1 – Relation between first cohabiting union and first birth

4 In Europe, family formation (first union formation, first marriage, and first child) occurs later than in the past, and the pathways followed are more diverse. The first union also less often begins with a marriage, and non-marital births are more frequent, especially in countries of Western and Northern Europe (Sobotka and Toulemon, 2008; Billari and Liefbroer, 2010; Coleman, 2013). While fewer first unions now result in a first birth (Winkler-Dworak and Toulemon, 2007; Hayford et al., 2014), first union formation and first childbirth nonetheless remain closely linked (Baizán et al., 2003, 2004). Singlehood, on the other hand, remains associated with permanent childlessness (Robert-Bobée, 2006; Mikolai, 2017; Saarela and Skirbekk, 2020), and births outside a cohabiting union are rare, in France especially (Toulemon, 1994; Vergauwen, 2016). Childbearing is thus an expected component of a stable relationship as identified by co-residence in the same dwelling (Hobcraft and Kiernan, 1997; Régnier-Loilier and Solaz, 2010). Hence, in the life course of an individual, the first cohabiting union is seen as the first opportunity to start a family (whether or not a child is born).

5 The relation between age at first cohabiting union and first birth in France has already been studied. The probability of remaining permanently childless increases, for example, with age at first union (Robert-Bobée, 2006). Using data from the Family History survey (INSEE, 1999), Bessin et al. (2005) showed that ‘older’ parents (who have a first or higher-order birth above age 40) form a first union 2 years later, on average, than younger parents. Mazuy (2006) used these same data to show that the interval between entry into first union and first birth is shorter when men form a union with a woman who is already a mother (2.5 years on average) than when she is childless (3–4 years). This interval is also longer for the cohorts born after the 1940s than for those born before (Mazuy, 2006). These studies are based on descriptive analyses and cannot provide insights into the net effect of age at union formation on the occurrence and timing of first births. The research presented in this article follows on from these earlier studies, applying a multivariate approach to provide a more in-depth analysis of this relation for both women and men.

2 – Early first cohabiting unions and first births

6 The occurrence and timing of a first birth varies according to the moment in the life course when the first cohabiting union is formed. There may be many reasons for this, linked to the biological, social, and normative dimensions of age. A young age at first cohabiting union (around 20 years) may, for example, reflect a desire to have children early (Baizán et al., 2003, 2004) and to have a large family (Tomkinson, 2019). These familialist aspirations (to marry and have children) may also reflect an environment where early childbearing is the norm—among families with religious convictions (Prioux, 2003; Philipov and Berghammer, 2007; Régnier-Loilier and Prioux, 2009) or with a low educational level (Prioux, 2003), for example. These factors suggest that relatively early union formation will lead to a rapid first birth. If so, it is not age that influences births directly, but rather the characteristics of the individuals who form early unions.

7 Conversely, early union formation may also be associated with a longer interval before the first birth, when the partners have not completed their education, for example, or have not yet found stable employment. Being unemployed or in education contributes to birth postponement (Skirbekk, 2008; Pailhé and Régnier-Loilier, 2016): young couples may wait for longer than others before having a first child. Likewise, between completion of education and labour market entry, couples may wish to take advantage of their freedom before achieving the necessary material and economic stability to start a family. All these factors are contributing to the trend towards later childbearing (Régnier-Loilier, 2007). Early cohabitation today may also be more closely linked to economic and material issues than to childbearing plans (Costemalle, 2015).

3 – Late first cohabitation and first birth

8 While cohabitation is relatively disconnected from childbearing in early adulthood, this is less true at later ages when couples more frequently have a child soon after moving in together (Mazuy, 2006). Age 30 is often a watershed beyond which living with a partner and having children is relatively frequent. Examining singlehood in France, Bergström et al. (2019), for example, showed that between ages 26 and 65, the lowest rates of singlehood (not in a cohabiting or non-cohabiting union) are observed between ages 30 and 35. Compared to younger singles, fewer partnerless men and women in this age group report being single through choice, and they often feel out of step with their peers. The desire to ‘be in a couple’ may also stem from a desire to have children, with age being a source of pressure, for women especially (Bergström et al., 2019). When a couple forms at around age 30 with the aim of having a child, the pressure of passing time and of approaching biological limits may lead the partners to bring forward a conception, even if the union is recent, in response to social norms defining the ‘right age’ to start a family. According to the 2010 Fecond survey (INED–Inserm), the ideal age for having a child reported by respondents is 26 years on average for women and 28 years for men. Once a couple has become ‘established’ (in material and affective terms) and comes close to an age at which childbearing is the norm, pressure to have a first child, exerted by family and friends, for example, may grow stronger (Bernardi, 2003). Wanting to have a child, and being urged to do so by social norms, may encourage individuals reaching their 30s to form a union in order to start a family at an age close to the ideal, well before it becomes too late (Billari et al., 2011).

