Assessing the Impact of Birth Spacing on Child Health Trajectories - - PDF document

Assessing the Impact of Birth Spacing on Child Health Trajectories

Ray Miller Harvard Center for Population and Development Studies hrmiller@hsph.harvard.edu Mahesh Karra Boston University mvkarra@bu.edu September 29, 2017

Abstract We investigate the effect of birth spacing between siblings on child health tra- jectories and investments using longitudinal data from the Young Lives Study. We analyze a birth cohort of 8,000 children with data collected between 2002 and 2013 from four low-and middle-income countries. We find increased height among children who are more widely spaced relative to children who are narrowly spaced (within two years from an older sibling). However, we also find evidence

• f compensatory growth (converging height and HAZ scores) for closely spaced

children as they age. There is a positive association between birth spacing and prenatal investments, which suggeests that the emergence of height gaps could be driven by parental behavior in addition to biological factors. However, we find little evidence that compensatory growth is driven by remedial health in- vestments in closely spaced children after birth, suggesting catch-up growth as a biological phenomenon. Available data suggests financial constraints and care- giver perceptions of child health as possible explanations for lack of remedial investments.

JEL classifications: I10, O57 Keywords: birth spacing, health, nutrition, child investments, compensatory growth 1

1 Introduction

The importance of birth spacing for maternal and child health has been a long-standing source of interest to researchers and policymakers alike (World Health Organization, 2005). Evidence from systematic reviews and other empirical studies have suggested that short birth intervals (less than two years between births) may be associated with increased risk of maternal and child morbidity, including pregnancy-related complica- tions (high blood pressure, pre-eclampsia), preterm birth, low birthweight, and small for gestational age, as well as increased risk of mortality for both women and children (Conde-Agudelo et al., 2006; DaVanzo et al., 2004; Winikoff, 1983). Some studies have also examined the relationship between birth spacing and longer-term cognitive and educational outcomes in children and have found longer intrapartum spacing to be associated with improved school test scores in older siblings, while the effects of longer spacing were found to be minimal for younger siblings (Broman et al., 1975; Buckles and Munnich, 2012). By the same token, studies have also begun to explore the role

• f birth order on child health and socioemotional development and have found that
• lder siblings are likely to be more socially outgoing and persistent than their younger

siblings (Black et al., 2016). When pooling the evidence together, however, results from this body of research have concluded that many of the findings on both birth spacing and birth order, par- ticularly those related to child morbidity and adverse health outcomes, are either weak

• r mixed (Dewey and Cohen, 2007). Moreover, to our knowledge, no studies have in-

vestigated whether adverse health outcomes associated with short intrapartum spacing persist in older children, especially as they transition into adolescence. In this study, we investigate the effect of birth spacing between siblings on child health outcomes (height and weight) using longitudinal data that was collected on a cohort of children and their sibling in four low-and middle-income countries. As part of our analysis, we assess the impact of short and long birth intervals for the pooled sample of children and separately by country, and we also assess whether and how this estimated impact changes for the sample as children aged.

2 Background

The relationship between short birth intervals and high infant and child mortality has been well established in a wide range of populations (DaVanzo et al., 2004). On the other hand, there is very little empirical evidence that directly assesses the links between birth intervals and child morbidity, which is surprising considering the mech- anisms through which birth intervals may operate to influence child health and well- being have been extensively discussed in the literature (DaVanzo et al., 1983; Miller, 1991). In particular, the consequences of a short birth interval for child health out- comes, particularly at younger ages, have often been attributed to the physiological effects related to the “maternal depletion syndrome,” which postulates that the woman 2

may not have fully recuperated from one pregnancy before supporting the next one (Conde-Agudelo et al., 2012; Dewey and Cohen, 2007). Other mechanisms that have been hypothesized to contribute to a detrimental effect of a short preceding interval include: 1) behavioral effects that are associated with competition between siblings, which may include competition for parental time or resources; 2) depleted parental investments or household resources that were used for the earlier birth, which may include a lack of physical resources or even a psychological or emotional inability to provide the later child with adequate attention if its birth came sooner than desired; and 3) larger morbidities through higher disease transmission among closely spaced siblings, particularly at younger ages (DaVanzo et al., 2004; Conde-Agudelo et al., 2012). Several recent studies have examined the extent to which birth order may contribute to child height and weight, nutritional status, and other measures of child development and attribute differences in health outcomes between siblings to sibling competition, gender bias (son preference), and resource constraints (Jayachandran and Pande, 2015; Black et al., 2016; Nuttall and Nuttall, 1979). However, none of these studies directly estimate the extent to which birth intervals play a role and, in the best of analyses,

• nly control for birth interval effects by examining the effects of birth order among

populations in which siblings are similarly spaced. The closest approximation of po- tential child morbidity effects from birth spacing is provided by studies that examine the relationship between indicators of childhood malnutrition (stunting, wasting, un- derweight) and family formation patterns, but evidence of the effect from this literature was found to be mixed (Winikoff, 1983). A more recent systematic review by Dewey and Cohen (2007) assessed the evidence on the effects of birth spacing on child nutri- tional status from 52 studies and noted that approximately half (25) of these studies found that a previous birth interval of at least 36 months was associated with a 10 to 50 percent reduction in childhood stunting (with similar findings for wasting), whereas the remaining studies either found no association or were inconclusive. A study by Rutstein (2008), which pooled birth history data from 52 Demographic and Health Surveys (DHS) that were conducted from 2000 to 2005, observed a positive association between birth interval length and child nutritional status outcomes. Similarly, a more recent study by Fink et al. (2014), which pooled 153 cross-sectional DHS surveys across 61 countries conducted between 1990 and 2011, found that birth intervals of less than 12 months and between 12 and 23 months were associated with higher relative risks for stunting (relative risks of 1.09 and 1.06, respectively) as compared to a 24–35 month inter-pregnancy interval. Due to the cross-sectional nature of the data, however, both the Rutstein (2008) and the Fink et al. (2014) studies were limited in their ability to make inferences on the persistence of these associations in children over time. Our study aims to address the methodological shortcomings in the literature in two ways. Firstly, we improve on prior methodologies that have almost exclusively relied on cross-sectional data by exploiting a longitudinal dataset that allows us to effectively observe trends of the estimated birth spacing effect across cohorts and over

• time. Secondly, we employ a secondary statistical model that relies on within-family

3

sibling comparisons of birth spacing for identification. This approach serves to minimize residual confounding by adjusting for all time-invariant factors that remain constant within the family.

