Introduction
Investments in girls’ education globally have been motivated in part by an expectation that more-educated women will have smaller and healthier families (Behrman
2015; Grant
2015; Psaki et al.
2019; Weitzman
2017). However, despite dramatic gains in women’s attainment globally, the pace of fertility decline is slower than expected in some countries (Bongaarts and Casterline
2013; Nguyen and Wodon
2012), raising the question, under which conditions is women’s education most likely to translate into improved demographic outcomes?
Some researchers have speculated that academic skills are the primary pathway through which grade attainment influences fertility (LeVine et al.
2012). Literacy may improve access to information about the benefits of birth spacing, the advantages of contraception, and the opportunity cost of high fertility, particularly for women (Jejeebhoy
1995). Literacy and numeracy may enhance women’s abilities to navigate health bureaucracies (LeVine et al.
2012), access family planning services, and successfully use contraception.
Despite theory, and some empirical evidence (Smith-Greenaway
2013), around the important role that literacy and numeracy play in improving demographic outcomes, little research has explored what happens to these skills after girls leave school (for exceptions, see Gorman and Pollitt
1997; Soler-Hampejsek et al.
2018). In low- and middle-income countries (LMICs), given increasing school enrollment and widespread early marriage and childbearing, many adolescent girls transition rapidly from school to motherhood (Lloyd
2005). Yet, research on the effects of adolescent childbearing on education in LMICs has been limited and has focused narrowly on grade attainment.
We hypothesize that adolescent childbearing in the years after school leaving contributes to the deterioration of already weak academic skills. If true, the assumption that skills remain constant after leaving school may lead to an overestimate of the potential effects of education on economic productivity, fertility, and other demographic outcomes. For example, if literacy affects fertility through women’s access to the labor market—a commonly hypothesized pathway (Psaki et al.
2019)—rapid loss of literacy skills after leaving school may mean that adolescent girls no longer have the skills needed to join the labor market after they begin childbearing. We test our hypothesis using data from three longitudinal studies in low-income settings that followed adolescent girls during the transition from school to motherhood: the Malawi Schooling and Adolescent Study (MSAS) (2007–2013); the Bangladeshi Association for Life Skills, Income and Knowledge for Adolescents (BALIKA) study (2013–2015); and the Adolescent Girls Empowerment Program (AGEP) in Zambia (2013–2017). We compare results across these settings in part because of data availability. However, despite important social and economic differences between these settings, all three countries are facing the challenges of low-quality schools, amid rapid expansions in access to school, and high levels of adolescent childbearing. To our knowledge, this is the first study to examine the effect of adolescent childbearing on academic skills in LMICs.
Background
Much research has documented the association between grade attainment and reproductive behavior among women in low-income countries (Bledsoe et al.
1999; Cochrane
1979): those with more years of schooling are likely to marry and give birth later and to have fewer and healthier children (Basu
2002; Cleland
2010; Gakidou et al.
2010; Glewwe
1999). This relationship may operate in either direction: for example, more schooling may contribute to delayed childbearing (Breierova and Duflo
2003; Osili and Long
2008; Smith-Greenaway
2013), and adolescent childbearing may disrupt schooling (Geronimus and Korenman
1992; Kane et al.
2013; Stange
2011). Given the many shared determinants of schooling and reproductive behavior, researchers continue to debate the extent to which these relationships are causal (Psaki et al.
2019).
Studies in low-income countries have largely focused on this relationship in one direction: the effect of education on reproductive behavior. The research that has been conducted on the effects of childbearing on education—both in low-income and high-income settings—has focused narrowly on grade attainment and has overlooked the potential effects of childbearing on academic skills.
Mechanisms Linking Childbearing and Skills
Research and policies in low-income settings are often designed based on the assumption that academic skills remain constant after school leaving (LeVine et al.
2012), despite evidence that skills may improve (Gorman and Pollitt
1997; Hartley and Swanson
1986; Wagner et al.
1989) or deteriorate (Abadzi
2003; Durgunoğlu et al.
2003), depending on whether they are reinforced through regular use (Wagner
1994).
