Elsevier

Journal of Health Economics

Volume 64, March 2019, Pages 93-107
Journal of Health Economics

Seeing and hearing: The impacts of New York City’s universal pre-kindergarten program on the health of low-income children

https://doi.org/10.1016/j.jhealeco.2019.01.004Get rights and content

Highlights

  • Examines the impacts of New York City’s universal pre-kindergarten program (UPK) on the health of low-income children.

  • Adopts a difference-in-regression discontinuity design to address the potential endogeneity of UPK eligibility.

  • Finds positive effects of UPK on diagnoses with vision problems, treatment for hearing or vision problems, and immunizations.

  • Finds no effect of UPK on injuries, infectious diseases or overall utilization.

  • Suggests that UPK might benefit children by identifying conditions that could delay learning and cause behavioral problems.

Abstract

We examine the effect of New York City’s universal pre-kindergarten program (UPK) on the health and utilization of children enrolled in Medicaid using a difference-in-regression-discontinuities design. We find that UPK increases the probability that a child is diagnosed with asthma or with vision problems, receives treatment for hearing or vision problems, or receives an immunization or screening during the pre-kindergarten year. These effects are not offset by lower rates in the kindergarten year, suggesting that UPK accelerates the rate at which children are identified with and treated for conditions that could potentially delay learning and cause behavioral problems.

Introduction

A growing body of literature suggests that early childhood education improves lifetime outcomes, especially for disadvantaged children (Heckman, 2011; Muennig, 2015). In response, many states have expanded funding for pre-kindergarten programs. By the 2014–2015 academic year, 42 states and District of Columbia had established state funded pre-K programs serving 4-year-olds. Across the nation 1.2 million 4-year-old children, who account for about 30% of the total population of 4-year-old children, were enrolled in a state funded pre-K program (Barnett et al., 2016).

While evidence for the long-term benefits of preschool is strong, it is more challenging to identify how these benefits are achieved. Many studies find that pre-K programs improve children’s academic skills (literacy, language and math), cognition, and test scores in the short run, but often find these effects fade out over time (Weiland and Yoshikawa, 2013; Gormley and Gayer, 2005; Fitzpartick, 2008; Lipsey et al., 2013; Wong et al., 2008; Cascio and Schanzenbach, 2013; Hill et al., 2015; Gormley et al., 2008). One potential pathway is through the impact of early childhood programs on non-cognitive skills, though the evidence in support of this is equivocal ((Zigler et al., 1994; Janvier et al., 2016) (Magnuson et al., 2007; Baker et al., 2008). There is very little evidence, however, about how UPK programs affect physical health, and particularly aspects of physical health that may be correlated with subsequent school performance, such as vision (D’Onise et al., 2010). This is surprising because most early childhood programs include some form of physical health screening or a co-located medical provider in their designs.

In this paper, we focus on the potential of UPK to address physical health conditions, and, in particular, identify and address sensory problems (such as vision deficits) early. Vision problems are prevalent among children in poor urban environments (Gould and Gould, 2003), are correlated with low academic performance and subsequent well-being (Krumholtz, 2000; Polack et al., 2008; Rein et al., 2006), are associated with social and emotional difficulties (Zaba, 2001; Johnson et al., 1996) and are readily addressed through screening and treatment Basch (2011).1 Prior research has likewise suggested that hearing problems may negatively impact academic performance, self-esteem, and social functioning (Daud et al., 2010; American Speech-Language-Hearing Association, 2017; Theunissen et al., 2014) and that early intervention is essential for closing the developmental gap (Mellon et al., 2009; Eriks-Brophy et al., 2012). The proactive diagnosis and treatment of vision and hearing problems among pre-K children may provide them with early advantages for succeeding in social environments that in turn shape their lives down the line.

The introduction of universal pre-kindergarten (UPK) in New York City, the nation’s largest school system, offers an opportunity to assess how a highly regulated, standardized UPK program affects health outcomes in the short-run. The substantial size of the program; the design and financing of the program, which suggest that it offers standardized care of relatively high quality, including standardized protocols for screening and follow-up of sensory problems; and the timing of the roll-out make the NYC case a valuable case-study.

As with other US UPK programs, New York’s program is currently voluntary. This means that enrolled children may differ from those whose parents choose other options. An analysis that does not take into account the endogeneity of enrollment (and the difference in children’s age by design) is likely to lead to biased results (Gormley and Gayer, 2005; Lipsey et al., 2015). We exploit two distinct sources of variation to avoid this endogeneity: the introduction of the program and the rules governing eligibility, which use a strict January 1 birthdate cutoff. The introduction of the program provides variation for before-and-after comparisons (as in a first-difference design) and the age cutoff allows us to compare outcomes for children who are just a little too old (young) to be eligible for the program with those for children who are just young (old) enough to be eligible (as in a regression discontinuity design). Each of these sources of variation alone, however, could mask other confounders.

