The Multifaceted Impact of Education on
Entry into Motherhood
TRUDE LAPPEGA° RD* and MARIT RNSEN
Division for Social and Demographic Research, Statistics, PO Box 8131 NO - 0033, Oslo,
Norway (*Author for correspondence, E-mail: lap@ssb.no)
Received 28 June 2004; accepted in final form 15 September 2004
Lappega° rd T. and Rnsen M., 2005, The multifaceted impact of education on entry into
motherhood, European Journal of Population 21: 31­49.
Abstract. This article studies the composite effect of education on young women's entry into
motherhood, using longitudinal data from Norway from 1971 to 2001. In line with previous
research, we find that school enrolment delays motherhood, but having finished education
there is a catching-up effect, as women who have completed at higher levels have their first
child sooner than women who have completed at lower levels. Contrasting behaviour between
women within various fields of education further indicate a career-adjustment effect related to
differences in opportunity costs and/or preference heterogeneity. Finally, increasing educational
differences in the timing of motherhood among younger cohorts suggest that long
parental leaves and generous family benefits may fit better with a career track in some jobs
than others.
Key words: fertility, first-birth, education, multivariate hazard model
Lappega° rd T. et Rnsen M., 2005, Les multiples facettes de ĺeffet de ĺinstruction sur la
maternité , Revue Europé enne de Dé mographie, 21: 31­49.
Ré sumé . Cet article é tudie ĺeffet composite de ĺinstruction sur ĺentré e en maternité des jeunes
femmes, en s'appuyant sur des donné es longitudinales norvé giennes couvrant la pé riode 1971­
2001. Dans la ligne de pré cé dentes recherches, nous trouvons que la poursuite des é tudes
retarde la maternité . En revanche, de` s ĺarre^ t des é tudes, il y a un effet de rattrapage et les
femmes qui ont atteint des niveaux ďinstruction plus é levé s ont leur premier enfant plus tô t
que celles qui ont fini leurs é tudes a` des niveaux plus bas. Selon les filie` res suivies, les femmes
n'ont pas les me^ mes comportements, ré vé lant un effet ďajustement de la carrie` re dé pendant
des cou^ ts et/ou de pré fé rences hé té roge` nes. Enfin, pour les plus jeunes gé né rations, ĺaccroissement
des diffé rences dans le calendrier des naissances selon le niveau ďinstruction tend
a` prouver que les congé s parentaux et des prestations familiales consé quentes conviennent
davantage a` certains cursus de carrie` re professionnelle qu'a` ďautres.
Mots clé s: fé condité , premie` re naissance, instruction, mode` le de risque multivarié
European Journal of Population (2005) 21: 31­49 Ó Springer 2005
DOI 10.1007/s10680-004-6756-9
1. Introduction
During the last decades Norway and other western countries have witnessed
a pronounced trend in the postponement of entry into motherhood. The
mean age at first birth among women in Norway was 27.9 years in 2003,
which is 2.7 years higher than it was 15 years ago (Statistics Norway, 2004a).
At the same time there has been a rapid educational expansion, during which
the proportion of Norwegian women with a university degree has about
doubled, from 11.4% in 1988 to 22.1% in 2002 (Statistics Norway, 2004b).
The connection between education and timing of first birth has been widely
analysed, and it is a well known descriptive finding that women with higher
education are older at first birth than those with lower education (see e.g.
United Nations 1996­2001, Table 17). When analysed in an event-history
framework, however, there is less consistent evidence on the postponing
effect of higher education on the entry into motherhood.
In much previous research, information on education has only been
available at the time of interview or data extraction, which may be from a life
stage far beyond the age period preceding first birth when fertility plans are
presumably made. Used in this way, education may at best be regarded as a
proxy for educational goals and strategies that are formed earlier in life, and
as such assumed to be exogenous to the fertility process. At worst, the results
may be biased, as has recently been demonstrated by Kravdal (2004) for
Norway. This is because the causality is also likely to run the other way, i.e.
fertility may have affected the educational level obtained later in life.
In event-history models where education is as a fixed covariate measured
at some point after childbirth, the results corroborate the findings of
descriptive statistics: women with higher education postpone motherhood
longer than women with lower education. When longitudinal information on
education is available and education is treated as a time-varying variable,
however, the results are less uniform. This is because education as a timevarying
covariate is very highly correlated with age and other life events
closely linked with age (see e.g. Liefbroer and Corijn, 1999; Santow and
Bracher, 2001). The results are thus quite sensitive to the specification of the
model. When school enrolment is controlled for in such models, the effect of
educational level is sometimes even estimated to be positive (Blossfeld and
Huinink, 1991; Kreyenfeld, 2000; Hank, 2002). Educational activity, on the
other hand, is consistently found to postpone motherhood (e.g. Hoem, 1986;
Blossfeld and Huinink, 1991; Kravdal, 1994; Blossfeld, 1995; Liefbroer and
Corijn, 1999; Hoem, 2000; Andersson, 2000; Hank, 2002).
