) lead economic
trajectories in secondary and tertiary city-regions (Friedmann, 1986; Sassen, 2001) and
establish global production processes (Smith, 1997). Infrastructure and services, such as
Downloaded by [Library Services City University London] at 20:50 01 July 2016

airports and freeways, both shape growth and are allocated on the basis of growth
trends (Cidell, 2013; Duranton & Turner, 2012). As such, world and global cities both
support global information flows, supply chains, and production and establish built
form norms for other city-regions (Smith, 1997).
Globally relevant economic activities have diverse needs which may spatially man-
ifest themselves in different ways, as illustrated by the field of knowledge-intensive
innovation. Some knowledge-intensive innovation depends very weakly on localized
accessibility (Shearmur, 2012), while in other cases, polycentric city regions support
specialized knowledge production (Romero et al., 2014; Shearmur, 2012). Globalization
has strengthened the discourse of policy intervention to improve city competitiveness
(Banister & Hall, 2004). Thus, governments may adjust spatial planning processes to
meet global economic goals, whether or not a given spatial structure is actually “better”
(Decoville & Klein, 2014). For example, many policymakers view polycentricity as
necessary for global economic competitiveness (Egermann, 2009; Maier, 2009; Sýkora,
Mulíček, & Maier, 2009; Meijers, Hoekstra, & Aguado, 2008; Gordon & Richardson,
1996).
Some have documented the links between global production and job dispersal,
including Shearmur (2012), highlighting that for some industries, production relies
on individual talent and global knowledge sharing with little need for proximity
(Storper & Venables, 2004). Others have highlighted that production processes have
induced developers and city planners to replicate suburban design typologies in
multiple contexts (Filion et al., 2004). Thus, whether globalization leads to more
dispersal or polycentricity depends on the strength of synergies between global and
local growth forces (Hoyler, Kloosterman, & Sokol, 2008; Romero et al., 2014;
Sassen, 2001).
Impacts of globalization on city structure are highly complex, but there are two
reasons globalization may be more likely to lead to polycentric cities. First, observed
trends for world or global cities to become polycentric could establish norms for smaller
city-regions. Both theory (Fujita et al., 1999; Fujita & Ogawa, 1982) and evidence
(McMillen & Smith, 2003) indicate that cities become more polycentric as they become
larger and more congested (Arribas-Bel & Sanz-Gracia, 2014; Egermann, 2009; Hoyler
et al., 2008; Lopez, 2005; Maier, 2009). Second, if policymakers continue to encourage
polycentricity on the basis of economic competitiveness (regardless of whether such
logic is well founded or not), one might expect polycentricity both as a result of land
market processes and policy action. Accordingly, polycentricity may become stronger in
URBAN GEOGRAPHY 7

regions with strong global economic ties and with policymaking authority to encourage
polycentricity.

Spatial structure in Canadian cities and study expectations


Testing existing models of urban spatial structure has been limited in the Canadian
context because studies more often focus on the distribution of population (Filion,
Bunting, Pavlic, & Langlois, 2010) rather than of jobs due to data availability.
Nevertheless, Canadian research on spatial structure has led to three fundamental
findings: (1) central areas remain strong, (2) forces of dispersion parallel to United
States regions, and (3) unique contexts preclude generalization about Canadian
Downloaded by [Library Services City University London] at 20:50 01 July 2016

cities. Filion and coauthors suggest that Canadian city-regions follow United States
trends (Filion et al., 2004) and are becoming more dispersed (Bunting, Filion, &
Priston, 2002), while other regions’ cores (Vancouver and perhaps Toronto and
Montreal) are strengthening (Filion et al., 2010). While CBD shares of regional
employment have declined, their absolute importance remains (Coffey & Shearmur,
2002; Simmons & Jones, 2003). Bourne (1989) finds evidence of complex changes in
urban spatial structure across Canadian cities, while Shearmur and Coffey (2002) and
Shearmur, Coffey, Dubé, and Barbonne (2007) similarly find that no single model of
spatial structure dominates the Canadian context.
Methods employed to study Canadian cities have varied greatly, so it is challenging
to interpret whether the unique conditions of Canadian cities are substantive or
functions of alternative methodologies (Filion et al., 2004). Differences between
Canadian and the United States’s metropolitan growth stem from historical patterns
in industrialization and growth, varying roles of federal planning interventions, and
lower incomes in Canada (Filion et al., 2004). Shearmur et al. (2007) argue that
Canadian urban spatial structure exhibits unique attributes. The authors employ enu-
meration areas as units of analysis (which are smaller zones than those conventionally
used) and assert that the decision leads to much noise which might better be addressed
by looking at larger zones, such as census tracts. To better understand spatial structure
in Canadian cities, comparable methodologies need to be applied for generalization—a
gap to which this research contributes.
Two chief expectations may be outlined based on urban economic theory and
previous studies of Canadian regions. First, although Canadian regions exhibit unique
elements, one might expect larger regions to be less monocentric than smaller regions.
Job centers are expected to remain important components of Canadian urban economic
fabric, see Coffey and Shearmur (2002) or Simmons and Jones (2003). But the unique
dominance of the CBD and monocentric model is expected to diminish as regions grow
and the costs in the urban core (land, congestion, and regulation) may grow more
rapidly relative to CBD agglomeration benefits.
Second, one might expect large regions to become more polycentric, especially when
there are many potentially competing municipalities. But whereas more potentially
competing municipalities appear to be linked with polycentricity, strong regional
growth management policies are expected to have a more ambiguous impact. Growth
management controls focus on regional scales and may not necessarily lead to localized
subcenters through which agglomeration economies are realized.
8 M. N. SWEET ET AL.

Methods for measuring spatial structure


Identifying the degree to which jobs have been centered in the CBD, spatially dispersed,
or clustered in job subcenters depends on the issue of locating job centers and
subcenters. Four replicable methods for delineating job subcenters have been most
common, of which this study employs two. First, stemming from the work of Giuliano
and Small (1991) and others (Cervero & Wu, 1997; McMillen, 2003), subcenters can be
defined as contiguous zones exceeding particular job density thresholds and if contig-
uous, would need to exceed a given threshold of absolute job mass. Different thresholds
have been used, but fundamental findings about the strength of regional polycentricity,
monocentricity, or dispersion appear to be stable regardless of differences in chosen
thresholds (Giuliano et al., 2007). Threshold methods have been modified in subsequent
Downloaded by [Library Services City University London] at 20:50 01 July 2016

models to allow more flexibility in density and absolute employment mass (Shin, 2005).
Second, job subcenters have been identified by comparing expected job density
estimates with observed job densities in zones. These metrics are relative, similar to
those estimated using spatial statistics, but are determined relative to a specific norma-
tive expectation that the city-region functions as a monocentric city. Song (1992)
empirically validates these methods by estimating polycentric job density functions in
which the zone-level job density is estimated as a function of distance to the CBD and a
host of candidate job subcenters. Likewise, McMillen and Smith (2003) propose a
nonparametric method for identifying candidate job subcenters using the residuals of
a nonparametric density function which are verified using a parametric polycentric job
density function. Similarly, Redfearn (2007) specifies a model of subcentering based on
more flexible and localized expectations of job densities.
Third, job subcenters have been identified by some on the basis of commuting flows
and travel behavior. Gordon and Richardson (1996) are among the earliest to employ
this method and their results suggest that Los Angeles is not only a polycentric
metropolis but is increasingly a dispersed metropolis with no clear centripetal forces.
Likewise, Casello and Smith (2006) use commuting data to estimate the locations of job
subcenters in the greater Philadelphia region. More recently, Roca et al. (2009) use this
method to study Barcelona, arguing that because it better reflects the interactions
between spatially fixed activity locations and the functional structure of a metropolis
(and not only the locations of jobs), this method is preferred.
Fourth, spatial statistics have been employed by others to identify the spatial
distributions of job subcenters (Lopez, 2005) and represent an additional delineation
method (Roca et al., 2009). In contrast to the previous metrics, the normative expecta-
tion with which spatial statistics can inform the delineation of job centers focuses on
identifying spatially contiguous clusters of jobs which vary from an assumed (usually
random) distribution. Job subcenters are characterized by critical spatial masses of jobs
and activities which influence broader economic function. So, while spatial statistics are
well suited to identify the geographies of employment, the normative expectation is not
based on economic or functional theories’ urban function. For example, Baumont,
Ertur, and Le Gallo (2004) and Guillain, Le Gallo, and Boiteaux-Orain (2006) use
spatial analysis techniques, including the Moran I index, to respectively study Dijon,
France, and Ile-de-France.
URBAN GEOGRAPHY 9

