Changing Gender Roles, Shifting Power Balance and Long-distance Migration of Couples

Urban Studies, Vol. 40, No. 3, 603–613, 2003
Changing Gender Roles, Shifting Power Balance
and Long-distance Migration of Couples
Jeroen Smits, Clara H. Mulder and Pieter Hooimeijer
[Paper ?rst received, June 2002; in ?nal form, August 2002]
Summary.
Long-distance migration of couples requires joint decision-making within the house-
hold. The uneven power balance between men and women and traditional gender roles have
given rise to the concepts of ‘tied stayer’ (usually the male partner) and ‘tied mover’ (usually the
female). Since these concepts were introduced, women have gained economic power and gender
roles have changed. The paper analyses the effects of these changes on the determinants of the
long-distance migration of couples, using data for the Netherlands. For 1977, the results re?ect
the ‘classic’ pattern of family migration: the male’s human capital counted more than the
female’s and a male age advantage led to more migration. Women seemed to use their power
mostly to prevent migration. By 1996, the dominance of the male’s human capital had largely
disappeared and the effect of an age advantage had become more symmetrical. The results point
to the emergence of a new pattern, with a more equal power balance within couples.
1. Introduction
Evidence from various countries has shown
force participation; and the overall shift in
that the incidence of long-distance migration
gender roles and in the balance of power
is lower among two-earner couples than
between men and women. It seems that a
among one-earner couples (Lichter, 1982;
new pattern is on the rise in which career-
Mincer, 1978; Mulder, 1993; Wagner, 1989).
oriented women with good labour market
The classic explanation for this phenomenon
prospects not only use their power to prevent
(Mincer, 1978) is that in one-earner couples
family moves for the career of their partner,
the wife is a ‘tied mover’; she moves for the
but also use it to initiate such moves for their
sake of the husband’s career. In two-earner
own career, thereby turning the husband into
couples, the labour force participation of the
a ‘tied mover’.
wife may inhibit the move, making the hus-
This paper studies the consequences of the
band a ‘tied stayer’. However, several devel-
changes in gender roles and in the balance of
opments are affecting that pattern: the rise in
power within relationships for long-distance
earning capacity of women as a result of
migration of couples on the basis of data for
increasing educational levels and labour
the Netherlands in 1977 and in 1995/96. The
Jeroen Smits is in the Department of Economics, Nijmegen School of Management, University of Nijmegen, PO Box 9108, 6500 HK
Nijmegen, The Netherlands, Fax: 31 24 361 2379. E-mail: jeroen.smits@nsm.kun.nl. Clara H. Mulder is in the Department of
Geography and Planning, University of Amsterdam, Nieuwe Prinsengracht 130, 1018 VZ Amsterdam, The Netherlands. Fax: 31 20
525 4051. E-mail: C.Mulder@frw.uva.nl. Pieter Hooimeijer is in the Urban Research Centre Utrecht, Utrecht University, PO Box
80115, 3508 TC Utrecht, The Netherlands. Fax: 31 30 254 0604. E-mail: P.Hooimeijer@geog.uu.nl. Part of Jeroen Smits’ and Clara
Mulder’s research was undertaken while they were employed at the Urban Research Centre Utrecht of Utrecht University.
0042-0980 Print/1360-063X On-line/03/030603–11 ? 2003 The Editors of Urban Studies
DOI: 10.1080/0042098032000053941

