Association for Public Policy Analysis & Management

In the last decade, campaigns to provide high quality prekindergarten (Pre-K) to four year olds have achieved remarkable success. The fraction of four year olds attending Pre-K reached 28% in 2013, double what it was in 2002 (Barnett et al, 2013).  Supporters of increasing funding for early childhood education make the case that major economic and social problems, such as crime and teen pregnancy, can be traced to low cognitive and socio-emotional skill levels.  Differences in these skill levels in advantaged and disadvantaged children appear early in childhood, but these differences can be alleviated by intervening early, leading to substantial reductions in negative later-life outcomes and therefore very high benefit-cost ratios and rates of return (Heckman, 2011).  These arguments have won the support of many policymakers at the state and federal level, including President Obama. In his 2013 State of the Union address, he announced the Preschool for All Initiative to expand high-quality preschool access to every child in America by allocating federal funds to finance states' provision of Pre-K.  

Despite the growing momentum behind universal Pre-K, there is little evidence to date regarding its long-run effects on adult outcomes. Instead, evidence of large long-run effects comes from studies of preschool programs that are targeted at more at-risk populations and are often more resource-intensive than Pre-K (e.g. Head Start and Perry Preschool).  Existing evaluations of state-run Pre-K programs do not yet observe outcomes past eighth grade. The results of these studies are mixed, with some finding evidence of substantial “fade out” in early test score effects. This does not rule out the possibility of large long-run impacts, however, as similar patterns of fade out in cognitive effects have been observed in a number of other early childhood interventions that nonetheless produced large long-run effects (e.g. Heckman et al, 2013; Deming, 2009). 

I estimate the impact of Oklahoma’s universal Pre-K program (UPK), introduced in 1998, on an important later-life outcome: teenage criminal activity.  I assemble data on criminal charges in the state of Oklahoma and use a regression discontinuity design which leverages the birthdate cutoff for UPK eligibility in the program’s first year of implementation.  This approach yields estimates of the effect of UPK availability (or the intent-to-treat effect of state Pre-K) compared to the prior mix of preschool services.  I compare the effect of UPK availability differentially by race, as black children in Oklahoma are four times more likely to be charged at age 18 or 19 than are white children (31% vs. 7%).  I find a significant negative impact of UPK availability on the likelihood that a black child is later charged with a misdemeanor or felony at age 18 or 19 (7 and 5 percentage points, respectively ), but no impact on the likelihood of later charges for white children.  This suggests that, like more targeted preschool programs, UPK has a large and important impact on a measure frequently associated with socio-emotional skills, but the impact on this measure is concentrated within a higher-risk population.