A rank-based measure of educational assortative mating

ABSTRACT

Measures of educational assortative mating must contend with the fact that the marginal distributions of education differ by sex and change over time. I propose a new measure of assortative mating, which is based on Spearman’s rank correlation coefficient but instead uses sex-by-birth cohort education ranks. I show that this adjusted measure is more robust than other measures to changes over time in the underlying education distributions.

KEYWORDS:

• Marriage market
• assortative mating

JEL CLASSIFICATION:

• J12
• J10

Acknowledgments

For many helpful comments, I thank David Green, Ilyana Kuziemko, Katie Shester, audiences at the Society of Labor Economists meetings and the Southern Economic Association meetings, and the anonymous reviewers of this journal.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

¹ In the 1950 census, education was only recorded for every fifth person, so we never observe the education of both members of a couple. Therefore, I begin my sample period in 1960.

² This is clearly not intended to be realistic – in practice, not everyone marries, and spouses are usually of similar ages but not exactly the same age – but it provides a clear setting to assess measures of assortative mating.

³ Repeating the exercise for the 11-category ‘linearized’ scheme yields a similar figure.