In addition to educational level and educational activity, some studies
have also shown that the field of education is important for women's fertility
behaviour (Hoem, 1994; Kalmijn, 1996; Lappega° rd, 2002). In this study we
take all three aspects of education into account when analysing entry into
TRUDE LAPPEGA° RD AND MARIT RNSEN32
motherhood. We hypothesize that timing of first birth among young women
are dominated by their current educational activity and their labour market
aspirations. Since most Norwegian women today return to work when their
youngest child is quite small (Rnsen and Sundstro¨ m, 2002), it has become
increasingly important to get a good foothold in the labour market before
becoming a mother. How much time young women need in order to get
established in a job and settled on a career track after completing education
will again depend on where in the labour market they are headed, which is
probably closely linked with both level and type of education. Information
on the field of study should thus give added insights into the multifaceted
impact of education on entry into motherhood.
Our analysis is based on longitudinal data on fertility, educational activity,
and level and field of education from administrative registers covering the
whole population of Norway. The large amount of data makes it possible to
construct several educational categories along all three dimensions (activity,
level and field), rendering detailed information about educational differentials
in the timing of first birth. The long time series of individual data further
provide us with complete and real time-varying variables for education. This is
an important improvement relative to imputing educational level and enrolment
based on standard progress through the educational system, which is
sometimes the only resort if the educational histories are incomplete. As shown
in Kravdal (2004), the latter approach may give quite misleading results.
The rest of the study is organized as follows: in the following section we
introduce the theoretical framework for the analysis and discuss the main
hypotheses to be tested. Next, we give a brief description of the development
of fertility and education in Norway during the last couple of decades, followed
by a closer presentation of the data and the empirical model. Finally,
we report the main findings and conclude with a short summary and dis-
cussion.
2. Theoretical Framework
When a traditional family structure with strict gender-specific division of
labour is challenged in modern societies, the reconciliation of childbearing
and female employment becomes an urgent issue. The timing of first birth is
clearly an important consideration in this respect, as the economic loss (the
opportunity cost) of taking a break from the labour market constitutes a
large part of the costs involved in having a child (Joshi, 1990; Kravdal, 1994;
Walker, 1995). It is generally believed that the incompatibility of work and
family is higher among highly educated than among other women, as they
have more to lose in terms of foregone earnings than women with lower
education. Theoretical research on fertility did not originally address the
THE MULTIFACETED IMPACT OF EDUCATION 33
timing of fertility, however, but rather completed family size (Easterlin, 1966;
Becker, 1981, 1991; Pollak and Watkins, 1993). Later, authors have also
developed models that more explicitly address the timing and spacing of
fertility (for a summary: see e.g. Gustafsson, 2001). Taking all other things
equal, the optimal time of birth is then the one that maximizes the wife's
lifetime earnings, and included in the considerations of foregone earnings are
not only the current loss of wages during a career break, but also future losses
due to the lack of human capital accumulation and depreciation of job skills
(Happel et al., 1984; Cigno and Ermisch, 1989; Walker, 1995). An important
determinant of the timing of fertility is thus the woman's life-cycle earnings
profile, depending among others on her initial human capital accumulation
and the profile of further investments, the rate of return to these investments,
and the rate at which her job skills decay (Gustafsson, 2001).
Since the opportunity cost of leaving the labour force are particularly high
for better educated women, it can be argued that they might postpone
motherhood to a later stage in their employment career, when they consider
themselves more established in a career-track and when taking a break from
the labour market may be less damaging to their future labour market career
(Kreyenfeld, 2000). On the other hand, if the lifetime earnings profile of
highly educated women is relatively steep, it may be less costly to have the
child earlier in the career rather than later (Walker, 1995). The predictions of
dynamic fertility models are thus not entirely unambiguous. Furthermore,
since highly educated women are older upon leaving the educational system,
they have a shorter time left of their reproductive period. They may therefore
want to catch up on childbearing to fulfil their fertility plans. Blossfeld and
Huinink (1991) argue e.g. that both increasing medical problems connected
with late births and societal age norms may induce highly educated women to
catch up on childbearing when they have finished their education.
Based on the above reasoning and empirical evidence, we shall argue that
the postponing effect on first-birth rates of a higher level of education mainly
operates through prolonged participation in the educational system. Having
completed education, however, differences in opportunity costs may first and
foremost be reflected through different fields of education that lead to different
occupations and employment sectors. These jobs may be associated
with working conditions that are more or less compatible with childbearing
and childrearing, as well as with differences in earnings and career prospects
that imply that the optimal time to have a child and take a break are assessed
differently.
Field of education may, however, also convey differences in preferences
and priorities, independent of economic resources and opportunity costs.
Hakim (2000, 2003) argues, for example, that heterogeneous preferences and
priorities lead to a polarization of work-lifestyles and family models. On the
one side there are work-centred women who are committed to work or
TRUDE LAPPEGA° RD AND MARIT RNSEN34
equivalent activities, while on the other side there are home-centred women
who throughout life give priority to family and children. Between these two
groups there is a larger group of adaptive women who prefer to combine
employment and family without giving a fixed priority to either (Hakim,
2000). Since heterogeneous work-life strategies are likely to lead to differential
occupational choices, we must assume that field of education also
reflects preference heterogeneity.
