Elsevier

Economics of Education Review

Volume 36, October 2013, Pages 41-59
Economics of Education Review

The long-run impacts of early childhood education: Evidence from a failed policy experiment

https://doi.org/10.1016/j.econedurev.2013.05.003Get rights and content

Highlights

  • We examine the effect of duration in Kindergarten on grade repetition and test scores.

  • We identify variation in Kindergarten length with a unique policy in British Columbia, Canada.

  • Longer kindergarten increases probability of grade repetition.

  • Longer kindergarten reduces grade 10 test scores.

Abstract

We investigate short and long-term effects of early childhood education using variation created by a policy experiment in British Columbia, Canada. Our findings imply being in kindergarten longer increases the probability of repeating the third grade, and decreases tenth grade math and reading scores. Effects are highest for low income students and males. Estimates suggest that more time in kindergarten may have a detrimental effect on future outcomes.

Introduction

Early childhood programmes are an important part of intellectual development, as they allow children to build a solid learning base, accumulate human capital at a higher rate in the future, and therefore perform better in school and in the labour market. The benefits may be especially important for disadvantaged children, since the early skill gap between them and their more advantaged peers can appear early and persist through time (Cunha, Heckman, & Lochner, 2006, chap. 12). Using early childhood programmes to close early skill gaps will help to prevent future skill gaps from forming. There are many empirical analyses of small scale early interventions like the Perry Pre-School experiment, and larger ones like Head Start, that support these ideas by showing positive impacts on a variety of short and long term outcomes. On the other hand, fewer researchers have examined the effect of universal programmes like kindergarten or pre-kindergarten (pre-k). Kindergarten is arguably the most widespread early childhood education programme, and is continuing to expand. For example, in the U.S. the availability of kindergarten expanded rapidly across the country in the 1960s and 1970s. More recently, several states have launched universal pre-k programmes. In Canada, Ontario and British Columbia (BC) moved from half day to full day kindergarten in 2010, while similar expansions are being considered in other provinces.

This paper examines the effects of a quick sequence of policy changes in BC that had the unintended effect of altering the length of time students spent in kindergarten. In the 1990–1991 school year, BC mandated a new school entry scheme called “dual entry,” whereby students entered school at two different points in the school year, rather than the traditional single entry system. Under dual entry, students whose birthdays fell between May and October would enter in September and those whose birthdays fell between November and May would enter in January. The first affected cohort – students born between May, 1985 and April, 1986 – entered under these rules, but by the end of that school year the government unexpectedly terminated the programme. This left all students who entered in January with only 6 months of kindergarten completed. To unwind the programme, financial support was provided for November–December born children to move to first grade at the start of the next school year, but not for January–April students. The end effect was that students born between November and December, 1985 were in kindergarten for as little as 6 months, while those born between January and April, 1986 were in kindergarten for as many as 16 months. Other students stayed in kindergarten for the normal 10 months. The effect of mandating, then quickly repealing, dual entry on expected length in kindergarten is summarized in Fig. 1. Students born at the end of 1985 get short kindergarten, those born at the beginning of 1986 get long kindergarten, and all others get normal kindergarten.

We estimate the impact of both short and long kindergarten on the probability of grade repetition at age nine and fifteen, and on tenth grade math and reading scores, using a difference-in-differences approach. The estimator compares differences in outcomes of short and long kindergarten cohorts and unaffected adjacent cohorts within a policy year to differences between similar cohorts in non-policy years. We find that long kindergarten increases the probability of repeating a grade by up to 2.3 percentage points by age nine, and by up to 2.8 percentage points by age fifteen. We also find that long kindergarten worsens tenth grade reading scores by up to 6.7% of a standard deviations. We find no significant effect of long kindergarten on math scores, and no effect of short kindergarten on any outcomes. The size of the long kindergarten effects is largest for students in the lowest household income quartiles, and for males.

Taken at face value, the estimates imply that spending more time in kindergarten is harmful to student performance. There are several possible explanations for the negative estimated effects. First, all else equal, increasing the length of kindergarten necessarily involves entering school earlier or leaving school later. In our experiment, long kindergarten students spend more time in school by virtue of entering earlier. One explanation for our result is therefore that the negative effect of entering school earlier outweighs any positive impacts of longer kindergarten. In the absence of entry age effects, our results are also explained by considering the counterfactual to spending more time in the school system. Without long kindergarten, some children would stay at home, and others would enroll in day care. If these alternative uses of time are better for student outcomes than kindergarten, it would support the effects we observe. In support of this explanation, our results are consistent with some studies in the child care literature, which show that preschool care can negatively impact on a variety of child outcomes.1

While our results are generated from a fairly specific policy experiment that did not originally intend to alter kindergarten length, we believe they are useful in at least two ways. First, the variation created by this policy experiment approximates the effect of introducing a universal pre-k programme where one did not previously exist (i.e. increasing kindergarten length, and reducing entry age). Thus we can generalize our estimates beyond the context of BC to help inform other jurisdictions of the possible impacts of introducing universal pre-k. This is important because much of the evidence on early childhood programmes is drawn from more targeted experiments (e.g. Head Start), so in spite of the relatively unique nature of the BC policy, we add to the literature by providing estimates on the impact of a more universal programme. The estimates are also useful as an interesting case study. In unwinding the dual entry policy, the government likely did not fully anticipate the impact their actions would have on students who were held back or moved ahead. The results show a long term effect of the government action on the affected students.