---

604
JEROEN SMITS ET AL.
Netherlands is a very interesting country in
female. This means that, even if the educa-
this respect, because there was a spectacular
tional level of the partners is the same, the
increase in the labour force participation of
male partner will be somewhat further along
married women over a rather short time-
in his career at any point in time and hence
span. In the early 1970s, only 16 per cent of
will earn more or have higher occupational
married women were engaged in paid em-
prestige than the female partner (Markham
ployment. The traditional family type—of
and Pleck, 1986). In traditional marriages,
the breadwinner husband and the stay-at-
the male partner also tended to start the
home wife—was still very dominant. By the
marriage with a higher educational level than
mid 1990s, however, more than half of the
the female, so that, in many such marriages,
married women were gainfully employed.
the female partner suffered from a double
Over the same period, women caught up with
disadvantage.
men and even overtook them in terms of
This paper seeks to establish the extent to
educational level. People increasingly ac-
which male dominance in the long-distance
cepted that married women had a right to
migration of couples was caused by either of
pursue a career of their own (Hooghiemstra
these disadvantages. For this purpose, the
and Niphuis-Nell, 1993; Van der Lippe,
migration propensity of the more traditional
1997).
couples is compared with the migration
One would expect these changes to in-
propensity among the small number of cou-
crease the in?uence of women on the mi-
ples in which the wife is older or has a higher
gration decisions of families. However,
educational level than the husband. This will
recent ?ndings suggest that, in the Nether-
be done by adding variables indicating the
lands, most long-distance moves are still
age and educational difference between the
prompted by the career of the male partner
partners to our logistic regression models.
and still affect the female partner’s career
negatively (Smits, 1999, 2001). This classi-
2. Theoretical Background and Hypo-
cal pattern may persist because women have
theses
not yet caught up completely with men in
occupational achievement and earning ca-
Three more or less overlapping theoretical
pacity. However, it is also possible that the
perspectives can be distinguished that lead to
‘tied mover’ and ‘tied stayer’ phenomenon
somewhat contradictory hypotheses about
remains gendered. That is, women may be
how the characteristics of the partners and
unable to push through a move for the sake
their household situation affect a couple’s
of their own careers, even if their earning
propensity to move: human capital theory,
capacity is equal to or higher than that of
marital power theory and gender role theory.
their husband. To explore these dynamics,
According to human capital theory, long-
we analyse the migration decision of couples
distance migration is an investment in human
with logistic regression analysis, using char-
capital intended to generate returns in the
acteristics of both partners and their house-
form of increased income, employment
hold as explanatory variables. This will give
prospects and/or occupational status (Sjaas-
an impression of the relative importance of
tad, 1962; Blau and Duncan, 1967; Green-
each partner’s characteristics and how it has
wood, 1975). The human capital perspective
changed between 1977 and 1996.
yields a number of factors that have a posi-
An important issue to be addressed in the
tive effect on migration at the individual
analysis is the degree to which the migration
level: a low age (because wage gains de-
decision is in?uenced by age and educational
crease with age as do the remaining number
differences between the partners. To a certain
of years to recoup the costs), a high level of
extent, the dominance of the male partner in
education (because highly educated workers
long-distance migration might re?ect the fact
tend to have faster career development and
that in most couples he is older than the
need job changes to step up the career lad-