9 These desires may also be linked to awareness of the growing risk of sterility as age increases, for women especially. Between ages 20 and 30, women’s capacity to bear a child remains practically constant, but it starts to decline from around age 35 (Dunson et al., 2002). Over age 40, and especially over age 45, few women seeking to conceive can bring a pregnancy to term (Toulemon, 2004). Women in their 30s who try to have a child soon after forming a union are thus more likely to become mothers, while those who wait longer are at greater risk of remaining childless (Leridon, 2017).

10 It is also interesting to examine the characteristics of individuals who form a ‘late’ first union. This tardiness may be due to difficulties in finding a stable partner with whom to start a family. The time taken to find the ‘right’ person is often the main reason given for being childless or having a first child after age 35 (Cooke et al., 2012; Schytt et al., 2014). A lack of desire for parenthood may also explain later union formation and a greater propensity to remain childless. Social background also plays a role: the most highly educated women form unions at later ages and more often remain permanently childless than the least educated women (Berrington, 2004; Robert-Bobée, 2006). Last, the structure of the marriage market evolves with age, and the probability of forming a union with a partner who already has children increases (Goldscheider et al., 2009; Beaujouan, 2011), reducing the likelihood of an additional birth. The partner who is a parent may not want another child, while for the other partner, assuming the role of step-parent may lead to second thoughts about having a child of their own. Childbearing after late union formation is thus closely linked to the characteristics of the partners and of the union. These characteristics may explain the timing of a first birth, rather than the pressure of age per se. Multivariate analyses will be used to examine these various factors.

4 – Man’s age, woman’s age, partners’ age

11 While an individual’s age at union formation is an important factor, so is that of the partner (Fraboni and Rosina, 2006). As age increases, while women tend to form heterosexual unions with partners of a similar age, men more frequently choose younger women (Bozon, 1990; Beaujouan, 2011). These gendered patterns of union formation may affect both the probability of a birth after a later first union and the timing of this birth.

12 In addition, the decrease in fecundity with age concerns men as well as women, although much later in life (La Rochebrochard et al., 2006). Late union formation (around age 30) is not always a ‘choice’, for women especially (Bergström et al., 2019), while for men, the social and biological pressures associated with advancing age are less keenly felt (Billari et al., 2011). This suggests that age may be a weaker factor of accelerated parenthood for men than for women. However, while men face fewer biological pressures than women, they may want to have a first child soon after a late union to avoid being ‘too old’, for example, or because they want to have enough ‘energy’ to raise a child. Their partner’s age may also be a factor (Bessin and Levilain, 2012). This article contributes to discussion on these aspects by focusing on the fertility behaviours of men and women after union formation.

1 – The EPIC survey

13 This study of the link between age at first cohabiting union and first birth is based on data from the EPIC survey conducted on a sample of 16,000 dwellings drawn from the 2012 annual census survey. A total of 7,825 men and women aged 26–65 were interviewed, whether single or in a union. The respondents are our study units and are treated separately by sex. The EPIC survey contains information on their characteristics, their fertility, and their union histories, giving the number, the order, and certain characteristics of their intimate relationships (Rault and Régnier-Loilier, 2019). For the first time, it also records the interval between the start of the relationship and the start of cohabitation, and the partners’ education and employment situation at the time they first met, making it possible to explore the link between these factors and the arrival of the first child.

14 While individuals may have several intimate relationships before moving in with a partner, many see cohabitation as a necessary condition for starting a family. First births to non-cohabiting couples or to women without a partner are rare (Vergauwen, 2016). For this reason, our study focuses on cohabiting unions, [1] above all the first union, seen as a key stage in family formation. This corresponds to the respondent’s first experience of cohabitation but may be a second or higher union for his or her partner. In what follows, entry into union or union refers to the first cohabitation. Fertility is studied in relation to the respondent’s first birth, according to his or her age at first cohabiting union.

15 To study the link between age at first cohabitation and first birth, our analysis is limited to cohorts who have reached the end of their reproductive lives, [2] especially those born between 1948 and 1970 who were aged between 43 and 65 years at the time of the survey (n = 4,901). [3]

16 As our study focuses on first births after a first entry into union (whether the birth occurs within this union or afterwards), people who had a child before this union are excluded (n = 354). To limit the number of censored observations, people with no experience of a cohabiting union are also excluded (n = 359), as are those who entered a union towards the end of their reproductive life. [4] Our final samples thus included 2,352 women and 1,764 men.

