Вы находитесь на странице: 1из 19

Clark University

Interaction in an Urban System: Aspects of Trade and Commuting


Author(s): John B. Parr
Source: Economic Geography, Vol. 63, No. 3 (Jul., 1987), pp. 223-240
Published by: Clark University
Stable URL: http://www.jstor.org/stable/143951 .
Accessed: 22/01/2014 17:56
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
.
Clark University is collaborating with JSTOR to digitize, preserve and extend access to Economic Geography.
http://www.jstor.org
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
INTERACTION IN AN URBAN SYSTEM:
ASPECTS OF TRADE AND COMMUTING
JOHN B. PARR
University of Glasgow, United Kingdom
A hierarchical urban system is examined with respect to certain economic link-
ages or flows which exist among the different levels of the hierarchy. The underly-
ing locational basis of the urban system is represented in terms of a standard central
place model in which flows of central place goods are specified. In order to
incorporate important facets of contemporary urban systems, this framework is
extended to consider first the existence of specialized (non-central place) goods,
and second the presence of commuting, the flows associated with each phenom-
enon being separately discussed. Attention is then directed to the coexistence of
different types of flows, not only in terms of the interrelations among them but also
in terms of the balance-of-payments implications for the various levels of the
hierarchy.
Treating the economy of a region or
nation in terms of an urban system can
often yield additional insights into its
structure and operation. Yet location
theory only offers the researcher rather
unsatisfactory or at best incomplete mod-
els of the urban system. It will be argued
here, however, that central place theory,
as developed by Christaller [5; 6] and
L6sch [13], can be used as a basis for a
more realistic representation of the urban
system than has hitherto been available.
Although central place theory is not with-
out a number of conceptual and technical
weaknesses, these are generally not seri-
ous enough to preclude the use of the
central place system as the locational
foundation of a broader urban system.
Building on the structure of the standard
model of a central place system, addi-
tional economic linkages or flows are
introduced. These are concerned with
trade in specialized goods and with com-
muting. The focus throughout the paper
will be on the equilibrium nature of a
closed urban system which may be re-
garded as regional or national in scope.
Furthermore, emphasis will be placed on
levels of the system, rather than on indi-
vidual centers which comprise a particu-
lar level, although this latter facet will be
examined where appropriate.
THE STANDARD MODEL OF A
CENTRAL PLACE SYSTEM
The central place framework is fre-
quently presented in terms of a shopping
model. This is partly for pedagogical rea-
sons and partly because the provision of
consumer-oriented services represents an
undeniably important facet of an urban
system. Here, however, the concern will
be with the broader perspective of trade,
of which shopping is a part. Attention is
first directed to the locational structure of
central place goods, i.e., that economic-
activity grouping which forms the basis
of the central place system. Central place
goods consist of services and manufac-
tured goods which are produced for con-
sumer demand and/or intermediate de-
mand. The demand for central place
goods tends to be widely dispersed, while
the supply of these goods is heavily
market-oriented, to the virtual exclusion
of other orientations. As a consequence,
supply points are fairly centrally located
within their respective market (or
demand) areas. There, thus, exists a cor-
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
224 ECONOMIC GEOGRAPHY
respondence between the spatial distri-
bution of supply and the spatial distribu-
tion of demand, the strength of this
correspondence being determined by the
interaction between high transportation
costs (which favor many supply points
and small market areas) and economies
of scale (which favor few supply points
and large market areas). These represent
the essential locational characteristics of
central place goods which distinguish
them from other types of economic activ-
ity. An account of the manner in which
individual central place goods are com-
bined to form the overall urban system is
found in the works of Christaller [5; 6]
and L6sch [13], which are discussed by
King [12] and Berry and Parr [4].
SPATIAL STRUCTURE OF STANDARD MODEL
The standard model of the central
place system contains a specified number
of hierarchical levels of settlement which
are labeled 0,1, 2,.. ., . . . , N. where: 0
indicates the rural community; 1 is the
lowest level of central places consisting of
villages; m is a representative level; and N
is the highest level of central place and
also refers to the total number of urban
levels. Within such a hierarchy, sets of
central place goods are supplied, and
these are labeled Cl, C2,
. . ., Cm, .. , CN.
In actual central place systems the higher
the value of m, the greater tends to be the
relative importance of intermediate
goods within set Cm. Level m > 0 of the
hierarchy is assumed to be self-sufficient
in sets C1 through Cm and also supplies
sets C1 through Cm to other parts of the
system. In terms of central place goods
the hierarchy is successively inclusive.
Each center of level m > 0 has m market
areas: a level 1 market area (which is
wholly rural) to which set C1 is supplied; a
level 2 market area (which contains a
rural part as well as centers of level 1) to
which set C2 is supplied; a level 3 market
area (which contains a rural part and cen-
ters of levels 1 and 2) to which set C3 is
supplied; and so on up to the level m
market area (which has a rural part and
centers of levels ] through m - 1) to which
set Cm is supplied. An example of this
market area structure is shown in Figure
1, which represents a central place system
with five hierarchical levels, i.e., a rural
level and four urban levels.
It is important to note that the market
of level m represents a composite market
area for all goods of set Cm, and there may
be variation in the nature of the market
area for individual goods of this set. This
becomes apparent if we consider a con-
sumer good and an intermediate good,
both of set C3. For the consumer good of
set C3, which is supplied from each center
of levels 3 and 4 of the hierarchy, the level
3 market area will be continuous and,
thus, comprise the rural area and all cen-
ters of levels ] and 2. For the intermediate
good of set C3, however, which is also
supplied from each center of levels 3 and
4, the market area may be punctiform,
perhaps consisting in each case of centers
of level 2 only.
The structure of market areas and cen-
ters in Figure I represents a central place
system of the type proposed by Chris-
taller [5; 6] and L6sch [13, p. 130-32].
Unlike their systems, however,
this one is
based on a square lattice of supply points
[11]. The spatial structure of the system
corresponds to the K =
9 case, so that
within a level m market area of a center
of level m or higher (where m > 1) there
are 9 market areas of level m - 1. This
gives rise to a sequence of market area
frequencies (starting with the highest
level) of 1, 9, 81, 729 and a corresponding
sequence of center frequencies of 1, 8, 72,
648. Those interested in spatial organiza-
tion of central place systems will recog-
nize that this particular K 9 structure
adheres to the "administrative principle"
by which: (a) administrative areas coin-
cide with market areas; (b) administra-
tive centers coincide with market centers
(central places); and (c) an administrative
center of level m and its administrative
area of level m both lie wholly within the
level m + 1 administrative area of an
administrative center of level m + 1 or
higher.
