1 – Europe’s highest rate of natural growth

8On 1 January 2019, the population of France was almost 67 million (66,992,699), including 2.18 million in the overseas departments and regions (Papon and Beaumel, 2019). In 2018, the population of France increased slowly, by 3.1 per 1,000 overall and by just 2.8 per 1,000 in metropolitan France (mainland France and Corsica) (Appendix Table A.1). [1] Natural growth still accounts for a large share of population increase, although the difference between numbers of births and deaths fell again in 2018 and is now below 150,000. This historic low is the result of a decrease in births and an increase in deaths that stem from age-structure effects (population ageing and fewer women of childbearing age) and a change in demographic behaviour (notably fertility decline).

9France is the second most populated country of the EU, behind Germany (83 million) and ahead of the United Kingdom (since 1986 only) (Table 1). On 1 January 2019, France accounted for 13.1% of the population of the 28 European Union countries and 15% if we exclude the United Kingdom. The four most populated countries (Germany, France, United Kingdom, and Italy) are home to 54% of the EU-28 population in 2019, and the top seven account for three-quarters of the total. France has Europe’s highest level of natural growth in absolute terms and ranks among European leaders in relative terms, after Ireland, Cyprus, Luxembourg, and Sweden (Table 1). France is an outlier in this respect, as natural growth in 2018 was negative in most EU-28 countries, notably in Eastern Europe and the Baltic countries (Fihel and Okólski, 2019).

10The contribution of migration to French population growth is difficult to interpret. Since 2015, the National Institute for Statistics and Economic Studies (Institut national de la statistique et des études économiques [INSEE]) has had to adjust the net migration figures each year. In 2018, estimated net migration was +58,000, but this figure was corrected by −100,000 to achieve consistency with the population estimates for 1 January based on census data (Papon and Beaumel, 2019). This makes European comparisons difficult. According to the indicators published by Eurostat, French adjusted net migration is negative, and France is the European country with the second lowest absolute level of net migration (Table 1). [2]

Table 1

Population size, natural growth, and net migration, EU-28 countries

Population on 1 January 2019 Natural increase Net migration Number (millions) Growth 2018–2019 (per 1,000) Total (thousands) Rate (per 1,000) Total (thousands) Rate (per 1,000) Germany 83.0 2.7 –167.4 –2.0 394.2 4.8 France 67.0 1.5 144.5 2.2 –42.6* –0.6 United Kingdom 66.6 5.6 115.2 1.7 258.3 3.9 Italy 60.4 –2.1 –193.4 –3.2 69.0 1.1 Spain 46.9 5.9 –56.3 –1.2 332.4 7.1 Poland 38.0 –0.1 –26.0 –0.7 22.1 0.6 Romania 19.4 –6.6 –75.3 –3.9 –53.7 –2.8 Netherlands 17.3 5.9 14.7 0.9 86.4 5.0 Belgium 11.5 6.1 7.6 0.7 61.7 5.4 Greece 10.7 –1.8 –33.9 –3.2 15.0 1.4 Czech Republic 10.6 3.7 1.1 0.1 38.6 3.6 Portugal 10.3 –1.4 –26.0 –2.5 11.6 1.1 Sweden 10.2 10.8 23.6 2.3 86.3 8.5 Hungary 9.8 –0.6 –37.8 –3.9 32.2 3.3 Austria 8.9 4.1 1.6 0.2 34.9 4.0 Bulgaria 7.0 –7.1 –46.3 –6.6 –3.7 –0.5 Denmark 5.8 4.3 6.2 1.1 18.6 3.2 Finland 5.5 0.9 –7.0 –1.3 11.7 2.1 Slovakia 5.5 1.3 3.3 0.6 4.0 0.7 Ireland 4.9 15.2 29.9 6.1 44.0 9.0 Croatia 4.1 –7.1 –15.8 –3.9 –13.5 –3.3 Lithuania 2.8 –5.3 –11.4 –4.1 –3.3 –1.2 Slovenia 2.1 6.8 –0.9 –0.4 14.9 7.2 Latvia 1.9 –7.5 –9.5 –4.9 –4.9 –2.5 Estonia 1.3 4.3 –1.4 –1.0 7.1 5.3 Cyprus 0.9 13.4 3.6 4.1 8.1 9.3 Luxembourg 0.6 19.6 2.0 3.2 9.9 16.3 Malta 0.5 36.8 0.8 1.6 17.1 35.3 EU-28 513.5 2.1 –354.2 –0.7 1,456.7 2.8

Population size, natural growth, and net migration, EU-28 countries

* Net migration published by Eurostat, which does not take into account INSEE’s correction (+100,000).
Interpretation: The seven countries with the highest figures in each column are in bold, and the seven with the lowest are in italics. The countries are listed in decreasing order of population size in mid-2018.
Coverage: EU-28.

2 – The base of the population pyramid continues to narrow

11For the last 8 years, the number of births has been decreasing. In metropolitan France, 82,000 fewer children were born in 2018 than in 2010 (−10%), and the base of the population pyramid has narrowed as a result. This decrease is due to lower fertility and, above all, to an age-structure effect (fewer potential mothers). The narrowing base of the pyramid corresponds to the drop in numbers of women born between 1986 and 1994 (points 6 and 7 in Figure 1). Births in 2018 for the whole of France (760,000) were almost equivalent to that in 1993 (741,000), the year with the lowest recorded number since 1946 (Festy, 1995). [3]

Figure 1

Population pyramid of France on 1 January 2019 (numbers)

Population pyramid of France on 1 January 2019 (numbers)

Coverage: Whole of France (including Mayotte).

12While the base of the French population pyramid is much wider than that of EU-28 (Figure 2), the difference is much smaller than it was in the early 2010s (Prioux and Barbieri, 2012). The shapes of the French and European pyramids differ because the French share of the European population varies by age (Figure 3). More than 1 in 6 Europeans under age 18 live in France (15%), but only 1 in 10 of those aged 75–79, born during the Second World War. German and Italian populations, on the one hand, and that of the French and UK populations, on the other, present differences in the timing and speed of fertility decline (Sauvy and Ledermann, 1946; Pison, 2012). The contribution of Italy, with the fourth largest EU-28 population, has been declining steadily; the Italian cohorts aged 0–5 now account for less than 10% of the European population at these ages.

Figure 2

Population pyramid of EU-28 and France on 1 January 2018 (per 100 inhabitants)

Population pyramid of EU-28 and France on 1 January 2018 (per 100 inhabitants)

Coverage: Whole of France (including Mayotte) and EU-28.

Figure 3

Share of the inhabitants of Europe’s four most populated countries in the EU-28 population by age in 2018

Share of the inhabitants of Europe’s four most populated countries in the EU-28 population by age in 2018

Coverage: Whole of France (including Mayotte) and other countries.

3 – Fewer men than women above age 25

13With 93.6 males per 100 females, the sex ratio (proportion of males to females) in France is among the lowest in Europe, well below the European average (95.7) but above the very low levels observed in the Baltic countries (Table 2). These countries’ populations are declining rapidly due to mass emigration (Breton et al., 2018; Fihel and Okólski, 2019), of men especially. This might explain the positive correlation between the sex ratio in 2018 and the population increase between 1998 and 2018 (r² = 0.63 / p < 0.01).

Table 2

Structural characteristics of the EU-28 populations in 2008 and 2018

Mid-year population (millions) Median age Proportion aged 65+ Sex ratio Country Increase 2008 2018 (%) 2008 2018 2008 2018 2008 2018 Germany 82.1 82.9 1.0 43.2 46.0 20.1 21.4 96.1 97.4 France 64.2 67.0 4.4 39.3 41.6 16.4 19.7 93.8 93.6 United Kingdom 61.8 66.5 7.5 39.1 40.1 15.9 18.2 96.1 97.5 Italy 58.8 60.4 2.7 42.7 46.3 20.2 22.6 94.0 94.8 Spain 46.0 46.8 1.8 39.2 43.6 16.4 19.2 97.8 96.2 Poland 38.1 38.0 –0.4 37.3 40.6 13.5 17.1 93.5 93.8 Romania 20.5 19.5 –5.2 38.7 42.1 15.4 18.2 95.0 95.8 Netherlands 16.4 17.2 4.8 40.0 42.6 14.7 18.9 97.8 98.5 Greece 11.1 10.7 –3.1 40.3 44.6 18.7 21.8 97.1 94.2 Belgium 10.7 11.4 6.8 40.7 41.6 17.1 18.7 96.0 97.1 Portugal 10.6 10.3 –2.6 40.4 44.8 17.7 21.5 92.4 89.8 Czech Republic 10.4 10.6 2.4 39.3 42.3 14.6 19.2 96.0 96.8 Hungary 10.0 9.8 –2.6 39.4 42.6 16.2 18.9 90.4 91.5 Sweden 9.2 10.2 10.4 40.6 40.6 17.5 19.8 98.8 100.9 Austria 8.3 8.8 6.2 40.9 43.2 17.1 18.7 94.8 96.8 Bulgaria 7.5 7.0 –6.2 41.9 44.1 17.8 21.0 94.9 94.3 Denmark 5.5 5.8 5.5 40.2 41.8 15.6 19.3 98.1 99.0 Slovakia 5.4 5.4 1.3 36.3 40.2 12.1 15.5 94.6 95.3 Finland 5.3 5.5 3.8 41.5 42.7 16.5 21.4 96.1 97.3 Ireland 4.5 4.9 8.4 33.4 37.3 10.8 13.8 99.8 98.1 Croatia 4.3 4.1 –5.1 41.5 43.7 17.8 20.1 92.9 93.3 Lithuania 3.2 2.8 –12.4 39.6 43.9 17.0 19.6 86.2 85.8 Latvia 2.2 1.9 –11.5 39.9 43.3 17.6 20.1 85.0 85.2 Slovenia 2.0 2.1 2.6 41.0 43.8 16.3 19.4 96.3 98.8 Estonia 1.3 1.3 –1.1 39.8 42.0 17.5 19.6 86.8 89.0 Cyprus 0.8 0.9 10.6 35.4 37.5 12.4 15.9 96.1 95.2 Luxembourg 0.5 0.6 24.4 38.6 39.4 14.0 14.3 98.1 101.0 Malta 0.4 0.5 18.4 39.3 40.4 13.9 18.8 98.8 102.4 EU-28 501.2 512.9 4.1% 40.4 43.1 17.1 19.7 95.2 95.7

Structural characteristics of the EU-28 populations in 2008 and 2018

Interpretation: The seven countries with the highest figures in each column are in bold, and the seven with the lowest are in italics. The countries are listed in decreasing order of population size in mid-2018.
Coverage: EU-28.

14In both France and Europe, the sex ratio varies by age (Figure 4). It is close to 105 in infancy because more boys than girls are born, reaches equilibrium at around age 45, and then falls steadily to a level well below 100 beyond age 70 due to excess male mortality at advanced ages. The imbalance at intermediate ages (more women at ages 25–39) observed in France (but not in Europe as a whole) is more difficult to interpret. Portugal and Ireland are the only other countries that follow this pattern. What is the reason for the ‘female surplus’ at these ages? In demographic terms, it might be explained by excess male mortality and/or sex-selective migration (Brutel, 2014; Wisser and Vaupel, 2014; Góis and Marques, 2018), which are specific to certain countries. Another explanation might be that more young men are omitted from the censuses, notably during periods of residential instability after leaving the parental home or separating from a partner (Toulemon, 2017). This phenomenon has been observed in several European countries (Smallwood and De Broe, 2009).

Figure 4

Sex ratios by age in France and in EU-28 on 1 January 2018

Sex ratios by age in France and in EU-28 on 1 January 2018

Coverage: Whole of France (including Mayotte) and EU-28.

4 – One in five people aged 65 or above on 1 January 2019

15The French population is continuing to age; one-quarter of the population was aged over 60 on 1 January 2019. The proportion of over-60s is now 2.0 percentage points higher than that of the under-20s (2.6 points in metropolitan France) (Appendix Table A.2). Nonetheless, France is still a relatively young country (9th youngest in Europe), with a median age of 41.6 years on 1 January 2018, up from 39.3 years 10 years earlier (same European ranking) (Table 2). Measured in terms of median age, the population is ageing much faster in countries such as Spain (which has fallen from 8th to 20th place), Greece (18th to 20th place), and Lithuania (13th to 23rd place). The two countries at the extremes in 2018 are Ireland (37.3 years) and Italy (46.3 years), now the oldest country in Europe, ahead of Germany (46.0 years).