In comparative research on family formation, the delaying influence of
educational attainment has been found to vary across countries (Blossfeld,
1995). This is contributed to differences between countries in so-called `family
systems' which include both cultural values, family and religious traditions
and family policies, and it is argued that the negative effect of educational
attainment on family formation will be stronger in societies in which the
incompatibility between female employment and family formation is larger
than in societies in which the incompatibility is small (Blossfeld, 1995).
Liefbroer and Corijn (1999) emphasize that the compatibility of female
productive and reproductive work has both a cultural and a structural
dimension, where the cultural dimension is related to ideology, values and
norms concerning the role of women on society, while the structural
dimension is related to actual societal opportunities and constraints on the
role of women. They further argue that family systems do not only differ
between countries, but also change within countries, as the incompatibility
between family life and female labour market commitments has weakened in
many countries throughout the last decades. Thus the impact of educational
attainment on family formation can be expected to be weaker for younger
cohorts than for older ones (Liefbroer and Corijn, 1999).
Based on the above discussion and previous research we outline four
hypotheses to be tested in our analysis. The first hypothesis can be referred to
as the student-effect-hypothesis:
H1: Women who are still enrolled in education will have lower first-birth
rates than women who are no longer enrolled.
The argument for this hypothesis is that school enrolment and childbirth is
especially incompatible, either for practical reasons1
or as a result of normative
views that students should not become parents before they have finished
education. Students may also have aspirations for a future career-track
that they want to fulfil before they become mothers. After finishing school,
however, we assume that women with higher education will give birth sooner
than other women who are no longer students. Thus we formulate additionally
the catching-up-hypothesis:
H2: Women with higher levels of education will have higher first-birth rates
upon finishing school than women with lower education.
THE MULTIFACETED IMPACT OF EDUCATION 35
The argument for this hypothesis is that women with higher education will
have stayed longer in school and therefore have a shorter time left of their
reproductive period. Further, we argue that different types of higher education
will give different opportunities, status and remuneration in the labour
market, which implies that the cost of a career break in connection with
childbirth will differ. This motivates the career-adjustment-hypothesis:
H3: Women within different fields of education will have different first-birth
rates, regardless of educational level.
Women within fields where it takes longer to get established on a career
track and where the costs of a withdrawal from the labour market are higher
will thus postpone motherhood more than women within other fields. In
addition, differences in postponement may indicate that women with certain
types of education have different family and fertility preferences than other
women.
Finally, we propose the cohort-effect-hypothesis:
H4: Educational differences in the timing of first birth have become smaller
among women in younger cohorts.
This hypothesis is based on the underlying assumption that the incompatibility
between family life and female labour market commitments has
become weaker, and that the negative effect of educational attainment on
family formation has thereby weakened.
3. The Norwegian Setting
At the beginning of the 1970s the total fertility rate (TFR) in Norway was
well above the replacement level with 2.5 children per woman. By 1980 it
had fallen to 1.72 and by 1983 it reached an all-time low of 1.66. In
contrast to most other industrialized nations, Norway and other Nordic
countries experienced a rise in fertility from the mid 1980s, and at the
beginning of the 1990s the TFR in Norway was again close to the
replacement level (1.93). Since then it has declined slightly and stabilized
around a level of about 1.8.
The last couple of decades have also witnessed a vast educational
expansion, with a larger increase in the proportion that has completed college
or university among women than among men (Table 1). From 1980 to 2000
the percentage of women aged 16 and older with education at this level more
than doubled, from 9.3 to 21.8% (1st and 2nd stage higher education in all),
while the corresponding percentage among men increased from 13.1 to
21.9%. Thus, in Norway today, the proportion with higher education is
practically the same for women and men. There are still more men than
TRUDE LAPPEGA° RD AND MARIT RNSEN36
women with an upper (2nd stage) university education, however. But, since
these numbers reflect the proportion among all adults, it partly conceals the
educational revolution that is happening among younger birth cohorts. If we
look at the percentage of 19­24 year olds registered in higher education,
there has been a spectacular increase especially among women, whose
enrolment rate more than tripled from 1980 to 2000. Female enrolment
surpassed that among men already in 1990, and at the turn of the century
almost one out of three women aged 19­24 years were registered in higher
education, compared to 21.6% among men.
In tandem with the educational expansion, there has also been a strong
postponement of motherhood. The aggregate statistics in Table 1 show that
women at all educational levels have delayed childbearing over time. For
example, among women with primary and lower secondary education the
mean age at first birth was 24.0 years in 1998, an increase of 0.7 years since
1980. Among women with a lower and upper university education the mean
age in 1998 was 29.5 and 31.6 years, respectively. In both university groups
Table 1. The percentage 16 years and older that completed higher education and percentage
19­24 years registered in higher education, men and women. Total fertility rate and mean age
at first birth, women. 1980, 1990, 2000
1980 1990 2000
Percentage 16 years and older that completed higher
education, 1st stage
Men 8.9 11.6 15.3
Women 8.6 12.6 19.0
Percentage 16 years and older that completed higher
education, 2nd stage
Men 4.2 5.0 6.6
Women 0.7 1.3 2.8
Percentage 19­24 years registered in higher education
Men 11.8 16.2 21.6
Women 9.5 20.0 31.4
Total fertility rate
Per woman 1.72 1.93 1.85
Mean age$
at first birth ­ women
Primary and lower secondary 23.3 23.9 24.0#
Upper secondary 24.0 25.3 26.4#
University, 1st stage 26.5 28.1 29.5#
University, 2nd stage 28.5 30.0 31.6#
Source: Statistics Norway.