---

CHANGING GENDER ROLES
605
der) and being unemployed (because of the
pational prestige). Because the other partner
lack of returns and the depreciation of the
is dependent on him/her, this partner may
human capital of the unemployed) (Bartel
exert a disproportionately large in?uence on
and Lichtenberg, 1987; Simpson, 1992;
major family decisions, like the migration
Topel and Ward, 1992; van Ham, Mulder
decision (Blood and Wolfe, 1960; Rodman,
and Hooimeijer, 2001).
1972). In con?icts about whether or not to
Living together with a partner is expected
migrate, this power may give this partner the
to increase the costs of migration and hence
possibility of pushing through a move for
to reduce the migration propensity. This is
their own career and thus turning the other
especially true for partners who are em-
partner into a tied mover.
ployed and have to give up their job for the
Although traditionally the male partner
move. However, even though the overall spa-
was the breadwinner and brought in more
tial mobility of couples is presumably lower
income than the female partner, the predic-
than that of singles because the costs of
tion of marital power theory is in essence
moving are higher, the individual factors
gender aspeci?c. The balance of gains and
suggested by human capital theory probably
losses favours a move for the sake of the
also stimulate the migration of couples.
partner with the highest earning capacity,
When both partners’ earning capacity and
who also has more power to push through a
power balance are equal and their gender
move. When their earning capacity is about
roles non-speci?c, one would expect these
equal, on the other hand, the gains of a move
factors to contribute equally to a couple’s
are less likely to outweigh the costs and
propensity to move, regardless of whether
neither partner has the power to push through
they pertain to the male or the female part-
a move. The second prediction, which is
ner. So, the ?rst prediction is that couples in
called the ‘power-balance hypothesis’, is that
which one partner is younger, has a higher
the migration propensity of couples will be
educational level or is not employed will
low if the partners are equal with regard to
have a higher migration propensity than
such human capital factors as age, educa-
other couples. Because the effects of the
tional level or labour market situation. On
characteristics of the partners are cumulative
the other hand, if only one of the partners is
(the propensity to move is expected to be
employed or if one is clearly older (and
higher if both partners have a high educa-
hence is on average further in his/her career)
tional level than if one of the partners does),
or has less human capital, we expect the
this is termed the ‘additive hypothesis’.
couple to be more likely to move (and in
favour of the partner with more capital) than
if both partners are equal in these respects.
2.1 Differences between the Partners
The only exception is the situation in which
In addition to the direct effects of human
both partners are unemployed, because then a
capital variables on the couple’s migration
move for either partner’s career is most
propensity, the economic theory of ‘tied
likely to favour both of them. In this situ-
movers’ and ‘tied stayers’ in combination
ation, it would be expected that the couple is
with marital power theory leads us to expect
more likely to move.
that the difference in earning capacity be-
tween the partners will affect their spatial
2.2 Gender Roles
?exibility or inertia (Mincer, 1978; Boyle et
al., 1999). According to marital power the-
Whereas the additive and power balance
ory, the balance of power within conjugal
hypotheses do not differentiate between men
units lies with the partner who brings more
and women, gender role theory would predict
valued resources into the marriage (that is,
the direct effects of the characteristics of the
the partner who earns more, has a higher
female partner on the migration decision to
educational level or has a higher occu-
be weaker than those of the male partner.

---

606
JEROEN SMITS ET AL.
The reason is that women are socialised to
Dutch population aged 15 years and over not
place family ?rst and personal goals second
living in institutions. The data are gathered
when it comes to critical household matters
by oral interviews, conducted by specially
(see, for example, Markham and Pleck, 1986;
trained interviewers of Statistics Netherlands.
Shihadeh, 1991). This means that in a situ-
The general aim of the survey is to acquire
ation of equal human capital—or even when
data on the size, composition, and in?ow and
the female partner has more human capital
out?ow of the working population, and on
than the male partner—the characteristics of
the short- and medium-term developments of
the male will have a greater in?uence. Fur-
the labour market. As well as information on
thermore, when there is a difference in hu-
the situation at the time of the survey, for
man capital, an advantage possesed by the
some
characteristics
(geographical
and
male partner will lead to more migration than
labour market) information on the situation
an advantage of the female partner. This
one year before the survey is also gathered.
prediction is called the ‘male-dominance
This makes it possible to use these data for
hypothesis’.
the study of family migration.
The data-sets of the 1995 and 1996 LFS
were pooled because the number of respon-
2.3 Children
dents in the recent 1995 and 1996 LFS is
The effects of the human capital, power bal-
lower than in the 1977 LFS and because the
ance and gender factors may work out differ-
number of long-distance moves is rather
ently for couples in different circumstances.
small in the Netherlands (Smits, 1999). Be-
An important condition in this respect is the
cause the 1977 and 1995/96 ?les had to be
presence of children in the household. Cou-
analysed at different locations, it was necess-
ples without children can migrate more eas-
ary to perform separate analyses for both
ily because they face fewer restrictions. If
time-periods. For reasons of simplicity, we
there are children, more persons are involved
will speak of (data for) 1996 when referring
in the migration decision. For older children,
to the combined 1995/96 data-set.
it may be a problem to change to another
The analyses are restricted to married and
school. Moreover, they probably do not want
cohabiting persons, aged 22–59 years. The
to give up their social network and start all
lower age limit is set at 22 because the focus
over again in another place. If there are
of this paper is on career migration and
young children, the number of restrictions on
therefore it is desirable to rule out migration
the female partner is high. Even in the most
for reasons of education as much as possible.
modern societies, females still bear most re-
The upper limit is age 60 because in 1996
sponsibility for the children (Gregson and
many persons of that age had already left the
Low, 1993; van der Lippe, 1997). In addition
labour market due to retirement or disability.
to the characteristics of the partners them-
The analyses are restricted to those respon-
selves, therefore, the presence and age of
dents who formed a couple before the poten-
children will also be taken into account in the
tial move. Those who made a move to start
analyses.
living with a partner are left out of the analy-
sis to avoid confounding effects of migration
for reasons of household formation.
3. Data and Measurement
3.1 Data
3.2 Method
The data used in this paper were derived
from the Dutch Labour Force Surveys (LFS),
To determine the effects of the characteris-
which were conducted in 1977, 1995, and
tics of the partners on long-distance mi-
1996 by Statistics Netherlands. The LFS are
gration, logistic regression analysis was used.
based on large samples representative of the
The dependent variable in these analyses is a

