A narrowing gap between birth and death rates

1On 1 January 2014, the population of France was close to 66 million (65.8 million), including 2.1 million in the overseas departments excluding Mayotte, and 0.2 million in Mayotte (Bellamy and Beaumel, 2014).

2Over the course of 2013, the population increased by 300,000, of which 240,000 in metropolitan France (mainland France and Corsica), where the growth rate was 0.42%. [1] This is a decrease with respect to 2012, when it was estimated at 0.49% (Appendix Table A.1). [2] This growth was due mainly to natural increase (3.4 per 1,000), which is the difference between the birth rate of 12.2 per 1,000 and the death rate of 8.8 per 1,000. The gap between these two rates has been progressively narrowing, from 4.6 per 1,000 in 2006 to 3.4 per 1,000 in 2013 (Appendix Table A.1).

The rate of natural increase remains one of the highest in the European Union

3The rate of natural increase in France is among the highest in the European Union (Figure 1). In 2012, just half of the countries in the EU had a positive rate. Ireland topped the rankings thanks to its high fertility and a younger age structure than other countries. In 2010, the median age of the Irish population was 34.3 years, versus 44.2 years in Germany ; the proportion of individuals aged 65 or older in the two countries was 11% and 20.7%, respectively (Mazuy et al., 2013).

Figure 1

Annual rates of natural increase, 2001-2012, European Union (per 1,000)

Annual rates of natural increase, 2001-2012, European Union (per 1,000)

Coverage : European Union (28 countries).

4Rates of natural increase vary from a maximum of 9.5 per 1,000 in Ireland to a minimum of –5.5 per 1,000 in Bulgaria. The rate is between 5.2 and 2.1 per 1,000 in six countries (Cyprus, Luxembourg, France, the United Kingdom, Sweden, and the Netherlands) and between 1.7 and 0.6 per 1,000 in seven countries (Belgium, Malta, Finland, Slovenia, Spain, Denmark, and Slovakia). The other 14 countries have zero or negative rates. Seven countries have a rate between 0 and –1.7 per 1,000 (Poland, Czech Republic, Austria, Estonia, Italy, Greece, and Portugal). Finally, the remaining seven countries have a rate between –2.3 and –5.5 per 1,000 (Croatia, [3] Germany, Romania, Lithuania, Hungary, Latvia, and Bulgaria).

5Net migration rates in certain countries are low or even negative, notably in the countries of central and eastern Europe, but also in Spain, Greece, and Ireland (Figure 2). This is speeding up population decline in certain countries, such as Bulgaria and Latvia, for example, which have negative rates of both natural increase and net migration, and which lost more than 10% of their respective populations between 1980 and 2010 (Avdeev et al., 2011). In situations of falling natural growth rates, the migratory component, when positive, has a stronger relative impact on levels of growth and on the ongoing processes of population ageing (Ambrosetti and Giudici, 2013).

Figure 2

Annual rates of net migration, 2001-2012, European Union (per 1,000)

Annual rates of net migration, 2001-2012, European Union (per 1,000)

Coverage : European Union (28 countries).

6In the countries of southern, central, and eastern Europe, the population under age 20 has fallen drastically, decreasing by more than 30%, for example, between 1980 and 2008 in Bulgaria, Latvia, the Czech Republic, and Romania, and by 35% in Italy over the same period. At the same time, the older population has increased to a varying extent (at the top of the age pyramid), due to mortality conditions that diverge across different European countries. In central and eastern Europe and in the Baltic countries, less favourable mortality conditions have slowed the growth of the older population. In Hungary and the Czech Republic, for example, it increased relatively slowly over this period (10%) due to high mortality.

Ageing of the French population is set to accelerate

7The transformation of the French population pyramid over the last century (1914-2014) reflects the progressive ageing of the population (Figure 3) (Pison, 2014). In 1914 the pyramid was in the shape of a haystack. In 2014 its base is still relatively wide, but the baby-boom generations will strongly accentuate the ageing process in the coming decades. The 1954 population pyramid clearly illustrates the effects of the two World Wars (in particular, the birth deficit) as well as longer-term effects such as the baby boom (the first large cohorts are quite visible at the base of the pyramid). In the near future, the arrival of these large cohorts in the upper age groups (above age 65) and the decrease in the proportion of women of reproductive age in the population will strongly affect crude birth and death rates, as well as the age distribution of the population. In 2014, a quarter of the population is below 20 years old. Another quarter is 60 years old or more, and nearly one person in ten (9.2%) is aged 75 or older (Appendix Table A.2). The over-60s will make up an increasing proportion of the population in the coming decades, and they could account for 30% of the total in 2035 (Blanpain and Chardon, 2010).

Figure 3

Population pyramids in 1914, 1954, and 2014

Population pyramids in 1914, 1954, and 2014

1 Shortfall of births due to the war of 1870
2 Exceptional infant mortality in 1911 due to a summer heat wave
3 Military losses of the 1914-1918 war
4 Shortfall of births due to the 1914-1918 war (depleted cohorts)
5 Depleted cohorts reach childbearing age
6 Shortfall of births due to the war of 1939-1945
7 Start of baby boom
8 End of baby-boom
Coverage : Metropolitan France.

A third of permits are long-term visas equivalent to residence permits

9Flows of foreigners [4] arriving legally in France to establish residence in the country can be estimated from statistics on long-term residence permits and long-term visas (valid for one year or more) valid as residence permits. These statistics only concern countries whose nationals require a residence permit to live in France, so they exclude migrants from within Europe (i.e. from countries listed in Footnote 5). They are based on data from the system used by the French Ministry of the Interior to track the status of foreigners residing in France (AGDREF) and were compiled at INED (Appendix Table A.3).

