In a number of developed countries, the delay in the onset of family formation is linked to later exit from the educational system and to the increasingly common desire of women to begin their working life before having children (in France, for example, the labour force participation rate of women aged 25-29 exceeds 80%). But what happens when these women encounter difficulties in obtaining a job, and experience periods of unemployment? In trying to answer this question, Monique Meron and Isabelle Widmer have used data from reproductive and work histories collected in a French national survey. Periods of unemployment at the beginning of a union appear to result in postponement of the first birth. Women who experience unemployment have very different family behaviour from homemakers, who have a first child more rapidly than women who are in the labour market.
What are the links between trends in family behaviour and changes in the labour market? Economic ﬂuctuations have a particularly important impact on the beginning of the working life of young people at the time when they may be starting a family. How will a young woman who is in a childless couple react when she is confronted with a period of unemployment? Will she accelerate her maternity schedule in view of the “free” time that is thus imposed on her? Or, on the contrary, will she temporarily forego those plans in consideration of the uncertain future and the material difﬁculties linked to the search for employment? Or are these two spheres — family and occupation — independent of one another? With respect to the link with fertility, is a period of unemployment comparable to a period out of the labour force? Is the behaviour of unemployed women closer to that of homemakers or to that of employed women?
Since the middle of the 1970s in France, the age at ﬁrst birth has risen while unemployment rates, and particularly those of young women, have increased considerably. The risk of unemployment, like the timing of fertility, depends on the place of women in society, on their birth cohort, and on the education and training that they have received. But the usual descriptive statistics do not enable one to establish a clear link between ﬂuctuations in unemployment and age of mother at time of ﬁrst birth, even when taking into account the differences in the above factors.
The methods used here are based on event histories. They make it possible to analyse individual life histories in all their diversity in order to better understand the interaction between the activity status of women and the decision whether to have a ﬁrst child. This biographical approach is used to study women who were surveyed in the Youth and Careers Survey carried out in 1997 by INSEE, the French National Institute of Statistics. In this nationally representative survey, there is a large sample of women, including both women who have and who have not lived with a partner. For reasons of homogeneity of the data and to allow sufﬁcient observation time in relation to the beginning of reproduction, we ﬁrst examine the experience of women aged 31 to 45 in 1997 (born from 1952 to 1966). Then we investigate whether the trends are apparent as well among younger women, aged 24 to 29 (born from 1968 to 1973).
Beginnings of family life and entry into the world of work: Changes in timing
The most recent cohorts of women, as compared to their elders, have taken the steps considered as constituting the passage to adult life (completing their schooling, leaving home, beginning life in union, seeking employment) at later ages. These events, although occurring earlier and closer together than for men, have been increasingly spread out in the lives of women since the middle of the 1970s (Galland, 1995, 2000; Galland and Meron, 1996).
At the same time, the average childbearing age of women has gone from close to 27 years in 1975 to over 29 years in 2000 (Prioux, 2001). This increase has taken place, at least for now, without threatening generation replacement, since all cohorts have had about 2.1 children per woman by the age of 40.
With respect to the ﬁrst birth, the average age of women at the time of delivery has gone from a low point of less than 24 years in 1972 to 25 years in 1983 and 26 years in 1989.
“Thus, at their 25th birthday, only 37% of the women born in 1965 were already mothers, as compared to 55% of women born in 1955 and 61% of those born in 1945”.
(Prioux, 1996, p. 16)
According to the INSEE surveys, the age at which half of the women had become mothers was 26 years for those born in 1963-1966 and 28.2 years for those born between 1968 and 1971 (Galland, 2000).
Postponement of the first birth is linked to family transformations, the prolongation of education, and the development of women’s labour force participation
Since the 1970s the time spent in a union prior to the arrival of the ﬁrst child has been prolonged, thus reversing the trend of the previous decade. Demographers explain this evolution by the diffusion of new methods of contraception in France. Better control over fertility contributes also to the lengthening of other birth intervals. But while the increase in age at ﬁrst birth observed in the mid-70s can clearly be attributed to the growing use of efﬁcient modern contraception, the continuation of the trend calls for other explanations.
The growing fragility of relationships and the increase of non-marital cohabitation have also characterized the recent period. In this context, the fact that 83% of children were living with both parents in 1994 (Villeneuve-Gokalp, 1999) indicates that today, children are, by and large, the result of well-considered decisions by the couple (Toulemon, 1994).
Since the end of the Second World War, the increase in the duration of formal schooling in France has been spectacular, as it doubled in ﬁfty years. The share of secondary school graduates (those with a baccalauréat) has grown from 4% in the 1946 cohort to more than 60% today, and the proportion holding a technical or vocational diploma (CAP, BEP) among persons aged 25-34 has tripled during the period (Estrade and Minni, 1996). Women have especially beneﬁted from this trend: since 1996, they study for longer than men and they have a higher level of educational attainment, which was far from being the case among earlier cohorts.