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
AN URBAN SYSTEM 225
0 .0 _ _
0
* 0 0 0 000 0 0
* 0 - _
0 0
01-
* 0 0 0 0 0
0 10~~* 00 0 0 0 * 0
* ~ ~ ~ ~ ~~ 0 0 0 * .- Q _0
00 *. . . 0 _ _
center level market area
@ ~4_
.0 * .0 2
o kmS 40
Fig. 1. Centers and Market Areas in the Standard Central Place Model.
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
226 ECONOMIc GEOGRAPHY
It is emphasized that Figure 1 is a dia-
grammatic representation of a central
place system which is presented purely
for illustrative purposes. Such a scheme is
based on a set of strong assumptions re-
lating to homogeneity conditions, trans-
portation costs, f.o.b. prices, consumer
behavior and producer rationality [1].
The resulting market area structure for a
given good is such that any location is
supplied from the nearest supply point,
i.e., the nearest relevant center. Rushton
[23] and Saey [24], among others, have
argued that this is not always the case, and
Figure 1 could be modified (with some
difficulty) to take this into account. The
tendency for certain locations within the
system not to be served by the nearest
center is, in effect, a manifestation of the
Reilly [21] Law of Retail Gravitation,
although its relevance is not confined to
retailing. Within the context of a central
place system this would require the mar-
ket area boundary (between two compet-
ing centers of different levels) to lie
beyond the perpendicular bisector of the
two centers and toward the lower-level
center. In such cases, however, the loca-
tion in question is invariably supplied
from the second- or third-nearest center,
so that the spatial relationship between
demand and supply is broadly main-
tained. This issue arises again, and rather
than use the cumbersome term "from the
nearest center or the second- or third-
nearest center," the term "from the near-
est center" will be employed, with the
understanding that the latter term is an
abbreviation of the former.
TRADE IN CENTRAL PLACE GOODS
The pattern of trade in central place
goods among the various levels is indi-
cated by the left-hand side of Figure 2,
where it can be seen that level 3, for
example, exports set C1 to level 0, set C2
to levels 1 and 0, and set C3 to levels 2, 1,
and 0, and imports set C4 from level 4.
This pattern of central place trade is
based on the not unreasonable assump-
tion that for any level m or higher, at least
one good of set Cm is exported to levels m
-1 through 0. Such a pattern of trade has a
number of distinct features. First, exports
of central place goods always take place
to lower levels, and imports of central
place goods are received from higher
levels. Thus, it is impossible for level 1,
for example, to export to level 4 (since
level 1 only exports to level 0) or for level
3 to import from level 2 (since level 3 only
imports from level 4). With some justifi-
cation Pred [19; 20] found this pattern of
interlevel interaction unsatisfactory as a
framework for modeling innovation and
information flows within an urban sys-
tem. Second, and this is implicit in Figure
1 though not obvious from Figure 2, there
are no intralevel or lateral trade flows of
central place goods, so that a center of
level 2, for example, will never export to
another center of level 2. This follows
from the characteristic of the standard
model that centers of a given level are
functionally identical, so that there is no
basis for trade among them. Such an
absence of intralevel or lateral trade was
also noted by Pred [19; 20]. Third, the
nature of trade balances for central place
goods of the various levels exhibits a par-
ticular pattern.
Because of the downward exporting to
lower levels, the following trade relations
exist: level N exports central place goods
to each lower level but imports no central
place goods from them and, therefore,
has a central place trade surplus with
every lower level; level 1! m < N exports
central place goods to each lower level
and imports other central place goods
from each higher level and, thus, has a
central place trade surplus with every
lower level and a deficit with every higher
level; level 0 exports no central place
goods but imports from each higher level
and, therefore, has a central place trade
deficit with every higher level. Letting
X represent the value of exports of cen-
tral place goods from level m, and M
t
the value of imports of central place
goods into level m, the pattern of interac-
tion in the standard model may be sum-
marized for each level m in terms of the
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
AN URBAN SYSTEM 227
Central - place-trade flows Level Specialized -trade flows
C4
< C3_ 4 ~ ~
~~~S --2
C
3
C2
C 2
S C
4
- '-'111
S
4
C
C2
S
2 _
2
2 ~~~4-S12
1 So S 'I ,_,
Fig. 2. Trade Flows in the Urban System.
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
228 ECONoMIc GEOGRAPHY
balance of central place trade X
'
- Mm'.
This balance of central place trade will
necesarily be positive for level N and
negative for level 0. For levels 1
!
m <
N,
however, the balance will be positive,
negative, or neutral, as will become ap-
parent later. Such a structure of balances
of central place trade can only be sus-
tained if other flows are present, although
these are not usually specified in presenta-
tions of the standard model. This balance-
of-payments aspect will be considered in
the Appendix and will draw on the mate-
rial presented later.
THE EXISTENCE OF SPECIALIZED GOODS
Attention has been focused on trade
flows based on central place goods.
Within modern urban systems, however,
a further type of trade flow is also present,
and in order to take account of this, we
introduce the existence of a further class
of goods, which is referred to here as
"specialized goods." The terms "non-
central" or "non-nodal" might have been
employed in this connection, but the term
"specialized" is used, in order to empha-
size the locational particularity of the
goods in terms of their demand and/or
supply characteristics. As with central
place goods, specialized goods may con-
sist of services or manufactured goods,
and their production may be for con-
sumer demand or intermediate demand.
There are, however, fundamental loca-
tional differences between the two types
of good. Whereas a central place good
belonging to set Cm is supplied from lev-
els m and higher, is exported to levels i
through m -1 (where O< i
-
m -1), and is
also exported to any given location in the
urban system from the nearest relevant
center, one or more of these features will
be absent in the case of a specialized
good.