10To ensure consistency of comparisons over time, the statistics presented below are established for a constant geographical area. They therefore exclude residence permits granted previously to immigrants from countries whose nationals no longer need a residence permit. [5]

11The residence permits considered here have two important characteristics. First, they are valid for a period of more than one year, so all short-term permits are excluded. Second, among the permits of more than one year granted to a given immigrant, only the first is taken into account, to avoid counting the same person more than once. These methodological choices enable us to focus on permanent migration and to count the inflow of foreigners with long-term migrant status. In other words, migrants who were granted two successive seven-month permits and then left the country, for example, are not counted. Moreover, flows are characterized on the basis of the permits themselves : validity start date, period of validity, and the reason for granting the permit. Two of the permit-holder’s characteristics are also taken into account : sex and age at the time the permit was granted. The principal advantage of the AGDREF database is its exhaustive coverage of migrants receiving a long-term permit.

12Additional figures are also published by the Ministry of the Interior, whose statistics include all permits granted (including short-term permits), and by INSEE, which estimates migrants’ actual date of arrival in France and duration of stay. The latter estimates thus correspond more closely to international standards, notably those of Eurostat, which recommends that estimates be based on the actual duration of stay, rather than on the period of validity of the residence permit. To produce its estimates, INSEE uses a census question on year of arrival in France. However, flows estimated from census data are also based on a constant geographical area, and are comparable to those calculated using the AGDREF database (Brutel, 2014 ; Arbel and Costemalle, 2015).

13Table 1 presents the flows, between 2007 and 2012, of migrants who were granted a residence permit with a duration of one year or more for the first time. The number of permits granted to foreigners remains very stable around a mean of approximately 182,000 per year. In the last few years, there has been a slight downward tendency in this figure. Nearly 90% of these first permits are valid for less than 10 years.

Table 1

Number of first permits of one year or more granted to nationals of third countries (constant geographical area) by first year of validity and period of validity

Period of validity 2007 2008 2009 2010 2011 2012 364 to 3649 days 152,684 159,984 167,175 163,629 157,784 159,209 More than 3649 days 24,727 24,345 22,326 20,905 19,957 20,868 Total 177,411 184,329 189,501 184,534 177,741 180,077

Number of first permits of one year or more granted to nationals of third countries (constant geographical area) by first year of validity and period of validity

Coverage : Permits granted in France and abroad to nationals of countries not listed in Footnote 5. Permits granted in year n and recorded in the data extracted in July of the year n + 2, except for 2009, when extraction took place in July 2012.

14Residence permits are issued in France, whereas long-term visas valid as residence permits are issued in French consulates abroad. The AGDREF only takes into account holders of long-term visas who present themselves at the prefecture on arrival in France. This is an obligatory requirement for migrants wishing to live in France for more than one year. Since 2010, long-term visas valid as residence permits have represented more than 36% of all permits granted.

A majority of women and of adults below age 35

15The distribution of permits granted in 2012 by age and sex reveals the concentration of migrants in the 20-35 age group (Figure 4). More women than men were granted residence permits, and their mean age was slightly higher.

Figure 4

Distribution of permits granted in 2012 by age and sex

Distribution of permits granted in 2012 by age and sex

16The age distribution of adult permit holders remained very stable over the period (Table 2). The proportion of minors, who are generally not legally required to apply for a permit, has been decreasing steadily since 2005. In 2012, 17,509 permits were granted to minors. Among adults, the age distribution was highly concentrated in the youngest age group : two thirds of permits were granted to persons aged 18-34.

Table 2

Distribution of holders of a first residence permit of one year or more by age group and first year of validity (%)

Age group 2007 2008 2009 2010 2011 2012 0-17 14.0 11.2 9.8 9.7 9.9 9.7 18-34 63.3 64.2 65.3 65.1 64.5 64.4 35-64 21.1 23.1 23.4 23.7 24.2 24.5 65+ 1.7 1.5 1.5 1.4 1.4 1.5

Distribution of holders of a first residence permit of one year or more by age group and first year of validity (%)

Coverage : Permits granted to foreigners. See Table 1.

17The majority of residence permit holders are women (Table 3). The trend towards an increasing proportion of women among migrants (Beauchemin et al. 2013), which has been clearly marked since 2000, continued over the period covered here.

Table 3

Distribution of adult holders of a first residence permit of one year or more by sex and first year of validity (%)

2007 2008 2009 2010 2011 2012 Men 47.5 49.7 49.0 48.7 48.6 47.8 Women 52.5 50.3 51.0 51.3 51.4 52.2

Distribution of adult holders of a first residence permit of one year or more by sex and first year of validity (%)

Coverage : Permits granted to foreigners. See Table 1.

18African nationals represent a large majority of recipients of a first residence permit, although their proportion has decreased slightly since 2002, and the proportion of immigrants from other continents has increased (Table 4). The principal countries of origin of recipients were Algeria (24,460 permits granted in 2012), Morocco (21,616 permits), Tunisia (11,374 permits), and Turkey (6,626 permits).