While ad-hoc methods continue to be employed in planning exercises, McMillen and


Lester (2003) argue that methods using absolute thresholds or methods using relative
thresholds in comparison with the monocentric expectation are, respectively, the best
means of assessing changes over time or comparing regions in a cross-sectional frame-
work. In fact, while the early work of Giuliano and its Small (1991) arguably employed
arbitrary cutoffs used to identify job subcenters, the method and its cutoffs have
remained resilient and lead to the same substantive conclusions in a cross-sectional
context. To demonstrate their respective findings in the cross-sectional case, this study
adopts both metric types to explore spatial structure in major Canadian regions.

Research design
Downloaded by [Library Services City University London] at 20:50 01 July 2016

This study employs thresholds to measure job subcenters using two established
means of assessing job subcenters: absolute thresholds when applying the methods
of Giuliano and Small (1991) and relative thresholds based on residuals identified
using a monocentric log-linear job density model. Models are estimated for eight
Canadian city-regions using 2011 job data from InfoGroup aggregated at the census
tract level. Those city-regions correspond to the CMA boundaries of Toronto,
Hamilton, Montreal, Winnipeg, Vancouver, Calgary, Ottawa, and Edmonton, as
defined by Statistics Canada.
By estimating the extent of monocentricity, polycentricity, and dispersion in the
eight major Canadian city-regions, the primary types of spatial structure in major
Canadian cities can be explored and interpreted relative to the forces of accessibility,
municipal governance, and globalization which shape spatial structure. If cities con-
tinue to be largely monocentric, this would indicate that accessibility and central
proximity-related forces continue to be relatively the strongest. Alternatively, if regions
are either highly polycentric or dispersed, the extent to which either municipal compe-
tition or globalization plays important roles must be assessed by comparing the location
of job subcenters with municipal borders and with production nodes of high global
accessibility (such as airports). But if each of these factors appears to be relatively
weaker, this would be consistent with either polycentricity or dispersion stemming from
the relative weakness of central proximity-related benefits and the relative strength
(polycentricity) or weakness (dispersion) of proximity-related benefits at outward lying
nodes. In sum, one would expect evidence of a dispersed spatial structure if proximity-
related agglomeration benefits are weak, one would expect a strong role for subcenter-
ing in large regions with strong proximity-related agglomeration benefits (perhaps
Toronto and Montreal), and one would expect evidence of monocentricity in smaller
regions with a stronger core (perhaps Calgary).

Data
This study focuses on 2011 job data collected from InfoCanada, a division of
InfoGroup. Data are compared with Canadian Business Pattern (CBP) data and indicate
similar patterns. But while CBP data provide much more generalized estimates of firm
counts grouped based on sizes (which are categorical) at generalized spatial levels,
InfoGroup data include not only the count, type, and locations of firms but also
10 M. N. SWEET ET AL.

point estimates of the number of employees physically situated at particular office


locations. The data are updated quarterly by InfoCanada staff telephone calls to firm
representatives to verify change in firm size, change in location, or other changes in
firm characteristics on a rotating basis. Based on conversations with their staffs,
InfoCanada monitors public records documenting when firms are established or dis-
solved. These data are collected to represent a rolling census of firms and these data are
sold to other private entities for marketing, business analytics, and location decisions.
Basic descriptive statistics of the datasets are provided in Table 1, showing that the
Toronto, Montreal, and Vancouver city-regions are the largest agglomerations of people
and jobs among the eight major Canadian regions (and within Canada more broadly).
Regions have between 0.36 and 0.50 jobs per resident and broadly speaking, the job-rich
Downloaded by [Library Services City University London] at 20:50 01 July 2016

regions correspond to high-growth areas of the western provinces.


Upon having linked firm-level job data with census tracts (henceforth referred to
simply as zones), the basic descriptive statistics for job density estimates are shown in
Table 2.1 Aggregating data to the census tract level generalizes the geographic patterns
of jobs, but is preferred over alternate methods for the purposes of comparison,
generalization, and reasonableness. Census tracts are defined by Statistics Canada on
the basis of the geographic distributions of residents and not of jobs and are reasonable
to neither overly generalize nor to capture spatial differences in too fine a granularity.
Job density maps of the three largest regions, Toronto, Montreal, and Vancouver (see
Figure 1), illustrate the spatial distribution of zones to which the jobs are aggregated.
Mean job density estimates are subject to the CMA boundaries defined by Statistics
Canada. If CMA boundaries were expanded, average densities would decline, so the
descriptive statistics should be viewed cautiously.2

Table 1. Jobs, firms, and population in eight study regions.


Jobs (data source: Firms (data source: Population (data source: Jobs per Jobs per
Region InfoGroup, 2011) InfoGroup, 2011) Statistics Canada, 2011) firm resident
Toronto 2,506,523 218,520 5,769,759 11.47 0.43
Montreal 1,670,507 143,718 3,885,709 11.62 0.43
Vancouver 959,024 93,563 2,373,045 10.25 0.40
Calgary 636,808 54,286 1,264,460 11.73 0.50
Edmonton 577,269 48,046 1,206,040 12.02 0.48
Ottawa 469,731 42,534 1,270,232 11.04 0.37
Winnipeg 354,576 26,055 746,059 13.61 0.48
Hamilton 266,674 27,008 742,498 9.87 0.36
Total 7,441,112 653,730 17,257,802 11.38 0.43

Table 2. Job data basic descriptive statistics at 2011 census tract level.
Job density at 2011 Census tract level (jobs per Acre)
Region Mean Median Standard deviation Maximum Number of tracts
Toronto 8.47 2.42 38.75 904.20 1,088
Montreal 9.27 3.60 22.55 356.76 921
Vancouver 7.42 1.99 22.10 296.24 457
Calgary 4.12 1.09 16.95 244.11 248
Edmonton 4.00 1.25 17.01 264.01 252
Ottawa 3.86 1.49 8.88 91.22 271
Winnipeg 5.62 2.17 15.05 157.96 173
Hamilton 3.91 2.08 6.81 49.17 188
URBAN GEOGRAPHY 11
Downloaded by [Library Services City University London] at 20:50 01 July 2016

Figure 1. Job density at census tract level for Toronto, Montreal, and Vancouver census metropolitan
areas.
12 M. N. SWEET ET AL.

Absolute threshold method


To estimate the extent and intensity of subcentering in each of the eight Canadian
regions, two methods are employed, an absolute threshold method and a relative
threshold method. To identify employment centers, the absolute threshold method
uses two thresholds: one based on job density and one based on the absolute number
of contiguous jobs, as follows (Casello & Smith, 2006):
Equation 1. Employment cluster job density threshold
Ez

Az
Downloaded by [Library Services City University London] at 20:50 01 July 2016

Equation 2. Employment cluster total jobs threshold


X
Ez  
z

where Ez represents employment within zone P z, Az is the area of zone z, and φ


represents the employment density threshold; z Ez is the sum of all employment
for adjacent zone grouping z and  is a minimum total employment threshold. Casello
and Smith (2006), Giuliano and Small (1991), and Giuliano et al. (2007) employ a
threshold of 10 jobs per acre and 10,000 total employees in a contiguous cluster.
Giuliano et al. (2007) further argue that the share of total regional employment in
subcenters is relatively insensitive to different density and absolute employment thresh-
olds. As a result, the 10 employees per acre and 10,000 absolute employment thresholds
are retained to maintain as much comparability with previous research as possible.
The strength of the monocentric, polycentric, and dispersed models of spatial
structure can be explored using the absolute threshold method to estimate job sub-
centers. The relative strength of central agglomeration and the monocentric model is
estimated as the share of regional jobs which are within the CBD. The strength of job
subcentering is estimated as the share of regional jobs which are within the sum of all
the job subcenters, while the relative strength of the dispersed spatial structure rests on
the share of jobs which are neither in the CBD nor in one of the subcenters.

