You are on page 1of 14

Zygmunt Tkocz and

Gustav Kristemen

Commuting Distances and Gender:


A Spatial Urban Model

In any urban center the commuting distances are a function of the spatial structure
of the center and of the characteristics of the commuters. In this paper theoretical
relationships between commuting distances and distances of residences to city
centers are derived for monocentric and polycentric cities. These relationships are
then linked to the sociological determinants of commuting distances. An econometric
model encompassing both spatial structure variables and social variables is con-
structed and estimated using data for sixteen urban centers. Gender differences are
focused upon. The expansion method is used.

1. INTRODUCTION
The classical urban models, such as Alonso’s (1964), generally assume that the
jobs are located in the city centers. In more recent studies this assumption is
relaxed somewhat. For instance, Durbin and Sung (1987) used the presupposition
that at least one member of each household works in the city center. The
assumption that jobs are located at city centers is crucial for the derivation of price
profiles for housing and land in the monocentric urban models. The need to be
close to the center to reduce the daily commuting time is the main argument for
declining land prices with increasing distance to the center. In this paper the
relationships between commuting distances and urban structure is investigated for
both monocentric and polycentric cities.
In modern societies often both husbands and wives work outside the home. In
Denmark, in 1987, the labor force participation rates (calculated over the total
number of people aged 15 to 74) was 74.3 for men and 62.3 for women. The
gender differentials in commuting distances have been extensively studied. A
number of investigations have shown that women tend to hold jobs nearer to their
homes then men (Kuhl et d. 1966; White 1977; Madden 1981; McLafferty and
Preston 1991; Johnston-Anumonwo 1992). Lopata (1980) discussed the difficulties
that women without cars experience in Chicago’s transit facilities that are not
designed for women needing to move children and goods. Choldin (1985) notes the
changes in the women’s transportation and commuting needs as they go through

Zygmunt Tkocz and Gustav Kristensen teach in the Institute of Economics at Odense
University, Denmark.
Geographical Analysis, Vol. 26, No. 1 (January 1994) 0 1994 Ohio State University Press
Submitted 12/30/91. Revised version accepted 6/15/93.
2 / Geographical Analysis
the family’s life cycle. After the birth of the first child the women feel restricted in
their ability to move. However, this does not reduce their commuting significantly,
especially single mothers (Madden 1981). Rosenbloom (1989) emphasized that the
combined transport time for women is longer than for men, due to the women’s
socially mandated obligation to take children to kindergarten and to school.
The children’s effect on commuting distances is generally inconclusive
(Johnston-Anumonwo 1992). Madden (1981) remarks that part of the children’s
effect manifests itself in the choice of residential locations. According to Madden,
age reduces commuting distances. For married people Madden concludes that the
changes of residence with age “slightly favor the wife’s job in the residential
location choice.” Singles are more mobile and tend to live closer to the city center
(Madden 1981; Johnston-Anumonwo 1992). White (1977) notes that married
women are more likely to work in the suburbs. McLafferty and Preston (1991)
remark that “Women make up a significant share of. . . central place type activities.”
In this article the social determinants of commuting distances are linked to the
determinants of commuting distances arising from the spatial structure of urban
centers. Specifically, a relationship between commuting distances and distances
between residences and city centers is defined, and its parameters are then
respecified into functions of social variables. Gender differences are focused upon
throughout. The expansion method (Casetti 1972; 1986) is employed. The urban
models used in this paper are related to those used by Krakover (1985), Kristensen
(19911, and Hansen and Kristensen (1991).
The empirical data used in the investigation presented in this paper were
collected from different sources. The number of inhabitants, POP, and the share of
manufactured labor force, SM, for sixteen Danish urban centers were extracted
from the 1991 Statistical Yearbook of Denmark. LP will denote the logarithmic
transformation of POP. Other variables were obtained from telephone interviews.
To this effect, a total of 2,300 addresses were extracted from the telephone books
of the sixteen urban centers in the study. Specifically, the top and midpage
addresses from each printed column in telephone books were selected. Since in
Denmark almost all households have telephones, this sample can be considered
representative of Danish households. In total 2,300 numbers were selected. For
each address its distance to the city center, t, was determined.
Telephone interviews were attempted at all the addresses selected, and 1,276
responses were obtained, which corresponds to a 55 percent response rate. In the
1,276 respondent households there was a total of 689 men and 715 women who
were in the labor force. The following variables were extracted from the inter-
views: age (AGE and LAGE, where LAGE denotes the logarithmic transformation
of AGE); number of children, CH;a dummy variable, S, identifying the “status” of
the head of the household, coded 1 for single and 0 for nonsingle; annual income
before tax (MY for men and WY for women); education level (ED), in five levels
with 1 lowest, 5 highest); commuting distance (in kilometers, labelled 0,” for men,
and D, for women); and a car dummy (CAR, coded 1 for households with a car,
and 0 otherwise).
The models estimated from these data were developed in stages. The starting
point was a relationship between commuting distances (namely, the distance
between a person’s residence and his/her work place), and distances of residences
to city center (namely, distances between a person’s residence and the center of
the city in which the residence is located). The actual model building can be
divided into three steps. The first step focused on the relationship between
commuting distance and its sociological determinants. The second one is con-
cerned with the relationship between commuting distances and distances of
residences to city center. The third step involved combining these relationships by
the expansion method.
Zygmunt Tkocz and Gustav Kristensen / 3
2. THE SOCIOLOGICAL DETERMINANTS OF COMMUTING DISTANCES
Intercity Commuting
Intercity commuting (ICC) needs to be considered separately. Many people who
live in small towns work in towns other than the ones in which they reside.
Obviously, widespread intercity commuting weakens the relationship between
commuting distances and distances between residences and city centers. For
example, the town of Roskilde in the sample is predominantly inhabited by people
who work in Copenhagen, thirty kilometers away. In manufacturing towns like
Horsens and Herning (see Table 2) intercity commuting seems to be predomi-
nantly oriented toward small manufacturing towns.
For the two biggest towns (Copenhagen and Aarhus), intercity commuting was
operationally defined as involving a commuting distance at least twice as large as
the largest measured distance from a residence to the center of these cities. For
small towns, intercity commuting was operationally defined as a commuting
distance of more than fifteen kilometers. The intercity commuting in the sample
calculated using these definitions is shown in Table 1.
People such as sailors, salesmen and others who travel “all over the country”
and “all over the region” or more than two hundred kilometers away are totally
disconnected from the urban system and cannot be considered as commuters.
Consequently, they were excluded from the study.
The probability of a person not being an intercity commuter was estimated by a
logit model. The result was for men,