Table 4

Distribution of holders of a first residence permit of one year or more by continent of origin and first year of validity (%)

2007 2008 2009 2010 2011 2012 Africa 58.3 58.7 57.7 57.3 56.9 57.0 America 10.8 10.8 10.7 12.6 11.9 11.5 Asia 24.4 24.3 25.4 24.1 24.3 24.5 Europe 5.9 5.6 5.6 5.5 6.3 6.4 Oceania 0.4 0.4 0.4 0.4 0.5 0.4

Distribution of holders of a first residence permit of one year or more by continent of origin and first year of validity (%)

Coverage : Permits granted to foreigners by nationality of origin. Turkey is included in Asia. Europe includes all countries not listed in Footnote 5. The total does not necessarily sum to 100 due to rounding and missing values.

Half of permits are granted for family reasons, a quarter for educational reasons

19The proportion of permits granted for family reasons seems to have strongly declined over the period considered (Table 5), returning to the level of the early 2000s. Analysis of recent changes is difficult, however, given the large increase in permits granted for unspecified reasons. Half of permits were granted for family reasons, and a quarter for educational reasons. In 2012, 9,753 permits were granted for work-related reasons (including 915 for seasonal work) and 17,338 for humanitarian reasons.

Table 5

Distribution of holders of a first residence permit of one year or more by reason for admission and first year of validity (%)

Reason 2007 2008 2009 2010 2011 2012 Work 3.5 3.8 9.3 8.8 7.3 6.0 Family 60.7 58.8 52.2 51.3 51.4 50.7 Education 20.7 23.0 24.4 25.1 25.8 25.2 Humanitarian 8.3 8.3 9.0 9.3 9.3 9.4 Various and unspecified 6.7 6.1 5.1 6.2 9.4 9.2

Distribution of holders of a first residence permit of one year or more by reason for admission and first year of validity (%)

Coverage : permits granted to foreigners, by reason for admission listed in AGDREF.

A downward tendency since 2010

20The estimated number of births in 2013 is 810,000 for the whole of France and 780,000 for the metropolitan départements. There were fewer births in 2013 than in 2012, when 821,000 births were registered, including 790,000 in metropolitan France (Appendix Table A.1). This decrease is due to the combination of a slight decrease in women’s fertility, which fell from 2.01 children per woman in 2012 to 1.99 in 2013 (1.99 in 2012 and 1.97 in 2013 for metropolitan France), and a 0.7% decrease in the number of women of reproductive age over the year 2013 (Bellamy and Beaumel, 2014). The number of births registered in 2012 is close to that of 2005.

21The crude birth rate decreased from 12.4 to 12.2 births per 1,000 inhabitants, a drop of 1.2% between 2012 and 2013. The number of births is gradually decreasing even as the total population continues to increase ; the result is a decrease in the crude birth rate.

Moderate continuation of the trend towards later childbearing

22Only the fertility of women aged 35 or older increased slightly between 2012 and 2013, reaching 338 births per 1,000 women at ages 35-39, and 87 births per 1,000 women aged 40 or above (versus 332 and 84, respectively, in 2012) (Table 6). The fertility rates of all other age groups decreased. The sharpest fall was at ages 20-24, followed by ages 25-29. The shift toward later fertility continues.

Table 6

Fertility by age group since 2008 (per 1,000 women)^*,(a)

Age reached in the year Sum of age-specific rates Absolute variation* 2008 2009 2010 2011 2012 2013 (a) 2008- 2009 2009- 2010 2010- 2011 2011- 2012 2012- 2013 Below 20 36 35 35 34 34 33 -1 -1 -1 0 -1 20-24 276 271 272 262 257 247 -5 1 -10 -5 -10 25-29 643 639 642 634 627 615 -4 +3 -8 -7 -12 30-34 650 653 665 657 660 654 +3 +12 -8 +3 -6 35-39 308 314 322 327 332 338 +5 +8 +5 +6 +6 40+ 76 76 80 83 84 87 0 +4 +3 +1 +3 Total* (TFR) 1,990 1,989 2,016 1,997 1,994 1,974 -1 +27 -19 -3 -20

Fertility by age group since 2008 (per 1,000 women)^*,(a)

* due to rounding, the total may differ slightly from the sum, and the variations may not correspond to apparent differences.
(a) Provisional data.
Coverage : Metropolitan France.

23Looking at changes since 1960, it can be seen that modal values fell sharply from the 1960s to the 1980s, as did fertility rates at later ages (Figure 5). In 1990, the curve of age-specific fertility rates began shifting to the right. This pattern reflects progressive increases in age at childbirth. Since the 1990s, the year-on-year differences in age-specific fertility rates have decreased ; the tendency towards later childbearing has continued, but at a more moderate pace. After decreasing in the 1960s and 1970s, mean age at childbearing (calculated using age-specific rates) began to increase in the late 1970s, and continued to do so over subsequent years. Women had children at a mean age of 27.6 years in 1960, 26.8 years in 1980, 29.4 years in 2000, and 30 years in 2010.

Figure 5

Age-specific fertility rates in 1960, 1970, 1980, 1990, 2000, 2010, and 2013 (per 1,000 women)

Age-specific fertility rates in 1960, 1970, 1980, 1990, 2000, 2010, and 2013 (per 1,000 women)

Coverage : Metropolitan France.

24This recent pattern of change is due mainly to the widespread diffusion of contraception, which has offered women and couples greater control over the timing of childbirth. Age at first childbirth (which has been above 28 since 2010, see Appendix Table A.4), intervals between births, and number of children (notably with the diffusion of the two-child norm) are no longer left to chance.

