Impact of Households’ Income Sources on Child’s Multidimensional Poverty in South Africa

Abstract

Poverty is extensively acknowledged as a multidimensional phenomenon and has been the subject of extensive research all over the world. When it comes to child poverty, the impacts differ significantly due to the critical stages of physical, cognitive and socioemotional growth which make them vulnerable to the negative impacts of poverty. This paper analysed the impact of households’ income sources on child’s (0–17 years) fuzzy multidimensional poverty indicators in South Africa. The South African General Household Survey (GHS) datasets for the years 2017, 2018 and 2019 were used. The data were collected using the stratified two-stage sampling method with sampled households comprising 25,915, 25,224, and 20,083 children in 2017, 2018 and 2019, respectively. The fuzzy set methodology was used to compute child’s multidimensional poverty indicator (MPI), which was later analysed with the Tobit regression model and treatment effects potential outcome framework using regression adjustment estimator. The results of the Tobit regression indicate that geography type, residence in Eastern Cape, North-West and Limpopo provinces, parental presence and income sources significantly (p < 0.01) influenced child’s fuzzy multidimensional poverty. Additionally, children identified as Coloured, Indian/Asian and White and those who were biological children of household heads had significantly lower (p < 0.01) poverty. With respect to the Average Treatment Effect (ATE) and the Average Treatment Effect on the Treated (ATET), the results showed that children from households that were receiving social grants, salaries, income from business, remittances, pensions and other sources of income had significantly lower (p < 0.01) fuzzy MPI, when compared to the control group. In contrast, children from households that were receiving once off grants had significantly higher fuzzy MPI significantly (p < 0.01) higher compared to the control group. It was concluded that some income sources possess MPI reducing tendency. Therefore, development programmes to enhance households’ access to them are recommended.

Background

Child poverty in South Africa remains an important development challenge, highlighting the urgency for targeted and sustainable policy interventions [35]. This challenge cannot be overlooked due to the significant and lasting adverse effects it poses on children’s physical health, cognitive development, educational achievement and their overall wellbeing [77]. The current poverty narratives and prevalence in South Africa can be closely linked to the legacy of apartheid that systematically denied the majority of Black South Africans access to quality education, adequate healthcare services and economic opportunities [34, 40, 56].

Since inception of a democratically elected government in 1994, different policy options had been utilized to promote households’ incomes [10]. These include programmes to facilitate employment equity, income stability and sustainability, such as the Reconstruction and Development Programme (RDP), Growth, Employment and Redistribution (GEAR), Skills Development framework, Expanded Public Works Programme, National Development Plan, Employment Tax Incentive and Youth Employment Service, and Presidential Employment Stimulus [51, 58]. Although these policies have expanded access to employment opportunities, over the years, many households are still relying on income sources that are characterized by low sustainability, insufficiency and seasonality [50].

In addition, the need to drastically reduce poverty among extremely vulnerable population in South Africa brought some social protection interventions for supporting children, old and disabled people, and war veterans, among others [71]. The 1997 White Paper for Social Welfare contains the policy framework that established social grants [81]. Moreover, some legislative emphases on the need to promote child’s and old people’s welfare had been addressed by the Children’s Act (2005) and Older Persons Act (2006), respectively [3]. Over the years, there have been revision of the amounts paid for all these grants to cater for inflationary pressures. Specifically, in April 2009, the approved amounts were R1010 for old age and disability, R1030 for war veterans, R240 for grant-in-aid and child support [57]. Government had recently increased these grants to R2320 for old age/disability, R2340 for war veterans and R560 for child support [67].

Despite these policy interventions, significant multidimensional poverty, which evaluates the levels of deprivation in different welfare indicators such as health, education, and living standards, prevails among South African households [63]. The conventional logic that poverty begets poverty also underscores a notion of child’s poverty increase. Specifically, some basic welfare indicators for South African under-5 children reveal that 29% were stunted, 8% underweight, 5% wasted, and about 13% overweight or obese. Also, school dropout rate was 25% and 23% drink unsafe water, all of which contribute to multidimensional child poverty [24]. According to Reyneke [76], children who experience poor health often face academic performance challenges and are at increased risk of dropping out of school. This reduces their opportunities for skill development and threatens their future employment opportunities [79]. Moreover, these challenges contribute to intergenerational poverty, where poor children will likely become poor adults [8, 49, 69].