Several studies in LMICs have examined the trajectories of academic skill development and loss. Using a longitudinal sample from Ladino communities in Guatemala, in which respondents were first interviewed as children and followed up 20 years later, Gorman and Pollitt (
1997) investigated skill loss after leaving school. They found a strong positive relationship between early grade attainment and literacy skills. In contrast to more recent data from low-income settings (World Bank
2018), all children in the Guatemala sample were literate once they had completed three years of schooling. Study respondents’ skills continued to improve after leaving school, especially among those with the lowest levels of grade attainment. More recently, research using data from the MSAS found evidence of gender differences in deterioration in literacy skills: adolescent girls had significantly lower levels of literacy in the several years after school leaving than their male peers, even after skill level at school leaving and grade attainment were controlled for. The authors suggested that students who leave school earlier, particularly females, may not have the same opportunities to apply and reinforce skills as those who leave school later (Soler-Hampejsek et al.
2018). The several years after leaving school may be a critical period for securing or strengthening academic skills for adolescent girls, who often experience the disruptions of marriage and childbearing soon after leaving school, and whose social roles may permanently change once those transitions begin (Lloyd
2005; Santhya and Jejeebhoy
2003).
Some evidence supports the idea of a threshold level of skill acquisition that protects against subsequent deterioration. Chall (
1996) proposed a widely used model of language acquisition as a multistage process, from basic letter and word recognition to comprehension of single and multiple viewpoints in complex texts. This model describes a critical transition—from “learning to read” to “reading to learn”—that lays the foundation for continuous learning after leaving school or other formal learning environments. In the U.S. literature, this transition is sometimes termed the
fourth grade slump, referring to the grade at which struggling readers often face challenges in developing more advanced comprehension skills. Others have argued that assuming a similar model of language acquisition in low-income countries, four to six years of schooling form an essential foundation to build the literacy skills necessary for lifelong learning (Wagner
1994). Therefore, even among those who leave school with basic literacy and numeracy skills, if they are unable to apply those skills because of changing roles (e.g., responsibility for childcare and housework) for a period of months or years, they may lose the skills permanently unless circumstances are altered.
Findings from the limited literature on skill loss and gain provide support for the relevance of this theory to low-income settings. In explaining skill improvements after leaving school among those with low levels of attainment, Gorman and Pollitt (
1997) hypothesized that children remain in school until they gain a certain level of skills (e.g., basic literacy) in order to apply and strengthen those skills after leaving school. In the Malawi study, the authors highlighted the consistent relationship between grade attainment and skill loss, arguing that those with higher levels of grade attainment were more able to apply and strengthen their skills after leaving school (Soler-Hampejsek et al.
2018). If true, this may reflect not only stronger skills at school leaving—perhaps because learners made the critical transition to reading to learn—but also a higher likelihood of accessing employment opportunities that require the application of those skills.
Although research has shown a link between grade attainment and academic skills, albeit not as strong as expected (Psaki et al.
2019; Smith-Greenaway
2013), less evidence exists on the effects of events after school leaving on literacy and numeracy. For adolescent girls, numerous life events may occur shortly after school leaving and affect their ability to apply and strengthen the skills gained in school. We know little about the potential effects that becoming a parent—an event that often occurs earlier for women than men (Lloyd
2005) and is likely to affect the regular use of skills—has on the retention of academic skills. The question of whether adolescent childbearing affects retention of academic skills after leaving school is particularly urgent in LMICs (like those included in our study) where despite recent progress in school enrollment, attainment levels remain low (Psaki et al.
2017), school quality is poor (Kendall
2007; World Bank and UNICEF
2009), and skill levels are low overall (Chimombo et al.
2005; Pritchett
2013; Smith-Greenaway
2013; World Bank
2013,
2018).
Effects of Adolescent Childbearing on Grade Attainment
Numerous studies, especially in high-income settings, have sought to examine the effect of adolescent childbearing on grade attainment; this research offers important lessons. Estimating the effects of childbearing on education presents a challenge given the shared underlying factors, such as socioeconomic status, that likely determine both the timing of childbearing and performance in school. Adolescent childbearing is not a random event but is instead selective by important characteristics that also affect school performance, including literacy and numeracy (Geronimus and Korenman
1992; Kane et al.
2013; Stange
2011). As a result, naïve estimates of these relationships are likely to overstate the effects of childbearing on skills. The key challenge, therefore, is to construct an appropriate counterfactual: what would have happened in terms of grade attainment and learning if a pregnancy (or birth) had not occurred (Hotz et al.
2005)?