Relative threshold method


The relative threshold method applies the same absolute employment threshold as
above, according to which total contiguous employment must exceed 10,000 employees
but in addition identifies potential job clusters as those with significantly higher
densities than expected according to the monocentric job density model. Using residual
error estimates for each metropolitan area’s monocentric job density model, we focus
on zones with significantly higher employment density at the p = 0.10 confidence level
and at the p = 0.05 confidence level using a one-tailed test. We explore clusters of
significant positive residuals, such that an employment cluster is composed of contig-
uous, significantly positive residuals, with a total of 10,000 employees or more. Thus, we
define relative employment clusters using one-tailed tests based on both the p = 0.10
URBAN GEOGRAPHY 13

and p = 0.05 confidence level thresholds. Naturally, the 0.10 p value threshold generates
more and larger clusters.
To estimate job subcenters using the relative threshold method, a monocentric
job density model is first estimated for each of the city-regions. To estimate the job
density model, the very densest zone identified as being part of the CBD (as defined
using method 1) is assigned as the center of the region. Each identified CBD zone
appeared to reasonably mirror local knowledge by the study authors and others.
Using the job density model, the degree of central agglomeration (the intercept
estimate) and the degree of employment containment (the steepness of the slope
estimate) are estimated. Estimates of employment density from the monocentric
model function as a point of comparison to identify relative employment activity
Downloaded by [Library Services City University London] at 20:50 01 July 2016

centers—in essence, those places where employment is significantly denser than one
would expect.
The job density curve is estimated as follows:
Equation 3. Estimating a job density curve
 
Emz
ln ¼ B0m þ B1m Dmz
Amz

where ln (Emz/Amz) represents the natural-logged employment (Emz) density per square
kilometer (Amz) for census tract (zone) z in city-region m; B0m is an intercept parameter
to be estimated for each city-region; B1m is a distinct slope parameter to be estimated
separately for each city-region; and Dmz represents the distance between zone z and the
CBD in city-region m. The intercept coefficient (B0m) corresponds to the estimated job
density at the CBD, while the job density gradient or slope estimate (B1m) captures the
rate at which expected job density declines as one is further from the CBD (in percent
density decrease per unit distance).
Results from the relative threshold methods can be interpreted by identifying the
number and locations of job subcenters and by interpreting the monocentric job
density model coefficients, which together illustrate the relative distribution of jobs.

Results
Job subcenters are estimated using both absolute and relative thresholds. Using
these metrics, the extent to which each region’s jobs are contained within the
CBD and subcenters are estimated—providing evidence of the relative roles of job
centralization, subcentering, and dispersal. By estimating the monocentric job den-
sity model, results provide insight into the relative strength of the monocentric
model in explaining spatial structure and a sense of how strong the CBD is in
shaping job density. Upon introducing the results, they are further interpreted in
light of the two expectations outlined in the literature review: one would expect (1)
a reduced role of monocentricity in larger city-regions and (2) an increased role of
polycentricity (not dispersion) in large city-regions, particularly in those with many
municipalities.
14 M. N. SWEET ET AL.

Absolute job subcenters


First, the number, location, and intensity of centers and subcenters are estimated
using the 2011 datasets for each of the eight regions by applying the absolute job
subcenter method. Results suggest that there are between 1 and 15 centers, including
the CBD, in each region. Results are shown for eight regions in Table 3, but subcenter
maps are only shown for the three largest regions: Toronto, Montreal, and Vancouver
(see Figures 2–4).

Table 3. Job subcenter results using absolute thresholds*.


Downloaded by [Library Services City University London] at 20:50 01 July 2016

CBD
No. of centers share Subcentre Dispersal CBD area Subcentre Population (Statistics
Region (including CBD) (%) share (%) (%) (acres) area (acres) Canada, 2011)
Toronto 15 19.2 32.6 48.3 6,956 55,508 5,769,759
Montreal 6 28.8 9.7 61.5 13,190 11,287 3,885,709
Vancouver 5 26.1 11.3 62.6 5,487 7,009 2,373,045
Calgary 4 33.9 5.9 60.3 6,512 2,171 1,264,460
Edmonton 4 21.0 18.3 60.7 394 196 1,206,040
Ottawa 1 13.8 0.0 86.2 1,351 0 1,270,232
Winnipeg 2 29.9 3.5 66.7 2,515 849 746,059
Hamilton 2 10.3 8.3 81.4 887 1,988 742,498
*Models are estimated using jobs per acre.
CBD: Central business district.

Figure 2. Toronto CBD and subcenters estimated using absolute and relative thresholds (0.10 level
of significance).
URBAN GEOGRAPHY 15
Downloaded by [Library Services City University London] at 20:50 01 July 2016

Figure 3. Montreal CBD and subcenters estimated using absolute and relative thresholds (0.10 level
of significance).

Figure 4. Vancouver CBD and subcenters estimated using absolute and relative thresholds (0.10
level of significance).
16 M. N. SWEET ET AL.

CBDs contain between 10.3% and 33.9% of regional jobs (see Table 3)—implying
significant variation in urban spatial structure. Calgary, Winnipeg, and Montreal CBDs
have the largest shares of regional jobs, while Hamilton’s CBD has the lowest—implying
weak CBD-related agglomeration economies. The Toronto CBD (19.2% of regional
jobs) has the lowest share of regional jobs among the three largest regions, despite it
having the largest absolute number of jobs located in the CBD.
The number of subcenters in each of the eight regions ranges from 0 to 14, while the
subcenter share of regional jobs ranges from 0% to 32.6%. Three regions have one or
fewer subcenters, four have between three and five subcenters (excluding the CBD), and
Toronto (14 subcenters) is most subcentered. Subcenters make up less than 12% of
regional jobs in most regions, while Toronto (32.6%) and Edmonton (18.3%) subcen-
Downloaded by [Library Services City University London] at 20:50 01 July 2016

ters contain the highest shares of regional jobs.


Approximately one-third of regional jobs are located either in the CBD or the job
subcenters of five of the eight regions, while the remainder of jobs are dispersed
throughout each of the city-regions. At the most extreme, jobs are very centered in
Toronto: just over half of Toronto jobs are in the CBD or 1 of the 14 job subcenters. But
in Hamilton and Ottawa, less than 20% of jobs are located in a center of any kind,
indicating a relatively weaker role of job centers in these two Ontario regions.
Maps of the subcenter locations (see Figures 2–4) are shown for Toronto, Montreal,
and Vancouver and results for these largest regions are discussed in turn. Findings are
most consistent with two of the three causes of subcentering: the globalization process and
services available in alternate municipalities. If subcenters are expected to play a stronger
role in global cities with strong knowledge and innovation industries (Romero et al., 2014),
one might expect the forces of subcentering to be strongest in these three cities relative to
the others, and among these three cities to be strongest in Toronto.3 Consistently with this
expectation, Montreal, Vancouver, and particularly Toronto appear to be both the most
globally important Canadian city-regions and the most subcentered—especially in close
proximity to international airports (Toronto, Montreal, and Vancouver), major freeways
(Toronto and Montreal), transit (Vancouver, but also Toronto and Montreal), and in
municipalities outside of the core (or formerly outside of the core).
First, at least one job subcenter is located near the international airports of each of
these major Canadian city-regions. This implies that not only is the modern geography
of these Canadian regions shaped by conventional downtown agglomeration econo-
mies, but that airport proximity and global access is a significant and valued service
which shapes spatial structure, as further described by Cidell (2013).
Second, freeway access appears to play a key role in job subcenter locations within the
Toronto, Montreal, and Vancouver (although to a lesser degree). In fact, with one
exception (Richmond Hill), most of the job subcenters in Toronto are located along the
401 freeway’s corridor (the east–west freeway located immediately north of Toronto’s
core), which has direct service to the airport, indicating that access to globalized networks
via airports appears to play a critical role in subcenter formation and location—at least for
the largest Canadian city-regions. Like with Toronto, freeways appear to play an important
role in the locations of all five Montreal subcenters. In Vancouver, only the subcenters of
Burnaby and North Burnaby are located within 2 km or less of major freeways, while
Richmond (near the Vancouver International Airport) and North Vancouver (like the
CBD) are relatively poorly served by freeways. In fact, Vancouver’s freeway network is
URBAN GEOGRAPHY 17

