Once Homogamous, Always Homogamous? Meeting in School, Educational Homogamy and Partners’ Careers over the Life Course in France

Milan Bouchet-Valat, French National Institute for Demographic Studies (Ined)
Sébastien Grobon, French National Institute for Statistics (Insee)

Following the theory of educational systems as marriage markets (Blossfeld & Timm, 2003), it is often claimed that educational expansion favors an increase in educational homogamy, which in turn fosters increased occupational similarity in a context of growing female labor force participation, and eventually contributes to rising income inequalities between households. We put to the test this sequence of hypotheses using unique data from a recent French survey on individual and conjugal trajectories (EPIC).

First, even if the prolongation of studies has increased the proportion of first relationships in which partners met in school, this change is much more limited among current relationships at the time of the survey due to separations and repartnerings. Second, relationships initiated in school are more educationally homogamous than others, as expected by the theory. Third, couples in which partners have the same educational level have more similar occupations both at the beginning of the relationship and at the time of the survey. But the effect is quite weak, implying that an increase in educational homogamy has very limited consequences on socioeconomic homogamy, and therefore on income inequalities between couples. We show that gendered differences in the partners’ occupational careers are essential to take into account to understand this result.

Overall, while effects go in the direction expected by the theory at each step, only the second step (meeting in school increases educational homogamy) corresponds to a strong effect. The combination of all three steps implies a very limited effect of longer studies on socioeconomic homogamy over the life course. Our results confirm recent works on the relationship between educational homogamy and income inequalities, and indicate that other demographic determinants of increases in inequalities need to be investigated.


Educational expansion has often been considered as a factor of increased educational assortative mating, following notably H.-P. Blossfeld and A. Timm (2003)’s theory of “educational systems as marriage markets”. Stronger resemblance between partners in educational terms could in turn foster increased occupational similarity in a context of growing female labor force participation, and eventually contribute to rising income inequalities between households.

This theory is consistent with evidence from the United States, where educational homogamy has increased over time (Schwartz and Mare, 2005 ; Mare, 2016), though its consequences on income inequalities are disputed (Breen and Salazar, 2011). But it doesn’t square well with evidence from European countries, were educational homogamy did not clearly increase despite educational expansion – or even decreased in some countries, including France (e.g. Ultee and Luijkx, 1990 ; Birkelund and Heldal, 2003 ; Halpin and Chan, 2003 ; Bouchet-Valat, 2014).

We investigate how educational expansion can go hand in hand with a decrease in homogamy by putting to the test each of the mechanisms which should supposedly create a positive association between these two phenomena according to the theory of educational systems as marriage markets. To this end, we use recent data from the Survey on Individual and Conjugal Trajectories (EPIC) conducted by the French National Institute for Demographic Studies (Ined) in 2013-2014, and which offers exceptional information on meeting context, on successive conjugal histories and on characteristics of partners at different time points. This survey covers cohorts from 1948 to 1988 with a sample size of about 8,000 individuals (and 15,000 relationships), and constitutes in our view a more adequate source than those used by most existing studies.

Three steps are analyzed successively. First, has the proportion of couples in which partners met in school or before the end of the partners’ studies increased over cohorts, both in first and current unions? Second, are these couples more educationally homogamous than others? Third, does a higher educational homogamy lead to closer labor market returns over the life course?

We first observe that the proportion of first couples in which partners have met in school indeed increased significantly over cohorts: it went from 10% for cohorts born in the 1950s to about 30% for cohorts born in the 1980s. The proportion of first unions formed before the end of the respondent’s studies even went from 20% to almost 70%. Yet, these dramatic changes are only partially reflected when looking at unions at the time of the survey. Because of the increase in of separations and repartnerings, first unions constitute an ever decreasing proportion of all unions at a given point in time: considering all age groups together, only 45% of all unions are first unions in 2013-2014. All in all, at the time of the survey, the proportion of unions in which partners met in school increased only from 8% in for cohorts born in the 1950s to 12% for those born in the 1990s. The proportion of unions formed before the end of studies increased from 10% to 30% over the same cohorts.

We confirm in the second part of the analysis that couples who met in school are more homogamous than others: in 50% of these couples, the partners have the same educational level, while this is the case for only 35% of all couples. Despite this, the absolute difference of 15 percentage points does not appear large enough to have major consequences on the composition of couples. Considering that partners met in school in only 12% of all unions, this gap cannot account for a significant increase in educational homogamy. Finally, we observe that couples who met in school are less homogamous in terms of social class of origin, which goes against the picture of a generalized reinforcement of boundaries between social groups.

The third part of our analysis shows that, at the aggregate level, partners’ occupational positions are more similar in educationally homogamous couples than in other couples. This difference, measured using an occupational prestige score, is visible both at the beginning of the relationship (when both partners already had a job) and at the time of the survey, but it is quite weak: educational homogamy is only weakly linked to occupational homogamy.

Overall, our study confirms, using more adequate data than previously available, that educational systems can act as marriage markets. But this mechanism only has very limited effects on educational assortative mating in unions at a given point in time. This explains how educational homogamy can have decreased over time in France despite a strong educational expansion: trends towards increased social openness have been much stronger. We also stress that educational homogamy can have an equalizing effect on gender differences within couples, a phenomenon which should more often be acknowledged when considering these issues.

Presented in Session 1234: Posters