This dissertation studies the effect of premarital cohabitation on women’s propensity
to divorce and women’s divorce risk. The dissertation focuses on various selection
issues in a woman’s decision to start a relationship and the form of the relationship,
and to dissolve or continue the relationship. The 1995, 2002, and 2006-2010 waves of
the National Survey of Family and Growth (NSFG) are used. The sample sizes are
10,847, 7,643, and 12,279 women aged 15 to 44 living in households in the United
States in 1995, 2002, and 2006-2010, respectively. The dissertation finds that premarital
cohabitation has no effect on women’s propensity to divorce and women’s divorce
risk. Sorting into marriage according to level of religiosity generates the variations
among the same levels of religiosity between cohabiting and non-cohabiting couples.
The variations can explain the positive correlation between premarital cohabitation
and divorce. In absence of the information on both the wife’s and the husband’s
level of religiosity, this dissertation suggests and estimates a selection model. Three
different empirical approaches all suggest that living together before marriage does
not lead to divorce. Findings of positive correlation between premarital cohabitation
and divorce in the previous literature could be attributed to omitted variable bias or
sample selection bias.
Overall the rising in cohabitation cannot explain the rise and the fallen in divorce
rates. However, the prevalence of cohabitation will induce a thick ’cohabitation’
market, and hence improve the quality of cohabiting matches through lowering search
costs and increasing the turnover rates of cohabitation. As a result, the correlation
between premarital cohabitation and divorce is expected to negative and significant
on the basis of newly released survey data such as the 2011-2015 wave of the NSFG.