Population  2002/4-5


2002/4-5 (Vol. 57)


publisher I.N.E.D

E-mail Alerts

Please provide your e-mail address to receive a Table Of Content alert every time a new issue is published.

you're reading

Recent Demographic Developments in France

byFrance Prioux[*]By the same author

France Prioux, Institut National d’Études Démographiques, 133, bd Davout - 75980 Paris, Cedex 20, tel. 01 56 06 21 44, fax 01 56 06 21 99,

e-mail: prioux.at.ined.fr

Page 687-728

Population growth and age structure


On 1 January 2002, the population of metropolitan France (mainland and Corsica) was estimated at 59,344,000 inhabitants (Doisneau, 2002), which represents an increase of 307,000 persons in one year, and a growth rate of 5.2 per 1,000 (Table 1) [1]  Tables 1-14 are grouped in the Appendix; they are numbered...[1]. That is the highest rate since 1992 and reflects a conjunction of several factors. The first was a number of births equal to that observed in 2000 (775,000) combined with a slight decline in deaths (528,000), resulting in a natural increase of 247,000 persons — the largest since 1988. The second factor was net migration, estimated at 60,000 persons, up slightly from the previous year’s 50,000, but down from the 80,000-90,000 a year observed in the early 1990s. This explains why the total population growth does not match that of the early 1990s, despite a larger natural increase. However, these are estimates subject to revision; even more important, there is no full registration of migration flows. Only the arrivals of some aliens are documented, but their departures are not recorded; nor are the flows of French residents between metropolitan France and the rest of the world or the French overseas départements and territories (DOM-TOMs).


Natural increase remains one of the highest in the European Union, and France is no doubt the country where natural increase contributes the largest share (four-fifths) of total growth [2]  However, according to data published by the Council...[2]. Conversely, immigration has been the only source of growth in countries where deaths outnumber births: Germany since 1972, Italy since 1993, Greece since 1997, and Sweden since 1998.


Thanks to the increase in births in 2000, and to their stability in 2001, the base of the French population pyramid has broadened somewhat (Figure 1). The upturn in the birth rate is, however, too modest to prevent the decline in the proportion of under-20s in the total population (Table 3), as the cohorts leaving that age group (the 1981 and 1982 cohorts in 2001 and 2002) outnumber the newborn (Figure 1). The proportion of persons aged 60 and over has not grown for three years: this is because those ages are being reached by the smaller cohorts born during World War II, in particular the 1942 cohort. But among persons aged 60 and over, the share aged 75 and over is steadily increasing as larger cohorts enter the age group and the “depleted cohorts” of World War I die out.

Age pyramid of the French population on 1 January 2002

Note: Birth-cohort sizes are available on the INED website (http:// www. ined. fr).

Source: INSEE.


Immigration between 1994 and 1999


This year, INED did not have access to the Ministry of the Interior’s files on residence permits issued in 2000. We therefore reproduce a portion of our analysis from last year concerning the statistics for 1994-99 (Prioux, 2001).


INED estimates the number of arrivals of aliens using the tabulation of initial permits for stays of one year or more, in compliance with United Nations recommendations on statistical standardization (Thierry, 2001a). The number of aliens obtaining French residence permits [3]  In the remainder of the text we use the term “permits”...[3] fell from 120,000 in 1994 to 106,000 in 1995, stayed unchanged in 1996, rose to 127,000 in 1997 and to 156,000 in 1998, then dipped to 145,000 in 1999 (Table 2). The special “legalization” programme carried out in 1997 and 1998 largely explains the rise during those two years. In 1999, the authorities issued 39,000 more initial permits than in 1996, even as the special “legalization” programme was drawing to its close.


Another way to measure the increase in entries since 1996 is to look at the numbers in the lower part of Table 2. Among the initial permits issued each year, we can single out those issued to aliens who arrived in that year. This category grew by one-third, from 77,000 in 1996 to 101,000, in 1999. As not all entrants obtain a one-year permit in the first year of their stay, there is no doubt that the number of entries has risen sharply in recent years and that it now exceeds 100,000 persons annually.


Permits issued to aliens eligible for free circulation in the European Union (EU) have been fluctuating around 43,000 for several years, with a slight fall in 1997. The total increase is thus due to arrivals from outside the European Economic Space (EES) [4]  EU member states, plus Iceland, Norway, and Liecht...[4]. Including the special regularizations, the permits delivered to that category jumped to 86,100 in 1997 and 112,800 in 1998. Although they fell back to 102,300 in 1999, the numbers remain well above the 1996 figure of 62,700. Generally speaking, there is a diversification of the immigrants’ origins and an increase in the percentage of arrivals from Africa, Asia, and European countries outside the free-circulation EES (Thierry, 2001a).


An analysis of the grounds for issuing a permit also reveals that the fastest-growing category is migration for family reunification, which has doubled for non-EES citizens since 1996 (from 17,700 to 35,400). The presence of a family in France and mixed marriages facilitate the granting of permits to aliens, some of whom then enter the labour market. This doubtless explains the fairly moderate rise in the number of non-EES citizens admitted as “workers” from 6,000 in 1996 to 9,200 in 1999, after a surge in 1997 and 1998 because most legalizations were granted for that reason [5]  In principle, the direct entry of “workers” is officially...[5]. The number of students coming from non-EES countries has also risen, from 17,600 in 1997 to 25,200 in 1999. While immigration for family reunification originates mainly in North Africa and Turkey, most students come from developed countries such as Japan and the United States.

From immigration to net migration


Most initial permits are issued for one year’s duration [6]  As noted earlier, our analysis is confined to permits...[6]. Of a total of 130,000 permits for one year or more issued to adults in 1999, 81,000, or 62%, were valid for one year only (43% for citizens of the EES countries, 71% for non-EES citizens). While it is impossible to know exactly how many aliens leave France when their residence permit expires, analysis of renewal rates for one-year permits shows that returns to the countries of origin are far from negligible: on average, only six permits in ten are renewed. The proportion is slightly over one-third for EES-country citizens, and seven in ten for non-EES citizens (Thierry, 2001b). However, these results do not allow an estimate of net migration, especially since we would also need data on returns to home countries after longer stays, as well as on the flows of French residents between France and abroad, and between metropolitan France and the overseas départements and territories. But the statistics do show that a significant percentage of aliens legally resident in France do not settle in the country for extended periods.


