Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)



Committee Chair

Victoria Prowse

Committee Co-Chair

Timothy Bond

Committee Member 1

Kevin Mumford

Committee Member 2

John Barron


This dissertation is composed of three independent chapters in the field of family and public policies. The first chapter is a reassessment of the impact of Unilateral Divorce Laws on marriage dissolution. The second chapter studies the effects of a widespread negative incentive policy on teen behavioral and health outcomes. The third chapter studies the impact of post-secondary college education quality on marriage outcomes.

The first chapter evaluates the impact of unilateral divorce laws (UDLs) on the risk of divorce via two distinct channels: the effect on divorce of married couples (divorce effect), and the effect on divorce through initial marital sorting (sorting effect). The divorce effect affects the divorce probability of all married couples, while the sorting effect is only experienced by couples that married after the implementation of UDLs. I use differences in the timing of states’ enactment of UDLs as a source of exogenous treatment variation in a difference-in-differences approach. Using the Divorce and Marriage file of Vital Statistics from the NBER data collection, I find that UDLs have a profound impact on marital sorting. The sorting effect significantly increases the cumulative risk of divorce within any length of marriage, while the divorce effect is only significantly associated with increasing the risk of divorce within the first 9 years and has no effect on subsequent years. Moreover, unstable marriages dissolve faster due to changes in marital sorting. The sorting effect increases the risk of divorce in each of the first 5 years of marriage by 4.5%, while there is zero divorce effect for these same years. 31% of the initial increase in the overall divorce rate identified in previous studies is due to the sorting effect.

The second chapter studies how No Pass, No Drive policies affect teen health and labor outcomes. Using difference-in-differences estimation, we identify the causal effect of these policies on teen behavioral outcomes – teen births, traffic fatalities, and teen employment. We find that the largest group of NPND polices cause a small increase in teen births, with large effects on black teen births (6 percent) and his-panic teen births (21 percent). These policies also cause a reduction in teen-involved fatal traffic accidents by 3 percent, with larger effects for white teens (15 percent). This is caused by the revocation of driver’s licenses, as the reduction in fatal traffic accidents is constant during and outside of school hours, and is larger (12 percent overall) in later years as enforcement technology improved. NPND polices cause a shift in teen employment – overall teen employment rates are not affected, but white teen employment decreases due to a 1 percentage point increase in white school enrollment. This is offset by a 5 percentage point increase in black teen employment, where black teens fill the new job openings created by white teens enrolling in school. Our results demonstrate that increasing school retention and decreasing teen drivers’ licenses reduce fatal traffic accidents and increase teen births, and we provide suggestive evidence of existing racial frictions in youth labor markets, which can be reduced by lowering dropout rates.

The third chapter investigates the impact of post-secondary college education quality on marriage outcomes. Using NLSY79 data, an instrumental variable strategy is employed to identify the causal effect. College quality is measured as the average SAT score of the entering freshman class of the degree-granting school. A better college education is associated with stabler marriages, later first marriages, and a lower likelihood of remarriages. I also find that it has no impact on the probability of marriage. Among people who obtained a college degree before age 25, a 100-point increase in SAT score leads to an increase in the age at first marriage by 4 years. In addition, people with higher quality of college education are 25 and 23 percentage points less likely to get divorced and divorce before 40, respectively, than those with a 100-point lower quality of college education. Additionally, better-educated people have 26% fewer number of marriages than their counterparts. Since the number of college graduates have been growing up over time, this article provides a new perspective to understand the trends of marriage and divorce.