---

CHANGING GENDER ROLES
607
dummy variable which takes the value of 1
dren including under age 6; and, only chil-
for couples who migrated over a long dis-
dren aged 6 and older.
tance in the year before the survey and a
The labour force participation of the part-
value of 0 for couples who did not move or
ners before the move was measured with a
only moved over a short distance. The dis-
four-category variable indicating whether,
tinction between long-distance migration and
one year before the survey: only the male
other forms of migration is based on the
was employed; only the female was em-
division
of
the
Netherlands
into
12
ployed; both partners were employed; or,
provinces. All moves in which at least one
both partners were unemployed. The indus-
province boundary was crossed are con-
trial sectors of the partners one year before
sidered to be long-distance moves. This oper-
the survey were indicated by four categories:
ationalisation on the basis of geographical
manufacturing and construction; agriculture;
units instead of in kilometres or miles is not
commercial services; and, non-pro?t sector.
ideal, because it will inevitably include some
Dummies were used for these variables with
short-distance moves that just straddle the
‘manufacturing and construction’ as the ref-
boundary between two adjacent provinces.
erence category. The distance over which the
However, it was necessary to use it because
partners had to commute to their work one
in the data no other comparable information
year before the survey was measured with
about the distance of migration was avail-
dummy variables indicating whether (1) or
able. This operationalisation issue will be
not (0) the place of work was in the same
discussed further in the ?nal section of the
province as the place of residence. We also
paper.
included dummy variables to indicate em-
ployed individuals with missing values for
the industrial sector and commuting distance
variables.
3.3 Independent Variables
The occupational prestige of the partners
The characteristics of the partners that are
was measured with the U&S occupational
used to explain the couple’s migration be-
prestige scale for the Netherlands (Sixma and
haviour include their age and educational
Ultee, 1983). This variable was only avail-
level and several characteristics of their work
able for the moment of the survey and hence
situation. To indicate the household situation,
only applies to males and females who were
information on the presence and age of chil-
employed at that time. We substituted miss-
dren is used.
ing values on occupational prestige with the
Because of the high correlation between
average prestige of the persons of the same
the ages of married persons, the couple’s
sex and included dummy variables to indi-
‘age’ is indicated by the age of the male.
cate the couples for whom the prestige scores
This age variable is measured in years. The
were substituted. For individuals without a
educational levels of the partners are divided
job one year before the survey, the averages
into four categories: primary education;
of the values for individuals of the same sex
lower-level
secondary
education
(Dutch
with a job were substituted in the dummies
names vbo, mavo; duration until about age
for occupational sector, commuting distance
16); upper-level secondary education (mbo,
and occupational prestige. For reasons of
havo, vwo; duration until about age 17–18);
clarity, the coef?cients of the missing value
and, tertiary education (hbo, university; dur-
dummies are not presented in the tables with
ation until about age 21–22). These variables
the results.
are entered in the form of three dummies,
To estimate the effects of age differences
with the category ‘primary education’ as the
between the spouses, we use two different
reference category. The presence of children
variables: the age difference in years (the
in the household is classi?ed according to the
male’s age minus the female’s age) and the
following three categories: no children; chil-
absolute value of the age difference in years.