Recent trends


Because of changes in the population’s age structure (Figure 1) and the arrival of increasingly sparse cohorts at the childbearing ages, the mean number of women of fecund age has been gradually falling for ten years. The stabilization of the number of births in 2001 is therefore due to a very slight rise in fertility. The total fertility rate, which stood at 1.88 children per woman in 2000, is estimated to have reached 1.90 in 2001. If Irish fertility does not rise, France may thus well rank first in western Europe in 2001. In 2000, France lagged behind Ireland — traditionally the most fertile country — by a mere one-hundredth of a point (Table 6). The only other country that comes near is Norway, with 1.85 children per woman. The Mediterranean countries (Spain, Italy, and Greece) continue to have the lowest fertility rates — from Italy’s 1.23 to Greece’s 1.29 — followed by Germany and Austria at about 1.35. The United Kingdom, whose annual fertility has long remained on a par with France’s, seems not to have kept pace in recent years (1.65 in 2000). It is the only country, with Finland, to have registered a fall in 2000. In most other countries, fertility increased at the turn of the millennium, the steepest rise being in France.


Final data for 2000 confirm the rise in fertility at the younger ages, and the provisional data for 2001 further corroborate the trend [7]  We thank Lionel Doisneau, of the INSEE Demographic...[7]. After declining steadily for nearly twenty years, fertility among women under 25 has been edging gradually upwards since 1999 (Figure 2). With the rise in the fertility of older women having in the meantime continued or even accelerated, the total fertility rate jumped by nine points in 2000 (Table 4).

Age-specific and total fertility rates since 1975 in France (sum of individual years of age, per 1,000 women)
Source: INSEE, vital registration.

Fertility of younger women in western Europe


France’s experience is in line with that of many western European countries, where the decline in the fertility of the young stalled since the mid-1990s. The decline has stopped in Ireland, Finland, Germany, Switzerland, Spain, and Portugal, and has even reversed slightly in recent years in Luxembourg (in 2000) and the Netherlands (since 1997). It is true that this is a far cry from the recovery in the fertility of younger women in Sweden between 1984 and 1990 (more than 20% among women aged 20-24). That phenomenon contributed to the spectacular surge in Swedish fertility, which drove its total fertility rate to the top European rank in 1990, at 2.13 children per woman. However, the recovery of Swedish fertility was short-lived, and fertility in the 20-24 age group is now half what it was in 1990.


For the time being, this stabilization or slight upturn in fertility of the young in a few countries — or its much slower decline — may herald, if not a trend reversal, at least the end of a long-term tendency registered in all developed countries: the postponement of childbearing to older and older ages. Yet fertility rates of young women still diverge widely in the developed world (Table A). In the United Kingdom, for example, where fertility still starts very early and has stopped declining, women under 20 give birth to five times as many children as in Italy, Switzerland or Sweden (0.15 child per woman versus 0.03). The U.K. also leads Europe in the 2024 age group, with 0.35 children per woman in 2000, nearly three times as many as in Spain (0.12 in 1999). France is in a lower range under 20 (0.05), but for 20-24 it comes close to the U.K. with 0.32. The similarity in trends, therefore, does not imply a convergence in fertility levels or timing: witness the mean age of childbearing, which ranges from 28.2 years in Austria [8]  Despite its holding the EU record for early births,...[8] to nearly 31 years in Spain (30.7 in 1999).

Table A –  Fertility at 15-19 and 20-24 in western European countries, 1995 and 2000 (per 100 women)

Cohort fertility


The upturn in annual fertility does not actually undermine last year’s projections for completed cohort fertility (Table 5). This is because the slight recovery of fertility at younger ages involves cohorts born after 1975 or thereabouts for which a projection would be premature. For cohorts born in the 1960s, all of which had reached their thirties by 2000, the persistence of the increase entails only a very modest revision—one point, at most—of the trend projection [9]  The projection recalculated with constant rates, based...[9]. How should we now assess the completed fertility of these cohorts, which have been the most heavily affected by the low-fertility phase of the early 1990s? Our starting point will be the 1960 cohort, whose reproduction rate can now be determined with certainty: its completed fertility stands at 212 children per 100 women. Table B compares cumulative fertility, from age 28, for women in the 1960, 1963, 1966 and 1969 cohorts.

Table B –  Cumulated fertility of women by ages 28 to 38 in the birth cohorts of 1960, 1963, 1966, and 1969 (Average number of children per 100 women)

As we can see, the lag accumulated by women of the 1963 cohort before age 28 begins to narrow slightly only after age 32. This compensatory mechanism, which weakens with age, will never completely close the gap. According to the trend projection [10]  Although we have taken the provisional 2001 rates into...[10], completed fertility for the 1963 cohort may only reach 207 children per 100 women, five fewer than for the 1960 cohort.


At age 28, the women of the 1966 cohort start out with an even larger deficit relative to the 1963 cohort (minus 14 children per 100 women) than the latter’s deficit relative to the 1960 cohort (minus 12 children). But the compensatory process begins earlier, so the lag is smaller by age 31. At 35, however, there is still a gap of seven children per 100 women, which is also unlikely to be made up totally. Completed fertility for the 1966 cohort is now projected at 203 children per 100 women.


The birth deficit of the 1969 cohort relative to the 1966 cohort appears to be much smaller. At six children per 100 women at age 32, it is still quite likely to be closed. Nevertheless, the projection of the current trend indicates a small decline of two children per 100 women compared with the 1966 cohort, resulting in a completed fertility of 201. The increase in fertility at older ages would therefore have to continue and even accelerate in order to halt the decline in completed fertility. If this happens, completed fertility could level off at slightly over two children per woman beginning with the late-1960s birth cohorts.