relatively sparse by North American standards, leaving suburban areas with fewer freeway
services (Punter, 2003) from which to develop car-oriented job subcenters.
Third, transit access appears to play different roles in each of the three largest city-
regions. Only three Toronto subcenters are located within 1 km of a subway station
(North York, the Downtown Core, and the Danforth). Nevertheless, while the
Downtown Core and Danforth depend very clearly on subway access, the role of
subway access in North York is more ambiguous, given the auto-oriented built form.
Three additional stations are within 2 km of subway stations. In Montreal, the down-
town center is served by the subway, but only two subcenters, Laval/Chomedey and
Longueuil, are moderately close to subway stations. While Laval/Chomedey opened too
recently (2007) to have significantly impacted the job subcenter, the Longueuil subway
Downloaded by [Library Services City University London] at 20:50 01 July 2016

most directly serves the University of Sherbrooke. The link between transit and job
subcenters in Vancouver appears to be very different. Much of the transit in Vancouver
has been constructed in the last 20 years, so it is no coincidence that many of these
transit lines connect four of the five major job subcenters. Thus, it appears that rather
than transit shaping the presence or location of job subcenters, which are relatively
static even over 20-year timelines, the geographic locations of job subcenters may be
shaping the location of transit in Vancouver.
Fourth, municipal geography appears to play different roles in each of the three
major regions. If subcentering occurs in alternate municipalities, this suggests that
better service value for cost may play a role. But if subcenters are within the core
municipality, this would suggest that urbanization externalities may drive subcentering.
Subcenters are relatively static (Glaeser & Kohlhase, 2003; Redfearn, 2009), so subcenter
locations must be interpreted in light of current and historical municipal boundaries.
All Vancouver subcenters are outside the city of Vancouver. Recent amalgamations
(provincially mandated mergers of central and outer municipalities) in Montreal
(amalgamated in 2002) and Toronto (amalgamated in 1998) complicate the interpreta-
tion of municipal governance in shaping subcenter locations. All Montreal subcenters
lie outside of the city’s 2002 municipal boundaries, but two of the four subcenters are
now in the city since amalgamation. Similarly, all Toronto subcenters lie outside of the
1998 municipal boundaries, but all but three (Richmond Hill, East Oakville, and
Pearson Airport) are now located within the city of Toronto. These findings are
consistent with the hypothesis that the relative costs and value of municipal services
spurred subcentering. But the landscape has changed dramatically in Montreal and
Toronto, whereby subcenters which previously had been in other municipalities are
now located in the primary city.

Relative job subcenters


Next, the monocentric job density model is estimated for each of the eight study regions
and the degree of job subcentering is estimated using relative thresholds. While absolute
threshold methods do not have any ex-ante expectations of job density, relative thresh-
old methods identify candidate subcenters relative to the monocentric job density
expectation. Differences in findings between the two methods hinge on interpreting
clusters based on either absolute economic mass or higher density relative to an
expectation.
18 M. N. SWEET ET AL.

Results of the monocentric job models and the relative subcenter counts (using
both the 0.10 and the 0.05 one-tailed level of significance) are shown in Table 4.
Most of the regions have more than one center, but among all regions, the
exponential distance decay function is better at explaining spatial structure in
some regions than in others. The adjusted R-squared values (ranging between 0.17
and 0.76) for each of the monocentric job density models indicates the share of
variation in job density at the census tract level which is explained by the mono-
centric distance–decay model. These values should be interpreted not as the relative
importance of the CBD, but rather the extent to which proximity to the CBD
explains the variation in gross job density within the regional land market. While
the monocentric model explains most variation in some regions, notably Winnipeg
Downloaded by [Library Services City University London] at 20:50 01 July 2016

(0.76) and Ottawa (0.59), it is much less powerful in explaining spatial structure in
Toronto, Vancouver, and Hamilton.
Results from the monocentric model should be interpreted carefully—particularly
compared to results on subcenters. For example, while the monocentric model explains
most of Winnipeg regional spatial structure, its localized job centers feature less
prominently as a share of the regional job market. Hamilton similarly does not have
strong subcenters but, in contrast to Winnipeg, is poorly characterized by the mono-
centric job-density model—implying a dispersed urban form. Likewise, the mono-
centric model poorly characterizes spatial structure in Vancouver (adjusted R-squared
of 0.17), but in contrast with Hamilton, Vancouver exhibits strong subcenters. This
implies stronger localized processes in Vancouver through which land market inter-
ventions may yield localized rather than regional benefits.
The intercepts and slopes of the models estimate the expected job density at the
regions’ centers and the rate of density declines for each kilometer of distance from the
center. Model results are in a log-linear functional form, so when exponentiated, the
intercept indicates the expected job density at the CBD (jobs per acre). The slope
indicates the percent expected decline in job density for each kilometer from the CBD.
A higher intercept suggests the strength of central agglomeration, but the densest CBDs
do not always have the highest intercept. Instead, the intercept is a function of the full
model fit and reflects the job densities of CBD and other inner areas. As such, the extent
of central agglomeration should be interpreted through the joint results of the intercept,
slope, and R-squared values.

Table 4. Monocentric job density model results and subcenter count estimates*.
No. of centers (with CBD)
0.10 Sig. 0.05 Sig. Absolute Adjusted R
Region level level threshold Intercept Slope squared Observations RSE
Toronto 23 9 15 2.26 −0.06 0.34 1,088 1.2
Montreal 11 7 6 2.57 −0.10 0.48 921 1.16
Vancouver 4 3 5 1.67 −0.26 0.17 457 1.5
Calgary 6 3 4 1.89 −0.17 0.44 248 1.3
Edmonton 3 1 4 1.60 −0.06 0.51 252 1.40
Ottawa 2 2 1 2.03 −0.12 0.59 271 1.32
Winnipeg 2 1 2 2.47 −0.04 0.76 173 1.65
Hamilton 2 2 2 1.50 −0.12 0.20 188 1.33
*Models are estimated using jobs per acre.
CBD: Central business district; RSE: Residual standard error.
URBAN GEOGRAPHY 19