1 – An upward trend in inflows

18Table 3 shows inflow data for the years 2012 to 2017. In 2017, 237,742 people received a residence permit, the highest number since 2000 (Appendix Table A.3). Inflows in 2016 were 9% higher than in 2016 and 32% higher than in 2012.

Table 3

Inflows of third-country nationals by first year of validity and period of validity of first residence permit of 1 year or more

Period of permit validity 2012 2013 2014 2015 2016 2017 Less than 10 years 159,077 173,058 178,677 187,626 193,163 208,773 10 years or more 20,934 19,338 21,210 22,414 25,191 28,969 Total 180,011 192,396 199,887 210,040 218,354 237,742

Inflows of third-country nationals by first year of validity and period of validity of first residence permit of 1 year or more

Coverage: Permits granted in France and abroad to foreign nationals excluding citizens of the European Economic Area and Switzerland (constant geographical area from 2012 to 2017). Permits granted in year n and recorded in the data extracted in July of the year n + 2. Permits of less than 10 years are valid for between 364 and 3,649 days; permits of 10 years or more are valid for more than 3,649 days.

19Among the individuals counted, the share of immigrants receiving a residence permit valid for 10 years or more remains low, at slightly above 12% in 2017. These long-term residence permits, typically a resident’s card, are generally granted after one or more permits of less than 10 years.

20Inflows of foreigners can also be estimated using other statistical sources. INSEE uses population censuses and notably a question on the previous place of residence. According to Eurostat, which disseminates INSEE data, 241,991 foreign nationals entered France in 2017. [8] This total also includes nationalities not required to hold a residence permit (i.e. EEA countries and Switzerland) and, potentially, undocumented third-country immigrants. Applying the same geographical scope as that of Table 3 brings the total down to 163,235, a figure well below that obtained from the AGDREF database. The Eurostat underestimation is difficult to explain because no information is available on the method used to construct the series. Students may have been excluded, even if they stay in France for more than a year.

2 – Women no longer account for the majority of incoming foreigners

21Recent immigrants are young. People aged 18–34 accounted for 64.5% of all arrivals and 71.9% of adult arrivals in 2017 (Table 4). The share of minors is stable at 10.3%. Only minors receiving a residence permit are counted. [9] Foreign minors do not have to hold a residence permit but may need to obtain one if, for example, they wish to travel outside France. Minors born in France to foreign parents are not counted in the inflows. The first row of Table 4 thus only includes minors born abroad who hold a residence permit.

Table 4

Distribution (%) of inflows by age group, by first year of validity of first residence permit of 1 year or more

Age group 2012 2013 2014 2015 2016 2017 0–17 years 9.7 9.5 10.3 10.2 10.3 10.3 18–34 years 64.4 62.8 62.2 62.5 63.1 64.5 35–64 years 24.4 26.2 25.7 25.5 24.9 23.7 65+ years 1.5 1.5 1.7 1.7 1.8 1.6

Distribution (%) of inflows by age group, by first year of validity of first residence permit of 1 year or more

Coverage: See Table 3.

22Figure 5 gives a more detailed representation of the age-sex distribution of flows in 2017. The peak at age 18 is explained by individuals who arrived as minors and who waited until age 18 to apply for a residence permit. Students also contribute to the large share of young people in the distribution. The distributions of women and men are similar overall, although the male distribution is more markedly bimodal. The mean age at entry into France is 29.1 years for women and 28.4 years for men.

Figure 5

Distribution of inflows by age and sex in 2017

Distribution of inflows by age and sex in 2017

Coverage: See Table 3.

23Most incoming migrants (excluding those from the EEA and Switzerland) are African nationals. They accounted for 58.4% of the total in 2017, the highest percentage since 2012 (Table 5). The share of arrivals from Asia continued to increase slightly in 2017, while that of arrivals from Africa and America decreased.

Table 5

Distribution (%) of inflows by continent of origin, by first year of validity of first residence permit of 1 year or more

Continent of origin 2012 2013 2014 2015 2016 2017 Africa 57.0 57.0 58.0 58.2 57.8 58.4 America 11.5 10.8 10.5 10.4 9.4 8.9 Asia 24.5 25.3 24.5 24.4 25.6 25.8 Europe 6.3 6.2 6.3 6.3 6.7 6.3 Oceania 0.4 0.4 0.4 0.4 0.4 0.4

Distribution (%) of inflows by continent of origin, by first year of validity of first residence permit of 1 year or more

Coverage: See Table 3. Turkey is included in Asia. Europe includes all European countries outside the EEA and Switzerland. The total does not necessarily sum to 100 due to rounding and missing values.

24Women no longer account for the majority of inflows, and they represented only 48.9% of total entries in 2017. This is a sharp drop from the peak reached in 2014 (Table 6). In 2017, there were fewer women than men among inflows from Africa and Asia. They formed a majority among inflows from all other continents, however, especially from the Americas and Europe.

Table 6

Share of women (%) in inflows by continent of origin, by first year of validity of the first residence permit of 1 year or more

Continent of origin 2012 2013 2014 2015 2016 2017 Africa 49.0 49.2 49.9 49.3 48.3 46.7 America 58.3 58.3 57.7 56.7 57.3 58.1 Asia 54.7 54.1 53.8 53.0 51.3 48.5 Europe 60.4 60.4 60.2 60.0 58.6 58.1 Oceania 52.4 55.4 50.1 52.7 53.5 54.8 Overall 52.2 52.2 52.3 51.6 50.6 48.9

Share of women (%) in inflows by continent of origin, by first year of validity of the first residence permit of 1 year or more

Coverage: See Tables 3 and 5.

3 – A sharp decrease in the share of admissions for family reasons

25In 2017, 107,405 people were admitted to France for family reasons [10]—still the main reason for admission. In absolute terms, the level is equivalent to that of 2013, but the relative contribution of these admissions to total inflows has fallen sharply since then. They represented 45.2% of inflows in 2017, a share that dropped almost 4 percentage points in 1 year, while the shares of admissions for educational (26.8%), humanitarian (15.2%), and employment-related reasons (8.8%) increased (Table 7). Admissions for humanitarian reasons mainly concern two types of immigrants: foreigners who are ill (3,898 people in 2017) or admitted as refugees, stateless persons, or beneficiaries of territorial asylum or subsidiary protection (32,111 people). [11] After increasing by more than 40% in 2016, the number of residence permits granted for this second set of reasons rose again by almost 50% in 2017. This reflects the increase in asylum applications received since 2014 (d’Albis and Boubtane, 2018a). Among people admitted for employment-related reasons (20,967 in 2017), almost 67% are wage employees or self-employed. The remainder are seasonal or temporary workers, scientists, and artists.

26The decrease in the share of women entering France in 2017 mainly concerns those entering for humanitarian reasons. In 2017, women represented 36.8% of inflows for this reason, compared with 41.3% the previous year (Table 8). Women are still over-represented among immigrants admitted for family reasons and under-represented among those admitted for humanitarian and, above all, employment reasons. Among students, women are slightly outnumbered by men.

Table 7

Distribution (%) of inflows by reason for granting first residence permit valid for 1 year or more, by first year of permit validity

Reason for admission 2012 2013 2014 2015 2016 2017 Family 55.5 56.1 55.0 52.7 49.0 45.2 Education 23.8 24.0 23.8 25.3 25.9 26.8 Humanitarian 9.7 8.9 9.9 10.2 12.7 15.2 including refugee 6.1 5.6 6.4 7.2 9.8 13.5 Employment 6.6 6.7 7.2 7.7 8.1 8.8 Various and unspecified 4.5 4.4 4.2 4.2 4.3 4.0

Distribution (%) of inflows by reason for granting first residence permit valid for 1 year or more, by first year of permit validity

Coverage: See Table 3. The ‘refugee’ line covers permits granted on the following grounds: ‘refugee and stateless, territorial asylum and subsidiary protection’.

Table 8

Share of women (%) in inflows by reason for granting first residence permit valid for 1 year or more, by first year of permit validity

Reason for admission 2012 2013 2014 2015 2016 2017 Family 57.3 57.1 58.3 58.1 58.0 58.5 Education 51.1 50.4 50.0 49.0 49.4 48.1 Humanitarian 43.5 44.1 44.8 44.6 41.3 36.8 Employment 23.5 24.9 23.1 24.8 23.6 24.2 Overall 52.2 52.2 52.3 51.6 50.6 48.9

Share of women (%) in inflows by reason for granting first residence permit valid for 1 year or more, by first year of permit validity

Coverage: See Table 3.

27Reasons for admission are distributed differently from one continent of origin to another (Table 9), with family reasons still taking first place for immigrants from all continents. Admissions for this reason are over-represented among Africans (51% of permits in 2017 compared with 45.2% on average) and Europeans (51.6%), and under-represented among Asians (30.8%). Educational reasons are over-represented among Asians (28.2% vs. 26.8% on average) and Americans (30.4%) and under-represented among Europeans (11.1%). Humanitarian reasons account for a large share of permits granted to Asians and to Europeans (25.3% vs. 15.2% on average) mainly nationals from Russia, Kosovo, and Albania. The share is very small among Americans (3.5%), for whom employment-related reasons are over-represented (12.8% vs. 8.8% on average).

28Among migrants from Africa, the share of admissions for family reasons in 2017 continued to decrease in favour of other reasons. Among migrants from America, the ongoing decrease in admissions for family reasons is counterbalanced by an increase in admissions for education and employment reasons. Since 2013, the share of Asians migrating to France for family and educational reasons has fallen in favour of humanitarian reasons, due notably to the war in Syria. Last, since 2013, European migration has increased for humanitarian reasons while falling for educational and family reasons.

Table 9

Distribution of inflows by reason for granting first residence permit valid for 1 year or more, continent of origin, and first year of permit validity (numbers and percentages)

Continent of origin and reason 2012 2013 2014 2015 2016 2017 for admission Africa Family 66,486 70,654 73,586 74,871 70,746 70,912 Education 19,810 22,067 23,416 27,858 31,321 38,288 Humanitarian 8,017 8,081 9,275 9,115 12,210 15,373 Employment 4,996 5,608 6,405 6,919 8,096 10,236 Africa (%) Family 64.8 64.4 63.5 61.2 56.1 51.0 Education 19.3 20.1 20.2 22.8 24.8 27.6 Humanitarian 7.8 7.4 8.0 7.5 9.7 11.1 Employment 4.9 5.1 5.5 5.7 6.4 7.4 America Family 9,914 10,175 10,367 10,017 9,327 9,348 Education 5,920 5,933 6,062 6,504 6,113 6,427 Humanitarian 591 493 425 419 452 736 Employment 2,190 2,138 2,193 2,921 2,513 2,716 America (%) Family 48.0 49.1 49.4 45.9 45.6 44.2 Education 28.7 28.6 28.9 29.8 29.9 30.4 Humanitarian 2.9 2.4 2.0 1.9 2.2 3.5 Employment 10.6 10.3 10.5 13.4 12.3 12.8 Asia Family 17,200 19,860 18,539 18,129 18,679 18,855 Education 15,226 16,304 16,321 16,834 17,261 17,291 Humanitarian 5,828 6,077 7,183 8,799 11,268 16,038 Employment 3,616 4,140 4,682 5,241 6,050 6,803 Asia (%) Family 39.0 40.7 37.8 35.3 33.4 30.8 Education 34.6 33.4 33.3 32.8 30.9 28.2 Humanitarian 13.2 12.5 14.6 17.2 20.2 26.2 Employment 8.2 8.5 9.5 10.2 10.8 11.1 Europe Family 5,761 6,602 6,747 7,095 7,603 7,752 Education 1,682 1,657 1,627 1,743 1,685 1,674 Humanitarian 2,705 2,205 2,668 2,830 3,631 3,804 Employment 761 865 801 856 888 966 Europe (%) Family 50.5 55.3 53.8 53.5 52.3 51.6 Education 14.7 13.9 13.0 13.1 11.6 11.1 Humanitarian 23.7 18.5 21.3 21.3 25.0 25.3 Employment 6.7 7.2 6.4 6.5 6.1 6.4

Distribution of inflows by reason for granting first residence permit valid for 1 year or more, continent of origin, and first year of permit validity (numbers and percentages)

Coverage: See Table 3.