3 Data and methods

3.1 Data

For our analysis, we use longitudinal data from the Young Lives Study (YLS), which is an international study that aims to investigate the determinants of childhood poverty and well-being (Oxford Department of International Development, 2017). As part

• f the YLS, detailed health, nutrition, education, and other sociodemographic data

is collected on a cohort of children born between 2001 and 2002 from four low- and middle-income countries, Ethiopia, India, Peru, and Vietnam. The sampling design of the YLS included selecting 20 communities in each country and randomly selecting 100 children from each. Currently, data is available on approximately 8,000 children (2,000 from each country) over four survey waves that were conducted in 2002, 2006, 2009, and 2013, when children were approximately one, five, eight, and twelve years old. The study also collected information on household and child characteristics in each survey wave, including the anthropometric markers height and weight. Beginning in the third survey wave, anthropometric markers were also collected for a sibling of the primary cohort of children. We derive birth spacing and order by combining the household rosters collected during each wave of the survey. Date of birth was collected directly for the primary cohort and for siblings with anthropometric data. For the remaining siblings who were reported on the household rosters, reported age in years was combined with the date

• f interview to calculate an approximate date of birth. For those with an older sibling,

we group spacing to next oldest sibling into four categories: spaced under 2 years, spaced 3-4 years apart, spaced 4-7 years apart, and spaced 7 years or more apart. Birth order is top coded at having 3 or more older siblings. A number of household and child characteristics were also used in analyses to help control for demographic and socioeconomic effects on child health outcomes. These included sex, age in months at measurement, age squared, mother’s age at birth, mother’s age at birth squared, a wealth index, total number of siblings, and caregiver’s education. To construct the analytic sample, we first pool observations from all four survey waves of the YLS from each of the four countries for the primary cohort and their siblings with anthropometric data. This gives us a total of 41,596 observations. Of this sample, 1.7 percent of observations were dropped due to missing data on birth spacing and another 6.1 percent were dropped due to missing household or child characteristics. This leaves us with a pooled sample of 38,368 observations—29,352 observations from the primary cohort of YLS children and 9,016 observations from their siblings. 4

3.2 Outcomes

We focus on three primary child health outcomes—height (cm), height-for-age, and

• stunted. Height-for-age, which is measured by a child’s height-for-age z-score (HAZ),

captures a child’s restricted growth potential associated with the chronic or long-term effects of undernourishment. Stunted is a binary indicator that is defined by a child’s HAZ falling below two standard deviations from the WHO MGRS reference median height. In addition to our primary health outcomes, we are also interested in understanding the underlying mechanisms driving the results of our analysis. To this end, we examine the effect of birth spacing on additional measures related to parental investments in

• children. These include outcomes on prenatal and birth investments including level of

prenatal care1 and indicators for place of delivery (home birth), presence of a medical professional at birth, and whether the pregnancy was wanted. We also examine birth weight and indicators for premature birth and cesarean section. In order to understand parental investment responses to health over childhood, we also examined child weight-for-height (WFH). Weight-for-height—weight in grams divided by height in centimeters—is a measure associated with acute nutrition as it is more sensitive to short-term health inputs and environment. Lastly, we examine the variety and frequency of meals and parent’s perceptions of child health as a means to provide additional insight into parental investment behavior.

3.3 Panel analysis

We are primarily interested in examining the effects of birth spacing on child health and in the change in the birth spacing effect as children age. We exploit the panel structure of the YLS in addressing this question. First, we limit the sample to the primary cohort of children and estimate the following OLS model separately for each wave of the survey: Yi =

K

• k=2

δkI (Spacei = k) +

L

• l=2

αlI (Olderi = l) + ζFBi +

N

• n=2

γnI (Comi = n) +

M

• m=2

λmI (Y OBi = m) +

Z

• z=2

ηzI (Seai = z) + βXi + ui, (1) where Yi is an outcome for individual i; Spacei is years to next oldest sibling (being spaced under 2 years apart is the reference group); Olderi is the number of older siblings (having one older sibling is the reference group); FBi is a dummy for being

1We used the level of antenatal care variable provided in the first round data set. The YLS study

team constructed the variable as follows: a mother that reported no antenatal care was given a zero. For those who had antenatal visits, one was added if the first visit was when they were four months pregnant or before, one was added if the mother had five or more visits in total, and one was added if the mother was given tetanus injections. This gave a value between zero and three for all mothers.

5

the first-born child; Comi is community of residence; Y OBi is year of birth; Seai is season of birth2; Xi is a vector of child-specific characteristics; and ui is the random error term. The coefficient of interest is that on birth spacing, δ. This approach allows for comparison of effects at ages 1, 5, 8, and 12 estimated longitudinally for a single birth cohort. Interpretation of the coefficient of interest requires careful consideration. Effects estimated from this model can only be interpreted as causal if birth spacing and order are uncorrelated with any unobserved determinants of examined outcomes. It is clearly the case that geographic residence is likely to be correlated with both health outcomes and birth spacing, as access to family planning and other health services vary consid- erably across countries and locales. However, effects associated with geographic area are controlled for with the inclusion of community fixed effects. An additional concern is the existence of seasonal patterns of fertility that correlate with our independent variables of interest. If, for example, the pregnancies associated with shorter birth intervals were correlated with times of the year when food was relatively scarce, then results could be attributed to season of birth as opposed to birth spacing (e.g. Moore et al., 1999, 2004; Rayco-Solon et al., 2005; McEniry, 2011; Lokshin and Radyakin, 2012; Miller, 2017). Moreover, studies have documented seasonal patterns of fertility across a variety of countries (e.g. Rajagopalan et al., 1981; Panter-Brick, 1996; Buckles and Hungerman, 2013). However, the inclusion of month by country of birth dummies controls for seasonal effects that occurred at the country level and were independent

• f birth spacing.

3.4 Pooled analysis

While our panel analysis controls for many individual and household characteristics, it is still conceivable that fertility patterns could correlate with additional unobserved characteristics of children or their families. To address this concern, we examine the robustness of our results by running a family fixed effects model specification using the pooled sample of YLS children and their siblings. Specifically, we estimate the following model: Yifs =

K

• k=2

δkI (Spaceif = k) +

L

• l=2

αlI (Olderif = l) + ζFBi +

M

• m=2

λmI (Y OBif = m) +

Z

• z=2

ηzI (Seaif = z) + βXifs + µs + θf + uifs, (2) where Yifs is an outcome for individual i from family f measured in survey round s; µs is a survey round fixed effect; θf is a family fixed effect; and other independent variables are as previously defined. Due to collinearity with the family fixed effect, we drop the household wealth index, total number of siblings, and caregiver’s education from the

2Season of birth is controlled for with a month of birth by country dummy.

6

vector of individual characteristics, Xifs. This approach controls for any remaining permanent unobserved correlation between a child’s family and the spacing measures by comparing children within the same family. The primary cost of this approach is the loss of the panel structure to examine effects as children age. However, we also estimate the pooled analysis model with an interaction between birth spacing and age to compare with the results of our panel analysis.

4 Results

4.1 Descriptive statistics

Table 1 presents descriptive statistics for the sample used in our primary panel analysis. The panel sample included 7,375 individuals and 29,352 observations. Average height was quite low, with 27 percent of the sample classified as stunted. About 9 percent

• f the sample were spaced within two years of an older sibling, and 8 percent were

spaced seven or more years apart. The total number of siblings averaged 2.1, and 41 percent of the sample were first-born children. The sample was spread evenly over the four countries included in the YLS. Descriptive statistics for siblings used in the pooled analysis are given in Appendix Table A1. In total, data is available for 4,694 unique sibling pairs for the pooled analysis.

