Who cares for children? A quantitative study of childcare in South Africa

ABSTRACT

In this study we investigate the gender division of labour in the physical and financial care of children in South Africa, in the context of large race differences in rates of union formation and parent–child co-residence. Using national micro-data, we show that across a variety of household forms, women are the primary caregivers of children even when they are not a child’s biological mother. Women are also more likely to provide physical care than men are to provide financial support for children’s schooling. However, this gender imbalance is far larger in the case of African children, the majority of whom do not live with their father. For most African children, both their physical and financial care is provided by women who are typically the child’s biological mother, but also the child’s grandmother or another female relative.

KEYWORDS:

• Childcare
• family formation
• gender division of labour
• absent parents

1. Introduction

With a traditional gender division of labour in the care of children, women assume primary responsibility for the physical care of children, while men provide financial care. This division of labour has become more elastic as women’s participation in the labour market has increased and perceptions of fatherhood have widened beyond the role of financial provision. However, many studies identify that routine childcare activities (feeding, bathing, taking children to school and putting them to sleep) are still more likely to be provided by women than men (Monna & Gauthier, 2008; Craig & Mullan, 2011).

In this study,¹ we describe the provision of childcare in South Africa using nationally representative data collected in the National Income Dynamics Study (NIDS). Women’s greater responsibility for unpaid work, including caring labour and childcare specifically, has been well documented in the quantitative analysis of data collected in South African Time Use Surveys (Budlender et al., 2001; Charmes, 2006; Guryan et al., 2008). Using the NIDS data, we are able to contribute to this quantitative literature on childcare in a number of ways. First, we investigate how women are related to the children for whom they provide physical care and the extent to which women are the sole providers of childcare. Second, we compare race² differences in the provision of childcare, and we show how these differences are related to the type of household in which children live and to the status of their parents. As poverty rates are highest among African households, and are higher still in households that include more children (Posel et al., 2016), the comparisons we draw also reflect class differences in the nature of childcare. Third, in contrast to earlier studies, we explore also the ‘financial care’ of children and we examine whether women’s greater responsibility for the physical care of children is balanced by men’s greater responsibility for their financial care. In addition, we investigate the financial support provided by parents (both fathers and mothers) who do not live with their children.

In the next section, we briefly review research on childcare in South Africa, and in section 3, we outline the data used in the study. In section 4, we describe the physical care of children, and in section 5, we investigate the financial support of children and the contributions made by parents who do not live with their children. In the final section, we summarise and discuss the main findings from the descriptive analysis.

2. Caring for children in South Africa: A brief review

As in many other countries, the provision of childcare in South Africa has been described quantitatively using data collected in Time Use Surveys (TUS). To date, two Time Use Surveys have been conducted, in 2000 and 2010. Both surveys show that women spend considerably more time on childcare than men. Using the 2000 TUS, for example, Guryan et al. (2008) find that among adults (aged 21–55) who live with at least one child, women spend an average of 5.9 hours per week on childcare compared with 0.6 hours by men (see also Budlender et al., 2001; Charmes, 2006). Similar gender differences are identified in the more recent (2010) survey, where the average amount of time that women spend on childcare is more than eight-fold larger than the time spent by men (Statistics South Africa, 2013:36).

For a study of childcare, however, time use data have several limitations. First, physical childcare is described from the perspective of the person providing the care, and not from the perspective of the child. This means that it is difficult to identify who provides care to which child, particularly in the complex types of households that are found in South Africa. Second, with its focus on time use, information is collected only on physical care, and not on the financial care of children, or who pays for the direct costs of children.

The majority of African children in South Africa do not grow up in two-parent households, and where children do live with a parent, this is far more likely to be their mother than their father (Hall & Posel, 2012). Parental absence from the child’s household is partly the result of high rates of mortality (Ardington & Leibbrandt, 2010). However, most parents who are absent from the child’s household are not deceased but are living elsewhere (Posel & Devey, 2006). Fathers are more likely than mothers to live separately from their children partly because of male-dominated patterns of labour migration and the particular institutional constraints of the apartheid period, which prevented Africans from settling permanently in (white) urban areas with their partners and children. These restrictions, together with the contractual nature of employment, meant that many labour migrants (most often men) would retain a base in the household from which they had migrated, to which they would return each year, and which was their ‘permanent’ home. Although discriminatory apartheid legislation began to be removed in the 1980s, temporary labour migration is still an important part of labour supply in South Africa (Posel & Casale, 2003; Collinson et al., 2007). In 2008, almost 22% of African households in rural areas reported a labour migrant as a non-resident household member (Posel, 2010).