THE LOCATION OF SPECIALIZED GOODS
PRODUCTION
In order to describe the trade in spe-
cialized goods within an urban system, it
is first necessary to review their locational
pattern of production. Unlike the supply
of central place goods, which has an
overwhelming market orientation, the sup-
ply of specialized goods has a variety of
orientations. One such orientation is to-
ward sources of low-cost energy, while
other orientations of specialized goods
production involve sources of low-cost
labor, as well as locations of climatic or
other natural amenity. The orientation to
ports of certain kinds of economic activ-
ity represents a further example, although
a number of activities in larger ports
should properly be regarded as activities
of the central place type. Certain types of
specialized goods production are loca-
tionally oriented to the production of
other types of specialized goods. Copper
smelting and, historically, the iron indus-
try have been oriented to raw material
locations. By contrast, the steel industry
has become increasingly oriented to the
location of steel-using industries. Some-
times the nature of the association is diffi-
cult to unravel, particularly in the case of
groups of industries which appear to have
a locational affinity, the obvious example
being an industrial complex [7; 22;
25].
Finally, the production of specialized
goods may be oriented to particular size
classes of the urban hierarchy. This is
obvious for agricultural activity and re-
source exploitation (which are oriented to
the rural community) but less obvious
perhaps in the case of other activities.
Nevertheless, there is some evidence to
suggest that particular size classes repre-
sent preferred locations for certain activi-
ties. This type of orientation is usually
thought to be related to labor market
conditions, land price considerations, and
the existence of an appropriate set of net
agglomeration economies [10]. Textile
weaving in the U.S. for example, has its
highest locational incidence in centers
with populations of between 2,500 and
10,000
[15].
The food processing industry
appears to display a similar locational
sensitivity to various size classes of the
urban hierarchy. Fruit and vegetable can-
ning, for example, tends to be associated
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
AN URBAN SYSTEM 229
with lower levels of the hierarchy, while
flour milling and meat packing typically
occur in higher levels.
Taken as a whole, the production of
specialized goods involves an extremely
complex locational pattern. It gives rise to
a configuration of market areas lacking
any resemblance to the ordered structure
shown in Figure 1. In the case of those
goods which are shipped throughout the
rest of the system from level m only, in
such a way that any location is supplied
from the nearest level m center, a fairly
well-defined market area structure would
emerge, and a strong element of market
orientation could be said to be present.
And even if only some of the level m
centers were exporting the good, a
reasonably coherent market area struc-
ture would exist. For other goods, partic-
ularly intermediate goods, the market
area structure is often punctiform, frag-
mented, and overlapping. For example,
the market area structure for a given good
may be such that only four centers pro-
duce the good, each serving between
three and ten other centers in a widely
dispersed pattern. It is under these cir-
cumstances that the concept of the mar-
ket area loses its validity, and the analysis
should best proceed in terms of non-
spatial, hierarchial relationships.
Unfortunately, there exists no single
framework (corresponding to the Chris-
taller-L6sch framework underlying Fig-
ure 1) for analyzing the spatial structure
of specialized goods and the locational
interdependencies among these, both on
the supply and demand sides. Neverthe-
less, it is possible to impose some order on
the production of specialized goods (and
the associated trade flows) by examining
this within the framework of the standard
central place model. Thus, for a given
point in time, each center within an urban
system is identified on the basis of its level
within the central place hierarchy. In
relating the production of specialized
goods to the central place hierarchy, sev-
eral assumptions are introduced. Alterna-
tive, though somewhat more restrictive,
assumptions relating to the locational struc-
ture of specialized goods have been pro-
posed elsewhere [17;
18].
By the first
assumption there exists a group of spe-
cialized goods {Zi} where i = 1, 2, ., n,
with n representing the serial number of
the last specialized good, the goods being
arranged in no particular order. A second
assumption requires that each specialized
good of the group {Zi} be produced in one
or more but not all levels of the central
place hierarchy. The third assumption,
which is quite independent of the second,
is that a particular set of the group of
specialized goods 7Zi} is produced in level
m, this set being referred to as Sm,. and the
term So referring to that set of specialized
goods produced within the rural com-
munity (level 0). From the second assump-
tion, a specialized good within set Sm may
also be present in sets Sm,- and Sm+2, for
example.
TRADE IN SPECIALIZED GOODS
Having specified the locational struc-
ture of specialized goods, we are now
able to consider the trade flows based on
this. To do this, we make the further
assumptions that each specialized good
may be exported from a given level to
one or more other levels, and that each
specialized good may be imported into a
given level from one or more other levels.
One possible structure of specialized
trade flows is indicated on the right-hand
side of Figure 2. It is immediately appar-
ent that upward exports within the sys-
tem may now occur. For example, goods
of set S2 are being exported to levels 3 and
4, as well as to levels 0 and 1, although not
every good of this set is necessarily ex-
ported to all four levels. As already noted,
flows of central place goods in the stan-
dard model, which are shown on the left-
hand side of Figure 2, have no such
upward destination. On the right-hand
side of Figure 2, set Sm is exported to all
other levels. This is not necessary, how-
ever, and in an alternative structure of
specialized trade, certain flows may be
missing. There may, for example, be no
export of set S2 to level 3, which would
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
230 ECONOMIc GEOGRAPHY
imply that of all the goods belonging to
set S2, none has an export destination in
level 3.
The right-hand side of Figure 2 is only
concerned with specialized goods trade
among levels, and for this reason it is
unable to reflect certain inter-center fea-
tures of this trade. These features, none of
which is present in the standard model,
are now noted. First, individual centers of
level m may export different combina-
tions of goods belonging to set Sm. Also,
certain centers may export all goods of set
Sm,
while others may export only some,
while still others may export none at all, in
which case such centers may be regarded
as pure central places of level m. Second,
since a given specialized good of Sm is not
exported from every center of level m,
the possibility of lateral exports arises,
i.e., a specialized good of set Sm being
exported from a center of level m to one
or more other centers of level m. Third,
an export from level m of a specialized
good of set Sm to, say, level m + 1 will not
necessarily be shipped to all centers of
level m + 1. It may be shipped as an input
to production only to those centers of
level m + 1 which are themselves produc-
ing a specialized good of the set Sm+i.
Fourth, and in the nature of an implica-
tion of the previous point, it is possible for
cross-hauling between levels to occur.
Thus, a good belonging to sets Sm and Sm+1
may be exported from centers a, b, and c
of level m to centers r, s, and t of level
m + 1, while centers x, y, and z of level
m + 1 may export this same good to cen-
ters d, e, and
f
of level m. This cross-
hauling between pairs of levels is not to
be confused with the more conventional
cross-hauling that takes place as a result
of product differentiation or branding,
although crosshauling between levels and
even within a level could arise for this
very reason.
