Cet ouvrage fait partie de la bibliothèque YouScribe
Obtenez un accès à la bibliothèque pour le lire en ligne
En savoir plus

Hausse de l'activité féminine : quels liens avec l'évolution de la fécondité ? (version anglaise)

De
10 pages
L'évolution des taux d'activité féminins est parfois mise en parallèle avec celle de la fécondité. Ce lien soulève de nombreuses questions : est-il purement fortuit ou traduit-il un rapport causal, et de quel sens peut-il être ? Est-ce la désaffection pour les familles nombreuses qui aurait permis la hausse des taux d'activité, ou est-ce au contraire la hausse du désir ou du besoin d'activité qui ont conduit les ménages à limiter leur descendance ? Par ailleurs, ces évolutions ont-elles interféré avec une modification du degré d'incompatibilité entre activité et charge de famille ? Il est possible d'examiner ces interdépendances à l'aide d'un modèle simple qui n'en privilégie aucune a priori. Son application fait ressortir le rôle moteur de la préférence pour l'activité, davantage qu'une désaffection intrinsèque pour les familles nombreuses. Cette thèse est compatible avec une caractéristique importante de l'évolution de l'activité féminine : le fait qu'elle s'est fortement accrue à charge de famille donnée. Ce diagnostic se retrouve si l'on analyse les comportements démographiques et d'activité selon le niveau de diplôme atteint par la mère.
Voir plus Voir moins

Is the rise in female
participation linked to fertility
rate trends?*
Didier Changes in female participation rates are sometimes related to changes in
Blanchet and fertility rates. This association raises a number of questions. Is it pure
Sophie Pennec** coincidence or does it reflect a causal relation and, if so, in which direction?
Has the shift away from large families prompted the increase in the
participation rate or rather has the increasing desire or need to work
prompted families to limit the number of children they have? Furthermore,
have these changes interfered with the modification of the extent of
incompatibility between work and family size?
These interdependencies can be studied using a simple model that does not a
priori favour any one of them. The application of this model reveals the driving
role of preference for participation over an intrinsic disinclination for large
families. This theory is compatible with an important characteristic in the
growth in female participation: the fact that it has risen sharply for a given
family size. The same finding results from an analysis of demographic and
participation patterns by the mother’s level of qualifications.
** At the time of writing
this article, Didier
Blanchet was Head of
1INSEE’s Social Policy emale participation and fertility rates detailed measurements of these tren , but tods
and Redistribution Fhave changed a great deal since the interpret them at their aggregate level.Division and a
researcher at INED. mid 1960s. These trends are summarised in
Sophie Pennec is a chart I. The average number of children per
woman, as measured by the cyclical fertilityDebated CausalityThis article is an update
of Blanchet and Pennec indicator, dropped from a high of 2.9 in 1964
(1993). The authors to around 1.8 to 1.9 in 1975. More recently, itWhat conclusions can be drawn about the
would like to thank the
has fallen to a new level of approximatelysimultaneous increase in female participationtwo anonymous editors
for their constructive 1.65 to 1.7. Female participation, as measuredand decrease in fertility? A number of
comments on a previous by the percentage of the female populationcontrasting interpretations can be put forward.
version of this article.
aged 25 to 49 in the labour force, increased
1from 42% to 80% over the same period. The See, for example, Lelièvre (1987), Des Nétumières (1994), and
Djider and Lefranc (1995).purpose of this article is not to propose more
Names and dates in pa-
rentheses refer to the
bibliography at the end *Originally published as "Projections de population active et participation au marché du travail," Économie et Statistique , no. 300,
of the article. 1996 – 10.
INSEE STUDIES N° 9, November1997 1Chart I There could be a causal link running from
Growth in female participation (percentage fertility to participation: exogenous causes may
of the female population aged 25 to 49 in the have reduced family size and therefore made
labour force) and fertility (cyclical rate) more women available to endeavour to enter
the labour market. Alternatively, motivations
for employment could be growing and bringing
about a drop in fertility. This viewpoint is
typical of many economic approaches to
demographic behaviour and labour supply: the
general increase in wage earning labour
productivity has raised the opportunity cost of
not working. This in turn has led to a general
shift from the domestic sphere to the market
sphere. The drop in fertility can be seen as a
result of this shift. A third approach is to deny
any link between the two and to view the
phenomenon as a pure coincidence devoid of
any causal relation.