However, many African fathers are not part of their child’s household, even as non-resident household members. This paternal absence derives from low and declining marriage rates among Africans and high rates of non-marital pregnancy, with children then living with their mother rather than their father (see, for example, Casale & Posel, 2002; Hosegood et al., 2009; Hunter, 2010; Posel et al., 2011). Over the past two decades, the labour migration of African women has also increased, particularly from rural areas of the country (Posel, 2010; Camlin et al., 2014), and although children are far more likely to co-reside with their mother than their father, a sizeable share lives in households with neither parent (Hall & Meintjies, 2016). In these households, extended family networks offer an important source of childcare (Russell, 2003; Moore, 2013; Makiwane et al., 2016), and several qualitative studies describe the role, particularly of grandmothers, in providing care to children (see for example Schatz & Ogunmefun, 2007; Bohman et al., 2009).

By law, economic support from absent parents should be enforced by the South African Maintenance Act 99 of 1998, which stipulates that both parents are responsible for the financial support of children regardless of the relationship status of the parent (Department of Justice and Constitutional Development, 2010). Given that children are far more likely to live with their mother than their father, the Maintenance Act is intended primarily to secure economic support from fathers. However, many fathers, and particularly young fathers, do not acknowledge paternity (Varga, 2003), and with high unemployment rates in South Africa, men may not have the economic means to support children (Makusha et al., 2012). Moreover, the Act is very poorly enforced even where fathers acknowledge paternity and can afford childcare payments (Bonthuys, 2008; Carnelley, 2012).

A number of recent studies, however, challenge the presentation of non-resident fathers as ‘disengaged and shirking their responsibilities’ (Clark et al., 2015:576). This research highlights the contributions that non-resident fathers make to their children, both through financial support and through interactions with children, such as helping with homework and offering moral guidance (Madhavan et al., 2008, 2014; Makusha et al., 2012; Clark et al., 2015). However, many of these studies have been based on qualitative research or on micro-data drawn from regionally specific samples or specific cohorts of men and children. Moreover, the involvement of mothers who do not live with their children has received little attention in the South African literature.

In this study, we use national micro-data to examine the provision of physical and financial care to children. These data are collected from the perspective of children, and we can therefore identify the relationship between a child and the person, or people, providing physical care or economic support to the child. Our primary objective is to describe national patterns in childcare support, given the complexity of household formation in South Africa. We consider specifically the extent to which the care of children is provided by the children’s parents, whether the provision of both physical and financial childcare is gendered, and whether fathers and mothers provide economic support to, and maintain contact with, children with whom they do not co-reside.

3. Data and methods

The data for the study come from the baseline wave of the National Income Dynamics Study (NIDS) undertaken in 2008, which sampled approximately 7300 households and 28 000 individuals. Although subsequent waves of NIDS have been conducted, we focus our analysis on the first wave, where the nationally representative sample provides a benchmark against which analysis of the subsequent waves can be compared.

In contrast to many of the nationally representative household surveys conducted in the country, NIDS is distinctive because it adopts a broad definition of household membership and allows respondents to identify people as household members even if they lived away from the household for much of the year. In addition, the survey instrument includes a detailed child questionnaire for all household members younger than 15, as well as a household questionnaire (including the household roster), and an adult questionnaire for all members older than 14.

In this study, we use descriptive methods to analyse responses to three sets of questions included in the child questionnaire. The first concerns the physical care of children. Respondents were asked to report on ‘who is the main person responsible for making sure that this child is fed, bathed, goes to school if of school-going age, helped with homework, taken care of when the child is ill, etc?’ (Question E1 of the 2008 NIDS Child Questionnaire), and how this person was related to the child (Question E2). A subsequent question asked ‘who else helps to care for the child?’ (Question E3), with multiple responses permitted (the highest number of helpers for any one child was nine). The second set of questions concerns financial support for schooling. Respondents were asked who paid for the child’s educational expenses and the person’s relationship to the child, where up to three people could be identified (Question C14).