4.2 Main outcomes

We exploit the panel structure of the YLS dataset by estimating the evolution of child health outcomes for the primary birth cohort over time. The estimated association between birth spacing and height from OLS regressions at each age are presented in Figure 1 (point estimates are shown in appendix Table A2). By examining the estimates from each of the four time periods in sequence, we can observe how the associations between child health trajectories and birth spacing change for the cohort over time. As Figure 1 shows, wider spacing to next oldest sibling was associated with increased height at age one, though the relationship is only statistically significant at the 95% level for spacing greater than three years. For example, relative to being spaced less than two years apart from one’s next oldest sibling, being spaced 4-7 years apart is associated with an increase in height of 0.82 cm at age 1—about 17.5 percent of the standard deviation of age 1 height in the sample. However, the magnitude of the associations between birth spacing and child height trajectories, both in terms of raw height and HAZ, declines over time. Moreover, the effect of birth spacing only statistically persists to age 12 for children who are very widely spaced (at least 7 years apart). Taken together, the observed attenuation of birth spacing effects provides evidence of potential catch-up growth among more narrowly spaced children. We also run an analysis using child stunting as an outcome, which allows us to com- pare findings from our cohort analysis with the existing estimates from the literature. We re-estimate equations (1) using a logit regression (results are presented in Table 2). 7

Table 1: Descriptive statistics

N Mean SD Min Max Height (cm) 29086 109.53 26.33 54.70 174.20 HAZ 29086

• 1.33

1.17

• 6.49

3.00 Stunted 29086 0.27 0.44 0.00 1.00 Preceding birth interval <2 years 29352 0.09 0.29 0.00 1.00 2-3 years 29352 0.15 0.36 0.00 1.00 3-4 years 29352 0.11 0.31 0.00 1.00 4-7 years 29352 0.16 0.36 0.00 1.00 7+ years 29352 0.08 0.27 0.00 1.00 First-born child 29352 0.41 0.49 0.00 1.00 Older siblings 1 29352 0.30 0.46 0.00 1.00 2 29352 0.14 0.34 0.00 1.00 3+ 29352 0.15 0.36 0.00 1.00 Male 29352 0.52 0.50 0.00 1.00 Mom’s age at birth 29352 25.54 6.11 11.00 53.00 Age (months) 29352 78.99 48.36 0.53 164.00 Wealth index 29352 0.53 0.21 0.00 1.01 Total siblings 29352 2.07 1.76 0.00 11.00 Caregiver’s edu. 29352 4.98 4.54 0.00 17.00 Ethiopia 29352 0.24 0.43 0.00 1.00 India 29352 0.25 0.43 0.00 1.00 Vietnam 29352 0.26 0.44 0.00 1.00 Peru 29352 0.25 0.43 0.00 1.00 Observations 29352 Individuals 7375

Source: Young Lives Study, young cohort. Sample of children with non-missing child characteristic covariates.

We estimate that for children who are spaced within two years of an older siblings, the

• dds of stunting at age 1 are 22.7 percent higher than those children who are spaced

2-3 years apart and are 75 percent higher than those children who are spaced 7 or more years apart, respectively. We again find that the spacing estimates attenuate and become statistically insignificant as children become older.

4.3 Prenatal and childhood investments

We run a secondary analysis using alternate measures of prenatal and childhood in- vestments as dependent variables. This analysis provides some suggestive evidence on if observed effects are operating entirely through biological channels (e.g. “maternal depletion syndrome”) or if there are behavioral aspects at play as well (e.g. competition for resources). Table 3 presents the association between birth spacing and prenatal investments in 8

• .5

.5 1 1.5 2

Effect on Height

2-3 3-4 4-7 7+

Years to Next Oldest Sibling

Age1 Age5 Age8 Age12

• .2

.2 .4 .6

Effect on HAZ

2-3 3-4 4-7 7+

Years to Next Oldest Sibling

Age1 Age5 Age8 Age12

Figure 1: Birth spacing and child height Table 2: Birth spacing and stunting

Age 1 Age 5 Age 8 Age 12 Space 2-3 0.815∗∗ 0.929 0.974 1.100 (0.081) (0.096) (0.124) (0.123) Space 3-4 0.639∗∗∗ 0.756∗∗ 0.775∗ 0.905 (0.074) (0.097) (0.112) (0.114) Space 4-7 0.592∗∗∗ 0.778∗∗ 0.961 0.979 (0.069) (0.090) (0.127) (0.111) Space 7+ 0.571∗∗∗ 0.676∗∗∗ 0.672∗∗ 0.869 (0.089) (0.101) (0.114) (0.128) Obs 7193 7305 7299 7285 Odds ratios reported. Robust standard errors (clustered at the community level) in parentheses, p-values—*** p<0.01, ** p<0.05, * p<0.1. Additional independent vari- ables in all regressions: mother’s age at birth, mother’s age at birth squared, age (months), age squared, wealth index, number of siblings, caregiver’s education, and dum- mies for first-born, older siblings, sex, survey round, year and season of birth, and community.

the child and subsequent birth outcomes. Wider spaced children received more pre- natal care, were less likely to born at home, and were more likely to have a medical professional present at birth. Moreover, they were larger at birth, less likely to be pre- mature, and more likely to be from a reportedly wanted pregnancy. The differences in prenatal care suggests the effects of birth spacing could be partially driven by parental investment behavior. Table 4 presents the association between birth spacing and alternate nutritional investment measures over childhood. Wider birth spacing was associated with increased weight-for-height across all spacing groups. However, unlike height measures, this association did not attenuate with age. More direct measures of nutritional inputs based on the frequency and variety of meals in the past 24 hours did not reveal a strong statistical relationship with birth spacing (save meal variety for those children 9

Table 3: Prenatal investments and birth outcomes

Prenatal care Wanted C-sect Premature Pro at birth Home birth Birthweight Space 2-3 1.110 1.229∗ 0.781 0.844 1.018 0.818 73.364∗ (0.100) (0.149) (0.157) (0.133) (0.135) (0.108) (40.819) Space 3-4 1.179 1.766∗∗∗ 0.785 0.756∗ 0.990 0.855 83.509∗∗ (0.122) (0.256) (0.196) (0.125) (0.163) (0.135) (41.054) Space 4-7 1.234∗∗ 2.250∗∗∗ 0.777 0.829 1.660∗∗∗ 0.536∗∗∗ 63.208∗ (0.123) (0.345) (0.177) (0.140) (0.259) (0.076) (37.084) Space 7+ 1.434∗∗∗ 2.645∗∗∗ 0.950 1.213 1.893∗∗∗ 0.478∗∗∗ 57.044 (0.168) (0.447) (0.221) (0.210) (0.355) (0.083) (40.284) Obs 7104 7153 4024 7074 6586 7092 4410 Odds ratios reported (except for birthweight). Robust standard errors (clustered at the community level) in parentheses, p-values—*** p<0.01, ** p<0.05, * p<0.1. Additional independent variables in all regressions: mother’s age at birth, mother’s age at birth squared, age (months), age squared, and dummies for first-born,

• lder siblings, sex, survey round, year and season of birth, and community.

who are spaced for 7 or more years apart). However, point estimates were almost all positive suggesting, if anything, wider spaced children received more nutritional inputs around the time of each survey. Overall, weight-for-height and meal based measures show no evidence that closely spaced children received more nutritional investments

• ver childhood to allow them to catch up in height.

