Knowledge Brief Health, Nutrition and Population Global Practice BASIC PROFILE OF CHILD MARRIAGE IN BANGLADESH Chata Malé and Quentin Wodon March 2016 Child Marriage Series with Education Global Practice e KEY MESSAGES:  Measures of child marriage remain high in Bangladesh. The share of women ages 18-22 who married as children is 59.4 percent, but it has decreased rapidly over time. The share of girls marrying very early, before the age of 15, has declined as well.  Child marriage is associated with lower wealth, lower education levels, and higher labor force participation. These are however only correlations, not necessarily causal effects. In order to design programs and policies to reduce child Box 1: Brief and Series Primer marriage, information is needed on the trend in the How is child marriage defined? Child marriage is defined as a practice over time, where it is most prevalent in a country, marriage or union taking place before the age of 18. and what the characteristics of girls marrying early are. Why a series on child marriage? Child marriage has Measuring child marriage is needed to inform policy. significant negative impacts – not only for girls, but also for a range of development outcomes. Demonstrating these impacts Child marriage is recognized as a major development will assist governments and others to make the case for intervening to reduce the practice. issue that affects girls in many developing countries. The practice has been linked to a number of health risks, What are the topics discussed in the series? The series higher fertility, and lower education attainment, among looks at the impacts of child marriage on health, population, others. The negative impact of child marriage on a wide education, employment, agency, and violence, among other range of development outcomes explains why in many outcomes. The welfare, budget, and non-monetary costs of child countries child marriage is now prohibited by law, and marriage are estimated. Legal/institutional aspects and options why the elimination of child marriage is part of the new to reduce the practice are also discussed. Sustainable Development Goals. Yet more is needed to eliminate the practice than adopting laws. In order to What is the question asked in this brief? The question is: How widespread is the practice, not only in terms of the share of inform program and policies to reduce the practice, this girls marrying early, but also in terms of how early they marry? brief provides a basic profile of child marriage in Bangladesh. The brief is part of a series of standardized How is the question answered? Measures and a profile of briefs on this topic for several countries. child marriage inspired by the literature on poverty are provided. Page 1 HNPGP Knowledge Brief  Six in ten women in Bangladesh still marry early. but it will also have other negative consequences for her as well as for her children. The analysis is based on data from the 2011 Demographic and Health Survey (DHS) for Bangladesh. Most studies on child marriage report the incidence of This is the latest DHS available. Table 1 provides basic child marriage - the share of girls who marry early (before statistics on the age at first marriage for women. Two 18), sometimes also with the share of girls who marry samples are considered: women ages 18 to 22, which is very early, before age 15. Such statistics are useful, but the youngest age group that can be used to measure they do not capture the “depth” and “severity” of the child marriage in the country1, and women ages 18-49 practice very well. Better measures of child marriage can (the women’s questionnaire in the DHS collects data for be adopted from the poverty literature (Ngyuen and women up to age 49). A large share of women marry Wodon (2012). Three measures are used here: the below the age of 18 in both groups, but especially among incidence of child marriage or headcount index, the child the larger age group. Many women marry before the age marriage gap, and the squared child marriage gap. of 15, but again especially among women ages 18-49. Definitions of these measures is provided in the annex. This suggests that child marriage has decreased over The measures are estimated for child marriage as well as time. very early marriage defined as marrying before age 15. Table 1: Age at First Marriage for Women (%) The child marriage gap represents the “depth” of child 18-22 years 18-49 years marriage. It takes into account not only the share of girls Not married 24.5 7.4 who marry early, but also the mean number of years of 18 or Above 16.1 20.5 early marriage. When using the child marriage gap for the Below 12 0.5 1.7 evaluation of programs or policies, instead of simply 12 1.9 3.5 looking at the share of the girls who marry early, more 13 9.6 18.5 14 10.9 13.4 weight is placed on the girls who marry at a very young 15 13.1 14.0 age. While the child marriage gap takes into account the 16 12.2 12.1 average number of years of early marriage for girls who 17 11.2 9.0 marry early, the squared gap takes into account the Total 100.0 100.0 square of that number, thereby putting even more Mean age at first marriage 15.8 15.7 emphasis on girls who marry very early and taking into Source: Authors’ estimation. account inequality in the age of marriage among girls marrying early. The consequences of child marriage are not the same whether girls marry at 12 or 17. Measures inspired from The incidence of child marriage in Bangladesh in 2011 the poverty literature help in capturing better how early was high, but