1It has been frequently noted in the literature that fertility in Britain and in France has converged in quite different socio-economic environments (Calot, Girard and Leridon, 1976; Ekert-Jaffé, 1990; Hantrais, 1997). For nearly the past 20 years, the total fertility rates (TFRs) have been very similar in the two countries (Craig, 1992; Lelièvre, 1995; Coleman, 1996). They fluctuate in the region of 1.8 children per woman, and are among the highest TFRs of the European Union. Other demographic trends too are remarkably similar: for example age at marriage, non-marital fertility, and childlessness have all increased steadily since 1970. Yet as regards the role of the state, social organization, family policies and the role of women, the two countries diverge considerably. On the one hand, Britain constitutes the prototype of liberalism in Europe, with state intervention reduced to a minimum (Esping-Andersen, 1996). On the other hand, France, the cradle of the policy of pro-natalist family benefits in democracies, is ranked among the countries with strong involvement of the pro-family state in the social system (Kaufmann, 1997). Benefits are especially large for families with three children and for young children, and policy seeks to assist women with childcare so as to reconcile childbearing and career.
2In these conditions, similar demographic behaviour in the two countries might suggest that family policies have little influence on fertility. But the similarity of overall period indices hides some profound differences at more detailed levels. For example, Lelièvre (1995) emphasizes the divergence in the early stages of family formation. In Britain, the high level of births among women under the age of 20 (most frequently outside of marriage) is unique in Europe. Cohabitation usually involves young women with relatively little education and from the lower socio-economic groups (Lelièvre and Kiernan, 1995). In France, by contrast, cohabitation typically involves better-educated women with a socio-economic position superior to that of their partner (Ekert and Sofer, 1991). The distribution of families according to number of children is also very different in the two countries. As elsewhere on the continent, families with two children are the most prevalent in France. In the United Kingdom, however, the greater dispersion in the tempo of childbearing (Chandola, Coleman and Hiorns, 1999) is associated with a greater dispersion of family sizes (Coleman, 1996; Lelièvre, 1995; Pearce, Cantisani and Laihonen, 1999). Thus there are more childless women and more large families in Britain than in France. Finally, fertility differs by men’s social class (Munoz-Pérez, 1987).
3The purpose of our study is to pursue this analysis in comparing fertility in the two countries by birth order according to the age and socio-economic characteristics of women and of their partner. We thereby explore possible effects of state policy. We recall how, according to economic theory, family policy in France should have the effect of attenuating differences between social groups. Our research will seek to determine:
- if fertility and its timing are also strongly linked to socio-economic factors in both countries;
- if greater social polarization in Britain can explain the greater dispersion of births over the course of the life cycle;
- how the influence of socio-economic characteristics on fertility has evolved for the most recent cohorts.
4We compare the growth of families and the timing of births by order (Table 1) in France and in England and Wales [1] according to different characteristics: marital status, the socio-occupational category of women, and that of their partner, if any. We first study the cohorts of women born in the 1950s, observed up to ages 39-44 [2]. We then assess the changes that took place during the 1980s and the 1990s, by comparing this cohort with the history of women born ten years later, for whom the observations are limited to ages 29-34 [3].
Average age at first, second and third birth and proportion with one, two and three or more children, for three cohorts of women aged 40-44, 35-39 and 30-34, respectively, in 1995 in France and in 1998 in England and Wales
Average age at first, second and third birth and proportion with one, two and three or more children, for three cohorts of women aged 40-44, 35-39 and 30-34, respectively, in 1995 in France and in 1998 in England and Wales
5In a first part devoted to family policy and its theoretical effect on fertility, we discuss the major differences in the policies of the two countries. Then we present the point of view of economists on the influence on fertility of the income of each of the partners and we see how family policy can affect the timing of fertility and induce specific birth-spacing strategies.
6The second part describes the social environment in the two countries. We first discuss the differences that exist in the classification of occupations, and in the qualifications and the hierarchical positions, and then study the differences in the evolution of the social structure over ten years, for all the cohorts.
7The third part describes the method of analysis—estimation of logit models examining parity progression ratios and annual probabilities of giving birth by rank of child as a function of the woman’s characteristics—and discusses the results from the models.
I – Family policies and their effects on fertility
1 – Different family policies
8In the land of Lord Beveridge [4], despite his ideal of universality, public aid is concentrated on the very poorest and seeks to maintain their standard of living at the threshold of poverty. Retirement payments and universal allowances do not exceed this threshold; maternity leaves last 18 weeks [5], but are only compensated up to an average 45% of salary (Gauthier, 1996, p. 174). The British “social state” thus conforms to the liberal model of Esping-Andersen (1996). In this model, men are resource providers and women devote themselves to domestic work (Sainsbury, 1999). The state abstains from intervention in couples’ lives (Gauthier, 1996) and assistance to families remains small for each child, even though it represents a sizeable share of the social budget [6]. Until the end of the 1990s, childcare was primarily a private matter for couples. There were no or few day-care centres, assistance from the state was reduced to a minimum and the regular school system took children for 6 hours per day from age 5. In those cases where children were in school at age 3 or 4, it was often for only 3 hours per day. In these conditions, mothers of young children are generally constrained to work part-time, for hourly wages that are often below the minimum recommended by the European Union (Fouquin, Jean and Sztulman, 2000). In this system, only women in managerial [7] positions are able to pay for childcare and thus retain full-time employment (Joshi and Davies, 1992; Rake, 2000).