Table 4: Childhood nutritional investments

Weight-for-height Meal frequency Meal variety Age 1 Age 5 Age 8 Age 12 Age 5 Age 8 Age 12 Age 5 Age 8 Age 12 Space 2-3 1.694∗∗∗ 2.114∗∗∗ 2.602∗∗∗ 3.169∗∗ 0.021 0.017

• 0.024

0.123 0.054

• 0.072

(0.584) (0.629) (0.855) (1.460) (0.055) (0.039) (0.040) (0.082) (0.079) (0.112) Space 3-4 2.112∗∗∗ 1.362∗ 2.576∗∗∗ 3.627∗∗ 0.011 0.002

• 0.011

0.049 0.048

• 0.098

(0.548) (0.718) (0.930) (1.816) (0.057) (0.043) (0.046) (0.077) (0.078) (0.124) Space 4-7 2.176∗∗∗ 1.898∗∗∗ 2.862∗∗∗ 5.439∗∗∗ 0.069 0.009 0.045 0.069 0.106 0.057 (0.504) (0.690) (0.847) (1.593) (0.058) (0.034) (0.043) (0.079) (0.080) (0.128) Space 7+ 3.066∗∗∗ 2.542∗∗∗ 3.331∗∗∗ 6.861∗∗∗ 0.091 0.079 0.030 0.254∗∗∗ 0.224∗∗ 0.008 (0.688) (0.763) (1.231) (2.366) (0.059) (0.051) (0.057) (0.088) (0.094) (0.148) Obs 7114 7304 7287 7284 7339 6769 7266 7337 7329 7267 Robust standard errors (clustered at the community level) in parentheses, p-values—*** p<0.01, ** p<0.05, * p<0.1. Additional independent variables in all regressions: mother’s age at birth, mother’s age at birth squared, age (months), age squared, wealth index, number of siblings, caregiver’s education, and dummies for first-born, older siblings, sex, survey round, year and season of birth, and community.

Table 5 presents the association between birth spacing and caregiver perceptions

• f child size. Caregivers perceived wider spaced children to be larger at birth, though

the relationship is only statistically significant for those children who are spaced more than 4 years apart. Similarly, there is some evidence that caregivers believed wider spaced children to be heavier and (to a lesser extent) taller than other children at age

• 1. However, this relationships become weaker by age 5, particularly for height.

10

Table 5: Caregiver perceptions of child size

Birthsize Weight Height Age 1 Age 5 Age 1 Age 5 Space 2-3 1.155 1.120 1.148 1.035 0.882 (0.116) (0.098) (0.105) (0.089) (0.073) Space 3-4 1.107 1.101 1.103 1.084 0.965 (0.112) (0.108) (0.110) (0.092) (0.103) Space 4-7 1.215∗∗ 1.198∗ 1.025 1.187∗∗ 0.945 (0.115) (0.110) (0.093) (0.098) (0.085) Space 7+ 1.338∗∗∗ 1.327∗∗ 1.119 0.935 1.043 (0.148) (0.165) (0.135) (0.104) (0.117) Obs 7270 7313 7318 7303 7327 Odds ratios reported. Robust standard errors (clustered at the community level) in parentheses, p-values—*** p<0.01, ** p<0.05, * p<0.1. Additional independent variables in all regressions: mother’s age at birth, mother’s age at birth squared, age (months), age squared, wealth index, number of siblings, caregiver’s educa- tion, and dummies for first-born, older siblings, sex, survey round, year and season of birth, and community.

4.4 Heterogeneity

The appendix presents results for panel analysis separately for each of the four countries in the YLS (Tables A4 to A7). Findings from these analyses indicate that there is heterogeneity across countries, both in terms of the magnitude of the effect of birth spacing as well as the persistence of the effect in the country-specific cohorts over time. Overall, effects are strongest for Ethiopia followed by India and Vietnam. Effects are generally smallest in Peru. As expected, the confidence intervals around most of the country-specific coefficient estimates are wider, particularly among those children who are more widely spaced. Appendix Tables A8 and A9 present results when conducting analyses on sex- specific sub-samples. Overall, spacing effects are stronger for females, while males exhibited faster catch-up growth. Again, the confidence intervals are wider due to the smaller sample size. Finally, Table 6 provides results for height and weight-for-height when adding an interaction between spacing indicators and caregiver education. For children who are spaced between 2-7 years, the point estimates are all negative, which would imply that birth spacing was associated more strongly with increased child height and weight- for-height for less educated caregivers. The point estimates even suggest that spacing is negatively associated with weight-for-height at some ages for highly educated care-

• givers. For example, for an otherwise average caregiver with 10 years of education,

the implied association between age five weight-for-height and spacing of 2-3 years is −1.933 = 4.447 − 10 × 0.638. 11

Table 6: Heterogeneity by caregiver education

Height WFH Age 1 Age 5 Age 8 Age 12 Age 1 Age 5 Age 8 Age 12 Space 2-3 0.444∗∗ 0.400 0.630∗∗ 0.714∗∗ 2.556∗∗∗ 4.557∗∗∗ 5.954∗∗∗ 7.701∗∗∗ (0.195) (0.253) (0.293) (0.345) (0.608) (0.748) (1.047) (1.699) Space 3-4 1.008∗∗∗ 0.787∗∗ 0.905∗∗ 0.875∗∗ 2.289∗∗∗ 2.683∗∗∗ 4.727∗∗∗ 5.817∗∗∗ (0.253) (0.327) (0.366) (0.417) (0.685) (0.877) (1.121) (2.124) Space 4-7 1.148∗∗∗ 0.919∗∗∗ 0.817∗∗∗ 0.741∗ 2.496∗∗∗ 3.443∗∗∗ 4.733∗∗∗ 6.877∗∗∗ (0.238) (0.250) (0.305) (0.383) (0.572) (0.764) (0.934) (1.789) Space 7+ 0.928∗∗∗ 0.745∗∗ 0.752 0.725 3.102∗∗∗ 1.544 3.249∗ 4.418 (0.301) (0.356) (0.542) (0.586) (0.987) (1.128) (1.861) (3.231) Space 2-3 x Edu

• 0.044∗
• 0.071∗
• 0.098∗∗
• 0.162∗∗∗
• 0.223∗∗
• 0.638∗∗∗
• 0.870∗∗∗
• 1.194∗∗∗

(0.025) (0.036) (0.048) (0.055) (0.110) (0.146) (0.217) (0.334) Space 3-4 x Edu

• 0.042
• 0.053
• 0.039
• 0.119∗
• 0.036
• 0.335
• 0.556∗
• 0.561

(0.031) (0.053) (0.055) (0.062) (0.111) (0.213) (0.320) (0.475) Space 4-7 x Edu

• 0.071∗∗∗
• 0.075∗∗
• 0.103∗∗
• 0.109∗∗
• 0.072
• 0.344∗∗∗
• 0.422∗∗
• 0.324

(0.023) (0.031) (0.039) (0.051) (0.081) (0.116) (0.175) (0.303) Space 7+ x Edu

• 0.006

0.032 0.057 0.072

• 0.022

0.104

• 0.072

0.321 (0.035) (0.044) (0.059) (0.067) (0.143) (0.152) (0.286) (0.478) Obs 7197 7305 7299 7285 7114 7304 7287 7284 Robust standard errors (clustered at the community level) in parentheses, p-values—*** p<0.01, ** p<0.05, * p<0.1. Additional independent variables in all regressions: mother’s age at birth, mother’s age at birth squared, age (months), age squared, wealth index, number of siblings, caregiver’s education, and dummies for first-born, older siblings, sex, survey round, year and season of birth, and community.