---

608
JEROEN SMITS ET AL.
The variable for the absolute age difference
interprovincial moves is even somewhat
indicates the difference regardless of which
lower.
partner is older; it is denoted as the sym-
In 1977, couples in which only the male
metrical age difference. The variable for the
partner was employed tended to move the
age difference as such (which will be called
least, while (the rather small number of)
the asymmetrical age difference) indicates
couples in which only the female partner was
how many years older the male is, pointing
employed tended to move the most. Couples
to a greater importance of the male. To com-
with both partners employed and couples
pute the effects of these difference variables,
with both partners unemployed had an
separate models are estimated. We test the
above-average propensity to move. At ?rst
contribution to the baseline model of each
sight, it might seem surprising that dual-
variable separately and choose the variable
earner couples are more likely to move than
with the greatest contribution. For the effect
one-earner couples. It is hypothesised that
of an educational difference between the
this unexpected difference has to do with the
partners, two variables are also used: the
fact that dual-earner couples are over-
difference in educational level between the
represented among the young, the highly
partners (the male’s level minus the female’s
educated and the childless couples; it is
level; called the asymmetrical educational
therefore expected that it will disappear in
difference) and the absolute value of the
the multivariate analysis.
educational difference (called the symmetri-
In 1996, still about 1 per cent of couples
cal educational difference). In the same way
having only the male partner employed made
as with the age difference, the contribution to
an interprovincial move. However, this time
the baseline model of each of the educational
the proportion of long-distance migrants
difference variables is tested and the one
among the other couples is hardly higher.
contributing most is choosen.1
The couples with only the female partner
employed show the biggest change. Whereas
in 1977 almost 5 per cent of these couples
4. Results
moved interprovincially, by 1996 this had
decreased to only 1 per cent. The couples in
4.1 The Number of Long-distance Moves
which both partners are employed also
Table 1 shows that the incidence of long-
moved less in 1996 than in 1977. The ?gures
distance migration among married couples in
indicate that, in the period under study, the
the Netherlands is rather low. In 1977, only
in?uence of the female partner on the mi-
1.2 per cent of the couples in the data-set had
gration decision—and especially her like-
moved to a different province during the year
lihood
of
preventing
migration—had
before the survey. In 1996, the percentage of
increased considerably.
Table 1. Interprovincial migration in the year before the survey among couples aged
22–59 years in the Netherlands
1977
1996
Percentage
Total
Percentage
Total
migrated
N
migrated
N
Only male employed
1.0
44 934
0.9
29 976
Only female employed
4.8
944
1.0
3 643
Both employed
1.4
12 795
1.1
27 149
Both unemployed
1.4
4 102
1.3
7 592
Total
1.2
62 775
1.0
68 360