Vancouver and Calgary have the steepest job density gradients. According to the
Calgary model, one would expect the job density at the CBD to be 1,353 jobs per square
kilometer (1,353 = 247.1 acres per square kilometer × e1.89). Then for each kilometer
from the CBD, the job density would be expected to decline by 16.6%.
Estimates of subcenter counts are broadly consistent with estimates using the
absolute job density thresholds. Toronto is significantly more subcentered than the
other regions and other large city-regions, including Montreal, Calgary, and Vancouver,
are moderately subcentered (four 11 job centers based on the 0.10 level). Using different
p values, one would expect different subcenter counts—frequently on the order of a
twofold difference.
For comparison with the absolute threshold method, maps of job subcenters using
Downloaded by [Library Services City University London] at 20:50 01 July 2016

the relative threshold method are displayed for Toronto, Montreal, and Vancouver,
as shown in Figures 2–4. Results are similar between the two methods. Subcenters
are located near each of the three city-regions’ major airports and closely follow the
freeway network in Toronto. Job subcenters are in suburban municipalities accord-
ing to the current (Vancouver) pre-2002 (Montreal) and pre-1998 (Toronto) city
boundaries. This is consistent with the explanation of job subcentering being partly
a function of the relative value for price and cheaper land of suburban
municipalities.
When focusing on Toronto, there are 15 subcenters according to the absolute
threshold method, while there are respectively 23 or 9 subcenters depending on the
significance level chosen in the relative models. Subcenter locations are similar using
the two methods (see Figure 2), but the Golden Mile and Scarborough West are
identified as subcenters using the absolute but not using the relative threshold methods.
Using the relative thresholds with a 0.10 significance level, more subcenters are identi-
fied and these are located further from the downtown core. The locations of the
additional subcenters are outside of the city of Toronto boundaries and they follow
the freeway network very closely, with some exceptions.
Like in Toronto, the Montreal results using the relative thresholds are similar:
there are five centers (including the CBD) according to the absolute thresholds,
while there are seven (0.05 level) or eleven (0.10 level) when applying relative
thresholds. All five centers are retained in the relative models, but Longueil splits
in two: Longueuil North and Longueuil South. Additional subcenters in the relative
models include Montreal North, the Montreal Olympic Stadium and surroundings,
and Station Cote-Vertu. With the exception of the Montreal Olympic Stadium
subcenter, each of the additional subcenters using the relative method were outside
of the original city of Montreal boundaries before 2002 and are in close proximity to
freeways.
Finally, when focusing on subcentering in Vancouver, results suggest fewer centers
using the relative thresholds (four centers using the 0.10 level and three using the 0.05
level of confidence) than using the absolute thresholds (five centers). With the 0.10 level
of confidence, North Burnaby is not identified as a subcenter, while the North
Vancouver subcenter shifts further west.
20 M. N. SWEET ET AL.

Interpretation
Expectations from the review of previous work had been that (1) the monocentric
model would less effectively describe larger Canadian regions and (2) that the largest
Canadian regions would become more polycentric, particularly when there is a larger
market in potential regional municipalities. There are insufficient city-regions in this
dataset to build predictive models of spatial structure. However, results from this study
of eight Canadian city-regions provide context-specific information on the potential
predictors of spatial structure. Findings on the two core expectations outlined in this
study are discussed in turn.
First, findings on the extent to which the monocentric model better characterizes
smaller regions are mixed. The monocentric job density model does better characterize
Downloaded by [Library Services City University London] at 20:50 01 July 2016

smaller regions—implying that the CBD dominates regional land markets through the
density gradient. But CBDs have similar shares of regional jobs, regardless of city size.
The core’s influence on the regional land market may attenuate in larger regions, but its
role as a localized agglomeration economy remains important. Nevertheless, there
remain differences among both large and small city-regions.
Second, results suggest that larger regions have more subcenters but whether a larger
share of regional jobs are located in job centers rests on unique cases. Across subcenter
metrics, the two smallest regions, Winnipeg and Hamilton (approximately 700,000
residents), typically have one fewer subcenter than Ottawa, Edmonton, and Calgary
(1.2–1.3 million residents). In turn, these three have approximately three fewer sub-
centers than Vancouver and Montreal (2.3–3.9 million residents) and the Toronto
region (5.8 million residents) has more than nine subcenters, depending on metric.
Differences could stem both from market forces and the character of planning initia-
tives in bigger city-regions which frequently lead to larger scaled recommendations
which support subcenter formation.
Region size and the share of regional jobs in centers are linked only in unique cases.
One-third of regional jobs are in either a CBD or subcenter in five of the eight regions.
Significant variations are evident only in the most extreme cases: Toronto (51.7% of
jobs are in centers), Ottawa (13.8% of jobs), and Hamilton (18.6% of jobs). The Toronto
region is Canada’s largest, it houses more than 17% of Canada’s 2011 population and it
is highly subcentered. In contrast, the two smaller and less centered city-regions
comprise of Hamilton, a highly dispersed Toronto-adjacent region which is weakly
monocentric, and Ottawa, a highly monocentric region serving as the nation’s capital.

Conclusion
Previous studies of spatial structure in major Canadian cities have found that central
economic forces remain strong but job dispersion is occurring. Results from this study
indicate a variety of spatial structures and role for job centers in these eight study
regions—consistent with the findings of Shearmur et al. (2007). Urban cores remain
strong, the traditional monocentric model best reflects smaller city-regions, and job
subcenters play important roles in global cities such as Montreal, Vancouver, and
particularly in Toronto. Toronto is highly polycentric and more than half of regional
jobs are in either the CBD or in one of its subcenters. Both Montreal and Vancouver
URBAN GEOGRAPHY 21

have numerous subcenters and strong CBDs, but while the monocentric model better
characterizes job density patterns in Montreal, it is less effective in Vancouver—
respectively, implying regionally and locally scaled land market forces at work. In
contrast, subcenters play a weak role in both Winnipeg and Hamilton, but the mono-
centric model better characterizes Winnipeg. Finally, while polycentricity is most
pronounced in global city-regions, especially Toronto, Edmonton represents an alter-
native type of smaller polycentric city in which there are few subcenters (four), but
subcenters contain a very high share (almost 20%) of regional jobs.
Calls that job sprawl may stifle urban economic advantages may be premature—
especially, if subcenters play an important role in delivering agglomeration benefits
despite their frequent auto-oriented design. Findings are consistent with previous
Downloaded by [Library Services City University London] at 20:50 01 July 2016

studies indicating the importance of urban centers in Canada (Coffey & Shearmur,
2002; Simmons & Jones, 2003), but large city-regions rely not on a single urban cluster,
but a system of clusters which frequently locate in suburban municipalities near free-
ways and subsequently may be connected to public transit. Planners have long acknowl-
edged the critical economic role of urban centers, but this study finds that job centers
have materialized both in central and suburban locations. Understanding the role of
subcenters and whether they lead to agglomeration externalities is critical for shaping
better public policy.
Three common themes emerge from the largest Canadian regions, Toronto,
Montreal, and Vancouver, which contain more than 35% of the Canadian population.
First, the spatial location of job subcenters suggest significant variation in the role of
surface transportation infrastructure. In Toronto, 14 of the 15 job centers (using
absolute threshold methods and including the CBD) are very close to major freeways.
But in Vancouver and Montreal, some subcenters are near transit stations which have
very recently been constructed—some near freeways and some in isolated locations.
The potential importance for freeways has likewise been documented by others
(Giuliano, Redfearn, Agarwal, & He, 2012). However, it is notable that transit station-
proximate subcenters in Montreal (one of six in total) and Vancouver (five of six in
total) represent a development model in which the public transit investments have
followed job center location and formation. Similarly, plans call for Montreal’s Eastern
Center (the Anjou job subcenter) to be connected to the subway system (City of
Montreal, 2014) and the Union Pearson Express recently connected the Pearson
International Airport with Toronto’s CBD. Many of these subcenters are dense but
highly auto-centric, so exploring the role for policy in most effectively leveraging or
allocating transit infrastructure and services is important. For example, while Pearson
International Airport is served by transit in Toronto, most airport-proximate jobs are
outside of the airport.
Second, at least one job subcenter is located near the international airports in
Montreal, Toronto, and Vancouver, supporting assertions by others, such as Cidell
(2013), that global air-based transportation policy shapes regional growth. This is con-
sistent with the expectation that the forces of globalization may lead to localized nodes of
economic activity near global nodes of communication and transportation—most nota-
bly airports. Thus, while both freeway proximity and airport proximity appear to be
linked with subcenter locations, each of these two services likely serve different users
focusing on either global (airports) or regional and local (freeways) access and trips.
22 M. N. SWEET ET AL.