Research on the effects of adolescent childbearing on grade attainment, largely conducted in high-income countries, has sought to address this estimation challenge by exploiting natural experiments, such as miscarriages or twins with different ages at onset of childbearing (for reviews of common approaches, see Hotz et al.
2005; Kane et al.
2013). Most studies have found statistical evidence of endogeneity in this relationship, resulting in attenuated or null effects in models addressing this issue (Ashcraft and Lang
2006; Fletcher and Wolfe
2009; Geronimus and Korenman
1992; Hotz et al.
2005; Kane et al.
2013; Webbink et al.
2011). Recent analyses using data from the National Longitudinal Study of Adolescent to Adult Health in the United States found a negative effect (0.7 years of schooling lost) of teen childbearing on educational attainment, albeit smaller than estimates from models that fail to control for endogeneity (1.0 years of schooling lost) (Kane et al.
2013). Using the National Longitudinal Survey of Youth and miscarriages as a natural experiment, Hotz et al. (
2005) found no significant effect of adolescent childbearing on the probability of obtaining a high school diploma, but they found a positive effect on annual earnings.
1 Similarly, using an Australian sample of twins, Webbink et al. (
2011) found no effect of adolescent childbearing on grade attainment. Diaz and Fiel (
2016) investigated heterogeneity in the effects of teen pregnancy on education and earnings and found that negative effects on earnings are most pronounced for those who are least likely to experience a teen pregnancy. More broadly, the authors noted that the effects of teen pregnancy (and childbearing) are likely to vary by the pre-pregnancy characteristics of adolescents, and this heterogeneity should be taken into account (Diaz and Fiel
2016).
Studies in LMICs have also found evidence of the selectivity of adolescent childbearing (Azevedo et al.
2012; Grant and Hallman
2008; Timaeus and Moultrie
2015). For example, research in South Africa found that adolescent girls who are struggling in school are more likely to become pregnant several years later and then drop out than their peers who perform well (Grant and Hallman
2008; Timaeus and Moultrie
2015). However, few studies in these settings have effectively addressed this selectivity in order to estimate the causal relationship between adolescent childbearing and grade attainment or other aspects of education. One exception, a study by Azevedo and colleagues, used the Mexican
Encuesta Nacional de la Dinamica Demografica (ENADID) data set collected in 2006, which includes detailed birth, miscarriage, and abortion histories. The authors used miscarriage as a natural experiment to address endogeneity after demonstrating that women who became pregnant as adolescents in this sample are selective. Although naïve estimates found a negative effect of adolescent childbearing on educational attainment, after addressing endogeneity, the authors found a statistically significant positive effect: those who gave birth during adolescence completed 0.34 more years of schooling, on average, compared with those who became pregnant but had a miscarriage during adolescence. They also found, however, that those who gave birth during adolescence received significantly more social assistance than the comparison group (Azevedo et al.
2012).
Several studies in South Africa have also examined this question using longitudinal data, although they were unable to address unobserved determinants of childbearing and schooling (Ardington et al.
2015; Ranchhod et al.
2011). Using census data from South Africa, one study found no evidence of selectivity into teenage childbearing on observed characteristics, as well as a negative effect of childbearing on the probability of being enrolled in school at every age (Ardington et al.
2015). In contrast, using data from the Cape Area Panel Study (CAPS), Ranchhod et al. (
2011) found evidence of selectivity in adolescent childbearing; they also found that despite a negative effect on high school graduation rates at age 20, this difference had disappeared by age 22, indicating a catch-up effect.
Overall, existing literature on the effects of adolescent childbearing on grade attainment shows consistent evidence of selectivity: those who give birth as adolescents differ from those who delay childbearing in important ways that may affect literacy and numeracy. However, rigorous research on this question has been concentrated in high-income settings, where the determinants of adolescent childbearing and learning likely differ from those in low-income settings. For example, compared with LMICs, most high-income countries have more widespread access to quality sexual and reproductive health services and education systems, higher levels of educational attainment and literacy, and more widely available opportunities to apply academic skills (e.g., through low-cost reading materials).
Our study builds on existing literature in three important ways. First, we examine the effects of adolescent childbearing on academic skills—a question that, to our knowledge, has not yet been investigated in the scholarly literature. Second, we use statistical methods designed to address the endogeneity in the relationship between childbearing and academic skills. Third, we conduct these analyses using data from three LMICs in two regions—Bangladesh, Zambia, and Malawi—where the determinants of adolescent childbearing and academic skills may also vary.