To address these potential biases, we adopt a difference-in-regression-discontinuities design (DRD), coupling the age-cutoff RDD with a difference-in-differences analysis. We identify the effect of UPK as the difference between the estimated effects on health of an RDD around the age cutoff in the year before the UPK expansion and an RDD around the age cutoff in the year after the UPK expansion.2

Our study contributes to the literature in several ways. First, while most of the existing literature on UPK programs focuses on school readiness, we examine the effect of UPK on a range of short-run health outcomes during the pre-K year and in the Kindergarten year. We focus on low-income children who are enrolled in Medicaid, a group that is the target for most UPK programs.3,4 Second, the large-scale UPK program in NYC, with clear guidance and regulation, provides us with evidence from a relatively homogenous, regulated-quality program, as well as a large data set. Third, we adopt a difference-in-regression-discontinuities design to identify plausibly causal effects.

We find that the NYC UPK program led to increases in rates of diagnosis of asthma and vision problems, to increased rates of screening for immunization or infectious disease, and to increased rates of treatment of hearing and vision problems. UPK eligibility (attendance) increased the probabilities of being diagnosed with asthma and vision problems by 1.3 (3.8) and 1.9 (5.6) percentage points respectively, which correspond to percentage increases of 9.4% (27.5%) and 18.8% (55.4%) from baselines. UPK eligibility (attendance) also increased the probabilities of receiving treatment of vision and hearing problems and immunization or infectious disease screening by 2.2 (6.5), 0.9 (2.6), and 2.5 (7.4) percentage points respectively, which correspond to percentage increases of 15.3% (45.3%), 21.7% (62.7%), and 6.2% (18.5%).5 Our findings indicate that UPK may accelerate the rate at which children are identified with and treated for conditions that could potentially delay learning and cause behavioral problems.

Our main findings, especially the effects on the diagnosis and treatment of vision problems, are quite consistent across specifications and robust to a series of robustness checks. We find similar results for the further roll-out of UPK in 2015, for children who were just old enough for pre-K, and for children who were just young enough for pre-K. We follow the first UPK cohort into the second academic year after the UPK expansion, when they are in kindergarten. We find that the increased screenings and treatments during the pre-K academic year are not offset by a reduced rate of diagnosis or treatment during the kindergarten year. This implies that our UPK findings reflect more than differences in the timing of screenings associated with school entry -- children eligible for UPK receive both earlier and more treatment for these problems. We show that UPK increases the propensity for children to get timely care after a diagnosis. Getting treatment earlier for vision and hearing problems, which might otherwise interfere with learning and cause behavioral problems, could generate short-run benefits for children development, and might contribute to lifetime benefits as well.

Section snippets

Institutional background and data description

One of the greatest challenges to studies of pre-K programs has been the heterogeneity of program components and quality (D’Onise et al., 2010). Many of the programs that have generated evidence of long-term effects are of higher quality than the typical program in the field today. Current spending per child per year in pre-K averages only about $4500 (Barnett et al., 2016, measured as 2015 dollars), less than half the $11,009 spent on public education per pupil per year across the US (U.S.

Analytical framework

For children born in 2009 and 2010 (the post-UPK control group 1 and treatment group in Fig. 1), the relationship between our 2014–2015 outcome measures and the introduction of UPK can be modeled as:yi=α+θDi+eiwhere yi is the health outcome of child i during academic year 2014–2015; Di is an indicator of eligibility for UPK, which equals 1 if child i’s date of birth DOBi is after the age cutoff c (DOBic, where c is January 1st, 2010), α is the intercept, and ei is the error term. θ is the

Graphical analysis

In Appendix A, Fig. A1 (about the cohort-oldest sample) illustrates how health outcomes and healthcare utilization in the 2013–2014 and the 2014–2015 academic years change by age relative to the cutoffs, which are January 1st, 2009 for children born in 2008 and 2009 (before the UPK expansion), and January 1st, 2010 for children born in 2009 and 2010 (after the UPK expansion), respectively. Similarly, Figure A2 (referring to the cohort-youngest child sample) illustrates how health outcomes and

Robustness checks

In our main analysis, we use a 60-day bandwidth around the cutoff. We also check whether our results are robust to alternative bandwidths: 15 days, 30 days, 90 days, 120 days, outcome-specific IK optimal bandwidth (Imbens and Kalyanaraman, 2012) and CCT optimal bandwidth (Calonico et al., 2014) (Table D1, Appendix D). We also check whether our main results are robust to different model specifications (column (a)-(e), Table D3, Appendix D). We check whether the current specification for the

Effects over time

Virtually all children enrolled in UPK who remain living in NYC progress to Kindergarten a year later. Asthma and hearing and vision problems tend to be chronic conditions. In the absence of UPK, children might have been diagnosed with these conditions a year later, when they entered kindergarten. To assess these possibilities, we follow the cohort-oldest children and analyze whom we can follow through their entry into kindergarten in the 2015–2016 academic year, the year following UPK.

