A Proposal for State, Income-Targeted, Preschool Vouchers

Abstract

This article proposes that states create low-income targeted voucher programs for 3- and/or 4-year-olds. The basis for this proposal is considerable research demonstrating long-term effects for a number of random assignment and quasi-experimental preschool programs. Benefit–costs rates of return for these programs are between $2 and $16 for every dollar invested. These results are also consistent with developmental and neuro-biological studies supporting early childhood interventions. Other research shows that low-income students are less likely to be in preschool, and if they are, they are more likely to receive low-quality services. Taken together, and to control costs, I propose an income-targeted program. I argue for vouchers based on improved equal opportunity and higher efficiency in the delivery of preschool services. Design details are generally to be the province of state decision makers. My one caveat is that it seems to make sense to require programs, at least for 4-year-olds, to be center based.

Notes

¹An exception is a recent summary chapter by Robert Haveman and Barbara Wolfe (2001) that reviews studies of social returns rather than market or income returns.

²There are a number of cautions in this literature, especially in the manner in which data were collected. They often rely on wage and salary surveys, usually of firms. This creates considerable inconsistency across countries. However, Psacharopoulos and Patrinos argued that these problems have lessened in recent years. In addition, although some modeling issues remain, the major issue of selection bias seems to not matter as much as first thought in that ordinary least square and instrumental variable estimates often produce quite similar estimates. If anything the instrumental variable estimates indicate slightly greater returns.

³As with the worldwide data, using more elaborate estimations than simple ordinary least square specifications does not seem to shift this rate very much.

⁴Mathematica is currently evaluating a random assignment, early Head Start experiment that should provide a larger scale random assignment experiment than currently exists.

⁵A Google search of the Perry Preschool Project came up with more than 630,000 cites.

∗ = Statistically significant at P > .05.

⁶The noneffect on crime is worrisome in that it was a major benefit of both the Perry program and the CCPC program.

^a Percentages do not add to 100 because children participate in more than one form of preschool childcare.

^a Black includes African American and Hispanic includes Latino. Race categories exclude Hispanic origin unless specified. Included in the total, but not shown separately, are children from other racial/ethnic groups.

^b “Poor” is defined to include those families below the poverty threshold; “nonpoor” is defined as 200% or more than the poverty threshold.

⁷That program is important for this proposal. In 2002 Florida voters amended the constitution to mandate required provision of education for 4-year-olds. In 2005, Florida enacted a preschool voucher program.

⁸Named after early 20th century Congressman James Blaire who, fearing public funding of Catholic schools, toured the country imploring states to amend their constitutions to prohibit funding of religious institutions, especially schools.