h o b ( ZCC = 0)

1/ 1 + exp (2.97 - .35LP - 1.OlLAGE + .13CH + .15ED)


=
( (2.24) (-3.78) (-3.19) (1.35) (2.35)
Obs = 853. Cases with ICC = 1 are 128.
Cases with ICC = 0 are 725.
Here and hereafter t-values are shown in parentheses under the corresponding
regress ion coefficients.
The probability of a person not being an intercity commuter was for women,

Prob( zcc = 0)

+ exp (4.63 +
= 1/ 1
( (2.32)
- .53LP - 1.92LAGE - .21CH
(-3.22) (-3.99) (-1.20)
.55ED)
(4.05)
Obs = 975. Cases with ZCC = 1 are 47.
Cases with ZCC = 0 are 928.
These results should be interpreted to mean that the probability of being an
intercity commuter increases with positive coefficients and decreases with negative

TABLE 1
Inter-City Commuting ( I C C ) in the Sample
Men Women

Observations 129 49
Commuting distance &m) 46 33
4 / Geographical Analysis
TABLE 2
Basic Statistics
Inhabitants Share mcu ation Commuting distance (incl ICC)
(looos) inmade. men women

Copenhagen 1.337 15% 12 (13) 8 (8)


Aarhus 200 16% 6 (7) 5 (7)
Odense 139 21% 6 (10) 4 (6)
Aalborg 114 20% 5 (10) 5 (5)
Esbjerg 72 26% 4 (10) 4 (4)
Randers 55 27% 3 (15) 3 (7)
Horsens 48 31% 3 (11) 4 (6)
Vejle 46 25% 5 (12) 4 (5)
Kolding 45 24% 3 (12) 4 (7)
Helsingor 43 18% 5 (17) 4 (10)
Roskilde 40 11% 5 (19) 5 (10)
Naestved 38 18% 5 (17) 3 (5)
Viborc 29 17% 3 (7) 2 (3)
Hernhg 29’ 33%
Hjoning 24 21%
Nyborg 15 19%

coefficients. In a capsule, the results indicate the following. Women have a lower
ZCC probability than men. It is more difficult for more educated people to find job
opportunities in smaller urban centers. Women (but not men) with children tend
to have within-city jobs. Children hinder womens’ intercity commuting but en-
courage intercity commuting for men.