In view of this complexity, little can be
said about the structure of specialized
flows among levels. If X represents the
value of exports of specialized goods
from level m, and M
'
the imports of
specialized goods by level m, the balance
of specialized trade is given by X
'
-
M .
The introduction of specialized goods
production into the standard model can
be expected to have the effect of intensi-
fying the central place trade flows, both
in terms of consumer demand and inter-
mediate demand. For example, a center
of level 3 may have within its market area
a center of level 2 which is supplying one
or more specialized goods of set S2. As a
result of this, the level 3 center has a larger
retail trade and more extensive banking,
wholesaling, and transportation sectors
than would be the case if the specialized
goods production had not been present.
The existence of specialized goods pro-
duction may also lead to the appearance
of additional central place goods. Sup-
pose that a good is only supplied from
levels 3 and 4 (the highest level) as a pro-
duction input for a specialized good be-
longing exclusively to set S2. As long as
each level 2 center is supplied from the
nearest center exporting the input, the
good in question can be considered a cen-
tral place good of set C3; if this was not
the case, it would be regarded as a spe-
cialized good belonging to sets S3 and S4.
These central place goods, which are not
present in the standard model, only
emerge because of the existence of spe-
cialized goods. If Figure 2 is regarded as a
representation of all trade flows within an
urban system, the left-hand side now
comprises the central place trade flows
covered by the standard model, as well as
the central place trade flows arising from
the production of specialized goods,
these latter flows resulting from intensifi-
cation of flows in the standard model and
flows of central place goods not pre-
viously present. To take account of this,
the balance of central place trade is now
expressed as X M'
-
.
THE PRESENCE OF COMMUTING
Within a central place hierarchy, not all
individuals employed (or self-employed)
at a given level reside at that level, and not
all employed (or self-employed) individ-
uals residing at that level are employed
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
AN URBAN SYSTEM 231
there. This locational separation of em-
ployment and residence, and the com-
muting to which this gives rise, is
common enough within urban or metro-
politan areas,
but here the concern is with
the situation among levels of the urban
system. In order to examine this,
we
return to the standard model which in its
original form says nothing about com-
muting. There exist a number of factors
which lead to inter-level commuting. The
first involves an individual's preference
for residing at a level of the hierarchy
other than the one at which he is em-
ployed. This preference may be based on
the set of perceived amenities
available,
or it may be related to the presence of
relatives, friends, and social organiza-
tions. To cite a couple of examples, an
individual employed in a relatively popu-
lous level 4 center may prefer to reside
some distance away, in a less crowded,
level I center. An individual may be
employed in level 0 or level I but may
prefer a residence in a center of level 3
with its wider range of services. Obvious-
ly, in both cases the preference has to be
backed by an ability and willingness to
bear the cost of commuting, including
time cost.
A second factor encouraging commut-
ing concerns the fact that an individual
employed at a given level may not be able
to afford the cost of living (particularly
the cost of housing) at that level, but can
afford the cost of living, as well as the
associated cost of commuting, at a center
of some other level. It is possible, of
course, for these two factors to reinforce
each other. Thus, an individual employed
in a level 4 center may not be able to
afford to live there, but may actually
prefer to live in a center of level 1. There
is a third factor which may encourage
commuting, although its importance is
difficult to ascertain. This involves the
positive utility derived either from the
physical separation of employment and
residence or from commuting, itself, so
that if, for example, the location of employ-
ment is given, the location of residence is
chosen in such a way as to involve a diff-
erent level. Since inter-level commuting
exists, it is reasonable to expect there to
be intra-level or lateral
commuting,
involv-
ing commuting from one center of level
m to another center of this level. Similar
factors underlie this type of commuting,
but since we are concerned with the
standard model (in which all level m cen-
ters could be expected to have similar
cost-of-living levels), the second factor is
likely to be unimportant.
The pattern of inter-level commuting
within the hierarchy is likely to be com-
plex. From what has been discussed so
far, it is perfectly possible for a given
level to have incommuters and outcom-
muters, and even for there to be cross-
commuting between pairs of levels.
Nevertheless, it is the lower levels of the
hierarchy which are always prominent in
the overall pattern of commuting: as resi-
dence levels (e.g., commuting from level
1 to level 4); as employment levels (e.g.,
commuting from level 3 to level 1); or as
both (e.g., commuting from level 0 to
level 1). This is because commuting be-
tween two higher levels of the hierarchy
is unlikely due to the relatively large dis-
tances involved, and it can be expected
that there is an upper limit on the distance
that people are. willing to or can afford to
commute. A typical pattern of commut-
ing flows among levels (based on the dis-
tances within Figure 1) is indicated in
Figure 3. The flow T2, for example, indi-
cates the commuter flow from level 2 to
other levels (levels 0 and I in this case). It
will be seen that there are no flows
between level 4 and level 3 (distance 90
km) and no flows between either level 4
or level 3 and level 2 (minimum distance
30 kmi). This would imply an upper limit
on commuting of, say, 28 km. Such a rela-
tively low upper limit might reflect a poor
level of private transportation and/or the
traditional practice of a relatively long
working day.
While Figure 3 is concerned with flows
among levels, it says nothing about the
particular centers involved. Let us as-
sume that commuting takes place from
level I to levels 3 and 4, and for the sake of
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
232 ECONOMIC GEOGRAPHY
simplification no other commuting flows
exist. From the standpoint of a given level
1 center, commuting will tend to take
place to the nearest center of level 3 or 4,
but only if it is within the upper limit of
commuting travel. From the standpoint
of a particular center of level 3 or 4, only
certain level 1 centers will be sending
commuters to this center: those which are
located within the upper limit of com-
muting. Similar considerations apply to
intra-level commuting. A given level m
center may have residents commuting to
other level m centers, and although a par-
ticular level m center may be bypassed as
an employment center, there is an ob-
vious restriction on the other level m cen-
ters to which commuting will take place.
