Overview

This project focused on investigating the relationship between fertility and marriage within an economic framework. A number of intriguing patterns have been identified in marriage and birth hazard rates by age from French Census and Vital Statistics reports for 1900-1940. For instance the hazard rate for marriage, defined as as the number of annual marriages per single person in a given age group, attain spectacular peaks in the years following World War I, for both sexes. This is surprising because: 1) the male-female sex ratio is far below normal in the mobilized birth cohorts after the war, so we should expect female marriage rates to be lower than normal, not higher, and; 2) the deviation of young men's marriage rates above normal persists into the 1930s, when the men in question were far too young to have been directly affected by the war.

The hypothesis is that the disruption of marriage markets that occurs during war (the marriage bust) has large and long-lasting effects on the composition of the singles pool, and that the unusual composition of post-war single pools explains the deviations in marriage hazards. As evidence for this, two further facts are relied upon: 1) the ratio of marriage rates post/pre war is increasing in age for women, but flat for men, and; 2) the birth hazard for married women by age is higher after the war, and remains so well into the 1930s.

A model of marriage and births has been designed to formalize this approach. In this model, the fundamental variation across women is the propensity to have births. The propensity to have a birth is a function of underlying fecundability and tastes. The propensity for births will influence a woman's willingness to marry and (implicitly) to use contraception or abortion. Women enter into the model as singles with high fecundability, who differ in their preferences for children; each year they age stochastically, hence their average fecundability declines, so that eventually they become infertile. This allows us to model long life-cycles using annual data, as compared to the two-period approach common in the literature.

Each year, single men and women participate in a marriage-matching process with search frictions; this means that each year there are men and women who would have been happy to marry each other but did not because they did not meet. Marriage rates are determined by the gains from marriage, these will increase according to the expected number of children produced by the marriage, so high-propensity women will marry at a higher rate because their marriages produce more children.