$
Age measured at the end of the year.
#
Figures from 1998.
THE MULTIFACETED IMPACT OF EDUCATION 37
the mean age had increased with 3 years since 1980. The educational differences
in the mean age at first birth have thus increased over time (for
further details see Lappega° rd, 2000; Rnsen, 2004).
4. Data, Methods and Classification
4.1. DATA
Our analyses are based on longitudinal data from the Norwegian Central
Population Register and the Norwegian Educational Database. The data
have been linked in order to obtain complete fertility and educational histories
for all females born 1955­1984, recorded on a monthly basis from
1971­2001. The analyses are restricted to women living in Norway at the end
of the year 2001, numbering 827,494 in total. We excluded women with an
immigrant background in order to get as homogeneous data as possible and
also because information on education is largely missing for immigrants.2
4.2. METHODS
The analyses are based on a discrete hazard rate model. In discrete time, the
hazard rate is the conditional probability that an event (in our case, the birth
of the first child) will occur at a particular time to a particular individual
given that the individual has not experienced the event before (see e.g. Allison,
1984). Besides depending on the current age, the hazard rate is assumed
to vary with education and other personal characteristics. Using a logit
transformation, the discrete hazard rate function can be expressed as
logPt=1  Pt  bXt 1
where Pt is the conditional probability that a birth occurs at time t, 1)Pt is
the probability that no birth occurs at time t, b is a vector of coefficients, and
Xt is a vector of covariates that may or may not vary with time. We follow
the women from age 16, until entry into motherhood or to the end of 2001, if
no birth occurs. The data have events dated by month, but we use quarters in
the model to reduce the otherwise vast number of records obtained (person-
months).
The problem with education as a determinant of first birth is that the
educational process is linked so closely to the birth process. It may therefore
be preferable to model the two processes simultaneously, but as this is a more
complex estimation procedure, we did not attempt to do so. Our approach
has been to formulate a model that primarily reflects the manifold dimensions
of education as discussed above. To highlight the close connection
between the educational activity and the educational level we have first
TRUDE LAPPEGA° RD AND MARIT RNSEN38
combined the two into a single variable (see Table 2). Then we have split
both educational activity and educational level into different types or fields of
education, as described in more detail below. To address the various aspects
Table 2. Educational effects on first-birth rates of Norwegian women born 1955­1984.
Discrete hazard model (odds ratio estimates). Model A
Age
16­20 1
21­25 1.83
26­30 2.17
31­35 1.37
36­40 0.53
41­45 0.09
Birth cohorts
1955­1959 1
1960­1964 0.90
1965­1969 0.89
1970­1974 0.81
1975­1979 0.64
1980­1984 0.48
Education
In education 0.36
Not in education: Primary and lower secondary 1
Not in education: Upper secondary 1.05
Not in education: University, 1st stage 1.20
Not in education: University, 2nd stage 1.42
Social background (father's and/or mother's education)
Low 1
Medium 0.88
High 0.67
Unknown 0.77
Regional background
Oslo and surrounding country 1
Eastern 1.17
South Eastern 1.23
South 1.45
Western 1.38
Middle 1.50
Northern 1.47
Unknown 1.15
THE MULTIFACETED IMPACT OF EDUCATION 39
of education, we present three versions of the model in which activity, level
and field have been aggregated in different ways (Models A­C).
4.3. CLASSIFICATION
The large amount of data enables us to divide education along many
dimensions according to educational activity and type, and completed educational
level and field. Educational level and field are classified using the
Norwegian standard classification of education (Statistics Norway, 2001).
The level is divided into four main groups:
1. Primary and lower secondary (­9 years)
2. Upper secondary (10­12 years)
3. University, 1st stage (13­16 years)
4. University, 2nd stage (17 years and more)
Combining level with different fields of education, we get 18 groups:
1. Primary and lower secondary
2. Upper secondary
3. University: humanities and aesthetics 1st stage
4. University: humanities and aesthetics 2nd stage (e.g. languages, history,
musicians, pictorial artists)
5. University: teaching, 1st stage (e.g. pre-school teaching, primary school
teaching)
6. University: teaching, 2nd stage (e.g. secondary school teaching)
7. University: social science, 1st stage (e.g. journalism)
8. University: social science, 2nd stage (e.g. psychology, sociology)
9. University: law, 2nd stage
10. University: administration and economics, 1st stage
11. University: administration and economics, 2nd stage
12. University: engineering, 1st stage
13. University: engineering, 2nd stage (e.g. civil engineering)
14. University: nursing, 1st and 2nd stage
15. University: physicians, 2nd stage
16. University: health care otherwise, 1st stage (e.g. welfare nursing, physio-
therapists)
17. University: health care otherwise, 2nd stage (e.g. dentists, pharmacists)
18. University: others/missing
Educational activity in its simplest form is just a dummy variable that
equals one if the woman is registered as a student. In addition, we have
divided those who are in education into seven different groups according to
TRUDE LAPPEGA° RD AND MARIT RNSEN40
type of study, based on the study-code in the Norwegian Educational
Database:
1. Primary and lower secondary
2. Upper secondary: academic
3. Upper secondary: vocational
4. University, 1st stage: female dominated fields (e.g. pre-school teaching,
primary school teaching, nursing)
5. University, 1st stage: male dominated fields (e.g. engineering, administration
and economics)
6. University, 1st stage: others
7. University, 2nd stage
Using register data, the availability of other fertility determinants is limited.
