«Kasey S. Buckles, University of Notre Dame Elizabeth L. Munnich, University of Notre Dame Abstract This paper investigates the effect of the age ...»
Birth Spacing and Sibling Outcomes
Kasey S. Buckles, University of Notre Dame
Elizabeth L. Munnich, University of Notre Dame
This paper investigates the effect of the age difference between siblings (spacing) on
educational achievement. We use a sample of women from the 1979 NLSY, matched to reading
and math scores for their children from the NLSY79 Children and Young Adults Survey. OLS
results suggest that greater spacing is positively associated with test scores for older siblings, but not for younger siblings. However, because we are concerned that spacing may be correlated with unobservable characteristics, we also use an instrumental variables strategy that exploits variation in spacing driven by miscarriages that occur between two live births. The IV results indicate that a one-year increase in spacing increases test scores for older siblings by about 0.17 standard deviations—an effect comparable to estimates of the effect of birth order. Especially close spacing (less than two years) decreases scores by 0.65 SD. These results are larger than the OLS estimates, suggesting that estimates that fail to account for the endogeneity of spacing may understate its benefits. For younger siblings, there appears to be no causal impact of spacing on test scores.
I. Introduction A large body of work in economics and other disciplines has found a relationship between family structure and children’s outcomes. For example, children from larger families generally have lower educational attainment, lower IQ scores, worse employment outcomes, and are more likely to engage in risky behavior (Kessler 1991; Hanushek 1992; Steelman et al. 2002;
Deschenes 2007; Black, Devereux and Salvanes 2010). A recent literature in economics has considered the effects of birth order and found that later-born children have lower educational attainment, receive less parental time investment, and in some cases have worse labor market outcomes (Black, Devereux, and Salvanes 2005; Price 2008). There is even evidence that the gender composition of one’s siblings affects educational attainment, though results are mixed (Butcher and Case 1994; Kaestner 1997; Hauser and Kuo 1998; Dahl and Moretti 2008).
However, the age difference between siblings (spacing) has received much less attention in the economic literature—despite the fact that child spacing “may well be the most important aspect of fertility differentials in low-fertility societies” (Wineberg and McCarthy 1989). The research that exists in other fields has focused primarily on the effect of small gaps (less than two years), and on very early outcomes such as birth weight and infant mortality. In this paper, we investigate the effects of birth spacing on one important later-life outcome: academic achievement as measured by performance on the Peabody Individual Achievement Tests for math and reading. Our focus on later outcomes is especially valuable given that many of the possible effects of spacing (described more in Section II) would occur after birth, meaning that studies focusing on perinatal outcomes could find effects that differ from long-run effects.
Evidence of the effect of spacing on later outcomes would add to our understanding of the effects of family structure. In fact, some of the hypothesized mechanisms for birth order effects, such as differential parental investments, could be mitigated by spacing (Zajonc 1976).
Furthermore, unlike birth order, spacing is a matter over which parents might have some control.
Empirical evidence of a causal effect of gap size on children’s outcomes would be helpful for parents making decisions about the timing of their fertility. 1 Additionally, policy makers in both developed and developing countries have advocated greater spacing between births as a means of improving maternal and infant health. For example, the Contra Costa County Health Services Department in California conducted a public health campaign in 2007, which encouraged greater spacing with the slogan “Just Us for Two Years” (Contra County Health Services 2007). Similarly, the United States Agency for International Development (USAID) has issued a policy brief stating that greater spacing is one of the best ways for women to achieve healthy pregnancies and safe births, citing evidence that “three to five saves lives” (USAID 2006). Programs informing women about the benefits of greater spacing have been implemented in countries including Nigeria, Zimbabwe, and Bangladesh (Olukoya 1986; Guilkey and Jayne 1997; Jamison et al. 2006). However, these policies may have unintended consequences (either positive or negative) if spacing affects outcomes beyond maternal and infant health. 2 Referring to birth spacing, Christensen (1968) notes that “Parents and prospective parents debate these questions, while at the same time being exposed to advice from physicians and varieties of child specialists. Obstetricians, with a primary concern for the mother's health, tend to recommend spacing intervals of from two to three years. Pediatricians and child development specialists look more toward what is best for the health and development of the offspring, but their counsel with reference to spacing seems less consistent.” Other policies may affect spacing indirectly; for example, Lalive and Zwiemüller (2005) show We begin by using OLS to estimate the relationship between spacing and academic achievement, using the sample of women with multiple children in the 1979 National Longitudinal Survey of Youth (NLSY79). We observe the spacing between each sibling pair, and match the data to detailed information about the siblings from the NLSY79 Children and Young Adults survey. We perform the analysis separately for the older and younger sibling in each pair. The OLS results indicate that longer gaps are associated with slightly better test scores for older children, while for younger children there is little relationship.