Fertility remains high

25According to current estimates, fertility levels in France continued to rank among Europe’s highest in 2013. In 2012, France was in third place, behind Ireland and Iceland (Appendix Table A.6). Fertility remained very low in the countries of southern and eastern Europe. Germany, Cyprus, Spain, Greece, Hungary, Poland, Portugal, and Slovakia recorded the lowest fertility in 2012, with total fertility rates below 1.4 children per woman. Estimates of completed fertility are also very low for these countries (between 1.4 and 1.6 children per woman for the cohorts born in the mid-1970s). [6] In France, the mean number of children has stabilized at two children per woman for all the cohorts born in the late 1960s and 1970s (Appendix Table A.5). Mean age at childbearing has increased less in certain European countries than in France. In some countries, however, notably in eastern Europe, fertility remains relatively early. This is the case in Bulgaria, Slovakia, and Lithuania, where the mean age is below 27 (Appendix Table A.7). The countries with the highest mean age at childbearing are Spain, Ireland, and Italy, where it is above 31 years.

Stable rates and concentration of demand at ages 20-24

26According to data based on medical statistics, the number of induced abortions remained stable in 2012 (Vilain et al., 2014). A total of 219,200 abortions [7] were registered in 2012, of which 207,120 in metropolitan France (Appendix Table A.8). The abortion rate is stable, at 14.5 induced abortions per 1,000 women aged 15-49. Abortions have become less frequent among minors and women below age 20, and are now concentrated at ages 20-24. In 2012, abortions performed at ages 20-29 accounted for nearly half of the total abortion rate (48%), those performed at ages 30-39 for 30%, those before age 20 for 15%, and those after age 40 for 7% (according to rates calculated from data in the PMSI medical statistics database). The mean age at abortion was 27.6 years in 2012. [8]

Fewer women seek induced abortions, but repeat abortions are more frequent

27Data from abortion notifications (Box) indicate the number of previous induced abortions and the date of the last abortion. Analysis of the data reveals that repeat abortions have become more frequent since the late 1990s. The stability of abortion rates thus results from two opposing trends : a lower lifetime probability of having an induced abortion, but an increase in the probability, among women who have an abortion, that they will do so more than once. Moreover, frequency of repeat abortion increases with decreasing age at first abortion : 48% of women who have a first abortion before age 26 subsequently have repeat abortions, versus 18% of those who have their first induced abortion after this age (Mazuy et al., 2014).

A renewed downtrend in marriages

28After a slight rise in 2012, the already long-established downtrend in marriages continued in 2013, with the number of marriages reaching a historical low of 231,000. From 1900 until the mid 1960s, the annual number of marriages oscillated around 300,000, with highly marked variations, notably linked to the two World Wars and to their knock-on effects over the short, medium, and long term (Figure 6A). The cohorts born in the 1950s, who reached adulthood in the 1970s, made up a numerically large “reserve” of marriageable men and women. A peak in marriages was recorded in the early 1970s, with more than 410,000 marriages in the year 1972. Up until that year, decreasing age at marriage had reinforced this effect. Beginning in 1972, the number of marriages began to decrease. In the 1990s, the pattern of change was more erratic, but the downtrend in marriages resumed in the early 2000s. In parallel, civil partnerships (pacte civil de solidarité, PACS) were introduced on 15 November 1999 and became increasingly popular. Although the PACS did not completely substitute for marriage, it very probably contributed to the decline in the annual number of marriages.

Figure 6A

Number of marriages (1901 to 2013) and civil partnerships (1999-2013)

Number of marriages (1901 to 2013) and civil partnerships (1999-2013)

29For some couples, civil partnership is seen as an alternative to marriage, while for others it may constitute a “trial marriage” (Rault, 2009). The PACS is thus contributing both to the drop in the number of marriages and to their postponement.

30The slight increase in marriages in 2012 (Figure 6B) could be linked to the 2011 tax reform : since 1 January 2011, married persons (and those in civil partnerships) no longer fill in three income tax declarations (as two singles and then as one couple) in the year of their marriage (or PACS), but instead must choose between one joint declaration or two separate declarations for the full year (i.e. their income tax is now calculated over the entire year and no longer separately for the periods before and after the marriage or PACS). This reform may have had a marked dissuasive effect in 2011, the first year of its application. The 2012 increase should thus be viewed more as a catch-up effect following the drop in marriages in 2011, with the additional marriages in 2012 being postponed weddings initially planned for 2011.

Figure 6B

Number of marriages and civil partnerships (close-up on 1999-2013)

Number of marriages and civil partnerships (close-up on 1999-2013)

Coverage : Metropolitan France.

31Source : INSEE and the Ministry of Justice.

The number of first marriages continues to decrease

32The decrease in first marriages is measured by the sum of rates (total first marriage rate) or the overall probability of first marriage. Between 1972 and 2012, the total first marriage rate fell from 91.7 to 46.6 first marriages per 100 men, and from 94.8 to 47.5 first marriages per 100 women. Probability data show a strong decrease in the proportion of marriages between never-married persons up to age 50 : it fell from 90 first marriages per 100 nevermarried men in 1972 to 53.5 in 2012, and from 93.4 first marriages per 100 never-married women to 56.3 for the same years (Beaumel and Bellamy, 2014). Women’s and men’s probabilities of first marriage follow a similar pattern between 1970 and 2012, in terms of both level and intensity, but men marry two years later on average due to the age gap between spouses (Figure 7). In 40 years, mean age at first marriage has increased by more than eight years for both sexes : never-married women married for the first time at a mean age of 22.5 years on average in 1972 and of 31.1 years in 2012 (according to agespecific probabilities) ; the respective figures for men are 24.6 and 32.8 years. Across cohorts, the delay in first marriages is also pronounced : the mean age was below 23 for women born in the mid-1950s and is above 28 for women born in the 1970s. The respective figures for men are age 25 (for those born in the mid-1950s) and above 30 for those born in the 1970s (Appendix Table A.10).