Using information collected in the first two sets of questions, we can match the person identifier of the primary physical or financial caregiver of the child, to the person identifier in the main household roster. In this way, we are also able to identify whether caregivers are resident in the child’s household, whether they are a member of the child’s household but not currently resident (as in the case of a temporary labour migrant), or whether they are ‘absent’ from, or not part of, the child’s household (for example, because the parents are divorced). Among all carers who are household members, we can identify their sex and their relationship to the child (including whether they are the child’s biological mother or father, or grandparent). However, we have no further information on caregivers who are not part of the child’s household (and therefore do not appear on the household roster).

Later in the child module, respondents were also asked to identify the residency status of the child’s biological mother and biological father, the frequency of contact between a child and a non-resident living parent, and whether the parent provided any financial support to look after the child (Questions E5 to E12). We use this third set of questions to probe the involvement of non-resident parents in the lives of their children.

The NIDS questionnaire is administered to the ‘mother/caregiver of the child or another household member who is knowledgeable about the child’. It is possible that respondents under-report the involvement of other individuals in the care of children. For example, mothers may misrepresent the extent of care provided by the child’s father, particularly when parents are not co-resident (see for example Coley & Morris, 2002; Mikelson, 2008). However, this concern may be mitigated somewhat by the possibility for respondents to list multiple individuals who assist, both with the physical care of children and with the payment of school fees, and by the inclusion of a series of questions which ask specifically about the contributions of parents not living in the household.

Because rates of union formation are far lower among Africans, while mortality rates from HIV/AIDS and rates of labour migration are far higher, we distinguish African children from other children. Included in the group of other ‘non-African’ children are Coloured, Indian and white children, but sample sizes in NIDS are too small to further disaggregate these other children by race (only approximately 16% of all children in the sample are non-African).³ We use the comparison between African and non-African children to highlight the distinctive characteristics of childcare among African children, and to link these characteristics to parental status and the nature of household formation.

4. The physical care of children

4.1. The characteristics of caregivers

Table 1, which describes the sex of primary caregivers in South Africa, shows that for the vast majority of children (14 years and younger), women are the primary caregivers, or the main person responsible for ensuring that the child is fed, bathed, attends school if eligible, and is taken care of when sick.⁴ However, the table also describes differences by the age and race of the child. First, among both African and non-African children, the share of primary caregivers who are female is highest among very young children (0–3 years). Second, women’s primary responsibility for childcare appears stronger among African children than other children. Regardless of the age of the child, less than 8% of African children were reported as having a male primary caregiver, compared with between 8 and 14% of non-African children.

Table 1. Sex of the primary physical caregiver of children, 2008.

Proportion of children where the caregiver is:	Ages 0–3	Ages 4–6	Ages 7–14
	African
Female	0.966 (0.007)	0.950 (0.008)	0.922 (0.006)
Male	0.034 (0.007)	0.050 (0.008)	0.078 (0.006)
	Non-African
Female	0.918 (0.028)	0.864 (0.040)	0.883 (0.025)
Male	0.082 (0.028)	0.136 (0.040)	0.117 (0.025)

Source: NIDS 2008.

Note: The data are weighted. Standard errors are in parentheses. The primary caregiver is the main person who is responsible for ensuring that the child is fed, bathed, goes to school if eligible, helped with homework and taken care of when sick. Proportions are calculated for the sample of children aged 0–14 for whom information on the sex and age of their primary caregiver is not missing.

For most children, other people are also identified as helping to provide care. Table 2 shows that approximately 78% of all children were reported as having a primary caregiver as well as at least one other person who provided childcare assistance. Perhaps surprisingly (given that African children live in households with more female adults than do non-African children (Posel et al., 2016)), African children are less likely than other children to receive care from more than one person. For example, whereas 20% of very young African children had a sole care-provider, this was the case for only 9% of very young non-African children. This may be explained, at least in part, by the far lower presence of fathers in the households of African children, and also because domestic workers (who may assist with childcare) are less likely to be employed in these households (this is discussed further below). However, among both African and non-African children, the sole provision of care is more likely as children grow older and the demands of childcare get smaller.

Table 2. Assistance in the provision of physical care to children.

