Child development and life outcomes are partly linked to prenatal and maternal conditions such as mother’s age at birth. Thus, the issue of teenage motherhood has attracted significant concern from researchers and policymakers because of its potential implications for children. The existing literature on effects of teenage motherhood on children is typically limited to weight at childbirth. Other studies are mainly descriptive in nature and do not account for selection bias associated with teenage mothers and their deprived environment resulting in their children also being brought up in similar environment.
This article examined the effects of teenage motherhood on child outcomes, specifically on children’s education, economic well-being, reported health status and body mass index (BMI).
Children (0–14 years) of teenage mothers (less than 20 years at first birth) in National Income Dynamics Survey (NIDS) data constitute the subjects under investigation in this study.
Using NIDS data, the study applied pooled regression, random effects model and propensity score matching (PSM) technique to examine the effect of teenage motherhood on child outcomes.
The study confirms that the PSM method is more robust to selection bias than pooled regression and random effect techniques. The findings from this study reveal that teenage motherhood significantly increases child grade repetition and economic dependency. However, teenage motherhood association with child health and BMI is found to be insignificant.
Teenage motherhood has far-reaching effects on children outcomes, thus proactive, reactive and post-active policies and programmes focusing on minimising the effect of teenage motherhood and enhancing children’s welfare are recommended.
It is well-documented that the physical, emotional, cognitive and social development that children experience from birth and childhood may have a long-enduring effect into their adult life (Excell
A report by Save the Children Organization (
For instance, the Bill of Right in the South African Constitution stipulates that ‘everyone has the right to basic education including adult basic education and further education, which the state through reasonable measures must make progressively available and accessible’ (Department of Education
The disruption in learners schooling because of teenage pregnancy may negatively impact their employability skills and financial well-being for themselves and their children. The high prevalence of early motherhood in the African and South African landscape has placed a compelling challenge on researchers to investigate the possible consequences and future implications.
A few empirical studies exist on the implications of teenage motherhood on children outcomes such as academic performance and health (Addo, Sassler & Williams
Existing literature on the impacts of teenage motherhood on children outcomes in the South African context is mainly limited to health (Branson, Ardington & Leibbrandt
In doing so, the study is designed to test the hypotheses that children of teenage mothers perform poorer in school, face economic challenges because of the circumstances of the mother and are often associated with poorer health outcomes than children born to delayed (non-teenage) mothers. It is believed that early childbearing poses high risks of health complications to the teenage mothers, which could affect foetal development and for that matter children’s health and cognitive ability later in life. Teenage childbearing also disrupts mothers’ schooling, which consequently deprives them of employable skills required to earn income to provide for the children’s need. Hence, the hypotheses of adverse effects of teenage childbearing on child outcomes are tested in this study.
This article is guided by the maternity theory that maternal characteristics such as age at motherhood, well-being and parenting affect child outcomes (ed. O’Reilly
The article is also guided by family system theory, which states that families are systemic and micro-social units (e.g. mothers, fathers and children) of interconnected relationships and action patterns where members grow, respond and interact with one another as partners (i.e. couple subsystem) and as sons and daughters (i.e. parent–child subsystem) (Burchinal, Vernon-Feagans & Cox
Becker (
One of the important early attempts to study the children outcomes of early motherhood was by Geronimus and Korenman (
Studies such as Hofferth and Reid (
Lanier and Zolotor (
It is pertinent to address the phenomenon in contexts where poverty and unemployment are higher such as South Africa while well-being in general is much lower compared with developed countries. In South Africa particularly, there is a limited empirical work on the impact of teenage motherhood on children outcomes. In his historical studies in South Africa, Macleod (
Firstly, by comparison, there is little knowledge on the intergenerational effects of teenage motherhood on the economic, educational and later health outcomes of children especially in South Africa where poverty and unemployment levels are high and well-being is lower in general. Secondly, the existing studies in South Africa, although largely limited to health outcome, drew differing conclusions on birthweight effect of teenage motherhood. Further study is therefore needed to consolidate the findings using a more robust estimation methodology. The limitations of existing literature have motivated this study with a methodology better suited to address selection bias (because of lack of randomisation) in order to answer the question of how teenage motherhood affects child outcomes (education, economic outcomes and health) in South Africa.