---

CHANGING GENDER ROLES
609
4.2 Logistic Regression Results
gration—although again more so when this
pertains to the male than to the female part-
Table 2 shows the results of the logistic
ner. This indicates that reducing commuting
regression analyses. The ?ndings for 1977
distance was an important reason for mi-
are very well in line with what could be
gration in the 1970s. In contrast, the presence
expected in a rather traditional society—as
of children, and especially of school-aged
the Netherlands was at that time. The male’s
children, tended to reduce the migration
human capital has the expected effects on
propensity of couples.
migration: a higher educational level and
The ?nding for 1977 that the male’s hu-
greater
occupational
prestige
lead
to
man capital and employment characteristics
signi?cantly more migration; a higher age to
have
consistently
stronger
effects
than
signi?cantly less. The male’s industrial sec-
those of the female ?ts in with the male
tor is also important, with higher migration
dominance hypothesis rather than with the
levels in commercial services and in the non-
additive
hypothesis.
However,
another
pro?t sector than in manufacturing.
?nding for 1977—that, when both partners
The educational level of the female partner
are employed, the migration propensity
has a positive effect on migration, although it
of the couple is lower than in any other
is not as strong as the effect of the male’s
situation—suggests that male dominance is
educational level. Couples in which the fe-
not prevalent in all circumstances, which is
male partner has upper secondary or tertiary
more in line with the power balance hypoth-
education migrate signi?cantly more than
esis.
couples in which the female has only primary
To test the power balance hypothesis fur-
education. However, the female’s occu-
ther, the models were re-estimated with vari-
pational prestige and industrial sector have
ables
for
the
effects
of
the
age
and
no effect on migration.
educational differences between the partners
With regard to the couple’s employment
added. This was done separately for the sym-
situation, asymmetry is seen in the effects. In
metrical and asymmetrical age and educa-
both situations where the male is not em-
tional difference effects, so that for each year
ployed (with and without employment of the
four extra models were estimated. The age
female),
the
migration
propensity
is
and educational difference parameters of
signi?cantly increased. Thus, the unemploy-
these models are presented in the lower part
ment of the male partner stimulates mi-
of Table 2. The parameters of the other
gration, regardless of the labour market
variables are not presented, because they are
situation of the female partner. When both
largely the same as in the models without age
partners are employed, the couple’s mi-
difference.
gration propensity is signi?cantly reduced.
In 1977, the asymmetrical versions of the
So, when the need to migrate is not very
educational and age difference variables con-
strong (because the male has a job), the
tributed more to the model than the sym-
female partner’s employment does reduce the
metrical versions of these variables. With
couple’s migration propensity.
regard to the educational difference effect,
This ?nding of a reduced migration
neither variable (asymmetrical version or
propensity among dual-earner couples in the
symmetrical version) is signi?cant. Thus, it
multivariate analysis seems to con?rm the
seems that educational differences between
expectation that the increased migration
the partners do not contribute very much to
propensity among these couples found in the
the explanation of the couple’s migration
bivariate analysis was caused by the overrep-
propensity over and above what is already
resentation of the young, highly educated and
explained by the main educational effects.
childless couples among them.
For age, the difference effect is quite sub-
Employment of either partner outside the
stantial in 1977. The coef?cient of the
province has a strong positive effect on mi-
asymmetrical age difference variable is

---

610
JEROEN SMITS ET AL.
Table 2. Logistic regression coef?cients (B) for the effects of selected characteristics of the male and
female partner on interprovincial migration of couples in the Netherlands, and differences between the
coef?cients
1977
1995/96
1995/96–1977
B
SE(B)
B
SE(B)
Differenceb
Constant
5.141***
0.404
3.588***
0.398
1.553***
Age male
0.049***
0.006
0.073***
0.006
0.024***
Education male
Primary
Reference category
Reference category
Lower secondary
0.478***
0.139
0.225
0.195
0.703***
Upper secondary
0.801***
0.145
0.088
0.178
0.713***
Tertiary
0.905***
0.166
0.549***
0.193
0.356
Education female
Primary
Reference category
Reference category
Lower secondary
0.085
0.109
0.118
0.194
0.033
Upper secondary
0.468***
0.127
0.526***
0.185
0.058
Tertiary
0.474***
0.166
0.724***
0.202
0.250
Employment situation couple
Only male employed
Reference category
Reference category
Only female employed
1.401***
0.199
0.036
0.193
1.437***
Both employed
0.578***
0.134
0.533***
0.103
0.045
Both unemployed
1.381***
0.155
0.882***
0.132
0.499**
Occupational prestige malea
0.014***
0.003
0.004
0.003
0.010**
Occupational prestige femalea
0.008
0.006
0.000
0.005
0.008
Industrial sector malea
Manufacturing/construction
Reference category
Reference category
Agriculture
0.194
0.287
0.245
0.306
0.051
Commercial services
0.570***
0.106
0.341***
0.117
0.229
Non-pro?t sector
0.532***
0.118
0.327**
0.133
0.205
Industrial sector femalea
Manufacturing/construction
Reference category
Reference category
Agriculture
0.384
1.043
0.617
0.479
1.001
Commercial services
0.107
0.239
0.161
0.215
0.054
Non-pro?t sector
0.218
0.233
0.124
0.214
0.342
Workplace male outside provincea
2.259***
0.093
1.968***
0.101
0.291**
Workplace female outside provincea
1.175***
0.230
1.052***
0.131
0.123
Children
None
Reference category
Reference category
Youngest
6
0.211**
0.107
0.350***
0.093
0.139
Youngest
6
0.670***
0.126
0.541***
0.128
0.129
Difference effects (separate models)
Asymmetrical age difference
0.032***
0.012
0.025**
0.012
0.007
Symmetrical age difference
0.027
0.015
0.033**
0.014
0.006
Asymmetrical educational difference
0.166
0.140
0.081
0.115
0.085
Symmetrical educational difference
0.064
0.057
0.071
0.058
0.007
N
62 775
68 360
N migrated
723
672
2 Log likelihood
6 443
6 312
Model chi-squared/DF
1 449/28
1 238/28
aThe dummy variables for cases with missing values on these variables are not presented.
bSigni?cance of the differences between the years is tested with t-tests for independent samples.
***indicates signi?cant at the 1 per cent level; **indicates signi?cant at the 5 per cent level.

