Married couples who began their relationship by cohabiting appear to face an increased risk of marital dissolution, which may be due to self-selection of more dissolution-prone individuals into cohabitation before marriage. This paper uses newly developed econometric methods to explicitly address the endogeneity of cohabitation before marriage in the hazard of marital disruption by allowing the unobserved heterogeneity components to be correlated across the decisions to cohabit and to end a marriage. These methods are applied to data from the National Longitudinal Study of the High School Class of 1972. We find significant heterogeneity in both cohabitation and marriage disruption, and discover evidence of self-selection into cohabitation.

Data from the National Longitudinal Study of the High School Class of 1972 are used to estimate a series of models of entry into marriage, entry into cohabitation, and nonmarital pregnancy. Our models account explicitly for the endogeneity of one outcome as a predictor of another by taking into account both heterogeneity across individuals due to unmeasured factors that may affect all these outcomes and the correlation in the unmeasured factors across processes. We find that these heterogeneity components are strongly and positively related across the outcomes. Women who are more likely to cohabit, marry, or become pregnant while unmarried are also more likely to do each of the others. Although black and white women differ in the likelihood of these behaviors, the interrelations of the behaviors are quite similar across groups.

The objective of this research is to further our understanding of how and why individuals enter and leave coresidential relationships. We develop and estimate an economic model of nonmarital cohabitation, marriage, and divorce that is consistent with current data on the formation and dissolution of relationships. Jovanovic's (Journal of Political Economy 87 (1979), 972-90) theoretical matching model is extended to help explain household formation and dissolution behavior. Implications of the model reveal what factors influence the decision to start a relationship, what form this relationship will take, and the relative stability of the various types of unions. The structural parameters of the model are estimated using longitudinal data from a sample of female high school seniors from the United States. New numerical methods are developed to reduce computational costs associated with estimation. The empirical results have interesting interpretations given the structural model. They show that a significant cause of cohabitation is the need to learn about potential partners and to hedge against future bad shocks. The estimated parameters are used to conduct several comparative dynamic experiments. For example, we show that policy experiments changing the cost of divorce have little effect on relationship choices.