Population (english edition)
I.N.E.D
I.S.B.N.sans
200 pages
p. 559 à 578
doi: en cours
Veille sur la revue
Veille sur l'auteur
I.N.E.D
I.S.B.N.sans
200 pages
p. 559 à 578
doi: en cours
Veille sur la revue
Veille sur l'auteur
Vous consultez
The Demographic Situation in France
Volume 58 2003/4-5
Age at First Union in France: A Two-Stage Process of Change
France Prioux [*] France Prioux, Institut National d’Études Démographiques, 133 bd Davout, 75980 Paris Cedex 20, Tel: 33 0(1) 56 06 21 44, Fax: 33 0(1) 56 06 21 99,
The “Study of Family History” Survey concurrent with the 1999 French census has for the first time made it possible to measure the proportions and ages of men and women forming first unions, and to observe the annual changes in first union formation.
The age at first union fell between approximately the 1930 and 1955 cohorts, in particular among men. The timing of first union has been substantially delayed. Over roughly fifteen cohorts, the median age has increased by more than two years for men and women, and this trend toward later first unions seems not to be complete.
Period indicators of union formation are calculated for the years 1960-1998 and used to date the reversal of trend to the mid-1970s and to link the annual variation of these indicators to levels of youth unemployment. The lengthening of education has had a fairly consistent delaying effect on union formation, but unemployment levels appear to have initiated the movement and modified its pace.
The postponement of first unions is accompanied, particularly among men, by an increase in the proportion of individuals reaching age 50 without ever having lived in a stable union. Several possible explanations are advanced for this: the rising frequency of non-cohabiting partnerships, the greater instability of unions, and the increasing difficulty, for some men, of finding a partner for life. Cette étude s’appuie sur l’enquête Étude de l’histoire familiale associée au recensement de 1999. L’âge à la première union a baissé entre les générations 1930 et 1955 environ, en particulier chez les hommes, mais ensuite le calendrier des premières unions a été beaucoup retardé : en une quinzaine de générations, l’âge médian a augmenté de plus de deux ans pour les hommes comme pour les femmes, et cette tendance au retard des premières unions ne semble pas achevée.
Le calcul d’indicateurs conjoncturels de formation des unions sur la période 1960-1998 permet de dater le retournement de tendance au milieu des années 1970, et de mettre en relation l’évolution annuelle de ces indicateurs avec l’évolution du chômage des jeunes. Par ailleurs, la prolongation des études pousse assez régulièrement au retard des unions ; mais l’évolution du chômage semble en avoir déclenché le mouvement et en conditionner les accélérations et les pauses.
Ce retard des premières unions s’accompagne, en particulier pour les hommes, d’une augmentation de la proportion de ceux qui atteignent la cinquantaine sans avoir jamais vécu en couple stable. Plusieurs pistes sont évoquées pour expliquer cette hausse : le développement de la vie en couple non cohabitant, l’instabilité accrue des unions et la difficulté croissante, pour certains hommes, de trouver une compagne pour partager leur vie. Este estudio se basa en la encuesta Estudio de la historia familiar asociada al censo de 1999. La edad a la primera unión disminuyó entre las generaciones de 1930 y 1955, especialmente entre los hombres, pero a partir de ésta el calendario de primeras uniones se retrasó fuertemente: en quince generaciones, la edad mediana ha aumentado en más de dos años tanto entre los hombres como entre las mujeres, y esta tendencia al retraso de las primeras uniones no parece haber finalizado.
El cálculo de los indicadores coyunturales de formación de uniones entre 1960 y 1998 permite fechar el cambio de tendencia a mediados de los años setenta y relacionar la evolución anual de estos indicadores con la evolución del paro juvenil. La prolongación de los estudios estimula el retraso de las uniones, pero la evolución del paro parece ser el propulsor de los cambios y también parece condicionar las aceleraciones y pausas del proceso.
La postergación de las primeras uniones va acompañada por un aumento de la proporción de individuos que llegan a los cincuenta años sin haber vivido una relación de pareja estable, especialmente en el caso de los hombres. El artículo sugiere varias hipótesis para explicar este aumento: el desarrollo de la vida en pareja no cohabitante, la inestabilidad creciente de las uniones y la dificultad creciente, entre ciertos hombres, de encontrar una compañera con quien compartir su vida.
The age at first union fell between approximately the 1930 and 1955 cohorts, in particular among men. The timing of first union has been substantially delayed. Over roughly fifteen cohorts, the median age has increased by more than two years for men and women, and this trend toward later first unions seems not to be complete.
Period indicators of union formation are calculated for the years 1960-1998 and used to date the reversal of trend to the mid-1970s and to link the annual variation of these indicators to levels of youth unemployment. The lengthening of education has had a fairly consistent delaying effect on union formation, but unemployment levels appear to have initiated the movement and modified its pace.
The postponement of first unions is accompanied, particularly among men, by an increase in the proportion of individuals reaching age 50 without ever having lived in a stable union. Several possible explanations are advanced for this: the rising frequency of non-cohabiting partnerships, the greater instability of unions, and the increasing difficulty, for some men, of finding a partner for life. Cette étude s’appuie sur l’enquête Étude de l’histoire familiale associée au recensement de 1999. L’âge à la première union a baissé entre les générations 1930 et 1955 environ, en particulier chez les hommes, mais ensuite le calendrier des premières unions a été beaucoup retardé : en une quinzaine de générations, l’âge médian a augmenté de plus de deux ans pour les hommes comme pour les femmes, et cette tendance au retard des premières unions ne semble pas achevée.
