«IZA DP No. 3452 Too Young to Leave the Nest? The Effects of School Starting Age Sandra E. Black Paul J. Devereux Kjell G. Salvanes April 2008 ...»
DISCUSSION PAPER SERIES
IZA DP No. 3452
Too Young to Leave the Nest?
The Effects of School Starting Age
Sandra E. Black
Paul J. Devereux
Kjell G. Salvanes
zur Zukunft der Arbeit
Institute for the Study
Too Young to Leave the Nest?
The Effects of School Starting Age
Sandra E. Black
University of California, Los Angeles,
NBER and IZA
Paul J. Devereux
University College Dublin,
CEPR and IZA Kjell G. Salvanes Norwegian School of Economics, Statistics Norway, CEP and IZA Discussion Paper No. 3452 April 2008 IZA P.O. Box 7240 53072 Bonn Germany Phone: +49-228-3894-0 Fax: +49-228-3894-180 E-mail: email@example.com Any opinions expressed here are those of the author(s) and not those of IZA. Research published in this series may include views on policy, but the institute itself takes no institutional policy positions.
The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center and a place of communication between science, politics and business. IZA is an independent nonprofit organization supported by Deutsche Post World Net. The center is associated with the University of Bonn and offers a stimulating research environment through its international network, workshops and conferences, data service, project support, research visits and doctoral program. IZA engages in (i) original and internationally competitive research in all fields of labor economics, (ii) development of policy concepts, and (iii) dissemination of research results and concepts to the interested public.
IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion.
Citation of such a paper should account for its provisional character. A revised version may be available directly from the author.
IZADiscussion Paper No. 3452 April 2008
Does it matter when a child starts school? While the popular press seems to suggest it does, there is limited evidence of a long-run effect of school starting age on student outcomes. This paper uses data on the population of Norway to examine the role of school starting age on longer-run outcomes such as IQ scores at age 18, educational attainment, teenage pregnancy, and earnings. Unlike much of the recent literature, we are able to separate school starting age from test age effects using scores from IQ tests taken outside of school, at the time of military enrolment, and measured when students are around age 18. Importantly, there is variation in the mapping between year and month of birth and the year the test is taken, allowing us to distinguish the effects of school starting age from pure age effects. We find evidence for a small positive effect of starting school younger on IQ scores measured at age 18. In contrast, we find evidence of much larger positive effects of age at test, and these results are very robust. We also find that starting school younger has a significant positive effect on the probability of teenage pregnancy, but has little effect on educational attainment of boys or girls. There appears to be a short-run positive effect on earnings of beginning school at a younger age; however, this effect has essentially disappeared by age 30. This pattern is consistent with the idea that starting school later reduces potential labor market experience at a given age for a given level of education; however, this becomes less important as individuals age.
JEL Classification: J1 Keywords: education, earnings, IQ, teenage childbearing
Sandra E. Black Department of Economics University of California, Los Angeles 8283 Bunche Hall Los Angeles, CA 90095 USA E-mail: firstname.lastname@example.org * Black and Devereux gratefully acknowledge financial support from the National Science Foundation and the California Center for Population Research. Salvanes thanks the Research Council of Norway for financial support. We would like to thank seminar participants at UCD Geary Institute, the ESRI, the University of Maryland, Irvine, Davis, and the Tinbergen Institute. We are indebted to Stig Jakobsen who was instrumental in obtaining data access to the IQ data from the Norwegian Armed Forces.
Does it matter at what age a child starts school? Older children do better on tests, but is this because they are older and, in fact, unrelated to the age they started school?
Despite the dearth of convincing evidence, the popular press seems to suggest that there are benefits to “redshirting” (holding back) children in kindergarten (See NY Times, June 3, 2007). But is this the case? Are the short run benefits in terms of better performance just that: short run? And are there costs associated with finishing school and starting work later?
Much research has shown a consistent pattern that children who start school later tend to score higher on in-school tests, even after accounting for the endogeneity of school starting age.1 However, a key limitation in the interpretation of these correlations is the inability to distinguish between the effect of school starting age and a direct age-attest effect, as they are perfectly collinear. As a result, it could be that children who start school when they are older do better simply because they are older when they take the tests and being older provides an advantage, or it could be because there are direct benefits to starting school at an older age.
Using data on the population of Norway, we are able to separate these two effects using IQ test scores measured outside of school, at the time of military enrolment when students are around age 18. The rule in Norway that children must start school the year they turn 7 provides a discontinuity in school starting age for children born around January 1st and provides an instrument for actual school starting age. Importantly, there is also variation in the mapping between year and month of birth and the year the test is 1 This includes a cross-country study by Bedard and Dhuey (2006) and country-level studies by Fredrikson and Ockert (2006) for Sweden, Puhani and Weber (2007) for Germany, Strom (2004) for Norway, Crawford, Dearden and Meghir (2007) for England, McEwan and Shapiro (2008) for Chile, and Elder and Lubotsky (2007) for the US.
Cognitive scores around age 18 are particularly interesting as it is about the time of entry to the labor market or to higher education and so these scores are more relevant to the labor market than scores in kindergarten or elementary school.