4 – Over 46,000 asylum seekers admitted for residence in 2017

29Asylum seekers may be admitted for residence in France (i.e. receive a residence permit of 1 year or more) in several ways. If their application is accepted, they obtain a permit on humanitarian grounds and are counted in the fourth row of Table 7. Some of those whose application is rejected are admitted for residence on different grounds, most often for family reasons. The rates of admission for residence by submission date of the asylum application are given in d’Albis and Boubtane (2018a). The perspective here is different. Table 10 shows the annual inflows of people having submitted an asylum application to the French Office for the Protection of Refugees and Stateless Persons (OFPRA). They totalled 46,174 in 2017 and represented 19.4% of overall inflows. The number of asylum seekers admitted for residence and their share of overall flows were at their highest levels since 2012. Asylum seekers are not admitted solely on humanitarian grounds, so the total number of admissions is higher than the number admitted for this reason alone; 24.2% were admitted for family reasons in 2017. The share of women among incoming asylum seekers is lower than among overall inflows; they represented 36.6% of the total in 2017, down sharply from previous years. Since 2013, the largest share of incoming asylum seekers has been of Asian origin, representing more than 45% of the total in 2017. Africans accounted for 38%.

Table 10

Inflows of asylum seekers by first year of validity of first residence permit valid for 1 year or more

2012 2013 2014 2015 2016 2017 Inflows 22,169 25,132 25,703 27,507 35,262 46,174 Share of women (%) 39.7 40.7 41.5 41.1 39.7 36.6 Continent of origin (%) Africa 41.1 37.8 37.9 35.9 36.9 38.2 America 4.9 5.2 4.7 4.1 3.0 2.8 Asia 36.6 41.6 41.4 44.0 43.9 45.5 Europe 15.7 13.8 14.5 14.9 15.3 12.9 Reason for admission (%) Family 36.8 45.1 38.1 33.1 28.7 24.2 Education 0.5 0.5 0.6 0.5 0.3 0.3 Humanitarian 57.3 48.4 54.4 59.5 64.3 70.1 Employment 4.7 5.3 6.2 6.2 6.3 5.1

Inflows of asylum seekers by first year of validity of first residence permit valid for 1 year or more

Coverage: Permits granted in France and abroad to foreign nationals who applied for asylum between 1985 and the first year of validity of the first residence permit valid for one year or more. Permits granted in year n and recorded in the data extracted in July of the year n + 2.

5 – Major differences across the European Union

30Immigration flows vary substantially across the 28 European Union member countries. They can be compared using Eurostat data (see note 8). In 2017, immigrants entering the 28 EU countries numbered 3,371,290. This figure includes Europeans, so it is higher than the flow of third-country foreigners entering the European Union. Figure 6 gives the share of each country in this total. France is the fifth receiving country behind Germany, the United Kingdom, Spain, and Italy. These four countries received almost 63% of the total number of immigrants to the European Union in 2017.

Figure 6

Share (%) of total immigration flows received by each European Union member country in 2017

Share (%) of total immigration flows received by each European Union member country in 2017

31The share of total flows received by each country has changed since 2012. The shares admitted by the five main receiving countries between 2012 and 2017 are given in Table 11. Inflows to Germany peaked in 2015 when the country admitted large numbers of asylum seekers, mainly from Syria. Numbers have since returned to the level of 2013. Conversely, a sharp increase is observed for Spain and, to a lesser extent, Italy and the United Kingdom.

Table 11

Share (%) of total migration inflows admitted by the main receiving countries, by year

Receiving country 2012 2013 2014 2015 2016 2017 Germany 20.6 23.7 27.4 38.4 27.4 23.5 United Kingdom 17.0 17.5 19.0 14.5 15.4 16.7 Spain 11.1 9.7 9.1 7.7 10.5 13.5 Italy 13.1 10.9 8.6 6.6 7.8 8.9 France 8.6 8.5 7.4 6.2 7.2 7.2

Share (%) of total migration inflows admitted by the main receiving countries, by year

32The country rankings are very different, however, when migration flows are expressed as a proportion of each country’s population (on 1 January of the year in question). In 2017, six countries (Malta, Luxembourg, Cyprus, Sweden, Austria, and Ireland) had immigration rates above 1%, while 12 countries, including France, had rates below 0.5% (Figure 7).

Figure 7

Immigration rates (%) to European Union member countries in 2017

Immigration rates (%) to European Union member countries in 2017

33Between 2012 and 2017, immigration rates increased most notably in Estonia (+725%), Lithuania (+333%), and Portugal (+218%), while they decreased in Belgium (−8%), Italy (−8%), and Poland (−6%). Figure 8 shows the mean annual growth in immigration rates for the 28 EU countries.

1 – A slower decline in period fertility

34Births have declined continuously over the last 8 years, and estimates for the first 5 months of 2019 do not suggest a trend reversal, although levels have somewhat stabilized (Figure 9). [12] For the 4th year in a row, this steady downtrend has been accompanied by a slight drop in the total fertility rate (TFR), which now stands at 1.87 children per woman for the whole of France versus 1.89 in 2017 (1.84 for metropolitan France alone) and 1.86 in 2016 (Appendix Table A.4) (Robert-Bobée and Volant, 2018). The year 2018 was thus marked by a slower decline in births and fertility. Between 2017 and 2018, fertility rates over age 35 increased slightly after 3 years of stability, but the rise was insufficient to offset the uninterrupted decline observed at ages 20–29 since at least 2013 (Table 12).

Figure 8

Mean (%) of annual rates of growth in immigration rates in European Union member countries between 2012 and 2017

Mean (%) of annual rates of growth in immigration rates in European Union member countries between 2012 and 2017

Figure 9

Monthly birth numbers from January 2011 to May 2019

Monthly birth numbers from January 2011 to May 2019

Interpretation: The moving averages are calculated over 5 months. The value for March 2019 (green curve) thus corresponds to the mean of the months from January to May 2019.
Coverage: Whole of France.

Table 12

Fertility by age group from 2013 to 2018

Age Sum of age-specific rates (per 1,000 women) Absolute variation* reached 2013 2014 2015 2016 2017 2013 2014 2015 2016 (p) 2017 (p) 2018 (p) – – – – – in the year 2014 2015 2016 2017 2018 Under 20 38 37 35 32 30 29 –1 –2 –3 –2 –1 20–24 257 252 241 232 224 215 –5 –12 –9 –8 –10 25–29 618 612 592 575 559 545 –6 –20 –17 –16 –14 30–34 650 658 648 645 638 636 8 –9 –4 –6 –2 35–39 338 347 347 345 345 348 9 0 –2 0 3 40–44 82 87 87 89 92 94 5 0 2 3 2 45+ 5 6 6 6 6 7 1 0 0 1 0 Total (TFR*) 1,988 1,999 1,955 1,924 1,895 1,873 11 –43 –32 –29 –22

Fertility by age group from 2013 to 2018

* Total fertility rate (sum of age-specific fertility rates), expressed as a mean number of children per 1,000 women. Due to rounding, the total may differ slightly from the sum, and the variations may not correspond exactly to apparent differences.
(p): Provisional data.
Coverage: Whole of France (including Mayotte since 2014).

35Between 2007 and 2017, [13] the curve of age-specific fertility in France shifted slightly to the right (postponement) and slightly downward (lower fertility), but its overall profile changed little. There is an identical trend towards later childbearing across all EU-28 countries. However, except for France, the increase in fertility at later ages largely offsets the decrease at younger ages (Figure 10). As a result, the European TFR remained stable between 2007 and 2017 while falling slightly in France (Table 13).

36The mean age at childbearing in France was 30.6 years in 2018 (30.7 years in metropolitan France), setting an all-time record. However, this is still well below the values observed in Southern Europe and Ireland, where the mean age is close to or over 32.0 years (Table 13).

37Why has the French TFR declined recently? The downtrend may reflect a decrease in fertility among the cohorts of reproductive age but also a postponement of childbearing to later ages. Over the 30-year period between 1975 and 2006, the TFR often fell below 2.0 children per woman—with a minimum in 1993 (1.66 children per woman)—while cohort fertility remained stable at around 2.0 children per woman. This inconsistency is explained by a steady increase in the age at childbearing (mainly at first birth), which rose by more than 3 years over the period. The decline in the TFR observed since 2010, which has accelerated in the last 4 years, may or may not be followed by a renewed increase. This will depend on whether later childbearing is also accompanied by a decrease in fertility intensity.

Figure 10

Age-specific fertility rates in France and in EU-28 countries, 2007 and 2017

Age-specific fertility rates in France and in EU-28 countries, 2007 and 2017

Coverage: Whole of France / EU-28.

2 – France is still a European ‘outlier’

38Since 2011, the TFR in France has been the highest in Europe, ahead of Ireland and Iceland (if we look beyond the EU-28 countries), two countries where fertility was traditionally higher than in France. But with the gradual convergence of period fertility levels across Europe, [14] France was less of a European outlier in 2017 than 10 years earlier. The TFR has risen most notably in countries where fertility was low (the Baltic States and certain countries of Eastern Europe: Hungary, Lithuania, Poland, and Slovakia; and of Western Europe: Germany, Austria, and Portugal). At the same time, it has fallen in all countries where it exceeded 1.6 children per woman (Table 13). France has joined the group of Northern and Western European countries (Belgium, Denmark, Ireland, United Kingdom, and Sweden) characterized by a high TFR and mean age at childbearing. In the countries of Southern Europe (Greece, Italy, and Spain), mean ages at childbearing are similar to those observed in the North and West, but period fertility is low, at close to 1.3 children per woman.

Table 13

Fertility indicators of the EU-28 countries in 2007 and 2017

Country Total fertility rate Mean age Mean age at first birth Proportion of births outside marriage 2007 2017 2007 2017 2007 2017 2007 2017 France 1.98 1.87 29.8 30.5 — 28.7 51.7 59.9 Sweden 1.88 1.78 30.6 31.1 28.8 29.3 54.8 54.5 Ireland 2.01 1.77 31.3 32.1 28.8 30.3 33.1 37.6 Denmark 1.84 1.75 30.4 31.1 — 29.4 46.1 54.2 United Kingdom 1.86 1.74 29.3 30.5 — 28.9 44.4 48.2 Romania 1.45 1.71 26.8 27.9 25.0 26.5 26.7 31.2 Latvia 1.54 1.69 27.9 29.7 25.2 26.9 43.2 40.4 Czech Republic 1.45 1.69 29.1 30.0 27.1 28.2 34.5 49.0 Belgium 1.82 1.65 29.5 30.6 27.8 29.0 43.2 49.0 (a) Lithuania 1.36 1.63 27.8 29.8 25.3 27.5 27.8 26.7 Netherlands 1.72 1.62 30.6 31.4 29.0 29.9 39.5 51.0 Slovenia 1.38 1.62 29.8 30.3 28.1 28.8 50.8 57.5 Estonia 1.69 1.59 28.4 30.4 25.3 27.7 57.8 58.6 Germany 1.37 1.57 29.9 31.0 — 29.6 30.8 34.7 Bulgaria 1.49 1.56 26.4 27.6 25.0 26.1 50.2 58.9 Hungary 1.32 1.54 28.8 29.8 27.1 28.0 37.5 44.7 Austria 1.38 1.52 29.4 30.7 27.7 29.3 38.3 42.0 Slovakia 1.27 1.52 28.1 28.8 26.2 27.1 28.8 40.1 Finland 1.83 1.49 30.0 30.9 28.1 29.1 40.6 44.8 Poland 1.31 1.48 28.4 29.5 26.0 27.3 19.5 24.1 Croatia 1.48 1.42 28.6 30.3 26.7 28.6 11.5 19.9 Luxembourg 1.61 1.39 30.2 31.9 28.9 30.8 29.2 40.8 Portugal 1.35 1.38 29.4 31.2 27.6 29.6 33.6 54.9 Greece 1.41 1.35 30.1 31.4 28.8 30.4 5.8 10.3 Cyprus 1.44 1.32 30.0 31.4 28.2 29.7 8.7 20.3 Italy 1.40 1.32 31.0 31.9 — 31.1 17.9 32.8 Spain 1.38 1.31 30.8 32.1 29.4 30.9 30.2 46.8 Malta 1.35 1.26 29.1 30.5 26.9 29.0 24.9 25.9 (b) European Union (28) 1.6 1.6 29.7 30.7 — 29.1 35.1 41.1 (b) Coefficient of variation (%) 14.4 10.6 4.1 3.6 5.2 4.5 39.1 31.9

Fertility indicators of the EU-28 countries in 2007 and 2017

(a) 2016
(b) 2013
Interpretation: The seven countries with the highest figures in each column are in bold and the seven with the lowest are in italics.
The countries are ranked in decreasing order of TFR in 2017.
The coefficient of variation is a measure of dispersion. It is the ratio between the standard deviation and the mean value of the indicator. The higher the coefficient of variation, the greater the dispersion between countries.
Coverage: EU-28.