The age at ﬁrst birth increases with women’s educational attainment. Women born between 1950 and 1954 were on average 22.6 years old at the birth of their ﬁrst child if they had only completed primary school, and 28.4 years old if they had completed four years of higher education. The differences by level of education and social class grew during the 1980s and the 1990s (Desplanques and de Saboulin, 1986). According to forecasts, women born between 1960 and 1964 with four years higher education will have, on average, further delayed the ﬁrst birth by 1.5 years as compared to those born ten years earlier. By contrast, the average age at ﬁrst birth of women with low educational attainment appears to be remaining more or less stable. F. de Singly explains how “women with career prospects delay their marriage for the duration of their studies and of a ﬁrst job” (1987, p. 182). “Moreover, as long as she is pursuing postsecondary education, a woman will avoid having children, and this results in later births” (Desplanques, 1996, p. 17). The general increase in the age at ﬁrst birth is thus all the more marked from one generation to the next as women become increasingly educated.
Women’s labour force participation rates have grown considerably since the beginning of the 1970s. In March 2001 nearly 80% of women aged 25-49 were employed or were seeking employment (full time or part time) whereas the corresponding ﬁgure in 1975 was less than 59% (Aerts and Mercier, 2001). But the labour force participation rates of women are all the lower as their parity is higher. In January 1999, 79% of women aged 25-49 were in the labour force; this rate was 88% among childless women in union, 83% for mothers of one child, 74% for those with two children and 52% for women with at least three children. With the exception of a very recent decline due to the inﬂuence of the extension of the parental educational allowance in July 1994 (Bonnet and Labbé, 1999), the economic activity of mothers of two or more children has increased the most.
Among younger women, inactivity (in the sense of not being in the labour market) has become increasingly uncommon. In 1995, 8% of women aged 15-29 were homemakers, whereas the corresponding ﬁgure in 1975 was 19% (Meron and Minni, 1995). Since 1995, however, some of these changes have slowed down. The duration of formal education has stopped increasing and the percentage of inactive women has stopped falling (Brunet and Minni, 2000).
The effect of women’s employment status has not been studied very much, except in relation to completed family size. In particular, little is known about the incidence of activity status (employment, unemployment, non-participation in the labour force) on the timing of family formation. Family sociologists and specialists of home economics have developed, however, the idea that the family and employment spheres are linked and are part of a broader gender division of labour (Barrère-Maurisson and Marchand, 2000).
Difficulties of economic integration and postponed family formation: What are the links?
The poor performance of the economy has resulted in difﬁculties of economic integration for young people. Unemployment, which has grown substantially since the end of the years of strong economic growth following World War II, hits them especially hard, and women are most affected (Maruani, 1998). Unemployment rates, which were holding at around 2.5%, rose sharply beginning in the ﬁrst half of the 1970s to exceed 10% in 1985. After a brief improvement at the end of the 1980s, unemployment reached record levels in 1994 and then again in 1997, exceeding 12% of the labour force before beginning a new decline since 1998. In March 2001 the overall unemployment rate was 8.8%, and unemployment rates still varied greatly by educational attainment, ranging from 14.1% for those without any diploma or only a primary school certiﬁcate, to about 5% for those with higher education (Aerts and Mercier, 2001).
Although young people are staying in school for longer, nearly half of them (48%) are in the labour market between the ages of 15 and 29. Despite an improvement in recent years, the unemployment rate of young workers remains very high, and their employment is especially sensitive to economic ﬂuctuations. This is especially the case for those with lower levels of educational attainment and for those who entered the labour market most recently (Brunet and Minni, 2000; Fondeur and Minni, 1999; Meron and Minni, 1995).
Thus, the context in which today’s young people begin their working and adult lives is much more difﬁcult than for their parents. The future prospects for young adults, in terms of careers and salary, are not good (Baudelot and Establet, 2000), and this is certainly not without inﬂuence on the family formation process. Young men especially mention these adverse prospects as a reason for continuing to live with their parents and delaying the onset of family formation (Leridon and Villeneuve-Gokalp, 1994; Galland and Meron, 1996; Dormont and Dufour-Kippelen, 2000). The inﬂuence of economic factors on the formation of couples has also been studied: among the most recent cohorts (born after 1968), “the effect of the ‘degree of precariousness’ is not the same for men and women. A ﬁxed-term employment contract does not have adverse consequences for women, but it does for men. But a more precarious status (subsidized employment, training course for integration into the labour market, unpaid training, unemployment) is a negative factor for both sexes” (Ekert and Solaz, 2000; see also Kieffer, Marry, Meron, and Solaz, forthcoming in 2002).