4.5 Pooled analysis

Results from the pooled analysis are presented in Table 7. The first columns shows results for the pooled sample without the inclusion of a family fixed effect. Relative to being spaced within 2 years of one’s next oldest sibling, being spaced 2 to 3 years apart is associated with a 0.43 cm increase in a child’s height, while being spaced at least 7 years apart is associated with a 1.42 cm increase. The second column shows results when the the family fixed effect is added to the previous model specification. There is a moderate increase in coefficient estimates when moving from the simple OLS to the family fixed effect specification. For example, being spaced 4-7 years of an older sibling is associated with a 1.18 cm increase in a child’s height under the family fixed effects analysis (compared to 0.78 cm under the pooled OLS analysis). However, the confidence intervals around these estimates are also wider than those from simple OLS, particularly for larger birth intervals where there are generally fewer

• bservations. Columns three and four show a similar pattern holds for HAZ scores.

Column five shows that spacing continues to have a strong positive association with weight-for-height using the pooled sample. However, in contrast to height measures, weight-for-height declines with the inclusion of the family fixed effect (see column six). The final two columns in Table 7 add an interaction between spacing indicators and child age to the pooled analysis with height as the outcome. There is strong negative interaction for the spacing groups between 2-7 years and a positive interaction for those spaced greater than 7 years. This is broadly consistent with our panel analysis suggesting that closely spaced children catch up in height to those who are more widely spaced as they age. 12

Table 7: Pooled sample: OLS vs. family fixed effect model

Height Height HAZ HAZ WFH WFH Height Height Space 2-3 0.430∗∗ 0.728∗∗∗ 0.084∗∗ 0.115∗∗∗ 2.747∗∗∗ 1.915∗∗ 1.475∗∗∗ 1.967∗∗∗ (0.176) (0.233) (0.034) (0.041) (0.664) (0.962) (0.192) (0.267) Space 3-4 0.958∗∗∗ 1.310∗∗∗ 0.195∗∗∗ 0.247∗∗∗ 3.490∗∗∗ 2.705∗∗ 1.856∗∗∗ 2.532∗∗∗ (0.194) (0.285) (0.037) (0.051) (0.750) (1.137) (0.219) (0.319) Space 4-7 0.788∗∗∗ 1.184∗∗∗ 0.172∗∗∗ 0.221∗∗∗ 3.869∗∗∗ 3.158∗∗ 1.252∗∗∗ 1.914∗∗∗ (0.191) (0.322) (0.037) (0.057) (0.757) (1.295) (0.203) (0.349) Space 7+ 1.427∗∗∗ 1.640∗∗∗ 0.293∗∗∗ 0.347∗∗∗ 5.074∗∗∗ 1.098 0.641∗∗∗ 0.800 (0.246) (0.534) (0.048) (0.094) (1.046) (2.298) (0.239) (0.549) Space 2-3 x Age

• 0.012∗∗∗
• 0.013∗∗∗

(0.002) (0.002) Space 3-4 x Age

• 0.010∗∗∗
• 0.012∗∗∗

(0.002) (0.002) Space 4-7 x Age

• 0.005∗∗∗
• 0.007∗∗∗

(0.002) (0.002) Space 7+ x Age 0.010∗∗∗ 0.009∗∗∗ (0.002) (0.002) Sibling FE No Yes No Yes No Yes No Yes Clusters 7460 7460 7460 7460 7460 7460 7460 7460 Obs 37788 37788 37788 37788 37680 37680 37788 37788 Robust standard errors (clusted at the family level) in parentheses, p-values—*** p<0.01, ** p<0.05, * p<0.1. Additional independent variables in all regressions: mother’s age at birth, mother’s age at birth squared, age (months), age squared, and dummies for first-born, older siblings, sex, survey round, and year and season of birth. OLS regressions also include wealth index, number of siblings, caregiver’s education, and community dummies.

13

5 Discussion

In this study, we use longitudinal data from the Young Lives Study on approximately 8,000 individuals and 4,600 of their siblings between 2002 and 2013 from four low-and middle-income countries. We assess the impact of birth spacing on height and health investments for the primary cohort over time and for the pooled sample. We find increased child height among children who are more widely spaced. However, we also find evidence of compensatory growth (estimated height and HAZ scores that converge to the null) for closely spaced children.

5.1 Summary of Main Outcomes

• 1. Spacing effects on raw height—birth spacing is associated with an increase in

child height, but the gap closes over time (except for the widest spaced group who are spaced at least 7 years apart). These results suggest that there may be compensatory growth where children catch up as they age.

• 2. Spacing effects on height-for-age (HAZ)

(a) Closely spaced children are out-growing their projected height-for-age tra- jectory - the HAZ results show they are out-growing the trajectory of the international reference population, while the increased variance in our sam- ple shows they are out-growing sample projections. Taken together, these results suggest that closely spaced children are out-growing wider spaced children relative to their respective projections, though wider spaced chil- dren start at a higher height. (b) Combined with point (1) above, our results on HAZ show closely spaced children are actually closing the absolute gap as they age (whether they completely catch up is unclear).

5.2 Mechanisms

Cross-country heterogeneity and the finding of stronger effects and slower catch-up growth for girls suggests the effects of birth spacing could be partially driven by parental investment behavior. The differences in prenatal care further supports this hypothe- sis, however additional research is needed to further disentangle the biological and behavioral channels.

• 1. Prenatal/birth investments—spacing associated with an increase in these invest-

ments—early (emergence of) effects may be behaviorally driven to some extent (a) but we do not know if birth spacing caused these behaviors (b) have at least the investments been independently linked to poor health after birth in other studies? 14

• 2. Weight-for-height shows no evidence that closely spaced children received more

nutritional investments over childhood to allow them to catch-up in height (a) This suggests substitutability (substantial concave curvature) in health pro- duction (b) But economic theory suggest substitutability should be accompanied by compensatory investments. Why don’t we see that here? 4 possibilities

• i. Unobservable characteristics are driving the patterns of investments
• A. Don’t find strong evidence of this based on family fixed effect model
• B. However, weight-for-height falls when fixed effects are added, sug-

gesting there could be some compensating investments across sib- lings — but association between spacing and WFH is still positive

• ii. Financial constraints
• A. Compensatory growth is more pronounced for the educated, WFH

is also potentially negatively associated with spacing for the highly educated — these both support financial constraints as a viable (partial) explanation

• iii. Adverse effects of spacing are latent to the parents
• A. Some (weak) evidence that parents viewed closely spaced children

as smaller/shorter at birth and age one — which does not support this explanation

• B. However, less evidence they perceived them as shorter at age five —

perhaps bulk of remedial investment occurred between age 1 and 5 and we just don’t have data in the correct time frame?

• C. Could argue that interaction effects with education support this

explanation — more educated caregivers more aware of child health deficiencies?

• iv. Weight-for-height too blunt of a short term investment measure
• A. Looking at food investments directly also suggests no additional

investments on average — which does not support this explanation

5.3 Limitations

There are several limitations to our study. While we find considerable evidence of catch-up growth over time, we are presently unable to say for certain whether children who are closely spaced are able to completely make up any initial growth differences that they may experience from these effects. We see considerable attenuation over time in both height and height-for-age growth gaps across birth spacing groups, and we observe the trajectory over time to indicate a potential convergences of growth to- wards the null. Given the relatively short (12-year) period over which our sample is

• bserved, however, we are unable to say whether or not a convergence is assured in the

15

long run, particularly as children enter into periods of potentially rapid growth and development during adolescence. A common methodological criticism in this literature is the inability to adequately account for within-family heterogeneity in unobservable characteristics, which in turn could bias the estimates of the effects of birth spacing (Rosenzweig and Wolpin, 1988; Rosenzweig, 1986). More specifically, the impact of birth spacing on child health outcomes is likely to be driven by a wide number of in- dividual, social, and contextual determinants, and these determinants may vary from birth to birth within the same family as parents choose to differentially invest in chil- dren who may be differentially spaced (Fink et al., 2014). For these reasons, our use of a family fixed effect would not be sufficient in adjusting for any residual confounding that is driven by differential birth timing decisions.