3 – Women born in 1980 will have a completed fertility of more than 2.0 children per woman

39Using data available in 2017, a reliable estimate of completed fertility can be obtained up to the 1980 birth cohort, aged 37 in that year. Beyond that age, the contribution to completed fertility remains marginal, although it has been increasing almost universally since the mid-1990s after more than 4 decades of decline (Beaujouan and Sobotka, 2019). [15] Our estimates are based on two assumptions: stable fertility of women aged 40–44 and 45–49 after 2017, and a continuation of the trend in age-specific rates observed between 2012 and 2017. Under both scenarios, women born in 1980 will doubtless have more than 2.0 children on average in France (Appendix Table A.5). Applying this same logic to women born in the 1990s is more risky (they were only 27 years old in 2017), but even under a pessimistic fertility scenario, [16] their completed fertility should be equal to or above 1.95 children per woman and more probably 2.0 children.

40Alongside Ireland, France is the only EU-28 country where the cohort of women born in 1980 will have more than 2.0 children per woman. In all countries, however, completed fertility will be equal to or above the TFR of 2012 (Figure 11, all countries are below the bisector). [17] The effect of birth postponement on the low levels of period fertility is real. In most countries, notably those of Southern Europe, completed fertility will nonetheless be well below cohort replacement level.

Figure 11

Completed fertility of the 1980 cohort (estimate) and TFR of the 28 EU countries in 2012

Completed fertility of the 1980 cohort (estimate) and TFR of the 28 EU countries in 2012

Interpretation: By the end of their reproductive lives, French women born in 1980 will have 2.07 children on average, and a woman who behaves throughout her childbearing years like the women of 2012 will have 2.0 children on average.
Coverage: EU-28, excluding Latvia and Croatia.

4 – A growing share of childless women?

41Some authors have recently described five combinations of fertility by birth order associated with a low fertility level of 1.6 children per woman: a reference model; [18] a high childlessness model; a model with a large share of only children; a fourth model where families with more than two children are rare; and a last model polarized between childless couples and families with three or more children (Zeman et al., 2018). Most European countries correspond to one of these models, but not France, which is close to the reference model but with a higher probability of having at least one child (0.85 vs. 0.80), a second child after the first (0.80 vs. 0.72), and a third child after the second (0.40 vs. 0.30). [19] Cohort fertility in France remains close to replacement level (2.1 children per woman) due to low levels of childlessness and more frequent third births than elsewhere (Breton and Prioux, 2005, 2009; Frejka, 2008; Toulemon et al., 2008; Ayerbe and Breton, 2015; Beaujouan et al., 2017).

42Should the decreasing TFR observed in France be considered the end of an atypical model within Europe? To answer this question, recent data on births by order are needed. Unfortunately, for historical reasons, birth order is not recorded accurately, as births were registered by order within the union and not by biological order (Breton and Prioux, 2009), even though all the mother’s live-born children must be mentioned on the birth certificate. [20] Nevertheless, we can estimate order-specific fertility by applying the ‘own children’ method to data from the annual census surveys. This method is used by INSEE to populate the Eurostat database (Desplanques, 2008; Davie and Niels, 2012). We have extended the results presented in the 2012 article on demographic developments (Prioux and Barbieri, 2012) up to the year 2016 (Figure 12A). The decrease in the TFR is attributable mainly to a decline in the first-order component (mean number of first births), which fell from 0.89 in 2010 to 0.80 in 2016 (65% of the decrease in the TFR). The second-order component (mean number of second births) fell from 0.71 to 0.67 and that of third and higher births from 0.41 to 0.40. The sharp drop in the first-order component may signal not only an increase in cohort childlessness but also a continued trend towards ever-later first childbearing. This second hypothesis is a reality: the age at first birth (measured using the same method) would be 28.7 years for a cohort that behaved in the same way as women in 2016 throughout their reproductive lives versus 28.1 years in 2010 (Figure 12B).

Figure 12

Decomposition of the TFR by birth order (A) and mean age at childbearing by birth order (B), 1970–2017

Decomposition of the TFR by birth order (A) and mean age at childbearing by birth order (B), 1970–2017

Coverage: Metropolitan France.

43First-birth postponement has a mechanical impact on age at birth of higher-order children. This trend towards later childbearing cannot continue indefinitely, however, as the risk of infecundity increases sharply beyond age 30. Indeed, very few countries in Europe have a mean age at first childbirth of 30 years or higher (Greece, Ireland, Italy, Luxembourg, and Spain; Table 13). That said, it is sometimes difficult to predict changes linked to societal factors, such as assisted reproductive technology. Age at childbearing may also increase if women are allowed to freeze their eggs and men to freeze their sperm for non-medical reasons. This possibility was included in the bioethics bill debated by the French National Assembly in 2019.

5 – Almost 1 in 4 children born to foreign parents

44A recent analysis based on various data sources, including the census, has revealed the small but growing contribution of immigrant women to French births. In 2017, in France (excluding Mayotte), 18.8% of births were to immigrant mothers (born as foreigners outside France) versus 16% in 2009, but only 0.1 point of the TFR is linked to higher fertility of immigrant women (Volant et al., 2019). Older studies measured the same tendency before 2005, taking as reference births not to immigrant women but to foreign women (Prioux, 2006a; Héran and Pison, 2007). Under this definition, the proportion of children born between 2009 and 2017 to at least one foreign parent rose from 20.3% to 23.8% [21] (of which 34% and 40%, respectively, were born to a mixed-nationality couple).

6 – More non-marital births, but most still take the father’s name

45In 2018, 60.3% of births occurred outside marriage. This percentage has been increasing steadily since the 1970s, and while fewer than half of children are now born to married parents, the proportion recognized by their father is not decreasing (Pison, 2018). France is the European country with the highest share of non-marital births (Table 13), overtaking the countries of Northern Europe that initiated this trend (Avdeev et al., 2011). Marriage as a prerequisite for childbearing is no longer the norm in France, and decreasingly so in Europe where the correlation between marriage and birth has practically disappeared (r² = 0.20 in 2017 vs. 0.28 in 2007).

46Another social gender norm persists: that of giving the child the father’s surname. Before 2005, it was a legal requirement for children to take their father’s name or (very rarely) that of their mother if the father did not recognize the child. Since 2005, parents have been free to give the father’s or mother’s name only, the mother’s name followed by the father’s, or vice-versa, in accordance with precise legal rules (Mazuy et al., 2013). Between 2012 and 2018, [22] the situation changed very slowly. Slightly more than 11% of children born in 2018 received the names of both their father and mother (vs. 9% in 2012), and 82.5% their father’s name only (vs. 83.0% in 2012). Among children born outside marriage, the proportion receiving their father’s name only (74.4%) has remained practically stable since 2012. In 2015, 14.4% of these children had a double name compared with 4.5% of children born to a married couple. The proportion also varies by the mother’s nationality and the department of birth, with more double names in the south-west of France and the Pyrénées orientales, for example (Bellamy, 2015).

1 – Increase in induced abortions in 2018

47For the whole of France, the number of induced abortions increased to 224,300 in 2018, up from 217,800 in 2017 (Vilain, 2019), while remaining below the levels recorded in 2014 (227,038) and 2013 (229,021). In metropolitan France, the total was 209,500 (Appendix Table A.8). Expressed as a number of induced abortions per 1,000 women aged 15–49, the figure rose from 14.8 in 2017 to 15.5 in 2018 for the whole of France (from 14.4 to 15.0 in metropolitan France). The total abortion rate (mean number of abortions per woman) also rose slightly to 0.56 (Mazuy et al., 2015). These various indicators all point to a rise in induced abortion in 2018.

48This increase is observed especially for women aged 25 and older (Figure 13). Among the youngest women, notably the under-20s, a continued downward trend reflects the increase in age at childbearing and a high level of contraceptive coverage; only a small minority of sexually active women below age 25 do not use any form of contraception (Rahib et al., 2017). Geographically, while recourse to abortion varies considerably across regions (Breton et al., 2017; Vilain, 2018, 2019), it is increasing everywhere, including in the overseas departments and in regions where abortion is least frequent, such as Pays de la Loire, where practitioners are also scarce.

Figure 13

Induced abortions per 1,000 women by age group, 1990–2018

Induced abortions per 1,000 women by age group, 1990–2018

Coverage: Whole of France.

49This slight increase coincides with a decline in fertility in 2018, although the decrease is smaller than in the 2 preceding years (see below). A new survey is needed to shed light on these contrasting trends in annual fertility rates and in abortions, which are doubtless multifactorial: fewer births at young maternal ages, concentration of births around age 30, a halt in the trend towards ever-later childbearing, changes in sexual and contraceptive behaviour, increasing economic insecurity, etc.

2 – Abortion in the European Union

50Most countries of the European Union have decriminalized, legalized, and/or lifted restrictions on access to abortion (Figure 14). The majority have enacted laws authorizing abortion on request, most often within a legal limit of 12 weeks of amenorrhoea (these limits range between 10 and 24 weeks). Abortion became legal only recently in Luxembourg (22 December 2014), Cyprus (March 2018) and Ireland (13 December 2018). In other countries, access is restricted. On 9 July 2019, the British parliament (in the absence of the Northern Ireland Assembly) voted amendments that included a law extending abortion rights in Northern Ireland. They were enacted by the House of Lords on 22 October 2019 without the local executive power. Access should be effective April 2020. In Finland, abortion on request is not strictly recognized under law, but in practice it is available to all women. Access to abortion in Poland is restricted to cases of danger to the mother’s health, foetal malformation, or rape. Malta is the only country where it is totally prohibited.

Figure 14

Abortion rights in the countries of the European Union

Abortion rights in the countries of the European Union

51Governments make regular attempts to limit access to abortion (through stricter conditions for health insurance reimbursement, specific restrictions for minors, bills to reverse the liberalization of abortion on request), as was the case in Spain in 2015. The Spanish government finally withdrew its bill while making abortion for minors conditional upon parental consent. The conscience clause evoked by physicians who refuse to perform abortions [23] and the lack of appropriate training are major obstacles to the availability of induced abortion. The situation in Italy is especially problematic in this respect. Provision has become so limited in certain regions that many women are forced to travel elsewhere in the country or abroad to access abortion.

52The abortion rates in EU countries are at an intermediate level with respect to other world regions. They are above those observed in North America and Oceania but below those of Africa, Latin America, and the Caribbean, where rates are highest (Rossier, 2014; Sedgh et al., 2016; Guillaume and Rossier, 2018). In Europe, rates fell from 48 to 27 abortions per 1,000 women between 1995 and 2008. They also fell elsewhere in the world over this same period, from 33 to 28 per 1,000 in Africa and Asia and from 37 to 32 per 1,000 in Latin America.

