Educational Saving Incentives for Low-Income Families : Experimental Evidence from the Michigan SEED Program

Low-income American families save very little in general, and especially for their children's education. This paper presents estimates for low-income families of educational saving incentives on college saving decisions. The evidence is drawn from the results of a field experiment in Pontiac, MI, known as the Michigan SEED Program. Qualified families were interviewed at baseline, when their children were roughly 4 years old, then were randomly assigned to treatment and control groups. The treatment group was offered an initial deposit of $1,000 in the child's name into the Michigan Education Savings Plan, the state-sponsored 529 college savings plan. Then, for three years, deposits by the treatment group were eligible for a 1:1 match up to a specified maximum. During the course of the experiment, treatment-group families could have accumulated $3,400, enough to purchase roughly two-thirds of the tuition cost of an associate's degree at the local junior college. Take-up was low: after four years, only 67% of children from treatment-group families had accepted the offer of the initial deposit. Only 22% of treatment-group families contributed their own funds, and, among these, the average family contribution was $440, or $9.16 per month over the course of the experiment. This study uses the random assignment of treatment status to generate a plausibly exogenous instrumental variable for 529 plan participation and identify consistent estimates of the causal effects of 529 plans on college saving. The instrumental variable estimates indicate 55% crowd-out. Therefore, much of the attempt to “seed” educational saving through the initial deposit and matching contributions was undone though reductions in other (non-529) forms of saving for college