Figure 7

Age-specific probability of first marriage, women and men, 1972-2012 (per 10,000)

Age-specific probability of first marriage, women and men, 1972-2012 (per 10,000)

Coverage : Metropolitan France.

7,000 same-sex marriages in 2013

33Law no. 2013-404 of 17 May 2013 authorized marriage between two persons of the same sex, and 7,000 same-sex marriages were contracted in that year (Bellamy and Beaumel, 2014). They took place in the second half of 2013, following enactment of the law on 18 May. A peak in such marriages was recorded a little more than 3 months later, in September 2013. Over the same period (second half of 2013), 3,100 same-sex couples entered into a civil partnership (PACS). Thus, around 20,000 men and women officialized their same-sex union in the second half of 2013. Married same-sex couples are older, on average, than heterosexual couples : age 50 years for men and 43 for women, versus 37 for men and 34 for women in heterosexual marriages. The age gap between partners is also larger : 5.5 years for women and 8 years for men (Bellamy and Beaumel, 2014). This age gap is also observed for civil partnerships : in 2009-2010 it was 5 years for women and 7 years for men (Mazuy et al., 2011).

168,000 PACS registered in 2013, including 13% by notaries

34Since it came into effect on 15 November 1999, the PACS has been widely adopted throughout French society, with a peak in 2010, when 205,500 PACS were registered. The number of PACS fell to 152,000 in 2011, the year of the tax reform, but increased again in 2012 and 2013 (Table 7, Figure 6B). In 2013, 168,000 PACS were registered, representing more than 40% of officialized unions (PACS and marriages combined). The PACS has become widespread among heterosexual couples, who account for 96% of all PACS unions.

Table 7

Number of PACS since 2009^*

2009 2010 2011 2012 2013* PACS registered (total) 174,584 205,561 152,176 160,732 168,223 including in overseas départements 1,404 1,602 1,376 1,537 1,656 Number of PACS by partners’ sex Man-man 4,894 5,208 4,156 3,750 3,333 Woman-woman 3,542 3,938 3,338 3,223 2,724 Man-woman 166,148 196,416 144,682 153,759 162,166

Number of PACS since 2009^*

* Provisional
Coverage : Whole of France, excluding Mayotte.

35In 2013, 13% of PACS were registered by a notary. According to a recent study by the Ministry of Justice, couples who register a PACS through a notary are older than those who do so directly at the magistrates’ court (Büsch and Timbart, 2014). These couples are perhaps more concerned to ensure mutual protection in case of the death of one of the partners than those who register a PACS at a magistrates’ court (the notary can also draw up a will for the couple [9]).

Divorces and PACS dissolutions

36Over the year 2013, 125,109 divorces were pronounced (Table 8), including 121,849 in metropolitan France (Appendix Table A.9). This is slightly fewer than in 2012, when 128,371 divorces were pronounced. More than half of divorces (direct divorces) are pronounced by mutual consent (53.5%). The total divorce rate reached 45 divorces per 100 marriages in 2012, a slight decrease with respect to 2011 (Appendix Table A.9).

Table 8

Number of divorces since 2010

Year 2010 2011 2012 2013 Divorce by mutual consent (A) 72,433 75,542 69,431 66,640 Uncontested divorce (B) 32,603 30,712 31,212 31,199 Divorce by breakdown of conjugal life (C) 14,107 14,045 15,569 16,038 Fault-based divorce (D) 13,117 11,065 10,685 9,835 Direct divorce, non-specified (E) 468 642 750 740 Direct divorces (3) (A+B+C+D+E) (1) 132,728 132,006 127,647 124,452 Conversion of legal separation to divorce (2) 993 779 724 657 Legal separation (3) 1,559 1,316 1,283 1,347 Total divorces and conversions (1+2) 133,721 132,785 128,371 125,109 Total union dissolutions (1+2+3) 135,280 134,101 129,654 126,456

Number of divorces since 2010

Coverage : Whole of France.

37From a longitudinal point of view, on the basis of probabilities by marriage duration, divorce has increased with successive cohorts. Between the 1970 and 2000 cohorts, the overall probability has doubled for a marriage duration of 10 years, and at almost all ages the overall probability has increased across successive cohorts (Figure 8).

Figure 8

Divorce by marriage duration and cohort

Divorce by marriage duration and cohort

Coverage : Metropolitan France.

38The number of PACS dissolutions has been increasing, and surpassed 53,000 in 2013 (Table 9). In nearly four out of ten cases (38%), the PACS is dissolved because the couple gets married. When PACS are dissolved due to separation, it is almost exclusively by mutual agreement.

Table 9

PACS dissolutions since 2010^*

Year 2010 2011 2012 2013* Number of dissolutions 35,627 42,290 48,841 53,655 Reason for PACS dissolution Mutual consent 20,817 24,117 27,745 31,167 Requested by one partner 1,153 1,295 1,473 1,613 Marriage 13,263 16,450 19,142 20,415 Death 366 417 451 428 Other cases and not recorded 28 11 30 32

PACS dissolutions since 2010^*

* Provisional.
Coverage : Whole of France.

A renewed increase in life expectancy

40Recent changes are consistent with the trends observed over the last 20 years. Since 2003, men’s mean length of life has increased by 2.8 years and that of women by 2 years, an annual increase of 0.28 and 0.20 years, respectively (Appendix Table A.11). Improvements were slightly more rapid over the past decade than over the preceding one. Between 1993 and 2003, the mean increase was 0.26 years for males and 0.16 years for females. As men’s life expectancy has increased faster than women’s over these two decades, the sex difference in mean length of life, which reached a peak of 8.2 years at the beginning of the 1990s, has since been steadily decreasing.