much lower than that observed 25 years girls marry (see the annex). The headcount (H) measures ago. There has been a reduction in how early girls marry. the share of girls who marry early. The child marriage gap (CMG) measures the “depth” of the practice, taking into Child marriage is still persistent. account how early girls marry. The squared gap (SG) puts even more weight on the girls who marry very early. Table 2 provides trends over time in the measures of child marriage inspired by the poverty literature. Consider first Beyond the share of girls who marry early, other the age group 18-22. In that age group almost six in ten measures of child marriage are also important. girls marry before the age of 18 (59.4 percent for the 18- 22 age group). The child marriage gap (CMG) is at 10.1 The negative impact of child marriage for a girl’s health, percent and the squared gap (SG) at 2.1 percent for that education, and well-being is often larger when the girl group. By estimating the same measures on older groups, marries very early. For example, child marriage is known the table provides the trend in child marriage over time. to have a negative impact on school enrollment and When considering the 18 years threshold, there has been attainment. The earlier a girl marries, the more likely it is a large decline in the headcount, and an even larger that she will drop out early and thereby have a low level of decline (in proportional terms) in other measures, education attainment. This will not only limit her suggesting that girls tend to marry less early when they employment and earnings potential for the rest of her life, marry as children. Nevertheless, the incidence of child remains fairly high. 1 Child marriage measures must be estimated on the population The fact that girls who marry early may marry less early is older than 18, because some younger girls not yet married in the survey could still get married by age 18. It is best to measure confirmed by the measures based on the 15 years age child marriage as early as possible after the age of 18 to provide threshold which suggest a larger decline in the headcount data on conditions as current as possible, which is why the age for those measures. The share of girls marrying as bracket 18-22 is used here. children has decreased by 23 percentage points over the Page 2 HNPGP Knowledge Brief  last 25 years (the approximate time gap between the first with men who have similar socio-economic profiles, so the and last age group). The decline for extreme child quintile after marriage may not be that different from the marriage (15 years threshold) is larger at 28 points2. quintile before. Also, for younger women, assets and wealth may be lower than for older women. In Table 2: Trend in Child and Very Early Marriage (%) Bangladesh, the measures of child marriage differ quite a 18 years 15 years bit by quintile, but it is only in the top quintile of wealth that H CMG SG H CMG SG child marriage is much less prevalent. All 18-49 years 72.1 14.1 3.3 37.0 4.5 0.7 Age group Table 4: Child Marriage by Quintile, Age 18-22 (%) 18-22 years 59.4 10.1 2.1 22.9 2.5 0.3 18 years 15 years 23-30 years 71.9 13.7 3.1 35.4 4.2 0.6 H CMG SG H CMG SG 31-40 years 76.4 15.6 3.8 42.0 5.3 0.8 All 18-22 years 59.4 10.1 2.1 22.9 2.5 0.3 41-49 years 82.3 17.8 4.5 50.5 6.6 1.1 Wealth quintiles Source: Authors’ estimation. Poorest 74.4 13.9 3.0 36.3 3.9 0.5 Poorer 67.1 11.8 2.5 27.8 3.1 0.4 Girls are more likely to marry early if they live in rural Middle 60.7 10.4 2.2 23.3 2.5 0.3 areas and are from poorer socio-economic groups. Richer 57.1 9.3 1.9 20.3 2.3 0.3 Richest 43.6 6.5 1.3 12.0 1.4 0.2 Child marriage is more prevalent in rural than in urban Source: Authors’ estimation. areas. There are also differences between regions, with the lowest measures observed in Sylhet and the highest Child marriage is associated with lower education measures observed (according to the headcount index for attainment and a lower likelihood of literacy. the 18 years threshold) in Rangpur, followed by Rajshahi, Khulna, and Dhaka. The ranking of the regions in terms of Table 5 provides data on child marriage by level of the measures obtained with the 15 and 18 years education of the women, as well as literacy. Child thresholds tends to be similar, but differences tend to be marriage affects education attainment negatively, magnified further when considering the lower age because girls often drop out of school when they marry. threshold for extreme child marriage. Apart from Sylhet, The causality goes the other way as well, as the ability to child marriage is less prevalent in Chittagong. pursue one’s education may help delay the age at marriage. This relationship between education and child Rural girls are more likely to marry early than urban girls. marriage is apparent in the data, in that the measures of Girls from the bottom four quintiles of wealth are much child marriage tend to be higher among women with lower more likely to marry than girls from the top quintile. levels of education. The same relationship is observed when considering literacy where three categories are considered: the woman cannot read at all, can read part Table 3: Child Marriage by Location, Age 18-22 (%) of a sentence, or can read a full sentence. 