However, as Rosenzweig (1986) observes, estimation techniques that fail to account for within- and across-family heterogeneity in unobservable characteristics could produce biased estimates of the effects of birth spacing. Therefore, we also use an instrumental variables strategy to identify the causal effect of spacing on sibling outcomes. The identification strategy exploits variation in spacing driven by miscarriages that occur between two live births; there are several caveats to consider when using this instrument, which will be discussed in detail in Section V. We show that a miscarriage between siblings is associated with an increase in spacing of about eight months, and decreases the likelihood that the siblings are less than two years apart by 19 percentage points.
The results using miscarriages as an instrument indicate that an increase in spacing of one year increases reading scores for the older sibling by 0.17 standard deviations (SD). This effect is comparable to estimates of the effect of birth order on IQ scores and larger than estimates of the effect of decreasing family size by one. 3 Spacing of less than two years decreases reading that an Austrian policy that increased paternal leave from one to two years increased the likelihood that a woman had another child within three years by fifteen percent.
Estimates of the effect of birth order on IQ scores range from 0.2 (Black, Devereaux, and scores by 0.65 SD; estimates for math scores are similar. The two-stage least-squares (2SLS) results are much larger than those obtained by OLS, suggesting that estimates that fail to account for the endogeneity of spacing may understate its benefits. We find no evidence of an effect of spacing on test scores for younger siblings.
II. Birth Spacing: Background A. Previous Research Social scientists have long been interested in the effects of birth spacing. Much of the research in sociology is built on the confluence model presented by Zajonc and Markus (1975), in which family size and birth order influence the intellectual environment of a household.
Zajonc (1976) argues that the effects of birth order “are mediated entirely by the age spacing between siblings” and that greater spacing between siblings can reverse the negative effects of birth order. The argument is that children born into families with older children are born into more favorable intellectual environments. In this model, larger gaps may also positively affect first-born children, who have more time to develop before the birth of an “intellectually immature” younger sibling. Empirical evidence is provided by Broman et al. (1975), who find that children born after longer intervals scored higher on the Stanford-Binet intelligence scale than those born after shorter intervals. However, Galbraith (1982) finds that sibling spacing was not related to intellectual development in a sample of college students.
Among economists, Rosenzweig (1986) develops a model of optimal child spacing in which spacing is an input into child quality. An important feature of the model is that the Salvanes 2007) to 0.25 SD (Bjerkedal et al. 2007). Increasing family size by one through twins decreases IQ scores by about 0.08 SD (Black, Devereaux, and Salvanes 2010).
endowments of older children affect the optimal timing of subsequent births. Empirically, he finds that having a healthier firstborn child significantly increases the likelihood of a closely spaced second child. This finding is confirmed in Rosenzweig and Wolpin (1988), who also estimate the effects of spacing using a procedure that uses lagged characteristics of parents and children as instruments. They show that greater spacing increases birth weight for younger siblings, and the effects are larger than those estimated with seemingly unrelated regression or fixed effects techniques. 4 Our paper builds on Rosenzweig and Wolpin in four ways. First, one might be concerned that lagged parental characteristics may be related to unobservable factors (such as parental tastes and abilities) that persist over time, which could affect the validity of their identification strategy. Here, we pursue a different identification strategy. Second, Rosenzweig and Wolpin’s study is based on a sample of 109 households from a village in Colombia; we have approximately 5,000 sibling pairs from a representative sample of the United States. Third, we focus on later outcomes, which may be valuable as many potential channels for a spacing effect would be realized after infancy. And finally, our strategy allows us to estimate the effect of spacing on older siblings, who are not considered by Rosenzweig and Wolpin.
B. Potential Mechanisms Birth spacing could affect child outcomes, including educational achievement, through a In other work in economics, Bhalotra and van Soest (2008) and DaVanzo et al. (2008) consider the effects of birth intervals on infant mortality in India and Bangladesh, respectively. Blalotra and van Soest estimate a structural model while DaVanzo et al. estimate a hazard model with controls for family characteristics. Both studies suggest that greater spacing reduces infant mortality.
number of channels. We now discuss several of these mechanisms, which we have organized into four broad categories.
1. Physiological Effects There is substantial evidence in the medical literature linking both short (typically less than 18 months) and long (more than 5 years) inter-pregnancy intervals to adverse infant health outcomes. 5 These include infant mortality, stillbirth, preterm delivery, and low birth weight.
Smits and Essed (2001) and van Eijsden et al. (2008) suggest nutritional depletion—in particular folate—as a mechanism through which short spacing might affect birth outcomes. On the other hand, the “physiological regression hypothesis” proposes that after long intervals, women’s reproductive capabilities regress (Zhu et al. 1999). There is also recent evidence linking spacing to conditions beyond the perinatal period. In a study of sibling pairs in California, Cheslack et al. (2011) estimate that second-born children conceived within 12 months of a previous birth have three times the odds of being diagnosed with autism than those conceived more than 36 months after a previous birth. 6 If spacing affects infant health or child development, this could produce a link between spacing and other outcomes like test scores.
2. Parental Investments Spacing may also affect parents’ investments in their children. Price finds that parents spend significantly more time with first-born than second-born children, and this translates into less time spent reading to the younger child and lower reading test scores (Price 2008, 2010).