Study Context
Table
1 provides nationally representative overviews of key economic, fertility, and education indicators for each country included in our study. None of the samples are nationally representative, but Table
1 provides context for our comparative analyses.
Table 1
Population, economy, fertility, and education characteristics of countries where each study occurred
| Bangladesh | Malawi | Zambia |
Population and Economy | 2016 | 2016 | 2016 |
Annual Population Growth (%) | 1.1 | 2.9 | 3.0 |
GNI per Capita, Atlas Method (current US$) | 1,330 | 320 | 1,300 |
Fertility | 2014 | 2015–2016 | 2013–2014 |
Fertility Rate, Total (births per woman) | 2.3 | 4.4 | 5.3 |
Adolescents (aged 15–19) Who Have Begun Childbearing (%) | 31 | 29 | 29 |
Education: Primary Enrollment Rate (% gross) | 2015 | 2015 | 2013 |
Female | 125 | 147 | 104 |
Male | 116 | 144 | 103 |
Education: Survival to Last Grade of Primary | 2009 | 2013 | 2012 |
Female | 71 | 55 | 57 |
Male | 62 | 54 | 54 |
Education: Literacy Rate (%), 15- to 24-Year-Olds (2015) | 2016 | 2015 | 2010 |
Female | 94 | 73 | 87 |
Male | 91 | 73 | 91 |
Given the challenges in finding longitudinal data that follow adolescent girls as they transition from school to childbearing, as well as data on changing skill levels, our choice to include these three countries partly reflects data availability. However, all three countries are also facing the challenges of low-quality schools, rapid expansions in access to school, and high levels of adolescent childbearing. Important differences among these countries exist as well. Although Malawi and Zambia share a border, they differ in terms of economic and educational development. In both Malawi and Zambia, adolescent childbearing often occurs outside marriage, whereas childbearing nearly always occurs in the context of marriage in Bangladesh. The comparison between these settings has the potential to shed additional light on the relationship of interest.
Economic Context
Bangladesh is a LMIC that has experienced sustained economic growth in recent decades and has reduced poverty levels (World Bank Group
2016). Low-skill, low-wage manufacturing jobs have driven growth, including those in the garment sector, where 90 % of employees are women. Malawi and Zambia are low-income countries that share a border in southern Africa. Malawi has experienced economic growth over the last decade with support from development partners. However, poverty continues to affect many Malawians, especially in rural areas because of the lack of opportunities outside the agricultural sector, which contributes 30 % of GDP but is subject to environmental shocks. Following a recent economic downturn in Zambia due to falling global copper prices, climate events affecting agriculture, and political uncertainty leading up to the 2016 elections, the economy of Zambia is rebounding (World Bank
2018). Growing youth populations in each country—accounting for 29 % of the population in Bangladesh and 45 % in both Malawi and Zambia in 2015—underline the urgency of ensuring that education provides the skills needed to contribute to future economic development in each country.
Education Context
Bangladesh operates on a 5-5-2-4 education system: 5 years of primary school, 5 years of secondary school, 2 years of higher secondary school, and 4 years of university. The official age of school entry is 6, meaning that students who enter on time and do not skip or repeat any years would complete primary school at age 10. Malawi operates on an 8-4-4 education system: 8 years of primary school, 4 years of secondary school, and 4 years of university. The official age of school entry is 6, with an expected age of 14 at the completion of primary school. Zambia operates on a 7-5-4 system, with an official age of school entry of 7, and expected primary school completion by age 14 (UNESCO Institute for Statistics (UIS)
2017).
In all three countries, alongside expanding access to school, the quality of schooling has remained poor overall and in some cases seems to have deteriorated as more students have enrolled. Following the adoption of free primary education in Malawi in 1994, enrollment increased dramatically, especially for girls and those living in the poorest households (World Bank
2018). However, school inputs have not kept up with increasing enrollment: between 2008 and 2015, as gross enrollment increased from 131 % to 145 %, the average number of students per class increased from 85 to 126 (World Bank
2018). Similar to Malawi, Zambia experienced a dramatic increase in enrollment following the elimination of primary school fees in 2002: secondary school enrollment increased by nearly 75 percentage points between 2000 and 2010, a faster rate than that experienced by any country in history. Despite increased investments in education by the government, the quality of schooling remains poor or has deteriorated amid rapid increases in enrollment. Of 14 countries, Malawi and Zambia had the highest proportion of students scored as “not competent” in both reading and mathematics based on the 2007 Southern and Eastern Africa Consortium for Monitoring Education Quality (SACMEQ 2007). By 2012, nearly 90 % of students in both countries were unable to read a single word by the end of grade 2 (World Bank
2018). Similarly, despite expanding access to education in Bangladesh, a National Student Assessment (NSA) conducted by the government in 2011 found low competency levels among grade 8 students: 44 % met standards in Bangla, 44 % met standards in English, and 35 % met standards in mathematics (World Bank
2013).