Sociological Factors
The relationship between commuting distances and distances of residences from
city centers will be sometimes in this paper referred to as “commuting distance
patterns” or “distance patterns.” The industrial structure plays an important role in
commuting distance patterns since manufacturing is more likely than, for example,
service and finance to be located in the urban peripheries. Relatively more men
than women are employed in manufacturing. Figure 1 shows the shares of men
and women in manufacturing as a fraction of total employment in manufacturing.
The data are for Denmarks sixteen regions. The manufacturing shares are calcu-
lated in percent of nonagricultural employment
A British study (Grieco et al. (1989) suggests that women’s use of means of
transportation differs from men’s, and that women spend more time commuting
because they travel by bus while men travel by car. A comparison of mean
commuting distances by gender and by transportation mode is shown in Table 3.

0.10 0.20 0.30 0.40

FIG. 1. Share of Gender in Manufacturing as Functions of Share of Total Employment in


Manufacturing SMM = share of men in manufacturing; SMW = share of women in manufacturing:
SM = share of total employment in manufacturing.
Zygmunt Tkocz and Gustav Kristensen / 5
TABLE 3
Means of Transport; Commuting Distance
Men Women
Obs Distance (km) Obs Di5tance (km)

Train 24 31 25 24
Car 461 16 340 9
Bus 43 14 108 6
Motorcycle 24 7 2 4
Cycle 132 7 211 3
Foot 53 2 88 2
Total 738 13 774 6

For all transportation modes the commuting distances for men are longer than for
women.
The estimated equations reported earlier suggest that children may have some
influence on the probability of intercity commuting and that such influence may
differ by gender, while income does not affect these probabilities. In the results to
be presented next, income does not appear to influence the commuting distances.
When the ZCC dummy is entered neither the children variable nor the income
variable is significant.
A model relating commuting distances to their sociological determinants was
defined and estimated. After some experimentation, the sociological variables
selected for inclusion in the model were S, ED, CAR, and LACE. The data for all
the sixteen cities were pooled, and the intercept was expanded into the logarithm
of population size, LP, in order to correct for the effect of city size on commuting
distances. The intercity commuting dummy, ZCC, was used, the manufacturing
share variable, SM, was not. The estimated equation for men was

D,, = 1.81 + 2.07LP - 2.60s + .13ED + 2.06CAR


(.37) (7.85) (-2.53) (.59) (2.92)
- 1.95LAGE + 35.38ZCC
( - 1.62) (37.82)

R2 = .69 Adj. R 2 = .69 Obs = 678

and for women,

0, = 5.68 + 1.12LP + .29S + .40ED + 2.16CAR


(1.81) (6.51) (.52) (2.50) (4.67)
- 2.27LAGE + 27.47ZCC.
(-2.91) (29.00)
R2 = .59 Adj. R 2 = .59 Obs = 713

These equations show that the educational level influences the commuting dis-
tances for women, but not for men. At higher ages the commuting distances
decline significantly for women. The same tendency seems to exist for men
although it is not significant at the 5 percent level. Bigger cities are characterized
by longer commuting distances than smaller cities. Car ownership is associated
with longer commuting distances. Single men are closer to their jobs than married
men. Single women are not significantly closer to their jobs than married women.
Children influence the commuting distances through the probability for ZCC to
become 1. Within the city they do not have a significant effect on commuting
distances.
6 / Geographical Analysis

3. SPATIAL STRUCTURE AND COMMUTING DISTANCES


In any given urban center the commuting distances depend upon the spatial,
distribution of jobs, the spatial distribution of residences, and the extent to which
people have a preference for short commuting distances. In this section we will
discuss a spatial theory of the relationship between commuting distances and
distances of residences to city centers.
The Monocentric Model
In a monocentric city to which all jobs are at or near the city center, the
commuting distance of a person will increase linearly with the distance of the
person’s residence from the city center. Namely, we will have the relationship:

D = a, + a,t
where D denotes commuting distances, and t distances between residences and
city center. In the simplest Alonso-type model with all jobs exactly located at the
city center, a , and a , will be respectively 0 and 1. If we relax this assumption and
assume that the density of jobs increases toward the city center and that people
have some degree of preference for shorter commuting distances, a , will be
greater than 0 and a , will decline below 1.
An empirical test of model (1) was carried out using the data for the city of
Aarhus. Equation (1) was estimated separately for men,