Implied treatment-on-treated (TOT) effect

Our analyses provide estimates of the ITT effect of UPK, which provides insights about the effect of UPK policy. In comparing these results with prior studies of the effects of early childhood education, it is useful to construct TOT estimates, which measure the effect of participation in UPK. We are not able to directly estimate the TOT effect because we do not have actual enrollment data, but we can calculate the approximate TOT effects implied by our estimated ITT effect using enrollment

Conclusion

Existing studies find that pre-K is an effective way to boost children’s development, especially cognition and academic performance (Weiland and Yoshikawa, 2013; Gormley and Gayer, 2005; Fitzpartick, 2008; Lipsey et al., 2013; Wong et al., 2008; Cascio and Schanzenbach, 2013; Hill et al., 2015; Gormley et al., 2008). This evidence supports the premise that a pre-K program’s key outcome is to increase readiness for kindergarten and the schooling that follows (Pianta et al., 2007). Our study

References (59)

  • W. Steven Barnett et al.

    The State of Preschool 2015

    (2016)
  • Charles E. Basch

    Vision and the achievement gap among urban minority youth

    J. Sch. Health

    (2011)
  • Jacob Bor et al.

    Regression discontinuity designs in epidemiology: causal inference without randomized trials

    Epidemiology

    (2014)
  • Sebastian Calonico et al.

    Robust nonparametric confidence intervals for regression-discontinuity designs

    Econometrica

    (2014)
  • Elizabeth. Cascio

    Does Universal Preschool Hit the Target? Program Access and Preschool Impacts. No. w23215

    (2017)
  • Elizabeth Cascio et al.

    The impacts of expanding access to high-quality preschool education

    Brookings Pap. Econ. Act.

    (2013)
  • Jing Cheng et al.

    Bounds on causal effects in three-arm trials with non-compliance

    J. R. Stat. Soc. Series B Stat. Methodol.

    (2006)
  • Alison Klebanoff Cohen et al.

    Education: a missed opportunity for public health intervention

    Am. J. Public Health

    (2013)
  • Mohd Khairi Md Daud et al.

    The effect of mild hearing loss on academic performance in primary school children

    Int. J. Pediatr. Otorhinolaryngol.

    (2010)
  • Greg J. Duncan et al.

    Investing in preschool programs

    J. Econ. Perspect.

    (2013)
  • Alice Eriks-Brophy et al.

    Communication, academic, and social skills of young adults with hearing loss

    Volta Rev.

    (2012)
  • Avi Feller et al.

    Compared to what? Variation in the impacts of early childhood education by alternative care type

    Ann. Appl. Stat.

    (2016)
  • Maria D. Fitzpartick

    Starting school at four: the effect of universal pre-kindergarten on children’s academic achievement. 2008

    B. E. Anal. Policy

    (2008)
  • Paul Glewwe et al.

    The impact of providing vision screening and free eyeglasses on academic outcomes: evidence from a randomized trial in title I elementary schools in Florida

    J. Policy Anal. Manag.

    (2018)
  • William T. Gormley et al.

    Promoting school readiness in Oklahoma: an evaluation of Tulsa’s Pre-K program

    J. Hum. Resour.

    (2005)
  • William T. Gormley et al.

    The Early Years. Preschool Programs Can Boost School Readiness

    Science

    (2008)
  • Marge Christensen Gould et al.

    A clear vision for equity and opportunity

    Phi Delta Kappan

    (2003)
  • Veronica Grembi et al.

    Do Fiscal Rules Matter?

    Am. Econ. J. Appl. Econ.

    (2016)
  • Elizabeth A. Harris

    Most Parents Got Top Choices for Pre-K, Mayor De Blasio Says

    The New York Times

    (2015)
  • Cited by (10)

    • The effects of universal child allowance on maternal health: Evidence from South Korea

      2022, Social Science and Medicine
      Citation Excerpt :

      More specifically, the age factors in child care needs can lead to different child care experiences for mothers, which can consequentially contribute to different maternal health outcomes. Second, the RDD assumes that all relevant variables other than CA eligibility status around the cutoff runs smoothly (Hong et al., 2019). However, if there are other relevant policies that use a similar age cutoff, then the RDD approach is problematic because we can't identify the true effects of CA.

    • The effects of a daycare reform on health in childhood – Evidence from Sweden

      2022, Journal of Health Economics
      Citation Excerpt :

      This may indicate that daycare has increased parents’ awareness for their children’s health and the necessity for health prevention. Such an explanation is in line with Hong et al. (2019) who show that a pre-kindergarten expansion immediately increases health screening and receiving an immunization among affected children. At age 6–7, affected children have about 0.56 fewer sickness visits per year than non-affected children, which corresponds to a reduction by about 17 percent at the mean.

    • Reframing High-Quality Public Preschool as a Vehicle for Narrowing Child Health Disparities Based on Family Income

      2021, Academic Pediatrics
      Citation Excerpt :

      In New York City's universal pre-k program for 4-year-olds, providers must work with the City's Office of School Health to conduct vision screenings. A recent evaluation found that children enrolled in pre-k were on average diagnosed with vision problems a year earlier than they would have been in the absence of pre-k.43 Pre-k enrollment also increased the chance that children diagnosed with a vision problem received treatment within six months.

    View all citing articles on Scopus

    ☆This research was supported by the Robert Wood Johnson Foundation Policies for Action Program.

    View full text