Completed fertility and family size


Thanks to the family history survey that accompanied the 1999 census [11]  The Study of Family History (Étude de l’histoire familiale...[11], we can now make a retrospective assessment of the distribution of completed family sizes in successive cohorts, and estimate it for the cohorts that have not yet reached the end of their fecund life (Table C) (Toulemon and Mazuy, 2001). Between the 1940 and 1950 cohorts, completed fertility has declined largely because of the sharp decrease in the number of large families (four children or more), mostly in favour of two-child families (from one in three women to two in five). There has been little change in the share of the other size categories, although we find a very small decline in the proportion of mothers of three children between the 1940 and 1945 cohorts, a temporary fall in the proportion of childless women, and a slight increase in the proportion of mothers with only one child. Between the 1950 and 1960 cohorts, the stabilization of completed fertility around 2.1 children per woman coincides with the end of two long-term trends. First, the number of large families has stopped declining: one woman in ten is the mother of four or more children. Second, the concentration on two-child families seems complete, at around four women in ten. Only small marginal shifts are observed, most notably between one- and three-child families. As for the subsequent fall in completed fertility, it appears to originate exclusively in the extreme sizes: the share of large families seems to be shrinking again, while childlessness, which has stayed close to 10% since the 1935 cohort, could reach 13% in the 1968 cohort [12]  Estimate obtained with constant first-birth probabilities,...[12].

Table C –  Distribution of women born since 1940 by final number of children born alive (per 100 women)

The relatively low level of infertility and the persistence of a fairly high proportion of families with three or more children (three women in ten since the 1950 cohort) explain why French fertility remains one of the highest in the European Union.

Completed fertility and mean age of childbearing in western Europe


Table 7 brings together estimates of completed fertility and mean age of childbearing for the 1950, 1955, 1960, and 1963 cohorts for the same countries as Table 6, using an identical method in all cases [13]  Calculations and estimates were made at the European...[13]. For the older ages not yet observed, the rates have been “frozen” at their last known values. The estimates are virtually certain for women born in 1950 and 1955, now at the end of their fecund life, and are near-certain for the 1960 cohort, aged 40 in 2000 [14]  The most recent observation year is usually 2000, or...[14]. But for the countries without up-to-date statistics, and especially for the 1963 cohort, the estimates are open to revaluation, as the method employed causes completed fertility to be understated when fertility at older ages is rising — a fairly general phenomenon across western Europe. We therefore do not regard a decline in completed fertility as certain unless the gap between the estimated completed fertilities of the 1960 and 1963 cohorts has reached at least 0.04 children per woman, and can therefore no longer be totally closed at these relatively old ages.


The fall in completed fertility that was already steep between the 1950 and 1960 cohorts persists at a brisk pace in later cohorts in Germany, Spain, Greece, Ireland, Italy, and the United Kingdom. It is also certain or near-certain in Austria, Belgium, France, the Netherlands, Portugal, Sweden, and Switzerland. The only exceptions to the general trend are the northern countries (apart from Sweden) and Luxembourg. Their completed fertilities either rise slightly or remain stable after the 1955 cohort; in Denmark and Luxembourg, we can already be certain that the rise in completed fertility after the 1960 cohort will continue.


France, whose fertility has long been nearly the highest in Europe, could thus be overtaken by Norway, owing to the contraction of completed fertility after the 1960 cohort. The country with the lowest fertility may turn out to be Germany (1.57 in the 1963 cohort), below Italy (1.60).


This fairly widespread decline of completed fertility in western Europe is accompanied by a delay in childbearing. Mean age of childbearing has risen in almost all of Europe by at least one year between the 1950 and 1960 cohorts; the increase is nearly two years in Luxembourg, the Netherlands, and Norway, and even slightly more in Denmark. Only Ireland and the southern countries lag slightly behind, as the rise in mean age did not begin there until after the 1955 cohort, but initial estimates for the 1963 cohort indicate that the trend persists and is even gathering momentum in that group. Delayed childbearing largely explains why the total fertility rates are particularly low in comparison with completed fertility values (Tables 6 and 7).



We do not have new data on induced terminations of pregnancy with which to update Table 8. The latter is based on an analysis of the abortion registration forms (IVGs, interruptions volontaires de grossesse [“voluntary interruptions of pregnancy,” as they are officially referred to in France]), which are still not available for the years after 1997. We cannot tell, therefore, if the characteristics of women who undergo abortions are changing, and in particular if multiple abortions are on the increase or if the teenage abortion rate is still rising. The most recent available data indicate that the greater frequency of abortion among minors was due not to a rise in conceptions but to a more frequent choice of abortion in the event of pregnancy (Prioux, 2001). How has the situation evolved since 1997?


As we do not even know the number of registration forms, it is equally hard to establish whether the increase observed in recent years in the annual statistics from healthcare institutions (SAE: statistique annuelle des établissements de santé) reflects reality or better record-keeping (Table D, column 4). The figures are based on annual data supplied by each institution on the procedures performed under specific rubrics, from which the abortion data can be extracted [15]  Apparently, these statistics include only abortions...[15]. Formerly, the comparison with the number of registration forms completed by each institution led the Health Ministry’s statistical office to revise the figures, as the number of forms sometimes exceeded the number of procedures reported by the institutions (Le Corre and Thomson, 2000). With these forms not being processed, that cross-check can no longer be carried out.

Table D –  Number of abortions by source since 1990, and number of live births

Is it plausible that the number of abortions performed in France in 1998 and 1999 is greater than in 1990, bearing in mind that medically-administered contraception has increased at all ages, and that the proportion of women using no contraception despite their stated wish to avoid pregnancy has remained stable (Leridon et al., 2002)? The rising prevalence, including among teenagers, of medically-administered contraception— deemed safer than the traditional methods it replaces — could, on the contrary, be expected to reduce the resort to abortion. It is therefore particularly important to have access to detailed statistics on the characteristics of women undergoing abortion, so as to better understand the increase — if there is one.


Number of marriages and PACS


The new millennium seems to be as propitious to marriages as it is to births: in 2001 as in 2000, the number of legalized unions is estimated by INSEE at about 305,000 (Table 9), the highest since 1982 (312,000). Yet, since the end of 1999, the “Civil Solidarity Pact” (PACS: Pacte Civil de Solidarité) provides an alternative to marriage for couples who do not wish to marry formally but nevertheless want to make their union official or take advantage of certain tax breaks. The number of PACS signed before district courts since the law came into force is far from negligible: between 15 November 1999 and 31 March 2002, 55,643 contracts were concluded in metropolitan France (Table E). Although the statistics do not indicate the partners’ sex, we may assume that a large proportion of the couples are heterosexual. Unfortunately, we cannot go any further in analysing this new choice available to couples, as there is no provision for statistical recording of the partners’ characteristics. At most, we can observe that PACSs seem to be settling into a seasonal pattern altogether different from that of marriages: the second and third quarters of the year are fairly slow months (just over one PACS in three signed in 2000 and 2001), whereas an overwhelming majority of marriages are celebrated between April and September (more than three-quarters for the same years). The first trimester is the peak season for PACS, and we even note an upturn in 2002 compared with 2001. If the tax incentive were a strong motivating factor, however, the peak would have been expected at the end of the year [16]  PACS partners must wait two calendar years before they...[16].

