Universal Child Care: A Risky Experiment with Our Nation's Children
By Katharine B. Stevens | Jenet Erickson
ARTICLE
April 27, 2021
A growing chorus of advocates are vigorously pushing for a large expansion of U.S. child care as a “win-win-win” that supports women’s careers and boosts the economy while promoting children’s healthy development. Those who raise questions about the potentially negative effects on very young children are quickly reassured that high-quality, nonparental, group care only advances child well-being or, at worst, simply has no effect.
For example, in Elliot Haspel’s recent Family Studies piece, he urged any who worry about the possible negative effects of nonparental group care to “set aside” concerns because, he argued, high-quality child care “works well for all families” and has conclusively been shown to be harmless. He further maintained on Twitter that “98% of solid studies” confirm this view, describing the discussion of reported negative effects as “so out of touch with the research consensus as to be almost a caricature.”
In particular, he dismissed out of hand our recent Family Studies essay, suggesting we cherry-picked research to make our “preferred ideological point” that universal child care is likely to have adverse effects for some children. In fact, we have not cherrypicked research to make that point but are eager to dispel any doubt. Here we respond to the claims Haspel made in his IFS blog post, relying entirely on the research literature he cites: reports on investigations of child care programs in Chile, Germany, Norway, Quebec, and the United States.
And as we show, far from confirming a “research consensus” that high-quality child care is harmless, this literature underscores that no such consensus exists. In fact, a careful read of the “more recent and sophisticated research” Haspel cites to support his claims reveals that it contradicts most of them.
Positive for Some, Negative for Others
In his post, Haspel makes the common claim that high-quality child care “helps many children from lower-income backgrounds, while having little direct effect on the already advantaged path of middle- and high-income children.” Yet, the research he cites fails to support this assertion. Rather, it shows that the effects of high-quality child care are decidedly mixed: positive or neutral for some children, and negative for others. Felfe and Lalive, researchers for the German study he cites (here and here), underscore: “Understanding effect heterogeneity is essential because early childcare may both help and harm children” (italics added).
In fact, several of the researchers Haspel references found that effects vary not only between often-compared groups, such as lower and higher SES children, but also among children within a particular group. These researchers stress that typically reported findings on average effects, even for specific demographic groups, thus provide an inadequate basis for general conclusions, warning that heterogeneous effects of child care must be seriously considered in deciding whether and how to scale up public programs. In their 2017 working paper on Quebec’s child care program, for example, Kottelenberg and Lehrer summarize:
Despite traditional emphasis in the applied literature to report only mean effects of a policy, the existence of treatment effect heterogeneity in education programs is now overwhelming … policy changes generate both winners and losers and as such it is important to report distributional treatment effects in empirical work.
So what are the heterogeneous effects found by the research Haspel cites? Several common themes emerge.
Child Care Does Help Some Children
Consistent with Haspel’s claim, all five studies report that child care benefits some children from low socioeconomic backgrounds. Kottelenberg and Lehrer found that Quebec’s child care program had positive impacts for children from low-income, single-parent families: on motor and social development for children from birth through age 3 and vocabulary for 4- and 5-year-olds. In their study of a high-quality child care program serving low-income children in Chile, Noboa-Hildago and Urzuafound “significant positive” effects in emotional regulation.
Three studies investigated the effects of carefully controlled, intensive quality programs for infants and toddlers, specifically. Felfe and Lalive found the German program particularly benefitted children with less-educated mothers and with immigrant parents. Similarly, the Norwegian study, conducted by Drange and Havnes, concluded that early enrollment in an intensive-quality child care program benefitted language and mathematics at age seven among children from low-income families. In the U.S., Chaparro et al. analyzed data from the Infant Health and Development Program, an experimental intervention for babies born low birthweight and premature carried out in eight US cities in the late 1980s. They, too, found that for low SES children the program had large positive effects on IQ at age three.
However, researchers for all three studies emphasize that the extraordinarily high quality of the programs they investigated was crucial to the outcomes produced, underscoring that their findings cannot be extrapolated to child care more generally.
Child Care Harms Some Children
While the research Haspel cites is consistent with his claim that high quality child care can benefit low SES children, he seems to have overlooked the concurrent finding of some negative effects. Notably, adverse impacts were found even for low SES children.
For example, Felfe and Lalive report that despite the German program’s exceptionally high quality, a substantial program expansion resulted in “harm[ing] the language skill development of children from more-educated families.” In the Quebec study, Kottelenberg and Lehrer report adverse impacts for children of two-parent families across the income spectrum, finding especially negative effects on motor and social development for girls in the bottom and boys in the top income quintiles (birth through age three); and on vocabulary for 4- and 5-year-old boys in the bottom half and very top of the income distribution.
Noboa-Hildago and Urzua report “severe negative” effects on child-adult interactions along with “negative effects in the areas of reasoning and memory” for the low-income children enrolled in the Chilean program. “This detrimental impact may potentially undermine the benefits of center-based care,” they conclude, because “child-adult interactions significantly influence many other tasks in early childhood and later years.”
