The impact of divorce laws on the equilibrium in the marriage market

This paper investigates how the adoption of unilateral divorce affects the gains from marriage and who marries whom. Exploiting variation in the timing of adoption across the US states, I first show that unilateral divorce increases assortative matching among newlyweds. To explain the link between divorce laws and matching patterns, I specify an equilibrium model of household formation, labor supply, private and public consumption, and divorce over the life cycle. Matching decisions depend on the anticipated welfare from marriage and divorce. The model has two key features (consistent with the data). First, working spouses whose partners do not work accumulate relatively more human capital during their lifetime, a fact that improves their outside value of divorce. Second, divorcees cannot sustain cooperation in public goods expenditures (interpreted as children’s welfare), leading to inefficiencies that are mostly harmful to the top educated. Under unilateral divorce, the value of divorce becomes a credible threat that shifts the bargaining power in marriage, making both household production and marriage less attractive. This pushes the marriage market equilibrium towards more positive sorting in education and lower welfare, particularly for the highest educated. I estimate the model using data from households that form and live under the pre-reform mutual consent divorce regime. Using the estimates, I then introduce unilateral divorce and solve for the new equilibrium. I find sizable equilibrium effects. First, the correlation in spousal education increases and people, particularly educated females, become more likely to remain single. Second, the gains from marriage decrease for the least and the most educated. Lastly, the marital gains from acquiring a college or higher degree decreases for women and men under unilateral divorce. These results reflect previously overlooked consequences of reducing barriers to divorce. ∗Department of Economics, Yale University, 37 Hillhouse Av., New Haven, CT 06511, [email protected]. I am extremely grateful to my advisors Joseph Altonji, Costas Meghir, Larry Samuelson, and Chris Udry for their guidance and support. I am also grateful to Pierre André Chiappori, Ilse Lindenlaub, Chiara Margaria, Cormac O’Dea, Esteban Peralta, Meredith Startz, and seminar participants at Yale for helpful comments, discussions, and suggestions. All errors are mine.