9In sharp contrast, France is considered as the prototype of a country practising a pro-natalist policy (Ekert-Jaffé, 1986), which stems from a Bismarckian vision. Sociologists categorize it as part of the group of conservative countries, together with others of continental Europe (Esping-Andersen, 1996). Assistance to families is very diversified (Ekert-Jaffé, 2001), including 15 allowances that accompany all stages of family life: moving allowance, beginning-of-school-year allowance, young-child benefit, housing allowance, etc. This assistance is concentrated, however, on families with three or more children and/or on those with young dependent children [8]. Beginning at the end of the 1980s these allowances were extended to poor individuals and families (social housing allowance in 1992 and RMI/minimum income allowance in 1988). With respect to women’s employment (Ekert, 1983), the system overtly favoured homemakers until 1974 [9]; such discrimination has not existed since 1983 [10]. Maternity leave lasts only 16 weeks, but the compensation, equal to 84% of salary, is much higher than in the United Kingdom, making the French system almost twice as generous as the English one. In addition, this leave is extended to 26 weeks beginning with the third child. At the end of the 1980s and in the 1990s, emphasis was placed on subsidizing different types of childcare (Fagnani, 1998): care by the mother, who may interrupt her working life in return for just under half the minimum wage [11], care at home or at the home of a child-minder [12]; and all these expenditures result in reductions of income taxes. At the same time, public institutions doubled the number of places in crèches between 1984 and 1994. Crèches now accommodate 16% of the childcare age population, while nursery schools take all three-year-olds and 40% of two-year-olds. For very different motives—the aim is to encourage childbearing among working women rather than to entice mothers to remain in the labour force—France thus shares with the Nordic countries a family policy designed to reconcile motherhood with the pursuit of a career.
10In addition, illegitimate and legitimate children now have nearly equal rights, and that is associated with a high proportion of births outside of marriage in both countries. In Britain, however, parents are more likely to separate and form single-parent families. Whether or not they have previously lived in a couple, British single mothers are mostly on the margins of the labour market. In the early 1990s two thirds were receiving Income
11Support [13]. In France, by contrast, women living alone with their children are less common and most are employed; 17%, or one in six, are beneath the poverty level.
2 – Family policy and fertility by social class according to economic theory
12When the level of education or the social class of the partners rises, their (potential) labour income increases. Economic theory suggests three mechanisms to explain the effect of income on fertility (Becker, 1981; Cigno and Ermisch, 1988).
Time cost or budget constraint?
13As women’s wages increase, the price of time increases correspondingly. Thus, while a woman is looking after her child, she cannot use that time to earn money in the labour market, and the opportunity cost is a function of her wage ( the time cost of children being borne principally by women). By a substitution effect, then, the greater the cost of the child, the larger the decline of fertility. This effect is thus stronger for women in higher socio-occupational groups. Moreover, career women may choose to limit their fertility in order to devote themselves more fully to their working life. In helping women to combine family life and working life, French family policy tends to minimize these two effects. The state (partially) compensates women for the loss of time and minimizes the reduction of fertility induced by women’s employment for all women, and especially for those in managerial positions (Ermisch, 1989).
14But a high wage also has the effect of relaxing the budget constraints that limit the number of children in a family: this is the income effect. For women, this effect is generally offset by the substitution effect; for men, the income effect dominates and fertility rises with the level reached on the social ladder. By reducing the monetary cost of children by a greater proportion for the lower and middle classes, French family policy diminishes fertility differences linked to men’s income.
The “quantity” versus “quality” effect
15As social class rises, couples seek to extend the schooling of their children in order to obtain for them the best possible social position, resulting in limitation of the number of children (Becker and Lewis, 1973). This explains the lower fertility of the middle class compared with the working class. At the same time, the upper class has sufficient resources not to limit the number of children. In the 1980s, empirical studies thus showed that 80% of families of “managers” in France sent their children to school up to age 20 and beyond, and that only a minority of working-class children stayed on at school until that age, while among the intermediate classes the duration of schooling depended crucially on family size (Ekert, 1982). These three effects combined produce the classic “U-shaped” (Calot and Deville, 1971) or “J-shaped” (Desplanques, 1985, 1994) curve of fertility by men’s socio-occupational group.
The timing of births
16Finally, with respect to the timing of births, the strategy of English women who must interrupt their employment involves spacing births closely together so as to reduce the time out of the labour market and the financial losses (Ní Bhrolcháin, 1987). If they are managers, they will postpone maternity until their position is consolidated and they have secured the salary increases that come at the beginning of a career (Abowd, Kramarz et al., 2000; Murphy and Welch, 1990). In contrast, women who combine full-time employment and motherhood can maintain a longer interval between births and thereby avoid having to look after several young children on top of their full-time employment. The timing of births should thus be more concentrated in England than in France (Cigno and Ermisch, 1988).
II – The socio-economic environment
1 – A common nomenclature
17The socio-economic environment in Britain is very different from that in France. In the 1970s the typical school-leaving age was much younger in the United Kingdom, especially for women. Young people entered the labour market quite early and usually became managers via promotions, obtained primarily by men. A detailed examination of the classification of occupations in social classes (Elias, 1997; INSEE, 1995; Desrosières and Thévenot, 2000) and of the composition of these categories by educational attainment makes it clear that there are substantial differences between the two countries. This is especially the case for managers and intermediate occupations. In Britain, managers may include “retail floor managers” or even “head nurses”. In these circumstances, managerial could not be clearly differentiated from intermediate occupations according to the definition used in France. We have thus developed a common nomenclature for both countries by distinguishing four broad social categories on the basis of the detailed three-digit classification of occupations and also taking into account levels of education (see Rendall et al., 2001, for details) [14].
18Men are categorized as “self-employed”, “managers”, “white-collar workers”, and “blue-collar workers” in accordance with the French definitions. Managers are further differentiated by whether or not they possess a higher-education diploma. For women, we first distinguished those occupations that, due to the high prevalence of women in the occupation or due to their special working conditions, are often more compatible with family responsibilities, following in this regard Rose and O’Reilly (1997). Thus, among the best-educated groups, female teachers have work schedules that allow them to look after their own (school-age) children, and women in the health professions sometimes have access to hospital crèches. At a lower occupational level, the working conditions of secretaries and other office employees are often compatible with childbearing (part-time and flexible work schedules, for example). Among other women in the labour force, upper-level occupations (managers, intermediate occupations and self-employed) were distinguished from the category of blue-collar and sales workers. Finally, women who were economically inactive as of the census could not be categorized in terms of their previous occupation, because of the poor quality or absence of this information in the French census; these women thus constitute a separate category.