Table E –  Civil Solidarity Pacts (PACS) and marriages recorded in metropolitan France since 15 November 1999

First marriages


In 1996, the rise in nuptiality involved the divorced as much as the never-married, but that is no longer the case. Since 1997, the nuptiality of the never-married has been rising faster (or declining less) [17]  The apparent decline in nuptiality for 1998 is probably...[17] than that of the divorced (Figure 3). If the preliminary estimates are confirmed, the period indicators of nuptiality for never-married women would be 0.63 for the sum of the age-specific first marriage rates, 0.69 for the overall probability of first marriage, and two or three hundredths of a point lower for men (Table 9). Depending on the indicator chosen, these are the highest values since 1982 (sum of rates) or 1991 (overall probability). The discrepancy is due to the methods of construction of the two indices, which give them different and complementary explanatory powers [18]  Our first indicator uses the sum of age-specific ratios...[18]. The annual frequency of first marriages (sum of rates) has fallen more sharply, and is currently rising faster, than the first-marriage probability for the never-married, as the long-term drop in nuptiality has substantially increased the “stock” of never-married exposed to the risk. Indeed, a rise in the annual frequency of marriages does not necessary imply an increase in the overall probability, as was the case in 1987-90. In recent years, the overall probability has been rising, as the frequency of marriage is no longer decreasing at the youngest ages, and is even starting to edge up again.

Total first-marriage rates (sum of age-specific rates) and total rates of remarriage of the divorced (sum of rates by duration since divorce) (per 100 men or 100 women)
Sources: INSEE, vital registration, and Ministry of Justice.

Cohort-specific marriage frequency


The sum of age-specific marriage rates, in a cohort, gives a direct view of the proportion of persons already married at each age (Figure 4). The fall of nuptiality is obvious. For example, by age 30 [19]  More precisely, at the end of the year following their...[19], 85% of women born in 1950 had already been married at least once; today, at the same age, only half of the 1971 cohort has been married. For men, the fall is even steeper, with the proportion of married men having been halved from 78% to 39% in the same cohorts. Yet the projected proportions ever-married at age 50 (which, however, it would be somewhat risky to extend to the 1971 cohort) indicate a much smaller drop. Above age 24 for women and above age 26 for men, the curves diverge gradually, with a steady rise in the number of marriages between successive ages (Figure 4). According to our projections, therefore, the proportion ever-married will reach 71% for women of the 1969 cohort (down from 90% in the 1950 cohort), and 68% for men of the 1967 cohort (down from 87% in the 1950 cohort) (Table 10). The hypotheses underlying these projections are conservative. While we have taken account of the provisional data for 2000 and 2001 to compute the proportions already married by the end of 2001, we assume that the marriage probabilities observed in 1996-99 will remain constant in the future. The projections are therefore likely to be revised upward if the provisional figures of 2000 and 2001 for the truncated cohorts are confirmed.

Proportions of ever-married men and women at selected ages, by birth cohort (cumulated first marriage rates per 100 men or 100 women)
Source: Calculations and projections from INSEE data.

In conclusion, therefore, despite the recovery of nuptiality indices in recent years, the fall in the proportion of persons married before age 50 could still prove to be very large.

The nuptiality of young cohorts


Another fact appears clearly from Figure 4. The proportions of both males and females ever- married have stopped declining since approximately the 1975 cohort. This should be viewed in conjunction with the small increase in the fertility of women under 25 noted earlier. The steady increase in age at marriage and childbirth is very likely coming to an end with the cohorts born in the second half of the 1970s.



After a peak of 119,000 in 1995, and a relative stability from 1996 to 1999 (between 116,000 and 117,000), the number of divorces fell to 114,000 in 2000 (Table 9). With 4,000 fewer divorces than in 1995, the crude divorce rate of the married population is unchanged, however, at 38.2 divorces per 100 marriages, because the fall in nuptiality has reduced the number of marriages exposed to the risk. Since 1995, this period indicator has fluctuated between 38 and 39 divorces per 100 marriages. Does this mean that, on average, over 38% of marriages end in divorce? In fact, this figure is not very far from the frequency of divorces that can be estimated today in the 1985 marriage cohort. Assuming stability of the average rates observed in 1996-2000 [20]  The hypothesis is supported by the stability of divorce...[20], nearly 37% of the 1985 marriages would end in divorce, as against 35% of 1980 marriages, 32% of 1975 marriages, and 29% of those of 1970. The frequency of divorce is therefore still rising, but much more slowly since the 1970 cohort. Only 16% of couples married in 1960 have broken up (Figure 5).

Proportion of marriages terminated by divorce for selected marriage durations (out of an initial 100 marriages)
Sources: INSEE, vital registration, and Ministry of Justice.

For couples married after 1985, the rise is likely to continue, as the proportion of marriages dissolved before the tenth anniversary is still rising (Figure 5). But the trend may stop with the cohorts of the 1990s. After peaking in the 1991 cohort, the proportion of marriages dissolved in the early durations decreased in the 1992-95 cohorts. This is the first time in 30 years that divorces have stopped increasing for the earliest marriage durations recorded. As with fertility and nuptiality, we may be witnessing the end of a long-term trend.


Mortality in France and Europe


The provisional number of deaths recorded in 2001 is 528,000, or 8,000 fewer than in 2000, and 10,000 fewer than in 1999 (Table 1). The decline raised the mean length of life by three-tenths of a year in 2001. This puts life expectancy at birth over 83 years for women, and over 75.5 years for men (Table 11) — a gap of 7.5 years, which appears to have stabilized in recent years. After a few years of slower increase, female life expectancy is again rising at the same pace as that of males [21]  Estimates for 2000 and 2001 are provisional.[21].


In most of western Europe, female life expectancy now exceeds 80 years (Table 12). Only Denmark, Ireland, and Portugal have not yet reached that threshold; their values ranged between 79.2 and 79.6 in 2000. With a mean length of life of 82.7 years, Spain is poised to catch up with France for first place. Men live longest in Sweden (77.4 years) and their life is shortest in Portugal (72.6 years), which ranks far below all the other European countries. France’s position — tenth in 2000 — is tending to deteriorate, as male expectation of life is now slightly higher in Austria and the United Kingdom, and this was not the case a few years ago.


