Knowledge Brief Health, Nutrition and Population Global Practice CHILD MARRIAGE AND FERTILITY IN ETHIOPIA Adenike Onagoruwa and Quentin Wodon June 2017 Child Marriage Series with Education Global Practice KEY MESSAGES: • In Ethiopia, women who marry as children have on average 21-34 percent more births over their life time as compared to women marrying after the age of 18. • Controlling for socio-economic and other characteristics, the average number of births per woman would be reduced by 0.97 births or about fifteen percent if child marriage could be eliminated. This in turn would have a substantial effect on demographic growth. Box 1: Brief and Series Primer Girls marrying early tend to have children earlier than girls who marry later. They also have more children over their How is child marriage defined? Child marriage is defined as a lifetime. Given the importance of population growth, marriage or union taking place before the age of 18. assessing the impact of child marriage on fertility matters. Why a series on child marriage? Child marriage has significant negative impacts – not only for girls, but also for a range of Child marriage is associated with higher fertility. development outcomes. Demonstrating these impacts will assist governments and others to make the case for intervening to The relationship between child marriage and fertility could reduce the practice. be due in part to the socio-economic and cultural context in which girls who marry early tend to live. But child marriage What are the topics discussed in the series? The series looks may also have a direct impact on fertility after controlling at the impacts of child marriage on health, population, education, for socio-economic and cultural context. Marrying early is employment, agency, and violence, among other outcomes. The often associated with a lack of agency for girls, including in welfare, budget, and non-monetary costs of child marriage are terms of access to family planning that could help delay or estimated. Legal/institutional aspects and options to reduce the practice are also discussed. reduce births if women so desire. For societies, higher total fertility rates lead to higher population growth, lower growth What is the question asked in this brief? The question is: To in GDP per capita, slower poverty reduction, and difficulties what extent would the number of live births that women have over for governments to provide basic services to a growing their lifetime decrease if child marriage were eliminated? population. This brief estimates the impact of child marriage on the number of children that women have over How is the question answered? Econometric analysis of their lifetime in Ethiopia, as part of a series of standardized Demographic and Health Survey data is used to estimate the briefs on this topic for multiple countries. impact of child marriage on the number of births women have. Page 1 HNPGP Knowledge Brief • Statistically, women marrying as children have 1.58 Controlling for socio-economic and other characteristics, more live births than women marrying after 18. women who marry as children have on average 21-34 percent more than women marrying after the age of 18. The analysis is based on data from the third Demographic and Health Survey for Ethiopia implemented in 2011. The With the baseline specification, table 2 suggests that focus is on the number of births that women have over their marrying at age 17 increases the number of births by 23 lifetime. For sample size reasons, we focus on women percent in comparison to marrying at age 18 or later. between 35 and 49 years of age. Some women continue Marrying earlier often has a larger impact. Overall, the to have children after age 35, so that the total number of births considered here may be slightly underestimated, but impacts vary from 33.8 percent for girls marrying at age 13 this should not affect too much the estimate of the to 20.6 percent for girls marrying at age 15. The marginal effects of early marriage do not change much when difference in the total number of births for women marrying additional controls are added to the regressions. There is as children (before age 18) or not. thus clear evidence that after controlling for a wide range of other variables, child marriage may contribute to higher Women ages 35-49 who married after 18 have on average fertility, but prudence remains needed when interpreting 5.25 live births by that age. For women who married before these results given the risk of omitted variable bias (see 18, the average is higher at 6.83 births. box 2). As shown in table 1, women ages 35-49 who married after Table 2: Impact of Child Marriage on Number of 18 have on average 5.25 births by that age. For women Children Ever Born by Age of Marriage, 2011 DHS who married before 18, the average is 6.83 births. The Age at first Baseline Extended difference (1.58 births) is statistically significant. Except for marriage model model girls marrying at 12 or earlier (these are extreme cases that Married <= 12 1.288*** 1.346*** may have specific circumstances), the number mostly Married at 13 1.338*** 1.325*** increases when girls marry earlier, as expected. Married at 14 1.301*** 1.317*** Married at 15 1.206*** 1.202*** Table 1: Mean Number of Children Ever Born by Age Married at 16 1.225*** 1.235*** of First Marriage, 2011 DHS Married at 17 1.230*** 1.224*** Age 35-49 Source: Authors. Sample as a whole 6.26 (0.088) Levels of statistical significance: *** 1%, ** 5%, * 10%. Married after 18 years of age 5.25 (0.150) Married before 18 years of age 6.83 (0.090) Box 2: Risk of Omitted Variable Bias Married at 17 6.81 (0.192) Married at 16 6.67 (0.184) Child marriage appears to be positively correlated with fertility Married at 15 6.69 (0.141) after controlling for other factors that may also contribute to Married at 14 7.12 (0.194) fertility. This could indicate a causal effect. However, other Married at 13 7.12 (0.202) variables correlated with both child marriage and fertility not Married at 12 /below 6.74 (0.186) included in the analysis could be at the source of the correlation Source: Authors. Standard error in parentheses. observed between child marriage and fertility. Because of the risk of omitted variable bias, the results cannot be considered as fully