At first glance, the data do not come out in
favour of any one of these different
Chart II interpretations. There is admittedly an
Growth in female participation for a given argument in support of the theory of
number of dependent children independence, which is that participation rates
have greatly increased, even for a given fertility
level. This can be seen in chart II, which shows
the increase in the participation rate by the
number of dependent children. This increase
actually does suggest that the increase in
participation is independent to some extent
(Véron, 1988). However, this finding does not
uphold the theory that the two phenomena are
totally independent. Although it implies that
the drop in fertility cannot totally explain the
increase in general participation, it does not
exclude the possibility of a partial explanation.
In addition, this finding tells us nothing about
the possible converse effects of the increase in
participation on fertility. We will thus examine
Coverage: women aged over 15 with 0, 1, 2 or 3 dependent children all the possible relationships using an approach
aged under 16 up to 1982 and aged under 18 in 1990. that does not a priori favour any of them and
Source: Population censuses (INSEE)
which, following quantification, ranks them.
Table 1
Breakdown* of the female population by A simple model for analysis ...
participation and the number of dependent
children Our analysis starts by stylising the problem. We
analyse the interdependencies between female
Three or more Two or fewer participation and family size using
children children
cross tabulations derived by describing these
two phenomena dichotomously. We divide the
In the labour force (1) a b
population into women with a high level of
involvement in the labour market, using aOut of the labour force (2) c d
criterion defined later in this article, and
(1) In the labour force: high level of involvement on the labour women with a low level of involvement. These
market.
we call “in the labour force” and “out of the(2) Out of the labour force: low level of involvement on the
labour market. labour force” respectively in order to simplify
*a + b + c + d = 1 the presentation. For fertility, we divide women
2 INSEE STUDIES N°9, November 1997 into two groups: those with more than two cross tabulations between successive censuses.
children and those with two or fewer children. These changes can be seen as the result of three
possible developments, which are not mutually
The population studied is thus divided into 2 exclusive, but are rather liable to overlap.2
categories: women in the labour force with
more than two children and those with two or - Take first of all the example of a change in the
fewer children, and women out of the labour parameter denoting incompatibility between
force using the same family size criterion (sepearticipation and family size. Let us
table 1). The observed frequencies a, b, c, and d suppose that it increases. The potential
in the table’s four cells are obviously assumedeffects are twofold, with one part of the
to satisfy the restriction tha + t b + c + d = 1.population mo ving towards the
Over time, the values of parameters a, b, c, and high participation/small family model and
d will change, but they must always sum to one: another part of the population moving towards
a decrease in the number of individuals in one the opposite low participation/large family
cell necessarily entails an increase in at leasmtodel. The ultimate effects on total
one of the other three cells. The idea is thus to participation and total fertility are therefore
analyse these shifts and understand what theyambiguous, since they depend on the levels of
represent in terms of links between the two the other two parameters and . This is in
types of behaviour. line with the intuition: if two behavioural
patterns become less compatible, we do not
The log linear model described in the normally know which one will increase at the
following box is particularly well suited to expense of the other.
analysing these shifts. This model circumvents
the problems posed by the restriction that the The main and unambiguous effect of a change in
four rates in table 1 must sum to one, since tithe parameter denoting preference for
entails reparameterising them using three participation is an increase in both general
mutually independent values. More participation and participation for a given family
importantly, it involves a more flexible causalsize. A closer look at chart II enables us to
framework than the simple search for anticipate that this type of phenomenon has
uni directional dependencies between actually been observed in France. We can thus
participation and fertility, provided that the expect a spillover effect on family size, provided
three parameters in this model are interpreted in that the parameter is not equal to zero. Although a
terms of exogenous factors of change: growing preference for participation might prompt
a large number of mothers to work without
the first parameter ( in the box) measures thechanging their choice of family size (compared
degree of preference for the first row of thewith previous generations of women), part of this
table: it can therefore be interpreted as a decision to participate will also translate into a shift
parameter of preference for participation as into the category of small families. Less
opposed to being out of the labour force; compatibility between work and family size will
strengthen this trend. Only i g f were equal to zero
the second parameter ( in the box) measures would a growing preference for participation have
the relative degree of preference for the firstno impact whatsoever on fertility.
column: it is thus interpreted as a parameter of
preference for large families; A drop in the parameter denoting a
decreased intrinsic preference for large
- the third parameter ( in the box) measures the families, on the other hand, results first and
degree of incompatibility between participation foremost in a drop in fertility. This could, at the
and family size. A zero value for this paramete same time, contribute to a rise in participation.r
would indicate total independence between
participation and family size at a given time. We
... which highlights the preference forknow that this is not the case, since, at a given
moment in time, participation decreases for participation
larger numbers of children: this parameter will
thus have a positive value. The reality is assumed to consist of a
combination of these three mechanisms.