The fall in infant mortality has shown signs of stalling in recent years (Table 11), but, with 4.5 deaths under one year for 1,000 newborn in 2000, France has one of the lowest rates in Europe (Table 13). Only the Scandinavian countries — Sweden, Finland, and Norway — have succeeded in reaching a level below four deaths per 1,000 newborn.

Fall in mortality by broad age groups and increase in life expectancy


Table F gives an overview by five-year periods of the ages responsible for the increase in female and male mean length of life over the past fifteen years, given the share of mortality that can be attributed to each age group in the computation of life expectancy at birth. To avoid random fluctuations, our computations are based on INSEE’s multi-year tables, the most recent of which covers 1997-99 [22]  To simplify the layout of Table F, we have only noted...[22].

Table F –  Gains In life expectancy in years, by large age group and sex, 1983-1988

Between 1983 and 1988, life expectancy at birth rose at almost the same pace for men (0.28 years per year on average) and for women (0.29). The fall in child mortality (mostly in the first year of life) and in young-adult mortality (ages 15-44) accounts for only 20% or so of the improvement. Men achieved the largest gains between ages 45 and 75. The female distribution is more even: the fall in mortality of the oldest old contributed twice as much to the improvement of life expectancy for women as it did for men (0.65 years per year as against 0.34).


In the following five-year period, 1988-1993, life expectancy at birth rose more slowly (0.22 years per year for both sexes). Young-adult mortality stopped declining or increased a little. Gains slowed appreciably for women over 45 and for men between 45 and 75, but most of the improvement was still provided by the fall in mortality among women aged 75 and over and among men aged 45-74. In the final five-year period, 1993-98, the progress accelerated for men (0.28 years per year) and continued to slow down for women (0.17 years per year). For the first time, male life expectancy began to catch up with female life expectancy. Most of the male gains were achieved between ages 15 and 75, with a marked fall in mortality at 15-44. For women, the improvement was still largely due to the decrease in mortality at the oldest ages, but the gains were progressively smaller.


In fact, this differential fall in mortality by age and sex is explained by the uneven success of the fight against the dominant diseases in each age group. AIDS, most widespread among men aged 25-44, was the main obstacle to the decrease in young-adult mortality in the early 1990s, until the introduction of new therapies led to a spectacular drop in deaths. But at those ages, and until age 75 or so, the leading cause of death is cancers, which are responsible for the largest reduction in mean length of life. Since the early 1990s, the gains recorded by women have slowed down markedly, in particular because of smoking-related cancers, even as France began to register a slight decline of those among men (Table 14). Men have thus achieved the largest gains when the decline in mortality from neoplasms was no longer counteracted by the rise in smoking-related cancer. At the same time, the progress achieved in cardiovascular diseases— responsible for the earlier sharp fall in mortality at the oldest ages — began losing momentum. This contributed to the slower rise in expectation of life among women, who had been the main beneficiaries of that progress during the 1980s.


The reduction in excess male mortality stems from a relative improvement in the position of men, due to the success of new therapies and a better prevention of the hazards that affect men more specifically: AIDS, traffic accidents, and all the alcohol- and smoking-related diseases. But it also reflects a deterioration in the position of women, since the resistance to the decline in certain cancers is on the contrary due to increasing female tobacco consumption.

Households and families in the 1999 census


Between censuses, mean household size, i.e. the mean number of occupants per dwelling, has been decreasing fairly steadily. The number fell from more than three in the 1968 census to 2.6 in 1990 and 2.4 in 1999 (Cristofari and Labarthe, 2001). The trend is conditioned by a wide range of behaviour: union formation and dissolution, birth of children, departure from (and, in some cases, return to) the family home, move into institutional accommodation, and death are the principal events, among others, that help create new households, cause the disappearance of others, and alter household composition and size. Most are closely linked to the changes in family behaviour in recent decades: more frequent union terminations, shift away from marriage, delay in the formation of a stable couple, reduction in family size, rise in age at first birth, and so on. These changes are also linked to the fall in mortality, to changing attitudes toward inter-generational cohabitation, and to residential independence of young people.

Living alone


An initial factor contributing heavily to the decrease in household size is the rise in the number of single-person households. In 1968, one dwelling in five was occupied by a person living alone; in 1982, nearly one in four; and by 1999, just over three in ten. Thus 7.4 million people, or 12.9% of persons living in private households, live alone at home (Chaleix, 2001) (Table G).


Women have consistently outnumbered men in single-person dwellings. Between 1968 and 1982, nearly twice as many women as men lived alone. This is not because women have a strong preference for living alone but mainly because of their longer life span, which compounds the effect of the usual age difference between spouses: more than half (52%) of women living alone in 1999 were aged 65 and over. But since the 1980s, the number of men living alone has generally risen faster than that of women living alone. Today, 40% of persons living alone are men.

Table G –  Changes in the proportion of one-person households in the census, 1968-1999

We can better grasp the trend’s underlying causes by observing the age-specific proportions of men and women living alone and living with a partner in the last two censuses (Figure 6). Between ages 25 and 50, it is men, not women, who are more likely to live alone [23]  These proportions are calculated with the general population...[23]. The trend actually gathered momentum between the 1990 and 1999 censuses, largely explaining the “catching-up” by the population of males living alone. Among men aged 25-49, the numbers living alone have risen so fast that the ratio now stands at 143 men for 100 women, up from 106 in 1990. One-half of all men living alone are aged 25-49; of women living alone, only 23% belong to that age group.

Proportion of men and women living alone and living with a partner in the censuses of 1990 and 1999 (per 100)

Note: Proportions calculated taking institutional households into account.

Source: INSEE, population censuses.

The rise in the proportions of men under 75 and of women under 60 living alone between the last two censuses has been accompanied by a fall — for men, mostly under 55 — in the corresponding proportions of persons living in union (Figure 6). Under age 25 or 30, the main factors involved are the increase in the age at first union and the greater number of young people leaving home to live on their own. Over age 25 or 30, the chief cause is the greater frequency of union terminations. When a couple with children breaks up, the man is more likely than the woman to end up living alone, if he does not form a new couple immediately. This explains why the rise in the proportion of singles is greater among men aged between 28 and 50, and particularly between 35 and 50. Women in the same age group register a smaller increase. They are more likely than men to live in a one-parent family after the breakup; after age 50, the departure of their children is more likely to leave them living alone, if they have not formed another couple.


