RESEARCH
Working Papers
The Effect of Abortion Legalization on Teenage Out-of-Wedlock Childbearing in Future Cohorts (Job Market Paper)
Abstract:
This paper examines the long-term impact of legalized abortion on teenage out-of-wedlock childbearing, which has been in constant decline since the early 1990s in the U. S. My argument is that, to the extent that it prevented unwanted births, legalized abortion could have reduced the likelihood of the teenage out-of-wedlock childbearing for the cohorts born after the legalization. This is analogous to the argument of Donahue and Levitt (2001) for crime but extends their analyses to a different context. I adopt a non-parametric approach that allows me to separately estimate the effects of the legal changes concerning abortion in the repeal states in 1970 and the Roe v. Wade ruling in 1973 on Whites and African-Americans. I find that for African-Americans, both changes lead to a long-term reduction in out-of-wedlock teenage childbearing. For Whites, there is no evidence supporting a long-term effect of the 1970 legalizations, but I find that teenage out-of-wedlock childbirth declined in the non-repeal states among the cohorts born after Roe v. Wade. My findings are consistent with Levine, Staiger, Kane, and Zimmerman (1999), who find that the early legalization in the repeal states had a much stronger effect on the immediate fertility of Non-Whites than Whites. My results also show that legalized abortion can potentially account for at least 30 percent of the decline in teenage out-of-wedlock childbearing among 15-17 years olds for African-Americans and 35 percent of this decline for Whites in the 1990s.
(with John C. Ham and Lara Shore-Sheppard)
Abstract:
Previous research on participation in Medicaid has estimated a constant treatment effect model where becoming eligible affects the take-up of each child in an identical fashion. This approach has two drawbacks if effects do differ across families. First, it provides no information about which groups have low take-up conditional on eligibility. Second, it may lead to misleading predictions of the effects of expanding eligibility if the newly eligible have different characteristics, and thus different take-up rates, from individuals who were previously eligible. We explore two approaches to allowing the effect of eligibility on participation in Medicaid to differ across demographic groups. The first incorporates interactions between eligibility and demographic variables in the now standard linear probability model approach, which does not require a distributional assumption. The second is based on a switching probit model and has the advantage of allowing newly eligible individuals to differ from those previously eligible in terms of both observed and unobserved characteristics. Using data from the Survey of Income and Program Participation and exploiting the exogenous policy changes in Medicaid eligibility that took place in the late 1980’s and early 1990’s, we estimate the average take-up rate of many demographic groups using these two models. We then measure how different demographic groups would respond to a hypothetical policy experiment of increasing eligibility in 1995 by raising the 1995 income limits by 10%. We find groups vary dramatically in terms of their average take-up rates and their response to the policy change; moreover we are able to quantify these differences. While our emphasis is on Medicaid participation, our approach is applicable to the crowding out of private insurance by public insurance, as well as to participation in other social programs.
Other Research in Progress
“How Did Legalized Abortion and Increase in the Availability of Oral Contraceptives Affect Single Motherhood in the 1970s?”