Up to this point commuting has been
examined largely in terms of the individ-
ual motivation, and it is therefore appro-
priate to consider the phenomenon in
more aggregate terms. One aspect of this
is the nature of overall commuter travel
within the system. It is likely that the
operation of the various factors that influ-
ence commuting will give rise to a pattern
of commuting which can be described in
terms of a gravity model of the following
form:
Tjk
=
AWjEk/djk
(1)
(i > 0;
dik
-dmax)
where:
Tik
is the number of individuals
residing in location j who commute to
location k;
Wj
is the resident workforce at
location j; Ek is the number of employees
at location k;
djk
is the distance (or travel
time) between j and k; dmax is the upper
limit on commuting; and A and i are con-
stants. Given the differences throughout
the urban system in the quality of trans-
portation systems and in the attitudes to
commuting, it is possible that dmax and i
(the distance exponent) may each display
a systematic variation with the level from
which commuting takes place. This rela-
tively simple formulation of commuting
could be extended by taking into account
the various intervening opportunities be-
tween j and k.
The existence of a commuting flow
between two levels of the hierarchy im-
plies a difference within each level be-
tween the size of employment and the
size of the resident workforce, i.e., an
inequality in the demand and supply for
labor at each level. For the outcommut-
ing level, the resident workforce will be
greater than employment, while for the
incommuting level, employment will be
greater than the resident workforce.
Since a level may have both outcommut-
ers and incommuters, as can be seen in
Figure 3, it is possible to speak in terms of
a level experiencing net outcommuting or
net incommuting, so that in the former
case the resident workforce exceeds em-
ployment, and in the latter case the oppo-
site is true. It is very difficult to generalize
about the nature of net commuting
among the various levels. Within the
urban systems of North America and
Europe, however, the higher levels of the
hierarchy tend to be associated with net
incommuting, the lower levels tend to be
levels of net outcommuting, while the
middle levels have a more even pattern of
commuting movements. The economic
impact of commuting on a particular
level is more accurately viewed in terms
of income flows rather than flows of indi-
viduals, and in this connection it is possi-
ble to make use of the "balance of com-
muter payments." Letting the incomes
associated with the outcommuters from
level m (i.e., residents of level m em-
ployed at other levels) be represented by
Jm and the incomes associated with incom-
muters to level m (i.e., employees at level
m residing at other levels) by Lm; the bal-
ance of commuter payments can be ex-
pressed as Jm
-
Lm. Unless there are signifi-
cant differences between the incomes of
incommuters and outcommuters at a
given level (and such a contingency
should not be disregarded), a positive
balance reflects net outcommuting, while
a negative balance reflects net incom-
muting.
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
AN URBAN SYSTEM 233
Commuting from Level
Commuting
from
levels 4, 3and2 levels
Qand1
+--
T1
T4
~~4
1
3 1
[ . 22 T
4?T0
1 F I FT-' 1
T4_
?
4- t
T- 0
T~~~~~~
Fi.3.Cmmtn Fos nth tndr Mdl
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
234 ECONOMIC GEOGRAPHY
COEXISTENCE OF FLOWS WITHIN
THE BROADER URBAN SYSTEM
In the two preceding sections, consid-
eration was given to specialized trade and
commuting, and in each case this was
examined in the context of the standard
central place model. We now deal with
the more realistic case of the broader
urban system in which there is usually a
coexistence of flows based on central
place trade, specialized trade, and com-
muting. Particular attention is directed to
the interrelation among these flows and to
the manner in which these shape the eco-
nomic linkages within the urban system.
INTERRELATIONS AMONG DIFFERENT FLOWS
It has already been suggested that the
locational pattern of specialized goods
may be influenced in part by the loca-
tional pattern of central place goods,
inasmuch as the supply of specialized
goods may be associated with different
size classes of the urban hierarchy. In this
case, it can be argued that the specialized
trade flows in question are influenced by
central place trade flows. It has also been
argued that specialized trade flows may
give rise to the intensification of central
place flows and to additional central
place flows not present in the standard
model. Another interrelation previously
alluded to is the influence of central place
trade on commuting, by which large cen-
ters supplying central place goods have
high living costs and so force commuting
to take place from centers of a lower
level. A further interrelation concerns the
influence of commuting on central place
trade. If commuters purchase central
place goods in the level at which they are
employed (e.g., lunch-time shopping or a
movie after work), these purchases repre-
sent an export of the employment level
and an import by the residence level.
Naturally some of these purchases might
take place in any event because the goods
in question are not supplied at the resi-
dence level, so that commuting travel also
serves as shopping or entertainment
travel. However, in the case of purchases
at the employment level, which are also
on offer at the residence level, there has
been what amounts to import penetration
of the residence level by the employment
level, and an element of inter-level cen-
tral place trade exists which would not
have been present, had there been no
commuting. One final example involves
the influence of specialized trade on com-
muting. The existence within a center of a
given level of substantial specialized
goods production may create sufficiently
undesirable living conditions or suffi-
ciently high costs of living so that certain
individuals employed there would prefer
or be obliged to reside in other levels or in
other centers of the same level in which
this specialized goods production is not
present.
The interrelation among flows is some-
times precipitated by a change in one set
of flows, as the urban system moves from
one equilibrium state to another. Suppose
that as a result of technological change
certain central place goods of set Ci now
have to be supplied from fewer locations.
Whereas, before the change the goods
would be supplied from centers of levels
1 through N, they are now supplied from
the fewer centers of levels 2 through N.
This upward transference of goods as a
result of technological change would obvi-
ously cause a redefinition of sets Ci and
C2. Assuming that the demand for labor
remains unchanged, additional jobs be-
come available at centers of level 2
through N as a result of the centralization
of production. One option for the level 1
employees associated with the supply of
these particular set C1 goods is to migrate
to levels 2 through N, each employee
moving from a level 1 location to the
nearest center of a higher level. However,
in view of the short distances involved in
this case, as well as the disruption and
possibly higher cost of living associated
with migration, an adjustment involving
commuting from level 1 to the higher lev-
els might be selected. Thus, changed cen-
tral place trade flows bring about the
onset or the extension of commuting
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
AN URBAN SYSTEM 235
flows. If, however, the goods involved set
C3, and if the centralization of production
was at level 4, the distances involved
would be sufficiently great that commut-
ing would not be an option. In this sense,
commuting and migration can be re-
garded as substitutes. To quote Termote
[26, p. 84]: "Once a certain distance is
reached, the individual will prefer one
type of spatial adjustment to the other,
and he will substitute migration for com-
muting." Commuting and migration may
also be complements, as would be the
case when commuting is a temporary
adjustment until migration (i.e., the phys-
ical move of the household) can be
accomplished.