53According to data published by Eurostat (non-exhaustive) [24] and national statistical offices, rates are lowest in Germany, certain Eastern European countries and the Balkans (Latvia, Lithuania, Poland, Czech Republic, Slovakia, and Slovenia), Finland, and the countries of Southern Europe (below 8 per 1,000). They are slightly higher, at close to 10 per 1,000, in Belgium, the Netherlands, Iceland, Hungary and Romania, and higher still in France, Bulgaria, and England and Wales.

54Overall abortion rates vary widely within the European Union, but they have fallen substantially since the 1990s thanks to the easing of abortion restrictions and growing contraceptive uptake.

55Rates can thus be correlated with contraceptive coverage, overall fertility, and conditions of access to abortion. Low rates may reflect a situation where abortion is restricted and highly stigmatized. For example, rates are intermediate or low in countries where abortion is legal but where the influence of religion is strong, such as Italy and Poland. The patterns observed in Eastern European countries are linked to their history of regime change; in Poland, abortion was legalized in 1956, then banned once again in 1993. Besides the legal barriers to abortion in some countries, practical access is not necessarily guaranteed, and the abortion services available to women do not always cover their diverse range of needs. Multiple obstacles may stand in the way of women seeking a safe abortion. These include a lack of accessible medical services at the local level; a conscience clause that reduces the number of available physicians; cost; worries about anonymity; [25] lack of provision for late abortions; a shortage of trained professionals, etc. (Guillaume and Rossier, 2018). France is no exception, with a shortage of practitioners in some regions, severe bottlenecks at certain times of year (in the summer months especially), and the closure of family-planning centres (Commission IVG, 2016).

56Yet legal and safe abortion is positively correlated with a lower mortality risk. The issues of legalization and quality of coverage are key to ensuring women’s sexual and reproductive health (WHO, 2012; Council of Europe, 2017).

3 – Improving provision

57New channels for providing non-hospital sexual and reproductive healthcare are developing in France and elsewhere, providing a means to extend abortion services or to complement existing provision in regions where it is poor or inadequate. [26] For example, web-based telemedicine can provide access to abortion pills for women living in areas where services and practitioners are lacking. [27] It also gives women greater independence in their reproductive health choices and practices (Aiken et al., 2016; Sheldon, 2016).

58In France, midwives can now perform medical abortions under a decree published in 2016 (Decree of 2 June 2016 on the capacity of midwives to perform medical abortions). In 2018, 248 midwives performed medical abortions in France. They represented 12% of all non-hospital practitioners who have performed such procedures (Vilain, 2019). Only early abortions are covered by these forms of non-hospital care, however; they cannot make up for the shortage of services for later abortions. The proportion of midwives who perform abortions will probably increase over time (especially with the rising share of medical abortions, which accounted for 69% of all abortions in 2018), and their legal scope of competence may be broadened in the future to include surgical abortions. In late September 2019, the French Ministry of Health announced its intention to extend authorization for midwives to perform surgical abortions (from the first quarter of 2020) and included abortion services among hospital evaluation criteria. The question of specialist training remains crucial. Belgium, for example, announced in August 2019 that abortion procedures are to be included in the training curriculum for medical students.

1 – Registered unions

Marriages nearing a record low, PACS unions close to peak numbers

59In 2017, 233,915 marriages and 193,950 PACS unions [28] were registered in France, a ratio of around 4 PACS per 5 marriages. Despite the opening of marriage to same-sex couples in 2013, [29] the number of marriages was close to its record low (232,725 marriages in 2016), while the number of PACS reached the second-highest level ever attained, after the peak of 205,000 in 2010 (Figure 15). [30] Compared with 2016, the number of marriages increased by 0.5% (+1,190), and PACS unions rose slightly more, by 1.3% (+2,413).

60According to provisional data, the 235,000 marriages in 2018 (Papon and Beaumel, 2019) signal a potential continuation of the ongoing slight uptrend. The number of PACS registrations in 2018 is not yet known. As of 1 November 2017, PACS unions are now registered and dissolved in municipal registry offices and no longer in magistrates’ courts, as had been the case since 1999. [31] Subject to confirmation of initial data published by the Ministry of Justice, among the 193,950 PACS unions concluded in 2017, 44,493 were registered in municipal registry offices in the months of November and December, representing 23% of the total for that year. This was an increase of 13,382 with respect to the same period in 2016, reversing the decline observed between 2015 and 2016 in PACS unions registered at magistrate’s courts in the last 2 months of the year (−4.4%). This new provision may have encouraged more couples to register a PACS at a municipal registry office rather than with a notary: in November and December 2017, 32,976 PACS unions were registered with notaries, slightly down from 2016, representing 17% of the annual total (17.4% in 2016).

Figure 15

Annual numbers of marriages and PACS unions by partners’ sex since 2000

Annual numbers of marriages and PACS unions by partners’ sex since 2000

Note: Provisional data for 2018.
Coverage: Whole of France (including Mayotte since 2014).

61Each year since 2013, marriages have outnumbered PACS unions among lesbian couples (Table 14), while for gay couples, PACS unions have again overtaken marriages since 2016. In 2017, same-sex unions represented 3.1% of all marriages and 3.8% of all PACS unions.

Table 14

Number of unions (PACS or marriage) registered between 2013 and 2017 by partners’ sex

Marriages PACS unions Year Between a man and a woman Between Between 2 men 2 women Between a man and a woman Between 2 men Between 2 women 2013 231,225 4,307 3,060 162,714 3,354 2,734 2014 230,770 5,666 4,856 167,487 3,519 2,745 2015 228,565 4,085 3,666 181,949 3,933 3,085 2016 225,612 3,672 3,441 184,444 3,863 3,251 2017 226,671 3,637 3,607 186,614 4,084 3,252 2018 (p) 229,000 6,000

Number of unions (PACS or marriage) registered between 2013 and 2017 by partners’ sex

(p): Provisional data.
Coverage: Whole of France (excluding Mayotte up to 2013 and including Mayotte from 2014).

62Although the number of intimate relationships and unions is increasing across birth cohorts for both men and women (Rault and Régnier-Loilier, 2015), the vast majority of marriages are still first marriages. The EPIC survey, covered in a special issue of Population (Rault and Régnier-Loilier, 2019), shows that more than half of the respondents in the 1970 birth cohort had experienced at least two important relationships before age 40. [32] Yet in 2017, almost 80% of marriages were first marriages: 79.8% of men and 80.9% of women who married a different-sex partner in that year were never-married. The proportions were higher for same-sex couples, at 91.6% and 89.3%, because while some people may have been married to a different-sex partner in the past, there have been few years in which to marry and divorce a same-sex partner since 2013.

Box 1. Total period rates

Calculating total period rates eliminates the effects of variations in population size and age structure, making it possible to describe changes in union formation behaviour over the life course. Taking account of all marriages (including remarriages) and at all ages (including after age 50), we calculate a mean number of marriages per 1,000 people under the conditions of a given year (Table 14). An identical calculation can be made for PACS unions. Marriage and PACS indicators can be compared for unions as a whole but not for distinct union orders because this information is not recorded when a PACS union is registered.
Standard total period first-marriage rates are obtained by summing probabilities or rates up to age 49 (Appendix Table A.9). However, as first marriages and remarriages after age 50 are becoming more frequent, it is useful to calculate indicators that take account of all marriages at all ages, including after 50, by summing rates at all ages (or age groups) from age 18, both for marriages and for PACS unions. This also makes it possible to compare marriages with PACS unions, which cannot be distinguished by order in available data.

63This downtrend in marriages in France over the last decade is only weakly related to changes in the population of marriageable age. While the number of marriages fell by 23.4% between 2000 and 2018, the population of France at the ages when marriage is most frequent (ages 25–54) decreased by only 3% for men and 1% for women over the period.

Convergence between numbers of PACS unions and marriages, but differences in timing

64Under the marriage conditions at all ages of 2017, the total period marriage rate (Box 1) across all marriageable ages was 571 per 1,000 men and 561 per 1,000 women (Table 15).

65Between 2014 and 2017, the total period marriage rate fell by 15 points for men (from 586 to 571) and by 12 points for women (573 to 561). This decrease stems not only from the decline in the total period first-marriage rate (by 2 points for men and 3 points for women) and remarriages (by 3 points for both sexes) but also from the stronger decrease in same-sex first marriages (from 24 to 16 for men, 20 to 15 for women) after the exceptionally high number recorded in 2014—the first full calendar year following the legalization of marriage for same-sex couples.

66Conversely, between 2014 and 2017, the total period PACS rate per 1,000 people increased sharply (from 438 to 495 for men, 423 to 479 for women), thus contributing to a convergence in the rate of PACS unions and marriage. This increase appears higher in relative terms for same-sex PACS unions (17.5% for men, 19.4% for women) than for different-sex PACS unions (12.8% for men, 12.9% for women).

67In terms of union timing, while both men and women enter a PACS union 2.5 years earlier, on average, than marriage, this age difference increased between 2014 and 2017. While mean age at entry into a PACS union remained stable at 35.0 years for men and 32.5 years for women between 2014 and 2017, the mean age at marriage increased by 0.5 years for men and 0.6 years for women. This increase is due mainly to an increase in age at first marriage, which rose by 0.8 years for men and women between 2014 and 2017, while the increase in mean age at remarriage of widow(ers) and divorcees is smaller, for men especially (+0.2 to +0.3 years).

68Last, same-sex couples still marry much later than different-sex couples, gay couples especially (43.2 years), despite a sharp decrease between 2014 and 2017 (2.1 years for men, 2.4 years for women) linked to the disappearance after 2013 of the ‘stock’ of older couples wishing to marry. This same ‘rejuvenation’ is observed for same-sex PACS unions (a decrease of 0.7 years to 37 years for men, and of 1.3 years to 35.2 years for women).

Table 15

Number of unions per 1,000 people (total period event rate*), all ages at entry into union, and mean age at union in 2014 and 2017 by marital status at time of union

Different-sex marriages All marriages Men Women Men Women Overall Single Widowed Divorced Overall Single Widowed Divorced Total period marriage rate (per 1,000 people) 2014 586 573 560 453 6 101 551 451 6 94 2017 571 561 554 451 5 98 544 448 5 91 Mean age at marriage (years) 2014 37.2 34.5 36.9 33.7 59.2 49.6 34.3 31.7 52.6 45.6 2017 37.7 35.1 37.5 34.5 59.4 49.9 35.0 32.5 53.3 46.1 Same-sex marriages Men Women Overall Single Widowed Divorced Overall Single Widowed Divorced Total period marriage rate (per 1,000 people) 2014 26 24 0 2 22 20 0 3 2017 17 16 0 1 17 15 0 2 Mean age at marriage (years) 2014 45.3 44.5 55.7 53.1 40.6 39.5 51.6 48.6 2017 43.2 42.3 59.4 52.5 38.2 37.2 51.3 46.9 All PACS unions Different-sex PACS unions Same-sex PACS unions Men Women Men Women Men Women Total period PACS rate (per 1,000 people) 2014 438 423 420 410 18 13 2017 495 479 474 463 21 16 Mean age at PACS (years) 2014 35.0 32.5 34.9 32.4 37.7 36.5 2017 35.0 32.5 34.9 32.4 37.0 35.2

Number of unions per 1,000 people (total period event rate*), all ages at entry into union, and mean age at union in 2014 and 2017 by marital status at time of union

* Aggregate number of events (marriages, PACS) at all ages and mean age calculated from all event frequencies at all ages (including beyond age 50).
Coverage: Whole of France.

69For different-sex unions, in 2017, the modal age at entry into a PACS union was around 25 years for women and 28 years for men compared with 28 and 30 years for marriages (Figure 16). Marriage becomes more frequent than PACS unions from these ages. Beyond age 40, marriage is still the preferred choice for formalizing a union (more than 1.5 times more frequent than the PACS).

Figure 16

Union formalization rates by age, sex, and type of couple

Union formalization rates by age, sex, and type of couple

Note: Number of persons who married (or registered a PACS) in 2017 per 1,000 people of a given age.
Coverage: Whole of France.

70Regarding the much smaller number of formalized same-sex unions, PACS unions are still highly concentrated around age 28, for both gay and lesbian couples, while age at marriage is quite heterogeneous. Lesbian couples marry mainly around age 30, while among gay couples the dispersion is wider, with relatively similar marriage frequencies at all ages between 28 and 50 in 2017. Among gay couples, there is a clear preference for the PACS at young ages and marriage later in life (after age 40), while for lesbian couples, marriage is chosen at much earlier ages.

