« Back to Results

Measuring the Effects of Early Childhood Education

Paper Session

Friday, Jan. 5, 2018 2:30 PM - 4:30 PM

Pennsylvania Convention Center, 104-B
Hosted By: American Economic Association
  • Chair: Chloe Gibbs, University of Notre Dame

Long-term Effects of Head Start: New Evidence From the PSID

Douglas Miller
,
Cornell University
Na'ama Shenhav
,
Dartmouth College
Michel Grosz
,
University of California-Davis

Abstract

The best available evidence of the long term impact of Head Start suggests that the program has a positive impact on participants’ economic and health outcomes. However, many of these studies are limited by small sample size or are only able to examine outcomes during early adulthood. This study uses the Panel Study of Income Dynamics and a family fixed effects approach to identify the effects of Head Start on an index of economic and health outcomes up to age 40. Using one of the largest longitudinal samples of Head Start participants and the detailed information gathered by the PSID, we find that Head Start participation increases the likelihood of completing some college. However, we find little evidence for a strong effect on overall economic or physical well-being. In our discussion, we highlight under-appreciated methodological limitations of the family fixed effects approach with a binary independent variable. The fact that estimated effects reflect a local average treatment effect among “switcher” families implies (1) significantly less variation than one might suppose, and (2) a disproportionate reliance on large families for identification. We also provide guidance regarding the (3) sensitivity to (typically innocuous) functional form assumptions. We conclude that alternative research design strategies should be pursued to gain additional evidence about the long term impact of Head Start.

The Impact of Full-day Kindergarten Expansions on Academic Achievement

Chloe Gibbs
,
University of Notre Dame

Abstract

In the United States, full-day kindergarten has proliferated in the past two decades as states and localities have rapidly expanded their provision of kindergarten in full-day settings. Participation in full-day kindergarten eclipsed half-day in 1995 and now constitutes approximately three-quarters of kindergarten students. In contrast to the existing, limited literature on full-day kindergarten impact, which focuses on participant effects, this study provides the first evidence on the systemic impact of provision by focusing on subsequent student achievement in places that expand full-day kindergarten offerings. Leveraging variation across states and—within one state, across districts—and over time, this research investigates the impact of expansions on mean student achievement, in the third through eighth grades, and on Hispanic-white and black-white test score gaps. Full-day kindergarten contributes to improved overall academic performance in both reading and math in the later grades, but may in fact exacerbate achievement gaps. As full-day kindergarten has evolved from a targeted to near-universal intervention, the findings have important implications for the broader policy discourse around early childhood investments and inequality.

A Reanalysis of Impacts of the Tennessee Voluntary Prekindergarten Program

Tyler Watts
,
University of California-Irvine
Mariela Rivas
,
University of California-Irvine
Greg Duncan
,
University of California-Irvine

Abstract

The current paper provides a number of estimates of the long-run effects of assignment to the Tennessee Voluntary Prekindergarten Program (TNVPK). In a previous analyses of these data, Lipsey, Farran, and Hoffer (2015) reported that the program produced substantial gains on student achievement at the end of preschool (b= .32 sd), but no impacts in kindergarten and first grade. Surprisingly, they found negative program effects at the end of grades 2 (b= -.15) and 3 (b= -.13). These results have garnered much attention as they suggest that academic preschool might harm the long-run achievement of participants. Although the TNVPK study was designed as a lottery-based RCT, Lipsey and colleagues (2015) used propensity score matching to analyze program effects. We rely on random assignment variation to estimate program effects through grade 3. We estimate ITT effects based solely on the lottery results as well as TOT effects using 2SLS models.
We find some evidence of baseline imbalance, with the treatment group less Hispanic (p = 0.021) and the control group scoring higher on 3 of the 4 baseline Woodcock Johnson reading tests (p values ranged from 0.003 to 0.066). The ITT effect of .19 sd at the end of preschool fades to 0 in kindergarten and first grade. Point estimates are negative but statistically insignificant in second and third grade. We also adopt a number of approaches to adjust for the considerable non-participation of families in the study, attrition, and to the definition of participation.

Does Universal Preschool Hit the Target? Program Access and Preschool Impacts

Elizabeth U. Cascio
,
Dartmouth College

Abstract

This paper uses the rich diversity in state rules governing access to public preschool programs in the U.S. to study the relative cost efficacy of universal programs for poor populations. Using age-eligibility rules to construct an instrument for attendance, I find that universal preschool generates substantial cognitive test score gains for poor 4-year-olds. Preschool programs targeted toward poor children do not. These findings are robust to the definition of poverty, comparison group, and controls for test scores earlier in life, and cross-state differences in demographics and alternative care options are not decisive factors. Benefit-cost ratios of universal programs remain favorable despite their relatively high costs per poor child. An auxiliary analysis suggests that peer effects are an important contributor to universal programs’ higher productivity.
Discussant(s)
Diane Whitmore Schanzenbach
,
Northwestern University
Maria Fitzpatrick
,
Cornell University
Christopher R. Walters
,
University of California-Berkeley
Douglas Miller
,
Cornell University
JEL Classifications
  • I2 - Education and Research Institutions
  • I3 - Welfare, Well-Being, and Poverty