Child Development Accounts and Mother's Educational Expectations: Impacts From a Statewide Social Experiment
Method: SEED OK is a policy experiment with randomly selected newborn children in Oklahoma. Study participants (N = 2704) were assigned randomly to treatment and control groups after completing a baseline interview and before the SEED OK intervention began. SEED OK intervention was offered to treatment group only. State-owned Oklahoma 529 accounts with an initial deposit of $1,000 were opened automatically for the treatment children and savings matches were available for income-eligible households. Also, information on college savings and educational materials were occasionally sent to treatment group. We use data from 2007 birth records, the baseline SEED OK survey conducted in 2007–2008, and the follow-up SEED OK survey conducted in 2011. The dependent variable is change in mother’s educational expectations between two surveys. Mother’s educational expectation is asked at the two surveys respectively, “How far in school do you think that [your child] will go?” with five response categories from “won’t finish high school” (=1) to “will go to graduate school” (=5). To measure change in educational expectations over time, we calculate the difference and recoded into two categorical levels: “remain the same or increase” (=1), and “decrease” (=0). Independent variable is treatment group status, and covariates include characteristics of child, parents, and family. Using the experimental design, we run chi-square tests to evaluate the impacts of CDAs for the full sample and across socio-economically disadvantaged groups. The regression-adjusted treatment effects controlling for demographic and socioeconomic characteristics are also reported.
Results: A significantly higher proportion of treatment participants has maintained their level of expectations or increased to a higher level than control group. Positive impacts of CDAs are more apparent in socioeconomically disadvantaged subgroups, that is, welfare recipients, those without high school diploma, those with income below 200% poverty line, and American Indians. The results are consistent in regression-adjusted analyses and other sensitivity tests.
Conclusion: Our study provides evidence of CDAs’ impact on fostering mother’s educational expectations, particularly the disadvantaged groups who might be discouraged from keeping expectations high without the intervention. By supplementing the existing early childhood intervention programs focusing on education and care—for example Head Start—CDAs may encourage the development of an early identity toward education, promote access to education-related information, and direct the efforts of parents and their children toward later academic achievement and investment.