Percentage of children:	One caregiver		More than one caregiver
All children	22.48 (0.006)		77.52 (0.006)
		African
Children (0–3 years)	20.39 (0.012)^a		79.61 (0.012)^a
Children (4–6 years)	23.14 (0.015)^a		76.86 (0.015)^a
Children (7–14 years)	26.96 (0.009)^a		73.04 (0.009)^a
		Non-African
Children (0–3 years)	9.41 (0.026)		90.59 (0.026)
Children (4–6 years)	10.09 (0.032)		89.91 (0.032)
Children (7–14 years)	13.92 (0.020)		86.08 (0.020)

Source: NIDS 2008.

Note: The data are weighted. Standard errors are in parentheses. The percentages are conditional on more than one caregiver being reported.

^aPercentages of African and non-African children are significantly different at the 95% confidence level.

Table 3 describes how primary caregivers are related to the children to whom they provide care. Across all ages of children, the majority of children are cared for by women who are their biological mothers.⁵ However, younger children are more likely than older children to receive primary care from their biological mother. Across all age groups, African children who do not receive primary care from their mother are far less likely to receive this care from their father than from other women in the household, and particularly from their grandmothers. Among non-African children, in contrast, fathers play a larger role in primary childcare provision than other women.

Table 3. Relationship of primary caregivers to children, 2008.

Proportion of children where the caregiver is:	Ages 0–3	Ages 4–6	Ages 7–14
	African
Mother	0.745 (0.013)	0.663 (0.017)	0.617 (0.010)
Father	0.029 (0.007)	0.033 (0.007)	0.054 (0.005)
Grandmother	0.159 (0.010)	0.204 (0.013)	0.183 (0.008)
Other female	0.061 (0.007)	0.085 (0.010)	0.126 (0.007)
Other male	0.005 (0.002)	0.015 (0.004)	0.020 (0.003)
	Non-African
Mother	0.846 (0.033)	0.794 (0.045)	0.782 (0.027)
Father	0.064 (0.026)	0.105 (0.036)	0.103 (0.024)
Grandmother	0.036 (0.011)	0.050 (0.018)	0.057 (0.010)
Other female	0.045 (0.018)	0.023 (0.008)	0.044 (0.009)
Other male	0.008 (0.007)	0.028 (0.022)	0.014 (0.008)

Source: NIDS 2008.

Note: The data are weighted. Standard errors are in parentheses. Proportions are calculated for children for whom there is non-missing information on the sex and age of their primary caregiver, and on the child’s relationship to their caregiver.

Employing a domestic worker to assist with household chores and childcare has been a common practice in South Africa. According to the 2008 NIDS data, approximately 17% and 37% of African and non-African households respectively indicated that they had paid for domestic labour and/or childcare in the past 30 days. Yet less than 1% of all children were reported as receiving primary care from someone who was unrelated to the child and was paid to provide care (a category which would include domestic workers). It may seem surprising that domestic workers are not identified as the primary caregivers of more children. However, the question on primary care asks for the main person who is responsible not only for the day-to-day care of the child but also for helping the child with homework and taking care of the child’s health, roles which domestic workers are less likely to perform. Domestic workers were considerably more likely to be identified as ‘assistant’ caregivers, particularly in the case of non-African children.⁶

4.2. Caregiving and household type

Race differences in the profile of caregivers are related to differences in the nature of children’s households⁷ as characterised by the status of parents. Table 4 shows that for both African and non-African children, there is little difference in the gender division of labour in childcare when households include both parents – approximately 11–12% of children receive primary care from their fathers, and 88–89% from their mothers. However, non-African children are far more likely to live with both parents (63%) than African children (27%).

Table 4. Parent mortality and primary caregivers of children by household type, 2008.