71As in previous years (Breton et al., 2017), a peak in marriages at ages ending in 0 is again observed in 2017 and is even becoming more pronounced. It mainly concerns marriages, both remarriages and first marriages, between different-sex partners, notably at ages 40 and 50. This preference for rounded ages is not observed for PACS unions, however, which is consistent with previous findings on the differences in importance attached to the two forms of union officialization (Rault and Régnier-Loilier, 2019).

At age 30, more than 1 in 5 newlyweds were in a PACS union before marrying

72Each year, a share of the couples who marry were already in a PACS union. This means that not all marriages represent a newly contractualized union. In 2016, 40,670 PACS unions were dissolved by a marriage; this means that the number of PACS unions was almost equal to that of marriages between couples not previously in a PACS. Altogether, from 1999 to the end of 2016, almost 270,000 PACS unions were dissolved by marriage. [33] With the growing number of PACS unions dissolved by marriage and the concomitant decrease in marriages, the share of different-sex marriages following a PACS is rising. They accounted for 17% of the total in 2016. Among same-sex couples, however, the share already in a PACS has been decreasing since 2013, accounting for 34.7% of lesbian marriages and 37.3% of gay marriages in 2016. Now that marriage has been open to same-sex couples for several years, the number already in a PACS union and wishing to marry is falling. These proportions also vary by the partners’ age; more than 1 in 5 newlyweds aged 30–34 were previously in a PACS union (21.9% of men and 23.1% of women who married at this age in 2016), but this share then decreases with age (Figure 17). It is at the ages where marriage is most frequent that the share of marriages between different-sex couples already in a PACS union is highest. This is not the case for same-sex marriages (for which the share varies between 35% and 40% at all ages beyond 25).

Figure 17

Percentage of marriages that dissolved a PACS by age at marriage and partners’ sex (marriages in 2016)

Percentage of marriages that dissolved a PACS by age at marriage and partners’ sex (marriages in 2016)

Coverage: Whole of France.

2 – Marriage and PACS trends across cohorts

PACS unions most frequent among the cohorts born in the early 1980s

73The marriage trends described above are the result of a steady decline in first-marriage rates across the cohorts born since the 1950s and of a continuous increase in age at marriage (Appendix Table A.10A). An estimated 55% of men and 61% of women in the 1980 birth cohort will marry before age 50, [34] at age 31.8 on average for men and age 29.7 for women. These trends appear to be continuing in subsequent birth cohorts, among whom a decrease in the proportion ever-married at ages 25 and 30 has already been observed.

74Introduced in 1999, the PACS union has gained in popularity across successive cohorts over the last 20 years (Figure 18). At the end of 2017, for both men and women, the total period PACS rate [35] (whatever the partner’s sex) was highest among the cohorts born in the early 1980s. These were the first cohorts able to enter a PACS union from age 18, giving them an alternative to marriage from the start of their conjugal life. Among 35-year-olds born in 1982, there were 245 PACS per 1,000 men and 243 per 1,000 women. If the behaviours observed in recent years beyond age 35 remain unchanged, the total period PACS rate at age 50 in this cohort should reach 359 per 1,000 for men and 333 per 1,000 for women and a slightly higher level (402 for men, 363 for women) if PACS unions after age 50 are included. The difference between marriage and PACS frequencies in this birth cohort remains large, especially considering that only first marriages are included here.

75For older cohorts, some couples were already married in 1999 and so could not choose between the two types of union. Hence, the older the birth cohort, the lower the total period PACS rate at a given age. Given the level already reached at age 30 in the 1982 birth cohort, the total period PACS rate should continue to increase in subsequent cohorts. That said, starting with the cohorts born in the late 1980s, the total period PACS rate at age 25 levels off at around 85 per 1,000 women and 47 per 1,000 men at that age, women entering unions at an earlier age than men.

Figure 18

Total cohort PACS rate (per 1,000 people)* at different ages in the birth cohorts, by type of PACS (all PACS, same-sex PACS)

Total cohort PACS rate (per 1,000 people)* at different ages in the birth cohorts, by type of PACS (all PACS, same-sex PACS)

* Total period PACS rate at all ages from 18 years.
Coverage: Whole of France

76Among same-sex PACS unions, the total period PACS rate in 2017 is highest among slightly older cohorts (born before 1982); same-sex partners are older on average at the time of entry into union, and this was especially so in the first few years after the PACS was introduced. Among 40-year-olds born in 1977, the total period same-sex PACS rate is 10.5 per 1,000 for men and 7.2 per 1,000 for women, with gay PACS unions largely outnumbering lesbian PACS unions since 1999.

3 – Divorces and PACS dissolutions

More than half the PACS unions registered before 2007 have been dissolved

80The annual number of PACS dissolutions fell for the first time in 2017, with 82,345 dissolutions, down by 2,317 (2.7%) with respect to 2016. This decrease may be linked to the new system of PACS registration in municipal registry offices [37] and to delayed and incomplete transmission of registration and dissolution figures. The breakdown by reason for dissolution is not yet known for the whole of 2017, but the database of all PACS unions registered before November 2017 enables us, for the first time, to follow the outcomes of PACS unions by cohort and with respect to several different variables (Box 2).

81More than half of the PACS unions registered before 2007 had been dissolved by 1 November 2017 (Figure 19), the largest share (from 25% to 34% across the cohorts) by mutual consent. A substantial share (15% to 20%) was dissolved by the marriage of one or both partners. This proportion increases for the PACS unions formed between 2007 and 2010 (20% to 25%), while the share of dissolutions by mutual consent decreases across the cohorts. For PACS unions registered after 2007, the rate of dissolution for whatever reason decreases for the more recent unions, as the period of exposure to the risk of dissolution is shorter.

Figure 19

Aggregate number of PACS unions dissolved by end 2017, by reason and year of PACS registration (%)

Aggregate number of PACS unions dissolved by end 2017, by reason and year of PACS registration (%)

Interpretation: Out of 100 PACS unions registered in 1999, 25.8 were dissolved by mutual consent before 1 November 2017.
Note: This figure illustrates the four most common reasons for dissolution. Others (e.g. by the request of a legal guardian) are recorded but remain quite rare.
Coverage: All PACS unions registered before 1 January 2017, situation on 1 November 2017.

Box 2. PACS data published by the Ministry of Justice

The French Ministry of Justice has developed an exhaustive database of all PACS unions registered and dissolved in the courts from 1999. It includes all the 2.1 million contracts registered in the courts and by notaries up to 1 November 2017. Accessed via the Secure Access Data Centre (CASD) to guarantee confidentiality, it contains the following information for each contract registered since 1999:

• date and place of contract registration (court seized);
• partners’ municipality of residence;
• sex and date of birth of each partner; and
• date and reason for PACS dissolution (if applicable).

Adjustments have been made for information not collected before 2007 (mention of the PACS in the partners’ civil records, notably their sex).

82For all PACS cohorts, analysis of dissolutions by sex reveals a higher dissolution rate (for all reasons other than marriage) for PACS unions between two women (Table 16) and, to a lesser extent, between two men. Various studies in European countries and in the United States (Wiik et al., 2014; Bennett, 2017) have already highlighted the greater fragility of these same-sex unions (civil partnerships or civil unions), particularly those between two women (Marteau, 2019).

83Moreover, the partners’ age (both under 35 vs. both over 35 at the time of PACS registration) is a determinant of dissolution risk for same-sex couples. Whatever the year of PACS registration, there are at least twice as many dissolutions of same-sex PACS unions when the partners are under 35 at entry into union as when they are over 35. Here again, these findings mirror those of certain Scandinavian countries (Bennett, 2017), which show a strong age gradient in the dissolution risk of same-sex unions. This age effect is not observed for heterosexual PACS unions, whose dissolution rate remains similar at all ages.

84It is worthwhile to examine differences in the stability of PACS unions and marriages. To make this comparison, we assume that PACS unions dissolved by marriage are not breakups but simply a change in the type of officialization preferred by the partners (Table 16). [38] This ‘change’, which was not available to same-sex PACS partners until 2013, accounts for a large share of dissolved unions, notably among couples who entered a PACS many years ago (between 2005 and 2013) and at young ages in the case of lesbian couples. A quarter of different-sex PACS partners in 2005–2007 were married 10 years later, and this proportion increases to 29% when both partners were under 35 when they entered the PACS union.

Table 16

Proportion (%) of dissolved PACS unions (on 1 November 2017) by year of registration, age, and sex of partners

Share of PACS unions dissolved by Share of PACS unions dissolved for Distribution of PACS marriage reasons other than marriage unions PACS cohort Both Both Both Both partners partners Partners partners partners Partners Both under Both over under over of all ages under over of all ages age 35 age 35 age 35 age 35 age 35 age 35 PACS union between 2 men 1999–2004 11.4 25.5 18.8 63.4 23.8 42.1 32.6 41.2 2005–2007 14.0 21.7 17.7 49.2 19.8 34.6 31.1 42.1 2008–2010 17.0 19.3 17.4 40.9 17.5 28.7 29.7 45.9 2011–2013 15.3 13.4 14.0 31.2 14.1 23.2 34.8 40.0 2014–2016 5.6 4.1 5.0 15.4 7.6 12.3 41.6 34.8 PACS union between 2 women 1999–2004 12.6 19.1 16.6 71.9 30.5 49.4 34.7 47.7 2005–2007 22.5 20.4 21.3 55.3 23.6 39.6 37.4 42.3 2008–2010 25.2 17.8 21.7 47.6 24.0 36.1 37.8 44.0 2011–2013 26.3 14.3 21.4 38.2 17.3 28.8 44.2 38.2 2014–2016 11.4 4.9 9.4 21.1 9.5 16.5 51.2 33.0 Different-sex PACS 1999–2004 18.9 15.9 17.8 34.2 32.9 34.8 70.2 17.0 2005–2007 28.9 16.4 25.3 26.5 26.4 27.0 67.2 18.8 2008–2010 26.1 13.9 21.9 21.3 18.7 20.8 60.3 26.0 2011–2013 19.2 10.0 16.1 17.0 14.5 16.4 60.8 25.9 2014–2016 7.2 4.3 6.3 7.9 6.4 7.5 60.0 26.4

Proportion (%) of dissolved PACS unions (on 1 November 2017) by year of registration, age, and sex of partners

Interpretation: 11.4% of PACS unions concluded in 1999–2004 between two men aged under 35 were dissolved by a marriage before 1 November 2017.
Note: The partners’ ages are those recorded at the time of PACS registration.
Coverage: Whole of France, situation on 1/11/2017 of unions registered between 15/11/1999 and 31/12/2016.

85With the divorce rate observed in the marriage cohorts (Appendix Table A.10B) and taking account of the PACS durations reached in 2017 by the different PACS cohorts (15 years on average for the 1999–2004 cohorts, 11 years for those of 2005–2007, 8 years for those of 2008–2010, etc.), the proportion of PACS unions dissolved for reasons other than marriage or death exceeds the rate of divorces at an equivalent marriage duration.

86These differences in the outcomes of PACS unions confirm the atypical profiles of the earliest PACS partners, as already highlighted in previous studies with regard to educational level and occupational category (Bailly and Rault, 2013). People who entered a PACS in the early days (before 2005), notably the youngest among them, and independently of the partner’s sex, married their PACS partner less often than those entering a PACS after 2005. This choice may reflect a stronger distaste for the institution of marriage. The PACS outcome thus varies considerably by the partners’ age and sex. Older gay couples more often dissolved their PACS to marry, while different-sex couples more often married at a young age, shortly after concluding a PACS union.

87Among those who dissolved their PACS after age 60, death is the most frequent reason for dissolution. In 2016, this reason accounted for 10% to 13% of dissolutions in this age group, depending on the partners’ sex (Figure 20). For different-sex couples, marriage is the most frequent reason for dissolving a PACS union at ages 25–35. For same-sex couples, the rate of dissolution for reasons of marriage increases with age. For both gay and lesbian couples, 64% of the PACS unions that end after age 60 are dissolved for reasons of marriage. Conversely, for PACS unions dissolved at a young age, dissolution by mutual consent is the most frequent reason among same-sex couples, but only among the under-25s for different-sex couples. No more than 3% of all PACS unions are dissolved by unilateral request of one partner.