19The presence of this heterogeneous category introduces some bias into our estimates since the fertility of economically inactive women should have been counted as part of the fertility of their original socio-occupational group. To estimate the incidence of this bias in the French-English comparison we have tried to determine the identity of economically inactive women (in terms of educational attainment, social class of origin, partner’s social class) and whether they are different in the two countries. The measures of social mobility between the censuses of the 1980s and those of the 1990s indicate the original socio-occupational categories of the women who were inactive in 1990-91. In France, they represented 21.5% of the sample, and those who had worked previously were primarily blue-collar or white-collar workers. In England, the inactive constituted almost 35% of the sample in 1991, and half of them were already in that category in 1981. The distribution of the others conformed roughly to the structure of the labour force in 1981, except that there were proportionately far fewer managers and teachers and more blue-collar workers than on average [15]. In other respects, the characteristics of the not economically active are similar in both countries. They resemble blue-collar and sales workers, but have more education on average. Relative to those with the same occupation who were in the labour force, a slightly higher proportion of economically inactive women in England had higher education [16]. Economically inactive women in England were also more likely not to have a partner. In both countries 40% of those not economically active are in a union with a blue-collar worker and 32% with a manager. Overall, the presence of this category, which seems closely linked to fertility and to position in the life cycle, should not greatly affect the comparisons.
20Finally, the absence in the English census of a question concerning secondary-education diplomas led us to distinguish only two higher-education groups (one or two years beyond secondary school, and three years or more) from all other levels of educational attainment. We combine higher-education diploma and socio-occupational group to differentiate managers with at least 3-4 years of higher education (bac +3 or more [17]), those with 1-2 years (bac +1 or +2) and those without any higher education; the same distinction is used for teachers and members of the health occupations.
2 – Evolution of the socio-occupational structure in France and in England: comparison of two cohorts of women
21Table 2 compares the distribution of women by socio-occupational category for three cohorts of women in the censuses of 1990 for France and 1991 for England [18]. The French women of cohort 1 (born 1952-56 and aged 34-38 in 1990) are more likely than their English counterparts to be in upper-level categories [19]. The gap changes as one moves across cohorts. When interpreting these changes, however, it is necessary to take account of three simultaneous effects that are pertinent here:
22— The timing of births influences the proportion of women who are not economically active, and the intermediate group (cohort 2) is in its period of peak fertility.
23— Promotions to managerial positions depend on work experience and are thus less common for the younger women (Chenu, 1998).
24— The younger cohorts are often better educated and more committed to a career, and this compensates in part for their lack of experience. In France, one must also take account of the high unemployment of young workers, which postpones the time when members of the last cohort, aged 24-28 in 1990, will join the ranks of managers (Ekert-Jaffé and Solaz, 2001). In England, there is a much more direct path from school to work for the most recent cohort (fewer interruptions and more managers among the better-educated women).
Socio-occupational groups for the three cohorts of women as of the 1982 and 1990 censuses in France, and as of the 1981 and 1991 censuses in England and Wales (%)
Socio-occupational groups for the three cohorts of women as of the 1982 and 1990 censuses in France, and as of the 1981 and 1991 censuses in England and Wales (%)
25The first effect is stronger in England, where the proportion of economically inactive women is more than one and a half times that of France (38% vs. 23%) in cohort 2 [20]. This may be linked to the age of children in this cohort, who are younger on average in Britain as of the time of the census, and it is especially apparent for the lower-level occupations (other white-collar workers and blue-collar workers) whose share is 3.6 percentage points lower than in cohort 1 (19.8% vs. 23.4%). The effect of promotions is more apparent in France, where the proportion of managers and teachers decreases across cohorts, with the youngest women substantially more likely to be other white-collar or blue-collar workers [21]. Moreover, delayed childbearing (Beaumel, Kerjosse and Toulemon, 1999) and increased activity among mothers contribute to the rise in the share of women in the lowest occupational group at the expense of those who are inactive in the youngest cohort. In England, a combination of the other effects—fewer inactive women, a greater cohort effect—moderates this increase for women in the lowest occupational category. Promotions, however, have the same impact in both countries. Among managers in the first cohort in 1990-91, 30% were office workers and 10% were blue-collar workers at the beginning of the 1980s [22]. Finally, the cohort effect operates principally in England, where the increase in the proportion of managers in the younger cohorts is substantial; in France, by contrast, some of the graduates in the youngest cohort had never held paid employment and are hence categorized as not economically active.
3 – Marital status, and partners’ socio-occupational category
26Another factor that influences fertility is women’s marital status. The great majority of women in cohort 1 were married and living with their first partner in 1990-91: this was the case for 70% of women in France and 62.6% in England (Table 3). This percentage decreases for the younger cohorts partly because of their age, but also and in particular because of the trend towards later marriage and even non-marriage [23]. The youngest women are mostly single, and they are more likely to be cohabiting in France than in England (Lelièvre, 1995) [24]. Women who are widowed, divorced or separated constitute a small minority that decreases from older to younger cohorts, and they are proportionately more numerous in England. In France, they represent 11.5% of women in the oldest cohort in 1990, compared with 15.4% in England in 1991.
Marital status for the three cohorts of women as of the 1982 and 1990 censuses in France and as of the 1981 and 1991 censuses in England and Wales (%)
Marital status for the three cohorts of women as of the 1982 and 1990 censuses in France and as of the 1981 and 1991 censuses in England and Wales (%)
27Despite slightly different categorization schemes, the socio-occupational categories of partners, married or not, are rather close in the two countries [25]. For the women in cohort 1, a third of partners are blue-collar workers, a third are managers, about 15% are self-employed, 7-8% are white-collar workers and 7-8% are categorized as “absent” or “other”. About a third of the women in the second cohort also are or have been in union with blue-collar workers, but the proportion of partners who are managers is reduced in favour of partners who are white-collar workers, or non-existent. Part of this reduction occurs because the partners, who are younger, have not yet benefited from promotions.
III – Results of the model: what is the influence of family policy and unemployment?
28The fertility history of French women born from 1952 to 1956 (cohort 1) is known through 1995, when they were aged 39-43. In England, women born between 1954 and 1958 are observed through 1998, when they were aged 40-44. We first analyse family formation, and then we examine the changes that took place between the 1980s and the 1990s, by comparing the first cohort up to 1986-88 and the cohort born 10 years later at the same ages until they were 30-34 years old at the end of the observation period.