This article uses data from National Income Dynamics Survey (NIDS) conducted by a research unit of the University of Cape Town, South African Research and Labor Research Unit (SALDRU). National Income Dynamics Survey that began its data collection in 2008 includes a nationally representative sample of individuals (over 28 000) and households (over 7300 families) across the country and this survey is repeated every 2–3 years, with the same and additional members. The NIDS data are partitioned broadly into household data, adult data and child data sets amongst others. The child data set contains information of children below 15 years of age (0–14 years). In child data set, NIDS defined children as those below 15 years of age, that is, 0–14 years, which is also a vulnerable age group (Pearson & Stone
The mother’s age at first birth was identified by finding the difference between the year at which she had her first child and the year in which she was born. Each first-born child in the same data set was classified as being born to a teenage mother (less than 20 years) or to a non-teenage mother (20 years and above).
By using the unique person code, each child’s mother is linked to the information in the adult data set. The data provide relevant variables of interest in analysing the effects of teenage motherhood on children outcomes. Firstly, the data provide information on children in the areas of economic status educational outcome, health, weight and other anthropometric measures and demographic characteristics. Information on mothers, those who gave birth at least once, includes maternal and background characteristics such as time of first birth. Furthermore, essential information on family structure environment including family types, household head composition, marital status, biological relation, family size and background factors is also contained in the data set. The study is restricted to women who have given birth at least once and their children.
The question ‘Does anyone currently receive a Child Support Grant (CSG), Foster Care Grant (FCG) or Care Dependency Grant (CDG) for this child?’, which yields a binary ‘yes’ or ‘no’ response, is used to assess the economic dependency of a child. The ‘yes’ response may be considered as a reflection of mother’s inability to meet child’s economic needs. The educational outcome of the children is assessed using the question ‘Has this child ever repeated a grade?’, again with a binary response. The health outcome of the child is assessed using indicators: reported health status and body mass index (BMI). The study calculates the BMI based on the height and weight reported in the child data set. The health status is obtained from the question, ‘Overall, how is this child’s health, would you say it is excellent, very good, good, fair or poor?’. The responses were then collapsed to binary form where 1 denotes good health (excellent health, very good health, good health) and 0 otherwise.
This study employs three progressive econometric strategies to investigate the effects of teenage motherhood on children outcomes. These are pooled regression (logit and OLS), random effect binary model and propensity score matching (PSM) technique.
Each Child outcome variable (CHME) is an explained variable in
The pooled models, however, do not account for the possible heterogeneous or child-specific effects that might be associated with each child; hence, a random effects binary model is used to account for the individual effects.
But
EM is the early motherhood variable. The model (2) is fully specified as:
In this analysis, maximum likelihood (ML) is used as the estimator for the random effects binary logic model while generalised least squared (GLS) estimator was used for the estimation for the BMI.
Although pooled regression models and random effects models are useful in their own merit, the confounding factors (such as income and common living environment of the mother and child) could bias the relationship between age of motherhood and child outcomes. Hence, the PSM technique is employed to account for the selection bias with pooled data (Leuven & Sianesi
Propensity score matching technique normally uses treatment estimators to estimate two parameters: The average treatment effect (
Average treatment effect on the treated (
The propensity score methodology follows probit wmodel for propensity estimation for treatment group (teenage mothers) using the nearest neighbour matching technique. Treatment
where
The ATET, is the mean of the difference between the treated and the controlled group, (
Thus, the probit model for the treatment is further specified here:
where
The treated and untreated groups can be compared for similarity by means or using medians of continuous variables and the distribution of their categorical counterparts (balancing property). The standardised form can also be used especially when different units are also involved. The standardised difference can be used to do the comparison between the mean of continuous variables and that also for binary variables between treatment groups (Austin
Summary statistics.