Poverty analyses have conventionally followed the unidimensional and multidimensional approaches [20]. The former is known as the income- or expenditure-based approach which concentrates adequacy of household income or consumption expenditure to provide a level of welfare given defined poverty lines, while the latter is the multidimensional or capability-based approach that targets a broader welfare deprivation such as health, happiness, education, and living standards [47, 62, 78, 89]. Although poverty analyses in South Africa have traditionally used the two approaches, majority of existing studies are based on households’ consumption and income [4, 25, 31, 42, 88, 92]. While the income approach gives a useful starting point for understanding poverty trends in South Africa, it offers a limited and narrow view.

Poverty transcends income, to include multiple and interrelated dimensions such as access to quality education, healthcare, adequate housing, sanitation, clean water, social inclusion, among others. Some studies have assessed child poverty through multidimensional lenses with South African datasets. Among these are Proudlock [73] who found that 18% of children resided in overcrowded households, 21% were located over 30 min from a healthcare facility and 30% resided in households with no employed adults. Similarly, Omotoso and Koch , [64] reported that over 38% of children lived in inadequate housing. Using Alkire-Foster approach, they also found a decline in the multidimensional headcount poverty ratio from 32.8% in 2002 to 22.2% in 2004, and a drop in the intensity of child poverty from 48.9% to 40.7% over the same period. Also, some studies have explored the impacts of social grants on poverty. Mtshali, [54] and Pasha [70] analysed the impacts of social grants on multidimensional poverty and found that social grants reduced multidimensional poverty. Using the income/expenditure approach, Satumba et al., [80] found that social grants had significant poverty reducing impacts. Other researchers with similar findings are Mackett [44] and Satumba et al., [80] while Vaaltein and Schiller [93] reported a contrary finding.

Despite the growing body of knowledge on poverty research in South Africa, there are several gaps that this study seeks to fill. Precisely, most of the studies rely on monetary approach that fails to capture the multidimensional nature of poverty, especially among children. Also, while some studies have explored the impacts of social grants, little is known on the impacts of other income sources. Therefore, this study is making a significant addition to the literature by analysing the impacts of income sources on child’s multidimensional poverty using the potential outcome framework to estimate the Average Treatment Effect (ATE) and Average Treatment Effect on the Treated (ATET). It hypothesizes that income sources do not significantly influence child’s multidimensional poverty. This is a step in the right direction for effectively addressing child poverty and achieving the Sustainable Development Goals 1 and 2. The findings are going to be essential for developing income-responsive and comprehensive policy contexts to understand and address child’s poverty in South Africa. This study can also provide policy benchmarking which can help in designing focused responses for child poverty alleviation as an integral component of the South Africa’s National Development Plan (2030).

Methodology

The Data

The datasets were obtained from the South African General Household Survey (GHS) for the years 2017, 2018, and 2019, which were collected by Statistics South Africa (SSA) to provide insights into the country’s economic development and challenges. The Statistics South Africa’s GHS used a Master Sample (MS) structure established in 2013 as a comprehensive framework for all household-based surveys that are conducted. This framework’s principal sampling units (PSUs) comprise of detailed sections within detailed PSUs, considering households’ geographic locations (urban, rural, or traditional) [90–92]. The sample design involved is a stratified two-stage sampling process for households. After the data was sorted and merged, the 2017–2019 datasets included 25,915, 25,224, and 20,083 South African children within 0–17 years, respectively [90–92]. These constituted the samples for this study.

Computational Framework for Child’s Multidimensional PovertyCom

Although multidimensional poverty indicators are often computed with three welfare dimensions which are standard of living, health and education, this study integrates the subjective notion of poverty. Conceptually, beyond the physical deprivation of material needs, poverty can be evaluated from the psychological angle that evaluates depressive moods and welfare perception. The adopted approach is in alignment with some previous studies such as Strotmann and Volkert [89], Nozaki and Oshio [62] and Mateu et al., [47].