THE SIZE DISTRIBUTION OF
THE URBAN SYSTEM
Not surprisingly, the introduction of
specialized trade flows and commuting
flows into the standard central place
model will cause a modification of the
overall size distribution of centers. In the
standard model, all centers of level m are
generally assumed to be of identical size
[2; 3; 8]. This follows from the fact that all
such centers are indistinguishable in
terms of the economic and social func-
tions performed. Such a size distribution
can be graphed, with the vertical axis,
scaled logarithmically, representing the
population of a given center and the
horizontal axis, also scaled logarith-
mically, representing the rank of that cen-
ter. Each center is plotted by population
and rank as a single point, and if all of
these points are now connected, the result
is a curve, which begins from a point on
the vertical axis, representing the popula-
tion of the single center of level N, and
falls downward to the right
in a series of
steps, each step corresponding to a level
of the hierarchy composed of identically-
sized centers, with the end of the curve
representing the last center of level 1. The
rural population is usually excluded from
consideration.
In the case of trade based on special-
ized goods, centers of a given level can be
expected to have differing outputs of spe-
cialized goods, leading to differing total
populations. Similarly, in the case of
commuting, the gravity model nature of
commuting, combined with the existence
of an upper limit on commuting, is likely
to cause a differential outcommuting pop-
ulation among centers of a given level
and, thus, result in differing total popula-
tions. If we now take into account the
effect of either specialized trade or com-
muting, the step-like curve of the stan-
dard model would be transformed into a
smoother curve, perhaps still retaining
the vestiges of steps. Naturally, the effect
of inclusion of specialized trade and com-
muting would cause the curve to become
even smoother, particularly in view of
what was said above regarding the interre-
lation among the various types of flows.
Such a smooth curve is an observable fea-
ture of the size distribution of urban
systems, the curve approximating a
downward-sloping straight line, if the
distribution is rank-size, or a downward-
sloping curve concave to the origin and if
the distribution is lognormal [16]. Ore
possible outcome of these additions to the
standard model is that a center of level m
may have a larger population than a par-
ticular center of level m + 1, by virtue of
the former's importance in specialized
goods production and/or because it also
has a high level of net outcommuting. In
other words, population is no longer a
reliable index of hierarchical level, as in
the standard model. Actual conditions are
such that many other influences could be
expected to blur the step-like curve of the
standard model. For this reason, no
doubt, Beckmann [2; 3] chose to treat
these numerous influences as "random
shocks." Here, however, the blurring is
due to two systematic influences and the
interactions to which these give rise.
THE BALANCE-OF-PAYMENTS EQUILIBRIUM
OF A LEVEL
At various points within this paper, ref-
erence has been made to the balance at a
particular level with respect to central
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
236 ECONOMic GEOGRAPHY
place trade (X
'
- M
'
), specialized trade
(Xm - Mu), and commuting payments
(Jm
-
Lm). It is now possible to see how
these various balances are related. For
any given level of the hierarchy, central
place exports, specialized exports, and
the income associated with outcommut-
ing each represent "payments received"
from the rest of the system, and they can
also be treated in a Keynesian sense as
injections into the overall economy of the
level. By contrast, central place imports,
specialized imports, and the income asso-
ciated with incommuting are all "pay-
ments made" to the rest of the system,
and they can also be regarded as with-
drawals from the economy of the level.
Before indicating how these various flows
are related to each other, we make the
simplifying assumption that there are no
other interlevel flows within the urban
system, e.g., government transfers, capi-
tal movements, property payments, etc.
Under these circumstances a balance-of-
payments equilibrium at a given level,
which would also correspond to a simple
macroeconomic equilibrium at that level,
would require the following general con-
dition to hold:
XMn
+
XMn
+ Jm
-
Mm' + Mm' + Lm
(m
=
0, 1, 2, . . ., N) (2)
In other words, for level m, payments
received are equal to payments made;
alternatively, the injections into the econ-
omy of level m are equal to the with-
drawals from the economy of level m.
Rearranging equation (2) into balances
yields equation (2a) where each balance
is shown in parentheses:
(XMC MC) + (XS - MS)
+
(Jm
-
Lm)
=
0
(2a)
In order to illustrate the nature of the
balance-of-payments equilibrium, we
may consider the following three possi-
bilities. First, a negative balance of cen-
tral place trade can be sustained by some
combination of a positive balance of spe-
cialized trade and a positive balance of
commuting payments. Second, a nega-
tive balance of central place trade can be
accompanied by a negative balance of
specialized trade, as long as there is a
positive balance of commuting pay-
ments. Third, a negative balance of com-
muting payments can coexist with a nega-
tive balance of central place trade, if
there is a positive balance of specialized
trade. Other combinations of balances
are possible, but it is always the case that
the various balances must sum to zero, as
in equation (2a).
This discussion of the balance-of-pay-
ments equilibrium serves to demonstrate
how the various flows may be related; it
represents something of a simplification,
since other flows would clearly be present
in an actual urban system [17]. The pay-
ments-received side for a given level, the
left-hand side of equation (2), would also
include non-level government expendi-
tures, capital imports, and property pay-
ment received in connection with past
capital exports from that level. Analo-
gously, the payments-made side for a
given level, the right-hand side of equa-
tion (2), would include non-level taxes,
capital exports, and property payments
made in connection with past capital
imports to that level. Within this more
complicated structure of flows the range
of possible combinations of balances to
achieve an overall balance-of-payments
equilibrium is considerably extended.
FURTHER COMMENTS
It has become apparent that while cen-
tral place theory may be adequate for
analyzing certain urban systems within a
contemporary or historical setting, and
while it is still extremely useful for under-
standing the increasingly market-
oriented nature of economic activity
within modern economies, it can never be
regarded as a general model of an urban
system. Valuable as the original models
of Christaller and L6sch are, these can
only be considered partial
in character, as
both authors were at pains to point out.
The regrettable fact is that central place
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
AN URBAN SYSTEM 237
theory has tended to be viewed as a self-
contained framework to be embraced or
rejected. Indeed, the numerous empirical
studies that have accumulated have
tended to be concerned with testing its
ability to describe actual systems,
and in
this connection the use of central place
theory as a "straw man" is not unknown.