Variables | Definitions | Children born to teen mothers-whole sample |
Children born to non-teen mothers-whole sample |
---|---|---|---|
Percentages (%) | Percentages (%) | ||
Economic well-being | = 1 child depends on CSG, otherwise 0 | 85.1 | 68.7 |
Education | = 1 repeated grades, otherwise 0 | 15.8 | 10.2 |
Excellent | =1 Excellent, otherwise 0 | 45.5 | 49.6 |
Very good | =1 Very good, otherwise 0 | 32.7 | 31.7 |
Good | =1 Good, otherwise 0 | 18.4 | 15.9 |
Fair | =1 Fair, otherwise 0 | 2.3 | 2.0 |
Poor | =1 Poor, otherwise 0 | 0.9 | 0.7 |
Binary health outcome; good health | =1 Excellent, otherwise 0 | 77.9 | 70.0 |
Good health | =1 Excellent, otherwise 0 | 20.03* | 20.05* |
Body mass index* | Measured in kg per meter square | (3.019) | (3.07) |
Intact married | = 1 Intact married, otherwise 0 | 24.9 | 25.1 |
Out-of-wedlock | = 1 Out-of-wedlock, otherwise 0 | 59.3 | 57.2 |
Divorced | = 1 Divorced, otherwise 0 | 1.1 | 1.5 |
Deceased parents | = 1 Deceased parents, otherwise 0 | 11.8 | 7.1 |
Extended family | = 1 Extended family, otherwise 0 | 81.1 | 67.1 |
Male head | = 1 Male head otherwise 0 | 24.1 | 26.3 |
Female head | = 1 Female head otherwise 0 | 41.8 | 43.4 |
Grandparent head | = 1 Grandparent head Otherwise 0 | 29.5 | 13.9 |
Family size* | Family size is measured as the number of members per household | 5.24* | 5.089* |
(2.39) | (2.36) | ||
- | - | - | |
African | = 1 African, otherwise 0 | 77.1 | 71.6 |
Asians | = 1 Asians, otherwise 0 | 0.4 | 0.9 |
Mixed race | = 1 Mixed race, otherwise 0 | 10.8 | 12.9 |
White Gender | = 1 White, otherwise 0 | 1.9 | 1.5 |
Female | = 1 female, otherwise 0 | 43.5 | 43.3 |
Mothers’ education* | Years of formal school completed | 11.681* |
13.891* |
Source of water | = 1 Has good water source, otherwise 0 | 81.8 | 86.8 |
Source of toilet | = 1 Has good toilet facility, otherwise 0 | 97.7 | 98.7 |
Household income (per capital)# | Household income in rand | 707.989 |
1094.1 |
- | - | - | |
Rural/traditional | = 1 Rural/traditional, otherwise 0 | 55.7 | 50.2 |
CSG, Child Support Grant.
Statistics of variables with asterisk ‘*’ are means and standard deviations are in parenthesis.
Results of the effects of teenage motherhood on children’s economic dependence and grade repetition are documented in
Results of pooled logistic and random effect models: Effects of early motherhood on economic dependency and grade repetition.
Variables | Economic dependency |
Grade repetition |
||||||
---|---|---|---|---|---|---|---|---|
Pooled logistic model: Marginal effect |
Random effect model: Marginal effect |
Pooled logistic model: Marginal effect |
Random effect model: Marginal effect |
|||||
Coefficients | Standard errors | Coefficients | Standard errors | Coefficients | Standard errors | Coefficients | Standard errors | |
Teen motherhood | 0.178 |
0.037 | 0.116 |
0.043 | 0.066 |
0.016 | 0.032 |
0.013 |
Family type: Extended | 0.001 | 0.008 | 0.004 | 0.010 | 0.002 | 0.005 | 0.001 | 0.006 |
Female-headed household | 0.021 |
0.009 | 0.027 |
0.010 | 0.030 |
0.018 | 0.025 |
0.015 |
Grandparent-headed household | 0.077 |
0.011 | 0.056 |
0.012 | −0.101 |
0.010 | −0.141 |
0.089 |
Intact married household | −0.061 |
0.015 | −0.056 |
0.014 | −0.079 |
0.017 | −0.101 |
0.014 |
Out-of-wedlock household | 0.040 |
0.014 | 0.030 |
0.014 | 0.076 |
0.022 | 0.079 |
0.026 |
Divorce/separate | 0.017 | 0.012 | 0.015 | 0.011 | 0.001 | 0.001 | 0.0001 | 0.001 |
Household size | 0.054 |