Le calcul d’indicateurs conjoncturels de formation des unions sur la période 1960-1998 permet de dater le retournement de tendance au milieu des années 1970, et de mettre en relation l’évolution annuelle de ces indicateurs avec l’évolution du chômage des jeunes. Par ailleurs, la prolongation des études pousse assez régulièrement au retard des unions ; mais l’évolution du chômage semble en avoir déclenché le mouvement et en conditionner les accélérations et les pauses.
Ce retard des premières unions s’accompagne, en particulier pour les hommes, d’une augmentation de la proportion de ceux qui atteignent la cinquantaine sans avoir jamais vécu en couple stable. Plusieurs pistes sont évoquées pour expliquer cette hausse : le développement de la vie en couple non cohabitant, l’instabilité accrue des unions et la difficulté croissante, pour certains hommes, de trouver une compagne pour partager leur vie. Este estudio se basa en la encuesta Estudio de la historia familiar asociada al censo de 1999. La edad a la primera unión disminuyó entre las generaciones de 1930 y 1955, especialmente entre los hombres, pero a partir de ésta el calendario de primeras uniones se retrasó fuertemente: en quince generaciones, la edad mediana ha aumentado en más de dos años tanto entre los hombres como entre las mujeres, y esta tendencia al retraso de las primeras uniones no parece haber finalizado.
El cálculo de los indicadores coyunturales de formación de uniones entre 1960 y 1998 permite fechar el cambio de tendencia a mediados de los años setenta y relacionar la evolución anual de estos indicadores con la evolución del paro juvenil. La prolongación de los estudios estimula el retraso de las uniones, pero la evolución del paro parece ser el propulsor de los cambios y también parece condicionar las aceleraciones y pausas del proceso.
La postergación de las primeras uniones va acompañada por un aumento de la proporción de individuos que llegan a los cincuenta años sin haber vivido una relación de pareja estable, especialmente en el caso de los hombres. El artículo sugiere varias hipótesis para explicar este aumento: el desarrollo de la vida en pareja no cohabitante, la inestabilidad creciente de las uniones y la dificultad creciente, entre ciertos hombres, de encontrar una compañera con quien compartir su vida.
Until the 1970s, civil registration data were largely adequate for studying union formation in France. Since then, the spread of unregistered consensual unions has made the use of surveys indispensable. In this article, France PRIOUX analyses the results of the 1999 “Study of Family History” Survey (EHF), covering 235,000 women and 145,000 men aged 18 and over. For the past twenty-five years, the age at first union formation—with or without marriage—has been steadily rising, but this delay is far shorter than that affecting first marriage, whose timing has receded by at least four years over the same period. The proportion of persons who have never lived in a stable union has also risen, but remains small at 7% and 11% respectively for women and men born around 1970. These figures too are well below the 15-20% drop in the proportion of single persons, in the legal meaning of the term. Lastly, the lack of an upper age limit for EHF survey respondents enables annual union formation indices to be computed for the 1960-1998 period, which permits a highly instructive comparison with the trends for youth unemployment.
The formation of a first stable union may be regarded as one of the key events in the transition to adulthood, as this is usually the moment when young people become truly independent from their parents (Bozon and Villeneuve-Gokalp, 1995; Galland, 1995 and 2000; Toulemon, 1994). Moreover, a stable partnership is generally viewed as a prerequisite for deciding to start a family. Any delay in the age at which men and women cross this threshold is therefore likely to impact on the next step: the birth of the first child.
Over the past fifteen years or so, a number of surveys by INED (1985 and 1994 “Family Situations” surveys) and INSEE, the French National Statistical Institute (annual Labour Force Surveys) have indicated a fall in the proportion of men and women living in union (see, for example, Leridon and Villeneuve-Gokalp, 1994; Toulemon, 1997); data from the 1999 Census confirmed the decline in unions among women under 60 and among men up to age 75 (Prioux, 2002). In the older age groups, this pattern is clearly due to the increase in union break-ups: for although re-partnering is common, it does not always occur immediately, and the proportion of persons enumerated in the census as living alone is rising. For younger individuals, however, it is very hard to disentangle the effects of later first union and of marital instability. Just thirty years ago, marriage statistics gave a reasonably accurate account of entry into first union, but recourse to surveys has since become indispensable for dating this major event in the lives of men and women, now that unmarried cohabitation has spread to become the most common way of starting life as a couple.
Around 1970, fewer than one in five couples began their life together without being married, compared with nine in ten today (Toulemon, 1997). Not only has it become almost exceptional to wait to be married to start living as a couple, but an increasing number of couples never legalize their union. Whereas marriage was once the near exclusive context for childbearing, living in union, with or without marriage, is today the usual preliminary condition. As a result, although analysis of nuptiality has lost none of its sociological interest, demographers are increasingly focusing on couple formation. At what age do men and women form their first cohabiting partnerships? Are the male and female timetables converging? Are couples being formed earlier or later than in the past? What proportion of men and women never live in a union? These are some of the questions addressed in this article, drawing on the data gathered in the INSEE-INED “Study of Family History” survey (Enquête Étude de l’Histoire Familiale, or, more simply, Enquête Famille: hereafter EHF) conducted jointly with the 1999 Census.