Additionally, we study the effects of school starting age on longer term outcomes including educational attainment, early fertility, and adult earnings. While this is methodologically less complicated than studying in-school tests because age of measurement and school starting age are not perfectly collinear, the literature has been hindered by a paucity of data. Given the complications created by school leaving age rules in the US, European data are attractive when studying education and earnings.2 Educational attainment has been studied in the literature and it has generally been found that older starters average modestly higher completed education.3 However, ultimately adult earnings are a very important outcome and we add to the literature by being the first study to track cohorts of men and women from ages 24 to 35 and analyze how the impacts of school starting age change with age. Also, we are the first study to examine the effects of school starting age on the probability of teenage childbearing by women.
There has been some more recent evidence that the timing of births may be manipulated by parents based on school starting age cutoffs; to the extent that this is true and that these parents may be “different” on other dimensions as well, estimates in the 2 Studies using U.S. data have suffered from the fact that compulsory schooling laws specify minimum school leaving ages rather than grades so early starters have completed more education at the minimum dropout age. Therefore, historically, persons whose quarter of birth predicts starting early have on average higher schooling and higher earnings (Angrist and Krueger 1991). Dobkin and Ferreira (2007) finds younger starters also obtain slightly higher education in more recent U.S. cohorts.
3 See Puhani and Weber (2007) and Fertig and Kluve (2005) for Germany, and Fredrikson and Ockert (2006) for Sweden.
effects to take account of this possibility.
There are a number of reasons why school starting age may have long-run effects on children’s outcomes, although the sign of this effect is theoretically ambiguous. The potential advantages of an early starting age could include the following: Starting school younger means finishing school at a younger age, which implies more time for the individual to earn returns on their investment in human capital. Starting school younger may also be advantageous to the extent that children learn more at school than at home (or be a disadvantage if the opposite holds true), which could affect their long run trajectory. Finally, parental investment in their children may also depend on school starting age – parents may provide more help to children who are young for their grade level.
There are also potential disadvantages of an early starting age (and some of these could work in the opposite direction as well). It is possible that children cannot learn as well in school earlier in their developmental life. In addition, social development may depend on a child’s age relative to that of his/her classmates; if being relatively older is advantageous, then it might be better to start children later (and vice versa). To the extent that older children have an advantage on exams in school (by mere virtue of the fact that they are older when they take the test and hence know more), they may do better in the long run. Given these potentially offsetting effects, the true causal relationship becomes an empirical question.
We find evidence for a small positive effect of starting school younger on IQ scores measured at age 18. In contrast, we find evidence of much larger positive effects
find that school starting age has a significant effect on teenage pregnancy among girls but no strong effect on education among girls or boys. We also find that there appears to be a short-run positive effect on earnings of beginning school at a younger age; however, this effect has essentially disappeared by age 30. This pattern is consistent with the idea that starting school later reduces potential labor market experience at a given age for a given level of education; however, this becomes less important as individuals age.
The paper unfolds as follows. Section 2 presents the relevant literature. Section 3 describes our methodology and contrasts it to other approaches in the literature. Section 4 discusses relevant institutional details in Norway, and Section 5 provides a data description. Section 6 presents our results and Section 7 concludes.
Despite the importance of the distinction, there is little solid evidence as to the role of school starting age (SSA) versus test age (AGE) in determining school test scores.
While children are in school, researchers are faced with the identity that
AGE AT TEST = SCHOOL STARTING AGE + YEARS OF SCHOOLINGMost of the literature has compared test scores of children who are in the same grade and so has in fact estimated the combined effects of SSA and AGE.
Given the difficulty with separating out the two effects, a number of recent papers try to infer the role of age versus school starting age by looking either at early test scores or at changes in scores over time.4 Elder and Lubotsky (2007) show that there are strong 4 For example, Datar (2006) finds that achievement changes between kindergarten and first grade are not highly correlated with age at school entry.
affected by formal schooling. Elder and Lubotsky (2007) and Cascio and Schanzenbach (2007) also show that effects of age-at-school-entry on test scores tend to get smaller as children move to higher grades. Together these papers imply that the estimated starting age effects partly reflect the endowment differences between students when school starts and that there is little evidence that students learn more in school if they are older when they start. However, none of this work is able to directly disentangle the direct effect of age at test from that of school starting age.
While most of the literature controls for time in school, there is another series of papers that controls for age at test. There is some evidence that, when tested at the same age, young children score better on in-school tests if they started school younger and hence have spent more time in school. Cahan and Cohen (1989) study Israeli elementary school children and Elder and Lubotsky (2007) compare U.S. children around age 6 but with different predicted starting ages (based on month of birth). However, the bulk of this evidence is for very young children in kindergarten and elementary school and it is not clear that these findings generalize to older ages relevant to the labor market Most similar methodologically to this paper are papers by Crawford, Dearden and Meghir (2007) and Cascio and Lewis (2006), both of which rely on multiple sources of variation to identify the effect of school starting age on children’s test scores. Crawford, Dearden and Meghir (2007) use the fact that there is variation in school starting age across local education authorities (LEAs) in Britain to separately identify the effect of school starting age from age at test effects on in-school tests. While some LEAs have only one entry point (with one cutoff date), other LEAs have two entry points (with some