17 Among the selected cohorts, 92% of the women and 91% of the men with experience of cohabitation become parents after their first entry into union, and 90% of first births occurred within this first union rather than within a subsequent union or outside a union. A large share of individuals forms a first cohabiting union for the first time between ages 20 and 23 (43% of women and 46% of men, Table 1). Just 4% of women’s and 7% of men’s first unions occur between ages 32 and 40. The timing of cohabitation and parenthood is later for men: they are 2.2 years older than women, on average, when they form their first cohabiting union, and 2.8 years older when they have a first child.

Table 1. Distribution of age at entry into cohabiting union and mean age at first union and first birth by sex

Description

Women Men Age at entry into union, % (n) Under 20 32.8 (702) 10.9 (190) 20–23 42.5 (1,034) 46.0 (816) 24–27 15.3 (398) 25.6 (460) 28–31 5.6 (130) 11.1 (189) 32–40 3.9 (88) 6.5 (109) Total 100.0 (2,352) 100.0 (1,764) Mean age (years) First union 22.2 24.4 First child 25.5 28.3

Table 1. Distribution of age at entry into cohabiting union and mean age at first union and first birth by sex

Note: Weighted percentages (unweighted numbers).
Coverage: Men and women who were childless at entry into first cohabiting union (1948–1970 cohorts).

2 – Methods for analysing the frequency and timing of first births, and associated factors

18 Here we take account of both the occurrence of a first birth after entry into cohabiting union and the timing of this are linked birth. In both cases, the aim is to determine whether age at cohabitation and first birth are linked after controlling for other variables. Having a child or not is estimated using logistic regressions that model the occurrence of a first birth up to the end of reproductive life. As people who form a first union later in life have less time to conceive than the others, the analysis is also based on identical observation windows of 2 years and 5 years following entry into union. Hence, individuals who enter a union at age 40 (latest age considered here) are 45 years old at the end of the observation period.

19 The timing of the first birth after first entry into union, whether it occurs within this union or afterwards, is studied using discrete-time parametric duration models with a log-normal specification among individuals who have had a first child. [5] By selecting only people who have had a first child, we can obtain coefficients that represent duration, without being affected by the occurrence or not of a first birth. These models also take account of time-varying characteristics, such as union status.

20 These models estimate the association between the respondent’s first union characteristics and the first birth, in relation to the variables described in the literature review. The following are taken into account: age at first cohabitation; partners’ educational and employment status when they first met (this information is not available for start of cohabitation); partner’s parental status (already a parent); age difference between partners; time between start of relationship and cohabitation; and status of cohabiting union when it was formed (direct marriage or not). Direct marriage may be linked to more traditional family values, and religious practice likewise. This variable is also taken into account. The models also include year of birth and social background (father’s occupational category) to take account of possible cohort effects and control for any intergenerational transmission of family preferences.

21 To take account of union dissolutions, the variables in the duration analysis are time-invariant, except for partnership status, which may vary from one year to the next. The models used do not consider unobserved characteristics liable to influence both union formation and fertility behaviour. These possible selection effects are discussed in the Conclusion.

1 – Less frequent first births among later first cohabiting unions

22 The descriptive analysis shows that first births become less frequent as age at entry into first cohabiting union increases (Figures 1A and 1C, shaded area), especially after age 27 for men. Among women, 65% of those who form a first union between ages 32 and 40 have a child versus 92% of those who form a union before age 20. This means that a larger number of women who form a late first union are childless at the end of their reproductive life. For men, the difference is smaller: the proportions are 73% and 92% at the same ages. According to estimates of the risk of sterility by age of childless women, more than 96% of these women could have children at age 20, 88% at age 32, 83% at age 35, and 70% at age 40 (Toulemon, 2004). We can thus consider that among the youngest women who formed a cohabiting union in the 1948–1970 cohorts, practically all those who were not sterile had a child during their reproductive life. At older ages, on the other hand, even after taking account of the time between union formation and first attempts at conception, the proportion of women who have children is always below what would be possible in purely biological terms. Other reasons, such as social background, weaker desire for children, partner already a parent, for example, may explain this decrease.

23 While transitions to parenthood during reproductive life become less frequent with increasing age at entry into union, the decline is less marked during the first few years after the start of cohabitation. For women who start a cohabiting union at ages 32–40, first births even appear to be more frequent in the 2 years following entry into union than for women who move in with their partner at ages 28–31 (Figure 1A), but this difference is not significant (following section). For men, no decrease in the frequency of first births with age is observed in the 3 to 5 years following union formation. This frequency increases until age 23 at first union formation (Figure 1C) and then levels off. Thus, age does not seem to have a strong impact on becoming a parent in the first 5 years.

24 Two mechanisms may explain these different observations, depending on whether the occurrence of births is estimated in the first years following union formation or at the end of reproductive life. A share of the people who form late unions may not want to have children. At the same time, these individuals are more often reaching their reproductive age limit, while younger people can postpone a first birth beyond 5 years after union formation.