A retrospective survey on a very large sample is the only way to measure accurately the rise in the age at first union, and most importantly to date the beginning of this shift with a view to identifying its causes. The 1999 EHF survey allows this, thanks to two important innovations. For the first time, the survey included men as well as women, and the questions were no longer confined to marriages but covered the periods respondents had spent “living as a couple”. Thanks to their special focus on the first experience of living in union, the questions enable us to measure exactly in what proportions and at what ages men and women enter into first unions, irrespective of their current union status.
We confine our study to observation of the temporal changes in the frequency and timing of this event, starting with the cohorts born in the 1930s which reached adulthood in the late 1940s. First unions of men and women are analysed separately, not jointly, and it should be borne in mind that for women in particular, some first unions are formed with a partner for whom this is a second union. We also observe the annual trend in first union formation, and suggest some explanations for the changes identified. The classification of unions by type (direct marriage or cohabitation), their duration, and the legalization (if any) of consensual unions will be analysed in a later study.
Defining partnership
We should begin by defining what we mean by “first union” or “first partnership” — terms used interchangeably in this article. Our definition derives from the 1999 EHF survey questionnaire, in which question 14 specifies living as a couple as “living together under the same roof, for six months or more, with or without marriage”.
By comparison with the INED “Family Situations” surveys of 1985 and 1994, the six months’ co-residence requirement is more restrictive. Not only had these two surveys not introduced a duration condition, but by not defining from the outset what was to be understood by “living as a couple”, they revealed that some respondents regarded themselves as living as partners even though they maintained two separate homes (Leridon and Villeneuve-Gokalp, 1994; de Guibert-Lantoine et al., 1994). Is this why, in the 1999 EHF survey, about 6% of men and 5% of women reported that they either were or had been living as a couple (question 13) but then provided no dates about their unions when given a more restrictive definition of the relationship (question 14)? In reality, this is not the main reason. As was established by the matching of 1,306 questionnaires from the 1999 EHF survey with those of the INED “Biographies and Contact Circle” survey in the Paris region in 2000-2001, when information on unions was collected by field workers, the responses to question 13 are almost 100% correct (Mazuy and Lelièvre, forthcoming). This apparent inconsistency therefore results mainly from a quite deliberate omission, related to the EHF survey questionnaire being self-completed. We have therefore relied primarily on question 13, and refer the reader to Appendix I for more details of the adjustments performed to allow for the under-reporting of dates relating to partnership histories.
Men and women born in the mid-1950s have been the youngest to form first unions
A quarter of men born in 1930 had formed a first union before age 23, half before age 25.3 years, and three-quarters before age 29. For men born twenty-five years later, each of these proportions was reached at least 1.5 years earlier (Figure 1a). Thereafter age at first union rose. The median age, which had fallen to 23.8 years in the 1951-1955 cohorts, rose by two years over fifteen cohorts and exceeded 26 years with the 1971 cohort. The mean age followed a very similar course to the median age. First, it fell by nearly 1.5 years, fluctuated around a minimum of 24.6-24.7 years in the 1944-1955 cohorts, then returned, in the 1969 cohort, to a level close to that of the 1930 cohort (26 years). The distribution of ages at first union around the median age was initially fairly stable and concentrated. The interquartile range
[1] varied between 5.2 for the 1932 cohort and 5.5 years for the 1950 cohort; then the age dispersion increased quickly, and the range reached 8 years in the 1968 cohort. The age at which three-quarters of men have already lived in union rose very quickly after the 1955 cohort, and exceeded 30 years by the 1966 cohort. The increase in age at first union was thus accompanied by a marked diversification of the ages at which this first union was experienced in the male cohorts.
Figure 1
Age at first union for men and women since the 1930 cohort
Source: INSEE-INED, 1999 EHF survey.
Entry into first union occurs almost three years earlier for women than for men (Figure 1b). A quarter of women born in 1930 have already lived with a partner at slightly over age 20 (20.3 years), half by age 22.5, and three-quarters by age 25.8. The changes are initially more gradual than for men, and the ages at which these various thresholds are reached fall by about one year: the median age is at a low of 21.5 years in the 1955-1959 cohorts. Thereafter it rises steeply, gaining two years over fifteen cohorts, to stand at 23.7 in the 1974 cohort. The mean age at first union falls less than the median age. It declines from 23 years in the 1930 cohort to 22.5 years in the 1952-1958 cohorts; it then rises by one year over nine cohorts to reach 23.5 years in the 1967 cohort, and subsequently increases more slowly. The concentration of ages at first union changes less than among men: the interquartile range varies between 4.7 and 5 years in the 1936-1953 cohorts, then increases gradually to 6.6 years in the 1971 cohort. The age at which three-quarters of women have lived in a couple has increased rapidly and currently stands at over 27 years.