Monographs and cross-sectional surveys give conflicting answers regarding the impact of unemployment on childbearing plans
In the social science literature, unemployment sometimes appears as a factor leading individuals to put off until better times their plans for childbearing, and sometimes as a period of “non-activity” to be taken advantage of in order to have a child.
Thus, according to the most recent results of INSEE’s permanent Survey on Living Conditions for 1998, “72% of women under the age of 25 consider that it is ‘very important’ for a woman to have a stable job prior to having a ﬁrst child” (Toulemon and Leridon, 1999). But other studies, based on the life histories of young women in precarious economic situations, support the idea that young women, when confronted with difﬁculties in entering the labour market, have a tendency to turn towards the family sphere.
“These young women will ‘make use’ of their professional precariousness to carry out various projects during the ‘interludes’ provided by these career interruptions (these projects range from voluntary mobility to formal training, via the ‘programming’ of a child”.
(Nicole-Drancourt and Roulleau-Berger, 1995, p. 79; see also Nicole-Drancourt, 1989)
Yet in the life histories, the proximity between a period of unemployment and the arrival of a child can lead the respondent to reconstruct, in good faith, an ex post facto explanation. Biographical interviews offer the opportunity “to invent discursive strategies for confronting the interview situation, for ‘saving face’ (Goffman, 1974) and afﬁrming their speciﬁcity as a subject” (Demazière and Dubar, 1997).
While both unemployment and age at ﬁrst birth have tended to increase since the mid-1970s, the comparison between economic ﬂuctuations and the timing of family events does not establish a direct link between unemployment and fertility (Figures 1 and 2). Although they are less affected by unemployment, the most educated women are most likely to delay the ﬁrst birth; in contrast, the least educated women have children earlier, despite their greater sensitivity to economic ﬂuctuations. Moreover, the curves showing age at childbearing do not exhibit ﬂuctuations similar to those on the curves showing unemployment, regardless of the level of education of the women, even with a time lag.
In view of the contradictory conclusions stemming from opinion surveys, monographs and statistics with respect to the impact of unemployment on childbearing plans, a biographical approach seemed appropriate.
Event history analysis allows a different approach
There are multiple and diversiﬁed consequences of unemployment for family life. They depend not only on the position of the unemployed individual in the household and the type of unemployment faced, but also on the timing of the spell of unemployment in the individual’s life. The consequences, then, vary according to age, place of origin, the spouse’s employment status, the duration of the union… In order to measure the impact of unemployment on women’s ﬁrst maternity, we have used event history analysis to study the time spent by couples in the childless state (Courgeau and Lelièvre, 1989).
The INSEE survey on Youth and Careers (Estrade and Thiesset, 1998) lends itself to this type of analysis because it describes retrospectively, year by year
The year is the unit of time used in this analysis...., the sequence of familial, residential and occupational events concerning the individual since the age of 16 (see Appendix 1).
Women are observed from the beginning of their ﬁrst union (marriage or cohabitation) until either the birth of their ﬁrst child, the separation from their ﬁrst partner if that takes place before the birth of a child, or (if neither of these events occurs) until the date of the survey. We limit the period of observation to a maximum of eight years of union.
Average age of mothers at the birth of their ﬁrst child, by level of education
Legend: see footnote 3 p. 312 for an explanation of the French educational categories.
Source: INSEE Family Survey, 1990.
Unemployment rates of women aged 15-29, by level of education
Legend: see footnote 3 p. 312 for an explanation of the French educational categories.
Source: INSEE Employment Surveys.
First, we consider here women aged 31 to 45 at time of survey (1997), who reported retrospectively at least one date of union or marriage. We eliminated those women who had a ﬁrst birth before the year of union. We will then study younger women (ages 24-29), to determine whether the trends observed have continued among the more recent birth cohorts.
Half of the women born between 1952 and 1966 had a first child during the initial three years of marital life
The ﬁrst sample includes 5,506 women born between 1952 and 1966, of whom 877 (16%) had no child during the period of observation (so-called “right-censored” data). They are representative of 5,660,000 women aged 31 to 45 in 1997. One woman out of two had a child during the ﬁrst three years of union (Appendix, section 2 and Table A).
The time spent in a childless couple depends on the woman’s own characteristics (her birth cohort, level of education, etc.), and also on the events that she experienced during this period and the context in which she lives. We have taken into account her social class background and some variables related to the circumstances of her studies and work experience. Other variables are missing from the data basis. Thus, the characteristics of the ﬁrst partner or housing conditions do not appear explicitly in the analysis, despite the fact that they play a role in the timing of fertility (Appendix, section 3). Moreover, unemployment may be correlated with some instability of the union. The fact that we take into account the date of separation from the ﬁrst partner when that happens during the period of observation tends to reduce this effect. It is logical to stop the observation at that time, since separation will, in all likelihood, alter the woman’s behaviour with respect to the timing of her fertility.