Seldom is it seen as one element, perhaps
the single most important locational ele-
ment, of a broader urban system. Here,
the underlying premise has been that
when the standard central place model
has specialized trade flows, and commut-
ing flows are grafted onto it, we obtain a
more realistic portrayal of the urban sys-
tern. The central place structure, thus,
becomes the underlying locational matrix
of a more complex urban system, of
which the central place structure is a
component or subsystem. This existence
of various subsystems within an overall
urban system has been considered by
Dziewonski [9], although in a somewhat
different context.
The concern throughout has been pri-
marily of a conceptual nature, and in the
application of this approach to the analy-
sis of actual urban systems, a number of
difficulties can be expected to arise. One
such difficulty involves the classification
of individual centers in hierarchical
terms, particularly in the case of centers
based primarily on specialized trade. A
further difficulty relates to the large data
requirements, and although the concern
is with the definition of hierarchical levels
and sets of goods, information is needed
on individual centers and individual
goods. A more general problem and one
which arises with any static-equilibrium
approach is the fact that the approach
does not deal with the emergence of the
urban system to its present state. The
prominence of the central place subsys-
tem in no sense implies that this existed
first, and only afterwards did that part of
the urban system based on specialized
trade emerge. Such may have been the
case throughout much of Europe and the
earlier-settled parts of North America,
but it is not hard to find examples of spe-
cialized trade (based on mineral exploita-
tion and manufacturing) leading to the
formation of an urban system and influenc-
ing the subsequent development of a cen-
tral place subsystem. Furthermore, even
in cases where the central place subsys-
tem emerged first, the subsequent devel-
opment of specialized trade may have
been sufficiently intense as to cause a
modification or hierarchical redefinition
of the central place subsystem. Thus,
the
development of centers based on special-
ized trade may result in substantial ele-
ments of central place trade being trans-
ferred to them. In the U.K. the nineteenth
century development of shipbuilding and
heavy engineering enabled Glasgow to
replace Edinburgh as the dominant com-
mercial capital of Scotland. Within the
U.S. during the course of the twentieth
century, Los Angeles emerged as the
dominant capital of the western states,
usurping this role from San Francisco,
largely as a result of growth of the aero-
space and defense-related industries. The
simulation model proposed by Morrill
[14] might be very helpful in approaching
this problem.
These various considerations notwith-
standing, the approach has a number of
advantages. The admittedly heavy reli-
ance on the central place subsystem,
apart from enabling central place trade
flows to be specified in terms of hierar-
chical levels, permits specialized-trade
flows and commuting flows to be viewed
in terms of these same hierarchical levels.
Viewing such flows at the scale of the
individual center would lead to an unman-
ageable complexity. A further advantage
of the approach is that it allows a distinc-
tion to be made between central place
trade and specialized trade. Such a dis-
tinction is of vital significance in under-
standing the upward and downward trans-
mission of growth within the urban
system, and thus in the analysis of such
policy issues as the development poten-
tial of different levels of the hierarchy, or
the selection of particular levels for invest-
ment as a means of stimulating the region-
al or national economy. More generally,
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
238 ECONOMIc GEOGRAPHY
the approach offers a framework for
demonstrating the interrelations of diff-
erent types of flows and how these coexist
in an urban system. It cannot be empha-
sized too strongly, however, that this dis-
cussion of flows is incomplete and merely
indicative of the direction that might be
followed. It should, therefore, be seen as
part of a more general pattern of eco-
nomic flows within a modern urban
system.
APPENDIX
The Appendix is concerned with the
balance-of-payments conditions within
the closed standard central place model
which was outlined earlier and extended
in the closed standard model to consider
commuting. A balance-of-payments equi-
librium in the standard model can only be
achieved if additional trade flows are
present, although these are generally not
made explicit. Such flows involve solely
the specialized goods of set So. These
goods, which would include agricultural
output and raw materials, are produced
exclusively within the rural community
(level 0). Since the initial concern is with
the standard model in the absence of
commuting, equation (2) is rewritten as
follows:
X
t
+Xm? = Mlt + MO? (3)
(m
=
01,2, . .., N)
which can be expressed in terms of bal-
ances as
(XI _M t) +(Xm? - Mm?)= (3a)
where the superscript t refers to the flows
of central place goods of the standard
model and the superscript 0 indicates
flows of the specialized goods of set So.
The trade flows thus consist of the central
place flows of the left-hand side of Figure
2, together with the flows from level 0
only, on the right-hand side. As earlier, it
is assumed that no other flows are present.
It is apparent from equation (3a) that
the balance-of-payments equilibrium cor-
responds to a balance-of-trade equi-
librium. Since set So goods are exported
from level 0 and no other specialized
goods are supplied within the system,
level 0 must have a positive balance of
specialized trade. This offsets its necessar-
ily negative balance of central place trade
referred to earlier. For the urban levels,
however, the situation is less clear cut,
and two situations are possible. In the first
situation, level 0 exports set So goods to
every urban level to fulfill consumer de-
mand, which is usually assumed to be
identical throughout the system. The re-
sulting negative balance of specialized
trade of each urban level will therefore
require a positive balance of central place
trade, so that for each urban level the
combined surplus of central place trade
with the lower levels will be greater than
the combined deficit with the higher lev-
els. Since level N by definition does not
import from higher levels, it will necesar-
ily have a positive balance of central
place trade, as referred to earlier. In the
second situation, level 0 exports set So
goods to certain urban levels only, as pro-
duction inputs for central place goods.
For example, a set So good may be
exported to levels m through N as a pro-
duction input for a central place good of
set Cm. (The possibility of level 0 export-
ing set 0 goods to a level m but not to
levels m + 1 through N also is excluded
because this would imply the existence of
specialized goods at urban locations, and
these are not present in the standard
model.) Each level m through N will have
a negative balance of specialized trade
which will require an offsetting positive
balance of central place trade. The other
urban levels not involved in this special-
ized trade (levels 1 through m
-
1) will
each have simply a zero (or neutral) bal-
ance of central place trade, so that the
combined surplus of central place trade
with the lower levels will be exactly equal
to the combined deficit with the higher
levels.
If the existence of commuting is now
introduced into this more explicit version
of the standard model, equations (3) and
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
AN URBAN SYSTEM 239
(3a) are replaced by equations (4) and
(4a):
Xnt + Xn? + Jm
= M
I
+MMO +Lm (4)
and
(XI tM o+ (X ? -MO?)
