Do the Perils of Universal Childcare Depend on the Child's Age?

Abstract

The rising participation of women in paid work has not only heightened demand for universal early education and care programs but also led to increased use of child care amongst children at earlier ages. Prior research investigating Quebec's universal highlysubsidized child care documented significant declines in a variety of developmental outcomes for all children aged 0-4 years. However, past analysis has not explored whether these effects vary for children of different ages. In this paper, we demonstrate substantial heterogeneity in policy impacts by child age. Children who gain access to subsidized child care at earlier ages experience significantly larger negative impacts on developmental scores, health and behavioral outcomes. The sole exception is the negative relationship between access to subsidized child care and hyperactivity scores which steepens with child age. Our analysis additionally provides significant evidence of treatment effect heterogeneity within ages, and reveals benefits from access to universal child care on developmental scores for those that are above three years of age. * We would like to thank Jonas Vlachos as well as participants at the CESIFO Economic Studies and UCLS Conference on Children, Families and Human Capital Formation and seminar presentations at York University and the University of Ottawa for many helpful comments and suggestions that improved this draft. This paper is a revised version of a portion of Kottelenberg's Queen's University 2009 Master's research paper. Lehrer wishes to thank SSHRC for research support. We are responsible for all errors.

Concern Over Internal, External, and Incidence Validity in Studies of Child‐Care Quantity and Externalizing Behavior Problems

“…Child‐care quantity has been studied in Australia, Canada, Denmark, Japan, the Netherlands, Norway, Portugal, Slovenia, and the United Kingdom (). Yet none of these countries has received even a quarter of the empirical attention that the United States has.…”

“…Among the articles that studied associations between child-care quantity (broadly defined as age of entry, hours per week, episodes per interval, child care or preschool vs. parental care, or any combination thereof), we included those using statistical methods more conservative than conventional covariate adjustment approaches. Thirteen studies met this criterion (13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25).…”

“…First, in large part because of differences between countries, we found variations in average age of entry ranging from 3 months (21) to 14 months (25), and in average weekly hours (e.g., at 36 months, ranging from 36.7 in one U.S. study [21] to 29.45 in a Norwegian study [23]). Second, although many studies used nationally representative samples (13)(14)(15)(16)(17)(18)(19), one examined families of predominantly low-SES (20) and others looked more at middle-and high-SES families (21)(22)(23)(24)(25). Even among those samples that were nationally representative, variations in treatment effects across SES or race/ethnicity were not addressed with methods other than multivariate regression.…”

“…Nonetheless, we did not fully explore sampling or design variations across these studies. For example, four of the six studies that report harmful consequences of time in child care (13,14,16,19) compared a treatment of publicly funded center-based child care or preschool with a mixed counterfactual of parental care or alternative nonparental care arrangements. And four of the six came from a single province in Canada (13)(14)(15)(16).…”

Cognitive and Non-Cognitive Costs of Daycare 002 for Girls

“…the Quebec extension are Kottelenberg and Lehrer (2014a) and Kottelenberg and Lehrer (2014b), which focus specifically on the heterogeneity of e↵ects at di↵erent ages in the 0-4 range and across genders. They show that the negative e↵ects of this intervention are particularly large among kids who start attendance at an earlier age and among boys.…”