	African
	Nuclear	Parents & other adults	Mother (no other adults)	Mother & other adults	Father (no other adults)	Father & other adults	Other adults only
Proportion of children	0.133 (0.006)	0.140 (0.006)	0.084 (0.005)	0.362 (0.007)	0.004 (0.001)	0.021 (0.002)	0.242 (0.006)
Mother is dead					0.498 (0.124)	0.243 (0.045)	0.258 (0.012)
Father is dead			0.207 (0.025)	0.155 (0.009)			0.263 (0.012)
Mother is caregiver	0.882 (0.019)	0.876 (0.015)	0.988 (0.005)	0.897 (0.008)	0.016 (0.012)^a	0.003 (0.003)^a	0.012 (0.003)^a
Father is caregiver	0.118 (0.019)	0.113 (0.014)	0.012 (0.005)^a	0.001 (0.000)^a	0.984 (0.012)	0.292 (0.046)	0.002 (0.001)^a
Grandmother is caregiver		0.006 (0.003)		0.077 (0.007)		0.273 (0.052)	0.590 (0.014)
Other female is caregiver		0.005 (0.002)		0.019 (0.003)		0.414 (0.056)	0.344 (0.014)
Other male is caregiver				0.006 (0.002)		0.017 (0.013)	0.053 (0.006)
	Non-African
	Nuclear	Parents & other adults	Mother (no other adults)	Mother & other adults	Father (no other adults)	Father & other adults	Other adults only
Proportion of children	0.407 (0.026)	0.224 (0.019)	0.075 (0.012)	0.174 (0.016)	0.009 (0.005)	0.025 (0.007)	0.073 (0.009)
Mother is dead					0.007 (0.008)	0.389 (0.146)	0.189 (0.045)
Father is dead			0.158 (0.062)	0.114 (0.024)			0.204 (0.048)
Mother is caregiver	0.891 (0.030)	0.841 (0.039)	1	0.916 (0.024)			0.015 (0.009)^a
Father is caregiver	0.109 (0.030)	0.103 (0.034)			1	0.743 (0.111)
Grandmother is caregiver		0.038 (0.016)		0.062 (0.021)		0.054 (0.029)	0.422 (0.065)
Other female is caregiver		0.018 (0.016)		0.010 (0.004)		0.204 (0.102)	0.370 (0.064)
Other male is caregiver				0.012 (0.012)			0.193 (0.076)

Source: NIDS 2008.

Note: The data are weighted. Standard errors are in parentheses. Proportions are calculated for children for whom there is non-missing information on the sex and age of their primary caregiver, and on the child’s relationship to their caregiver.

^aThese mothers (and fathers) were identified as non-resident household members.

Parental death is an important explanation for why children do not live with either, or both parents, and African children are more likely than other children to have deceased parents. As we would expect, children are the most likely to be maternal or paternal orphans when they live with neither parent. But even in this household type, the majority of children have parents who are reported as being alive (over 70% of African children, and over 80% of non-African children).

When children live with one parent, then this is typically their mother. In mother-only households (i.e. households that exclude fathers), there is again little difference in the profile of primary caregivers for African and non-African children: mothers provide primary care to 90% or more of children. However, a much larger share of African children lives in mother-only households (45%, compared with 25% of non-African children).

Less than 6% of children live with their father and not their mother. Where there are no other adults in the household, then fathers are the primary caregivers. However, most father-only households also include other adults. In these household types, most non-African children (74%) still receive primary care from their father; but fathers are the caregivers for only 29% of African children. Where fathers do not provide primary care, then this role is typically undertaken by women other than the child’s mother, with far more African children being cared for by their grandmother (27% compared with 5% of non-African children).

The provision of primary physical childcare in South Africa therefore remains highly gendered. When mothers live with their children, then they are mostly the child’s primary caregiver. A key reason explaining why African children are less likely than non-African children to receive primary care from their mother is that a considerably larger share lives without their mother (or their father) – 24% compared with 7% of non-African children. In these households, the child’s grandmother is most likely to be the primary caregiver, although this is more common among African children.

Only a very small share of children receives primary care from their father, but the share is even smaller among African children. This is mostly because African children are far less likely than other children to live with their father. In the next section, we explore whether the financial support of children is also gendered, and we consider particularly the contributions made by parents who do not live in the same household as their children.

5. The financial support of children

To describe the financial care of children, we first investigate who paid for, or contributed towards, the child’s education expenses in the year preceding the survey. Education expenses include fees, uniforms, books and stationery, transport to school and other school-related expenses, and they represent an important component of the direct costs of childcare, and of total household expenditure more generally. For example, in their analysis of the NIDS 2008 data, Branson et al. (2013) find that expenditure on education accounts for over 20% of total spending in poor households. In the analysis, we only consider children who are of school-going age (7–14 years) and who are enrolled in school (accounting for approximately 98% of children who are age-eligible for school).

Table 5 shows that for the majority of all children, the responsibility for schooling expenses falls on only one person, and this person is typically living in the child’s household. A larger share of ‘payers’ who are not co-resident are absent rather than non-resident household members. However, this is because children are more likely to have absent compared with non-resident parents. For example, over 40% of African children had a father who was absent from the child’s household, but less than 7% had a father who was a non-resident household member. For all children with absent parents, 13% received educational support from people who were absent from the household; while for all children with non-resident parents, 46% received support from non-resident household members.