The different situations relative to economic activity are identiﬁed in the woman’s year by year record, during the life as a couple preceding the arrival of the ﬁrst child, that is, from the beginning of union to the year before the birth of the ﬁrst child for those women who had a child, or until the end of the observation period for the others (Appendix, section 2). What we study is then truly the inﬂuence of the economic context on the date of birth of the ﬁrst child, free of the reverse causation implicit in the inﬂuence of the arrival of the child on the employment status of the woman (though there is still some haziness when the birth and the beginning of union occur the same year). In the survey, periods of unemployment, of economic inactivity and of long-term employment are only identiﬁed if they lasted at least six months. An “intermittent” period of employment designates shorter periods of employment alternating with periods when the woman is not employed. Continuing and resuming one’s studies are combined in a single category, “studies.” The different types of employment status are not necessarily mutually exclusive. Up to three (consecutive or concurrent) may be encountered for a single year. All the statements are entered in the same manner in the analysis, and were not ranked a posteriori.
The youngest and the most educated women were in union longest without a child…
While half of the women born between 1952 and 1966 had a ﬁrst child during the initial three years of union, the median duration was only 2.1 years for women with no diploma or only a primary school certiﬁcate, but reached 3.7 years for women with the baccalauréat and 4.7 years for those with more than 2 years of higher education.
Moreover, among the cohorts born between 1952 and 1954, one woman out of two waited less than 2.4 years before having a ﬁrst child, whereas among the younger women (born between 1964 and 1966) this interval was one and a half times longer, or 3.7 years (Appendix, Table A). Between these two extremes, the median duration increased regularly from one cohort to the next.
The duration of union without a child thus depends both on the woman’s cohort and on her educational level, which is to some degree a proxy for the socio-economic background of the young woman (Galland, 1995). The non-parametric survival function shows the proportion of women still childless for a given duration of union in the absence of exit from observation. The curves in Figures 3 and 4 clearly show that the younger and better-educated the women, the longer they are likely to remain in a union without becoming mothers.
Proportion of childless women by duration of union and birth cohort (survival function)
Source: INSEE Youth and Careers Survey, 1997.
Proportion of childless women by duration of union and level of education (survival function)
Legend: see footnote 3 p. 312 for an explanation of the French educational categories.
Source: INSEE Youth and Careers Survey, 1997.
When we try to examine the inﬂuence of other variables, we quickly encounter a problem of inadequate sample size. We are thus obliged, in certain analyses, to summarise the preceding effects by regrouping different educational attainment categories — secondary school graduates and non-graduates — and different cohorts: we distinguish women born during the 1950s (that is, from 1952 to 1960) from those born during the 1960s (that is, from 1961 to 1966) (Table 1).
Table 1 –
Time elapsed between beginning of union and birth of first child, by cohort and level of education, according to woman’s activity(1) in the intervening period (median durations in years)
Women aged 31-45 – cohorts: Women aged 24-29 – cohorts: 1952-1966 1952-1960 1961-1966 1968-1973 Total Total Βac(2) or higher Less than bac Total Βac or higher Less than bac Total Βac or higher Less than bac No long-term employment 1.84 1.74 2.23 1.62 1.99 3.22 1.84 3.59 5.48 2.87 At least one period of long-term employment 3.46 3.08 3.89 2.65 3.96 4.66 3.63 5.26 6.06 4.46 No short-term employment 2.89 2.69 3.58 2.26 3.26 4.11 2.79 3.84 5.41 2.89 At least one period of short-term employment 4.32 3.75 4.81 3.14 4.80 5.85 4.12 5.51 6.44 4.65 No long-term unemployment 2.92 2.69 3.60 2.25 3.36 4.26 2.86 3.95 5.66 2.96 At least one period of long-term unemployment 4.48 4.27 5.48 3.94 4.68 5.88 3.99 5.49 6.01 4.91 Not involved in studies 2.78 2.58 3.30 2.31 3.10 3.70 2.87 3.52 4.79 3.10 Involved in studies 4.47 3.88 4.43 2.56 5.17 5.63 4.26 5.82 6.39 4.75 No inactivity 3.21 2.87 3.66 2.44 3.69 4.46 3.22 4.73 5.95 3.80 At least one period of inactivity 2.32 2.26 3.62 1.97 2.42 4.25 2.14 2.63 4.13 2.41 Total 3.02 2.76 3.66 2.34 3.48 4.45 2.96 4.40 5.80 3.52 (1) As used in the survey: continuous periods of at least 6 months for long-term employment, unemployment, and inactivity; periods of short-term employment alternate short-duration employment and unemployment. Multiple statuses may be experienced in the course of a single year. (2) Bac: Baccalauréat (upper secondary school diploma). Reading: Half of the women with at least the secondary school diploma, born between 1952 and 1966, and having experienced at least one long period of unemployment since beginning of union, had a first child within the first 5.5 years. Source: INSEE, Youth and Careers Survey, 1997.