However, in addition to age, education and birth cohort, we also control for
social and regional background.
Age is a time-varying covariate (and baseline hazard), categorized into
intervals as follows: 16­20, 21­25, 26­30, 31­35, 36­40, and 41­46.
Birth cohorts are collapsed into 5-year groups, consisting of women born
1955­1959, 1960­1964, 1965­1969, 1970­1974, 1975­1979, and 1980­1984.
Social background is based on information on the parents' level of education,
divided into low (primary and lower secondary), medium (upper
secondary) and high (university).
Regional background is the women's residential region at age 16 defined as
Oslo and surrounding country (Oslo and Akershus), Eastern (Hedmark and
Oppland), South Eastern (stfold, Vestfold, Buskerud and Telemark), South
(Agder and Rogaland), Western (Hordaland, Sogn og Fjordane and Mre og
Romsdal), Middle (Trndelag), and Northern (Nordland, Troms and
Finmark).
5. Results
5.1. THE EFFECTS OF BEING A STUDENT
Not surprisingly, and in full accordance with previous research, student
enrolment is found to have a negative effect on first-birth rates (Table 2).
Thus our analysis further corroborates the student-effect-hypothesis (H1)
that childless women who are still in school will have lower first-birth rates
than other women.
There are, however, some noteworthy differences between various types of
educational activity (Table 3). The effect of being enrolled is still negative for
all groups, but there are significant differences in how inhibiting the student
role seems to be for childbearing. Among women in upper secondary
THE MULTIFACETED IMPACT OF EDUCATION 41
education the negative effect is much less pronounced within the vocational
than within the academic fields. This is probably due to higher aspirations for
future education in the latter group, as women in academic fields are more
likely to continue with university studies than women within vocational
fields. Thus they will have more to lose by having a baby while still enrolled
in upper secondary education.
At the university level the most pronounced differences are between those
in male versus those in female dominated fields, where the former group is
found to have the lowest first-birth rates. This may be related to greater
difficulties in combining studies and motherhood in male dominated fields,
but it may also be related to their future career aspirations. Women in male
dominated fields may, for example, feel a greater need to get a good foothold
on the labour market and get more established in their careers before they
become mothers.
5.2. THE EFFECTS OF LEVEL AND FIELD OF EDUCATION
The results in Table 3 show that university educated women who are no
longer enrolled and who have completed at the upper level (2nd stage) have
higher first-birth rates than non-enrolled women who completed at the lower
level (1st stage). Those who have postponed motherhood longer thus give
birth sooner, which support the catching-up-effect-hypothesis (H2) that
childless women who stay longer in school recuperate childbearing faster
Table 3. Educational effects* on first-birth rates of Norwegian women born 1955­1984.
Discrete hazard model (odds ratio estimates). Model B
Education
In education:
Primary and lower secondary
0.12
In education: Upper secondary: academic 0.16
In education: Upper secondary: vocational 0.47
In education: University, 1st stage: female dominated fields 0.48
In education: University, 1st stage: male dominated fields 0.24
In education: University, 1st stage: others 0.39
In education: University, 2nd stage 0.40
Not in education:
Primary and lower secondary
1
Not in education: Upper secondary 1.08
Not in education: University, 1st stage 1.24
Not in education: University, 2nd stage 1.46
*Controlled for birth cohort, age and social and regional background.
TRUDE LAPPEGA° RD AND MARIT RNSEN42
upon finishing education. Further, in Table 4 there are some noticeable
contrasts between women within different fields that support the careeradjustment-hypothesis
(H3).
Women with a lower university degree in humanities/aesthetics (e.g.
musicians, artists, actors) and social sciences (e.g. journalists), in particular,
are found to have relatively low first-birth rates. Other university groups
with relatively low first-birth rates are engineering 1st stage, and administration
and economics. The relatively low first-birth rates of women within
humanities and aesthetics can probably be related to their labour market
situation, as musicians, dancers, actors and pictorial artists often work
freelance and have loser ties to the labour market than women in other
occupations. Lower first-birth rates may then indicate that the greater
uncertainty connected with a possible job break is an obstacle for having a
baby. However, low first-birth rates in these and other groups (e.g. engineering,
administration and economics) may also indicate that women
within these fields constitute a select group who are less family oriented and
Table 4. Educational effects* on first-birth rates of Norwegian women born 1955­1984.