An analysis of the growth in these three The question is whether one of them is
parameters is therefore a simple way of dominant. To answer this, the proposed
interpreting the changes in the model has to be quantified, which we have
INSEE STUDIES N° 9, November1997 3
gbbagbgaBox
USING A LOG LINEAR MODEL TO ANALYSE INTERDEPENDENCE BETWEEN CHANGES
IN PARTICIPATION AND FERTILITY
The aim of this model is to explain the distribution (no one situation is preferred on average). It is
of the population among the four cells of table 1. A simply a marginal product if equals zero (no
premise shared by many qualitative models (see interaction or incompatibility between participation
Gourieroux, 1989) is that ind ividuals place and fertility). We can also check whether the
themselves in one of the four cells according to the content of each cell changes according to , and
“values” they associate with the different situations. with the expected sign. Therefore, the content of
These values may include certain economic factors the first cell, i.e. the probability of participating and
(positive economic value associated with having a large number of children, can also be
participation, negative value associated with large written as:
families when their cost is high) and non economic
factors (some parents attribute greater intrinsic
1
value to having a large family than others). We a = (1)
assume that the values and costs associated with e + e + e + 1
each type of behaviour can be quantified. The
reference behavioural pattern chosen is women out
of the labour force with small families (reference which is an increasing function of the preference for
participation , which also increases the greater thevalue of 0). We assume that, in terms of this
pattern, working represents an additional value of , preference for a large number of children .
having a large family represents a value , and However, it is a decreasing function of the
incompatibility between participation and a largecombining participation with a large family
represents a negative value (or cost) equal to . family . The probability of being in this first cell is
close to one when and are both close to + ,
while remains finite. Yet the probability of being inWith this as our starting point, we can calculate the
preference rates for the different options (see table this cell is close to zero when the degree of
A) and make the assumption that the behavioural incompatibility between participation and large
pattern with the highest preference rate will be the
one adopted.
Table A
Parameters , and are obviously not the same Relative preferences for the different
for all individuals in the population under
options
consideration. If this were the case, all the
individuals would adopt the same behaviour.
Instead, these values are average values and the Three or more Two or fewer
children childrenfluctuations relating to the individual parameters are
denoted by as many random factors as there are
different situations, i.e. X , X , X and X in thisa b c d In the labour +
case, with the indicators bearing the cell force
designations chosen for table 1. Given these
Out of the labour 0
circumstances, the percentage shown in a given
force
cell of the contingency table is equal to the
probability that the value attributed to the cell is
greater than the values attributed to the three other
cells. For example, frequency a is equal to the Table B
percentage of individuals for whom + - + X isa Contingency table of average preferences
greater than the values for +X , +X and X .b c d +( D = e + e + e + 1)
Calculating this percentage presupposes
hypotheses regarding the distributions of variables
X , X , X and X . These distributions generallya b c d Proportion of
Two or
Three or more families withrequire a complex expression, but a convenient fewer
children three or moreassumption is available regarding the shape of such children
children
1a distribution, which furnishes simple expressions.
These expressions and the expressions for the
+ participation rate by family size and family size by In the e ⁄ D e ⁄ (e +1) e ⁄ D
labourparticipation are shown in table B.
force
The formulae in table B have the simple qualitative
Out of the e ⁄ D 1 ⁄ D e ⁄ (e +1)
properties expected of a model that aims to explain labour
the structure of table 1. All the expressions are force
positive due to the use of exponentials and they
Participation e ⁄ (e + 1) e ⁄ (e + 1)sum to 1, since they are divided by D. The table is
rateevenly distributed if , and are all equal to zero
1 See Blanchet (1992) for a more detailed presentation.
4 INSEE STUDIES N°9, November 1997
aaag-b-b-bbga-gbaa¥-gb-b-agbbgabg--gbbagbaa-aagg-bbgabagabbagaabbagggBox (continued)...