D,, = 3.52 + .81t R2 = .118 Adj. R 2 = ,104 Obs = 66


(2.28) (2.93)

and for women

D, = 5.12 + .45t R 2 = .022 Adj. R 2 = .008 Obs = 71


(2.56) ( 1.25)

The influence of residential distances to city center on commuting distances is


obvious. The positive and significant intercept values indicate that, as it could be
expected, all jobs are not in the exact center of the city.
The Polycentric Model
In a nonmonocentric city the commuting patterns will be influenced by the
presence of secondary employment centers. Figure 2 shows the idealized urban
structure with secondary employment centers symmetrically located on concentric
rings around the city center that is employed in this study. The relationship
between commuting distances and distances between residences and the city
center can be discussed in purely theoretical terms as follows.
We wish to answer this question: In this system, what is the expected commut-
ing distance for people living at alternative distances from the city if we assume
that the probability for a person to be employed in a given employment center
declines with the commuting distance involved, and increases with the number of
jobs there?
This question can be addressed using a Monte Carlo approach. To this effect,
first we specify the distances of the rings to the city center, the number of jobs in
the employment centers, and a suitable probability function to assign residents to
employment centers. Then, based on these, we simulate the commuting distances
for a population of urban residents. Finally, using the results of the simulation we
Zygrnunt Tkocz and Gustav Kristemen / 7

FIG.2. Nonmonocentric Town Model with Secondary Employment Centers

estimate the relationship between commuting distances and distances of resi-


dences to the city center. This process can be easily repeated to explore a range of
alternative assumptions.
The spatial structure portrayed in Figure 2 encompasses the following special
cases that are worthy of separate consideration:
1. All jobs are concentrated in the city center (The Alonso case).
2. Jobs are located at the city center and in secondary employment points in the
O-ring.
3. Only the two x-rings contain employment centers, and no jobs are located at
the center or in the O-ring.
4. Jobs are located at the city center, in the O-ring, and the x-rings.
Figure 3 shows the simulated relationship between commuting distances and
distances of residences from the city center in a model with a center and one ring
of secondary employment centers, corresponding to case 2 above. The relationship
exemplified by Figure 3 will be referred to as a center-oriented distance pattern.
Figure 4 shows the simulated relationship in a model with employment points in
the two x-rings as in case 3 above. This relationship exemplifies a periphery-
oriented distance pattern.

40.

o+. . , , , , , . . . , , , . ,
t
2 4 8 a 10 IZ 14 18 18 20

FIG.3. Commuting Distance as a Function of Distance 'between Residence and City Center.
Calculated in a model with a center and one ring of secondary employment points. X1 = people with
low aversion against commuting distance; X2 = people with high aversion against commuting distance.
8 / Geographical Analysis

Fic. 4. Commuting Distance as a Function of Distance between Residence and City Center.
Calculated in a model with a center and two rings of secondary employment points. X1 = people with
low aversion against commuting distance; X2 = people with high aversion against commuting distance.

In Figures 3 and 4, the X 1 curve was produced assuming a low aversion to long
commuting distances, while the X 2 curve reflects a high aversion to long commut-
ing distances. These low and high aversions were entered into the simulations by
letting the probability of being employed in a certain point decline slowly (or
rapidly) with distance to that point.
The basic specification of the relationship between commuting distances, D , and
distances of residences to city center, t , in a nonmonocentric town is a polynomial,
as Figures 3 and 4 clearly indicate. For a third-degree polynomial the model is

D = a, + a,t + a 2 t 2 + a,t3. (2)


The coefficients of the initial model ( 2 ) can be expanded with respect to a variable
indexing the aversion to long commuting distances. The car dummy, CAR, was
used. Low aversion is defined as the possession of a car, high aversion as the
nonpossession of a car. This can be expressed as

aj = b, + b,,CAR. (3)
The terminal model then becomes

D = (bo1 + b,,CAR) + (bo2+ b,,CAR)t


+ (b,, + b,,CAR)t2 + (b,, + b,,CAR)t3 (4)
The estimation of equation ( 4 ) (and all the subsequent equations in this paper) was
carried out by removing any variables that do not increase the adjusted R2. Using
again the Aarhus data, equation ( 4 ) was estimated separately for men and for
women, after having excluded the intercity commuters. The following results were
obtained:

Dm = 5.36 + 4.06CAR - 1.38t + .16t2 .