Figure 20

Distribution (%) of people who dissolved a PACS union in 2016 by sex, age at dissolution, and reason

Distribution (%) of people who dissolved a PACS union in 2016 by sex, age at dissolution, and reason

Interpretation: 69% of men aged 18–24 in a different-sex PACS union who dissolved their union in 2016 did so by mutual consent.
Coverage: PACS unions dissolved in 2016.

4 – Unions in France and Europe

Fewer, later marriages in France

88The decline in marriage and the changing partnership behaviours observed in France are part of a well-documented overall trend that began in Northern Europe in the 1970s (Sardon, 2006; Sobotka and Toulemon, 2008; Avdeev et al., 2011). The timing of these changes varied across European countries, leading to large disparities in first-marriage rates in the 1980s and 1990s (Sardon, 1992), with the Scandinavian model at one extreme and the Mediterranean model at the other (Prioux, 2006a). With marriage in sharp decline (decrease in the share of persons ever-married at age 50, ever-later age at marriage, widespread non-marital cohabitation), France is moving closer to the Scandinavian model.

89Recent first-marriage indicators (Table 17) reveal a contrast between the countries of Southern and Western Europe, on the one hand, and of Eastern Europe, on the other. In 2017, the total period first-marriage rate does not exceed 0.5 first marriages per person in France (for men and women alike), and age at marriage is quite high in France, Spain, Italy, the United Kingdom, the Benelux countries, and Slovenia. In these countries, age at first marriage is over 30 years for women and 33 for men. France is moving away from the countries of Northern Europe, however, where total period first-marriage rates have increased in recent years (0.57 in Sweden and 0.65 in Denmark). The countries of Eastern Europe are very distinctive, with a relatively high total period first-marriage rate and an age at first marriage that remains low despite a clear trend towards postponement.

90In some countries, the total period first-marriage rate is very high (Latvia, Romania, and Slovakia) and in others very low (Luxembourg), depending on the direction of migration flows. Marital mobility enables immigrant couples to marry in their home country, while other couples prefer to marry abroad (lately in Greece, around 10% of marriages concern non-residents, on the Greek islands especially). These phenomena have a clear impact on total period first-marriage rates based on marriages registered in the country of residence, pushing them up in high-emigration countries and down in high-immigration countries. The rapid changes in marriage timing and the sharp increase in age at first marriage from the 1990s in countries where people traditionally married young (Eastern Europe) may also explain the recent increase in the total period first-marriage rate through a catch-up process of younger cohorts marrying at later ages. A longitudinal analysis of first marriages would be needed to confirm these hypotheses.

91Although difficult to verify, the existence of the PACS since 1999 in France may have exacerbated the marriage decline by enabling couples to choose an alternative way to formalize their union. Alongside France, the Netherlands is the only other European country where all couples, including different-sex couples, can choose between marriage and civil partnership (known as a registered partnership in the Netherlands) to formalize their union.

Table 17

First-marriage indicators in EU countries in 2010 and 2017

Total first marriage rate Mean age at first marriage (years) Country Men Women Men Women 2010* 2017** 2010* 2017** 2010* 2017** 2010* 2017** Belgium 0.42 0.42 0.44 0.43 31.7 33.5 29.4 31.2 Bulgaria 0.38 0.53 0.42 0.62 29.8 30.6 26.6 27.3 Czech Republic 0.41 0.54 0.46 0.60 30.8 31.8 27.9 29.1 Denmark 0.66 0.65 0.68 0.66 33.6 34.8 31.2 32.4 Germany 0.54 0.58 0.57 0.62 32.9 34.0 30.2 31.2 Estonia 0.38 0.48 0.38 0.54 30.5 32.2 28.0 29.7 Ireland — 0.60 — 0.59 33.1 33.8 31.3 32.0 Greece 0.58 0.65 0.67 0.70 32.6 33.3 29.3 30.3 Spain 0.38 0.44 0.42 0.48 33.1 35.4 30.9 33.2 France 0.49 0.47 0.49 0.46 32.8 34.4 30.7 32.2 Croatia 0.63 0.67 0.67 0.72 30.0 31.0 27.1 28.2 Italy 0.48 0.45 0.53 0.48 33.2 35.0 30.3 32.2 Cyprus — — — — — — — — Latvia 0.39 0.70 0.41 0.76 29.4 31.5 27.1 28.9 Lithuania 0.61 0.86 0.62 0.90 28.7 30.5 26.5 27.8 Luxembourg 0.35 0.31 0.39 0.34 32.8 34.0 30.2 31.6 Hungary 0.36 0.59 0.39 0.65 31.2 32.3 28.3 29.4 Malta 0.78 — 0.84 — 31.3 — 28.6 — Netherlands 0.52 0.45 0.53 0.45 32.4 33.9 29.8 31.4 Austria 0.48 0.58 0.50 0.61 — 34.2 — 31.5 Poland 0.64 0.59 0.68 0.64 28.4 29.6 26.1 27.3 Portugal 0.43 0.42 0.47 0.44 29.9 32.5 27.7 30.7 Romania 0.70 0.93 0.77 1.05 29.0 31.0 25.6 27.5 Slovenia 0.37 0.41 0.42 0.47 31.7 32.9 29.0 30.4 Slovakia 0.48 0.65 0.52 0.72 30.0 31.2 27.2 28.6 Finland 0.64 0.51 0.70 0.53 32.5 33.9 30.2 31.6 Sweden 0.59 0.57 0.66 0.58 35.6 36.6 32.7 33.8 United Kingdom 0.46 0.49 0.50 0.50 — 33.4 — 31.5

First-marriage indicators in EU countries in 2010 and 2017

* 2011 data for Germany and Austria, 2008 data for the United Kingdom.
** 2016 data for Estonia, and 2015 data for Ireland, France, and the United Kingdom.

A high rate of same-sex marriages in France

92With 3.1% of marriages between partners of the same sex in 2017, the rate of same-sex marriages in France is quite high relative to other European countries where same-sex marriage is legal. [39] The proportion ranges between around 1% and 4% of all marriages (Figure 21), and same-sex marriages are generally most frequent in the years following legalization, when existing couples are able to formalize a long-standing union, except in certain countries (Spain and Portugal) where the share has increased steadily over the years (Cortina et al., 2013). The high level in France is all the more remarkable given that a choice between marriage and PACS, which also provides considerable protection and benefits, is available to all couples wishing to formalize their union.

Figure 21

Share (%) of same-sex marriages among total marriages in selected European countries

Share (%) of same-sex marriages among total marriages in selected European countries

93In most countries, notably those of Northern Europe, lesbian couples account for more than half of the same-sex couples who marry. This is not yet the case in France, although their share is tending to increase (from 41.5% in 2013 to 49.8% in 2017). A similar increase has been observed in Spain and the Netherlands. In Finland, Denmark, and Sweden, a large majority of same-sex marriages are between women, encouraged perhaps by these countries’ more progressive gender norms and more egalitarian gender relations.

A relatively high rate of divorce in France

94Up to the early 2000s, divorce increased more slowly in France than in other European countries, and levels were below those observed in the pioneer countries of Northern Europe (Sweden and Denmark), the Baltic countries, and the United Kingdom, where the total period divorce rate [40] exceeded 40 divorces per 100 marriages from the early 2000s (Prioux, 2006b; Sardon, 2006).

95Because the annual distribution of divorces by marriage duration is not known for all countries, we estimate here a total divorce rate for each country as the ratio of the mean number of divorces (averaged over 3 years) to the mean number of marriages registered in the country between 14 and 10 years previously (divorce occurs at a marriage duration of around 12 to 15 years on average). [41] With a more appropriate denominator, this indicator is more meaningful than a crude divorce rate, even if the migration of married couples is liable to introduce bias.

96In recent years, divorce rates have been generally high in France but remain lower than those of the Northern European and Baltic countries, despite a decrease in the total divorce rate between 2008 and 2016 in many of these countries where divorce was very frequent (Figure 22). The rankings remain relatively unchanged, however, even though divorce has increased rapidly in some countries where it was once rare, such as Spain and Portugal. While it would be overstating the case to talk about a convergence of divorce behaviours, major legislative changes in countries traditionally opposed to divorce are leading to a rapid increase in divorce numbers. This is the case in certain countries of Southern Europe (Croatia and Slovenia) and notably in Italy, where the divorce rate has increased from 18% to 35% within a few years in response to two successive reforms in 2014 and 2015 (simplification of procedures and shorter waiting time between separation and divorce). A similar law was voted in Ireland in 2019, and its effects may well be the same. But the trend is towards a decrease in divorce in most other countries, including France.

Figure 22

Divorce indicator in the European Union, 2008 and 2016

Divorce indicator in the European Union, 2008 and 2016

Note: Ratio of mean number of divorces in years t – 1 to t + 1 to the mean number of marriages in years t – 14 to t – 10.

1 – Characteristics of mortality

A slower increase in life expectancy

99While life expectancy at birth rose between 2017 and 2018, reaching 79.5 years for males and 85.4 years for females in metropolitan France (79.4 and 85.3 years for the whole of France), the rate of increase has slowed recently. Independently of the excess mortality observed in 2015 due to a severe flu epidemic, [43] life expectancy has increased more slowly in the last decade than in the previous one, for females especially. In metropolitan France, life expectancy at birth rose by 1.9 years for males and 1.0 years for females between 2008 and 2018 versus 2.8 years and 2.0 years between 1998 and 2008. To understand the reasons for this slowdown, we analysed the pattern of gains in life expectancy at birth by age group using INSEE’s 3-year life tables, the most recent of which corresponds to the period 2015–2017 (Figure 24).

Figure 23

Total annual deaths in metropolitan France, 1946–2018

Total annual deaths in metropolitan France, 1946–2018

Coverage: Metropolitan France.

100Before age 45, gains were halved for both sexes between 2005–2007 and 2015–2017 compared to the previous decade (0.48 vs. 0.87 years for males and 0.23 vs. 0.42 years for females). Beyond age 45, gains were identical for males over both periods but fell from 1.5 to 1.0 years for females. As a result, France is not performing as well as its European neighbours in terms of premature mortality, despite its high ranking for life expectancy at birth.

2 – France well placed in Europe for life expectancy

101In 2017, the most recent year for which comparative data are available, France still ranked among the top third of European countries in life expectancy at birth for both sexes combined (Appendix Table A.12). Its position has not changed notably over the last 15 years. It remains a European leader for female life expectancy at birth (85.2 years), alongside Italy and just behind Switzerland (85.6) and Spain (86.1).

Figure 24

Decomposition of life expectancy gains by age group for males and females

Decomposition of life expectancy gains by age group for males and females

Coverage: Metropolitan France.

102France is less well placed for males, ranking 9th among the 28 EU members in 2019 (Figure 25). With a male life expectancy at birth of 79.4 years, it trails Switzerland (the European leader, which holds the record at 81.6) and is outranked by several countries of Western Europe (Ireland, Luxembourg, the Netherlands, and Switzerland), Northern Europe (Iceland, Norway, and Sweden), and even Southern Europe (Spain and Italy), a region that had long lagged behind. However, like Austria and the United Kingdom, France ranks above all the countries of Eastern Europe and even above several other economically prosperous countries such as Belgium, Denmark, and Germany.

High premature mortality

103Analysis of mortality by age group shows that France is well placed for survival at older ages but is below the mean for mortality before age 65. It performs poorly for mortality of children under 15, holding 19th place for boys and 20th place for girls in 2017 (out of 28 countries ranked by increasing order of probability of dying in each age group). While France ranked 7th in 2005 (6th for girls, 8th for boys), it has gradually lost ground over the last decade or so, due mainly to a lack of progress in infant mortality. While deaths before age 1 have continued to fall in other European countries, they have stagnated in France. The infant mortality rate was estimated at 3.9 per 1,000 in 2017 for the whole of France (3.6 per 1,000 in metropolitan France), while in 11 other EU-28 countries, excluding Eastern Europe and with the notable exception of the United Kingdom, the rate was below 3.0 per 1,000.