Discussion
Over the last few decades, government policies in LMICs, particularly those supporting free compulsory schooling, have led to dramatic expansions in school enrollment, especially for girls (Behrman
2015; Grant
2015; World Bank
2013,
2018). As a result, many more girls are now enrolling in school, and remaining there during adolescence, than several decades ago (Lloyd
2009; UNESCO
2016). Recent evidence has also demonstrated that despite progress in enrollment, young people in many settings are unable to gain even basic academic skills during primary school (Smith-Greenaway
2013; World Bank
2018), and many fail to complete primary school and continue onto secondary school (Psaki et al.
2017; UNESCO
2016). Beyond the immediate benefits of education, many policy-makers and researchers expect that expanded access to schooling will translate into lower fertility across diverse contexts and populations, with mixed empirical support for this assumption (Behrman
2015; Breierova and Duflo
2003; Grant
2015; Osili and Long
2008). Evidence on the mechanisms linking education and fertility in LMICs is also inconsistent (Psaki et al.
2019). Therefore, questions remain as to the circumstances under which expanded access to schooling for girls is most likely to translate into lower levels of fertility and other demographic benefits.
We find support for our hypothesis that adolescent childbearing has a negative effect on academic skills, with important differences by level of grade attainment and skill type. Among those with low attainment (grade 7 and below), childbearing has consistently negative effects on English literacy and numeracy. Among those with higher levels of attainment (grade 8 and above), childbearing tends to have smaller effects or no effects on skills overall. The effects of childbearing on local language literacy are smaller or nonexistent. Despite the apparent protective effect of secondary school against skill loss, we also find that basic academic skills are not universal, even among those who have attended secondary school. For example, in the Zambia study, among those enrolled in the first year of secondary school at baseline, only 54 % of the rural sample and 69 % of the urban sample could read two full sentences in English (see Fig.
1). This finding runs counter to the assumption that all those who have attended some secondary school are literate.
12
Our results suggest that attaining grade 8 or higher—reflecting secondary school in Malawi and Zambia, and upper secondary in Bangladesh—is protective against skill loss due to childbearing. This evidence of a threshold level of grade attainment that may secure academic skills—especially English literacy and numeracy—is consistent with existing theory on how literacy skills are acquired (Chall
1996; Wagner
1994). Our results may initially appear inconsistent with those of Gorman and Pollitt (
1997), who found that skills continued to improve for young people in Guatemala after leaving school, and the largest improvements were for those with the lowest levels of attainment. Importantly, however, every respondent in their sample who attained at least three years of schooling was literate, perhaps in part because Spanish (the language of assessments) is commonly spoken in Ladino communities in Guatemala, where the data used for this study were collected.
In contrast, in our samples, a much smaller proportion of respondents with low levels of attainment had basic English literacy and numeracy skills. For example, in the MSAS analytical sample, only 8 % of those with grade 4 attainment had basic English literacy skills, compared with 85 % of those with grade 8 attainment (not shown). Therefore, although the pattern of skill retention after school leaving differs between our study and the work in Guatemala (Gorman and Pollitt
1997), both are consistent with a threshold model of skill acquisition and loss, where thresholds differ between settings and by skill.