+
(Jm Lm)
= 0
(4a)
Obviously, a balance-of-payments equi-
librium can now be attained by addi-
tional means, and the possibilities may be
contrasted with conditions in the absence
of commuting. Level 0 still necessarily
has a positive balance of specialized trade
and a negative balance of central place
trade. Now, however, these two balances
need not be equal, provided that the bal-
ance of commuting payments is positive
or negative. For the urban levels, two
situations are again possible, the first
being one in which level 0 exports set So
goods to every urban level. For levels 1
!
m < N.
it is no longer necessary (with
commuting present) for the balance of
central place trade to be positive. A nega-
tive balance of central place trade may
coexist with the necessarily negative bal-
ance of specialized trade, as long as these
two negative balances are offset by a posi-
tive balance of commuting payments.
For level N, it is still the case that the
balance of central place trade must be
positive and the balance of specialized
trade negative, but the two balances need
not be equal, as long as the balance of
commuting payments is positive or nega-
tive. In the second situation involving
commuting, level 0 only exports set So
goods to levels m through N. For each
level m through N -
1, it is not necessary
for there to be a positive balance of cen-
tral place trade, so that the necessarily
negative balance of specialized trade
may be accompanied by a negative bal-
ance of central place trade, on condition
that both balances are offset by a positive
balance of commuting payments. For
levels 1 through m
- 1 the balance of
central place trade need no longer be neu-
trail, provided that the balance of com-
muting payments is negative or positive.
For level N the same conditions apply as
in the first situation involving commuting.
LITERATURE CITED
1. Beavon, K. S. 0. Central Place Theory: A Rein-
terpretation. London: Longman, 1977.
2. Beckmann, M. J. "City Hierarchies and the Dis-
tribution of City Size," Economic Development
and Cultural Change, 6 (1958), pp. 243-48.
3. Beckmann, M. J. Location Theory. New York:
Random House, 1968.
4. Berry, B. J. L. and J. B. Parr, with B. Epstein, A.
Ghosh, and R. Smith. Market Centers and Retail
Location: Theory and Applications. Englewood
Cliffs, N.J.: Prentice-flall, 1988.
5. Christaller, W. Die zentralen Orte in Siiddeutsch-
land. Jena: G. Fischer, 1933. Translated by
C. W. Baskin as Central Places in Southern
Germany. Englewood Cliffs, N.J.: Prentice-
Hall, 1966.
6. Christaller, W. "Das Grundgeriist der rdumli-
chen Ordnung in Europa," Frankfurter Geo-
graphische Hefte, 24 (1950), pp. 1-96.
7. Czamanski, S. "Some Empirical Evidence on
the Strengths of Linkages between Groups of
Related Industries in Urban-Regional Com-
plexes," Papers of the Regional Science Associa-
tion, 27 (1971), pp. 137-50.
8. Dacey, M. J. "Population of Places in a Central
Place Hierarchy," Journal of Regional Science,
6 (1966), pp. 27-33.
9. Dziewonski, K. "General Theory of Rank-Size
Distributions in Regional Settlement Systems:
Reappraisal and Reformulation of the Rank-
Size Rule," Papers of the Regional Science
Association, 29 (1972), pp. 73-86.
10. Evans, A. W. "The Pure Theory of City Size in
an Industrial Economy," Urban Studies, 9
(1972), pp.
49-77.
11. Hoover, E. M. An Introduction to Regional
Economics. New York: Alfred E. Knopf, 1971.
12. King, L. J. Central Place Theory. Beverly Hills:
Sage, 1984.
13. L6sch, A. Die riumliche Ordnung der Wirt-
schaft. Jena: G. Fischer, 1941. Translated from
the 2nd edition (1944) by W. H. Woglom and
W. F. Stolper as The Economics of Location.
New Haven: Yale University Press, 1954.
14. Nlorrill, R. L. "The Development of Spatial Dis-
tributions of Towns in Sweden: An Historic-
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions
240 ECONOMIC GEOGRAPHY
Predictive Approach," Annals of the Associa-
tion of American Geographers, 53 (1963), pp.
1-14.
15. Morse, R. M., 0. P. Mathur, and M. C. K.
Swamy. Costs of Urban Infrastructure as Re-
lated to City Size in Developing Countries. Palo
Alto, Calif.: Stanford Research Institute, 1968.
16. Parr, J. B. "A Class of Deviations from Rank-
Size Regularity," Regional Studies, 19 (1976),
pp. 285-92.
17. Parr, J. B. "Economic Flows in L6sch's Modified
Urban System," Space-Structure-Economy: A
Tribute to August L6sch. Edited by R. H. Funck
and A. Kuklinski. Karlsruhe: von Loeper Verlag,
1986.
18. Parr, J. B. "The Tinbergen Analysis of an Urban
System and Alternative Approaches," Envi-
ronment and Planning A, 19 (1987), pp. 187-204.
19. Pred, A. R. "Large City Interdependence and
the Pre-Electronic Diffusion of Information,"
Geographical Analysis, 3 (1971), pp. 165-81.
20. Pred, A. R. "The Growth and Development of
Systems of Cities in Advanced Economies,"
Systems of Cities and Information Flows: Two
Essays (Lund Studies in Geography, Series B,
No. 38). Lund: Gleerup.
21. Reilly, W. J. The Law of Retail Gravitation.
New York: Pilsbury Publishers, 1953.
22. Richter, C. E. "The Impact of Industrial Link-
ages on Geographic Association," Journal of
Regional Science, 9 (1969), pp. 19-28.
23. Rushton, G. "Postulates of Central Place Theory
and the Properties of Central Place Systems,"
Geographical Analysis, 3 (1971), pp. 140-56.
24. Saey, P. "Three Fallacies in the Literature on
Central Place Theory,"
Tijdschrift
voor Econ-
omische en Sociale Geografie, 64 (1973), pp.
181-94.
25. Streit, ME. "Spatial Associations and Economic
Linkages between Industries," Journal of Re-
gional Science, 9 (1969), pp. 177-88.
26. Termote, M. "Migration and Commuting in
Ldsch's Central Place System," The Analysis of
Regional Structure: Essays in Honour of August
LUsch. Edited by R. Funck and J. B. Parr. Lon-
don: Pion, 1978.
This content downloaded from 201.174.75.10 on Wed, 22 Jan 2014 17:56:43 PM
All use subject to JSTOR Terms and Conditions

Вам также может понравиться