The fuzzy set approach has been used by several authors. These include Bastos and Machado [5] who used the approach in Lisbon and found out that children who were in large families, Gipsy and Black and children without parental figure had increased risks of being deprived. In 2012, Gallant studied multidimensional poverty in South Africa since the transition and found out that people from female-headed households, Popogbe et al., [72] also used the same approach and found out that most of the households were deprived in the education dimension, followed by standard of living and health dimensions. Limpopo, Eastern Cape and KwaZulu-Natal provinces had higher probabilities of being poor under the fuzzy set methodology. In addition, Leu et al., [39] revealed that multidimensional poverty was mainly affected by housing conditions, medical care, social exclusion and education through the fuzzy set approach in Taiwan. The first step in computing the child’s MPI involved selection of relevant welfare dimensions, their indicators and cut-off points and associated weights. Table 1 outlines the selected dimensions and corresponding attributes of child’s multidimensional poverty, identifying four key areas: standard of living, health, education and perceived happiness. A total of ninety-one (91) welfare attributes were recorded, with the majority, seventy-six (76) falling under the standard of living category. Health accounted for eleven (11) attributes, while education and perceived happiness each two (2). The standard of living is further subdivided into six (6) indicators, whereas the health dimension was categorised into two (2) indicators.

Table 1.

Child MPI dimensions, indicators, deprivation cut-offs and weights

Dimension (Weight)	Indicator (Weight)	Deprived when…
Standard of living (76/91)	Assets (22/91)	Lacking the following: Ownership of a motor vehicle; ; Radio ownership TV Set; Swimming Pool; DVD Player/Blu-ray Player Pay TV (M-Net/DSTV/Top TV) Subscription; Air Conditioner (Excluding Fans); Computer/ Desktop/ Laptop; Vacuum Cleaner/Floor Polisher; Dish washing machine; Washing Machine; Tumble Dryer Deep Freezer – free standing; Refrigerator or Combined Fridge Freezer; Electric Stove; Microwave Oven; Built-in Kitchen sink; Home Security Service Home Theatre System; Geyser providing hot running water; Solar hot water geyser; Solar electrical panel
	Telecommunications (7/91)	Lacking the following: Internet connection in the household; Internet in a library or community hall/Thusong centre; Internet for students at a school/university/college; At place of work; Internet Café ≤2 km from the household; Internet Café and Telephone.
	Waste removal (6/91)	A child resides in a household that there is irregular or no waste removal, littering, water pollution, outdoor/indoor air pollution, land degradation and excessive noise.
	Housing characteristics (10/91)	A child resides in an informal/traditional dwelling type (shack, caravan or other), without bricks or cement for walls and without materials such as tile, corrugated iron, asbestos, and others for roof; with unimproved floor materials, more than two persons per room; using unimproved drinking water; distance of water source from the dwelling more than 30 min, using unimproved toilet facilities, and sharing toilet facilities.
	Safety (10/91)	Responded yes to the following: Motor vehicle injury – occupant; Bicycle related; Gun shots wounds; Severe trauma due to violence, assault, beating; Crime- related injury; Fire or burn; Accidental poisoning; Intentional poisoning; Sports related injuries; and other injuries.
	Energy (21/91)	A child resides in a household with the following features: lack access to electricity; uses paraffin for lighting; uses candles for lighting; uses no energy source for lighting; uses other unclean sources for cooking; uses paraffin for cooking; uses wood for cooking; uses coal for cooking; uses animal dung for cooking
Health (11/91	Nutrition/hunger (8/91)	Responded yes to the following: run out of money to buy food; run out of money 5 or more days in the past 30 days; cut the size of meal or skip any meals; cut size of meals 5 or more days in the past 30 days; Skipped meals; skipped meals 5 or more days in the past 30 days; smaller variety of food; smaller meals 5 or more days in the past 30 days
	Healthcare Facilities and Health Status (3/91)	A child is not covered under any medical aid scheme; A child resides in a place that takes more than 30 min to reach the health facility; A child has a fair or poor health status
Education (2/91)	Early Childhood Development (ECD) (2/91)	A child of 0–59 months is not attending any ECD centre; A child who is old enough to attend school (6–18 years) does not attend and school/education institution
Perceived happiness (2/91)	Perceived Happiness (2/91)	A child resides in a household where the head says they are poor; A child resides in a household where the head is not happy or same as before.

Source: Own Computation

The fuzzy sets allocate weights to attributes based on the overall severity of deprivations experienced by children. This process is essential for identifying children classified as poor and for aggregation of data into single composite poverty index. Notably, the fuzzy set approach applied in the study utilised the same attributes and established the threshold values for each indicator by assigning value of 1 to children identified as deprived and 0 to those considered non-deprived.

