Educational Homogamy and Income Inequality in European Countries

Positive assortative mating on education, or educational homogamy, means that individuals tend to couple, in marriage and/or cohabitation, with partners who have a similar level of education. Both homogamy and income inequality have increased in most developed countries over the past 30-40 years.  Homogamy is believed to contribute to rising inequality as highly-educated men and women increasingly opt to live together, are more likely both to be in gainful employment, and tend to have higher earnings than their less educated counterparts. However, the association may be more complicated than it seems at first sight. First, it is important to account for the degree of homogamy as well as the shares of low-, medium-, and highly-educated couples among all homogamous couples; i.e the effects of homogamy on income inequality may change over time due to changes in their distribution. Second, (employment) effects that can be attributed to individual educational attainment need to be distinguished from those that result directly from the individuals’ assortative mating. Furthermore, the actual pooling may be compensatory instead of complementary. For instance, higher educated individuals coupled with highly educated and well earning partners might in fact opt to reduce their hours and employment participation below what their educational attainment would lead us to expect. Third, by taking these two points into account, the relationship between homogamy and income inequality will be considered.

In the proposed paper we investigate the extent to which homogamy and income inequality can be related in European countries between 1994 and 2012. We emphasize the possible contribution of homogamy to rising income inequality as a result of pooling employment and earnings, and pay special attention to developments following the start of the financial crisis. For our analysis we focus on three questions:

(i)                 How important is educational homogamy?

In order to demonstrate the existence of homogamy in a cross-country (or within country over-time) perspective it is insufficient to compare factual shares of homogamous couples between countries, due to between-country differences in the underlying distributions of educational attainment of men and women. To account for this we develop a ‘Homogamy Index’ that controls for these underlying distributions and provides a standardized degree of homogamy on top of random educational pairing (the index gives 0% homogamy under random pairing and 100% when homogamy is at its potential maximum, i.e. when all couples are homogamous couples). The Index’s aggregate level informs about the international variation and the changes over time. Its division into (three) levels of educational attainment and broad birth cohorts serves to track and illustrate the contribution of increasing educational attainment, especially of women. For the general effect of homogamy on income inequality it is also important to account for the incidence of couples in the overall population, which may be under pressure from different and increasing levels of individualization.

Preliminary findings of the first part of this study suggest that most of the deviations from the random pairing distribution are estimated at being between 35% and 55%. This implies that a significant amount of the variation in the number of homogamous couples in Europe is the result of assortative mating. Further we see that the degree of homogamy is highest in the Southern countries and lowest in the Northern countries. Within countries the aggregate degree of homogamy is more or less constant over time for most countries. Dividing the Homogamy Index into the three educational levels (low, medium, and high) shows that it are specifically the low-educated homogamous couples that drive the high degree of homogamy in the Southern countries. Conversely, we find that in the Northern countries high- educated homogamous couples make up the highest share of the Homogamy Index.

(ii)               How does educational homogamy affect individual employment and earnings?

We analyse whether homogamy as such exerts a positive effect, independently from the individual attainment, on the probability of having a job, occupational level, pay level and annual hours of work.

Provisional results differ between countries but indicate that the level of educational attainment tends to be stronger determinant of being in employment than homogamy.

(iii)             To what extent does homogamy contribute to an increase in income inequality?

Even if homogamy does not directly affect individual employment participation and earnings, the pooling of resources could still enhance income inequality. However, this might depend on homogamy growing mainly amongst highly-educated couples. We will analyse the effect of the aggregate Homogamy Index as well as the effects of the separate components of the index (low, medium, and high) on income inequality. As a measure of income inequality we will primarily use the Theil Index, while checking against several other aggregate indicators as well as decompositions of the Theil Index. Regressing the Theil Index on a constant and on the Homogamy Index will identify income inequality under random pairing (measured by the constant) and income inequality as a result of homogamy (measured by the coefficient of the Homogamy Index). Adding relevant covariates (share of female employment, share of dual earners, etc.) and using other measures of inequality will test the robustness of the findings. Replacing the Homogamy Index by its separate components will investigate the effect of the composition of homogamy. Replacing the Theil Index by its decompositions (singles versus couples; within couples: single earner versus dual earners; within dual earners: both full-time employed versus one part-time employed) will show which part(s) of the distribution of income inequality (if any) are affected by homogamy. Finally, by adding the time component we will attempt to clarify dynamics, particularly those resulting from the recent financial crisis.     

The analysis uses data collected in the ECHP (European Community Household Panel) and the EU-SILC (European Union Statistics on Income and Living Conditions), two datasets based on similar questionnaires, well suited to cross-country analysis. Together these datasets cover the second half of the 1990s and the first twelve years of this century. The number of countries covered differs between the two datasets.