Universal Prekindergarten, Fertility, and Maternal Health: A Difference-in-Regression-Discontinuities Design with Extrapolations
Posted: 12 Oct 2017 Last revised: 11 Mar 2019
Date Written: November 21, 2017
Prior research suggests that high quality prekindergarten (pre-K) programs can promote school readiness and generate lifetime benefits for children. However, very little is known about how pre-K programs affect other family members. In this study, I examine the effects of the New York City pre-K program on the health of low-income mothers, their healthcare utilization, fertility, and well-being, by linking Medicaid data to 2014 program participants, the year the universal pre-K program was launched in NYC. I use a difference-in-regression-discontinuities design that exploits both the introduction of universal pre-K and the birthdate cutoff for enrollment, in order to address potential bias from the steep development trajectory found in even small age ranges in early childhood, as well as potential preexisting confounding policies that use the same age cutoff to determine eligibility. I find no significant effects from having a child who was born just inside the eligibility cutoff, versus a child who was born just outside the eligibility cutoff, except a small negative effect on the likelihood of prescriptions for those without younger children. In addition, I extend the regression discontinuity design extrapolation methods, based on local derivatives and matching on observables, to the difference-in-regression-discontinuities framework, in order to estimate the average effects on the entire pre-K and non-pre-K population, a year on either side of the cutoff date. The extrapolation results differ little from those at the cutoff. Additional extrapolations done in small age ranges indicate that the effects of universal pre-K on mothers are heterogeneous by children’s date of birth.
Keywords: Prekindergarten, Maternal Health, Regression Discontinuity Design, Early Childhood, Medicaid
JEL Classification: C21, H51, H52, I18, I28, J13
Suggested Citation: Suggested Citation