Discrete hazard model (odds ratio estimates). Model C
Education
In education 0.36
Not in education (NIE): Primary and lower secondary 1
NIE: Upper secondary 1.05
NIE: University: Humanities and Aesthetics, 1st stage 0.79
NIE: University: Humanities and Aesthetics, 2nd stage 1.25
NIE: University: Teaching, 1st stage 1.45
NIE: University: Teaching, 2nd stage 1.57
NIE: University: Social science, 1st stage 0.82
NIE: University: Social science, 2nd stage 1.51
NIE: University: Law, 2nd stage 1.23
NIE: University: Administration and Economics, 1st stage 1.13
NIE: University: Administration and Economics, 2nd stage 1.01
NIE: University: Engineering, 1st stage 1.05
NIE: University: Engineering, 2nd stage 1.33
NIE: University: Nursing, 1st and 2nd stage 1.54
NIE: University: Social work, 1st and 2nd stage 1.24
NIE: University: Physicians, 2nd stage 1.71
NIE: University: Health care otherwise, 1st stage 1.29
NIE: University: Health care otherwise, 2nd stage 1.54
NIE: University: Others and missing 1.01
*Controlled for birth cohort, age and social and regional background. Numbers in
italics = not significant.
THE MULTIFACETED IMPACT OF EDUCATION 43
more work oriented at the outset, and that these preferences guide both
educational and family and fertility choices. This corroborates Hakim's
argument (2000, 2003) that differences in preferences deserve a more
prominent place in causal explanations of female fertility and employment
patterns.
The highest first-birth rates are found among women educated as physicians,
followed by nurses, other healthcare workers, teachers and women
with an upper social science degree. Most of these fields of education qualify
for occupations that are involved with care for other people, but there are
differences in work prestige related to the occupations. The fact that physicians
have the highest first-birth rates show that a high family-orientation can
also be found in occupations that generally are related to high work-orientation
and high costs of withdrawal from the labour market. These findings
further suggest that there is no clear-cut relationship between high costs of
labour market withdrawal and postponed motherhood, and that preference
heterogeneity also plays a role in the differential timing of first birth.
5.3. COHORT TRENDS
In order to examine possible changes in educational effects across successive
female cohorts we estimated the model separately for three cohorts: 1955­1959,
1960­1964 and 1965­1969.3
The results are reported in Table 5. The general
impression is that the negative effects of educational activity have become more
negative and the positive effects of higher education upon school completion
have become less positive over time. This implies that younger cohorts postpone
fertility more when they are in education and recuperate slower when they
have finished their studies. Thus, the findings do not support the cohort-effecthypothesis
(H4) that educational differences in the timing of first birth have
Table 5. Educational effects* on first-birth rates of Norwegian women born 1955­1969.
Discrete hazard model (odds ratio estimates). Model A
Birth cohort
1955­1959 1960­1964 1965­1969
Education
In education 0.47 0.34 0.34
Not in education: Primary and lower secondary 1 1 1
Not in education: Upper secondary 1.24 1.00 0.95
Not in education: University, 1st stage 1.59 1.17 1.07
Not in education: University, 2nd stage 1.99 1.40 1.23
*Controlled for age and social and regional background. Numbers in italics = not significant.
TRUDE LAPPEGA° RD AND MARIT RNSEN44
become smaller among women in younger cohorts. However, the findings do fit
the descriptive pattern of increased educational differences across cohorts in
the mean age at first birth.
The argument behind the assumption of a weaker impact of education in
younger cohorts was that the incompatibility between family life and female
labour force participation had become weaker across time. A possible
explanation for the opposite finding is that because young women are more
educated, they have higher career aspirations and want to get a firm foothold
on the labour market before having a baby. At the same time the competition
on the labour market has become fiercer, with more job insecurity and extra
demands of work efforts in many jobs (see e.g. Srensen, 2002). Also longer
parental leaves, especially since the early 1990s, have made it more profitable
to work some time before taking leave, first and foremost to establish eligibility,
but also to secure higher earnings, as the parental benefit is proportional
to earned income.
Table 5 further shows that the main change in the effect of education
occurred between the 1955­1959 and the 1960­1964 cohorts. This implies
that increased postponement of well-educated women is losing momentum in
the youngest generations. In Table 6 the increased negative effect of educational
activity across cohorts is further seen not to apply to all types of study,
Table 6. Educational effects* on first-birth rates of Norwegian women born 1955­1969.
Discrete hazard model (odds ratio estimates). Model B
Birth cohorts
1955­1959 1960­1964 1965­1969
In education:
Primary and lower secondary 0.32 0.15 0.10
Upper secondary: academic 0.22 0.13 0.13
Upper secondary: vocational 0.57 0.42 0.42
University, 1st stage: female
dominated fields
0.48 0.44 0.45
University, 1st stage: male
dominated fields
0.26 0.26 0.23
University, 1st stage: others 0.51 0.45 0.39
University, 2nd stage 0.52 0.45 0.37
Not in education:
Primary and lower secondary 1 1 1
Upper secondary 1.14 1.03 0.99
University, 1st stage 1.35 1.22 1.12
University, 2nd stage 1.58 1.46 1.28
*Controlled for age and social and regional background. Numbers in italics = not significant.