Figure 1. Proportions of first births by time since entry into first cohabiting union and a possible first birth, by age at cohabitation

Figure 1. Proportions of first births by time since entry into first cohabiting union and a possible first birth, by age at cohabitation

Interpretation: Among women who entered a union at ages 20–23, 70.6% had a first child within 5 years, and 94.4% had a child at some time in their reproductive life (Figure 1A); 74.8% of fertile women who entered a union at ages 20–23 had a child within 5 years (Figure 1B).
Notes: The shaded area corresponds to becoming a parent, whatever the timing of first births. The lines correspond to the probabilities of having a first child by time since first union formation. The data are weighted.
Coverage: The left-hand figures concern men and women who were childless at first cohabiting union (1948–1970 cohorts). The right-hand figures concern individuals who became parents.

25 Looking solely at persons who have had a child, we see that for fathers, the proportion of births occurring 3 to 5 years after first entry into union increases sharply with age at union formation. At shorter durations (0 to 2 years), a significant acceleration is only observed between ages 27 and 32 (Figure 1D). For mothers, the proportion of births in the 3 to 5 years after moving in with a partner increases gradually with age. At short durations, the proportions of births decrease up to age 27 and then increase at older ages (Figure 1B). Beyond age 30 at union entry, it appears that men and women who want to have a child seek to do so without delay.

26 These proportions were calculated without taking account of union dissolutions as we are interested in the occurrence of first births and their timing with respect to the start of the partnership trajectory. In the cohorts studied, most births occurred during the first reported cohabiting union. These proportions of first births by time since union formation were also calculated for intact couples using union ‘survival’ tables [6] (Appendix Figure A.1) and show comparable tendencies by age at union formation. If separation is taken into account, the low proportion of births following entry into union disappears for men below age 20 at union entry. This suggests that a separation postpones the occurrence of a birth beyond 5 years after first union formation more frequently at young ages. Apart from this exception, whether or not separations are taken into account, the conclusions of the descriptive analysis are similar.

2 – First birth after first entry into cohabiting union by age difference between partners

27 The probability of having a child depends not only on age at first cohabiting union but also on the partner’s age. Both the man’s and the woman’s age appear to influence entry into parenthood and the timing of a first birth. For late parents, is the birth accelerated when the partner’s age is also advanced, and less so when individuals are partnered with someone younger or the same age as themselves? Does this have the same consequences for both men and women? Figure 2 shows the same analyses as before, but by the age difference between the man (be it the respondent or the partner) and the woman.

28 When the age difference between partners is small, or the woman is older, the shape of the curves representing the occurrence of a first birth up to 5 years after union formation are quite similar for men and women (Figures 2A and 2C). In both cases, they are flat and then start falling from age 32. In couples of this type, men and women seem to have similar fertility behaviours by age. Having a partner who is at least 3 years older, on the other hand, produces, for women, a rapid decrease in first births with age at entry into union (Figure 2B). This could be explained by the increase with age in the probability of forming a union with someone who is already a parent (Goldscheider et al., 2009), which could correspond to (or give rise to) a weaker desire for an additional child. Conversely, for men in a union with a younger woman, the proportion of births within 5 years increases with age (Figure 2D). Moreover, men who form a union with a younger woman after age 32 more often have children than those whose partner is of a similar age. In fact, the later men form their first union (Appendix Figure A.2), the more frequently they partner with a woman who is younger on average (while women tend to partner with men of a similar age). Men who form a late first union with a woman of a similar age may also have a weaker desire for children than men whose partner is younger, and women who are older than their partner may not want a child or may already have children and not desire any more.

Figure 2. Proportions of first births by time since entry into first cohabiting union and a possible first birth, by age at cohabitation and age difference between partners

Figure 2. Proportions of first births by time since entry into first cohabiting union and a possible first birth, by age at cohabitation and age difference between partners

Interpretation: Among women who entered a union at ages 20–23 with a partner of a similar age, 96.2% had a first birth, and 69.3% became mothers within 5 years.
Notes: Age difference is the difference between the man’s age (respondent or partner) and the woman’s age (respondent or partner). Partners are considered to have similar ages when the age difference is less than 3 years. Few women are more than 3 years older than their partner. We then identify the couples where the man is at least 3 years older than his partner.
The shaded area corresponds to becoming a parent, whatever the timing of these first births. The lines correspond to the probabilities of having a first child by time since first union formation. The data are weighted.
Coverage: Men and women who were childless at entry into first cohabiting union (1948–1970 cohorts).

29 The descriptive analyses show that the woman’s and the man’s age at first cohabitation influence the occurrence of first births and the timing of entry into parenthood. Certain results may reflect family situations that are more frequent at these ages (presence of the partner’s children, for example) or individual characteristics. These questions are examined in the following sections using multivariate analyses that control for different characteristics to verify the observed link between age and births.