A semi-parametric analysis (Cox model) was undertaken to control for the link between educational attainment and cohort. This makes it possible to study the duration spent childless and in union, taking account of the explanatory variables that we have introduced. The Cox model belongs to the family of proportional hazards models, and it offers the advantage of allowing the direct estimation of the effect of duration, without imposing a form other than the assumption that women’s characteristics exert a proportional effect on the chances of having a ﬁrst child. Its use thus requires only a (graphic) validation of the similarity of the curves representing the logarithms of the cumulative quotients as a function of duration
This condition was veriﬁed for the ﬁrst eight years....
In such an analysis, the curves can be inferred one from the other as a function of a multiplicative factor, or relative risk. The relative risks, which are interpreted in relation to a reference category that takes a value of 1, are presented in the tables of results of the models. This analysis conﬁrms that the differences due to each of the two variables (level of education and cohort) remain signiﬁcant when both variables are introduced together in the model.
…but this does not depend on the age at which the union began
For women born between 1963 and 1966, the median age at start of union ranges from 20.2 years for those who do not have even the CAP to 21.8 years for those with a CAP or BEP, 23.1 years for those with a baccalauréat and 24.4 years for women with higher education
The lowest diploma in the French educational system.... Yet the analysis shows that age at beginning of union does not have a signiﬁcant effect on the time elapsed before ﬁrst birth once cohort and educational attainment have been controlled for (Table 2). This surprising result indicates that these two variables, in summarising the characteristics of the past and the social background of the young woman, are sufﬁcient to explain the age at beginning of union. Consequently, including this age in the analysis, in addition to cohort and level of education, does not add anything.
Table 2 –
Estimated coefficients and relative risks of having a first child since the beginning of union, by cohort, level of education, and age at the start of the union (Semi-parametric analysis) – sample of women aged 31-45
Coefficient Relative risk Cohorts 1952-1954 0.15** 1.16 1955-1957 0.02 1.02 1958-1960 Ref. 1.00 1961-1963 – 0.13* 0.88 1964-1966 – 0.16*** 0.85 Level of education: More than two years higher education – 0.46*** 0.63 Two years higher education – 0.22*** 0.80 Baccalauréat – 0.17*** 0.84 BEPC, CAP, BEP(1) Ref. 1.00 None or primary school certificate 0.20*** 1.22 Age at beginning of union 16-19 years 0.07 1.07 20 -21 years – 0.06 0.94 22-24 years Ref. 1.00 25 years or more – 0.03 0.97 * Significant at the 5% level; ** Significant at the 5‰ level; *** Significant at the 1‰ level. (1) For an explanation of the French educational categories, see footnote 3. Reading: All other things being equal, women born between 1952 and 1954 are 1.16 times more likely to have had a first child since beginning of union than women born between 1958 and 1960 (the reference group). Source: INSEE, Youth and Careers Survey, 1997.
In contrast, other variables that are known to inﬂuence fertility behaviour play a signiﬁcant role here. For example, given the level of education and the cohort, women from large families of origin become mothers at an earlier age than the average, and it is known that they tend to have more children than women from small families (one or two children) (Leridon, 1985, p. 517). Moreover, women whose father is a manual worker or whose mother is not in the labour force have a ﬁrst child more rapidly than those for whom the mother or father is an executive. Finally, women living in rural areas or in cities of fewer than 20,000 people have a ﬁrst child a little sooner than residents of larger cities or Paris.
Women experiencing unemployment postpone the birth of a first child more…
Three quarters of the women born between 1952 and 1966 held at least one job for six months or more during the period of observation (that is, between the beginning of life together and the year preceding the birth of a ﬁrst child or the year when observation ended). Intermittent employment mixed with unemployment is most common among the youngest and least educated women. The requirement in the survey that in order to be counted as periods of unemployment there must be six continuous months of jobhunting explains why this situation is relatively uncommon, even though it affects nearly one woman out of ten. One woman in ﬁve pursues her studies after the beginning of union, and this is observed mostly among the best-educated women (more than one in three). A spell of at least six months out of the labour force was reported by more than one in ﬁve of the least educated women (those who had not passed the baccalauréat) (Appendix, Table B).
