Skip over navigation

Journal Issue: The Next Generation of Antipoverty Policies Volume 17 Number 2 Fall 2007

Reducing Poverty through Preschool Interventions
Greg J. Duncan Jens Ludwig Katherine A. Magnuson

Potential Criticisms of the Proposal

It is only natural that taxpayers who are being asked to contribute $20 billion in new funding for our proposal would want to be sure that the program will accomplish its stated goals. We address here several of the more obvious doubts.

Is the program too expensive? Do we really need to require—and pay—college-educated teachers and insist on such small classes? The evidence on the effects of particular early childhood programs tells more about the net effect of these programs than about which program elements matter most for their success. Many—though clearly not all—successful programs involve highly educated teachers and small classes. These findings are consistent both with social science research that finds, for example, better life outcomes for children of highly educated mothers than for those whose mothers have less schooling or lower cognitive test scores and with research from class-size experiments studying slightly older children, in kindergarten through third grade.43 Even with the expense of highly educated teachers and small class sizes, this intensive intervention is still likely to pass a benefit-cost test quite easily.

Is program intensity too low? How do we know that half a day of educational instruction is enough? We have no direct evidence on whether a full-day early education program will yield larger or more lasting effects than a part-day program. Indirect evidence on the effects of full-day and part-day kindergarten may be informative, however. Studies have found that although students in full-day kindergarten programs learn more during the kindergarten year than students in part-day programs, the differential gains are relatively small and do not persist much beyond that year.44 Researchers have pointed out that additional time in the classroom does not necessarily translate into greater exposure to enriching opportunities, and thus it is important to know how programs structure children’s “extra” time.45 With these considerations in mind, we propose a part-day preschool program.

Why wait until age three to provide educational services, given that disparities in cognitive and noncognitive skills are apparent in even younger children? We certainly agree that the period between conception and age three is vital for children’s healthy development. 46 We choose to concentrate on ages three and four for several reasons. First, model programs such as Perry Preschool have generated lasting effects for poor children by waiting until ages three and four to provide services. Second, although Abecedarian started even earlier than age three and achieved more pronounced and longer-lasting effects on outcomes such as IQ scores, it cost twice or three times as much as our proposed program (in part because class sizes for very young children in Abecedarian were about half of what we propose for our intervention). Yet we are far from certain that if the proposed program spent that extra money by starting earlier, effects would be commensurately greater. Third, many low-income families seem wary of sending very young children to center-based care, so starting our early childhood education services at age three is likely to enhance take-up rates and fit better with the preferences of our target population. For example data from the Department of Health and Human Services show that around 43 percent of eligible low-income children from birth to age two received federal child care subsidies, compared with 56 percent for poor children aged three to five.47 Nevertheless, we agree that targeted health care and child maltreatment interventions are clearly warranted at an early age, although more work is needed to better understand what types of health and parent services are beneficial for particular populations.

Wouldn’t spending $20 billion more a year on income transfer programs do more to reduce poverty in the United States? The reason for devoting scarce resources to preschool education rather than to income transfer programs rests with cost-benefit analysis. It is true that income transfer programs will do more to reduce poverty than preschool interventions, in part because prevention programs such as those we propose here are imperfectly targeted. Although poor children are disproportionately likely to become poor adults, many of tomorrow’s poor adults come from families that are not poor today.48

In contrast, income transfer programs are by definition directly targeted toward those people who wind up poor during adulthood, and provide cash assistance precisely during the periods when people need it. However, income transfer programs are usually a zerosum game in a benefit-cost analysis, since they merely transfer resources from one group in society (taxpayers) to another (poor families). Consequently, the extent to which they are cost-effective depends on the benefits that accumulate to children from their parents’ enhanced income. Although recent research has found that work-conditioned income transfers do improve children’s achievement, the effects are relatively small.49 In contrast, the sort of preschool education program that we propose here reduces poverty by making children more productive when they grow up, as well as more likely to engage in pro-social activities, such as work, and less likely to engage in antisocial behaviors, such as crime, that impose substantial costs on society. And in fact, as noted, we believe that the benefits to society generated by our proposed program are likely to be as much as four times the program costs.

Finally, is this a targeted or a universal program? Although our program is available to all three- and four-year-olds, we recommend publicly subsidizing its cost only for families with incomes less than three times the federal poverty threshold. Thus, access may be universal, but we are targeting public support to relatively economically disadvantaged children. Some researchers are concerned that targeted programs are both less efficient and less popular than universal programs.50 But the “target” of our public funding is much broader than that of most targeted programs. More than 60 percent of all three-and four-year-olds would be eligible to receive some public support to attend our proposed early education program. Moreover, all children would be able to attend, even if they do not receive subsidies. One other concern about targeted benefits is that they create a disincentive for parents who qualify to increase their earnings. For this reason, we have recommended a sliding scale for subsidies to avoid penalizing a family with the loss of a valuable benefit when its earnings exceed the benefit eligibility level by just a little. But we also note that because a child would be attending the program for at most two years, any labor disincentives would be short lived.