The data from AGEP/Zambia also provide the opportunity to compare these relationships between urban and rural areas, where the contexts of adult literacy and opportunities for adolescent mothers differ. The English literacy results in both rural and urban Zambia are consistent with those from the other rural samples (MSAS and BALIKA): we find an overall negative effect of childbearing on English literacy among those with low attainment and less consistent effects among those with high grade attainment. Notably, however, time since school leaving in the urban Zambia sample has a significant and positive effect on English literacy, in contrast to the significant and negative effect found in the rural samples in all three studies (see Table
S2, online appendix). The experiences of young people after leaving school, and specifically their opportunities to apply and strengthen skills, are likely different in the urban and rural samples. Smith-Greenaway (
2017) explored links between community education context (specifically, the proportion of women who are literate) and child survival using DHS data from 30 sub-Saharan African countries. She found that the proportion of literate women in a community was even more important than mother’s literacy and that the relationship between community context and child survival did not depend on mother’s literacy level. Instead, she argued that this relationship may reflect the effects of the presence of literate women in a community on social learning and influence as well as shifting the institutional context. Our findings on the differences between rural and urban samples in Zambia provide support for the idea that community-level factors—such as access to reading materials, or expectations about mobility and social and economic engagement by mothers—may influence the effects of adolescent childbearing on literacy and numeracy. More research is needed on the role of community context in modifying this relationship.
We also find interesting differences in the effects of childbearing on English versus local language literacy. The results are particularly striking in the Malawi sample, where even grade attainment and current school enrollment status were unrelated to local language literacy. In the Zambia samples, levels of local language literacy were low at baseline, likely reflecting the fact that more than twice as many respondents spoke Bemba at home (65 %) than Nyanja (31 %), and the first sentence in the local language assessment was in Nyanja.
13 In both Zambia samples, the effects of childbearing on local language literacy are more similar to the results for English literacy, likely reflecting the fact that these skills were more tenuous in Zambia because of a greater diversity in languages spoken at home. In contrast, the high levels of local language literacy in Bangladesh and Malawi, coupled with the lack of effects of childbearing on these skills, indicate that those skills were likely acquired outside of school and were more easily maintained through daily activities after leaving school.
The education community, including UNESCO and national governments, has increasingly emphasized the importance of instruction in learners’ mother tongue (that is, language spoken at home) for several reasons: growing evidence that it improves learning outcomes, inclusion of diverse learners and community members, and maintenance of linguistic and cultural diversity (Trudell
2016; UNESCO
2007). At the same time, English language skills are often seen as not only economically valuable (given that English is often required for entry into higher education and the formal economy) but also symbolically appealing (Casale and Posel
2011; Chowdhury and Kabir
2014). Research on the economic returns to “dominant” language skills (often English) has focused largely on the value to immigrants looking for jobs in the formal labor market in high-income countries. However, Casale and Posel (
2011) found evidence of substantial economic returns to English literacy for Africans in South Africa and limited evidence of independent returns to local language literacy, although the two were closely related. In populations that are less likely to enter the formal labor market, such as those included in our study, local language literacy and numeracy skills may be more valuable than English language literacy for commerce, communication, and daily life (Trudell
2016).
Previous research has explored both biological and behavioral reasons that childbearing may contribute to deterioration of weak academic skills. Most of the research focused on the biological effects of pregnancy and childbearing on maternal cognition has been conducted in high-income countries and in lab settings. Researchers have described several plausible biological pathways contributing to deteriorating cognition during pregnancy and the postpartum period, including lack of sleep, depressed mood, and changes in hormone levels (De Groot
2006). Results have been mixed, with some studies finding negative effects of pregnancy and childbearing on cognitive function (De Groot
2006) and others finding differences only in reported cognition rather than objective performance (Casey
2000; Crawley et al.
2003). A 2007 systematic review found evidence of small but significant memory deficits among pregnant and postpartum women, particularly for tasks that placed high demands on respondents, such as free recall (retention in the absence of cues) and delayed free recall tasks (Henry and Rendell
2007). A prospective study with first-time mothers found substantial changes in brain structure during pregnancy in the region linked with social cognition, which has been found to be important for attachment. These changes were sustained for two years in new mothers. The same changes were not observed in first-time fathers, indicating a possible biological link.
14 The authors speculated that this restructuring process may contribute to cognitive deficits for pregnant and postpartum women (Hoekzema et al.
2017).
The included studies followed respondents for only a few years after they began childbearing, meaning they likely had not completed childbearing during follow-up. Although we are able to observe negative effects of childbearing on skills in that relatively short time frame, it is possible that those effects were temporary and that skills rebounded once childbearing was complete. A rebound in skills would be most plausible if negative effects were completely due to temporary biological changes, including those due to exhaustion. Given the time frame of follow-up—years rather than months for most study respondents—we believe it is more likely that the negative effects of childbearing on skills reflect changes in social roles, which are likely maintained after the first birth. Therefore, a rebound in skills seems less likely; rather, we may expect to see further deterioration in skills with longer follow-up. Some research in the United States (Hotz et al.