Compared with women of the same cohort and educational level, those who had always been in the labour force delayed having a ﬁrst child longer after the start of union. But this prolongation of the period as a childless couple is greater for the women who experienced intermittent periods of employment than for those who report only continuous employment. Continuous periods of unemployment postpone the arrival of a ﬁrst child the most (Table 1, ﬁgure 5).
Proportion of childless women by duration of union and various work-and education-related situations, for the birth cohorts of the 1950s and the 1960s (survival function)
Source: INSEE Youth and Careers Survey, 1997.
…than those who experienced stable or even intermittent employment
Among the women born between 1952 and 1966 who had at least one long spell of employment, half had a ﬁrst child within 3.5 years of life in union. This median duration amounts to 4.3 years for the women who experienced short or intermittent periods of employment, and 4.5 for those who experienced six months of continuous unemployment at least once. These results vary by cohort and educational level (Table 1).
As indicated earlier, these statuses with respect to employment are not mutually exclusive, and their effects may be combined. In order to better visualise the impact of unemployment (in the sense of the survey, i.e. for a spell lasting at least six months), we measure the joint effect (model A) and the net effect (model B) of the situations: “have been employed” (in the broad sense) and “have been unemployed”, holding constant the cohort and level of education, in a semi-parametric proportional hazards analysis (Cox model). The effect of unemployment on ﬁrst birth postponement is conﬁrmed for women in the labour force (Table 3).
Table 3 –
Estimated coefficients and relative risks of having a first child since the beginning of union, by cohort, level of education, and activity experiences (Semi-parametric analysis). sample of women aged 31-45
Model A Model B Coefficient Relative risk Coefficient Relative risk Cohorts 1952-1954 0.11* 1.12 0.11** 1.12 1955-1960 Ref. 1.00 Ref. 1.00 1961-1966 – 0.12*** 0.89 – 0.12*** 0.88 Level of education: More than two years higher education – 0.55*** 0.58 – 0.50*** 0.61 Baccalauréat to bac + 2 – 0.22*** 0.81 – 0.22*** 0.80 BEPC, CAP, BEP(1) Ref. 1.00 Ref. 1.00 None or primary school certificate 0.13*** 1.14 0.19*** 1.21 Activity status(2) Employment and unemployment – 1.16*** 0.32 Employment without unemployment – 0.55*** 0.58 No employment Ref. 1.00 Experienced unemployment – 0.41*** 0.67 Did not experience unemployment Ref. 1.00 Experienced inactivity 0.15*** 1.16 Did not experience inactivity Ref. 1.00 * Significant at the 5% level; ** Significant at the 5‰ level; *** Significant at the 1‰ level. (1 For an explanation of the French educational categories, see footnote 3. (2) Long periods (of at least 6 months) experienced since the year of the union and before the birth of a child. Reading (model A): All other things being equal, a woman with only a primary school education is 1.14 times more likely to have given birth to a child since the start of union than a woman with some secondary education: BEPC, CAP, BEP (reference group). By contrast, the chances of giving birth are 1.2 times smaller (relative risk of 0.81) for women with a baccalauréat or two years of higher education. Source: INSEE, Youth and Careers Survey, 1997.
These results may, however, be criticised: the longer the duration since the couple has been established, the greater the likelihood that the woman will have experienced one of the enumerated statuses (employment, unemployment…). There is thus a possible bias in considering variables for which the intensity is correlated with the observed duration. To address this issue, the event “experienced unemployment” can be analysed as a time-dependent variable in a Cox model, that is, the event is taken into account only from the year when it took place. The results are consistent with the earlier conclusions, but continuous unemployment, in the sense of the survey, is too rare an event when considered on a year-by-year basis, holding constant cohort and level of education, to yield statistically signiﬁcant results at the levels considered earlier.
Women rarely experienced a period of continuous unemployment during the year in which they entered a union (5% of the sample, Appendix, Table B). This observation is linked to the fact that unemployed women delay this entry (Ekert and Solaz, 2000). If one considers the activity status for the year in which the couple was established, unemployment does not have a signiﬁcantly different impact from that of continuous employment. In contrast, if the distinction is made between having experienced unemployment for the ﬁrst time before or after beginning of union, the results are unambiguous. Unemployment clearly delays the arrival of the ﬁrst child more markedly if the women have experienced unemployment when they were already living in a couple (Table 4).