The male and female timetables of union formation thus at first converged somewhat, until the cohorts born in the second half of the 1940s. The difference between the median ages of men and women is only 2 years in these cohorts, as against 2.7-2.8 years up to the 1939 cohort
[2]. The difference increases to 2.9 years for the 1961 cohort, then dips again to 2.4 years in the 1966-1967 cohorts, before climbing back to 3 years in the 1971 cohort. Despite the many areas of convergence in lifestyles between the sexes
[3], first union formation for women still follows a timetable that is earlier and more concentrated than that for men (Figure 2). In the 1955 cohort — that entering into first union at the earliest age — nearly one woman in seven formed her first union in the year of her twentieth birthday; the modal age is slightly higher among men (21 years) and less concentrated (one man in eight), with nearly as many first unions occurring in the years 22 and 23. Subsequently, some relaxing of this normative pattern occurs: ten cohorts later, about one woman in nine enters into a union in the year of her twentieth birthday, and almost as many do so in the following year, whereas for men the first union most often begins at age 23, and for slightly fewer than one man in ten. The distributions have widened and the distance between the curves is intact. In sum, ages at first union remain fairly sharply differentiated between young men and young women (Galland, 1995).
Figure 2
Age-specific rates of rates union formation in the 1955 and 1965 cohorts (per 100 men or 100 women of each age)
Source: INSEE-INED, 1999 EHF survey.
The precocity of first unions and first marriages used to be very closely linked
The fall in the age at first union among women and men across the 1930-1945 cohorts is not surprising. It confirms all previous analyses of first marriages for the cohorts in which these two events almost invariably coincided (see, for example, Festy, 1971; Munoz-Pérez, 1979)
[4]. We also know that the cohorts born in the mid-1940s are those which married youngest, and that the age at first marriage has since risen considerably. Here we see that this pattern is not repeated, at least initially, for the timing of first unions. Although the mean age declines only very slightly after the 1945 cohort, the first quartile and the median age continue to fall until the 1955 cohort, and even slightly longer among women; only later does the age at first union begin to rise.
The 1985 and 1994 INED “Family Situations” surveys had clearly shown the two stages of the change (see, for example, Leridon and Villeneuve-Gokalp, 1994; Toulemon, 1997). To begin with, the first consensual union replaced the first marriage — even though the latter was sometimes merely pushed back a few months or years; in the second stage, first union formation itself began to be delayed
[5]. The large sample of the 1999 EHF survey allows us to observe in precisely which cohorts the trends were reversed: 1955 seems to be a pivotal cohort for men, 1957 for women.
Period analysis of first union formation
The annual trend in couple formation can be observed by making period indicators comparable to those used for first marriages (Figure 3 and Table C in Appendix II). Let us begin by examining the indicator constructed from the sum of rates, which aggregates all the first unions formed each year after scaling them back to a constant number of men (or women) in each cohort (Figure 3a). Because of the way it is constructed, the indicator can exceed 100%, and it is then proof of a sharp fall in the age at first union. In practice, whenever the indicator nears 95%, assumed to be the maximum proportion of unions in a cohort, it is an unmistakable sign of a lowering of age at union. This is the case for men from 1960 to 1976; this is also observed for women until 1979 but with a brief interruption in 1967-1968 and at levels below the male indicators between 1962 and 1971. As noted earlier, the fall in the age at first union was sharper for men than for women (Figure 1). The first drop in the female indicator, between 1964 and 1967, was only transient. It was related to specific conditions in the supply of potential partners, when the arrival at the ages of union formation of the large post-war cohorts created a temporary imbalance between the sexes, with women in these cohorts having greater difficulty finding partners among the smaller male cohorts born during World War II (Munoz-Pérez, 1979). The fall in the female age at first union then resumed and, for both sexes, this period of high indices of union formation continued, whereas the first marriage indicators began to fall sharply after 1972
[6]: this is the period when consensual union replaced or delayed first marriage, but the downward trend in the age at union formation continued. With the legalization of contraception and the gradual mastery of the techniques by young people, it became possible to form a union without intending to start a family immediately.
Figure 3
Period indicators of first union formation, and of first marriages for women, 1960-1998 (per 100 men or 100 women)
Source: INSEE-INED, 1999 EHF survey.
But after 1975, and especially after 1979, the indicators of union formation for men and women began to fall, and have stayed consistently below 90%, signalling an unquestionable delaying of the formation of first unions. The fall stopped in 1984-1985, however, and for a few years the indices recovered strongly: the female indicator moved back over 90% in 1989 and 1990, and the male indicator levelled off at 87%. The recovery was only short-lived, and the indicators dropped back to their earlier levels, stabilizing this time at about 80% — slightly higher for women, slightly lower for men.
Before trying to explain this change of trend, let us examine the behaviour of another period indicator, the overall probability of first union, computed in a synthetic cohort of people who have never been in union (Figure 3b)
[7]. The trends are the same as with the previous indicator but distinctly attenuated: temporary difficulties for women around 1967, a fall in the probabilities of first union from the late 1970s, especially for men, a brief upswing in the late 1980s, and a relative stabilization in recent years.
As this indicator is less sensitive to changes in timing, its level is, in principle, closer to the intensity of first couple formation in the cohorts. A downward trend is visible throughout the period, particularly among men, for whom living in union seems to have become less common than for women.