0.013 | 0.053 |
0.012 | 0.026 |
0.008 | 0.013 |
0.007 |
Education of mother | −0.006 |
0.001 | −0.006 |
0.001 | −0.053 |
0.006 | −0.053 |
0.005 |
Log household income | −0.050 |
0.018 | −0.055 |
0.0183 | −0.051 |
0.005 | −0.053 |
0.007 |
Age of the child | −0.021 | 0.018 | −0.011 | 0.019 | 0.014 | 0.017 | 0.011 | 0.014 |
Gender: Female | 0.007 | 0.021 | 0.005 | 0.019 | −0.002 | 0.010 | −0.012 | 0.010 |
Geography: Rural | 0.074 |
0.009 | 0.080 |
0.011 | 0.024 |
0.006 | 0.024 |
0.007 |
Employment status: Unemployed | 0.059 |
0.009 | 0.037 |
0.009 | 0.047 |
0.017 | 0.024 |
0.015 |
African race | 0.078 |
0.012 | 0.088 |
0.015 | 0.042 | 0.059 | 0.029 | 0.059 |
Asians | 0.118 | 0.073 | 0.112 | 0.089 | −0.051 | 0.116 | −0.062 | 0.112 |
Mixed race | 0.250 |
0.046 | 0.292 |
0.055 | 0.058 | 0.061 | 0.051 | 0.062 |
LR chi2 | 639.4 | - | - | - | 525.91 | - | - | - |
Pseudo R2 | 0.0984 | - | - | - | 0.0552 | - | - | - |
Wald chi2 | - | - | 312.47 | - | - | - | 483.96 | - |
Probability > chi2 | 0.0000 | - | 0.000 | - | 0.0000 | - | 0.000 | - |
Number of observation | 11719 | - | 11719 | - | 11603 | - | 11603 | - |
LR, Likelihood ratio.
Standard errors are indicated in parentheses,
,
,
,
In terms of children’s educational outcomes, both the pooled logistic and random marginal effect models indicate teenage motherhood as significant positive predictors of grade repetition (
In addition, teenage mothers often are not able to engage in productive economic activities, thus limiting their ability to provide adequately for their children, which in turn impacts their performance at school. In many cases children born to out-of-wedlock teenage mothers, which is widespread in South Africa, stay with relatives other than biological parents and often do not receive the needed parental care and encouragement that motivate good academic performance (Makiwane
A significant positive effect of households headed by females and grandparents on the economic dependency of the children was identified. This is not surprising given the income criteria used by the South African Social Service Agency (SASSA) to determine beneficiaries of CSG. Female-headed households carry a higher risk of poverty in South Africa (Anakpo & Kollamprambil
On the other hand, children living in households headed by females have significantly higher probability of grade repetition. The high prevalence of out-of-wedlock births especially amongst young mothers in South Africa (see
Moreover, intact-married families have a significant negative relationship with grade repetition of the children (
Our results show household size to be a positive predictor of a child’s economic dependency (
The results do not show any significant effect of teenage motherhood on the children’s reported health status and BMI (
Results of pooled logit and random effect model: Effects of early motherhood on child health and body mass index.
Variables | Child perceived health |
Body mass index |
||||||
---|---|---|---|---|---|---|---|---|
Pooled logit model: Marginal effect |
Random effect model: Marginal effect |
Pooled |
Random effect model |
|||||
Coefficients | Standard errors | Coefficients | Standard errors | Coefficients | Standard errors | Coefficients | Standard errors | |
Teen motherhood | 0.031 | 0.036 | 0.027 | 0.031 | 0.257 | 0.262 | 0.287 | 0.266 |
Family type: Extended | 0.013 | 0.012 | 0.013 | 0.012 | −0.085 | 0.093 | −0.092 | 0.096 |