Controlling for other factors, child marriage still conclusive regarding a causal impact of child marriage on fertility, increases the number of live births substantially. even though in this specific case causality is very likely. In order to measure the impact of child marriage at the Because early marriage also has an impact on other margin on the number of births, regression analysis is variables used as controls in the regression, the overall needed. Details of the methodology is provided in the effect of child marriage on the number of births including annex. Table 2 shows the results for the variables of through indirect effects could be larger than the direct interest with two models: a baseline specification and the estimates observed from the regression coefficients. For specification with the largest number of added variables as example, for some girls marrying early, child marriage has controls. The interpretation of the coefficients is in terms of a negative effect on education attainment, and it may also incident rate ratios. A coefficient of 1.15 for girls marrying contribute to lower household wealth. Still, in terms of at the age of 17 suggests that if a girl marries at that age, magnitude, those indirect effects are likely to be small in the number of children she will have over her lifetime will comparison to direct effects. Only a relatively small share be 15 percent higher than the number of children she would of girls marrying early would have been able to complete have had if she had married at age 18 or later. their secondary education if they had married later and the coefficient of secondary education on the number of births in the regression is close to one, suggesting limited impact of a secondary education on fertility in comparison to a Page 2 HNPGP Knowledge Brief lower level of education. This implies that one may rely on woman in the country (first column in table 3), this suggests the estimated direct effects of child marriage on the number that the average number of births (essentially the total of births that women have over their lifetime in order to fertility rate) would be reduced by about fifteen percent if simulate the impact of the elimination of child marriage on child marriage could be eliminated. This in turn would have total fertility. Not factoring in indirect effects in the a substantial effect on demographic growth in the country. simulations does not entail large mismeasurement. Eliminating child marriage could reduce the country’s total A number of other factors not shown in table 2 lead to fertility rate by 0.97 births or about fifteen percent. This differences in the number of births for women. As already would help reduce population growth substantially. mentioned, better educated women, especially those with a higher degree, tend to have fewer children. The Table 3: Impact of the Elimination Child Marriage differences by wealth quintiles are not statistically under the Baseline Model for the Estimations significant. Younger women within the sample have fewer Women 35+ Observed Predicted Simulated births, but may still get more later on in their life. (n=4,373) Mean Mean Mean Differences according to religion are not statistically All 6.26 6.33 5.36 significant, but differences according to geographic areas Age at first marriage are. Most of the additional control are not associated with 30-34 1.86 4.29 4.29 statistically significant effects on fertility, especially in the 25-29 3.18 4.51 4.51 last model. 20-24 5.20 5.19 5.19 18 and above 5.25 5.24 5.24 Eliminating child marriage would help reduce the 18 6.02 5.59 5.59 number live births for women marrying early by 1.41. 17 6.81 6.81 5.54 16 6.67 6.67 5.44 15 6.69 6.69 5.55 The last step in the analysis consists in assessing the 14 7.12 7.13 5.48 impact of eliminating child marriage on the total fertility rate. 13 7.12 7.14 5.33 Results in table 3 are obtained by predicting the number of 12 or below 6.74 6.73 5.23 births that women who married as children would have had Never married 0.69 - - if they had married later. The first column in table 3 17 and below 6.83 6.83 5.42 provides the estimates of the number of births in the Source: Authors’ estimations. sample. The second column provides the predicted values under the baseline model (results are similar with other Conclusion models). The third column provide the simulated number of births without child marriage. Note that for all women who Child marriage is linked to higher and less controlled fertility marry after the age of 18, there are no differences between for girls who marry early. In Ethiopia, under the baseline the predicted and simulated number of births marriage specification, girls marrying before the age of 18 have one since these women did not marry early. fourth more births over their lifetime than women who marry after the age of 18. This translates on average for all girls For women who did marry early, the simulated number of who marry early in an increase in the number of children births are substantially lower than the predicted values ever born of 1.41 children versus the number of children taking into account the fact that they married early. With the born for girls who marry at 18 or later. For the country as a baseline model, women who married early have on whole, the average number of births (essentially the total average 1.41 more children than if they had married later. fertility rate) would be reduced by about 0.97 births or about The difference between predicted and simulated numbers 15 percent if child marriage were to be eliminated. of children for women who married early or not with the expanded model is similar, at 1.53 fewer children. References The top row in table 3 provides the difference that child Godha, D., Hotchkiss, D. R., & Gage, A. J. (2013). Association marriage makes for the average number of births nationally between child marriage and reproductive health outcomes and (this is similar to the total fertility rate), thereby factoring the service utilization: a multi-country study from South Asia. Journal of Adolescent Health, 52(5), 552-558. share of women who marry early at different ages. With the baseline model, the average number of births that women Kamal, S. M. (2012). Decline in child marriage and changes in its have