Table 4 –
Estimated coefficients and relative risks of having a first child since the beginning of union, by cohort, level of education, activity status(es) in year of beginning of union (model C), or exposure to unemployment before or after the beginning of union (model D) (Semi-parametric analysis) sample of women aged 31-45 years
Model C Model D Coefficient Relative risk Coefficient Relative risk Cohorts 1952-1954 0.13** 1.14 0.11* 1.12 1955-1957 0.02 1.02 Ref. 1.00 1958-1960 Ref. 1.00 0.01* 1.011961-1963 – 0.11* 0.90 – 0.11* 0.89 1964-1966 – 0.13* 0.88 – 0.13* 0.88 Level of education: More than two years higher education – 0.41*** 0.66 – 0.48*** 0.62 Two years higher education – 0.19*** 0.82 – 0.24*** 0.79 Baccalauréat – 0.18*** 0.83 – 0.18*** 0.84 BEPC, CAP, BEP(1) Ref. 1.00 Ref. 1.00 None or primary school certificate 0.14*** 1.15 0.21*** 1.23 Activity status in year of beginning of union Inactive 0.35*** 1.41 Unemployed – 0.01 0.92 Student – 0.12** 0.88 Short-term work – 0.16** 0.85 Long-term work Ref. 1.00 Experience of unemployment Prior to union – 0.10* 0.90 Since beginning of union – 0.54*** 0.58 None Ref. 1.00 * Significant at the 5% level; ** Significant at the 5‰ level; *** Significant at the 1‰ level. (1) For an explanation of the French educational categories, see footnote 3. Reading (model C): All other things being equal, a woman with only a primary school education is 1.15 times more likely to have had a first child since beginning of union than a woman with some secondary education: BEPC, CAP, BEP (reference group). By contrast, the chances of giving birth are 1.2 times smaller (relative risk of 0.83) for women with a baccalauréat. Source: INSEE, Youth and Careers Survey, 1997.
Unemployed women and homemakers do not have the same family behaviour
Conversely, the women who experienced periods of inactivity spent less time in childless union than other women. Half of them had a child after 2.3 years of life together (as against 3.2 years for women who were not inactive). Once again, this result depends also on the cohort and the educational attainment (Table 1).
The impact of inactivity is very strong (Table 4). This result seems to indicate that women who are not in the labour force during the year when they start their union plan, more than other women, to have a child quickly. As Laurent Toulemon has suggested (in Leridon, 1994, p. 174), “for women, forming a couple before starting to work reﬂects a choice that privileges family and children.” By contrast, continuing studies while in union delayed the ﬁrst birth for women aged 31 to 45 years. This result is more ambiguous when broken down by cohort and level of education and the inﬂuence of further studies is compared with that of employment or unemployment. Women born in the 1950s who pursued studies while they lived in a union postponed the arrival of their ﬁrst child less than those who were unemployed. In contrast, students born during the 1960s delay even longer than the members of the same cohorts who are in the labour force (Table 1).
Are women under 30 following in the footsteps of their elders?
Now let us look at the impact of employment status on the birth of a ﬁrst child for the women aged 24-29 interviewed in the Youth and Careers Survey of 1997. The employment histories of women under 30 were collected following a slightly more detailed procedure than for the older women, but these modest differences do not hinder the comparison of results with those of the older women (see Appendix, sections 1 and 2). This complementary study is based on data for 1,565 women born from 1968 to 1973 who were living or had lived in a union at time of survey. They represent approximately 1,700,000 women in these cohorts. In this sample, the “right-censored” observations (770, being 49.2% of the total) are clearly more frequent than in the preceding group. Being younger, these women are much more likely than their elders to have lived in a union for less than eight years at time of survey. They also remained childless more often during the period of observation, and have more frequently left their ﬁrst partner (Appendix, Tables A and B). Since the sample size is smaller for these younger cohorts, educational levels and cohorts had to be regrouped. We distinguish those who obtained the baccalauréat from those who did not, and those born between 1968 and 1970 from those born between 1971 and 1973 (Figure 6).
Proportion of childless women by duration of union, birth cohort, and level of education, for the younger women (survival function)
Source: INSEE Youth and Careers Survey, 1997.
Half of the women born between 1968 and 1973 waited at least 4.4 years before having a first child
Certain trends that may be observed among the women under 30 are in fact extensions of the trends described above for the older women. Thus, among the women aged 24 to 29 (born from 1968 to 1973), one out of two had a ﬁrst child after 4.4 years of life in union (as compared to 3.0 years for the cohorts born between 1952 and 1966), and the differences by cohort persist: women born at the end of the 1960s had a ﬁrst child more rapidly after the start of their union (4.1 years) than those born at the beginning of the 1970s (4.9 years) (Table A).
Having obtained a diploma appears to be a strong differentiating factor, and the time spent in a childless union tends to increase more markedly from one cohort to the next when the woman has graduated from secondary school. Yet pursuing studies deters a certain number of young women from forming a couple. In this regard, our observations are evidently incomplete. Moreover, at these ages (24 to 29 at time of survey), a certain number of secondary school graduates have not yet completed their higher education, and this may help explain the substantial delay preceding a ﬁrst birth among the youngest secondary school graduates in union.