---

CHANGING GENDER ROLES
611
signi?cantly positive, whereas the coef?cient
gration propensity. However, this conclusion
of the symmetrical age difference variable is
is not completely unequivocal, for the
not. This means that there is an asymmetrical
coef?cient of the symmetrical variable is
effect to the advantage of the male partner:
signi?cant too. It is therefore concluded that
the more years the male is older than the
there is indeed a trend towards more equal-
female, the higher the propensity of the cou-
ity, but that by 1996 the situation of male
ple to move. This result is more in line with
dominance has not yet completely disap-
the male dominance hypothesis than with the
peared.
power balance hypothesis.
The picture of male dominance that arises
5. Conclusions
from the results for 1977 is hardly discern-
ible in 1996. For education, occupational
The migration propensity of couples in the
prestige
and
employment
situation,
no
Netherlands has been analysed from a per-
stronger effect was found for males than for
spective of gender differences in human
females. Only for working outside the prov-
capital, marital power and gender roles.
ince of residence and industrial sector does a
Compared with previous research, this paper
stronger effect for the males remain.
contributes to the literature by making a
With regard to education and employment
comparison between 1977, when gender
situation, the changes are rather striking. The
roles were quite traditional and the share of
effect of the male’s education on the couple’s
married women in the workforce was very
migration
propensity
has
decreased
low, and 1996, when the number of two-
signi?cantly over time (as is clear from the
earner couples was much higher and the
difference parameters in the last column of
Netherlands had become a much more mod-
Table 2) and is no stronger than the effect of
ern society. From human capital theory,
the female’s education. Furthermore, the
power balance theory and gender role theory,
large difference in migration propensity be-
three alternative hypotheses were derived:
tween couples in which only the female part-
the additive hypothesis (stating that the
ner was employed and couples with only the
in?uence of the partner’s labour market re-
male partner employed has completely disap-
sources is cumulative); the power balance
peared by 1996.
hypothesis (stating that it is the difference in
At the same time, the negative effect of
resources that counts, regardless of whether
dual-earnership
on
migration
has
not
they belong to the male or the female); and,
changed at all in the period under study. As
the male dominance hypothesis (stating that
in 1977, in 1996 dual-earner couples were
the male’s resources count more than the
less likely to migrate than couples with any
female’s).
other pattern of labour market participation.
For 1977, the results are mostly in line
After controlling for the other explanatory
with the male dominance hypothesis. The
variables,
dual-earner
couples
are
only
human capital factors of the male partner
exp(
0.533)
0.59 times as likely to mi-
were more important at that time than the
grate as couples with only a male earner.
human capital factors of the female partner.
The parameters for the difference effects
The female’s employment situation only
also suggest that the dominance of the male
played a role when the need for migration
partner has decreased over time. The educa-
was not very high because the male was
tional difference variables are again not
employed. Only in that case did the female’s
signi?cant in 1996. However, both the asym-
employment reduce the couple’s migration
metrical and the symmetrical age difference
propensity. The effect of an age difference in
effects are signi?cantly positive. The positive
1977 was also in line with the male domi-
effect of the asymmetrical age difference
nance hypothesis: the more years the male
variable would indicate that a larger age
was older than the female, the higher the
advantage of either partner increases the mi-
probability of migration. These ?ndings sug-

