Policy debate has shifted away from the simple question of whether early childhood programs have an impact on children’s educational development toward questions of how long the effects  last. The recent evaluation of Head Start has ignited fears that effects might “fade out” over time as children enter under-resourced classrooms populated by peers who had not benefitted from high-quality early childhood programs.

North Carolina’s two flagship early childhood programs, Smart Start and More at Four, have been implemented since 1995 (for Smart Start) and 2002 (for More At Four). Smart Start provides state funding to improve the quality of childcare services at the county level for all children between the ages of 0-5. More at Four provides state funding for pre-school slots for disadvantaged four-year-olds.

In prior research ,we capitalized on naturally-occurring county-level variation in dosage (operationalized as dollars allocated per age-relevant child in a county’s population, including zero dollars prior to the introduction of the program in a county)  of each program in each year (1993-2009) to measure their effects on child outcomes in third grade, using models that incorporated individual and county covariates and fixed effects for county and year. In the first study, we  found that allocations to a county for each program had independent and positive effects on third grade test scores in reading and math. In a second study we found that both programs also reduced placements in special education program for third graders.

We propose the same basic approach for this paper but extend the analysis through fifth grade and examine a broader range of outcomes, including, for example retention in grade, Central to our approach is that we are measuring community level effects, not just effects on participants.

Preliminary results provide no evidence that the effects on test scores in math and reading of the two programs decline as children age, and in some cases the effects may get larger. Similar effects and patterns hold for special education placements and grade retention. In further  analyses we will examine heterogeneity of effects for groups of students whose backgrounds vary by mother’s immigrant status, education level, poverty status, and ethnicity. 

We conclude that the impacts of each of these two statewide early childhood programs remain positive across elementary school with no evidence of fading out of effects. We speculate that perhaps population-level penetration of each program might be responsible for the persistence of impacts, due to positive spill-overs and improved classroom-level teaching that is afforded when fewer students need remediation.