over their lifetime is reduced for the country as a effect on reproductive outcomes in Bangladesh. Journal of health, whole by 0.97 births without child marriage. The effect is population, and nutrition, 30(3), 317. again very similar when considering the expanded model (not shown), with a reduction of 1.05 in the average number Kamal, S. M., & Hassan, C. H. (2015). Child marriage and its of births that women would have. Given the observed association with adverse reproductive outcomes for women in average number of children of just above six births per Page 3 HNPGP Knowledge Brief • Bangladesh. Asia-Pacific Journal of Public Health, 27(2), NP1492-NP1506. A few additional explanations on definitions may be useful. In this brief, the focus is on the impact of child marriage on the number Central Statistical Agency [Ethiopia] and ICF International (2012). of children ever born to women towards the end of their fertile life. Ethiopia Demographic and Health Survey 2011. Addis Ababa, This is a useful statistic that summarizes the fertility behavior of Ethiopia and Calverton, Maryland, USA: Central Statistical women who are nearing the end of their reproductive period. If Agency and ICF International. fertility is stable over time, the mean number of children ever born is similar to the total fertility rate (TFR) which refers to the average Nasrullah, M., Muazzam, S., Bhutta, Z. A., & Raj, A. (2014). Girl number of live births a woman would have if she were subject to child marriage and its effect on fertility in Pakistan: findings from current age-specific fertility rates throughout her reproductive Pakistan Demographic and Health Survey, 2006–2007. Maternal years. In other words, under stable conditions, the number of and child health journal, 18(3), 534-543. children ever born is a good proxy for the TFR. If fertility levels have been falling, the TFR will however be (slightly) lower than Raj, A., Saggurti, N., Balaiah, D., & Silverman, J. G. (2009). the mean number of children ever born. For sample size reasons, Prevalence of child marriage and its effect on fertility and fertility- we consider women ages 35 to 49 to estimate the number of control outcomes of young women in India: a cross-sectional, children ever born, rather than women ages 40-49. This does not observational study. The Lancet, 373(9678), 1883-1889. affect our results substantially given the focus on the differential impact of child marriage Santhya, K. G. (2011). Early marriage and sexual and reproductive health vulnerabilities of young women: a synthesis As for the regression analysis, given that the variable of interest of recent evidence from developing countries. Current opinion in is a count variable (taking values such as 1, 2, 3…), Poisson obstetrics and gynecology, 23(5), 334-339. regressions are estimated. Different models are estimated to test for the robustness of findings to the specification used. Apart from Santhya, K. G., Ram, U., Acharya, R., Jejeebhoy, S. J., Ram, F., the age at first marriage or union, the baseline model controls for: & Singh, A. (2010). Associations between early marriage and (1) location (urban versus rural); (2) education level of the young women's marital and reproductive health outcomes: girl/woman; (3) wealth quintile (from the poorest to the richest); evidence from India. International perspectives on sexual and (4) religion; (5) geographic area (provinces); and (6) age group reproductive health, 132-139. (35-39, 40-44, and 45-49). In other models variables capturing agency for girls/women in the household and additional Annex: Methodological Note community-level controls are added. The community variables are the leave-out-means of contraceptive use and whether Previous research has been conducted on child marriage and women experienced a child death before the age of five (leave- fertility (e.g., Santhya, 2011; Kamal, 2012; Godha et al., 2013; out-means are estimated at the level of the primary sampling units Nasrullah et al., 2013; Kamal et al. 2015). This brief has three in the survey). Overall, the estimated impacts of early marriage features that differentiate it from previous work. First, the impact on the number of births for women over their lifetime are fairly of each additional year of early marriage on the number of births robust to the choice of specification. a woman has over her lifetime is estimated. Previous studies typically looked only at the impact of marrying early or not. Here, we look at the impact of each single additional year of early marriage. Second, we estimate the impact of child marriage on a This brief was produced by a World Bank team as part of the Economic woman’s number of child births over her lifetime. Previous Impacts of Child Marriage study. The synthesis report under the study was studies focused mostly on young women (ages 18 to 24). The produced jointly with the International Center for Research on Women. reason for considering older women is to account for potential The study benefitted from support from the Bill & Melinda Gates Foundation, the Children’s Investment Fund Foundation, and the Global substitution effects whereby if a woman does not have children Partnership for Education. Comments from colleagues and peer early in life, she may want more children later on. reviewers are gratefully acknowledged. The opinions expressed in this brief are those of the authors only and need not reflect the views of the In order to look at the impact of child marriage on total fertility, we World Bank, its Executive Directors, of the countries they represent. need to consider women towards the end of their fertile age. Third, we use the estimations from the regression analysis to conduct simulations about the magnitude of the reduction in the average number of children that women are expected to have that could be achieved through the elimination of child marriage. The Health, Nutrition and Population Knowledge Briefs of the World Bank are a quick reference on the essentials of specific HNP-related topics summarizing new findings and information. These may highlight an issue and key interventions proven to be effective in improving health, or disseminate new findings and lessons learned from the regions. For more information on this topic, go to: www.worldbank.org/health. Page 4