Such findings of adverse effects are in fact consistent with a large body of longitudinal research carried out in the U.S. by the National Institute of Child Health and Human Development (NICHD). One NICHD investigator whom Haspel cites later reported that by age eight, “more time spent in childcare was associated with lower social competence and poorer academic work habits” along with “more conflicted relationships with teachers and mothers.” Importantly, these negative effects were “independent of quality, quantity, and type of childcare.”
Of the investigations Haspel cites, only the Norwegian study and the U.S. analysis of the 1980s Infant Health and Development Program (IDHP) data reported no detrimental impact of child care. Yet the results of these two investigations shed little light on the potential effects of expanding publicly funded child care in the U.S. for three reasons:
First, neither study examined emotional or behavioral effects, which are precisely the outcome domains where negative effects have been found in other research.
Second, average attendance in the U.S. program was only part time—around 18 hours per week—while research has largely found negative effects associated with full-time attendance.
Third, the programs examined in these two studies were of unusually high quality. The Norwegian program, for example, included well-trained teachers, teacher-child ratios of 1 to 3, and maximum group sizes of 10—higher quality standards than any public program now operating in the United States. The IDHP program cost $17.61 per hour (in 2021 dollars): amounting to an annual cost of over $18,000 for an average of less than 20 hours per child per week. Researchers for both studies emphasize that the intensive quality of these programs were essential to the outcomes produced.
Overall, however, these reported findings—both positive and negative—represent only average effects on large numbers of children, obscuring what Kottelenberg and Lehrer describe as “substantial policy relevant heterogeneity” in the impact of high-quality child care on young children. The finding of adverse effects does not mean that child care harms all children. Similarly, the finding of positive or neutral effects does not “prove” that high-quality child care is always harmless.
Children’s Home Environments
What causes this widely observed heterogeneity in child care’s effects? Several studies Haspel cites suggest that variation in impact is driven by the quality of children’s home environments. “The causal effects of attending early care depend on the quality of maternal care,” Felfe and Lalive explain. Researchers report that children from unstable home environments can benefit from attending high-quality child care. But when child care displaces even higher quality parenting—or even diminishes parenting quality, as the Quebec study finds—it can have an adverse effect on children’s development.
Felfe and Lalive hypothesize, for example, that the negative effects they found for children from higher SES families occurred because the program “displace[d] time spent one-on-one with parents, particularly college-educated mothers. Kottelenberg and Lehrer, too, argue that while child care “may provide a remedy for children from the most disadvantaged home environments,” even high quality programs have an adverse effect on children when they displace higher quality parental care they would otherwise receive at home.
Consistent with this assessment, Kottelenberg and Lehrer identify “particularly striking” variation in the effects of Quebec’s child care program, based not on SES, but on whether children were living in one- or two-parent households. They found large developmental gains for lower SES children from single-parent families as child care “appears to substitute for lower levels of parental care or informal care arrangements.” But even low SES children from two-parent families were adversely affected: “Children from two-parent families between the 10th and 50th [income] quantile generally receive significant negative impacts,” the authors write, because child care is “less than a perfect substitute for investments that were previously made in the home.”
And, in contrast to Haspel’s assertion that child care only improves parenting, Kottelenberg and Lehrer report that its adverse effects for children from two-parent families occurred, in part, because the quality and quantity of time parents spent with their children declined. “We find large and statistically significant declines… on many parenting practices for two-parent families” associated with child care attendance, they write, although no such changes in parenting quality for single-parent households were found.
While child care appeared to provide additional resources for children from single-parent households, the authors summarize, children from two-parent households “shifted from receiving otherwise strong one-on-one parental care at home to reductions in many parent-child activities coupled with potentially less effective higher adult-child ratio caregiving away from home.” Thus, a “significant portion of the negative effects of subsidized child care on developmental outcomes derives from changes in how parents in two-parent families invest into their children,” they conclude.
In his Family Studies post, Haspel also emphasizes the suggestion made by Chaparro et al. that child care may improve parenting by reducing parental stress. However, he fails to note their specific finding that child care’s effects on the quality of maternal care “differ sharply” by maternal education. For low SES women, child care increased the quality of maternal care substantially. But among mothers with a bachelor’s degree or higher, it had no effect.
Moreover, Kottelenberg and Lehrer cite contrary research showing that participation in child care actually increases the likelihood that children are “subjected to working parents affected by potentially higher levels of stress.” Parental preference could partially explain these differential responses: Child care may reduce stress for parents who prefer fewer hours of parenting, while increasing it for parents who prefer parenting more. In other words, high-quality child care is not an intrinsic good, as Chaparro et al. show; rather, its value depends on the nature of the tradeoffs that families make between maternal and non-maternal care.
