Can Starting Childcare Earlier Reduce Achievement Gaps for Low-Income Children?

Almås, Drange, Meghir, and Zachrisson (2026) examined whether attending childcare at an earlier age improves academic performance later in childhood and adolescence. They used Norwegian administrative registry data covering entire national birth cohorts from 2002–2007, along with standardized math and reading test scores in grades 5 and 9. They found that starting childcare one year earlier increased ninth-grade math scores by about 9.7% of a standard deviation overall. The effects were much larger for disadvantaged children. For children whose mothers lacked a high school diploma, math scores increased by about 28% of a standard deviation, reducing educational achievement gaps substantially.

Why This Article Was Selected for The Policy Scientist

Early childhood education policy has become increasingly important as governments confront long-term concerns about inequality, labor force participation, social mobility, and the developmental consequences of economic disadvantage. This article is especially timely because many countries are rapidly expanding subsidized childcare systems while debates continue over whether large-scale programs produce durable academic gains or merely short-term improvements that later fade out. Almås, Drange, Meghir, and Zachrisson are established contributors to this literature, particularly in Scandinavian childcare research, and this study extends prior work by examining outcomes into adolescence rather than early elementary school. The Norwegian administrative data are exceptionally strong, covering entire national cohorts with high-quality longitudinal linkage and standardized testing outcomes. The quasi-experimental design exploiting staggered childcare expansion represents a credible causal inference strategy that is substantially stronger than conventional multivariate regression approaches. The findings are likely most generalizable to affluent democracies with relatively high-quality childcare systems and strong public institutions.

Full Citation and Link to Article

Almås, I., Drange, N., Meghir, C., & Zachrisson, H. D. (2026). Early childcare attendance and cognitive skills in adolescence. NBER Working Paper No. 35109. National Bureau of Economic Research. https://doi.org/10.3386/w35109

Central Research Question

This paper examines whether attending formal childcare at an earlier age produces lasting improvements in cognitive achievement during later childhood and adolescence. Specifically, Almås, Drange, Meghir, and Zachrisson evaluate whether Norway’s large-scale expansion of childcare slots for one- and two-year-old children beginning in 2003 improved later academic performance in mathematics and reading. The authors are particularly interested in whether the effects differ across socioeconomic groups, especially for children whose mothers have low levels of education and for children from immigrant families. More broadly, the article addresses a central policy debate in education and social policy research: whether universal early childhood programs generate durable developmental gains or whether early advantages fade over time as children progress through school. The study also evaluates whether high-quality childcare systems can reduce long-term educational inequality rather than merely increase maternal labor force participation.

Previous Literature

The article situates itself within a large international literature examining the developmental effects of universal or subsidized childcare programs. Prior studies have produced mixed findings, with outcomes varying substantially across countries, institutional settings, and measures of program quality. Research on Quebec’s universal childcare expansion, particularly the work of Baker, Gruber, and Milligan, found weak or sometimes negative effects on cognitive and behavioral outcomes, although later work identified some positive impacts on adolescent educational outcomes. In contrast, studies from Germany and Norway have generally reported neutral or positive effects, especially for disadvantaged children. The authors argue that differences in staff-child ratios, teacher qualifications, pedagogical orientation, and the quality of home environments likely explain much of the variation across studies.

The article also contributes to a newer strand of literature examining whether gains from early childhood interventions persist into adolescence. Much of the earlier research focused on preschool-age outcomes or early elementary school performance. The persistence question has become increasingly important because some meta-analyses and longitudinal studies suggest that early gains often diminish over time. The present study directly addresses this issue by following entire cohorts of Norwegian children into grades five and nine. The authors also build upon previous Norwegian childcare studies by Dearing, Havnes, Drange, and Zachrisson, many of whom have repeatedly examined Scandinavian childcare reforms using quasi-experimental methods. Unlike earlier work focused on younger children, this article evaluates outcomes much later in the educational trajectory.

Data

The study relies on exceptionally strong administrative registry data covering the full population of Norwegian children born between 2002 and 2007. These cohorts were directly exposed to Norway’s rapid childcare expansion for one- and two-year-old children that occurred between 2003 and 2008. The dataset links children to detailed parental demographic information, including education levels, immigrant background, birth order, maternal age, and municipality of residence. The comprehensive nature of the Norwegian registry system substantially reduces concerns regarding attrition, sampling bias, or inaccurate self-reporting that commonly affect survey-based studies.