29To distinguish the effects of completed fertility from pure timing effects, logistic models were used to analyse, on the one hand, the parity progression ratios for births 1, 2 and 3, and on the other, various parity-specific birth probabilities [26].
30By construction, the birth rates are underestimated in these cohorts, as some births could not be linked to their mothers. This non-linkage may also entail errors in the birth order: in a case where the first birth was not linked to the mother, the second would be considered as the first child. One consequence is that, relative to first births, higher-order births are under-estimated even more (see Appendix). We compared our results (Table 1, cohorts 1952-56 and 1954-58) with those of Pearce, Cantisani and Laihonen (1999) and Smallwood (2002) for the 1955 cohort up to age 45. In France, according to these two sources, 92% of women had at least one child, 72% had at least two children and 33% at least three (by comparison, our results in Table 1 show figures of 87.9%, 66.6% and 28.1%, respectively). In Britain, the corresponding figures are 85%, 72% and 30% from the other sources and our own results are 82.7%, 65.2% and 25.1%, respectively. A small part of this underestimation occurs because our cohorts are not all observed up to age 45. This study, then, should be seen as a comparison of the influence of socio-economic variables, and not as an estimate of rates. The following results show the influence of each variable “all other things being equal”. This is thus a simulation where we examine the effect of one variable at a time (marital status, socio-occupational category, etc.). The parity progression ratios were estimated from the corresponding odds ratios in the regressions. They depend on the distribution of women by socio-occupational category and by marital status (see Tables 2 and 3). In particular, it is useful to keep in mind the high proportion of economically inactive women in England, representing about 41% (as opposed to 27% in France) of women at risk of a third birth.
1 – Family formation and marital status of women born in the 1950s
31The effect of marital status on fertility is highly significant. Two principal conclusions emerge. First, despite the importance of non-marital births in both countries (apparent from registration data), being unmarried remains an obstacle to fertility (Table 4, A). The negative coefficients for this variable in the estimated models reflect low parity progression ratios for cohabiting unmarried women compared with married women, and much lower in England (48% versus 83.3% for parity zero) than in France (80% as against 94.5%). The relative handicap is greatest for women who had not cohabited before 1990. Second, for women whose union was terminated (widowed, divorced or separated in 1990-91), the number of children depends above all on the de facto marital situation: women in a new union (married or cohabiting) are not less likely to have additional children than their counterparts who did not experience marital disruption. Moreover, mothers of two children who are in union for (at least) the second time have a greater-than-average likelihood of having a third child by age 40-44. There are two ways to interpret this result: either mothers of large families who have experienced marital disruption seek (successfully) to form a new couple, or, more likely, new couples seek to have a child together, thereby raising their fertility. This association is stronger in France than in England, where the fact of not being married or cohabiting in 1990-91 exerts a more negative influence on the probability of having a third child.
Parity progression ratios to first, second and third birth in birth cohort 1 (women born in the mid-1950s), observed until 1995 in France and 1998 in England and Wales, by marital status and socio-occupational category of women in 1990/91 and their partners, as estimated by logistic regression model
Parity progression ratios to first, second and third birth in birth cohort 1 (women born in the mid-1950s), observed until 1995 in France and 1998 in England and Wales, by marital status and socio-occupational category of women in 1990/91 and their partners, as estimated by logistic regression model
2 – Influence of socio-economic position
Socio-occupational categories and first births
32French and English women begin their reproductive careers at different times, as shown by the annual sequence of first birth probabilities over the life cycle by educational attainment, “all other things being equal” (Figure 1 and Table 5). Among women with no higher education, the profile is flatter in England: there are somewhat more births prior to age 18, a plateau around 8% between ages 25 and 29, and a non-negligible probability until 35 and over. In France, the probability shows a much clearer peak, in the region of 17% about age 26, and from this age up to 35 the likelihood of having a first child remains higher than in England. Higher-education graduates delay having children and that, surprisingly, to the same extent in both countries. Figure 1 shows a similar trend for both countries when the curves are displaced. The effect of this delay on completed fertility, however, is very different: the proportion of women who have a first child remains much higher in France, where this delay is merely a timing effect [27], while in England it translates into a higher level of childlessness. For example, all other things being equal, among managers with three years or more of higher education, one woman in four remains childless in England (Table 4, C and Figure 2). Thus, a certain incompatibility appears to exist between upper-level work and motherhood in this country where the chances of being a mother diminish as the level of education increases. In France, the policy of providing childcare assistance appears to allow better-educated women to become mothers sooner than in England. On that side of the Channel, we also note that teachers and nurses, who often have more flexible work schedules or who have access to hospital crèches, become mothers more often than office workers (88% versus 83%) and nearly as often as blue-collar workers (92%). Whether greater compatibility or a stronger desire for children prompts them to choose these types of employment, well-educated women with the highest completed fertility are concentrated in these sectors. Finally, in both countries, neither the partner’s level of education nor his socio-occupational category have a significant influence on first births [28].
Annual first birth probabilities, 1952-1956 cohorts in France and 1954-1958 cohorts in England and Wales
Annual first birth probabilities, 1952-1956 cohorts in France and 1954-1958 cohorts in England and Wales
Results of the logistic regression model of the annual probabilities of having a first, second or third child by timing of births, marital status and socio-occupational category of women in 1990/91 and their partners (if any), for women born in the 1950s in France and in England and Wales
Results of the logistic regression model of the annual probabilities of having a first, second or third child by timing of births, marital status and socio-occupational category of women in 1990/91 and their partners (if any), for women born in the 1950s in France and in England and Wales
Socio-occupational categories and second births
33With respect to second births (Figure 2 and Table 4), women’s educational attainment tends to have less influence. Nevertheless, the effects remain significant, even though the women at risk of a second birth are already selected out by their decision to have a first birth. In France, for example, well-educated managers or white-collar workers are less likely to have a second child than either teachers and nurses or blue-collar workers (71% and 73% of the former versus 87% and 76% of the latter, respectively). This effect is even more apparent in England where only two-thirds (65%) of managers with one or two years of higher education have a second child. At the same time, however, among English women the progression to parity 2 among blue-collar workers (and teachers and nurses without a higher-education qualification) is especially high (81%). Is this because interrupted or part-time employment occasions a smaller loss for these women than for better-educated women? Whatever the reason, if English mothers are more likely to give birth to a second child, it is due to the behaviour of blue-collar workers and the greater proportion of economically inactive women.