families is very high, while and remain at stable The above model is part truly explanatory model
levels. Finally, the participation rate expressions for (where , and are expressed according to
a given family size also give standard objective measured variables), and part descriptive
logisticalexpressions, which can be applied in many model. We can therefore put forward just one
ways to the separate analysis of these participation possible reparameterisation of the initial contingency
patterns (see for example Lollivier, 1988). tables. However, this reparameterisation is
particularly well suited to the analysis of changes in
It is easy to estimate , and from the frequencies these tables. It is actually easier and more accurate
of a, b, c, and d, since the model used is completely to compare , , and g , which have behavioural
saturated (i.e. none of the parameters , or is a significance and are not, as are a, b, c, and d,
priori constrained to zero). We simply have to solve subject to the restriction that they sum to 1. This
the identity between tables 1 and B, which gives: restriction complicates the analysis of the initial
contingency tables, because there is no knowing
whether a change in one of the frequencies results
= log(b) log( d) (2) from a change in “pure” preference for the
= log(c) log( d) ( corresponding cell or the of3) fsetting of changes in
= log(b) + log(c) log( a) log( d) (4) preference for one of the other three cells.
Table 2done with table 2 and reoducing thpr e table 1
Breakdown of women aged 35 to 39structure for the last five census periods. We
by participation and family sizehave chosen a specific age group for this: 35 to
from 1962 to 1990 (census years)39 year olds. We have assumed that the average
participation rate for this age group is fairly
% ofrepresentative of participation choices for the
3 or 2 or women
entire life span, excluding temporary breaks to more fewer with 3 or
children children moretake care of very young children. This age
children
group is also the age at which the family has
2 1962 In the labour forcevirtually attained its ultimate size . 6.7 25.9 20.5
Out of the labour force 30.6 36.8 45.4
Table 2 sets out the overall female participation % in the labour force 17.9 41.3
and fertility trends mentioned in the
1968 In the labour force 6.9 28.2 19.6
introduction. From 1962 to 1990, the
Out of the labour force 30.0 34.9 46.2
proportion of women aged 35 to 39 with more
% in the labour force 18.6 44.7
than two dependent children decreased from
1975 In the labour force37.3% (6.7 + 30.6) to 27.8% (13.7 + 14.1). The 9.3 36.4 20.3
Out of the labour forceproportion of women in the labour force 26.0 28.3 47.9
increased from 32.6% to 72.4% (6.7 + 25.9 and % in the labour force 26.3 56.3
13.7 + 58.7 respectively). Finally, although few
In the labour force1982 9.3 52.3 15.1
changes are observed in family size for a given
Out of the labour force 15.6 22.8 40.7
participation rate, a phenomenon previously
% in the labour force 37.3 69.7
noted by Léry (1984), participation increases
In the labour forceregardless of the number of children (see chart 1990 13.7 58.7 18.9
Out of the labour forceII). In 1962, only 17.9% of mothers with three 14.1 13.5 51.2
or more children were in the labour force as % in the labour force 49.2 81.3
opposed to 49.2% by 1990. Similarly, the
Interpretation: In 1990, out of every 1,000 womaged 35 to 39,en
participation rate for women with two or fewer 137 were women in the labour force with three or more children.
The participation rate for mothers with three or more children waschildren rose from 41.3% to 81.3%.
49.2%. 18.9% of all women in the labour fo had three or morerce
children.
Table 2 can be used to reconstruct the values for Source: population censuses from 1962 to 1990 (INSEE).
the three parameters in the log linear model for
each census and show how they change over
2 Due to a change in the “dependent child” definition, thetime, as shown in chart III. These parameters
figures for 1990 cover children aged 0 to 18 rather than aged
are dimensionless, and we must therefore 0 to 16 as before. This change in definition appears to have a
confine our discussion to the signs and growth. minimal effect on the classifications used here.