Figure 25

Male and female life expectancy at birth in France and in other countries of the European Union, 1980–2017

Male and female life expectancy at birth in France and in other countries of the European Union, 1980–2017

Coverage: European Union members in 2019 (including the United Kingdom).

104France ranks slightly better for mortality at ages 15–24, 25–44, and 45–64 but is below the median (14th, 17th, and 15th positions for men; and 15th, 16th, and 12th for women) (Table 18). Its position has been relatively stable since 1980, except in female mortality at ages 45–64, which has declined much more slowly than in many other European countries. In the first half of the 2000s, France still ranked 5th or 6th for the female probability of dying in late middle age but has now fallen to 12th place (15th for men in 2000 and in 2017, with little fluctuation over the period).

Table 18

Probability of dying and life expectancy at age 85 in EU-28 countries, males and females, 2017

Males Probability of dying (per 1,000) Life Country Before At ages At ages At ages At ages At ages expectancy age 1 1–14 15–24 25–44 45–64 65–84 at age 85 Bulgaria 6.9 3.7 6.9 37.0 242.9 747.2 4.6 Croatia 4.7 2.2 5.4 22.3 172.4 687.1 5.0 Cyprus 1.3 1.5 4.0 13.8 92.5 522.4 5.3 Hungary 3.9 2.2 4.6 26.6 244.5 729.9 5.2 Slovakia 5.2 2.6 6.3 28.7 198.4 700.0 5.3 Lithuania 2.9 3.2 7.6 63.7 250.2 724.7 5.1 Latvia 4.5 2.8 9.8 64.3 269.7 743.7 5.2 Romania 7.6 3.5 7.2 35.8 248.1 727.8 5.5 Czech Republic 3.1 1.7 5.4 22.9 152.1 651.4 5.3 Greece 3.4 1.7 5.1 20.0 127.1 546.0 6.6 Netherlands 3.9 1.6 3.4 14.1 88.4 531.9 5.8 Austria 3.0 1.7 4.2 16.7 107.0 534.2 6.1 Denmark 4.2 1.4 3.6 14.9 110.1 551.6 5.7 Germany 3.5 1.4 3.5 17.5 121.0 558.7 5.9 Sweden 2.4 1.2 4.7 17.6 78.7 499.4 5.7 Poland 4.6 1.8 6.7 36.9 204.6 665.4 5.9 United Kingdom 4.2 1.5 3.8 22.0 103.0 529.7 6.2 Portugal 3.0 1.8 3.8 20.7 131.8 545.9 5.7 Estonia 2.7 1.8 6.3 41.5 198.2 681.0 5.8 Ireland 2.8 1.1 3.4 17.1 86.4 529.0 6.3 Slovenia 2.0 0.9 3.9 17.6 129.0 588.6 5.7 Malta 8.0 1.2 4.3 15.2 84.7 508.2 5.9 Finland 2.1 1.4 6.3 23.9 112.6 544.1 6.1 Belgium 3.9 2.0 4.2 20.1 108.7 537.3 6.1 Italy 2.8 1.4 3.4 13.5 85.5 506.3 6.1 Luxembourg 4.8 1.2 4.7 13.0 91.6 528.0 6.5 Spain 2.9 1.3 2.8 13.5 101.3 504.9 6.6 France 4.3 1.7 4.6 22.2 121.9 483.8 6.7

Probability of dying and life expectancy at age 85 in EU-28 countries, males and females, 2017

Females Probability of dying (per 1,000) Life Country Before At ages At ages At ages At ages At ages expectancy age 1 1–14 15–24 25–44 45–64 65–84 at age 85 Bulgaria 5.8 2.1 3.1 17.7 112.6 582.9 5.2 Croatia 3.3 2.1 1.5 10.2 77.0 524.6 5.6 Cyprus 1.3 2.4 1.0 6.6 47.4 367.4 5.6 Hungary 3.0 2.1 1.7 13.3 117.1 549.6 5.9 Slovakia 3.8 1.8 2.1 12.6 87.6 517.2 5.9 Lithuania 2.8 1.5 2.8 21.1 92.6 486.0 6.0 Latvia 3.5 1.7 3.9 20.8 103.4 512.2 6.0 Romania 6.0 2.8 2.6 15.8 106.0 568.6 6.0 Czech Republic 2.2 1.2 2.1 9.9 72.2 472.2 6.1 Greece 3.3 1.5 1.5 8.1 57.0 390.0 6.4 Netherlands 3.1 1.2 1.6 9.1 66.7 398.3 6.6 Austria 2.9 1.2 1.6 8.6 58.3 378.6 6.6 Denmark 3.3 0.8 1.4 8.5 68.4 423.1 6.7 Germany 3.0 1.3 1.7 9.2 67.1 397.3 6.7 Sweden 2.3 1.1 1.8 8.4 54.0 386.3 6.8 Poland 3.7 1.5 2.3 11.9 86.7 459.0 6.9 United Kingdom 3.4 1.2 1.8 12.4 68.3 412.0 6.9 Portugal 2.2 1.3 1.9 10.2 55.1 358.7 7.0 Estonia 1.8 2.0 3.0 11.6 76.8 437.2 7.0 Ireland 3.2 1.0 1.5 8.9 56.8 396.8 7.0 Slovenia 2.1 1.1 2.5 7.5 64.3 394.1 7.2 Malta 4.6 1.4 1.2 6.6 58.6 367.0 7.3 Finland 1.8 1.4 2.8 10.0 56.7 368.9 7.3 Belgium 3.2 1.3 2.0 10.3 67.0 376.4 7.4 Italy 2.6 1.1 1.4 7.4 49.7 349.2 7.4 Luxembourg 1.7 1.6 1.1 8.1 57.4 390.6 7.7 Spain 2.4 1.2 1.4 7.3 49.3 314.2 7.9 France 3.8 1.3 1.8 10.1 61.5 305.0 8.2

Note: In order of countries’ life expectancy of women.
Coverage: EU-28 in September 2017.

A narrowing of the life expectancy gender gap

106The gender gap in life expectancy (5.8 years in 2017) is still wider in France than in most other European countries except Croatia (6.1 years), Portugal (6.2 years), and those of Eastern Europe, where it ranges between 5.9 years in the Czech Republic and 9.9 years in Latvia (Appendix Table A.12). After reaching more than 8 years between 1976 and 1996, the gap has narrowed since the mid-1990s, however, due to a slowing of progress in female life expectancy (Meslé, 2006). Compared with 1992, when the gender gap was at its widest (8.3 years), men can expect to live 6.3 years longer and women just 4.0 years longer. Changes in the mortality gender gap by age across successive 10-year periods reveal a converging trend (Figure 26).

Figure 26

Excess male mortality by age in 1995–1997, 2005–2007, and 2015–2017

Excess male mortality by age in 1995–1997, 2005–2007, and 2015–2017

Coverage: Metropolitan France.

107Female mortality is lower than male mortality at all ages, but the gender gap varies considerably over the lifespan. For the two periods (1995–1997 to 2005–2007 and 2005–2007 to 2015–2017), the female advantage is especially pronounced at young adult ages (with a male excess mortality risk of more than 3.0 around age 25). It is also high around ages 60–65, with a ratio of more than 2.0 between ages 55 and 70. The gender gap is small, on the other hand, at the beginning and end of life, with a ratio of less than 1.2 in the first 2–3 years of life and after age 95. The ratio of probabilities has nonetheless decreased at all ages below 80, especially between ages 35 and 70 (when the difference between the curves corresponding to 1995–1997 and 2015–2017 in Figure 26 is largest).

108The narrowing of the life expectancy gender gap is due primarily to slower progress in reducing cancer mortality among women than among men. This trend is linked to gender differences in smoking behaviour. While the proportion of male smokers has been falling since the 1970s, it increased steadily among women until the 2010s. This narrowing is also due to a decline in deaths from external causes (road traffic accidents, suicides, and homicides), which has mainly benefited men, whose mortality rates from these causes were previously much higher than those of women (Breton et al., 2018).

3 – A stagnation of infant mortality in France

109As mentioned earlier, France is well down in the European rankings for child mortality, and this poor performance is due mainly to infant mortality (before age 1), which accounts for 65%–70% of overall mortality before age 15. Over the last 12 years or so, the probability of dying before age 1 has fluctuated around 3.7 per 1,000 births in France (3.5 per 1,000 in metropolitan France). In the early 2000s, the infant mortality rate in France was around 25% below the EU-28 average, but by 2017 it was slightly above, most other countries having achieved further progress in this area (Figure 27). In the Scandinavian countries (Finland, Iceland, Norway, and Sweden), where infant mortality rates are lowest, the probability of dying in the first year of life fluctuates around 2.0–2.5 per 1,000 births (Appendix Table A.13).

110The lack of progress concerns all components of infant mortality to varying degrees (Figure 28; Appendix Table A.11). After falling to a record low in 2005, at 1.55 per 1,000 births in metropolitan France, [44] early neonatal mortality (death in the first week of life) has increased steadily, reaching 1.85 per 1,000 in 2017. Late neonatal mortality (death in the 3 following weeks) has remained stable, at around 0.80 per 1,000, since 2000. Post-neonatal mortality (from the fifth week to the first birthday) fell progressively to a low of 3.3 per 1,000 in 2011 but has been rising steadily, reaching 3.6 per 1,000 in 2017. The increased contribution of mortality in the first week of life to total infant mortality (50% in 2017) is the result of these distinct trends that have not been studied in detail and whose determinants remain poorly understood (Papon, 2018).

Figure 27

Infant mortality rate in France and other EU-28 countries

Infant mortality rate in France and other EU-28 countries

Coverage: EU-28 in 2019 (including the United Kingdom).

4 – Causes of death in France: a comparative perspective

111The following analyses are based on World Health Organization data for the distribution of deaths by cause, and on Eurostat data for all-cause mortality rates. For France, INSERM (CépiDc) and INSEE data are also used. As long time series of mortality by cause are not available for the overseas departments, only metropolitan France is covered here. The comparison covers 2000–2015, a period for which we could obtain the necessary information for most EU-28 countries apart from Cyprus, Luxembourg, Portugal, and Slovakia. To facilitate comparison, deaths from ill-defined causes were redistributed proportionately for each country by year, sex, and age group. The proportion of deaths from ill-defined causes ranges from 0% to 11% (the highest figure is recorded in France), with a median at 2.2% and an interquartile difference of 2.7%.

Very low cardiovascular mortality in France

112Cardiovascular disease, the leading cause of death in Europe, accounts for 45% of all-cause mortality (50% for women, 40% for men), with three-quarters of cardiovascular mortality attributable to heart diseases. Mortality from this cause in France is particularly low compared to its European neighbours.

Figure 28

Trends in infant mortality and its components in metropolitan France since 1980

Trends in infant mortality and its components in metropolitan France since 1980

Coverage: Metropolitan France.

113While the mean rate was 500 deaths per 100,000 population for Europe as a whole in 2015, France recorded a level of just 220 per 100,000. France is slightly ahead of Spain for men, but more markedly so for women, with a rate of 180 per 100,000 in France versus 215 per 100,000 in Spain. Mortality from ischaemic heart disease is particularly low in France, with a rate 4 times below the European average in 2015.

114In France, as elsewhere (Ouellette et al., 2014), the decrease in cardiovascular mortality has been the main factor driving progress in life expectancy at birth over the last 50 years. Over the period 2000–2015, the decline in deaths from these diseases accounted for 35% of total life expectancy gains in France for men and 51% for women, i.e. one-third of the 4.0 years gained by men and half of the 2.5 years gained by women (Appendix Table A.14). The age-standardized mortality rate from cardiovascular diseases fell by 40% over this period for men and women alike. Because of this steep decline, cardiovascular diseases have been overtaken by cancers as the leading cause of death in France. But despite the steady decrease in smoking in most European countries, progress in the fight against cancer is slowing in Europe, perhaps due to the increased prevalence of risk factors such as diabetes and obesity (Wilkins et al., 2017).