6 Conclusions

When taken together, our results suggest that short birth intervals and high fertility are likely to contribute to poor child health and development. Interventions that aim to lower fertility and increase birth intervals, including family planning and reproduc- tive health services that reinforce healthy timing and spacing of pregnancy, therefore have the potential to substantially reduce the number of stunted children. Finally, it is important that we continue to investigate the biological, social, and behavioral mecha- nisms by which adequate birth spacing might improve health for children - a thorough understanding of these pathways is essential for the development of effective policies, programs, and evidence-based interventions that seek to address key determinants of poor child health.

7 Ethical considerations

Ethical approval for the evaluation was granted by the Harvard T.H. Chan School of Public Health Institutional Review Board (IRB), Protocol No. IRB17-0028.

8 Acknowledgments

The authors thank David Canning, Guenther Fink, Randall Kuhn, and seminar par- ticipants at the Harvard Center for Population and Development Studies and the 2017 PAA conference for their helpful comments and suggestions on the analysis.

References

Black, S. E., E. Gronqvist, and B. Ockert (2016). Born to Lead? The Effect of Birth Order on Non-Cognitive Abilities. 16

Broman, S. H., P. L. Nichols, and W. A. Kennedy (1975). Preschool IQ: Prenatal and early developmental correlates. Hillsdale, NJ: Erlbaum Associates. Buckles, K. S. and D. M. Hungerman (2013). Season of birth and later outcomes: Old questions, new answers. Review of Economics and Statistics 95(3), 711–724. Buckles, K. S. and E. L. Munnich (2012). Birth Spacing and Sibling Outcomes. Journal

• f Human Resources 47(3), 613–642.

Carneiro, P. and M. Rodrigues (2009). Evaluating the Effect of Maternal Time on Child Development Using the Generalized Propensity Score. Institute for the Study

• f Labor, 12th IZA European Summer School in Labor Economics.

Conde-Agudelo, A., A. Rosas-BermÃodez, and A. C. Kafury-Goeta (2006). Birth spacing and risk of adverse perinatal outcomes: a meta-analysis. JAMA: The Journal

• f the American Medical Association 295(15), 1809–1823.

Conde-Agudelo, A., A. Rosas-Bermudez, F. Castano, and M. H. Norton (2012). Effects

• f Birth Spacing on Maternal, Perinatal, Infant, and Child Health: A Systematic

Review of Causal Mechanisms. Studies in Family Planning 43(2), 93–114. DaVanzo, J., W. P. Butz, and J.-P. Habicht (1983). How Biological and Behavioural Influences on Mortality in Malaysia Vary during the First Year of Life. Population Studies 37(3), 381–402. DaVanzo, J., A. Razzaque, M. Rahman, L. Hale, K. Ahmed, M. A. Khan, G. Mustafa, and K. Gausia (2004). The effects of birth spacing on infant and child mortality, pregnancy outcomes, and maternal morbidity and mortality in Matlab, Bangladesh. Technical Consultation and Review of the Scientific Evidence for Birth Spacing. Dewey, K. G. and R. J. Cohen (2007). Does birth spacing affect maternal or child nu- tritional status? A systematic literature review. Maternal and Child Nutrition 3(3), 151–173. Fink, G., C. R. Sudfeld, G. Danaei, M. Ezzati, and W. W. Fawzi (2014, July). Scaling- Up Access to Family Planning May Improve Linear Growth and Child Development in Low and Middle Income Countries. PLOS ONE 9(7), e102391. Jayachandran, S. and R. Pande (2015). Why Are Indian Children So Short? The Role of Birth Order and Son Preference. Northwestern University Working Paper, Northwestern University, Evanston, IL. Knodel, J. and A. I. Hermalin (1984, October). Effects of birth rank, maternal age, birth interval, and sibship size on infant and child mortality: evidence from 18th and 19th century reproductive histories. American Journal of Public Health 74(10), 1098–1106. 17

Lokshin, M. and S. Radyakin (2012). Month of birth and childrens health in India. Journal of Human Resources 47(1), 174–203. McEniry, M. (2011). Infant mortality, season of birth and the health of older Puerto Rican adults. Social Science & Medicine 72(6), 1004–1015. Miller, J. E. (1991). Birth Intervals and Perinatal Health: An Investigation of Three

• Hypotheses. Family Planning Perspectives 23(2), 62–70.

Miller, R. (2017). Childhood health and prenatal exposure to seasonal food scarcity in Ethiopia. Moore, S. E., T. J. Cole, A. C. Collinson, E. Poskitt, I. A. McGregor, and A. M. Prentice (1999). Prenatal or early postnatal events predict infectious deaths in young adulthood in rural Africa. International Journal of Epidemiology 28(6), 1088–1095. Moore, S. E., A. J. Fulford, P. K. Streatfield, L. A. Persson, and A. M. Prentice (2004). Comparative analysis of patterns of survival by season of birth in rural Bangladeshi and Gambian populations. International Journal of Epidemiology 33(1), 137–143. Nuttall, E. V. and R. L. Nuttall (1979). Child-spacing effects on intelligence, person- ality, and social competence. The Journal of Psychology 102(1), 3–12. Oxford Department of International Development (2017). The Young Lives Study. Panter-Brick, C. (1996). Proximate determinants of birth seasonality and conception failure in Nepal. Population Studies 50(2), 203–220. Rajagopalan, S., P. Kymal, and P. Pei (1981). Births, work and nutrition in Tamil Nadu, India. In R. Chambers, R. Longhurst, and P. Arnold (Eds.), Seasonal Dimen- sions to Rural Poverty, pp. 156–162. London: Frances Pinter. Rayco-Solon, P., A. J. Fulford, and A. M. Prentice (2005). Differential effects of seasonality on preterm birth and intrauterine growth restriction in rural Africans. The American Journal of Clinical Nutrition 81(1), 134–139. Rosenzweig, M. R. (1986). Birth Spacing and Sibling Inequality: Asymmetric Infor- mation within the Family. International Economic Review 27(1), 55–76. Rosenzweig, M. R. and K. I. Wolpin (1988). Heterogeneity, Intrafamily Distribution, and Child Health. The Journal of Human Resources 23(4), 437–461. Rutstein, S. O. (2008). Further Evidence of the Effects of Preceding Birth Intervals on Neonatal, Infant, and Under-Five-Years Mortality and Nutritional Status in Devel-

• ping Countries: Evidence from the Demographic and Health Surveys. DHS Working

Paper No. 41, USAID, Washington, D.C. 18

Sekiyama, M. and R. Ohtsuka (2005, July). Significant Effects of Birth-Related Biolog- ical Factors on Pre-Adolescent Nutritional Status among Rural Sundanese in West Java, Indonesia. Journal of Biosocial Science 37(04). Silles, M. A. (2010, October). The implications of family size and birth order for test scores and behavioral development. Economics of Education Review 29(5), 795–803. Teti, D. M., L. A. Bond, and E. D. Gibbs (1986). Sibling-created experiences: Re- lationships to birth-spacing and infant cognitive development. Infant Behavior and Development 9(1), 27–42. Winikoff, B. (1983). The Effects of Birth Spacing on Child and Maternal Health. Studies in Family Planning 14(10), 231–245. World Health Organization (2005). Report of a WHO Technical Consultation on Birth

• Spacing. Technical Report, World Health Organization, Geneva, Switzerland.