Parity progression ratios to first and second births by women’s socio-occupational group, 1952-1956 cohorts in France (observed until 1995) and 1954-1958 cohorts in England and Wales (observed until 1998)
Parity progression ratios to first and second births by women’s socio-occupational group, 1952-1956 cohorts in France (observed until 1995) and 1954-1958 cohorts in England and Wales (observed until 1998)
34In these conditions, the influence on fertility of the man’s income, measured by the partner’s socio-occupational category, is also important [29] (Table 4, B). In both countries, being in a union with a manager who has three years or more of higher education increases the likelihood of having a second child (81.4% as opposed to 75.6% when the partner is a white-collar worker, in England). In this regard, English self-employed workers and French farmers are similar to upper-level managers. The gap with the lower-level occupations is greater in France.
35The birth interval is shorter in England (Figure 3 and Table 5), reflecting women’s strategy of short spacing between births in order to reduce the period during which work activity is interrupted [30]. In France, though, the timing of fertility varies more according to educational attainment; women who waited until the end of their studies to begin childbearing have their children at a more rapid pace.
Yearly second birth probabilities, 1952-1956 cohorts in France and 1954-1958 cohorts in England and Wales
Yearly second birth probabilities, 1952-1956 cohorts in France and 1954-1958 cohorts in England and Wales
Socio-occupational categories and third births
36The influence of family policy in France appears clearly when considering third births (Ekert-Jaffé and Maugüé, 1992). Progression to a third birth is much higher than in England for nearly all the socio-occupational groups (Figure 4). This is the case for office workers (28% vs. 22%), blue-collar workers and economically inactive women [31] (63% vs. 41%). For these groups, the parental education allowance represents a significant proportion of their income. Further, for teachers and nurses, as well as for the best-educated managers (44% have a third child in France vs. 23% in England), the effect of income is positive: high salaries and a policy supporting childcare at home reduce the marginal cost of the third child. The influence of family policy may also be observed on the timing of births (Figure 5 and Table 5). In France, beyond secondary education, the more education women have, the shorter the intervals between their successive births. In England, by contrast, women’s socio-occupational category has very little association with whether or not there is a third birth and very little effect on its timing. For all economically active women (except teachers), the proportion of women having three children is close to that for the lowest-fertility French group, white-collar workers [32].
Parity progression ratios to third births by women’s socio-occupational group, 1952-1956 cohorts in France (observed until 1995) and 1954-1958 cohorts in England and Wales (observed until 1998)
Parity progression ratios to third births by women’s socio-occupational group, 1952-1956 cohorts in France (observed until 1995) and 1954-1958 cohorts in England and Wales (observed until 1998)
Yearly third birth probabilities, 1952-1956 cohorts in France and 1954-1958 cohorts in England and Wales
Yearly third birth probabilities, 1952-1956 cohorts in France and 1954-1958 cohorts in England and Wales
37Despite these different environments, there are no notable differences according to the partner’s socio-occupational category (Table 4). In both countries, the quantity-quality trade-off results in a U-shaped curve according to the partner’s social group. But the high-fertility groups are not the same. In France, women whose partner is a manager with no higher-education diploma are significantly less likely to have a third child than the reference category, office workers. In England, there is instead a greater propensity to have a third child for women whose partner is either self-employed or a blue-collar worker. From an economic perspective, there is a U-shaped curve characteristic of the income effect. Contrary to the theoretical predictions concerning the influence of family allowances and benefits, this income effect does not seem stronger in England.
3 – Changes over ten years: the influence of unemployment and prolonged studies?
38For women born ten years later (cohort 3), the most remarkable fact is the postponement of marriage and childbearing, in England as well as in France (Tables 1 and 3). In England, for example, only 36% of women born in the 1960s were married by ages 23-27 [33], compared with more than 60% at the same ages for the cohorts born ten years earlier. This figure of 36% remains lower than the total of 60% even if the 18% of women in consensual unions are included. Moreover, among women born in the 1960s, 67% were already mothers at the end of the observation period when they were 30-34 years old, as against 72% of women at the same ages for the cohort born ten years earlier. For France, the changes are even greater: the proportion of mothers is 72% as opposed to 82% for the older cohort. This could be a timing effect, with the delay made up in the coming years (Toulemon and Mazuy, 2001).
39Another notable change between these two cohorts is the reduction in the share of women who are not economically active and the growth in the proportion of women who are either managers or blue-collar workers (see Table 2). The percentage of women not economically active fell from 42.0% to 32.6% in England and from 28% to 20.0% in France. The share of blue-collar workers grew more in France (from 19.5% to 28%) while that of managers increased more in England (from 8.0% to 15.5%). As we have already seen, unemployment of young workers in France has slowed access to managerial status in the youngest cohort, but the age difference (24-28 for cohort 3 and 26-30 for cohort 1) should also be taken into account.
40To study the relationship between partner’s occupational group and the parity progression ratios (Table 6), we take as reference group women who were married [34] as of the census, a group whose childbearing is a little earlier than other women and who have above-average fertility. They represent 54% of English women and 62% of French women in cohort 3. Among these women, 82% in England and 89% in France are mothers by the end of the observation period (Tables not shown). Somewhat surprisingly, these figures are declining for France but increasing for England when compared with those for the cohort born ten years earlier [35].
41We now examine the influence of marital status on the parity progression ratios. In the first place, for the most recent cohorts, not having been married by age 24-28 in France does not have the same negative consequences as in England. This observation is linked to the effect of marital status, which diminished more in France than in England (Table 6). Moreover, cohabitation has become widespread in France and these unions are increasingly fertile. It appears that while the probability of becoming a mother remains lower for unmarried than for married women, it is practically equal [36] to that of married English women [37]. Overall, for first births, the decline in the proportion of women who become mothers is primarily the result of a structural effect: there is more cohabitation, and there are more women without partners.
