Early childhood education has a problem in the research community, and it’s called “fade-out.” If you work in early childhood research, you’ve heard this term before. If you haven’t, the general problem is that some research suggests that gains from early childhood education programs realized by children as they enter kindergarten “fade away” in elementary years to the point where they are nothing by third grade. If this is true, then early childhood education programs could be expensive programs that ultimately yield little results for children or families.
The savvy policymaker, though, will ask the following question: why should I care about third-grade test scores? Well, third-grade test scores are nice because they allow us to evaluate a program after a couple of years rather than decades. The problem with third grade test scores is that they might be the least predictive results to extrapolate to life outcomes.
Timothy Bartik, a leading economist of job creation, has studied the impact of early childhood education on local wages. In his book From Preschool to Prosperity: The Economic Payoff to Early Childhood Education, Bartik tackles the question of fade-out. He looks at the four most high-profile studies in early childhood education: experimental studies the Abecedarian Project and the Perry Preschool Project, and quasiexperimental evaluations of Head Start and the Chicago Child-Parent Center Program.
As can be seen above, estimates of adult earnings based on third grade test scores are below the estimates of adult earning effects at end of the early childhood program. Thus, Bartik does find evidence for fade-out. However, Bartik also finds evidence of a strong bounce-back from fadeout, with actual adult earnings higher not only than third grade test scores would predict, but also higher than the original end-of-program test scores would have suggested.
These results are also reflected in cost-benefit results reported in a high-profile literature review by the Rand Corporation. The review reported the results of 12 evaluations and meta-analyses of early childhood education programs, some broken out into treatment group subcategories. While the evaluations that only included results from elementary school tended to show negative results, evaluations that followed up with participants in secondary school, early adulthood, and middle adulthood showed increasing net benefits.
Data from Karoly et al., Early Childhood Interventions: Proven Results, Future Promise, pp.xxvi-xxvii
According to Rand,
The largest benefit-cost ratios were associated with programs with longer-term follow-up because they allowed measurement at older ages of outcomes such as educational attainment, delinquency and crime, earnings, and other outcomes that most readily translate into dollar benefits (p. xxv).
This suggests that third grade scores could be missing latent human capital that was built by early childhood scores such as socialization, emotional intelligence, or verbal communication skills that then end up leading to future education, crime reduction, and labor market outcomes.
Ultimately, third grade test scores have little social relevance on their own: they are only a proxy for future outcomes with immense social importance: educational attainment, crime victimization, and labor market earnings. While third grade test scores let us more quickly assess the impact of policies, if these scores are not correlated with socially-relevant outcomes, they are not of much use to policymaking. Maybe the problem isn’t the policies: it’s the limitations of how we test their impacts.