Youth unemployment and delayed entry into union
A striking coincidence exists between the dates of the inflections in the period indicators of union formation and the curve of youth unemployment (Figure 4). The first fall in the indicators of union formation closely followed the first oil crisis, which made it harder for young people to enter a first job. The proportion of jobless young people was very low until the mid-1970s, then rose — at first gradually, then quickly — to a peak in 1985. The proportion of unemployed in the total 20-24 age group climbed from under 5% in 1975 to 17% in 1985 (Meron and Minni, 1995). Youth unemployment then declined very rapidly under the combined effect of "new employment policy measures (TUC, SIVP, Plan d’Urgence pour les Jeunes, Contrat d’Adaptation and Contrat de Qualification) and the subsequent economic recovery which created new jobs" (Meron and Minni, 1995, p. 19). By 1991, unemployment at 20-24 had eased to 9.7% for men and 12.2% for women. It was precisely in this period of falling unemployment that the indicators of union formation turned up; they fell again when unemployment resumed its upward course
[8]; later still, when the proportion of young unemployed stopped growing, the indicators stabilized then rose slightly. There is no doubt that the difficulties experienced by young people in entering the labour market exert a powerful influence on union formation
[9].
Figure 4
Proportion of unemployed young people aged 20-24 (per 100 men or 100 women of the same age), and period indicator of first union formation (sum of rates per 100, right-hand scale)
Source: Meron and Minni, 1995; INSEE-INED, 1999 EHF survey.
Longer education and delay in union formation
When we observe the large delay in first union formation, one of the possible causes that inevitably comes to mind is the longer time being spent in education. This is known to have increased considerably over the generations, with a steadily rising proportion of young people going on to higher education.
The 1999 EHF survey enables us to measure this “educational revolution,” thanks to a question on the age at which the respondent ceased regular attendance of a school or higher education institution. The age at which half of a cohort has already left the education system has risen almost continuously in the cohorts observed, with very distinct phases of acceleration, just after the 1934-1935 cohorts, and — roughly speaking — between the 1965 cohort (1967 for men) and 1973 cohort (Appendix II, Table B). This age has risen from a mere 15 years until the 1934 cohort to over 21 years since the 1974 cohort for men, and since the 1972 cohort for women, who now spend slightly longer in the education system than their male counterparts.
Thus, in total, there is no correlation between the changes in these indicators across the cohorts. The substantial increase in the length of studies appears to have had no incidence, initially, on the median age at entry into first union, since it did not prevent the fall in that age down to the cohorts born in the mid-1950s (Figure 1); furthermore, the reversal in the trend after the 1955 cohort is not linked to any acceleration of the rise in the median age at completion of education, a change that occurred at least ten cohorts later. However, the age at which three-quarters of a cohort have left the education system has risen sharply for men since the 1956 cohort, after a pause in the 1948-1955 cohorts: from 17.5 years in the 1930 cohort, it stabilized at 20 years in the 1948-1955 cohorts, then rose quickly to 23.8 years in the 1975 cohort. On the whole, the change was faster for women, and the pause less marked: the age at which three-quarters of women have completed their education has risen from 17.1 years for the 1930 cohort to 24 years for the 1975 cohort.
At first, therefore, the lengthening of education seems to have had no influence on the median age at first union, because in the older cohorts several years usually elapsed between finishing education and forming a union. The divergent trends in the ages at which these two events occurred has brought them closer, and this has certainly made the latter more dependent on the former, at both the collective and individual levels.
The conjunction of rising unemployment and longer studies has thus undoubtedly played an important role in delaying union formation. While the increase in the length of studies was a fairly constant factor tending to delay union formation, the changes in unemployment seem to have initiated the movement and modified its pace. In fact, the two phenomena are not totally independent, since many young people remain in education to obtain a qualification and avoid unemployment. The increase in the length of studies — sometimes analysed as a “retention” of the young in the education system — was particularly large between 1987 and 1993 (Poulet, 1996). Since the second half of the 1990s, however, the duration of education has stopped increasing (DARES, 2001). At the time of the survey, we are still too close to this change in trend to observe its possible effect on the timing of union formation.
It should be noted, however, that the influence of unemployment and length of education on union formation differs between the sexes. For women, the first union often comes soon after completing education, whereas men usually have to wait until they have found a job before forming a union (Bozon, 1991; Galland, 1995; Toulemon, 1994).
A slightly higher proportion of men have never lived in union
The proportions of men and women who have never lived in union by the time they reach age 50, which were close to 8% for men and 7% for women born in the early 1930s, are gradually diverging (Figure 5). Among women, the proportion fell in the first cohorts, settled below 6% in the 1944-1964 cohorts, then rose slightly; it is estimated at 7% in the 1971 cohort
[10]. Among men, the proportion initially fluctuated around 8%, then in the 7-8% range between the 1940 and 1957 cohorts. Subsequently, it rose quickly, stabilized at 10% in the 1960-1965 cohorts, then resumed its upward course, and is estimated to reach 11% in the 1969 cohort.
Figure 5
Proportions of men and women never having lived in union before age 50 by cohort (%)
Source: INSEE-INED, 1999 EHF survey.
A sizeable difference (4 points) of frequency has therefore developed between male and female experiences of forming first unions. In principle, this difference cannot be explained by a sharp imbalance between the available male and female populations, such that a large excess of men would make union formation harder for men and easier for women
[11]. On the contrary, the sex ratios at the ages of union formation are increasingly favourable to men, because immigration, which in the past was essentially male, now has a larger female component.
Whether it reflects a conscious decision never to form a partnership or results from some sort of exclusion from the “marriage market” of a slightly higher proportion of men, this difference between male and female frequencies of entry into first union can come about only because a higher proportion of men form several successive unions and sometimes re-partner with women who have never previously lived in a union (Cassan et al., 2001). And the pattern is identical for legal unions. The same survey (like the civil registration statistics) shows that, in the cohorts born in the 1930s, the proportion single (in the legal sense of the term) was twice as high among men (14%) as among women (7%). The gap narrowed to 5 percentage points in the 1965 cohort, in which the proportion of men never married at age 50 is estimated at 30%, that of women at 25% (Prioux, 2003).