41In 2012, although female life expectancy at birth in France (84.8 years) was below that of Switzerland and Spain (84.9 years and 85.5 years, respectively), France remained close to the top of the European ranking (Appendix Table A.12). For males, it was in a less favourable twelfth position (out of a total of 29 countries). Contrary to the situation two decades ago, excess male mortality in France, which at the time was among the highest in western Europe, is now in the lower middle of the distribution, with a difference of 6.3 years between the sexes. Mean sex differences in life expectancy are much higher in some other countries – as high as 10 to 11 years in Sweden, Switzerland, and Iceland. For infant mortality, France stands exactly in the middle of the ranking, at 3.5 per 1,000 in 2012 (Appendix Table A.13).

Rising life expectancy thanks to lower mortality above age 60

42The steady increase in mean length of life conceals unequal improvements over time by age and cause of death. Figure 10 shows, for both sexes, how age-specific probabilities of dying changed between the three-year life table of 1990-1992 and that of 2010-2012. The general trend is identical for males and females. The greatest improvements are seen for young people, with probabilities of dying halved up to the ages of 30-35 years. Improvements taper off gradually with increasing age, with a particularly small decrease around the ages of 50-55. There are nevertheless notable differences between the sexes. Male mortality decreased more than female mortality between the ages of 25 and 75, but a little less above age 75. The most marked difference between the sexes over this period was in the 45-50 age group, where, again, a larger mortality decrease was observed for men than for women. As we will see below, an examination of causes of death sheds light on these differences.

Figure 10

Ratios between the three-year life table of 2010-2012 and that of 1990-1992 for age-specific probabilities of dying (smoothed over 3 years of age), by sex

Ratios between the three-year life table of 2010-2012 and that of 1990-1992 for age-specific probabilities of dying (smoothed over 3 years of age), by sex

43Decreases in mortality at different ages have unequal effects on changes over time in life expectancy at birth. Child mortality has reached such low levels that any changes now have little effect on life expectancy at birth : between 1990-1992 and 2010-2012, a decrease of 6.7 deaths before age 15 per 1,000 newborns in 20 years only increased mean length of life by 0.5 years for males and by 0.4 years for females (Figure 11). In contrast, the decrease in mortality at ages 15-45 over the same period played an important role, for males at least, contributing 20% of the total gain in life expectancy at birth (one out of the total of 5.3 years of life expectancy gained between 1990-1992 and 2010-2012, versus only 0.4 out of 3.7 years, only slightly above 10%, for females). However, the greatest gains are due to improvements at ages 45-80 in men (+3 years) and beyond age 65 in women (+2.6 years).

Figure 11

Contributions by age group to gains in male and female life expectancy at birth between 1990-1992 and 2010-2012 (5.3 and 3.7 years in total)

Contributions by age group to gains in male and female life expectancy at birth between 1990-1992 and 2010-2012 (5.3 and 3.7 years in total)

Massive decrease in women’s cardiovascular diseases

44Contributions by cause of death to improvements in life expectancy at birth over the period from 1989-1991 to 2009-2011 [10] vary greatly by age and sex (Table 10). Note, however, that while decreasing male mortality can be attributed to decreases in various pathologies (notably external causes in young adults, cancers around ages 45-75, and diseases of the circulatory system at age 60 and beyond), improvement in females is overwhelmingly due to decreases in cardiovascular diseases (both heart diseases and cerebrovascular diseases) (Appendix Table A.14).

Table 10

Contribution of age groups and causes of death to gains in life expectancy at birth (in years) between 1989-1991 and 2009-2011, by sex

Age group Cause of death Infectious diseases Cancers Cardiovascular diseases Respiratory diseases Digestive diseases Other diseases External causes All causes Males 0-14 0.01 0.02 0.01 0.02 0.01 0.31 0.13 0.51 15-24 0.01 0.01 0.01 0.01 0.00 0.01 0.29 0.34 25-44 0.18 0.13 0.09 0.01 0.05 0.02 0.32 0.80 45-64 0.04 0.59 0.39 0.06 0.11 -0.03 0.14 1.30 65-79 0.01 0.46 0.82 0.15 0.10 0.02 0.10 1.66 80 + 0.00 0.10 0.48 0.14 0.04 -0.04 0.05 0.77 Total gain 0.25 1.31 1.81 0.38 0.32 0.28 1.03 5.38 Females 0-14 0.01 0.02 0.01 0.02 0.01 0.23 0.09 0.38 15-24 0.00 0.01 0.01 0.01 0.00 0.01 0.09 0.12 25-44 0.03 0.06 0.03 0.01 0.04 0.01 0.14 0.31 45-64 0.00 0.11 0.14 0.01 0.08 -0.02 0.08 0.40 65-79 0.01 0.17 0.74 0.07 0.07 0.04 0.09 1.20 80 + 0.00 0.07 1.01 0.17 0.09 -0.09 0.12 1.37 Total gain 0.06 0.43 1.95 0.29 0.28 0.18 0.61 3.78

Contribution of age groups and causes of death to gains in life expectancy at birth (in years) between 1989-1991 and 2009-2011, by sex

Note : The method used to calculate the contribution of each age group and cause to gains in life expectancy at birth between two periods is the one proposed by Andreev, Shkolnikov and Begun (2002).

