Is the French PACS Similar to Cohabitation or Marriage? Recently Merged Census and Tax Data on Family Situations and Transitions

Laurent Toulemon, Institut National d’Études Démographiques
Giulia Ferrari, INED

We have recently observed consistent changes in couple configurations in France, with a dramatic decline in marriage rates and a recent increase in Pacs (i.e., French civil partnerships) as a new form of legal union. Although marriages and Pacs are both occurring at older ages, age at first cohabiting union is stable. Unmarried unions are thus becoming more frequent, especially at young ages, but not only.

The French demographic panel conducted by the INSEE, the French Institute for Statistics, has recently be dramatically enlarged, with a sample rate increasing from 1% to 4%, and the inclusion of tax data, in addition of census data and vital statistics. As the census is now based on annual census surveys, 14% of the sample – as of the whole population – is included in the census each year. The census allows us distinguishing, among unmarried and unpacsed adults living with others, those who are living as a couple and those who are just sharing their dwellings with relatives, friends or roommates, while tax data inform about the situation in other years, based on the identification of the partner in the dwelling.

Applying multistate life table methods to these data, we aim at investigating the distribution and the evolution from one year to the next by age and sex of different couple situations; we show that unmarried cohabitation as well as Pacsed couples are likely to become more common. All ages 20-70 together, the proportion of wen and women living as a couple would remain constant around 65%. Among couples, the part of married couples could decline from 70% to 48%, while Pacs would double, from 7% to 15%. We will next further develop the analysis by extending the observation time, stratifying by socioeconomic strata and distinguishing between same-sex and different-sex couples.

Is the French PACS similar to cohabitation or marriage?Recently merged census and tax data on family situations and transitions

Note: this is an extract from the text of alonger abstract uploaded as a PDF Introduction

In recent years we have observed consistentchanges in couple configurations of men and women in France, with a dramaticdecline in marriage rates and a recent increase in Pacs (i.e., French civilpartnerships) as a new form of legal union (Mazuy et al. 2014). With respect tomarriage, Pacs is simpler to start and to end, and is less protective in termsof inheritance and reversion pension. In 2015, 235,000 marriages and 192,000Pacs were celebrated, with a very high total of 427,000 “legalized unions” (INSEE2016). As Pacs has become popular only recently, very few Pacs are transformedinto marriage but the current couple behaviors carry the potential for majorchanges.

Previous research has investigateddifferent couple situations, either based upon administrative data (thusmissing unmarried cohabitations) or survey data (thus suffering of attritionand memory bias). However, to our knowledge there is not yet any study lookingat the evolution of different couple situations in France (including cohabitingunions and Pacs).

In particular, we aim at answering toquestions such as: (1) What is the distribution by age and sex of differentcouple situations? (2) How many new couples are formed each year? (3) How manypeople transit to another couple situation from one year to the next? (4) Dodistributions and transition probabilities vary across socioeconomic strata?(5) How do same-sex couples differ from the others? Data and methods

In order to answer these questions, we takebenefit of the availability of the French Demographic Panel, Échantillondémographique permanent, EDP, conducted by the INSEE, the French Instituteof Statistics. For a 4% sample of all inhabitants of France, EDP is mergingdata from population censuses, vital events registration, salaries and, since2011, tax data (income and housing tax data, including all source of income). Taxdata allow identifying Pacsed and married couples, but does not distinguishbetween unrelated persons sharing the same dwelling ad unmarried couples. Basedon these data, we build multistate life tables based on the followingtransitions, between the states Singlehood (no parner), Cohabitation (apartner), Pacs, and Marriage. The transition probabilities are estimatedseparately for women and men for ages 15 to 70. We present here preliminaryresults for the years 2013-4.

  Preliminary results

On average, transitions from the state ofsingle or cohabitor to married or pacsed are less likely than transitions fromsingle to unmarried cohabitations. Further, cohabitations are also more likelyto end with respect to the other forms of union. Few Pacs are transformed intomarriages (less than 1% a year), however it is more likely than people getmarried if already in an unmarried couple (either pacsed or cohabiting) than ifsingle (i.e. direct marriage has declined). 

Figure 1 shows that age at marriage or Pacsis generally higher than age at cohabitation. Comparing the current situation(observed, obs.) and the distribution by sex and age deduced from themultistate life tables (estimates, est.) shows that the current transitions,would they remain stable, would lead to a further decrease in the prevalence ofmarried couples. The proportion of adult men or women living as a couple wouldnot change much (maybe slightly less single adults around age 40 and, for men,more singles after age 55). The decline in marriage would thus be compensatedfor by an ongoing increase in unmarried cohabitation and, at ages above 30, adramatic increase of the prevalence of Pacs. Of course, the transitions fromunmarried cohabitation to Pacs or marriage, as well as from Pacs to marriage,may change with legal and taxation rules, but our analysis shows that thecurrent situation is far from equilibrium: the proportion of married couples islikely to decrease in the future.

Additional analyses on the transitionprobabilities (to be presented in the full paper) show that marriageprobabilities among Pacsed couples are higher than among cohabitors, especiallyat young ages but also at ages above 45. Considering union disruptions (muchhigher for unmarried couples than for married couples), pacsed are much moresimilar to married couples.

Figure 1Observed and estimated distribution by sex and age of the couple situation in tand the transitions to t+1.

Note: weighted to French resident population

Source: census and tax data collected in2013 and 2014 Discussion and further steps

The EDP allows conducting many analyses. Wewill differentiate men’s and women’s behavior by level of education, in orderto see how the increasing propensity of highly educated women to live in aunion is related to the spread of Pacs and unmarried cohabitations.


Presented in Session 1108: Families and Households