Parity progression ratios to first and second birth in birth cohorts 1 and 3 (born 10 years apart), observed until aged 30-34 in France and in England and Wales, by marital status and socio-occupational category of women in 1981/82 and their partners, if any, as estimated by logistic regression model
Parity progression ratios to first and second birth in birth cohorts 1 and 3 (born 10 years apart), observed until aged 30-34 in France and in England and Wales, by marital status and socio-occupational category of women in 1981/82 and their partners, if any, as estimated by logistic regression model
The influence of women’s educational attainment and social position
42In France as in England, the effect of a secondary or higher education diploma remains important, all other things being equal. In ten years, from one cohort to the next, the gap widens with lesser-educated women and those not economically active, for whom the delay in childbearing does not appear. In France, it even seems that the latter women, selected by marriage, have children sooner than their elders. The differential effect of socio-occupational group on the timing of fertility becomes closer in the two countries for the cohorts born in the 1960s, with a greater polarization. But the changes are small. This timing difference is reflected, for English women, in greater differences in the number of children born by ages 34-38 (Table not shown). In France, the proportion of women with a first child changed little among blue-collar workers and economically inactive women relative to the older cohort. Women in the other groups are somewhat more likely to be childless, especially among teachers (this is the case for 8% of teachers with a higher-education diploma compared with 2.5% in cohort 1). Finally, in general, the decline observed is especially linked to the absence of a partner and to the woman’s labour force status [38]. In England, the differences between socio-occupational groups widen: teachers lose some of their advantage. Blue-collar workers and economically inactive women are even more likely to give birth to a first child, in contrast to women managers who are less likely to become mothers (reduction of six percentage points for managers and self-employed workers with three or more years of higher education). Is this due to the inability to reconcile career and motherhood? The greater access to managerial positions for women born in the 1960s and the decline in the proportion of teachers (Table 2) may well be associated with the decline of fertility in England.
43The same changes are observed for second and third births. The differences have a tendency to be attenuated in France, where only teachers with one or two years of higher education stand out clearly, while other teachers and blue-collar workers move closer to the average. Again, the differences widen in England, where young managers are even less fertile than their elders, in contrast to blue-collar workers and economically inactive women, who for the most part are former or future blue-collar workers.
The influence of the partner’s socio-occupational group
44In both countries, the role of the partner’s socio-occupational group increased over the ten-year period. This result holds for both first and second births. In England we find a classic U-shaped curve for first births, with self-employed workers and blue-collar workers having the highest fertility, whereas in France only blue-collar workers are more likely than others to be fathers. In France the U-shaped curve appears beginning with the second birth—but again, this is only a timing effect. Thus, the increase in the third parity progression ratio for the younger cohort concerns the most educated managers and the blue-collar workers (as well as the self-employed). Among them, the most educated retain above-average fertility. The income effect thus becomes more important for this cohort, which reached age 30 during the 1990s.
45The increased influence of marriage in England and of the partner’s social status in both countries is somewhat surprising. Income effect? Are we returning to more traditional ideas about the roles of men and women with regard to childbearing? Women born in the 1960s, though, are more likely to pursue professional careers at the expense of their fertility. In this case, the effect of poor economic conditions can explain this behaviour. The youngest cohort has been hit directly at peak childbearing ages; economic factors and the security provided by the partner’s employment thus take on increased importance (Blossfeld and Drobnic, 2001; Ekert-Jaffé and Solaz, 2001; Meron, 2002).
Conclusion
46The study of the family formation process by women’s socio-occupational category for cohorts born during the 1950s in France and in England and Wales illustrates several points regarding the influence of family policies on fertility. It also shows that social polarization of fertility is larger in England and that the disparities between women who leave the labour market and those who do not are larger still.
47In England, the common view is that market forces should determine the compatilility of motherhood and women’s employment. This laissez-faire approach appears to result in the selection of mothers, from first birth on, by educational attainment. In particular, among well-educated managers, one woman in four is childless, whereas well-educated women with high fertility are concentrated in teaching and the health sector, where employment is more compatible with family life. Despite this selection of mothers, blue-collar workers, teachers and nurses have a higher probability than university-educated managers of having a second child, but somewhat less so than for first births. This social polarization of fertility disappears for third births. Thus, only economically inactive women have a greater propensity to give birth to a third child, but to a lesser degree than in France.
48In France, nearly all women eventually become mothers, whereas in England, motherhood is much more strongly linked to occupation and in particular to withdrawal from the labour market. In France, selection begins at the second or even third birth. While the propensity to have a second child is weaker than in England, it depends very little on the woman’s social group. Family policy thus seems to erase fertility differentials by social class. All mothers have almost the same chance of having a second child as English white-collar workers. The fertility difference between the two countries is mostly due to the blue-collar workers (their fertility is much higher in England), who are penalized less in their hourly wage and opportunities for advancement by a career interruption or part-time employment.
49The following comparison illustrates this greater social polarization in England. Overall, in the two countries (Table 4), for first births, the women with the lowest fertility are upper-level managers with high levels of education, while those with the highest fertility are the economically inactive women. The difference between these two extremes is only 4 percentage points in France (92% vs. 96% of women have at least one child), while it is 19 percentage points in England (76% vs. 95%). The probability of having a second child is lowest for managers with one or two years of higher education, and highest, here again, for economically inactive women. Between these two categories, the parity progression ratios vary from 71% to 89% in France and from 65% to 86% in England, the differences thus being 18 and 21 percentage points, respectively. With respect to the third child, social polarization appears to be stronger in France, where behaviour vis-à-vis interruption of employment becomes more similar to that found in England (Davies and Joshi, 1994). The likelihood of having a third child is lowest for white-collar workers and highest for the economically inactive women. The difference comes from the higher fertility of economically inactive women, managers and blue-collar workers, compared with white-collar workers whose behaviour is similar to the average in England. Moreover, third births are more sensitive to family policy in France, where interruptions of employment are compensated. Overall, social polarizations are crystallized in France by the third child, whereas they are evident in England beginning with the first birth.