(3.25) (3.71) (-1.94) (2.72)
R2 = .32 Adj. R2 = .29 obs.= 63

D, = .39 + (3.77 + .48CAR)t - .94t2 + .062t3


(2.30) (2.64) ( -2.63) (2.88)
R2 = .30 Adj. R2 = .25 obs. = 67
Zygrnunt Tkocz and Gustav Kristensen / 9

FIG. Commuting Distance as a Function of Distance between Resi-Lnce and City Center.
Predictc by the individual town model estimated on data for Aarhus. WCARl = women, nonintercity
commuter, car driver; WCARO = women, nonintercity commuter, non-car driver.

Figures 5 and 6 show the distance patterns for men and women produced by the
two estimated equations for Aarhus. These estimated patterns are qualitatively
similar to the ones produced by simulation and portrayed in Figures 3 and 4. They
suggest that more women then men are employed in the city center and subcen-
ters, while men to a greater degree are employed in the areas around those centers
and at the periphery. The same distance patterns were also obtained, although with
lower t values, for the other fifteen towns in the sample.

4. THE EXPANDED MODEL


In the previous analysis the coefficients of equation (2) were expanded only with
respect to the variable CAR. However, it seems appropriate to expand the
coefficients with respect to all the variables appearing in the ‘‘sociological’’model
plus the industrial structure variable. Specifically, the following expansion equa-
tions were defined:

aj = bj, + bj,LP + bjzSM + bj,S + bj,ED


+ bj,CAR + bj,LAGE + bj,ICC. (5)

The terminal model generated from the initial model (2) and the expansion

0.0 , . . , . , , . , , , . .,...,,.. . , , . , , . , , , . , , t
5 10

FIG.6. Commuting Distance as a Function of Distance between Residence and City Center.
Predicted by the individual town model estimated on data for Aarhus. MCARl = men, non-ICC, car
driver; MCARO = men, non-ICC, non-car driver.
10 / Geographical Analysis

equations (5)is as follows:

D = (boo+ b,,LP + b,,SM + bO3S+ b,,ED


+b,,CAR + b,,LAGE + b,,ZCC)

+ ( b , , + bllLP + b,,SM + bl3S + b,,ED

+b,,CAR + b,,LAGE + b,,ZCC)t

+ (b,, + b,,LP + b,,SM + b,,S + b,, E D


+b,CAR + b,,LAGE + b2,ZCC)t2

+ (b30 + b31LP + b22SM + bz3S + b,, ED


+b,CAR + b3,LAGE + b,,ZCC)t3. (6)

Equation (6) was estimated using the data for all the sixteen urban centers in the
sample, and excluding the variables that do not increase the adjusted R2. The
estimated for men are

D,,, = (18.71 + .79LP - 1.80s - 5.06LAGE + 39.49ZCC)


(1.74) (2.11) ( - 1.71) ( - 1.76) (16.84)

+ (-9.53) + .58CAR + 2.57LAGE - 3.5OZCC)t


( - 1.40) (4.11) ( 1.40) ( - 1.97)

+ (1.47 - .38LAGE + .61ZCC)t2


(1.49) (-1.46) (1.88)

+ (-0.042 - .0021S + .0011LAGE - .024ZCC)t3


( - 1.11) (-2.13) (1.10) ( - 1.63)

R2 = .72 Adj. R 2 = .71 Obs = 676

a0

D,
1s

10

FIG. 7. Commuting Distance as a Function of Distance between Residence and City Center.
Predicted by the overall model. SM = .16%, POP = 200,000 inhabitants. WCARl = married women,
non-ICC, thirty-year-old car drivers; WCARO = married women, non-ICC, thirty-year-old non-car
drivers.
Zygmunt Tkocz and Gustuv Kristensen / 11

and for women

D, = (-22.61 + 1.55LP + 78.47SM + 2.45s


(-3.74) (2.34) (3.84) (1.67)

+ .41ED + .61CAR + 43.27ZCC)


(2.24) (1.09) (13.31)

+ (32.03 - 1.51LP - 92.45SM - 1.66s


(4.31) ( -2.79) (-3.30) ( - 1.65)

- 1.50LAGE - 17.99ZCC)t
( -2.55) ( - 3.87)
+ (-8.89 + .49LP + 29.87SM + .29S - .024ED
(-3.47) (3.24) (2.67) (1.77) (-1.88)