At all the other ages except between 25 and 50, women are more likely to live alone than men, and the phenomenon increases with age, at least until 83 to 85, when the proportion of women living on their own peaks around 55%. That is a classic effect of women’s longer life span and of the age difference between spouses. Nevertheless, and contrary to the general trend, the proportion living alone has fallen among women aged 62-78 and men aged 77-86, while the proportions of women over 60 and men over 75 living in union has increased. With the rise in life expectancy, widowhood occurs later on average. As a result, life as a couple extends to older ages. We also note that this translates into a slight decline in the proportions of individuals living in institutional households, at least before age 85 (Prioux, 2001, pp. 578-80). In the future, however, the increased frequency of divorce and separation will probably more than offset the beneficial effect of lower mortality on the “residential solitude” of the elderly (Gaymu and Pennec, 2001). Indeed, there are rather few divorced people in today’s older cohorts. In the following cohorts — i.e. among men and women now in their fifties — the increase in union terminations is bound to contribute to the rise in the proportion of persons living alone. Given that individuals are less likely to form a new couple as they grow older, the proportion of persons entering old age alone may eventually rise.


Beyond 80 years, the proportion of women living alone rises sharply, despite the rise or stability of the corresponding frequencies of unions. This may be viewed as the effect of the near-disappearance of intergenerational cohabitation. Thanks to the improvement in health and living conditions among the elderly, a growing number of older women can live alone. Moreover, the expansion of home help services and of institutions such as sheltered housing provide an alternative to old people’s homes, as long as the individual does not become dependent.

Living in a family


In 1999, 16 million French households — or two out of three — comprised at least one “family” according to the census definition [24]  i.e. a couple, with or without children, or a father...[24] (versus 71% in 1990), and slightly over 85% of the household population lived in a family (down from 88% in 1990) (Table H). The cohabitation of several families in a single dwelling has become fairly rare in France. These households represented less than 1% of the total in the 1982 census. In 1999, as in 1990, 0.6% of households included two or more families, and less than 1.5% of the population lived in such households. The number and population of family households comprising “isolated” members [25]  Under the census definition, “isolated” persons may...[25] have decreased since the census of 1990, owing to the decline in intergenerational cohabitation. In fact, many of the persons considered to be isolated in the census are the reference person’s relatives in the ascending or descending line [26]  The census housing schedule does not allow an identification...[26].

Table H –  Households and population in households by family composition

Despite their decline in relative terms, the number of family households and their population have continued to increase between censuses, even as their average size has decreased: between 1990 and 1999, the mean size of family households has fallen from 3.10 to 2.99 persons. While the decrease in the number of large families has contributed to this steady decline between censuses, two other factors are now mainly responsible: the increase in cohabiting childless couples, and the sharp rise in one-parent families.

An increase in the number of childless couples


In 1999, 6.3 million households (or 39% of families), consisted of childless couples. The category includes couples who have not yet had children and — for the most part — couples whose children have left home. Back in 1990, only 5.5 million households fell into that category (Table I). This steep increase essentially concerns elderly couples, among whom widowhood occurs at later and later ages. The number of childless couples in which the woman is over 65 has risen by one-quarter since the previous census. But the numbers are also rising at other ages, owing to shifts in fertility patterns. Among the youngest couples, later first births are lengthening the pre-parental phase, thereby increasing the number of childless couples. For the intermediate cohorts, the reduction in completed fertility means a longer period in which couples live together after their children have left home.

Table I –  Families by marital status of adults and presence of children

The vast majority of couples — 82% in 1999 — report themselves as married; but the predominance of marriage is weakening rapidly, since the number of married couples has fallen, while that of unmarried couples has risen by 46% (Table I). The latter increase has been particularly rapid for couples with children. As a result, while most unmarried couples used to be childless, the number of unmarried couples raising children is now just as high as that of unmarried couples living without children. Meanwhile, the number of married couples with children has fallen by almost one million between the two censuses; the only category of married couples to have expanded (by nearly 500,000) is that of couples living without children. They increasingly consist of elderly couples whose children have left home. The marriage slump since the late 1970s initially led to a sharp increase in the number of childless unmarried couples in the 1980s. The persistent rise in fertility outside marriage explains the rapid rise, in the 1990s, in the number of unmarried couples with children. But the married couple remains by far the dominant model, as changes in behaviour take a long time to produce structural effects. Finally, unmarried couples are still, on average, younger than married couples, and will continue to be so in the future, not only because earlier generations were more committed to marriage, but because many young people now choose to live together several years before legalizing their union.

More one-parent families


Nearly two million French families were enumerated as one-parent families in 1999 [27]  Including when the one-parent family is not the principal...[27]. This is an increase of almost 400,000, or 24%, since 1990 [28]  For consistency with all the previous analyses, we...[28]. More than five million people live in one-parent families, versus four million in 1990 (Table I). If the count is limited to families raising children under 18 years old — the definition adopted by most European countries — the number rose from 894,000 in 1990 to 1,212,000 in 1999. That represents an even steeper relative growth of 36%. Of this total, 60% have only one child under 18.


More than 85% of one-parent families consist of a mother and her children: after a divorce, child custody is usually awarded to the mother. But while divorce has become the leading cause of one-parent families, it is not the only one. There is also early widowhood, the break-up of unmarried couples, or the birth of a child to women without a partner. The latter two situations are reflected in the rapid increase in the number of fathers and, even more so, of mothers describing themselves as unmarried. (For more details on one-parent families, see Algava in this issue).

The parental environment of children seen through the censuses


Unfortunately, the parental universe of children cannot be described fully on the basis of censuses, which include no questions on step-parenthood. However, we can determine whether the child lives alone with its father or mother, or with a couple that includes at least one of its parents. We can refine the analysis by examining the marital status reported by the parents.