19

Appendix A: Tables and Figures

Table A1: Descriptive statistics: siblngs of primary YLS cohort

N Mean SD Min Max Height (cm) 8702 125.12 20.73 70.80 182.00 HAZ 8702

• 1.43

1.22

• 6.47

2.92 Stunted 8702 0.30 0.46 0.00 1.00 Preceding birth interval <2 years 9016 0.17 0.38 0.00 1.00 2-3 years 9016 0.25 0.43 0.00 1.00 3-4 years 9016 0.18 0.39 0.00 1.00 4-7 years 9016 0.18 0.38 0.00 1.00 7+ years 9016 0.02 0.12 0.00 1.00 First-born child 9016 0.20 0.40 0.00 1.00 Older siblings 1 9016 0.39 0.49 0.00 1.00 2 9016 0.19 0.39 0.00 1.00 3+ 9016 0.23 0.42 0.00 1.00 Male 9016 0.50 0.50 0.00 1.00 Mom’s age at birth 9016 25.71 5.79 12.00 52.00 Age (months) 9016 113.15 49.44 0.16 353.28 Wealth index 9016 0.51 0.21 0.00 1.00 Total siblings 9016 2.58 1.80 1.00 11.00 Caregiver’s edu. 9016 4.09 4.36 0.00 17.00 Ethiopia 9016 0.30 0.46 0.00 1.00 India 9016 0.35 0.48 0.00 1.00 Vietnam 9016 0.19 0.39 0.00 1.00 Peru 9016 0.16 0.36 0.00 1.00 Observations 9016 Individuals 4694

Source: Young Lives Study, young cohort. Sample of children with non-missing child characteristic covariates.

20

Table A2: Birth spacing and child height

Height HAZ Age 1 Age 5 Age 8 Age 12 Age 1 Age 5 Age 8 Age 12 Space 2-3 0.268 0.124 0.252 0.088 0.105 0.028 0.048 0.013 (0.161) (0.214) (0.265) (0.331) (0.063) (0.044) (0.046) (0.047) Space 3-4 0.844∗∗∗ 0.582∗∗ 0.738∗∗ 0.422 0.343∗∗∗ 0.130∗∗ 0.132∗∗ 0.063 (0.198) (0.268) (0.314) (0.361) (0.078) (0.056) (0.055) (0.052) Space 4-7 0.824∗∗∗ 0.576∗∗∗ 0.343 0.248 0.326∗∗∗ 0.128∗∗∗ 0.062 0.037 (0.178) (0.217) (0.269) (0.304) (0.069) (0.046) (0.047) (0.043) Space 7+ 0.941∗∗∗ 0.993∗∗∗ 1.165∗∗∗ 1.259∗∗∗ 0.374∗∗∗ 0.214∗∗∗ 0.205∗∗∗ 0.182∗∗∗ (0.215) (0.283) (0.361) (0.410) (0.085) (0.059) (0.063) (0.059) Obs 7197 7305 7299 7285 7197 7305 7299 7285 Robust standard errors (clustered at the community level) in parentheses, p-values—*** p<0.01, ** p<0.05, * p<0.1. Additional independent variables in all regressions: mother’s age at birth, mother’s age at birth squared, age (months), age squared, wealth index, number of siblings, caregiver’s education, and dummies for first-born, older siblings, sex, survey round, year and season of birth, and community.

Table A3: Birth spacing and child height, never missing sub-sample

Height HAZ Age 1 Age 5 Age 8 Age 12 Age 1 Age 5 Age 8 Age 12 Space 2-3 0.281∗ 0.150 0.313 0.136 0.109∗ 0.031 0.057 0.020 (0.161) (0.207) (0.246) (0.318) (0.063) (0.043) (0.043) (0.046) Space 3-4 0.813∗∗∗ 0.564∗∗ 0.659∗∗ 0.323 0.330∗∗∗ 0.122∗∗ 0.116∗∗ 0.049 (0.200) (0.271) (0.306) (0.361) (0.079) (0.057) (0.053) (0.052) Space 4-7 0.835∗∗∗ 0.483∗∗ 0.261 0.233 0.330∗∗∗ 0.105∗∗ 0.045 0.035 (0.182) (0.217) (0.271) (0.310) (0.071) (0.046) (0.047) (0.044) Space 7+ 0.962∗∗∗ 1.053∗∗∗ 1.168∗∗∗ 1.307∗∗∗ 0.383∗∗∗ 0.223∗∗∗ 0.204∗∗∗ 0.189∗∗∗ (0.216) (0.292) (0.365) (0.411) (0.085) (0.061) (0.064) (0.059) Obs 7000 7000 7000 7000 7000 7000 7000 7000 Robust standard errors (clustered at the community level) in parentheses, p-values—*** p<0.01, ** p<0.05, * p<0.1. Additional independent variables in all regressions: mother’s age at birth, mother’s age at birth squared, age (months), age squared, wealth index, number of siblings, caregiver’s education, and dummies for first-born, older siblings, sex, survey round, year and season of birth, and community.

Table A4: Birth spacing and child height: India

Height HAZ Age 1 Age 5 Age 8 Age 12 Age 1 Age 5 Age 8 Age 12 Space 2-3 0.282

• 0.186

0.011

• 0.343

0.107

• 0.039

0.005

• 0.049

(0.281) (0.339) (0.393) (0.522) (0.110) (0.071) (0.069) (0.075) Space 3-4 1.045∗∗∗ 0.614 1.052∗∗ 0.744 0.423∗∗∗ 0.131 0.185∗∗ 0.109 (0.317) (0.428) (0.465) (0.622) (0.125) (0.089) (0.081) (0.090) Space 4-7 0.645∗ 0.179

• 0.051
• 0.323

0.257∗ 0.040

• 0.008
• 0.042

(0.319) (0.394) (0.471) (0.567) (0.128) (0.085) (0.083) (0.080) Space 7+ 0.429

• 0.166

0.236

• 0.515

0.187

• 0.033

0.040

• 0.070

(0.432) (0.535) (0.794) (0.877) (0.169) (0.110) (0.139) (0.126) Obs 1800 1829 1833 1837 1800 1829 1833 1837 Robust standard errors (clustered at the community level) in parentheses, p-values—*** p<0.01, ** p<0.05, * p<0.1. Additional independent variables in all regressions: mother’s age at birth, mother’s age at birth squared, age (months), age squared, wealth index, number of siblings, care- giver’s education, and dummies for first-born, older siblings, sex, survey round, year and season of birth, and community.