+ .10CAR + .20LAGE + 6.05ZCC)t'


(2.99) (1.69) (3.40)

+ (.75 -.04OLP - 2.89SM + .0012S + .0016ED


(2.75) (-3.23) (-2.14) (-1.74) (2.12)

- .0033CAR - .0090LAGE - .65ZCC)t3


( - 1.77) ( - 1.72) ( - 3.38)
R2 = .66 Adj. R 2 = .64 Obs = 710
The Predicted Distance Patterns
In the unexpanded initial model (2) the coefficients are fixed. In the terminal
model (6) the coefficients of the relationship between commuting distances and
distances between residences and city centers is a function of a number of
expansion variables. In this section the estimated terminal models are analyzed. To
this effect the expansion variables in the estimated terminal model are set at
representative values, and then the distance patterns thus produced are graphed
and discussed.
Figure 7 shows a distance pattern for a thirty-year-old married woman with a
college education, living in a city with 200,000 inhabitants and in which the labor
force share in manufacturing is .16. The two curves in the figure distinguish
between car drivers and non-car drivers. The pattern obtained is very similar to
that obtained by simulation assuming that jobs are located at the city center and in
the secondary employment centers in the O-ring (Figure 3), and to the pattern for
all women in Aarhus (Figure 5).
In Figure 8 the distance pattern is likewise shown for a thirty-year-old married
man with a college education (with and without a car) living in an identical city.
The distance pattern in this figure is closest to that produced by simulation for the
x-rings case. The curve for car drivers is comparable to that produced for low
aversion to long commuting distances (Xl) and the curve for non-car drivers is
comparable to that produced when a high aversion to long commuting distances is
assumed (X2).
The estimates of equation (6) indicate that the share of the manufacturing
variable does not influence significantly the distance pattern for men. For women
12 / Geographical Analysis

Dm
15.0.
-MCARl _____ YCARO
12.5.

10.0

2.5 ...,....,... , , , . . . , . . ..................... t


5 10

FIG. 8. Commuting Distance as a Function of Distance between Residence and City Center.
Predicted by the overall model. SM = .16%, POP = 200,000 inhabitants. MCARl = mamed men.
non-ICC, thirty-year-old car drivers; MCARO = married men. non-ICC, thirty-year-old non-car drivers.

7.

I_ WSMlJ _____ WSM241 '-,

FIG. 9. Commuting Distance as a Function of Distance between Residence and City Center.
Predicted by the overall model. POP = 50,000 inhabitants. WSMl5 = married women, non-ICC,
thirty-year-old car drivers, SM = .15%.WSM24 = married women, non-ICC, thirty-year-old car drivers,
SM = 24%.

it is different. Figure 9 shows, for a city with 50,000 inhabitants, the effect of
varying the manufacturing shares. The distance pattern for women is center-ori-
ented when SM = .15 and periphery oriented when SM = 24.
Age seems to influence the distance patterns differently for men and women.
For women higher ages are associated with smaller commuting distances (Figure
10); for men higher ages not only decrease their commuting distances but also
makes them center-oriented (Figure 11).
Single men experience shorter commuting distances (Figure 12), single women
do not (Figure 13).

FIG. 10. Commuting Distance as a Function of Distance between Residence and City Center.
Predicted by the overall model. SM = .16%, POP = 200,000 inhabitants. W70 = married women,
non-ICC, seventy-year-old car drivers; W20 = married women, non-ICC, twenty-year-old car drivers.
Zygmunt Tkocz and Gustav Kristensen / 13

FIG. 11. Commuting Distance as a Function of Distance between Residence and City Center.
Predicted by the overall model. SM = .16%, POP = 200,000 inhabitants. M70 = married men, non-
ICC, seventy-year-old car drivers; M20 = married men, non-ICC, twenty-year-old car drivers.

FIG. 12. Commuting Distance as a Function of Distance between Residence and City Center.
Predicted by the overall model. SM = .16%,POP = 200,000 inhabitants. MS1 = single men, non-ICC,
car drivers; MSO = married men, non-ICC, car drivers.

