The Effects of Cohort Size on Marriage Markets in Twentieth Century Sweden

Large, short-run fluctuations in the birth rate have been an important demographic feature of industrialized, low- fertility populations in the twentieth century. Since females normally marry men who are two or three years older than themselves, these fluctuations result in large imbalances between the size of male and female cohorts who would normally marry each other. These imbalances must somehow be resolved, either by a change in traditional patterns of age at marriage or by changes in the proportions of the population of one sex or the other who ever marry. The authors of this paper have developed an implementable general equilibrium model of marriage assignments, which can be used to predict the way in which marriage patterns adjust to change in the numbers of males and females in each cohort. In this model, the efficient assignment of marriage partners across birth cohorts can be interpreted as a linear programming assignment problem. In the paper, they examine Swedish marriage patterns during the twentieth century in order to test their model. Dramatic fluctuations in fertility in Sweden have created large differences between the size of male and female cohorts who would normally marry each other. Nevertheless, the median age difference between husbands and wives has remained in the range of two or three years. Using the linear program to find the optimal set of marriage assignments between chorts, the authors are able to almost perfectly replicate the rapid fluctuations in the age difference between spouses actually observed in Sweden for the cohorts born between 1915 and 1925. This is the period of the most dramatic fluctuations in cohort size in recent Swedish history. The success of their model in replicating actual Swedish marriage patterns during this period indicates that they have captured fundamental features of the way the marriage market responds to changes in the relative supplies of males and females. While the model was less successful at tracking the steady decline in the age difference between spouses beginning with the late 1920's cohorts, the authors explain what adjustments need to be made in order to do this more accutrately.