Figure 7 shows the speed with which the shift away from marriage has transformed children’s parental environment. The situation that used to be the virtual rule — a child living with married parents — lost ground rapidly between 1990 and 1999. The main gains are for unmarried couples: of children born in the census year or in the preceding one (ages 0-1), three in ten were living with unmarried parents in 1999, up from two in ten in 1990. Among older children, the situation is slightly less common. In 1999, two children in ten were living with an unmarried couple at age four; at age 13, one in ten. This is due to an age effect (some parents marry after the birth of their children) and a cohort effect (births outside marriage were less common some years ago). The progression of the phenomenon across the cohorts is also visible in the comparison of values observed at the same age in the two censuses. At age 10, 6.5% of children born in 1980 lived with unmarried parents, a proportion that has risen to 12.5% for children born in 1989.

Living arrangements of children between ages 0 and 18,in 1990 and 1999 (per 100 children living in a family)
Source: INSEE, population censuses.

As a result of the increase in divorces and in the breakup of couples (married or not), a growing number of children live in one-parent families. Again, the progression is notable. In the youngest age group (0-1), the percentage rose from 6.2 in 1990 to 9 in 1999; at age 10, the proportion went from 10% for the 1980 cohort to 15% for the 1989 cohort. As the children grow older, the proportion tends to increase with separations, although the blending of families attenuates the rise: at age 15, 12.2% lived with a single parent in 1990, and 16.6% in 1999. The curve’s gentler slope in 1999 indicates that children find themselves in this situation at an ever-earlier age. Among them, a growing number live with mothers describing themselves as never-married: whether the mother has never lived with a partner or has left him, we are still dealing here with a consequence of the retreat from marriage.


In sum, the proportion of children living with a married couple has fallen sharply at all ages, and particularly among the very young. At age 0-1, it dropped from 74.3% in 1990 to 60.7% in 1999 [29]  These proportions are slightly higher than those of...[29]. At age 10, 83.5% had married parents in 1990; in 1999, 72.7%. The age effect (due to marriage of parents after birth) and the cohort effect (births outside marriage less common a few years ago) exert their influence in the same direction, but it is dampened by union terminations. Here as well, at all ages, the married couple remains the norm, even if it is sometimes a blended family.

The Statistical Data

Table 1 –  Population change (in thousands) and crude rates (per 1,000)(a)
Table 2 –  Legal long-term immigration by geographic origin
Table 3 –  Age distribution of the population of metropolitan France on 1 January (%)
Table 4 –  Fertility since 1970
Table 5 –  Cohort fertility: cumulative fertility to selected ages, estimated completed fertility (mean number of children per 100 women), and mean age of childbearing (in years and tenths of years)
Table 6 –  Total fertility rates for western europe (average number of children per woman)
Table 7 –  Cohort fertility in Western Europe
Table 8 –  Number of abortions and annual indices since 1985
Table 9 –  Characteristics of nuptiality and divorce since 1985
Table 10 –  Characteristics of nuptiality by birth cohort
Table 11 –  Characteristics of overall mortality since 1985
Table 12 –  Life expectancy at birth in Western Europe in 2000
Table 13 –  Infant mortality in Western Europe (rate per 1,000 live births)
Table 14 –  Standardized death rates by sex and groups of causes of death, 1990-1998
Table 14 –  Standardized death rates by sex and groups of causes of death, 1990-1998


  • Algava É., 2002, “France’s one-parent families in 1999, Population, 57(4-5), pp. 729-752.
  • Beaumel C., Doisneau L., Vatan M., 2002, “La situation démographique en 1999” INSEE Résultats, Société, no. 3, 4 p.
  • Cassan F., Mazuy M., Clanché F., 2001, “Refaire sa vie de couple est plus fréquent pour les hommes”, INSEE Première, no. 797, 4 p.
  • Chaleix M., 2001, “7,4 millions de personnes vivent seules en 1999,” INSEE Première, no. 788, 4 p.
  • Council of Europe, 2001, Recent Demographic Developments in Europe, 2001, Strasbourg, Council of Europe Publishing.
  • Cristofari M.-F., Labarthe G., 2001, “Des ménages de plus en plus petits”, INSEE Première, no. 789, 4 p.
  • Doisneau L., 2002, “Bilan démographique 2001. Le regain des naissances et des mariages se confirme”, INSEE Première, no. 825, 4 p.
  • Gaymu J., Pennec S., 2001, “Does life expectancy as a couple increase? The example of the elderly”, paper at European Population Conference, EAPS, Helsinki, Finland, June 7-9, 2001.
  • INSEE 2002, Ménages – Familles – Population totale. Tableaux thématiques. Exploitation complémentaire (March 1999 population census), 224 p.
  • Le Corre M., Thomson É., 2000, “Les IVG en 2000,” DREES, Études et Résultats, no. 69, 4 p.
  • Leridon H., Oustry P., Bajos N., and the COCON team, 2002, “La médicalisation croissante de la contraception en France”, Population et Sociétés, no. 381, 4 p.
  • Prioux F., 2000, “L’évolution démographique récente en France”, Population, 55(3), pp. 441-476.
  • Prioux F., 2001, “L’évolution démographique récente en France”, Population, 56(4), pp. 571-610.
  • Thierry X., 2001a, “Les entrées d’étrangers en France de 1994 à 1999”, Population, 56(3), pp. 426-50.
  • Thierry X., 2001b, “La fréquence de renouvellement des premiers titres de séjour”, Population, 56(3), pp. 451-68.
  • Toulemon L.,Mazuy M., 2001, “Les naissances sont retardées mais la fécondité est stable”, Population, 56(4), pp. 611-44.


[*] Institut National d’Études Démographiques, Paris.

[1] Tables 1-14 are grouped in the Appendix; they are numbered in identical order from year to year, not necessarily the same order in which they are mentioned in the text.

[2] However, according to data published by the Council of Europe (2001), Ireland registered negative net migration in 2000 and owed its population growth that year to natural increase alone.

[3] In the remainder of the text we use the term “permits” in connection with all aliens authorized to reside in France, even though minors are not required to possess a residence permit.

[4] EU member states, plus Iceland, Norway, and Liechtenstein.

[5] In principle, the direct entry of “workers” is officially authorized only for occupational groups where it is warranted by demand (as measured by the number of unemployed).

[6] As noted earlier, our analysis is confined to permits for a duration of one year or more.

[7] We thank Lionel Doisneau, of the INSEE Demographic Surveys and Studies Division, for providing us with the provisional age-specific fertility and nuptiality rates for 2000 and 2001.

[8] Despite its holding the EU record for early births, the United Kingdom posts a mean age of 28.5 years, since fertility at older ages is higher there than in Austria, a country where the timing of fertility is highly concentrated.