Female-headed household | −0.018 | 0.015 | −0.018 | 0.015 | 0.248 |
0.111 | 0.241 |
0.111 |
Grandparent-headed household | −0.077 |
0.013 | −0.079 |
0.027 | 0.538 |
0.224 | 0.555 |
0.228 |
Intact Married household | 0.085 |
0.033 | 0.083 |
0.033 | −0.295 | 0.237 | −0.637 | 0.591 |
Out-of-wedlock household | −0.069 |
0.018 | −0.069 |
0.019 | 2.034 |
0.145 | −0.023 | 0.393 |
Divorce/separate | −0.018 | 0.012 | −0.020 | 0.012 | −0.386 | 0.471 | −0.717 | 0.697 |
Household size | −0.042 | 0.053 | −0.043 | 0.053 | −0.039 | 0.061 | −0.040 | 0.073 |
Log household income | 0.022 |
0.078 | 0.044 |
0.035 | 0.618 |
0.209 | 0.723 |
0.217 |
Child support grant | 0.076 |
0.023 | 0.075 |
0.023 | 1.276 |
0.153 | 1.293 |
0.157 |
Age | −0.017 | 0.029 | −0.018 | 0.021 | −0.345 |
0.064 | −0.348 |
0.072 |
Gender of the child | −0.027 | 0.024 | −0.024 | 0.025 | 0.068 | 0.309 | 0.049 | 0.333 |
Toilet | 0.016 | 0.035 | 0.015 | 0.035 | 0.138 | 0.864 | −0.107 | 0.519 |
Sanitation water | 0.022 |
0.013 | 0.023 |
0.014 | −0.317 | 0.340 | −0.297 | 0.458 |
Education of Mother | 0.005 | 0.001 | −0.005 | 0.001 | 0.067 |
0.028 | 0.056 |
0.021 |
Geography: Rural | 0.008 | 0.010 | 0.007 | 0.011 | −0.539 | 0.887 | −0.595 | 0.525 |
Employment status: Unemployed | −0.011 | 0.010 | −0.011 | 0.010 | −0.277 |
0.303 | 0.3210 | 0.368 |
African race | 0.069 |
0.017 | −0.068 |
0.017 | −1.567 |
0.445 | 1.503 |
0.574 |
Asia | 0.025 | - | 0.157 | 0.105 | −0.544 |
0.234 | 3.627 | 3.652 |
Mixed race | 0.155 | 0.104 | 0.099 |
0.051 | - | - | - | - |
Constant | - | - | - | - | 6.022 |
1.628 | 8.016 |
1.961 |
Likelihood ratio chi2 | 135.83 | - | - | - | ||||
Pseudo R2 | 0.0197 | - | - | - | ||||
Wald chi2 | - | 122.44 |
- | - | ||||
R2 | - | - | 0.240 | - | ||||
Adjusted R2 | - | - | 0.215 | - | ||||
- | - | 6.69 | 2.28 | |||||
> |
- | - | 0.0000 | 0.0044 | ||||
Number of observation | 11710 | 11710 | 11605 | 11605 |
OLS, Ordinary least squares.
Standard errors are indicated in parentheses,
,
,
,
Propensity scores and summary result of matching assessment are reported in
Propensity scores and summary result of matching assessment.
Propensity score | Teen mothers | Non-teen mothers | Total |
---|---|---|---|
0.0555 | 28 | 175 | 203 |
0.1 | 29 | 331 | 360 |
0.15 | 519 | 2395 | 2914 |
0.2 | 93 | 331 | 424 |
0.25 | 1352 | 3245 | 4596 |
0.3 | 1020 | 2362 | 3382 |
0.35 | 122 | 201 | 323 |
Propensity scores and summary result of matching assessment.
Sample | Ps R2 | chi2 | MeanBias | MedBias | % var | |||
---|---|---|---|---|---|---|---|---|
Unmatched | 0.019 | 266.67 | 0 | 4.8 | 4.1 | 21.5 | 0.68 | 0 |
Matched | 0.0001 | 2.78 | 0.997 | 0.8 | 0.6 | 4.2 | 1.3 | 0 |
LR, Likelihood ratio.
Results from propensity score matching model: Effects of teen motherhood on child outcomes.
Parameters | Economic effect |
Educational effect |
Health effect |
Body mass |
||||
---|---|---|---|---|---|---|---|---|
Coefficients | Standard errors | Coefficients | Standard errors | Coefficients | Standard errors | Coefficients | Standard errors | |
ATE | 0.167 |
0.080 | 0.058 |
0.023 | 0.009 | 0.013 | 0.013 | 0.015 |
ATET | 0.175 |
0.078 | 0.048 |
0.024 | 0.082 | 0.008 | 0.011 | 0.011 |
Γ = 1 prob(1) | 0.031082 | 0.018953 | 0.243951 | 0.256044 | ||||
Γ = 1.5 prob(1) | 0.0169 | 0.001670 | 0.304600 | 0.227894 | ||||
Γ = 2 prob(1) | 0.001767 | 0.000117 | 0.33698 | 0.216473 | ||||
Observation | 7237 | 7358 | 7358 | 7358 |
ATE, Average treatment effect; ATET, Average treatment effect on the treated.