Section snippets

Existing literature

Most of the available evidence on the introduction of kindergarten or pre-k shows positive effects on student outcomes. In Tulsa, Oklahoma, a comparison of students of a similar age who are just about to start pre-k to students who just finished shows that pre-k finishers had improved test scores by 0.24–0.39 standard deviations (σ) (Gormley and Gayer, 2005). The introduction of universal pre-k in Georgia increased student test scores by 0.12σ, and increased probability of being on-grade for

BC's dual entry experiment

The variation in kindergarten length comes from a policy implemented in BC in the 1990–1991 school year.2 Just prior to the start of that school year, the

Data and sample description

The primary data are administrative student records from the Ministry of Education in BC. On a yearly basis since 1990, the Ministry collects data on all public and private school students in the BC school system. The data contain information on the school each student attends, their grade level, special education status, English as a second language (ESL) status, a small set of personal characteristics, date of birth, and home postal code (for all records after 1996).9

Grade repetition

Table 3 presents the reduced form and 2SLS results. Columns 1 and 4 are a baseline specification with no controls, columns 2 and 5 add demographic covariates, and columns 3 and 6 add a 3rd degree polynomial in day of birth. Starting with the reduced form in Panel A, across the three specifications the long kindergarten cohort is roughly 2.1 percentage points more likely to repeat a grade at age nine, while short kindergarten has virtually no impact. Relative to the average grade repetition rate

Conclusion

Early childhood education is important for the development of human capital, and kindergarten is arguably the most widespread early childhood programme. Early learning can establish a base upon which children will build as they age, and the more solid the base the easier it will be to accumulate human capital in the future. In this paper, we examined the effect of kindergarten length on the likelihood that students repeat a grade, and their high school test scores. We found that students who

References (39)

  • F. Bula

    January starters a dilemma

    The Vancouver Sun, July

    (1991)
  • F. Bula

    Hagen kills dual entry: Mixed reacion to end of twice-a-year admittance

    The Vancouver Sun, May

    (1991)
  • J.S. Cannon et al.

    Is full better than half? examining the longitudinal effects of full-day kindergarten attendance

    Journal of Policy Analysis and Management

    (2006)
  • E. Cascio et al.

    First in the class? age and the education production function

    (2013)
  • E.U. Cascio

    Do investments in universal early education pay off? long-term effects of introducing kindergartens into public schools

    (2009)
  • Children's Interests Must Come First

    The Vancouver Sun, July

    (1991)
  • H. Cooper et al.

    Effects of full-day kindergarten on academic achievement and social development

    Review of Educational Research

    (2010)
  • J. Currie

    Early childhood education programs

    Journal of Economic Perspectives

    (2001)
  • J. Currie et al.

    Does head start make a difference?

    American Economic Review

    (1995)
  • Cited by (13)

    • Education Research and Administrative Data

      2016, Handbook of the Economics of Education
    • The impact of a Caribbean home-visiting child development program on cognitive skills

      2014, Economics of Education Review
      Citation Excerpt :

      But they are in line with ample evidence documenting the potential benefits of center-based ECD programs. Longitudinal studies find that the impact of center-based interventions can be long-lasting and reach far into the future (Currie, 2001; Engle et al., 2011; Garces, Thomas, & Currie, 2002; Schweinhart et al., 2005), although the effects of early childhood education are sometimes found to be detrimental, especially with respect to behavioral skills (e.g. DeCicca & Smith, 2013; Loen, Bridges, Bassok, Fuller, & Rumberger, 2007). Studies that evaluate center-based interventions in Latin-America and the Caribbean have found beneficial cognitive effects in for example Argentina (Berlinksi, Galiani, & Gertler, 2009) and Bolivia (Behrman, Cheng, & Todd, 2004).

    View all citing articles on Scopus

    We thank Oscar Bedard, Elizabeth Dhuey, Steve Pischke, Fabian Waldinger, seminar participants at the University of Manitoba, the University of Winnipeg, and the Society of Labour Economists (SOLE) for valuable comments. All errors are our own.

    View full text