[9] The projection recalculated with constant rates, based exclusively on 2000, is necessarily higher.

[10] Although we have taken the provisional 2001 rates into account to establish the initial proportions, the projection is based on 1998-2000 trends.

[11] The Study of Family History (Étude de l’histoire familiale – EHF) survey was conducted on a sample of 145,000 men and 245,000 women.

[12] Estimate obtained with constant first-birth probabilities, which assumes that the increase in age at first birth is ending (see Toulemon and Mazuy, 2001, Figure 15).

[13] Calculations and estimates were made at the European Demographic Observatory and published in Council of Europe (2001).

[14] The most recent observation year is usually 2000, or 1999, except for Belgium (1995) and Italy (1996).

[15] Apparently, these statistics include only abortions charged at the standard rate (Le Corre and Thomson, 2000).

[16] PACS partners must wait two calendar years before they can file a joint tax return.

[17] The apparent decline in nuptiality for 1998 is probably due to defective marriage registration statistics (Prioux, 2001, footnote 21, p. 589). In 1999, INSEE had to adjust its marriage statistics to allow for forms that were not returned—mostly by rural municipalities (Beaumel et al., 2002, p. 351).

[18] Our first indicator uses the sum of age-specific ratios of the first marriages to the total population of the cohort concerned, irrespective of marital status. The sum could be interpreted directly as the cohort-specific first-marriage frequency, if the timing of nuptiality were invariable. But when age at marriage rises, the sum’s value may be markedly smaller than the cohort-specific intensity of nuptiality. For the second indicator, the probabilities are computed by relating first marriages to the persons at risk of marriage only, i.e. the never-married. The overall probability summarizes the age-specific probabilities of marriage throughout life in a hypothetical cohort, yielding a mean number of marriages per person. While it is impossible to establish a simple relationship with a cohort’s first-marriage intensity, the variations in the second indicator can be read as a summary of the changing risk of marriage for the never-married.

[19] More precisely, at the end of the year following their thirtieth birthday.

[20] The hypothesis is supported by the stability of divorce rates between 5 and 25 years of marriage in the past five years. For longe durations, however, the rates are still rising.

[21] Estimates for 2000 and 2001 are provisional.

[22] To simplify the layout of Table F, we have only noted the central years of the tables, which actually cover the years 1982-84, 1987-89, 1992-94 and 1997-99.

[23] These proportions are calculated with the general population in the denominator, unlike the data reported in Table G, which have the population of “private” households only.

[24] i.e. a couple, with or without children, or a father or mother living with his or her children (one-parent family). To be regarded as a child of the couple, or as living in a single-parent family, the individual must be reported as never-married and without a child of his or her own. Since the 1990 census, there is no age limit for being classified as a child. This change has led to an increase in the number of families classified as one-parent families. No overall comparison can be made with the 1982 census from the published series, as INSEE released tables using the old definition of children of families in 1990, but did not do so in 1999.

[25] Under the census definition, “isolated” persons may be related, including in direct line: a mother and her daughter form a household with two isolated persons if the daughter is widowed or divorced, or a one-parent family if she was never married.

[26] The census housing schedule does not allow an identification of kinship ties with the reference person’s partner.

[27] Including when the one-parent family is not the principal family.

[28] For consistency with all the previous analyses, we use the new definition of “children” of families (without age limit), which yields a higher estimate of the number of single-parent families than in Algava’s article in this issue.

[29] These proportions are slightly higher than those of children registered as born in marriage in the corresponding periods (71.5% and 59.7% respectively). This is logical, as some parents have married after the birth of their children. Moreover, the proportions shown on the chart do not include the 1-2% of children living in a non-family environment.



The flow of migrants into France has been rising since 1996, mainly from countries outside the European Economic Space (EES).
The total fertility rate registered another mild gain in 2001, to 1.9 children per woman. The fertility of women over 30 is still increasing, and that of women under 25 is recovering. Despite these developments, the completed fertility of the birth cohorts of the 1960s is declining, and may settle at slightly over two children per woman.
The introduction of the Civil Solidarity Pact (PACS, Pacte Civil de Solidarité) did not prevent the number of marriages in 2000 and 2001 from exceeding 300,000, the highest figure since 1983. Yet the proportions never married (according to the legal definition) are progressing strongly across cohorts.
In 2001, life expectancy at birth passed the threshold of 83 years for women and 75.5 years for men. At present, male gains are most significant in adulthood, whereas female gains are concentrated at the older ages.
Mean household size continues to fall; the mean number of persons per dwelling in the 1999 census was 2.4. Slightly more than three dwellings in ten are occupied by one person only. A large majority (60%) are women, but their predominance is weakening, as the proportion of men living alone is increasing rapidly at nearly all ages. By contrast, the proportion of men and women living as partners is declining at all ages except the oldest. Children are increasingly less likely to live with married parents, and more likely to live with unmarried parents or in one-parent families.


  1. Population growth and age structure
  2. Immigration
    1. Immigration between 1994 and 1999
    2. From immigration to net migration
  3. Fertility
    1. Recent trends
    2. Fertility of younger women in western Europe
    3. Cohort fertility
    4. Completed fertility and family size
    5. Completed fertility and mean age of childbearing in western Europe
  4. Abortion
  5. Nuptiality
    1. Number of marriages and PACS
    2. First marriages
    3. Cohort-specific marriage frequency
    4. The nuptiality of young cohorts
  6. Divorce
  7. Mortality
    1. Mortality in France and Europe
    2. Fall in mortality by broad age groups and increase in life expectancy
  8. Households and families in the 1999 census
    1. Living alone
    2. Living in a family
    3. An increase in the number of childless couples
    4. More one-parent families
    5. The parental environment of children seen through the censuses

To cite this article

France Prioux "L'évolution démographique récente en France", Population 4/2002 (Vol. 57), p. 687-728.
URL  www.cairn.info/revue-population-2002-4-page-687.htm

Back to Top


Why is this article not available in English?

Cairn International Edition is a service dedicated to helping a non–French–speaking readership to browse, read, and discover work published in French journals. You will find English full–text translations, in addition to French version already available on Cairn regular edition. Full text translations only exist for a selection of articles.

If you are interested in having this article translated into English, please enter your email address and you will receive an email alert when this article has been translated.

Message sent

Your email address has been saved.
We will notify you when this article becomes available in English.