Table 5. Person(s) paying for educational expenses of children aged 7–14, 2008.

	African children	Non-African children
Proportion of children enrolled	0.979 (0.003)	0.985 (0.006)
More than one person pays^a	0.083 (0.006)^c	0.251 (0.033)
Status of payer:
Not a household member: absent	0.072 (0.005)	0.059 (0.016)
Household member: not-resident	0.056 (0.004)^c	0.009 (0.003)
Household member: resident:	0.871 (0.006)^c	0.932 (0.017)
Sex of payer if household member:
Female^b	0.723 (0.011)^c	0.500 (0.034)
Male^b	0.277 (0.011)^c	0.500 (0.034)

Source: NIDS 2008.

Note: The data are weighted. Standard errors are in parentheses.

^aProportions are calculated conditional on the child attending school.

^bProportions are calculated only where payers were (resident or non-resident) members of the household, and aged 15 years or older.

^cProportions of African and non-African children are significantly different at the 95% confidence level.

The table also highlights a number of differences between African and non-African children. First, there is more sharing of costs for non-African children, 25% of whom have more than one person who contributes to their educational expenses, compared with 8% of African children. Second, where payers are household members, the responsibility for non-African children is evenly shared among men and women. Among African children, however, women are far more likely than men to be identified as paying for the child’s education expenses: approximately 72% of African children have payers who are female. Third, non-African children are less likely to be financially supported through school by someone who is not living in the child’s household (accounting for 7% of non-African children compared with 13% of African children).

The education expenses of the large majority of children are paid for by adults who are (resident or non-resident) members of the child’s household (93% of African children and 94% of non-African children). Table 6 shows that payers who are household members are most likely to be one, or both, of the child’s parents. However, non-African children are more likely to receive support from their parents compared with African children. Approximately 92% of the payers for non-African children are the child’s mother, father, or both, with fathers being more likely to pay (although not significantly so). In contrast, about 74% of the payers for African children are one, or both, of the child’s parents, with mothers being almost three times more likely to pay than fathers. Furthermore, both parents are more likely to contribute to the schooling expenses of non-African children (21%) compared with African children (7%). A greater share of African children therefore receives schooling support from someone who is not their parent. This is most likely to be the child’s grandmother (13% of payers) or another woman (8% of payers).

Table 6. Relationship of payers to children, 2008.

	African	Non-African
Mother	0.489 (0.013)^a	0.313 (0.032)
Father	0.185 (0.011)^a	0.398 (0.039)
Mother & father	0.068 (0.006)^a	0.207 (0.035)
Grandmother	0.134 (0.007)^a	0.018 (0.004)
Mother & grandmother	0.005 (0.002)	0
Father & grandmother	0	0
Other woman	0.075 (0.006)^a	0.039 (0.012)
Other man	0.030 (0.004)	0.015 (0.005)
Other combination of individuals	0.013 (0.003)	0.011 (0.004)

Source: NIDS 2008

Note: The data are weighted. Standard errors are in parentheses. The sample includes only those children where the payer is a member (resident or non-resident) of the child’s household, and the payer is 15 years or older.

^aProportions of African and non-African children are significantly different at the 95% confidence level.

Among all children aged 7–14, a smaller percentage receives financial support from someone not living in the household than the percentage without a co-resident parent. This difference reflects parents who do not live with their children and who are reported as not contributing towards their children’s schooling. However, given the fungibility of income, it is possible that income received from parents living elsewhere is not reflected as a payment towards schooling expenses specifically. We therefore analyse the third set of questions asked about children, concerning how much contact they have with parents who do not live in the child’s household, and whether financial support is received from these parents. These data are summarised in Table 7. Because very few non-African children are reported as having parents who are non-resident household members, the table describes only non-African children with absent parents.

Table 7. Contact with, and support by, absent and non-resident parents, 2008.