Infant mortality : a growing contribution of diseases of early infancy

45Following a long-term trend, infant mortality from infectious diseases (above all respiratory diseases) continued to decrease steadily between 1990 and 2010 [11] (see also Meslé, 1995). Mortality due to external causes (mainly accidents, at this age) also decreased steadily up until 2008, but seems to have stabilized since, particularly for girls.

46The weight of sudden infant death syndrome in infant mortality statistics grew considerably from the mid-1970s and peaked in the late 1980s, before declining very sharply for a few years, and more slowly in subsequent years, thanks to campaigns advising parents not to lay infants on their stomachs (Barbieri, 1998). The mortality rate in the first year of life from this cause decreased from nearly 220 deaths per 100,000 births around 1990 to only 27 per 100,000 in 2010 for the two sexes combined – a spectacular drop of nearly 90% in twenty years.

47The decline in mortality due to diseases of early infancy, in contrast, was very limited over this period, with a relatively small drop in deaths due to congenital abnormalities and an absence of progress for other causes associated with the perinatal period. Consequently, the contribution of these two groups of causes has increased very rapidly over the last twenty years : while they accounted for 50% of infant mortality in 1990, they now represent more than 75%. If the recent stagnation of mortality from these diseases specific to early infancy persists, it may slow, or even stop, the long-standing trend of declining infant mortality.

Mortality at ages 15-24 : deaths from external causes predominate

48For males, deaths from external causes account for most deaths at these ages : in 2010, more than three quarters of the standardized mortality rate in this age range could be attributed to this cause (versus roughly half for women). Road accidents and suicide are the principal causes for both sexes (Appendix Table A.15). Mortality due to traffic accidents decreased between 1990 and 2010, and particularly after 1999, but trends in mortality by suicide and other external causes have been less favourable. Since 2005, mortality due to these causes seems to have stabilized, and suicide has even recently risen slightly among women.

49Change in mortality for the other major categories of causes has not been much more favourable, which explains this age group’s particularly small contribution to progress in life expectancy between 1990 and 2010 (0.34 years out of a total gain of 5.4 years in men, 0.12 years out of 3.8 years in women). Mortality from infectious diseases is a noteworthy exception : it fell sharply until 1998-2000 thanks to better control of the HIV-AIDS epidemic from the early 1990s.

Mortality at ages 25-44 : contrasting trends in the two sexes

50Certain characteristics of the preceding age group are also found among persons aged 25-44. In this age group, the pattern of deaths by infectious disease due to the evolution of HIV-AIDS mortality was just as remarkable as in the younger age group, with both a strong increase up until 1995 and a sharp drop after that year. Whereas at the peak of the epidemic infections were the third-leading cause of mortality after deaths from external causes and cancers, in 2010 they were the least frequent cause.

51As in younger people, mortality due to external causes, which was the leading cause of death before 1990, has been steadily declining ever since. The decrease since that time (and especially since 2000) has been very rapid for traffic accidents, with a 57% drop in the standardized rate for men and a 72% drop for women between 1990 and 2000. For other external causes in men, however, this downtrend was interrupted from 2003-2004 onward. External causes are still responsible for the majority of male deaths at these ages, representing 51% of the total standardized rate, versus 26% for women.

52In women of these ages, cancer was the leading cause of death in 2010 (Appendix Table A.15). Twenty years earlier, the standardized mortality rate for cancer was one third higher in men (40 per 100,000 versus 30 per 100,000 in women). Due to sustained decreases up until the second half of the 2000s in men, the standardized rate is currently very similar for both sexes, at around 22 per 100,000. This situation results in part from contrasting trends in lung cancer : men’s standardized mortality rate from this cause decreased by half in 20 years, whereas it increased in women until around 2005. It also results from rapid falls in cancers of the upper aerodigestive tract (lips, mouth, and pharynx) and the œsophagus thanks to lower alcohol consumption, which have primarily benefited men as the incidence of these pathologies has always been much lower in women. In contrast, breast cancer, the leading cancer in women in terms of mortality, has decreased very slowly.

Mortality at ages 45-64 : cancer strongly predominates

53More than at all other ages, cancers constitute the leading cause of mortality at ages 45-64 : in 2010, they represented, respectively, 47% and 56% of male and female standardized mortality rates in this age range, with highly marked contrasts between the sexes by site.

54The most lethal cancers in men (in order of frequency) are those of the lung ; the upper aerodigestive tract ; the intestine, the colon, and the rectum ; the oesophagus ; the blood-forming organs ; the stomach ; and the prostate. In women they are : the breast ; the lung ; the intestine, colon, and rectum ; the uterus ; the blood-forming organs ; and, to a lesser extent, the upper aerodigestive tract, the stomach, and the oesophagus (Figure 12). There is a very marked contrast between the sexes : for men, deaths from almost all types of cancer decreased over the observed period, while for women, apart from slight decreases in deaths from cancers of the breast and the blood-forming organs, they remained stable or increased. The increase in lung cancer was very pronounced in women, with a standardized rate that more than doubled between 1990 and 2010 ; this cancer seems to be on the verge of becoming the leading cause of cancer death at ages 45-64. These different trends in men and women are largely explained by differences in smoking behaviours (decreasing consumption among men since the 1970s, regular increases among women, particularly at ages 45-64). [12]

Figure 12

Standardized mortality rates at ages 45-64 for the most lethal cancers by sex, from 1989-1991 to 2009-2011 (three-calendar-year moving average)

Standardized mortality rates at ages 45-64 for the most lethal cancers by sex, from 1989-1991 to 2009-2011 (three-calendar-year moving average)

UADT : Upper aerodigestive tract

55Cardiovascular diseases are the second most frequent cause of death at these ages, far behind cancers, with a standardized rate that continues to decline, due to factors including changes in behaviour (notably smoking), decreases in infectious diseases which are risk factors for heart diseases, and medical advances, which have led to considerable improvements in the prevention and treatment of diseases of the circulatory system (Appendix Table A.15).