Finally, the role of municipal boundaries is important in shaping the geography of


job subcenters. This is consistent with the expectation that municipal competition could
lead to the suburbanization of jobs (Boarnet, 1994) and when agglomeration benefits
are high, those suburban jobs would locate in subcenters (McMillen & Smith, 2003).
Almost all job subcenters are located outside of the city of Vancouver (current bound-
aries), city of Montreal (2002 boundaries before amalgamation), and city of Toronto
(1998 boundaries before amalgamation) before recent amalgamations for two of the
three regions. As the real estate development process takes time to adjust to changing
policy environments, this suggests that subcenters may have formed outside of the
previous central municipality—perhaps to seek preferred public services and tax poli-
cies and cheaper land. But now that many of the Toronto and Montreal subcenters have
Downloaded by [Library Services City University London] at 20:50 01 July 2016

been amalgamated into the core city limits, the environment appears to be rife for
additional tension between CBD and subcenter interests, as anticipated by Filion (2000)
and Keil and Boudreau (2005).
From a normative perspective, exploring whether polycentricity, monocentricity, or
dispersion should be more desirable hinges on how different spatial structures are linked
with other outcomes (Chowdhury, Scott, & Kanaroglou, 2013; Veneri, 2010; Zolnik, 2012).
This has been a central debate in the field of urban planning—see Gordon and Richardson
(1997) and Richardson and Ewing (1997)—and extends beyond the scope of this paper.
Urban economic theory holds that economic agglomeration externalities may accrue from
particular types of spatial structure, but this needs to be better understood in the context of
regional urbanization benefits, local economic agglomeration benefits, inter-regional
growth, and local urban design. If subcentering continues through the joint forces of the
suburbanization of jobs and globalization in the face of continued high benefits from
proximity-related accessibility, the more important policy question may be how to best
serve and shape localized agglomeration through policymaking. Answering this question
requires an astute understanding of growth dynamics between and within city-regions.
Although planners’ normative enthusiasm for monocentric cities with strong urban cores
reflects well-documented urban agglomeration benefits, such a model of growth may
imply encouraging growth in medium-size Canadian cities in which the monocentric
model remains relatively stronger. In contrast, in global cities such as Toronto, Montreal,
or Vancouver, it appears to be critical to understand the capacity for subcentering to
preserve agglomeration benefits despite their frequent auto-oriented urban design
(Cervero & Wu, 1997). Municipal competition appears to encourage subcentering, so
the policy task of facilitating agglomeration economies will remain rooted in the challenges
of reconciling the perceived needs of urban and suburban residents and firms. Despite the
links between suburban and urban economic interests (Inman, 2003), history suggests that
policymakers frequently ignore one at the expense of the other (Nelson, Sanchez, Wolf, &
Farquhar, 2004).

Notes
1. Using census tract boundaries from the 2011 Statistics Canada census, jobs are aggregated to
the census tract level using a geographic information system. While census tract are
delineated on the basis of tracking population changes by Statistics Canada, tracts provide
useful means of categorizing and describing the geography of jobs as well. In fact, according
URBAN GEOGRAPHY 23

to Shearmur et al. (2007), tracts represent sufficiently large zones with which to track job
patterns to allow meaningful comparisons of differences across space, with much less local
noise. Moreover, tracts are chosen as reasonable units of observation for the purposes of
comparability with studies of spatial structure and subcentering in other contexts.
2. Minimum job densities are not shown, as they are all very close to zero because some rural
or exurban tracts within each of the CMAs contain very few jobs.
3. Based on the concepts and ranking methods discussed in Taylor et al. (2009), Derudder,
Hoylder, Taylor and Witlox (2012), according to the 2012 inventory of the Globalization
and World Cities (GaWC) Research Initiative, Toronto is an Alpha global city, while
Montreal (Beta+) and Vancouver (Beta) are arguably less globalized (GaWC, 2012).
Downloaded by [Library Services City University London] at 20:50 01 July 2016

Acknowledgements
Errors and omissions are solely the responsibilities of the study authors. Many thanks to several
reviewers and the journal editor for their input on this article; it has improved, thanks to their
feedback. The authors would also like to thank Micki Chana, Jamie Vernon, and Raj Padda for
their support in applying InfoCanada data for this project. Sadly, Dr Pavlos Kanaroglou passed
away on May 13, 2016. He was one of the foremost experts on travel behavior modeling and
transportation geography and he dedicated much of his life to bridging the links between
scholarship and practice, for example, by founding and leading the McMaster Institute for
Transportation and Logistics.

Disclosure statement
No potential conflict of interest was reported by the author.

Funding
This research was completed with financial support from the Social Sciences and Humanities
Research Council (SSHRC) in Canada under grant [435-2013-1120], which was managed by
Dr Pavlos Kanaroglou as Principal Investigator.

References
Alcacer, Juan., & Chung, Wilbur. (2013). Location strategies for agglomeration economies.
Strategic Management Journal, 35, 1749–1761.
Alonso, William. (1964). Location and land use: Toward a general theory of land rent. Cambridge,
MA: Harvard University Press.
Anas, Alex, Arnott, Richard, & Small, Kenneth. (1998). Urban spatial structure. Journal of
Economic Literature, 36(3), 1426–1464.
Anas, Alex, & Xu, Rong. (1999). Congestion, land use and job dispersion: A general equilibrium
model. Journal of Urban Economics, 45, 451–473.
Arribas-Bel, Daniel, & Sanz-Gracia, Fernando. (2014). The validity of the monocentric city
model in a polycentric age: US metropolitan areas in 1990, 2000, and 2010. Urban
Geography, 35(7), 980–997.
Banister, David, & Berechman, Joseph. (2000). Transport investment and economic development.
London: UCL Press.
Baum-Snow, Nathaniel. (2007). Did highways cause suburbanization? The Quarterly Journal of
Economics, 122(2), 775–805.
24 M. N. SWEET ET AL.

Baumont, Catherine, Ertur, Cem, & Le Gallo, Julie. (2004). Spatial analysis of employment and
population density: The case of the agglomeration of Dijon 1999. Geographical Analysis, 36(2),
146–176.
Beauregard, Robert. (2007). More than sector theory: Homer Hoyt’s contributions to planning
knowledge. Journal of Planning History, 6(3), 248–271.
Bel, Germa, & Fageda, Xavier. (2006). Between privatization and intermunicipal cooperation:
Small municipalities, scale economies and transaction costs. Urban Public Economics Review,
6, 13–30.
Boarnet, Marlon. G. (1994). An empirical model of intrametropolitan population and employ-
ment growth. Papers in Regional Science, 73(2), 135–152.
Bourne, Larry S. (1989). Are new Urban forms emerging? Empirical tests for Canadian Urban
Areas. The Canadian Geographer, 33(4), 312–328.
Brueckner, J K. (2007). Urban growth boundaries: An effective second-best remedy for unpriced
Downloaded by [Library Services City University London] at 20:50 01 July 2016

traffic congestion? Journal of Housing Economics, 16, 263–273.