INSEE STUDIES N° 9, November1997 5
babagagbabaabbg
g Growth remains relatively level for the two increase that serves to explain the upturn in the
parameters reflecting preference for families general participation rate and participation for
with more than two children and a given family size.
incompatibility between participation and
family size. The consistently negative value ofThese findings are consistent with an economic
the first parameter reflects the fact that a family explanation of demographic and labour supply
with more than two children is a minority patterns. The wages offered on the labour
choice for all these periods. This value presents market increase with the level of education and,
some fluctuations with a trough around 1982,for a given level of education, with the growth
but no clear trend. The positive value of thein overall productivity. In both cases, this
incompatibility parameter obviously confirms phenomenon results in a greater loss of incomepatibility, but we also observe that the in the event of an exit from the labour force. It
visible degree of incompatibility varies only therefore leads to a higher participation rate,
slightly over time. Conversely, there is a sharp which may remain compatible with sustained
rise in the implicit preference for the family size for some, but leads others to limit
participation parameter. It is clearly this family size. This analysis does not imply that
all changes in demographic patterns can be
reduced to this type of explanation. TheChart III
analysis does, however, appear to be relativelyEstimated and extrapolated (post-1990)
likely and play an important role in explainingvalues for parameters , andg.
3the situation in France .
To further demonstrate the explanatory value of
this method, an evaluation can be made of the
spillover effect of this factor on the drop in the
number of large families. This simply involves
reconstructing what the combined growth in
(a) participation and fertility would have been if
this factor alone influenced it. In other words, it
entails calculating the cross tabulations of
(g) participation and family size that would have(b)
been observed since 1962, by keeping
parameters and at their 1962 levels and
leaving to change at its actual rate. This is
shown on the left hand side of chart IV for both
the general participation rate and the general
frequency of large families. As expected, the
Chart IV growth in parameter alone perfectly explains
Reconstructed and predicted growth in the the increase in the general participation rate. It
general participation rate and the prevalence also significantly explains part of the drop in
of large families when influenced solely by the frequency of large families, via the
the increase in preference for participation above described selection effect.
( andg remain constant)
Might the continuation of this trend bring about
the demise of families with three or more
children? Probably not. Given that a certain
number of households are able to cope with a
family of more than two children with both
parents in the labour force, there is no reason
for this to change when preference for the
labour force reaches the stage where couples
3 Note that this analysis does not necessarily forecast the
chronological order of the birth of children and exits from the
labour force. It could very well be in line with the observation that
leaving the labour force precedes starting a family (Lollivier,
1988): for example, a low level of preference for participation can
lead to an exit from the labour market before children are born.
6 INSEE STUDIES N°9, November 1997
bbaaagbwith just one member in the labour force no1962 levels and the value tending towards an
longer exist. arbitrarily high level. Although participation
tends towards 100%, the percentage of families
The right hand side of chart IV represents thiwsith more than two children remains above
theory with an imputed extension of the curveszero at 22%. This exercise is not a real
for participation and frequency of large prediction. The highly aggregate nature of the
families, with the and values frozen at their model precludes its use in this form for an
Table 3
Participation and family size by level of education* of the man and the woman,
with estimated values for parameters , and
In the labour In the labour Out of the Out of the