18 years 15 years H CMG SG H CMG SG The relationship between child marriage and schooling is All 18-22 years 59.4 10.1 2.1 22.9 2.5 0.3 Region important for policy as the causality goes both ways. Child Barisal 60.0 9.5 1.9 20.0 2.1 0.3 marriage may lead to dropouts and lower education Chittagong 50.9 7.9 1.5 16.1 1.7 0.2 attainment. But the reverse is true as well: keeping girls in Dhaka 60.2 10.2 2.2 22.7 2.6 0.4 school is often one of the best ways to delay marriage. Khulna 64.7 11.0 2.3 22.9 2.7 0.4 Rajshahi 68.2 12.6 2.8 32.8 3.5 0.5 Rangpur 72.4 13.4 3.0 35.4 3.9 0.5 Marrying between the ages of 15 and 17 tends to affect Sylhet 36.1 5.2 1.0 9.6 0.9 0.1 primarily secondary education enrollment or completion, Residence and may not necessarily affect the completion of primary Urban 51.1 8.6 1.8 18.8 2.2 0.3 education. But marrying even earlier can also prevent Rural 62.6 10.7 2.2 24.5 2.6 0.3 girls from completing their primary education (primary Source: Authors’ estimation. school takes in principle six years to complete, but some students start primary school late and may also repeat Household welfare is measured through a wealth index grades, so the actual age of completion may be delayed). with households categorized in five quintiles from poorest to richest. For most women the level of wealth observed is that of the household in which they married, not their household or origin, but it is likely that many women marry 2 These measures have standard errors (not shown to save space). Some differences may not be statistically significant. Page 3 HNPGP Knowledge Brief  Table 5: Child Marriage by Education Level and Conclusion Literacy Status, Age 18-22 (%) 18 years 15 years This brief has provided a basic profile of child marriage in H CMG SG H CMG SG Bangladesh. Measures of child marriage are still high. All 18-22 years 59.4 10.1 2.1 22.9 2.5 0.3 The share of women ages 18-22 who married as children Education has declined over time, but remains substantial at 59.4 No education 72.8 14.1 3.2 34.3 4.2 0.6 percent. The share of girls marrying very early, before the Primary, some 72.8 13.9 3.1 37.1 4.1 0.5 age of 15, has declined too, but also remains high at 22.9 Primary, compl. 68.8 12.7 2.8 32.8 3.7 0.5 percent. Child marriage is associated with lower wealth, Secondary, some 66.8 11.0 2.3 23.7 2.5 0.3 lower education levels, and a higher likelihood of work Secondary, compl. 46.5 5.6 0.9 5.0 0.6 0.1 without cash earnings. These are however only Higher 15.8 1.5 0.2 0.9 0.1 - Literacy correlations, not necessarily causal effects. Other briefs in Cannot read 72.8 13.9 3.2 35.7 4.2 0.6 this series look at potential causal effects. Limited ability 69.4 13.5 3.0 35.9 4.0 0.5 Full sentence 55.7 9.0 1.8 19.0 2.1 0.3 References Source: Authors’ estimation. Values rounding to 0.00 not shown. Foster, J., J. Greer, and E. Thorbecke, 1984, A Class of Relationships between child marriage and labor force Decomposable Poverty Measures, Econometrica 52: 761–776. participation can be complex and depend on context. Nguyen, M. C., and Q. Wodon, 2012, Measuring Child Marriage, Economics Bulletin 32(1): 398-411. Table 6 provides data on labor force participation. In some countries child marriage may reduce labor force Annex: Methodological Note participation through higher fertility. In others, if child marriage is associated with poverty, women may leave The headcount index, child marriage gap, and squared child little choice but to work. Other effects could be at work, so marriage gap are the first three measures of the so-called FGT that the relationship between child marriage and labor class (Foster et al., 2014). Denote by q the number of girls who force participation is complex. In Bangladesh, child marry early and by n the number of girls in the overall marriage measures are similar whether women are population. Denote by yi the age of marriage of girl i and by z working or not. However, the type of work associated the age threshold defining child marriage (18 years of age, but a most with child marriage is work without cash earnings, lower age threshold can also be used to measure extreme child whether unpaid or with in kind payment, which may be marriage). The general formula for the FGT class of measures depends on a parameter α which takes a value of zero for the work with low productivity. These basic statistics however headcount, one for the child marriage gap, and two for the do not imply causality. squared child marriage gap in the following expression:  1 q  z  yi  Table 6: Child Marriage by Labor Force Participation Status, Age 18-22 (%) P   z  n i1   18 years 15 years H CMG SG H CMG SG All 18-22 years 59.4 10.1 2.1 22.9 2.5 0.3 This brief was produced as part of the Economic Impacts of Child Working Marriage study, a joint project of the International Center for Research on Women (ICRW) and the World Bank, which is supported by the Bill & No 59.2 10.0 2.1 22.3 2.4 0.3 Melinda Gates Foundation and the Children’s Investment Fund Yes 60.9 11.0 2.5 27.5 3.2 0.4 Foundation (CIFF). More details on the research can be found at the Type of work project’s website: www.costsofchildmarriage.org. Partial funding for the Not paid 82.8 15.0 3.0 21.2 2.8 0.4 work related to child marriage and education, labor force participation, Cash only 60.3 10.8 2.4 26.7 3.1 0.4 earnings, and program responses has been provided by the Global Cash and in-kind 67.2 14.0 3.1 40.7 4.1 0.4 Partnership for Education. Comments from Jeff Edmeades and In-kind only 79.0 19.0 4.7 63.6 7.0 0.8 Margareta Norris Harrit are gratefully acknowledged. The opinions Source: Authors’ estimation. expressed in this brief are those of the authors only and need not reflect the views of the World Bank, its Executive Directors, of the countries they represent. 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