Bunting, Trudi E., Filion, Pierre, & Priston, Heath. (2002). Density gradients in Canadian
metropolitan regions, 1971–96: Differential patterns of central area and suburban growth
and change. Urban Studies, 39(13), 2531–2552.
Carlino, Gerald. (2005). The economic role of cities in the 21st century. Business Review, Q3, 9–15.
Casello, Jeff M., & Smith, Tony E. (2006). Transportation activity centers for Urban transporta-
tion analysis. Journal of Urban Planning and Development, 132, 247–257.
Castells, Manuel. (1996). The rise of the network society. Oxford: Blackwell Publishers.
Cervero, Robert. (1986). Suburban gridlock. New Brunswick, NJ: Center for Urban Policy
Research.
Cervero, Robert, & Wu, K-L. (1997). Polycentrism, commuting, and residential location in the
San Francisco Bay area. Environment and Planning A, 29, 865–886.
Chowdhury, Tufayel A., Scott, Darren M., & Kanaroglou, Pavlos S. (2013). Urban form and
commuting efficiency: A comparative analysis across time and space. Urban Studies, 50(1),
191–207.
Cidell, Julie. (2013). Air transportation, airports, and the discourses and practices of globaliza-
tion. Urban Geography, 27(7), 651–663.
City of Montreal. (2014). Montreal urban agglomeration land use and development plan, draft.
Montreal, QC: City of Montreal.
Cladera, Roca, Josep, Duarte, Carlos, R., & Moix, Montserrat. (2009). Urban structure and
polycentrism: Towards a redefinition of the sub-centre concept. Urban Studies, 46(13),
2841–2868.
Coffey, William, & Shearmur, Richard. (2002). Agglomeration and dispersion of high-order service
employment in the montreal metropolitan region, 1981–96. Urban Studies, 39, 359–378.
Combes, Pierre-Philippe, Duranton, Gilles, & Gobillon, Laurent. (2011). The identification of
agglomeration economies. Journal of Economic Geography, 11, 253–266.
Decoville, Antoine, & Klein, Olivier. (2014, February). The limits of polycentrism at the city-
regional scale: The case of Luxembourg. European Journal of Spatial Development, 1(54).
Retrieved from http://www.nordregio.se/en/European-Journal-of-Spatial-Development/
Refereed-articles/
Derudder, Ben, Hoylder, Michael, Taylor, Peter J., & Witlox, Frank. (2012). International hand-
book of globalization and world cities. Cheltenham, UK: Edward Elgar Publishing Limited.
Duranton, Gilles, & Turner, Matthew A. (2012). Urban growth and transportation. The Review of
Economic Studies, 79, 1407–1440.
Egermann, Markus. (2009). The Saxon Triangle - a polycentric metropolitan region from an
actor-oriented perspective. Urban Research & Practice, 2(3), 269–286.
Ewing, Reid. (1997). Is Los Angeles sprawl desirable? American Planning Association Journal, 63
(1), 107–126.
Filion, Pierre. (2000). Balancing concentration and dispersion? Public policy and urban structure
in Toronto. Environment and Planning C: Government and Policy, 18, 163–189.
URBAN GEOGRAPHY 25

Filion, Pierre, Buntin G, Trudi, McSpurren, Kathleen, & Tse, Alan. (2004). Canada-U.S. Metropolitan
density patterns: Zonal convergence and divergence. Urban Geography, 25(1), 42–65.
Filion, Pierre, Bunting, Trudi, Pavlic, Dejan, & Langlois, Paul. (2010). Intensification and sprawl:
Residential density trajectories in Canada’s largest metropolitan regions. Urban Geography, 31
(4), 541–569.
Freestone, Robert, & Murphy, Peter. (2007). Metropolitan restructuring and suburban employ-
ment centers: Cross-cultural perspectives on the Australian Experience. Journal of the
American Planning Association, 64(3), 286–297.
Friedmann, John. (1986). The world city hypothesis. Development and Change, 17, 69–83.
Fujita, Masahisa, Krugman, Paul, & Mori, Tomova. (1999). On the evolution of hierarchical
urban systems. European Economic Review, 43, 209–251.
Fujita, Masahisa, & Ogawa, Hideaki. (1982). Multiple equilibria and structural transition of non-
monocentric urban configurations. Regional Science and Urban Economics, 12(2), 161–196.
Downloaded by [Library Services City University London] at 20:50 01 July 2016

Garreau, Joel. (1992). Edge City: Life on the New Frontier. New York, NY: Anchor Books.
Gaspar, Joel, & Glaeser, Edward L. (1998). Information technology and the future of cities.
Journal of Urban Economics, 43, 136–156.
GaWC. (2012). The World according to GaWC 2012. Retrieved October 9, 2014, from Globalization
and World Cities Research Network: http://www.lboro.ac.uk/gawc/world2012t.html
Giuliano, Genevieve, Redfearn, Christian, Agarwal, Ajay, & He, Sylvia. (2012). Network acces-
sibility and employment centres. Urban Studies, 49(1), 77–95.
Giuliano, Genevieve, Redfearn, Christian, Agarwal, Ajay, Li, Chen, & Zhuang, Duan. (2007).
Employment concentrations in Los Angeles, 1980-2000. Environment and Planning A, 39,
2935–2957.
Giuliano, Genevieve, & Small, Kenneth A. (1991). Subcenters in the Los Angeles region. Regional
Science and Urban Economics, 21, 163–182.
Glaeser, Edward, & Kahn, Matthew E. (2003). Sprawl and urban growth. National Bureau of
Economic Research, Working Paper 9733. Cambridge, MA: National Bureau of Economic
Research. Retrieved from http://www.nber.org/papers/w9733.pdf
Glaeser, Edward, Kallal, Heidi D., Scheinkman, Jose A., & Shleifer, Andrei. (1992). Growth in
cities. Journal of Political Economy, 100(6), 1126–1152.
Glaeser, Edward, & Kohlhase, Janet E. (2003). Cities, regions and the decline of transport costs.
National Bureau of Economic Research Working Paper Series. Retrieved from http://www.nber.
org/papers/w9886
Golob, Thomas F., & Regan, Amelia C. (2001). Impacts of information technology on personal
travel and commercial vehicle operations: Research challenges and opportunities.
Transportation Research Part C, 9, 87–121.
Gordon, Peter, & Richardson, Harry W. (1997). Are compact cities a desirable planning goal?
APA Journal, 63(1), 95–107.
Gordon, Peter, & Richardson, Harry W. (1996). Beyond polycentricity: The dispersed metropolis,
Los Angeles, 1970-1990. Journal of the American Planning Association, 62(3), 289–295.
Gordon, Peter, Richardson, Harry W., & Yu, G. (1998). Metropolitan and non-metropolitan employ-
ment trends in the US: Recent evidence and implications. Urban Studies, 35(7), 1037–1057.
Guillain, Rachel, Le Gallo, Julie, & Boiteaux-Orain, Celine. (2006). The evolution of the spatial
and sectoral patterns in Ile-de-France Over 1978-1997. Urban Studies, 43, 2075–2098.
Hamilton, Bruce W. (1975). Zoning and property taxation in a system of local governments.
Urban Studies, 12, 205–211.
Harvey, David. (1990). The condition of postmodernity: An enquiry into the origins of cultural
change. Cambridge, MA: Blackwell.
Henderson, J. Vernon, Kuncoro, Ari, & Turner, Matt. (1995). Industrial development in cities.
Journal of Political Economy, 103(5), 1067–1090.
Henderson, J. Vernon, & Mitra, Arnindam. (1996). The new urban landscape: Developers and
edge cities. Regional Science and Urban Economics, 26, 613–643.
26 M. N. SWEET ET AL.

