Revisiting No Child Left Behind: The Surprising Effects of Child Care on Longer-Run Outcomes

• In Event: Long-Term Outcomes of Early Supporting Environments and Subsequent Developmental Contexts

Abstract

Background
With increasing policy attention on providing affordable, high quality child care, the role of child care programs in shaping early child development and its effect on later-life outcomes continues to be an important research field. Studies have found effects of childcare programs on measures of adult outcomes such as earnings, health, and crime (Gray-Lobe et al 2021, Havnes & Mogstad 2011, 2015, Deming, 2009; Heckman et al., 2010, 2013, Silliman & Mäkinen 2022, Caneiro & Gilja 2014). Among them, two influential papers from Norway used a child expansion reform (Havnes & Mostad 2011, 2015), finding that it led to increased education and labor participation and a reduction in income inequality due to gains at the bottom and losses at the top. However, as those studies were unable to follow individuals beyond early adulthood, we do not know whether these effects persist or change. In this paper, we replicate their findings and then extend their analysis by following their sample of young adults (aged 30-33) into adulthood (aged 40-43).

Study Population
We use Norwegian register data and follow Havnes & Mogstad (2011, 2015) as closely as possible in defining the estimation sample and treatment. We start with the entire population of children born 1967-1976, residing in Norway when our outcomes are measured. Our sample is further restricted to children who’s mothers were married at the end of 1975 and were aged 16-49 when giving birth. We also exclude children from families that move between treatment and comparison municipalities during the expansion period. Parental education is measured in 1970 when the children are no older than three, with indicators for missing parental education, and sibling/family information relates to siblings born to each mother. The final sample consists of 499,938 children from 318,702 families compared to 499,026 children from 318,367 families in the original Havnes & Mogstad sample. Our main outcome variables are years of education and income, measured for a given year and fully controlling for age effects.

Methods
We use a difference-in-differences approach comparing adult outcomes between children from municipalities exposed to large and small expansions in child care coverage (first difference) before and after the child care reform (second difference). Quantile treatment effects allow us to study the impact of the reform on the income distribution.

Results
We are able to replicate the qualitative findings by Havnes & Mogstad (2011) for income and education measured when children are young adults (aged 30-33), see left and middle panels of Table 1. We then extend our analysis following the children into adulthood (aged 40-43), see right panel of Table 1. The main findings for education still hold, however, in contrast to Havnes & Mogstad (2015), we find gains at the top of the income distribution rather than the bottom, illustrated in Figure 1. The dramatic and somewhat surprising change is largely due to the timing of education and family formation for young adults and returns to educational investments, demonstrating the importance of using longer run outcomes to draw conclusions about heterogeneous program benefits.

Authors

• Astrid Marie Jorde Sandsør, Universitetet i Oslo
  Presenting Author

• Nicolai Topstad Borgen, Universitetet i Oslo

• Henrik D. Zachrisson, University of Oslo