The SEED for Oklahoma Kids (SEED OK) experiment is a large-scale study of universal and progressive Child Development Accounts, with random selection from the full population of newborn children in the state. SEED OK aims to investigate the policy innovation of giving every child an account at birth, and to test whether the initiative has an impact on ownership of accounts, savings for children from all sources, family attitudes and behaviors, and child development outcomes. Study participants (N=2,704) were randomly assigned to treatment and control groups. Using an automatic enrollment (or “opt out”) strategy, SEED OK opened a state-owned 529 college savings plan account for every treatment child, with an intial deposit of $1,000. In addition, treatment families were encouraged to open their own Oklahoma 529 accounts, and low- and moderate-income treatment families qualified for matches on deposits into these participant-owned accounts. 

In this study, we focus on early participation in 529 plans.  Existing research has shown that SEED OK increased 529 account ownership, individual 529 savings, and total 529 assets (including SEED OK subsidies and incentives).  Here, we examine these outcomes for a variety of socioeconomic subgroups, asking: (1) whether early participation varied by socioeconomic status (SES), and (2) whether the impact of the SEED OK treament varied by SES.  Data come from birth records, a baseline survey, and 529 account data.  We have multiple measures for each outcome of interest.  We examine these outcomes for subgroups categorized by income-poverty status, parent’s education, race/ethnicity, banked status, home ownership, financial assets, receipt of public assistance, and primary language.

Findings show that socioeconomically advantaged children had much more favorable outcomes than disadvantaged children.  The SEED OK treatment had a clear impact on some of the outcomes examined (e.g., whether a child had a 529 account in her name, whether parents had set aside college savings for these young children, whether a child had any 529 assets, and the amount of 529 assets in a child’s name.  SEED OK did not clearly increase the amount of individual 529 savings set aside for these children.  These patterns hold for diverse SES subgroups.  Even with the SEED OK treatment, advantaged groups tended to have more favorable outcomes, except when outcomes were directly related to automatic components of the treatment.  Automatic opening of state-owned 529 accounts eliminated virtually all variation by SES in 529 account holding, and automatic initial deposits eliminated most (but not all) variation by SES in 529 asset accumulation.  These outcomes were predictable, of course, but this does not make them meaningless.  If universality is a goal—that is, if we want children from disadvantaged families, not just children from advantaged families, to grow up with accounts and savings for post-secondary education—the evidence clearly favors automatic account opening and some automatic subsidies.