	African				Non-African
	Absent parents		Non-resident parents		Absent parents
	Mother	Father	Mother	Father	Mother	Father
Parent sees child:
Daily	0.043 (0.009)	0.054 (0.005)	0.011 (0.008)	0	0.199 (0.063)	0.141 (0.030)
Several times a week	0.141 (0.014)	0.124 (0.008)	0.098 (0.024)	0.172 (0.031)	0.189 (0.051)	0.113 (0.024)
Several times a month	0.394 (0.018)^a	0.246 (0.009)	0.552 (0.037)	0.512 (0.034)	0.276 (0.061)	0.168 (0.031)
Several times a year	0.342 (0.017)^a	0.262 (0.010)	0.317 (0.034)	0.310 (0.029)	0.169 (0.049)	0.204 (0.037)
Never	0.080 (0.009)^a	0.314 (0.010)	0.022 (0.010)	0.007 (0.004)	0.167 (0.060)^a	0.373 (0.040)
Proportion with parents who pay	0.496 (0.019)^a	0.377 (0.011)	0.733 (0.033)	0.831 (0.023)	0.449 (0.072)	0.460 (0.043)
Last payment received (Rands)	312.521(13.121)^a	483.869 (44.176)	324.465 (32.009)^a	641.247 (57.831)	438. 651 (106.482)	589.742 (75.920)

Source: NIDS 2008.

Note: The data are weighted. Standard errors are in parentheses. The residency status of parents is determined by information collected in the household roster. The sample includes all children whose mother or father is not co-resident with the children. Two children who were reported as receiving R130 000 as their last maintenance payment from their father were dropped from the sample.

^aDenotes significant differences between children’s mothers and fathers within each of the three categories that indicate absent or non-resident parents.

The majority of children with a parent living elsewhere see this parent at least several times a year. However, parents who are not resident but still considered to be household members (non-resident parents), are far more likely to have some contact with their children during the year, than parents who are not part of the household (absent parents). There are also no significant differences in the extent of contact by the sex of the non-resident parent. Among children with absent parents, in contrast, children are significantly less likely to have contact with an absent father than an absent mother: over 30% of both African and non-African children with absent fathers did not see their father during the year; while where mothers are absent, the percentages are 8% and 17% of African and non-African children respectively.

Children with an absent parent are also substantially less likely than children with a non-resident parent to receive any financial support from the parent. For example, whereas 38% of African children with an absent father receive financial support from their father, this is the case for 83% of African children with a non-resident father. Among African children (but not among non-African children), absent fathers are also significantly less likely than absent mothers to provide financial support: almost 50% of African children with an absent mother received income from their mother. Conditional on a financial contribution, fathers contribute more income than mothers, and these differences are significant among Africans. However, we cannot establish what share of the payer’s income these contributions represent, and gender differences in payments mirror a gender gap in earnings (see for example Posel, 2014).

6. Summary and discussion

The objective of this study was to investigate the gender division of labour in childcare in South Africa, not only in the physical care of children but also in their financial support. Family formation is distinctive in South Africa because the majority of children, and specifically African children, do not live with both their parents. Although it would be very difficult for parents who live separately from their children to be their primary caregivers, the provision of financial support is not similarly constrained by physical distance. Part of the investigation therefore explored whether parents who live apart from their children contribute financially to their upkeep.

We distinguished African children from other children because of the particular features of household formation among Africans (and small sample sizes make it difficult to distinguish among non-African children). African children are far less likely than other children to live with both parents, and more likely to live with only one parent (almost always their mother), or with neither parent. Temporary labour migration, which was enforced under apartheid, together with low marital rates, help to explain these living arrangements. Despite the removal of apartheid laws, temporary migration persists today at least partly because of high rates of unemployment as well as the lack of formal and low-cost housing in urban areas (Posel & Marx, 2013). High rates of parental mortality also contribute to parental absence from the child’s household. These differences in household formation help to explain why, although there are common features in the provision of childcare across all children, there are also differences in who provides physical and financial care to African and non-African children.

The nationally representative household survey data which we analysed show that the provision of primary childcare in South Africa is highly gendered. Although children typically receive physical care from more than one person, women are identified as the primary caregivers (or the main person responsible for providing care) to the large majority of children. Men are more likely to be the primary caregivers to non-African than African children, but still, only a small minority of children (at most 13% of non-African children and 8% of African children) receives primary physical care from a man.

Children who live with their mother are typically cared for by their mother, but mother–child co-residency is far lower among African children than other children. Children who live apart from their mothers mostly still receive primary care from women, and particularly from grandmothers or other co-resident female household members (such as the child’s aunt or older sister), highlighting both the importance, and the gendered nature, of extended family networks in the physical care of children.