21

Table A5: Birth spacing and child height: Ethiopia

Height HAZ Age 1 Age 5 Age 8 Age 12 Age 1 Age 5 Age 8 Age 12 Space 2-3 0.643 0.250 0.917 0.379 0.248 0.054 0.162∗ 0.056 (0.378) (0.550) (0.535) (0.654) (0.146) (0.112) (0.092) (0.095) Space 3-4 1.421∗∗ 0.834 1.209 0.765 0.574∗∗∗ 0.182 0.212 0.115 (0.506) (0.659) (0.767) (0.785) (0.200) (0.137) (0.133) (0.114) Space 4-7 1.832∗∗∗ 1.108∗∗ 1.690∗∗∗ 0.817 0.713∗∗∗ 0.238∗∗ 0.297∗∗∗ 0.121∗ (0.420) (0.518) (0.573) (0.482) (0.164) (0.107) (0.100) (0.070) Space 7+ 2.202∗∗∗ 1.872∗∗ 2.558∗∗ 1.347 0.874∗∗∗ 0.399∗∗ 0.446∗∗ 0.198 (0.602) (0.801) (1.016) (0.989) (0.234) (0.169) (0.177) (0.143) Obs 1687 1772 1760 1772 1687 1772 1760 1772 Robust standard errors (clustered at the community level) in parentheses, p-values—*** p<0.01, ** p<0.05, * p<0.1. Additional independent variables in all regressions: mother’s age at birth, mother’s age at birth squared, age (months), age squared, wealth index, number of siblings, caregiver’s education, and dummies for first-born, older siblings, sex, survey round, year and season of birth, and community.

Table A6: Birth spacing and child height: Vietnam

Height HAZ Age 1 Age 5 Age 8 Age 12 Age 1 Age 5 Age 8 Age 12 Space 2-3 0.063 0.365 0.383 0.444 0.032 0.079 0.072 0.057 (0.306) (0.490) (0.674) (1.019) (0.119) (0.103) (0.119) (0.144) Space 3-4 0.676∗∗ 0.538 1.079 0.273 0.277∗∗ 0.122 0.194 0.034 (0.315) (0.538) (0.694) (0.886) (0.125) (0.114) (0.122) (0.125) Space 4-7 0.622∗∗ 0.932∗∗ 0.363 0.790 0.250∗∗ 0.200∗∗ 0.065 0.108 (0.253) (0.367) (0.520) (0.824) (0.095) (0.078) (0.092) (0.116) Space 7+ 0.743∗∗ 1.109∗∗ 0.961 1.812∗ 0.298∗∗ 0.237∗∗ 0.169 0.256∗ (0.296) (0.401) (0.607) (0.873) (0.113) (0.083) (0.107) (0.123) Obs 1904 1910 1902 1867 1904 1910 1902 1867 Robust standard errors (clustered at the community level) in parentheses, p-values—*** p<0.01, ** p<0.05, * p<0.1. Additional independent variables in all regressions: mother’s age at birth, mother’s age at birth squared, age (months), age squared, wealth index, number of siblings, caregiver’s education, and dummies for first-born, older siblings, sex, survey round, year and season of birth, and community.

Table A7: Birth spacing and child height: Peru

Height HAZ Age 1 Age 5 Age 8 Age 12 Age 1 Age 5 Age 8 Age 12 Space 2-3 0.030 0.178

• 0.135

0.151 0.021 0.039

• 0.023

0.023 (0.311) (0.350) (0.605) (0.679) (0.127) (0.075) (0.106) (0.097) Space 3-4 0.010 0.322

• 0.273
• 0.417

0.011 0.072

• 0.048
• 0.059

(0.272) (0.555) (0.549) (0.654) (0.108) (0.116) (0.095) (0.093) Space 4-7 0.064 0.073

• 0.602
• 0.381

0.045 0.018

• 0.104
• 0.054

(0.294) (0.422) (0.486) (0.672) (0.116) (0.090) (0.084) (0.097) Space 7+ 0.428 1.413∗ 1.417∗ 2.046∗∗ 0.168 0.298∗ 0.250∗ 0.293∗∗ (0.464) (0.681) (0.697) (0.842) (0.189) (0.145) (0.122) (0.122) Obs 1806 1794 1804 1809 1806 1794 1804 1809 Robust standard errors (clustered at the community level) in parentheses, p-values—*** p<0.01, ** p<0.05, * p<0.1. Additional independent variables in all regressions: mother’s age at birth, mother’s age at birth squared, age (months), age squared, wealth index, number of siblings, caregiver’s education, and dummies for first-born, older siblings, sex, survey round, year and season of birth, and community.

22

Table A8: Birth spacing and child height, females only

Height HAZ Age 1 Age 5 Age 8 Age 12 Age 1 Age 5 Age 8 Age 12 Space 2-3 0.510∗ 0.270 0.529 0.139 0.205∗ 0.055 0.094 0.022 (0.274) (0.292) (0.383) (0.501) (0.105) (0.059) (0.066) (0.073) Space 3-4 0.803∗∗ 0.531 1.026∗∗ 0.742 0.312∗∗ 0.114 0.180∗∗ 0.109 (0.311) (0.395) (0.455) (0.554) (0.119) (0.081) (0.079) (0.081) Space 4-7 1.025∗∗∗ 0.913∗∗∗ 0.776∗∗ 0.775∗ 0.393∗∗∗ 0.191∗∗∗ 0.135∗∗ 0.113∗ (0.258) (0.308) (0.362) (0.449) (0.098) (0.063) (0.062) (0.066) Space 7+ 1.225∗∗∗ 1.377∗∗∗ 1.940∗∗∗ 1.859∗∗∗ 0.460∗∗∗ 0.289∗∗∗ 0.337∗∗∗ 0.271∗∗∗ (0.364) (0.427) (0.531) (0.646) (0.139) (0.088) (0.092) (0.094) Obs 3438 3483 3485 3475 3438 3483 3485 3475 Robust standard errors (clustered at the community level) in parentheses, p-values—*** p<0.01, ** p<0.05, * p<0.1. Additional independent variables in all regressions: mother’s age at birth, mother’s age at birth squared, age (months), age squared, wealth index, number of siblings, caregiver’s education, and dummies for first-born, older siblings, survey round, year and season of birth, and community.

Table A9: Birth spacing and child height, males only

Height HAZ Age 1 Age 5 Age 8 Age 12 Age 1 Age 5 Age 8 Age 12 Space 2-3 0.090 0.018 0.108 0.156 0.033 0.013 0.019 0.023 (0.196) (0.272) (0.345) (0.491) (0.080) (0.058) (0.061) (0.069) Space 3-4 0.945∗∗∗ 0.555∗ 0.455 0.131 0.397∗∗∗ 0.131∗ 0.082 0.021 (0.228) (0.321) (0.417) (0.488) (0.092) (0.069) (0.074) (0.069) Space 4-7 0.688∗∗∗ 0.247 0.004

• 0.176

0.285∗∗∗ 0.065 0.002

• 0.023

(0.215) (0.281) (0.403) (0.492) (0.088) (0.060) (0.072) (0.069) Space 7+ 0.747∗∗∗ 0.632∗ 0.596 0.911 0.325∗∗∗ 0.146∗ 0.107 0.129 (0.232) (0.357) (0.478) (0.585) (0.096) (0.076) (0.085) (0.082) Obs 3759 3822 3814 3810 3759 3822 3814 3810 Robust standard errors (clustered at the community level) in parentheses, p-values—*** p<0.01, ** p<0.05, * p<0.1. Additional independent variables in all regressions: mother’s age at birth, mother’s age at birth squared, age (months), age squared, wealth index, number of siblings, care- giver’s education, and dummies for first-born, older siblings, survey round, year and season of birth, and community.

23