While not reaching such high values, the proportions of persons never having lived in a cohabiting union for six months or more are rising, particularly among men. To explain this increase in the “single state” two additional possibilities can be mentioned: the difficulty of forming a stable union, and the increasing proportion of couples who do not cohabit continuously. Partial support for these hypotheses comes from the 1994 INED “Family Situations” survey, which clearly showed the growing fragility of unions. And although the proportion of couples not sharing the same dwelling at the time of the survey had not risen significantly since the previous survey, the period of semi-cohabitation seemed to be getting longer for couples who reported “not having immediately lived together on a continuous basis” (de Guibert-Lantoine et al., 1994; Toulemon, 1997).
Overview
The decision to marry and the attendant formalities (publication of banns, organization of the church service — if one was required — and the wedding reception, etc.) all took time, whereas the formation of an unmarried union is in theory a faster process. In addition, the spread of contraception means that young people can experience partnership at an early age without the risk of having a child. As a result, the age at first union fell somewhat more than the age at first marriage. This decline initially continued, while marriage began to be postponed as unmarried cohabitation became the most frequent way to start living together as a couple. But the trend was reversed in the mid-1970s, and men and women born since the late 1950s have entered into union slightly later. The delay is at present very large, and the age at first union is now higher in the younger cohorts than in the cohorts born in the 1930s. The two main factors responsible for triggering this new trend seem to be the difficulties of entering the labour market and the longer time spent in education. The beginning of the trend coincided with the rise in youth unemployment from the mid-1970s, while in contrast, successive economic upturns appear to favour union formation. These factors appeared important from the perspective of the period analysis, but they do not preclude other factors that are harder to measure or that are related to broader social change. In this respect it is relevant to mention the generalization of female participation in the labour force (acquiring employment experience is doubtless becoming a prerequisite of union formation for women too), and the transformation of relations between the generations: in the 1960s and 1970s, living in a couple and marrying were the only means to achieve independence, especially for women. Today, it is no longer necessary to leave one’s parents in order to have stable sexual relations. Also worth mentioning is the growing rigidity of the rented housing market, which makes it hard to obtain an independent dwelling in the absence of adequate financial guarantees. The increase in the age at first union is in fact just one of the elements of the delayed entry into adult life among the young generations (Galland, 1995 and 2000).
The postponement of first union formation is accompanied, particularly for men, by an increase in the proportion of individuals reaching age 50 without ever having lived in a stable partnership. Several possible explanations have been offered for this increase: the growing number of non-cohabiting couples, the greater fragility of unions, and the increasing difficulty, for some men, of finding a partner for life. In a forthcoming article we shall see that for men educational qualifications are now more than ever a prerequisite for successful entry into the marriage market.
APPENDIX I
The 1999 “Study of Family History” survey and the adjustments for non-responses on union dates
In the 1999 “Study of Family History” Survey, 145,000 men and 235,000 women aged 18 and over (born before 1981) completed a four-page questionnaire containing questions on their children (their own and those raised by them), their union history, and languages spoken in childhood.
The three questions on unions that we have used in our study are worded as follows:
- (even if your spouse/partner lives in another dwelling for work-related reasons)
- Yes … 1 → go to question 14
- No, but you have lived as a couple previously … 2 → go to question 14
- No, you have NEVER lived as a couple … 3 → in that case, go to next page
- Main dates of living as a couple
- By “living as a couple,” we mean living together under the same roof, for six months orlonger, with or without marriage.
- If you have had only one period of living as a couple, use the first line.
- If you have had several such periods, refer only to the first and latest.
(In a two-row table, respondents were asked to list dates (months and years) of the successive stages of their first (or only) union, and of their latest: start of union and, as appropriate, date of marriage, end of union, divorce, and death of spouse/partner.)
- How many times have you been married? … times
So question 13 does not define “living as a couple” while suggesting that there are some couples who do not always share the same dwelling. Next, question 14 is assumed not to have an answer unless the co-residence has lasted at least six months.
Among the persons born between 1930 and 1980 who are the subject of this study, approximately 6% of the men and 5% of the women did not complete the table in question 14 after replying to question 13 that they were currently living in union (2.2% of men and 2.1% of women) or had done so in the past (3.9% and 2.8% respectively). In addition, a fairly small proportion of respondents (0.3% of men and 0.4% of women) reported at least one marriage under question 15, but reported no union or union history dates.
These responses seem to result primarily from forgetfulness or from a conscious desire not to provide information about a past union. As a corroboration of this, the INED “Biographies and Contact Circle” Survey—conducted in 2000-2001 with a network of home interviewers—allowed a matching of the questionnaires of 1,306 persons resident in the Paris region born between 1930 and 1950 who had also responded to the 1999 EHF survey (Mazuy and Lelièvre, forthcoming). The comparison of their responses in the two surveys demonstrates that the answers to question 13 are reliable, and that the persons who did not provide dates for unions exhibited no specific profile. We therefore imputed the missing ages at first union in each cohort using the distribution of known ages at first union in the same cohorts. This concerns 4-5% of women in the 1930-1960 cohorts, and sometimes a slightly higher percentage of men in the same cohorts, without ever exceeding 6% (see Table A).