---

612
JEROEN SMITS ET AL.
gest that married women who were older
likely to migrate than their one-earner coun-
than their husband used their power to pre-
terparts. There are no signs that the effect of
vent migration, instead of using it to migrate
dual earnership is changing—the estimated
for their own careers.
effect is about the same for 1996 as for 1977.
By 1996, the picture had changed consid-
Combined with the increase in dual earner-
erably. For education, occupational prestige
ship that has occurred in the Netherlands,
and employment situation the effect was not
this result indicates that, over time, more and
stronger for the male than for the female. The
more individuals have become restricted in
most remarkable difference, however, was
their migration possibilities because of the
that couples in which only the female was
presence of a working partner. Our ?ndings
employed did not migrate any more fre-
also suggest that the ‘tied mover’ phenom-
quently than couples in which only the male
enon has become less gendered and that
was employed. The results for 1996 seem to
males with less human capital than their
be mostly in line with the prediction of the
partners are becoming ‘tied movers’. How-
power balance hypothesis that an unequal
ever, the likelihood of becoming a ‘tied
balance in earning capacity will contribute to
stayer’ seems to have increased for both
the couple’s migration propensity, regardless
males and females, because of the greater
of which partner has this higher capacity.
equality in their relationship.
Support for this hypothesis is also given by
One last point should be made in interpret-
the ?nding that the effect of the age differ-
ing the ?ndings of this study. As was stated
ence has become more symmetrical over
in the method section, it is possible that the
time. In other words, any age difference—
operationalisation of long-distance migration
regardless of which partner was older—had
on the basis of geographical units instead of
more predictive power than an age advantage
distances has introduced bias in the results.
ascribed to one of the partners. However,
Some of the interprovincial moves may in
with respect to the effect of an age differ-
fact have been short-distance moves for
ence, the dominance of the male partner,
housing reasons of people living close to the
which was found for 1977, has not yet disap-
provincial border. No information exists con-
peared completely. The asymmetrical age
cerning the extent of this bias. However, it is
difference variable (age of male minus age of
known that the effects of education and
female) also had a signi?cantly positive ef-
labour
market
characteristics
on
short-
fect in 1996 (in a separate analysis). This
distance migration in the Netherlands are
somewhat contradictory result suggests that,
weaker than their effects on long-distance
when the male is older, the positive effect of
migration (Mulder, 1993). A possible bias,
an age difference on migration is stronger
therefore, is most likely to have led to an
than when the female is older.
underestimation of the true effects of these
The analyses show that in both periods
characteristics.
reducing the commuting distance of the part-
ners was an important reason for long-
Note
distance migration in the Netherlands. When
one of the partners was employed outside the
1.
We would have preferred to use more (and
province of residence, the couple’s propen-
more sophisticated) measures of power bal-
ance—for example, the number of hours
sity to migrate was considerably greater. This
worked by each partner or the difference in
effect was strongest if the male had a long
occupational prestige. However, this was not
commuting distance, but it was also quite
feasible. Information about the number of
substantial if the female had a long journey
hours worked was not available for the year
to work.
before the survey. In the 1977 data, the share
of women with known occupational prestige
The results of the multivariate analyses
was too small to allow inclusion of both the
con?rm earlier ?ndings that, ceteris paribus,
prestige itself and the difference between the
dual-earner couples and families are less
male’s and the female’s prestige.

---