56Mortality from other causes has evolved less favourably, as the downtrend has halted for almost all categories since the early 2000s. However, as these causes account for a much smaller proportion of general mortality, this age group contributed positively to gains in life expectancy between 1990 and 2010, particularly for men (Table 10).

Mortality at ages 65-79 : a large drop due to a decline in cardiovascular diseases

57Mortality decreased considerably over the period in this age group, which accounts for nearly a third of the years of life gained between 1990 and 2010. These improvements are predominantly due to control of cardiovascular diseases (50% of total gain in men and 62% in women) and notably ischaemic heart diseases, thanks to improvements in prevention, treatment, and surgery, associated with decreasing tobacco consumption in men.

58Decreasing cancer mortality also played an important role in men : it was responsible for a quarter of years of life gained by men, versus only a sixth in women. Whereas the standardized rates for these two groups of causes of death were similar until the end of the 1980s, mortality from cardiovascular diseases has decreased by 50% in twenty years, versus 15% and 25% for male and female cancer mortality.

59The ranking of cancers by site in this age group is slightly different than in the preceding one. In men, lung cancer is the leading cause of death, but it is followed by cancers of the colon and rectum, the prostate, the blood-forming organs, and, at a much lower level, the œsophagus, the stomach, and the upper aerodigestive tract. The decreasing trend in mortality for all cancers in men beginning in 1990 seems to have been interrupted in recent years, with two exceptions : blood cancer (cancer of the blood-forming organs), mortality from which has stabilized since the early 1980s, and prostate cancer, which has continued to decrease, with a standardized rate in 2011 at half its 1990 level. For women, the ranking is very similar to the one for ages 45-64 : cancer of the breast, followed by the lung, the colon and rectum, the blood-forming organs, and, further behind, the stomach, the œsophagus, and the upper aerodigestive tract. Mortality from all cancers combined has levelled off since the early 2000s, except for lung cancer, which has been in constant increase due to the high prevalence of smoking in these generations, and stomach cancer, which continues to decrease.

Mortality at ages 80 and above : cardiovascular diseases predominate

60This age group is the only one in which cardiovascular diseases are still the leading cause of death. Decreasing cardiovascular mortality explains most of the decrease in the all-causes rate, accounting for 74% of female life expectancy gains at age 80 and above between 1990 and 2010, and 62% of male gains. This pattern explains the gradual convergence with cancer mortality, the second cause of death at this age, which has decreased very slowly in men and virtually stagnated in women due to increasing lung cancer mortality.

61Mortality from respiratory diseases has decreased considerably since the late 1990s, thanks to the spread of influenza vaccination, which has had a beneficial effect not only on influenza but also on asthma and other chronic respiratory illnesses. The standardized mortality rate from influenza has fallen from 50 to less than 5 per 100,000 in just 15 years. The mortality rate from asthma, which was only slightly below that of influenza in 1990, has also dropped markedly, decreasing fourfold in 20 years. Respiratory illnesses still ranked third among specific causes of mortality in 2010, however, as they did in 1990, but with a lower standardized rate than the residual category of “other diseases”.

62Mortality from senile dementias, which has progressively increased over the last 20 years, represents a growing proportion of these “other diseases”, accounting for 20% of male deaths and 27% of female deaths from “other diseases” in 1990, but 40% and 50%, respectively, in 2010. The corresponding mortality rate has increased from around 350 to 800 per 100,000 in men and from a little over 400 to 900 per 100,000 in women over the period. Alzheimer’s disease, which is responsible for 60% of deaths attributable to senile dementias, has been rapidly increasing since it was first included in the 9^th edition of the International Classification of Diseases, published in 1980. It is difficult, however, to distinguish between the effects of diagnostic improvements – which result from changes in medical and certification practices – and the actual growth in this pathology among the elderly population.

Life expectancy in good health

63A new question now arises. Do the years of life gained in the last 20 years reflect further years in a state of disability or dependence, or a continuous increase in life expectancy in good health ? This is a question of particular importance in the contemporary context of population ageing. In recent years there has been a growing literature on the subject, and an ever wider range of data has been collected with a view to establishing more diverse indicators to capture the phenomenon of disability.

64The concept of disability-free life expectancy, without activity limitation or chronic morbidity, was developed in the 1980s (Robine et al., 1986). This indicator is calculated by combining mortality data from vital records and health data from general population surveys. In France, disability-free life expectancy at age 50 increased at the same rate as life expectancy in the 1990s, but more slowly in the following decades. The proportion of years of severe disability, and notably of dependence, has decreased for the cohorts born before the Second World War. There is less clear improvement for the subsequent cohorts, particularly women, for whom the years lived with certain types of severe disability appear to have increased (Cambois et al., 2012). However, it is difficult to determine whether this results from higher survival rates among persons with chronic diseases or disabilities, improved reporting of health problems in surveys, or increased disability in these cohorts in comparison to previous ones. The post-Second World War cohorts also have a higher prevalence of risk behaviours (such as smoking). In addition, the strain of working life may have been greater for women born during the baby-boom years, who have a dual workload in both the workplace and the home (Cambois and Robine, 2012).