Woman’s force with force with labour force labour force
Man’s level
level of 3 or more 2 or fewer with 3 or morewith 2 or fewerab
of education
education children children children children
(a, in %) (b, in %) (c, in %) (d, in %)
0 0 20.3 37.5 27.2 14.9 0.921 0.601 1.216
1 17.2 49.0 15.2 18.5 1.765 0.191 1.559
2 19.0 45.7 18.7 16.6 1.657 0.072 1.221
3 15.9 46.5 14.7 22.8 2.021 0.386 0.729
4 12.4 37.8 19.3 30.5 2.308 0.057 1.229
5 10.6 33.0 23.5 33.0 2.211 0.693 0.824
1 0 15.6 61.2 12.7 10.5 0.973 0.194 0.850
1 14.3 61.3 9.5 15.0 1.415 0.451 1.002
2 16.8 60.9 7.5 14.7 1.406 0.820 0.728
3 14.1 54.6 12.1 19.2 2.510 0.117 1.817
4 14.5 50.8 14.0 20.7 2.553 0.251 1.615
5 12.3 40.4 18.2 29.1 2.890 0.405 1.791
2 0 18.5 58.4 12.0 11.1 1.014 0.119 0.997
1 13.6 63.9 6.9 15.7 1.420 0.673 0.611
2 16.9 60.9 10.2 12.0 1.622 0.165 1.116
3 12.4 59.3 9.8 18.5 2.293 0.377 1.121
4 12.4 59.6 11.3 16.8 2.138 0.659 0.626
5 9.5 47.4 18.1 25.0 2.538 0.693 0.865
3 0 21.1 64.5 5.8 8.6 0.714 0.434 0.636
1 13.5 73.8 6.7 6.0 1.044 0.465 0.886
2 16.0 71.8 5.0 7.2 1.165 0.629 0.937
3 14.5 68.0 7.8 9.7 1.951 0.213 1.330
4 17.2 57.4 7.0 18.4 1.887 0.340 0.921
5 14.4 48.6 17.0 20.0 2.625 0.451 0.934
4 0 18.8 68.0 6.4 6.8 0.214 0.456 0.654
1 17.8 69.8 7.0 5.4 0.896 0.393 0.859
2 19.0 68.7 4.2 8.1 1.269 0.397 1.175
3 18.3 64.8 7.0 9.8 1.136 0.969 0.238
4 21.1 60.6 9.5 8.8 1.930 0.080 1.137
5 17.8 45.4 18.6 18.2 2.036 0.271 0.959
5 0 15.8 72.3 4.0 7.9 0 0.339 0.793
1 18.0 72.0 6.0 4.0 0.329 0.470 0.722
2 15.8 75.2 3.0 5.9 0.639 0.323 1.279
3 18.3 73.1 3.4 5.3 0.887 0.160 1.057
4 19.2 65.7 6.5 8.6 0.916 0.021 0.959
5 23.6 59.5 9.0 7.9 2.018 0.125 1.052
*The level of education of each member of the couple is measured on a scale of 0 to 5, where 0 is no qualifications or primary education; 1 is
the BEPC (first degree certificate) taken at the age of 14; 2 is the vocational education certificate or the technical school certificate (CAP, BEP);
3 is the baccalauréat; 4 is the baccalauréat plus two years of higher education; and 5 is the baccalauréat plus four years of higher education.
Source: Family Survey, 1990 (INSEE).
INSEE STUDIES N° 9, November1997 7
gbaaabggbgChart V operational prediction of family structures and
Values of the , and parameters by the participation. It does, however, proffer a way of
level of education of both spouses* making integrated predictions that would make
general use of the logistical extrapolation
procedures normally applied solely to
participation rate predictions.
Another application: participation
and family size by level of education
This analytic framework can be used to
address other aspects of the relationship
between female participation and fertility. To
illustrate this, we take the example of the
cross tabulation of fertility and participation
by the level of education of both spouses. The
figures used are taken from the 1990 Family
survey. This survey differs from the census in
that the “number of children” variable
reflects the actual number of children born
rather than just dependent children.
Therefore, we can study the case of slightly
older women who have finished, or nearly
finished, forming their family. Here again,
fertility is divided into women who have
given birth to more than two children and
those with two or fewer children.
Participation is derived from the respondent’s
working status when the survey was carried
out. This status is assumed to accurately
reflect the distinction between a high and low
or zero level of female involvement in the
labour market. The population studied
consists of married women aged 40 to 44 at
the time of the survey. The level of education
of each spouse is measured on a scale of 0 to
5, where 0 is no qualifications or primary
education; 1 is the BEPC (first degree
certificate) taken at the age of 14; 2 is the
vocational education certificate or the
technical school certificate (CAP, BEP); 3 is
the baccalauréat; 4 is the baccalauréat plus
two years of higher education; and 5 is the
baccalauréat plus four years of higher
education.
A table similar to table 1 is drawn up for each
intersection between a couple’s levels of
education. The , and parameters
associated with each intersection between a
man’s level of education and a woman’s level
of education are deduced by equations (2), (3)
and (4) in the model (see box).