For cohorts born after 1960, adjustments were made on a decreasing scale, after it was observed that the proportion of men and women reporting that they had already lived in unions but not completing the table in question 14 rose appreciably in the younger cohorts. For example, 9.5% of men and 6.7% of women born in 1976-1980 reported having lived in union but then did not complete the table, compared with 3.3% and 2.4% respectively in the 1961-1965 cohorts. Conversely, among persons currently in unions, the number omitting the date of union formation falls steadily (see Table A). Our hypothesis is that a certain proportion of respondents reporting a past union in question 13 then left question 14 unanswered once they realized that their broken union did not satisfy the duration or co-residence criteria. This proportion rises in the younger cohorts, as they are more likely to have remembered a union that was certainly more recent. Accordingly, we have assumed that 100% of persons born in 1960 who reported a past union had actually lived in union. We then reduced this percentage by 5 points with each cohort down to 0% in the 1980 cohort; this is tantamount to assuming that none of the young persons reaching age 18 in 1999 had effectively lived in a cohabiting union for at least six months followed by a separation. On the other hand, our adjustments do take into account all persons — regardless of cohort — who reported that they were currently living in union or that they had married.
Table A
Proportions of men and women reporting at least one union or marriage, and not giving date of entry into union (per 100 men or 100 women in each group of cohorts)
APPENDIX II
Table B
Age at completion of education, age at first union, and frequency of first unions since the 1930 cohort
Table C
Period indicators of first union formation, 1960-1998 (per 100 men or 100 women)
BIBLIOGRAPHIE
·
Bozon M., 1991, “Women and the age gap between spouses. An accepted domination ?”, Population: An English Selection, 3, pp. 113-148.
·
Bozon M., 2003, “At what age do women and men have their first sexual intercourse? World comparisons and recent trends”, Population et Sociétés, No. 391.
·
Bozon M., Villeneuve-Gokalp C., 1995, “L’art et la manière de quitter ses parents”, Population et Sociétés, No. 297.
·
Cassan F., Mazuy M., Clanché F., 2001, “Refaire sa vie de couple est plus fréquent pour les hommes”, Insee première, No. 797.
·
DARES, 2001, “L’évolution récente des scolarités et de l’insertion professionnelle des jeunes (1998-2000)”, Premières informations et premières synthèses, No. 26.1.
·
Festy P., 1971, “Évolution de la nuptialité en Europe occidentale depuis la guerre”, Population, 26(2), pp. 331-380.
·
Galland O., 1995, “Une entrée de plus en plus tardive dans la vie adulte”, Économie et Statistique, No. 283-284, pp. 33-52.
·
Galland O., 2000, “Entrer dans la vie adulte : des étapes toujours plus tardives mais plus resserrées”, Économie et Statistique, No. 337-338, pp. 13-36.
·
Guibert-Lantoine C. de, Leridon H., Toulemon L., Villeneuve-Gokalp C., 1994, “La cohabitation adulte”, Population et Sociétés, No. 293.
·
Henry L., 1966, “Perturbations de la nuptialité résultant de la guerre de 1914-1918”, Population, 21(2), pp. 273-332.
·
Leridon H., Villeneuve-Gokalp C., 1994, Constance et inconstances de la famille. Biographies familiales des couples et des enfants, Ined, Travaux et documents, Cahier No. 134, 342 p.
·
Leridon H., Villeneuve-Gokalp C., 1994, “Les nouveaux couples : nombre, caractéristiques et attitudes”, in Constance et inconstances de la famille. Biographies familiales des couples et des enfants, Ined, Travaux et documents, Cahier No. 134, pp. 17-53.
·
Mazuy M., Lelièvre É., forthcoming, “Déclarer ses enfants, déclarer ses conjoints : rationalité des locuteurs et mode de questionnement”, in an INED Cahier on the results of the Family History Survey (EHF).
·
Meron M., Minni C., 1995, “Des études à l’emploi : plus tard et plus difficilement qu’il y a vingt ans”, Économie et Statistique, No. 283-284, pp. 9-31.
·
Munoz-Pérez F., 1979, “L’évolution récente des premiers mariages dans quelques pays européens”, Population, 34(3), pp. 649-694.
·
Poulet P., 1996, “Allongement de la scolarisation et insertion des jeunes : une liaison délicate”, Économie et Statistique, No. 300, pp. 71-81.
·
Prioux F., 2002, “Recent demographic developments in France”, Population-E, 57(4-5), pp. 689-728.
·
Prioux F., 2003, “Recent demographic developments in France”, Population-E, 58(4-5), pp. 525-558.
·
Toulemon L., 1994, “Les étapes vers l’âge adulte : vers un nouveau statut des femmes”, in Leridon H., Villeneuve-Gokalp C., Constance et inconstances de la famille. Biographies familiales des couples et des enfants, Ined, Travaux et documents, Cahier No. 134, pp. 165-181.
·
Toulemon L., 1996, “Cohabitation is here to stay,” Population: An English Selection, 9, pp. 11-46.
NOTES
[*]
Institut National d’Études Démographiques, Paris.Translated by Jonathan Mandelbaum.
Institut National d’Études Démographiques, Paris.Translated by Jonathan Mandelbaum.
[1]The interquartile range is the difference between the ages at which three-quarters and one-quarter of men in a cohort have already lived in union.