5 . CONCLUSION

The investigation reported in this paper suggests the following:


a. The commuting distances are higher for persons who reside farther from city
centers.
b. Because men tend to work in the periphery, while women tend to work in the
city centers, the women’s distance patterns are closer to that for the monocentric
Alonso-type of city than is the case for men’s distance patterns.
c. It can be shown experimentally, by simulation, that differences in center
versus periphery employment by themselves can produce distance patterns similar
to those produced by our empirical analyses.
25

D,
D, -w s 1 .....wso

20 -

01.. , , , ... , ..,... , , . , . ..., . , .,... I t


5 10

FIG. 13. Commuting Distance as a Function of Distance between Residence and City Center.
Predicted by the overall model. SM = .16%, POP = 200,000 inhabitants. WS1 = single women,
non-ICC, car drivers; WSO = married women, non-ICC, car drivers.
14 / Geographical Analysis
d. Men in general experience longer commuting distances than women. Married
couples tend in their residential location choices to adjust to the wife’s job. This
latter conclusion is supported by the following. First, consider the theoretical
urban models considered here: even if we assume that men work in the periphery,
this by itself does not imply that the commuting distances have to be longer for
men than for woman. It is only if a married couple adjusts to the wife’s job that we
can explain the differences between the theoretical distance patterns (Figures 3
and 4) and the estimated distance patterns (Figures 5 and 6). Second, single men
experience shorter commuting distances than married men, while married women
do not experience longer commuting distances than single women. Third, in the
periphery of manufacturing towns, women acquire periphery-oriented distance
patterns; women living in the periphery nevertheless adapt to their work place,
unlike men (Figure 9). These points confirm the conclusions reached by Madden
(1981).
e. Women’s commuting patterns seem to be more dependent on the city’s
structure than those of men, and are more influenced by changes in city structure
than men’s, for example, by a change in the industrial structure of the town.
f. Age clearly reduces the commuting distances for women and men.
g. In the aggregate, these results and conclusions convey the impression that the
land price profiles a la Alonso depend more on women than on men. The typical
classic land price profile mirrors the women’s distance patterns.

LITERATURE CITED
Alonso, W. (1964). Location and Land Use: Toward a General Theory of Land Rent. Cambridge, Mass.:
Harvard University Press.
Casetti, E. (1972). “Generating Models by the Expansion Method: Application to Geographical
Research.” Geographical Analysis 4, 81-91.
(1986). “The Dual Expansion Method: An Application for Evaluating the Effects of Population
Growth on Development.” ZEEE Transactions on Systems, Man, and Cybernetics, SMC-16, January-
February, 29-39.
Choldin, H. M. (1985), Cities and Suburbs. New York: McGraw-Hill.
Durbin, R. A., and C. Sung (1987). “Spatial Variation in the Price of Housing, Rent Gradients in
Non-monocentric Cities. Urban Studies 24, 193-204.
Hansen, J. D., and G. Kristensen (1991). “Price Profiles for Land in Danish Urban Areas.” Urban
Studies 28, 277-87.
Johnston-Anumonwo, I. (1992). “The Influence of Household Type of Gender Differences in Work Trip
Distance.” Professional Geographer 44, 161-69.
Krakover, S. (1985). “Spatio-Temrral Structure of Population Growth in Urban Regions: The Cases of
Tel Aviv and Haifa, Israel.” Ur an Studies 22, 317-28.
Kristensen, Gustav (1991). “Household and Business Firm Densities in the Danish Urban Pattern.”
Geography Research Forum 11, 51-65.
Kuhl, P. H., I. Kock-Nielsen, og K. Westergtrd (1966). Fdtidsvaner i Danmark. Kebenhavn.
Lopata, H. Z. (1980). “The Chicago Woman: A Study of Patterns of Mobility and Transportation.” Signs:
]ouml of Woman in Culture and Society 5, 161-69.
Madden, J. F. (1981). “Why Women Work Closer to Home.” Urban Studies 18, 181-94.
McLaf€erty, S., and V. Preston (1991). “Gender, Race, and Commuting among Serivce Sector Workers.”
Professional Geographer 43, 1-15.
Rosenbloom, S. (1989). “Trip-Chain Behaviour: A Comparative and Cross-cultural Analysis of the
Travel Pattern of Working Mothers.” In Transport and Employment, by M. Grieco. L. Picup, and R.
Whipp-Gender. Avenbury.
White, M. J. (1977). “A Model of Residential Location Choice and Commuting by Men and Women
Workers. * Journal of Regional Science 17, 41-52.
(1986). “Sex Differences in Urban Commuting Patterns.” AEA Papers and Proceedings 76,
368-72.

You might also like