THE MULTIFACETED IMPACT OF EDUCATION 45
as the estimated effects for female and male dominated fields at lower university
level are more or less constant. The general picture of decreasing
recuperation across cohorts among those who have completed education
within different fields, is on the other hand, quite unambiguous (Table 7), as
the more mixed trend among teachers and physicians at upper university level
(inverted and ordinary U-shape, respectively) is not significant.
6. Conclusion
Our analysis of the timing of first birth among women in Norway demonstrates
that education influences young women's entry into motherhood in
manifold ways. In full accord with previous research, our results confirm that
there is a student-effect as educational activities clearly delay motherhood.
The differences in first-birth rates between young women enrolled at different
levels and within different fields of education further suggest that women's
Table 7. Educational effects* on first-birth rates of Norwegian women born 1955­1969.
Discrete hazard model (odds ratio estimates). Model C
Birth cohorts
1955­1959 1960­1964 1965­1969
In education 0.47 0.34 0.34
Not in education:
Primary and lower secondary 1 1 1
Upper secondary 1.24 1.00 0.95
University: Hum. and Aesthetics, 1st stage 1.13 0.83 0.75
University: Hum. and Aesthetics, 2nd stage 1.72 1.27 1.06
University: Teaching, 1st stage 1.74 1.29 1.25
University: Teaching, 2nd stage 1.53 1.81 1.51
University: Social science, 1st stage 1.26 0.93 0.81
University: Social science, 2nd stage 1.82 1.40 1.34
University: Law, 2nd stage 2.06 1.33 1.29
University: Adm. and Economics, 1st stage 1.44 1.19 1.03
University: Adm. and Economics, 2nd stage 2.02 1.33 1.10
University: Engineering, 1st stage 1.38 1.03 0.97
University: Engineering, 2nd stage 1.95 1.37 1.13
University: Nursing, 1st and 2nd stage 1,83 1.42 1.39
University: Physicians, 2nd stage 2.34 1.43 1.62
University: Health care otherwise, 1st stage 1.72 1.29 1.11
University: Health care otherwise, 2nd stage 1.98 1.63 1.30
University: Others and missing 1.44 1.01 0.93
*Controlled for age and social and region background. Numbers in italics = not significant.
TRUDE LAPPEGA° RD AND MARIT RNSEN46
fertility behaviour is influenced by future educational aspirations as well as by
aspirations related to the future occupational career. After finishing their
studies women with a higher-level education proceed to motherhood faster
than women at a lower level. This confirms that there is a catching-up-effect
among women who have delayed childbearing longer. But at a given educational
level, we also find contrasting behaviour between women within different
fields of education, which indicates that there is a career-adjustmenteffect
in women's entry into motherhood. These contrasts may be related to at
least two factors. First, there may be differences in the economic opportunity
costs of a career break in different sectors that are independent of educational
level. Besides, women with education directed at different sectors may need
different amounts of time after school to get a good foothold in the labour
market and therefore proceed to motherhood in different speed. Secondly,
there may be differences in women's preferences towards family and work that
might be reflected in their educational choices. Family-orientation and workorientation
are not necessarily opposites: Some women might have high
aspirations for both a family and a work career. In order to achieve both, the
opportunity costs of withdrawal from the labour market in connection with a
childbearing need to be compensated for to some extent. In modern welfare
states, social policies directed at working mothers are helping to alleviate this
need and are making family life and female labour market commitments more
compatible. However, social policies may generate different responses from
women with different educational backgrounds. Long parental leaves and
generous family benefits may fit better with a career track in certain jobs, and
may thus be perceived to reduce the opportunity costs of childbearing more
for some women than for others. Together with a more competitive labour
market, this may have contributed to the increasing educational differences in
timing of motherhood observed in younger cohorts.
Acknowledgement
This research was funded by the Research Council of Norway and Statistics
Norway. Comments from participants at the 2003 European Population
Conference and helpful suggestions from two anonymous referees have been
greatly appreciated.
Notes
1
In Norway, the economic situation for female students who give birth to a child is relatively
good, as they receive a grant for 42 weeks equal to the maximum annual amount they would
receive as a loan otherwise. However, this is still far less than what they can expect to earn
after completing education.
THE MULTIFACETED IMPACT OF EDUCATION 47
2
Women who have died or emigrated are excluded as information on their education is also
often lacking. This could, in principle, bias the results as fertility is related to both migration
and mortality. Fertility differentials should, however, only be very marginally affected.
3
We did not run separate analyses for younger cohorts, as a high proportion have not yet
entered motherhood, especially among the highly educated.
References
Allison, Paul D., 1984. Event History Analysis. Regression for Longitudinal Event Data. Sage
University Paper series on Quantitative Applications in the Social Sciences, 07-046. Sage
Publications, Beverly Hills and London.
Andersson, G., 2000. `The impact of labor-force participation on childbearing behaviour: procyclical
fertility in Sweden during the 1980s and the 1990s', European Journal of Population
16: 293­333.
Becker, G. S., 1981, 1991. A Treatise on the Family. Harvard University Press. London/
Cambridge.