The data and conclusions are shown in table 3
and chart V respectively. They demonstrate
that, once again, it is the preference for
8 INSEE STUDIES N°9, November 1997
bbaaggbagparticipation that appears to determine the This explains the increase in participation for a
differences in behaviour between couples withgiven family size and the widespread
different levels of education. The preference foroccurrence of the participation/number of
participation increases the higher the woman’schildren combination. Another part of the
level of education, and decreases the higher the female population changes its participation
man’s level of education (this latter phenomenonbehaviour by restricting family size, hence the
is clearly demonstrated only between levels 4 and spillover effect on the distribution of families
5). By comparison, the preference for large by size.
families and the degree of incompatibility
between participation and fertility bear no clearIt is important not to misinterpret the
trends, but simply fluctuations, mainly due to the significance or implications of this conclusion.
sampling fluctuations. In particular, the large Firstly, it applies to a summary and synthetic
variations observed in the p arameter concern description of the behaviour studied. A more
sub-samples too small to be of any significance.detailed study would probably find more
These results are comparable with results complex interrelations. Secondly, it cannot
previously obtained using the 1982 Family necessarily be applied elsewhere. International
survey, suggesting that this explanatory comparisons for example, as opposed to
framework has a certain permanence (Blanchetcomparisons over time or across
and Pennec, 1993). socio economic groups, find variations in
fertility that have no apparent link with the
Statistics can therefore help shed light on the level of participation (Blanchet and Pennec,
4 4 See the comparison apparently indecidable causal link between 1993) . Finally, an emphasis on the driving role
of France and Germany female participation and fertility. The of the preference for participation in no way
by Fagnani (1992) for the
findings posited here place the emphasis onmeans that an increase in fertility, assumingdiversityof factors explai-
ning differences in partici the preference for participation, whether a that this is desired, can only be brought about
pation and fer tility spontaneous preference or a more by a decrease in participation. An increase in
between countries
compulsory preference in terms of the needfertility, when a high level of preference for
for a second wage. The scenario then plays participation is present, can just as easily result
out in a straightforward manner more or lessfrom reduced incompatibility between
consistent with the intuition. One part of the participation and family size. This could be the
female population is able to shift its aim of a family policy. Once again, the findings
participation behaviour away from previous based on this model are consistent with
generations without changing its family size.expectations.
INSEE STUDIES N° 9, November1997 9
gREFERENCES
Blanchet, D. (1992), “ Interpréter les évolutions tiques familiale en France et en Allemagne de
temporelles de l’activité féminine et de la fécondi-l’Ouest ,”Recher ches et Prévisions , no. 28, Cnaf,
té ,” Population, no. 2, pp. 389 408. pp. 23 38.
Blanchet, D. and Pennec, S. (1993 , “ A) Simple Gourieroux, C. (1989) , Econométrie des variables
Model for Interpreting Cross Talations of Fabu - qualitatives, Economica.
mily Size and Women’s Labour Force Participa-
tion ,” European Journal of Population , no. 9, Lelièvre, E. (1987) , “ Activité professionnelle et
pp. 121 143. fécondité : les choix et les déterminations des fem
mes françaises entre 1930 et 1960 Cahiers Qué ,”
Bordes, M. M. and Guillemot, D. (1994) , Marché becois de Démographie , vol. 16, no. 2,
du travail. Séries longues , “ Insee Résultats ” collec pp. 207 236.
tion, “ Emplois Revenus ” series, no. 62 63.
Léry, A. (1984) , “ Les actives de 1982 n’ont pas
Des Nétumières, F (1994) , “ L’arrêt de travail des moins d’enfants que celles de 1968 ,” Économie et
femmes. Mariage et maternité DEA,” paper , Univer Statistique, no. 171 172, pp. 25 34.
sité René Descartes, Paris Sorbonne.
Lollivier, S. (1988) , “ Activité et arrêt d’activité fé
Desplanques, G. (1993) , “ Activité féminine et vieminine, le diplôme et la famille ,” Économie et Sta-
familiale ,” Économie et Statistique, no. 261, tistique, no. 212, pp. 25 30.
pp. 23 32.
Sardon, J. P. (1994) , La population européenne en
Djider, Z and Lefranc, C. (1995), “ Femme au foyer chiffres Fécondité générale , INED.
: un modèle qui disparaît Insee Pr,” emière, no. 403.
Véron, J. (1988) , “ Activité féminine et structure fa
Fagnani, J. (1992) , “ Les Françaises font elles des miliale, quelle dépendance ? ,”Population , no. 1,
prouesses ? Fécondité, travail professionnel et poli pp. 103 120.
10 INSEE STUDIES N°9, November 1997

Un pour Un
Permettre à tous d'accéder à la lecture
Pour chaque accès à la bibliothèque, YouScribe donne un accès à une personne dans le besoin