In contrast, the responsibility for the financial support of children, at least in terms of schooling expenses, is not borne disproportionately by men. Children’s schooling expenses are typically paid for by only one person, but this is far more often the case for African than non-African children. Men are considerably more likely to pay for children’s schooling than they are to provide primary care; but even among non-African children, the majority of whom live with their father, fathers are only slightly more likely than mothers to assume primary responsibility for the costs of schooling. Among African children, a far larger share is supported financially through school by women, and not men. Where financial support for schooling comes from a household member, then this is almost three times more likely to be the child’s mother than father.

In recent years, there have been a number of studies which have probed the contribution that fathers make to the care of their children, and particularly fathers who are not co-resident in the child’s household. These studies, which are based often on qualitative research or regionally specific micro-data, challenge the assumption that fathers who are absent from the child’s household are also absent from the child’s life, and they call for more recognition of fathers’ involvement in the care of children (Clark et al., 2015; Madhavan et al., 2016).

The NIDS data which we analysed are nationally representative, and an important advantage of these data is that we can distinguish among fathers (and mothers) according to whether they are viewed as non-resident household members or as absent from the child’s household altogether. We found significant differences in the involvement of fathers among these two groups. The majority of fathers who are non-resident household members see their children and contribute financially towards their upkeep. Moreover, non-resident fathers are more likely than non-resident mothers (although not significantly) to provide support.

However, a far greater share of children has fathers who are classified as absent from the household, rather than as non-resident household members, and the contribution of absent (living) fathers is considerably smaller. Among African children, there are also clear differences in contact and support from absent mothers and absent fathers: absent mothers are significantly more likely than absent fathers to see their children regularly and to provide financial support.

It is possible that men are carers of children in ways that are not reflected in the data which we analysed. Studies from the USA, for example, have found that mothers’ reports of fathers’ childcare contributions are typically lower than fathers’ self-reports, and particularly when fathers are not resident in the household (Cooley & Morris, 2002). However, the possibility for respondents to list multiple individuals who assist, with both physical childcare and the payment of school fees, as well as the inclusion of a separate module on the contributions of parents living elsewhere, should reduce the likelihood of this under-reporting.

Overall our findings suggest that, rather than a gender division of labour in the provision of care particularly to African children, both the primary physical care and the financial support of children are most often provided by women. This in turn has significant implications for the economic wellbeing of women and the children they support – for women continue to earn significantly less than men, and this is at least partly because women’s childcare responsibilities limit the nature and extent of their labour force participation (Aguero & Marks, 2011). Children are far more likely to live in poverty if they live in households with their mother and not their father (Posel & Rudwick, 2013). Policy initiatives that seek to reduce high rates of child poverty in South Africa therefore need to recognise the consequences for both women and children of women’s disproportionate responsibility for the care of children, and they need to consider ways in which the contributions of fathers can be encouraged (through the provision of paternity leave and social education programmes for example) or ‘enforced’ (through an effective maintenance system).

Disclosure statement

No potential conflict of interest was reported by the authors .

Additional information

Funding

This work was partly supported by funding received from the National Research Foundation. Any opinion, finding and conclusion or recommendation expressed in this material is that of the authors and the NRF does not accept any liability in this regard.

Notes

1 An earlier version of this paper was presented at the 24th IAFFE conference (Posel & Hatch 2015).

2 In our analysis, we retain the racial classification used by the official statistical agency (Statistics South Africa) and adopted in the Population Censuses and a wide range of national household surveys: Black African, Indian/Asian, Coloured and white. Specifically, we distinguish between ‘Africans’, representing ‘Black Africans’, and ‘non-Africans’.

3 This is not to suggest that non-African children are an homogeneous group.

4 Among all African children, 99.9% of primary caregivers are adults (15 years and older), while all primary caregivers of non-African children are adults. Three observations for African children identify caregivers who are themselves younger than 10 years, and these observations have been dropped from the sample.

5 Comprehensive information on ‘non-biological’ parents is not collected in the survey. Where they can be identified, we have classified foster, adoptive and stepmothers as ‘other female’; and foster, adoptive and stepfathers as ‘other male’.

6 For example, for approximately 25% of non-African children, the third person identified as providing childcare assistance was someone who was ‘unrelated and for pay’.

7 In identifying household type, we consider only resident household members (members who spend at least four nights a week in the household).