Blossfeld, H.-P., 1995. `Changes in the process of family formation and women's growing
economic independence: A comparison of nine countries', in H-P. Blossfeld (ed.) The New
Role of Women-Family Formation in Modern Societies, Westview Press, Oxford.
Blossfeld, H-P. and Huinink J., 1991. `Human capital investments or norms of role transition?
How women's schooling and career affect the process of family formation', American
Journal of Sociology 97: 143­168.
Cigno, A. and Ermisch, J., 1989. `A micro-economic analysis of the timing of births', European
Economic Review 33: 737­760.
Easterlin, R. A., 1966. `On the relation of economic factors to recent and projected fertility
changes', Demography 3: 131­153.
Gustafsson, S., 2001. `Optimal age at motherhood. Theoretical and empirical considerations
on postponement of maternity in Europe', Journal of Population Economics 14: 225­247.
Hakim, C., 2000. Work-Lifestyles Choices in the 21st Century: Preference Theory, Oxford
University Press.
Hakim, C., 2003. `A new approach to explaining fertility patterns: Preference theory', Population
and Development Review 29(3): 349­374.
Hank, K., 2002. `Regional social contexts and individual fertility decisions: A multilevel
analysis of first and second births in western Germany', European Journal of Population 18:
281­299.
Happel, S. K., Hill, J. K. and Low, S. A. 1984. `An economic analysis of the timing of
childbirth', Population Studies 38: 299­311.
Hoem, B., 1994. `Lä rare fo¨ der flere barn' (Teachers give birth to more children), Vä lfä rdsbulletinen
3: 17­19.
Hoem, B., 2000. `Entry into motherhood in Sweden: The influence of economic factors on the
rise and fall in fertility, 1986­1997', Demographic Research, Vol. 2: Article 4. Available at
www.demographic-research.org/Volumes/Vol2/4.
Hoem, J. M., 1986. `The impact of education on modern family-union initiation', European
Journal of Population 2: 113­133.
Joshi, H., 1990. `The cash opportunity costs of childbearing', Population Studies 44: 41­60.
Kalmijn, M., 1996. `Effecten van opleidingsniveau, duur en richting op het tijdstip waarop
paren hun eerste kind krijgen', (Effects of educational level, school enrollment and type of
schooling on the timing on the first birth), Bevolking en Gezin 1: 41­71.
Kravdal, ., 1994. `The importance of economic activity, economic potential and economic
resources for the timing of first births in Norway', Population Studies 48: 249­267.
TRUDE LAPPEGA° RD AND MARIT RNSEN48
Kravdal, ., 2004. `An illustration of the problems caused by incomplete education histories
in fertility analyses', Demographic Research, Special Collection 3: Article 6, 133­154.
Available at: www.demographic-research.org/special/3.
Kreyenfeld, M., 2000. `Educational attainment and first births: East Germany before and after
unification', MPIDR Working Paper, WP 2000-011.
Lappega° rd, T., 2000. `New fertility trends in Norway', Demographic Research Vol. 2: Article
3. Available at: www.demographic-research.org/Vol2/3.
Lappega° rd, T., 2002. `Education attainment and fertility pattern among Norwegian mothers',
Documents 2002/18, Statistics Norway.
Liefbroer, A. C. and M. Corijn, 1999. `Who, what, where and when? specifying the impact of
educational attainment and labour force participation on family formation', European
Journal of Population 15: 45­75.
Pollak, R. A. and Watkins, S. C., 1993. `Cultural and economic approaches to fertility, Proper
marriage or mesalliance? `, Population and Development Review 19: 467­495.
Rnsen, M., 2004. Fertility and family policy in Norway ­ A reflection on trends and possible
connections. Demographic Research Vol. 10: Article 10, 263­286. Available at:
www.demographic-research.org/Vol10/10.
Rnsen, M. and Sundstro¨ m, M., 2002. `Family policy and after-birth employment among new
mothers ­ A comparison of Finland, Norway and Sweden', European Journal of Population
18: 121­152.
Santow, G. and Bracher, M., 2001. `Deferment of the first birth and fluctuating fertility in
Sweden', European Journal of Population 17: 343­363.
Srensen, B. Aa., 2002. `Det gra° dige arbeidsliveť (The greedy workplace), in A. Roness and S.
B. Matthiesen (eds), Utbrent: krevende jobber ­ gode liv? (Burnout: demanding jobs - good
lives?), Fagbokforlaget, Bergen, 116­143.
Statistics Norway, 2001. Norwegian Standard Classification of Education, Official Statistics of
Norway, C676.
Statistics Norway, 2004a. Population statistics. Births 2003, www.ssb.no/english/subjects/02/
02/10/fodte_en/
Statistics Norway, 2004b. `Sosiale indikatorer' (`Social indicators'), Samfunnsspeilet 4: 76­103.
United Nations, 1996­2001. Fertility and Family Surveys in Countries of the ECE Region.
Standard Country Reports, Economic Studies No. 10a-v, United Nations, New York and
Geneva.
Walker, J. R., 1995. `The effect of public policies on recent Swedish fertility behaviour',
Journal of Population Economics 8: 223­251.
THE MULTIFACETED IMPACT OF EDUCATION 49