Summarizing the Research: Evidence from a Statewide Randomized Experiment Testing CDA Policy

A growing number of studies examine the relationships between children’s savings accounts and their educational and financial outcomes. (Please see the Center for Social Development’s website or read this review article for more information on the studies examining these relationships). Consistently, children who have savings accounts early in life have significantly better educational and financial outcomes compared to children who do not have savings accounts early in life. These findings hold true even after taking into consideration things like academic achievement, household income, and household net worth.

These findings have led researchers and policy makers to articulate policy innovations like Children's Development Accounts (CDAs), which would establish savings accounts for every newborn in the U.S. In policy terms, this means that CDAs are intended to be universal. Countries like Singapore and the United Kingdom have already implemented CDA policies, suggesting that they can be viable policy innovations.

However, very few studies test the efficacy of CDA policies themselves. The SEED OK study, a statewide experiment that randomly assigned 529 college savings plans to participating newborns across the state of Oklahoma, is one of the first applied tests of CDA policies in the U.S. The results from the first 18 months of SEED OK have recently been made available as a paper on the Center for Social Development’s website. The results described in the paper are quite comprehensive; however, the results are condensed in this blog post to provide a summary of the main findings.

Researchers undertook the SEED OK study to determine whether CDAs could be successfully implemented (e.g., can universal savings accounts be opened for newborns), the extent to which participants saved, how much assets participants accumulated, and whether CDAs, including savings accounts and amounts saved, affected parents' attitudes and behaviors and children's outcomes.

Study Design

The SEED OK study began with newborns born during 2008-2009. Out of 7,115 eligible newborns, 2,704 parents and caregivers agreed to participate. Participants were randomly assigned into two groups: the treatment group (1,340) and the control group (1,330). These numbers exclude participants who were not from Oklahoma, who had missing data, or who were not parents or caregivers of SEED OK newborns. Random assignment means that participants in both the treatment and control groups should be roughly equal based on characteristics that are observable, like income and education level, and characteristics that are unobservable, like attitudes toward saving. The only thing that should be different between the two groups is receipt of treatment.

In this case, the treatment group received a few things that the control group did not. The participants in the treatment group automatically received a state-owned 529 college savings plan with a $1,000 deposit, were encouraged to open another participant-owned 529 college savings plan with a time-limited incentive of $100 deposited into the account and matched deposits based on income eligibility, and told that anyone could open another private college savings account without financial incentives. Participants from the control group could have opened a 529 college savings plan or other private college savings account on their own. However, by nature of being in the control group, these participants were not told they could open college savings accounts and were not provided any information or financial incentives.

Researchers examined (1) the total assets held in the state-owned 529 college savings plans, (2) whether another participant-owned 529 college savings plan was opened and the assets held in these accounts, (3) whether another private college savings account was opened and the assets held in these accounts, and (4) all three types of accounts combined and the assets held in these combined accounts.

What Researchers Found

Participants in the treatment group—nearly 100% of whom agreed to open state-owned 529s—saved significantly more compared to the control group. The total assets accumulated in the state-owned 529 college savings plans were 996 times the assets accumulated by the control group, after taking demographic characteristics into consideration. This means that SEED OK incentives from being in the treatment group accounted for almost all the assets accumulated in the state-owned 529 college savings plans. (Remember, the accounts were automatically opened with a $1,000 deposit).

Participants in the treatment group opened participant-owned 529s and accumulated significantly more assets compared to the control group. The probability of opening another participant-owned 529 college savings plan was 20 percentage points higher for treatment group participants compared to control group participants after taking demographic characteristics into consideration. The amount saved in participant-owned 529s was 1.4 times more for the treatment group compared to the control group.

There is no significant difference in opening another private college savings account or the amount saved in these accounts between treatment and control groups. Since these accounts did not have any financial incentives attached to them, the non-significant difference between treatment and control groups was expected.

Overall, participants in the treatment group opened significantly more accounts and accumulated significantly more assets compared to the control group. After taking demographic characteristics into consideration, the probability of having accounts was 95 percentage points higher for the treatment group compared with the control group. The treatment group also accumulated 1.3 times more assets compared with the control group.

What Does this Mean?

The results from the SEED OK study suggest that participating in the treatment group—that is, having a 529 college savings plan automatically opened at birth with an initial $1,000 deposit, being encouraged to open a participant-owned account with financial incentives, and opening private college savings accounts—is related to significantly better savings outcomes 18 months later. This means that a universal CDA policy in the U.S. may be possible. Also, almost all of the participants in the treatment group opened state-owned 529s, suggesting interest in a universal CDA policy.

Things to Keep in Mind

In applied social sciences, the SEED OK study was rigorous by most measures. However, there are a few things we should keep in mind. First, the results in the paper are based on the first 18 months of a study intended to examine lifelong impacts of CDAs on educational outcomes and asset building. In other words, these results are a first glimpse into the potential impacts of CDAs. It will be a few more years before we know, for example, whether CDAs significantly impacted educational outcomes like math and reading scores or college attendance. Second, asset building research suggests that savings accounts have a cumulative effect: once participants start saving in basic or college savings accounts, they may become familiar with and gain access to other types of mainstream financial products like stocks, bonds, mutual funds, and IRAs. However, the results are not able to tell us whether SEED OK participation leads to overall household asset diversification and accumulation. Third, we should keep in mind the context in which the SEED OK study was undertaken. SEED OK began in 2008-2009 when the U.S. was experiencing an economic recession, the impacts of which are not measured in this study. Fourth, SEED OK is an extension of a previous research initiative that piloted savings programs at approximately 10 locations across the U.S., called SEED, which stands for Saving, Education, Entrepreneurship and Downpayment. Please see the suggested reading for more information about the original SEED initiative.