Those under 30 have been much more markedly affected by the difﬁcult state of the labour market than their elders. Thus these young women are proportionately three times more likely than the older women to have experienced unemployment or periods of intermittent employment between the beginning of union and the year preceding the birth of a ﬁrst child or the end of observation (Table B). These precarious situations in the labour market are even more frequent among the less-educated women and the younger women, born between 1970 and 1973. Conversely, having held a job for a long time is much less common (47% of women born from 1968 to 1973 during the period observed following establishment of the couple, as compared to 77% for the women born between 1952 and 1966). Women under age 30 are also twice as likely to continue their studies after the start of the union. In contrast, only one woman out of ten has been inactive during the period observed and only 8% were out of the labour force during the year when they began their union, whereas 16% experienced unemployment that year (Appendix, Table B).
Among the younger cohorts, unemployment has a stronger influence on the less educated women, comparable to the effect of intermittent employment
As was the case for their elders, the median duration between beginning of union and birth of a ﬁrst child is clearly longer for the women from the younger cohorts who are in the labour force or still in school, as compared to the economically inactive. Half of the young women who experienced a spell of inactivity had their ﬁrst child within 2.6 years of the beginning of the union, while this median is longer by two years for those who did not experience this situation (Table 1, Figure 7).
Proportion of childless women by duration of union, employment experience and level of education for the youngest cohorts (survival function)
Source: INSEE Youth and Careers Survey, 1997.
The time spent in childless union is shorter when the woman has little education and when she has withdrawn from the labour force. The effect of inactivity is especially notable for the women who were already non-working homemakers during the year when the union began (Table 5, model F). Other activity statuses during that year are not signiﬁcant in the models, despite the fact that the frequency of unemployment, studying, and short employment spells is more important for the young women of this sample than for their elders.
Table 5 –
Estimated coefficients and relative risks of having a first child since the beginning of union, by cohort, level of education, activity experiences since the start of the union (model E), or activity status(es) in year of beginning the union (model F) (Semi-parametric analysis) sample of women aged 24-29
Model E Model F Coefficient Relative risk Coefficient Relative risk Cohorts 1968-1970 Ref. 1.00 Ref. 1.00 1971-1973 – 0.19* 0.83 – 0.23** 0.80 Level of education More than two years higher education – 1.30*** 0.27 – 1.23*** 0.29 Baccalauréat to bac + 2 – 0.46*** 0.63 – 0.42*** 0.65 BEPC, CAP, BEP(1) Ref. 1.00 Ref. 1.00 None or primary school certificate 0.22* 1.25 0.15 1.16 Activity status in year of beginning of union Inactive 0.79*** 2.19 Unemployed 0.02 1.02 Student Ref. 1.00 Short-term work – 0.07 0.93 Long-term work – 0.04 0.96 Activity experiences Experienced unemployment -0.45*** 0.64 Did not experience unemployment Ref. 1.00 Experienced inactivity 0.46*** 1.59 Did not experience inactivity Ref. 1.00 * Significant at the 5% level; ** Significant at the 5‰ level; *** Significant at the 1‰ level. Reading (model E): All other things being equal, a woman with only a primary school education is 1.25 times more likely to have had a first child since beginning of union than a woman with some secondary education: BEPC, CAP, BEP (reference group). By contrast, the chances of giving birth are 1.6 times smaller (relative risk of 0.63) for women with a baccalauréat or two years higher education. Source: INSEE, Youth and Careers Survey, 1997.
As for the preceding sample, we tested the inﬂuence of different activity statuses (unemployment, stable employment, intermittent employment, studies) on the duration between beginning of union and the year preceding the birth of a ﬁrst child. It is clear that each of these situations lengthens substantially the time spent in a childless couple, for a given educational level and cohort, as compared to inactivity. The women with the highest educational attainments remain childless for longest, regardless of their employment status.
The inﬂuence of unemployment is very clear for the least educated w omen (Table 1). Women without a secondary school diploma (baccalauréat) who have experienced unemployment since beginning of union delay the birth of their child by almost 2 years more than those who have not experienced unemployment: the median durations are 4.9 years and 3 years.
The delaying effect of long-term employment is also stronger among those without a secondary school diploma than among those with one. As for periods of intermittent employment, rather than having an intermediate effect between that of long employment and unemployment as was the case among women born between 1952 and 1966, they delay the arrival of a ﬁrst child among the younger cohorts born between 1968 and 1973 by as much as unemployment: the median duration before a ﬁrst birth reaches 5.5 years (Table 1, ﬁgure 7).