Standard errors in parentheses
,
,
,
Sensitivity analysis results are also reported in
The PSM analysis results show that the children of teenage mothers have a significantly higher probability of dependency on the meagre CSG from government. This further reduces the prospects of transitioning from poverty because the CSG amount in South Africa is lower than the food poverty line. Therefore, although the grant may ensure sustenance and survival, it is barely minimum and insufficient to provide the child with a fair opportunity in life. This is reflected in the educational outcomes, where the children of teenage mothers are found to have a higher probability of repeating grades. This is not surprising given the established fact that the strongest predictor of a child’s educational attainment is the mother’s education (Abuya et al.
It is well-documented that the life outcomes of children are partly linked to prenatal and maternal conditions such as mother’s age at birth. Thus, the issue of early motherhood has received considerable attention from researchers and policymakers because of its potential implications not just for the mother but also for the next generation. The high prevalence of early motherhood in South Africa has placed a compelling challenge on researchers to investigate the possible consequences and future implications. The existing literature on effects of teenage motherhood on children are either limited to child birthweight or are descriptive studies that do not account for selection bias associated with teenage mothers and their deprived environment resulting in their children also being brought up in similar environment. Hence, based on the NIDS data, this article examined the effects of teenage motherhood on children outcomes, especially on children’s economic well-being, education, health outcomes and BMI using, pooled regression, random effects model and PSM technique. The study confirms that the latter estimation strategy (PSM method) is more robust to selection bias than the pooled regression and logit panel models. The findings from this study thus reveal that teen motherhood significantly increases child grade repetition and economic dependency. Furthermore, teenage motherhood association with child health and BMI is, however, insignificant.
Based on the findings from the study, it is recommended that the Department of Education’s Policy on measures for the prevention and management of learner pregnancy should be implemented in full scale (at all schools as stated) to minimise teenage pregnancy prevalence along with the enforcement of the SASA (RSA
Secondly, welfare programmes should be designed within the framework of the development policy keeping in mind the need to avoid the perverse incentive of increased fertility to access the CSG (Kollamparambil
Lastly, given the finding that intact married consistently predicts better children outcomes, it is recommended that marital education programme be instituted with particular focus on the existing couples and those preparing for the institution of marriage to support healthy marriages to minimise divorce and its associated problems. This will greatly help the existing couples and prepare the prospective couples for married life, and to be aware of some of the typical marital challenges and ways to resolve them and this will also go a long way to provide complementary economic support and congenial environment for better children outcomes.
Firstly, the finding from teenage motherhood effects on children outcomes is not to be generalised to all the children in South Africa. The findings only reflect teenage mothers as contained in the data used in the analysis and the effects on the outcomes of children born to them. The study is limited by unavailability of long-time data to thoroughly investigate the dynamics of child outcomes over long period and to disentangle these effects from other confounding factors other than early motherhood. The PSM does not deal with the problem of unobservable characteristics; however, the analysis uses the rich information available in the data and performs sensitivity checks to ensure the reliability of results.
The authors acknowledge the University of the Witwatersrand.
The authors declare that they have no financial or personal relationships that may have inappropriately influenced them in writing this article.
G.A. was responsible for conceptualisation, methodology, formal analysis, investigation and original draft writing. U.K. was involved in supervision, writing reviews and editing and suggestions that enhance the conceptualisation, methodology and write-up.
This article followed all ethical standards for research without direct contact with human or animal subjects.
The support of the DST-NRF Centre of Excellence in Human Development and African Economic Research Consortium towards this research/activity is hereby acknowledged.
Data can be accessed via
Opinions expressed and conclusions arrived at, are those of the authors and are not necessarily to be attributed to any of the two institutions, DST-NRF Centre of Excellence in Human Development and African Economic Research Consortium.
In this article, child’s economic dependence or outcome means that parents or caregivers do not have the financial means to care for the children and so depend on CSG for the child’s needs.