2005) and South Africa (Ranchhod et al.
2011) has found that the negative effects of childbearing on grade attainment are short-lived and that women are able to catch up to their peers by adulthood. In both of these settings, adolescent mothers commonly return to school after giving birth, allowing women to catch up with their peers. In all three settings included in these analyses, it is uncommon for young women to return to school after beginning childbearing.
Adolescent pregnancy and childbearing often leads to immediate and long-term shifting in social roles (Lloyd
2005), which may be most pronounced and sustained in settings with unequal gender norms around childrearing (Santhya and Jejeebhoy
2003). Few studies have investigated the social consequences of adolescent childbearing in LMICs. One exception, a qualitative study in India, found that married adolescent girls reported increased restrictions on their mobility and social interactions after marriage. The authors argued that compared with their unmarried peers and older women, married adolescents were less likely to be exposed to new ideas, resources, and even diverse topics of conversation (Santhya and Jejeebhoy
2003). Although the experience of adolescent childbearing is almost certainly selective, once a birth occurs, motherhood and school attendance are incompatible in most LMICs (Eloundou-Enyegue
2004; Lloyd and Mensch
2008). Therefore, narrowing opportunities to apply skills and limited exposure to new information may explain negative effects of adolescent childbearing on academic skills. If so, expanding adolescent mothers’ mobility and opportunities to apply their skills may protect against skill loss.
Regardless of the mechanisms explaining this relationship, skill loss after school leaving among adolescent girls may have important repercussions for adult literacy, economic productivity, and fertility. Literature in high-income countries has explored the motherhood wage penalty, seeking to identify the factors that account for lower earnings among women with children (compared with their childless counterparts). Common explanations have focused on (1) loss of work experience, (2) lost productivity at work, (3) trade-off of higher wages to focus on mother-friendly jobs, and (4) discrimination by employers (Budig and England
2001; Correll et al.
2007). Although conducted in very different economic and social contexts, this area of research raises questions about the longer-term economic impact of motherhood in LMICs, especially when childbearing begins during adolescence. Potential effects of childbearing on loss of skills are not explicitly captured by lost productivity theories, which tend to focus more on behavioral differences (less sleep, more distraction) (Budig and England
2001). Even in settings with less formal labor markets, however, lost skills very likely translate into a wage penalty for mothers (or perhaps a more general economic penalty), which may be most pronounced for adolescent mothers who were unable to secure employment before beginning childbearing.
In terms of effects on child health, young mothers with weakened skills may be those with the most pressing need to access accurate information on child health; to navigate services, including family planning services; and to successfully use contraception. As a result, despite progress in expanding women’s educational attainment in LMICs, expectations about economic empowerment and declines in fertility—and other wide-ranging improvements—may be overstated.
There are important limitations to our study. Our measures of literacy and numeracy are fairly simple and do not assess important domains. For example, the Programme for the International Assessment of Adult Competencies (PIAAC) defines literacy as “understanding, evaluating, using and engaging with written texts to participate in society, to achieve one’s goals, and to develop one’s knowledge and potential” (OECD
2012:19). The ability to read two simple sentences does not capture the more advanced components of understanding and engaging with written texts, but it also does not capture more foundational literacy skills, such as identifying letters or reading simple words. Our numeracy measure perhaps captures basic numeracy skills more effectively but fails to capture more advanced skills, in accordance with the PIAAC definition of numeracy as “the ability to access, use, interpret and communicate mathematical information and ideas, in order to engage in and manage the mathematical demands of a range of situations in adult life” (OECD
2012:34). If we conceptualize literacy and numeracy on a continuum, the assessments included in our study capture a snapshot of skills at the lower end of that continuum. More nuanced assessments may provide additional insight into the types of skills that are lost or gained after childbearing as well as more direction for effective interventions.
Although comparing findings across three settings and four study samples provides rich insights into consistent patterns in the effects of childbearing on academic skills, we are unable to investigate the important contextual variations within each site in a way that may further elucidate our results. None of the study samples are nationally representative, so our results should not be interpreted as indicating variations between countries, although the patterns in each sample likely reflect country-level differences, in part. Last, because these three studies were designed and implemented separately, they did not measure all independent variables, including academic skills, in exactly the same way. It is even more striking, then, that the results remain consistent between settings.