50In both countries, the influence of men’s income is seen to start with the second birth, and this income effect increases for the third birth. Consequently, we observe a U-shaped curve, arguably reflecting the trade-off between the number and the education of children.
51The changes observed over a ten-year period, for the cohorts that reached age 30 by the middle of the 1990s, may well reflect an influence of the economic situation on the family formation process. The very high proportion of women not in unions, which is known in France to be a consequence of increased unemployment and prolonged studies, is a major factor contributing to birth postponement and even infertility among these cohorts. The social position of the partners becomes more important, for understandable reasons, in an uncertain environment. The other notable element in the changes seen over the ten-year period is the tendency for women not to jeopardize their professional lives. All other things being equal, the differences across socio-occupational groups decrease in France while they increase in England. Better-educated women there attempt to reconcile having a small number of children (if they have children at all) with their careers, opting under constraint for child quality rather than numbers.
52Overall, this article has enabled us to measure the influence of the respective socio-economic positions of men and women. The econometric analysis (Tables 4 to 6 and Figures 1 to 5) establishes that in all cases, the most pertinent variable, with the strongest links to fertility, is for women their educational attainment, and for men their socio-occupational category. We also show that fertility is above all linked to women’s characteristics.
Acknowledgements
The authors thank INSEE and the Office for National Statistics (ONS) for giving them access to the Echantillon démographique permanent (Permanent Demographic Sample) and the Longitudinal Study (LS), respectively. This study benefited from the assistance of the Longitudinal Study Support Programme at the Institute of Education. The opinions expressed in this article are those of the authors and do not necessarily reflect the views of INSEE or ONS. The authors are also grateful to the Nuffield Foundation for having supported the participation of Michael Rendall and Kevin Lynch during their presence at the Institute of Education.The Permanent Demographic Sample
53The French Permanent Demographic Sample (EDP) is a longitudinal panel that contains essentially demographic information on a representative sample of the French population. For the individuals in the sample, it links up census data from 1968, 1975, 1982 and 1990 with vital registration data for major demographic events (births, marriages, deaths, acknowledgements of paternity) recorded between 1968 and 1995. This link has been achieved for a sample of a little over 1 in 200. The quality of the links between census data and birth registration data is acceptable. However, 10-12% of the births are missing in the EDP, a figure very similar to the omission rate in the Longitudinal Study (Babb and Hattersley, 1992). The omission rate is not constant with respect to age: it is greater at young ages, particularly for adolescents. After correction for missing data (using information provided on other birth records available for the current marriage), the total omission rate for the sample is reduced to 5-6% (estimates and corrections by the authors). Thus, despite these corrections, the EDP tends to underestimate fertility, which is important for our study. Nevertheless, since it is women in the lower-level socio-occupational groups who tend to have their children at young ages, the biases that might result from this omission in no way change our commentaries, since the differences would be even greater than those we found.
54The EDP’s method of data collection implies certain gaps in the individual files. It must be kept in mind that the certificates (of births and marriages) may be missing for reasons related to the management of these files or to the omission of the corresponding certificate at the time of data extraction. In addition to events concerning individuals who migrated temporarily or permanently from metropolitan France, we can also cite births of children who were not legally recognized by the father at the time of the birth (since the birth certificate did not mention the individual in the EDP, it could not be included).
55In some cases information obtained from other records of births during the marriage could be used to determine the missing birth dates. Other examples of data recovery: the father’s birth date (for a non-acknowledged child) on marriage certificates legitimizing the birth; date of marriage from legitimate child birth certificates.
56Once these corrections were made, we decided to retain in the sample only those individuals for whom all the registration and census data were complete and correct. To check the information on births, we used variables concerning birth order for the children of individuals in the EDP. To give a representative example, an individual for whom we observe two birth certificates and for whom the birth orders identified on the certificates are one and three is excluded from the sample. Finally, for the reasons cited above, we use a sub-sample of women born in France between 1945 and 1975 and present for all the censuses.
Notes
-
[*]
Olivia Ekert-Jaffé: Institut National d’Études Démographiques, Paris.
Heather Joshi: Centre for Longitudinal Studies, Institute of Education, London.
Kevin Lynch: Office for National Statistics, London.
Rémi Mougin: Institut National d’Études Démographiques and Institut d’Études Politiques, Paris.
Michael Rendall: Office for National Statistics and Centre for Longitudinal Studies, Institute of Education, London.
Translated by David Shapiro. -
[1]
Our study is focused on England and Wales since data for Scotland and Northern Ireland were not available. For brevity, we sometimes use “Britain” or “England” below, but always analyse data for England and Wales.
-
[2]
More precisely, the fertility history of French women born in 1952-56 (cohort 1) is known through 1995, i.e. up to ages 39-43. In England, the cohorts born between 1954 and 1958 are observed through 1998, corresponding to ages 40-44 (older by one year). The age of these cohorts is 34-38 in France as of the 1990 census and 33-37 in England as of the 1991 census.
-
[3]
In this second comparison, the socio-occupational grouping of the cohorts born during the 1950s (cohort 1) will be that from the 1982 census in France, when the women were 26-30 years old, and from the 1981 census in England, when the women were 23-27. To make the observation periods comparable, this cohort is observed until 1986 for France and 1988 for England.
-
[4]
Lord Beveridge is known as the architect of the post-war welfare state in Britain. Its blueprint was his 1942 report on “Social Insurance and Allied Services”.
-
[5]
An unpaid parental leave of 40 weeks, with a maximum of 29 weeks after the birth, was available up to 1994 to women who had worked full-time for two years with the same employer before the birth (Heitlinger, 1993). Since 1995, women who worked part-time have also been eligible for this leave.
-
[6]
In 1996, this assistance amounted to 47 ecus per child per month, plus an additional 11 ecus for the first child. Ecu was the name originally given to the European Union currency unit, before it was officially renamed the euro in 1999. (1 euro = 1 ecu).
-
[7]
Translator’s note: For convenience, the French “cadres” category will be translated as “managers”. In fact, this category includes not only managers and executives, but also senior civil servants and members of the liberal and higher intellectual professions.