Hoyler, Michael, Kloosterman, Robert C., & Sokol, Martin. (2008). Polycentric puzzles -
Emerging mega-city regions seen through the lens of advanced producer services. Regional
Science, 42(8), 1055–1064.
Inman, Robert P. (2003). Should Philadelphia’s suburbs help their central city? Business Review,
Q2, 25–36.
Jacobs, Jane. (2000). The nature of development. New York, NY: Modern Library.
Keil, Roger, & Boudreau, J-A. (2005). Is there regionalism after municipal amalgamation in
Toronto?. City, 9(1), 9–22.
Lang, Robert E. (2003). Edgeless cities: Exploring the elusive metropolis. Washington, DC: The
Brookings Institution.
Levine, Jonathan. (2006). Zoned out. Washington, DC: Resources for the Future Press.
Llewelyn-Davies, Banister, David, & Hall, Peter (2004). Transport and city competitiveness -
literature review. London: U.K: Department for Transport.
Downloaded by [Library Services City University London] at 20:50 01 July 2016

Lopez, Miguel A. (2005). Employment descentralisation: Polycentric compaction or sprawl? The case
of the barcelona metropolitan region 1986-1996. Barcelona: Universitat Autonoma de Barcelona.
Maier, Karel. (2009). Polycentric development in the spatial development policy of the Czech
Republic. Urban Research & Practice, 2(3), 319–331.
Maoh, Hanna, & Kanaroglou, Pavlos. (2007). Geographic clustering of firms and urban form: A
multivariate analysis. Journal of Geographical Systems, 9, 29–52.
McMillen, Daniel P. (2003). Employment subcenters in Chicago: Past, present, and future.
Chicago: Federal Reserve Bank of Chicago.
McMillen, Daniel P., & Lester, T. William. (2003). Evolving subcenters: Employment and
population densities in Chicago, 1970-2020. Journal of Housing Economics, 12, 60–81.
McMillen, Daniel P., & Smith, S C. (2003). The number of subcenters in large urban areas.
Journal of Urban Economics, 53, 321–338.
Meijers, Evert. (2008). Summing small cities does not make a large city: Polycentric Urban regions and
the provision of cultural, leisure and sports amenities. Urban Studies, 45(11), 2323–2342.
Meijers, Evert, Hoekstra, Joris, & Aguado, Ricardo. (2008). Strategic planning for city networks:
The emergence of a basque global city? International Planning Studies, 13(3), 239–259.
Mills, Edwin S. (1967). An aggregative model of resource allocation in a metropolitan area.
American Economic Review, 57(2), 197–210.
Mokhtarian, P L., & Meenakshisundaram, Ravikumar. (1999). Beyond tele-substitution:
Disaggregate longitudinal structural equations modeling of communication impacts.
Transportation Research Part C: Emerging Technologies, 7, 33–52.
Muth, Richard F. (1969). Cities and housing: The spatial pattern of Urban residential land use.
Chicago: University of Chicago Press.
Nelson, Arthur, Sanchez, Thomas W., Wolf, James F., & Farquhar, Mary Beth. (2004).
Metropolitan planning organization voting structure and transit investment bias: Preliminary
analysis with social equity implications. Transportation Research Record, 1895, 1–7.
Orfield, Myron. (2008). The new suburban typology: From American metropolitics: The new
suburban reality. In Eugenie Birch (Ed), The Urban and regional planning reader (pp. 47–54).
Milton Park: Routledge.
Parr, John. (2004). The polycentric Urban region: A closer inspection. Regional Studies, 38(3),
231–240.
Partridge, Mark, Olfert, M R., & Alasia, Alessandro. (2007). Canadian cities as regional engines
of growth: Agglomeration and amenities. Canadian Journal of economics/Revue Canadienne
D’économique, 40(1), 39–68.
Polese, Mario, & Champagne, Eric. (1999). Location matters: Comparing the distribution of
economic activity in the Canadian and Mexican Urban systems. International Regional Science
Review, 22, 102–132.
Porter, Michael E. (1990). The competitive advantage of nations. New York, NY: Free Press.
Punter, John. (2003). The vancouver achievement. Vancouver, BC: The UBC Press.
Redfearn, C L. (2007). The topography of metropolitan employment: Identifying centers of
employment in a polycentric urban area. Journal of Urban Economics, 61, 519–541.
URBAN GEOGRAPHY 27

Redfearn, C L. (2009). Persistence in urban form: The long-run durability of employment centers
in metropolitan areas. Regional Science and Urban Economics, 39, 224–232.
Romero, Vincente, Solis, Eloy, & De Urena, Jose M. (2014). Beyond the metropolis: New
employment centers and historic administrative centers and historic administrative cities in
the Madrid global city region. Urban Geography, 35(6), 889–915.
Ross, Catherine L. (2009). Megaregions: Planning for global competitiveness. Washington, DC:
Island Press.
Sassen, Saskia. (2001). The global city: New York, London, Tokyo. Princeton: Princeton University Press.
Shearmur, Richard. (2012). The geography of intrametropolitan KIBS innovation: Distinguishing
agglomeration economies from innovation dynamics. Urban Studies, 49(11), 2331–2356.
Shearmur, Richard, & Coffey, W J. (2002). A tale of four cities: Intrametropolitan employment
distribution in Toronto, Montreal, Vancouver, and Ottawa-Hull, 1981-1996. Environment and
Planning A, 34, 575–598.
Downloaded by [Library Services City University London] at 20:50 01 July 2016

Shearmur, Richard, Coffey, William, Dubé, C., & Barbonne, R. (2007). Intrametropolitan
employment structure: Polycentricity, scatteration, dispersal and chaos in Toronto, Montreal
and Vancouver, 1996-2001. Urban Studies, 44(9), 1713–1738.
Shin, Jungyeop. (2005). The statistically and economically significant clustering method for eco-
nomic clusters in an Urban region. Journal of the Korean Geographical Society, 40(2), 187–201.
Simmons, Jim, & Jones, Ken. (2003). Growth and change in the location of commercial activities
in Canada: With special attention to smaller urban places. Progress in Planning, 60, 55–74.
Smith, Neil. (1997). The satanic geographies of globalization. Public Culture, 101, 169–189.
Soja, Edward. (2011). Regional urbanization and the end of the Metropolis Era. In Gary Bridge &
Susan Watson (Eds.), The new blackwell companion to the city (pp. 679–689). West Sussex:
Wiley-Blackwell.
Song, Shunfeng. (1992). Monocentric and polycentric density functions and their required com-
mutes. Berkeley: University of California Transportation Center.
Statistics Canada. (2011). Table 051-0056 Estimates of population by census metropolitan area, sex
and age group for July 1, based on Standard Geographical Classification (SGC) 2011 annual
(persons), CANSIM. Retrieved September 26, 2014, fromhttp://www5.statcan.gc.ca/
Storper, Michael. (2010). Why does a city grow? Specialisation, human capital or institutions?
Urban Studies, 47(10), 2027–2050.
Storper, Michael. (2011). Why do regions develop and change? The challenge for geography and
economics. Journal of Economic Geography, 11, 333–346.
Storper, Michael, & Venables, A. J. (2004). Buzz: Face-to-face contact and the urban economy.
Journal of Economic Geography, 4, 351–370.
Sýkora, Ludek, Mulíček, Ondrej, & Maier, Karel. (2009). City regions and polycentric territorial
development: Concepts and practice. Urban Research & Practice, 2(3), 233–239.
Taylor, P. J. (2004). Regionality in the world city network. International Social Science Journal,
56(181), 361–372.
Taylor, P. J., Ni, Pengfei, Derudder, Ben, Hoyler, Michael, Huang, Jin, Lu, Fengyong . . . Shen,
Wei. (2009). The way we were: Command-and-control centres in the global space-economy
on the eve of the 2008 geo-economic transition. Environment and Planning A, 41(1), 7–12.
Tiebout, C. M. (1956). A pure theory of local expenditures. The Journal of Political Economy,
64(5), 416–424.
Veneri, Paolo. (2010). Urban polycentricity and the costs of commuting: Evidence from Italian
Metropolitan Areas. Growth and Change, 41(3), 403–429.
Wheaton, W. C. (1974). A comparative static analysis of urban spatial structure. Journal of
Economic Theory, 9, 223–237.
Wheaton, W. C. (1998). Land use and density in cities with congestion. Journal of Urban
Economics, 43(2), 258–272.
Wheaton, W. C. (2004). Commuting, congestion, and employment dispersal in cities with mixed
land use. Journal of Urban Economics, 55, 417–438.
Zolnik, E. J. (2012). The costs of sprawl for private-vehicle commuters. Journal of Transport
Geography, 20, 23–30.

You might also like