The primary outcome measures are standardized national test scores in mathematics and reading administered in grades five and nine. These exams are mandatory for nearly all students in Norway and are designed to capture the full distribution of academic ability. Because the authors observe entire national cohorts, they are able to conduct subgroup analyses for children from low-education households and immigrant families while still maintaining substantial statistical power. The article also uses administrative records associated with Norway’s “cash-for-care” policy to infer the age at which children entered formal childcare. These data allow the researchers to reconstruct childcare participation patterns with considerable precision. Overall, the data infrastructure represents one of the strongest features of the paper and provides unusually high external validity within the Norwegian context.

Methods

The authors employ a quasi-experimental instrumental variables strategy designed to identify causal effects rather than simple correlations between childcare attendance and later academic achievement. Their identification strategy exploits substantial variation across Norwegian municipalities in the pace and intensity of childcare expansion following the 2003 policy reform. Because municipalities expanded childcare coverage at different rates due to local administrative and logistical constraints, children were exposed to differing levels of childcare availability that were plausibly unrelated to their underlying academic potential.

The core empirical approach instruments age of childcare entry using municipal childcare coverage rates during the years when children were eligible for care. This strategy is intended to address the selection problem that higher-income or more educated families may systematically choose earlier childcare enrollment. The authors provide extensive evidence supporting instrument validity, including balancing tests showing little relationship between the instrument and observable family characteristics once municipality and year fixed effects are included.

The statistical methods are rigorous and appropriate for the research question. The use of instrumental variables represents a substantially stronger causal inference framework than standard multivariate regression. The authors also cluster standard errors at the municipal level, apply bootstrap procedures for inference, and adjust for multiple hypothesis testing using Romano-Wolf stepdown corrections. These methodological choices increase confidence that the reported findings are not statistical artifacts. Although the study is not a randomized controlled trial, the research design approximates experimental conditions reasonably well within the constraints of population-level social policy research.

Findings/Size Effects

The study finds that earlier childcare attendance generated meaningful long-term improvements in mathematics achievement, particularly for disadvantaged children. Across the full sample, starting childcare one year earlier increased ninth-grade mathematics scores by approximately 9.7% of a standard deviation. The overall effects for reading achievement were small and statistically insignificant, although the authors also found no evidence that early childcare caused harm to reading outcomes.

The most important findings involve socioeconomic heterogeneity. For children whose mothers lacked a high school diploma, starting childcare one year earlier increased fifth-grade mathematics scores by roughly 21% of a standard deviation and ninth-grade mathematics scores by approximately 28% of a standard deviation. The authors estimate that these gains reduced the math achievement gap between low-education and higher-education households by approximately one-third. Reading scores for disadvantaged children also improved modestly, although these estimates were less precise statistically.

The paper also reports large estimated effects for immigrant children. Among children with immigrant mothers, earlier childcare attendance improved ninth-grade mathematics scores by more than 90% of a standard deviation, although these estimates were less stable and lost significance after corrections for multiple hypothesis testing. Even so, the reduced-form estimates strongly suggest that increased childcare access substantially improved academic outcomes for immigrant children.

An important substantive finding is the absence of negative effects for children from more advantaged households. Unlike some prior studies from Quebec and Italy, the Norwegian childcare expansion did not appear to reduce cognitive outcomes among higher socioeconomic status children. The authors argue that this likely reflects Norway’s relatively high-quality childcare system, characterized by low staff-child ratios, strong teacher qualifications, and a play-based pedagogical framework.

Conclusion

The article provides substantial evidence that high-quality early childcare can generate durable academic benefits extending well into adolescence, particularly for disadvantaged populations. The findings are especially important because they directly address ongoing debates regarding the persistence of early childhood intervention effects. Rather than fading out entirely, the mathematics gains observed in this study remained detectable through ninth grade, suggesting that early cognitive advantages may persist under certain institutional conditions.

The study also contributes to broader debates regarding educational inequality and social mobility. The large effects observed for children from low-education households imply that universal childcare systems may function not only as labor market infrastructure for parents but also as developmental interventions capable of narrowing achievement gaps. At the same time, the article avoids claiming that such results automatically generalize across all national contexts. Norway’s childcare system combines unusually high public investment, strong institutional quality, highly trained staff, and relatively low child-to-teacher ratios. These features may limit the direct applicability of the findings to lower-capacity childcare systems elsewhere.

Methodologically, the article represents one of the stronger recent contributions to the childcare literature because of its use of population-level administrative data and a credible quasi-experimental identification strategy. Although the study cannot fully replicate the certainty associated with randomized assignment, its instrumental variables framework provides considerably stronger evidence than conventional observational studies relying solely on regression adjustment. Overall, the article strengthens the case that the quality and structure of childcare systems matter substantially in determining whether early childhood interventions produce lasting educational gains.