-
[8]
In 1996, for the 80% of all families whose income did not exceed the ceiling for provision of a family income supplement (complément familial) the allowance amounted to 265 ecus monthly (family allowance plus family supplement) for the third child. For higher-income families the amount was reduced to 131 ecus (family allowance only). There was a family allowance of 91 ecus per month for the second child, but nothing for a first child. Over and above these family allowances, families with a child under the age of 3 were also entitled to a young-child benefit, amounting to 147 ecus per month in 1996.
-
[9]
At that time there was an allowance for one-salary families, and no child-care allowance.
-
[10]
The conditions for attribution of the family income supplement were eased for two-earner couples: the ceiling was raised by an amount equal to 25% of the ceiling for families with two children.
-
[11]
Complementarily, the parental education allowance—a paid parental leave—was created in 1985 for the third child, and extended to the second child in 1994.
-
[12]
Created in 1986, the allowance for childcare at home covers the social contributions for a home child-minder, up to a maximum amount (300 ecus per month in 1996). Created in 1990, the family assistance for the employment of a person to look after the child covers all social contributions so long as an approved child-minder is employed; in addition, the amount was increased by 75 ecus per month in 1991, raised to 120 ecus in 1994, to contribute to the payment of wages.
-
[13]
Income Support is a type of assistance destined for non-working families whose income is less than a specified minimum.
-
[14]
In particular, in this classification scheme two-digit occupations in France are matched up with three-digit occupations in England. In Britain, there were 371 categories in the SOC809 in 1981 and 549 in 1991.
-
[15]
Among women in France aged 34-38 who were not economically active in 1990, 62% were either not economically active or without identified employment in 1982, 16% were blue-collar or sales workers, 12% were office workers, 3% were teachers or nurses, and 5% were in managerial positions. In England, 52% were economically inactive as of the preceding census, and the other percentages were 16, 21, 5 and 6, respectively.
-
[16]
In England as in France, 10% of those not economically active had higher education, compared with 6% of white-collar workers in England and 3% of white-collar workers in France.
-
[17]
To facilitate comparisons, the abbreviations “bac +1 or 2” and “bac +3 or more” will be used to indicate one or two years, and three years or more of higher education, respectively. In France, the baccalauréat (bac) is the upper secondary-school diploma that leads to higher education.
-
[18]
These cohorts were born in the years 1952-56, 1957-61 and 1962-66 respectively for France, and in 1954-58, 1959-63 and 1964-68 for England and Wales.
-
[19]
While French women are more likely than English women to be in upper-level categories, among managers there are more working in France and more not economically active in England. Among all women not economically active in 1990/91, proportionately as many women in both countries had been recorded as managers in the previous census.
-
[20]
The percentage of women economically inactive increases by 2.7 points in England as against 2.0 points in France between the first and the second cohort. The difference is greater for women with a first birth. At the beginning of the 1990s, nearly 90% of French mothers remained employed after the birth of their first and even their second child, compared with only about half of English mothers.
-
[21]
Average age at leaving school rose throughout the period, which means that the reduction in the proportion of women in upper-level occupations for the youngest group cannot be imputed to a cohort effect.
-
[22]
These changes took place over a 10-year period in Britain and an 8-year period in France.
-
[23]
In England, among women aged 23-27, 58% in the oldest cohort were married as of the 1981 census, while among women of the youngest cohort in the same age range only 36% were married as of the 1991 census.
-
[24]
In France, women living in consensual unions are often managers or otherwise economically active, while in England they are often blue-collar workers or economically inactive.
-
[25]
Partner’s socio-occupational category (SOC) is determined as follows. If the women indicate that they have a partner in the census being considered, we use the reported SOC. In the 1982 French census, however, the SOC is only known for a quarter of the sample. For the analysis of second or higher-order births, when the women do not have a partner or if the partner is reported to be economically inactive or unemployed as of the census, we use the socio-economic category reported by the father as of the first birth.
-
[26]
The denominator is constructed to exclude individuals who are no longer at risk of having a child, thus eliminating the selection bias. For France, we took into account the structure of residuals, decomposed into an error term for the individual and an error term for the year of observation. This procedure did not generate results different from those of the simple model. Finally, therefore, we estimated the model without taking into account the correlations between observations for the same individual.
-
[27]
Nearly all French women aged 44 had at least one child. The delay is thus made up for, meaning that in France the probability of having at least one child increases more with age.
-
[28]
The timing of fertility does, however, show a distinct U-shaped curve: wives of blue-collar workers and of managers or self-employed men tend to have their first child a little sooner (Table 5).
-
[29]
In economic terms, these results suggest that the male income effect is stronger than the female substitution effect.
-
[30]
Teachers and nurses, for whom interruptions are less common, and blue-collar workers, who lose less from working part-time or interrupting work, tend to have longer spacing between births.
-
[31]
Those who are economically inactive are mainly white-collar and blue-collar workers who have interrupted their work activity and are eligible for the parental education allowance.
-
[32]
While the proportions of women with three children are not very different in France and in England (Table 1), this is partly due to the share of blue-collar or not economically active women among those at risk of a third birth: 66% in England versus 47% in France.
-
[33]
Marital status and educational attainment of French women born between 1952 and 1956 were observed in the 1982 census when they were 26-30 years old. English women born between 1954 and 1958 were 23-27 years old as of the 1981 census. Women born ten years later (cohort 3) were born between 1962 and 1966 for France and between 1964 and 1968 for England. They were aged 24-28 and 23-27 respectively as of the 1990/91 censuses.
-
[34]
The reference category consists of married women when we study the impact of marital status, and of women in union when the focus is on partner’s occupational group.
-
[35]
For cohort 1, the figures were 93% for France and 71% for England.
-
[36]
Even if the relative effects are the same in both countries (the coefficients are similar in each), the reference category is different.
-
[37]
The comparison here is between the sum of the constant term and the coefficient for cohabiting unmarried women in France, with the constant term from the regression for English women.
-
[38]
Blue-collar workers born between 1962 and 1966 were more likely than their elders to be in the labour force, but had not really changed their fertility, all other things being equal. This exerts a positive effect on the progression to parity two compared with economically inactive women, for whom the ratio decreases (Table 6).