The interquartile range is the difference between the ages at which three-quarters and one-quarter of men in a cohort have already lived in union.
[2]The same observation applies to the difference between the mean ages at first union formation for men and women. These differences between mean ages and median ages should not be confused with the age differences between partners (mean difference and median difference), which must be computed within couples, not by observing men and women separately.
The same observation applies to the difference between the mean ages at first union formation for men and women. These differences between mean ages and median ages should not be confused with the age differences between partners (mean difference and median difference), which must be computed within couples, not by observing men and women separately.
[3]A case in point is the median age at first intercourse, currently almost identical for boys and girls whereas formerly it was one or two years lower for boys (Bozon, 2003). Another example is the length of education, which is fairly similar for both sexes, although girls now study slightly longer than boys (Table B of Appendix II).
A case in point is the median age at first intercourse, currently almost identical for boys and girls whereas formerly it was one or two years lower for boys (Bozon, 2003). Another example is the length of education, which is fairly similar for both sexes, although girls now study slightly longer than boys (Table B of Appendix II).
[4]The impact of the Algerian war and the gradual reduction in the length of military service remain to be studied.
The impact of the Algerian war and the gradual reduction in the length of military service remain to be studied.
[5]Other INSEE surveys, and in particular the 1992 Enquête Jeunes (Survey on the Young), also highlighted, among other stages of the transition to adulthood, this delay in first couple formation in the younger generations (see for example Galland, 1995).
Other INSEE surveys, and in particular the 1992 Enquête Jeunes (Survey on the Young), also highlighted, among other stages of the transition to adulthood, this delay in first couple formation in the younger generations (see for example Galland, 1995).
[6]To avoid overloading Figure 3, we have reproduced only the female first marriage indicators; these, too, have been calculated from the 1999 EHF survey data.
To avoid overloading Figure 3, we have reproduced only the female first marriage indicators; these, too, have been calculated from the 1999 EHF survey data.
[7]The probabilities are computed for each age as the ratio of first unions to the number of single persons exposed to the risk, i.e. persons never having lived in union at that age. The period indicator is calculated by assigning to a synthetic cohort the probabilities of first union observed at each age in the course of a year, to obtain a mean number of first unions per 100 men or 100 women.
The probabilities are computed for each age as the ratio of first unions to the number of single persons exposed to the risk, i.e. persons never having lived in union at that age. The period indicator is calculated by assigning to a synthetic cohort the probabilities of first union observed at each age in the course of a year, to obtain a mean number of first unions per 100 men or 100 women.
[8]The parallel is even more striking when we look at the change in the unemployment rate among persons who have completed their education in the previous one or two years (Meron and Minni, 1995, diagram X, p. 25), especially for those with the lowest educational qualifications (diagram XI).
The parallel is even more striking when we look at the change in the unemployment rate among persons who have completed their education in the previous one or two years (Meron and Minni, 1995, diagram X, p. 25), especially for those with the lowest educational qualifications (diagram XI).
[9]Despite the disaffection of young people for marriage, the improvement in the labour market also seems to have produced a slight upturn in nuptiality between 1987 and 1990 (Figure 3). In contrast, the indices of first union formation merely stabilized during the nuptiality recovery of 1996-1997, which was driven largely by a change in tax laws for unmarried couples with children.
Despite the disaffection of young people for marriage, the improvement in the labour market also seems to have produced a slight upturn in nuptiality between 1987 and 1990 (Figure 3). In contrast, the indices of first union formation merely stabilized during the nuptiality recovery of 1996-1997, which was driven largely by a change in tax laws for unmarried couples with children.
[10]All the estimates of “final” proportions and the resulting mean ages have been calculated on the assumption that the age-specific probabilities of union formation were stable (in practice, to minimize random variations, we have averaged the probabilities of the preceding five years at the same ages). These estimates are reported here only when the observation covered at least about 80% of the cohort’s estimated first unions. Because the male timetable is later than that of women, the estimates for men can be reported only up to the 1969 cohort.
All the estimates of “final” proportions and the resulting mean ages have been calculated on the assumption that the age-specific probabilities of union formation were stable (in practice, to minimize random variations, we have averaged the probabilities of the preceding five years at the same ages). These estimates are reported here only when the observation covered at least about 80% of the cohort’s estimated first unions. Because the male timetable is later than that of women, the estimates for men can be reported only up to the 1969 cohort.
[11]The effect on nuptiality of imbalance between male and female populations was studied by Louis Henry (1966). Structural imbalances — arising, for example, from sex-selective migrations or heavy wartime losses — explain the differences in the proportions of single persons at age 50; short-term imbalances, if not too large, mainly give rise to adjustments of timing.
The effect on nuptiality of imbalance between male and female populations was studied by Louis Henry (1966). Structural imbalances — arising, for example, from sex-selective migrations or heavy wartime losses — explain the differences in the proportions of single persons at age 50; short-term imbalances, if not too large, mainly give rise to adjustments of timing.
© Cairn 2007
Vie privée |
Conditions d’utilisation |
Conditions générales de vente
À propos | Éditeurs | Bibliothèques | Aide à la navigation | Plan du site | Raccourcis
À propos | Éditeurs | Bibliothèques | Aide à la navigation | Plan du site | Raccourcis
Retour en haut de page
Ici s'achève le contenu de la page, la suite ne concerne que les attributs d'affichage