Lack of Research on the Effects on Infants and Toddlers
Haspel refers to a large body of research yielding the consensus finding that child care has no harmful effects on children. Yet, in addition to the findings of negative effects described above, of the five program investigations he cited, two explicitly note that very little research has been conducted on the effects of child care on infants and toddlers—clearly precluding any meaningful consensus on its effects for children under age three. Felfe and Lalive characterize existing research as “inconclusive,” noting that scholars are “hotly” debating “whether ECC helps or hinders children's development.” Drange and Havnes, too, emphasize that “evidence on how childcare affects the development of toddlers is largely missing,” describing researchers’ persistent concern that separation of infants and toddlers from a primary caregiver “may cause stress and anxiety… with potentially adverse effects on children's development.”
What About Universal Child Care?
Similarly, we know little about the effects of universal child care programs, specifically. As Drange and Havnes note, most current research on child care is on targeted and smaller-scale programs, while “the literature on universal programs is smaller and findings are mixed.” Furthermore, the scant existing research on universal programs provides insufficient information because, as discussed above, studies have focused on child care’s average causal effects, obscuring heterogeneity in its impact. Finally, as Felfe and Lalive show, the effects of universal care cannot be predicted by research on smaller programs because even demographically identical children added in an expansion differ in unobservable, but crucial ways from children who were already attending.
Using a marginal treatment effect (MTE) analysis, Felfe and Lalive found that benefits for lower SES children of even the exceptionally high quality German program diminished as the program expanded: A small increase in access to child care benefitted low SES children, while a large expansion had no significant effect. This occurred not because program quality declined, they explain, but because the expanded program included children whose parents had a weaker preference for enrolling them in a child care program. While child care “works particularly well for low-SES children whose parents have a high preference for sending their child to early care” (italics added), benefits were not seen in children whose parents were less motivated to put them in child care. “One cannot simply extrapolate findings from one context to another,” the researchers conclude, because “heterogeneity… with respect to unobserved characteristics matters.”
This finding also lends support to an argument for funding targeted rather than universal programs: providing truly high-quality child care to the children—and parents—whom it helps the most, while avoiding counterproductive public expenditures on programs that might harm others. Indeed, several authors emphasize targeted child care as the best approach. Drange and Havnes suggest that, “child care policies may be more effective if targeted at low income households." Kottelenberg and Lehrer, too, write that, “targeting child care coverage is a more effective policy option relative to universal coverage.”
No Evidence That High Quality Can Be Maintained At Scale
Finally, Noboa-Hildago and Urzua highlight the enormous challenge of maintaining quality at scale, especially for younger and lower SES children, writing: "Our study demonstrates the complexity associated with the design, implementation, and evaluation of public ECE interventions in developing as well as developed countries."
This is a warning we must heed very carefully because the U.S. has never maintained consistently high quality in a large-scale government-funded program, particularly for disadvantaged children. The K-12 public school system is a longstanding, free, universal program operating at a large, national scale. Yet its quality for lower SES children has been notoriously low for decades, despite high teacher standards, average teacher salaries well over twice those for child care workers (roughly $60,000 compared to $24,000 annually), and decades of steadily increasing spending. Similarly, Head Start, another nationwide, public program has failed to ensure high quality, even while enrolling just a small fraction of the children projected to attend universal child care programs.
Could we achieve in universal child care what we have been incapable of accomplishing in the nation’s public schools and Head Start? We have no credible reason to believe the answer to that question is yes.
A Risk Not Worth Taking
Last summer, several researchers published an article in Epigenomics discussing what they see as the growing conflict between science and an increasingly widespread use of child care—a conflict that remains unexamined, they argue, because of a “taboo on open debate.” Noting that “record numbers of mothers are working and children under three are spending longer in day care than ever before,” they write:
We have identified around 1000 research reports, in different sciences spread over 30 years, that separating small children from their mothers has a variety of adverse effects. However, we have not identified a systematic review in any leading general medical journal and as far as we know, this is the first editorial on this topic.
Haspel argues that high-quality child care has “enormous benefits” for parents “as well as for gender equity and the economy.” But we simply know too little about the effects of nonparental, group care on children—especially infants and toddlers and those spending long hours in child care.
“Each child has only one first 3 years,” Chaparro et al. write, and “the quality of the care environments they experience over this short time has life-long consequences.” And as James Heckman recently explained, the environment provided by a young child’s family cannot easily be replaced by institutional care:
The ‘intervention’ that a loving, resourceful family gives to its children has huge benefits that, unfortunately, have never been measured well. Public preschool programs can potentially compensate for the home environments of disadvantaged children. No public preschool program can provide the environments and the parental love and care of a functioning family and the lifetime benefits that ensue.
A strong body of research clearly shows that high-quality child care has significant benefits for low SES children, particularly from single-parent households—the very children with the very least access to high quality care. Boosting both access and quality for low SES children merits substantially increased public funding.
But a large, non-targeted expansion of publicly-funded child care would constitute nothing less than a massive, national experiment with our nation’s youngest, most vulnerable children. Notwithstanding potential benefits to family finances, women’s careers, and national GDP, taking such a risk with millions of children’s lives would be a gravely irresponsible and dangerous mistake.
See Also
Op-Ed ~ February 10, 2015
Event ~ March 2, 2021
Blog ~ April 14, 2021