Child’s Welfare Indicator Computation Using the Fuzzy Set

This study adopted the fuzzy set approach that was proposed by Dagum and Costa [14]. Fuzzy set is an effective and rigorous method for aggregating binary or ordinal variables. In the context of poverty analysis, fuzzy set is advantageous because it constructs an indicator from several welfare attributes without the need for integrating arbitrary weights and defining poverty lines [28, 59]. The fuzzy set methodology is characterized by a class of continuous membership ranking. However, instead of using , the 0 and 1 coding method was used. The multidimensional poverty ratio , which highlights the level of welfare deprivation is defined as the weighted average of ,

1

The weights attached to each  jth characteristic, denoted as , play a crucial role in measuring the intensity of deprivation with respect to . The weight is an inverse function of the degree of deprivation; hence, the smaller the number of households and the lesser their extent of deprivation, the larger the assigned weight . This methodology safeguards that the  jth characteristic with fewer deprived households receive higher importance in the multidimensional poverty measurement. In practice, the methodology for assigning the weights in fuzzy sets follows the proposition by Cerioli and Zani [11], which addresses the justification of material goods in the context of poverty measurement. This can be mathematically expressed as follows:

2

First, and is the relative frequency characterized by the sample observation in the total population. Therefore when , the welfare characteristic is to be removed.

The poverty ratio of the population is obtained as a weighted average of the poverty ratio of the  ith household .

Number this equation.

Tobit Regression Model

The children’s fuzzy MPI (multidimensional poverty index) was analysed with Tobit regression. The model is specified in the equation below:

3

= 0 if y ≤ 0, y = Y* if y > 0.

Y* = Multidimensional poverty.

βs = estimated parameter or coefficient.

ei = error term and is normally distributed with zero mean and constant variance.

Table 2 presents the coding format used in this study. this is to ensure consistency in data entry and analysis.

Table 2.

Coding format on explanatory variables

Explanatory Variable	Coding format
Functioning Impairments of Household Member(s)
Seeing problem	(yes = 1, otherwise = 0)
Hearing	(yes = 1, otherwise = 0)
Walking	(yes = 1, otherwise = 0)
Remembering	(yes = 1, otherwise = 0)
Selfcare	(yes = 1, otherwise = 0)
Communication	(yes = 1, otherwise = 0)
Provinces
Eastern Cape (EC)	(yes = 1, otherwise = 0)
Northern Cape (NC)	(yes = 1, otherwise = 0)
Free State (FS)	(yes = 1, otherwise = 0)
KwaZulu Natal (KZN)	(yes = 1, otherwise = 0)
North-West (NW)	(yes = 1, otherwise = 0)
Gauteng (GP)	(yes = 1, otherwise = 0)
Mpumalanga (MP)	(yes = 1, otherwise = 0)
Limpopo (LP)	(yes = 1, otherwise = 0)
Child’s and Households’ Demographic Characteristics
Child gender	(male = 1, female = 0)
Child age	(years)
Coloured	(yes = 1, otherwise = 0)
Indian/Asian	(yes = 1, otherwise = 0)
White	(yes = 1, otherwise = 0)
Son or daughter of household’s head	(yes = 1, otherwise = 0)
Father alive	(yes = 1, otherwise = 0)
Father part of the household	(yes = 1, otherwise = 0)
Mother alive	(yes = 1, otherwise = 0)
Mother part of the household	(yes = 1, otherwise = 0)
Domestic or household workers services	(yes = 1, otherwise = 0)
Household size	Number
Traditional	(yes = 1, otherwise = 0)
Farms	(yes = 1, otherwise = 0)
Income sources
Salaries/wages commission	(yes = 1, otherwise = 0)
Income from a business	(yes = 1, otherwise = 0)
Remittances	(yes = 1, otherwise = 0)
Pensions	(yes = 1, otherwise = 0)
Once off grants	(yes = 1, otherwise = 0)
Social grants	(yes = 1, otherwise = 0)
Sales of farming products and service	(yes = 1, otherwise = 0)
Other income sources	(yes = 1, otherwise = 0)
Gardening Activities
Backyard garden	(yes = 1, otherwise = 0)
School garden	(yes = 1, otherwise = 0)
Communal garden	(yes = 1, otherwise = 0)

It should be noted that although income sources may be endogenous in the proposed model, utilization of standard instrumental-variable Tobit estimators in Stata requires continuous endogenous regressors [86]. Therefore, the model could not be assessed for endogeneity given the binary nature of the income-sources. Consequently, we interpret our estimates as conditional associations.

Treatment Effects Computation

The impacts of income sources on child’s MPI were estimated using the potential outcome framework. The data contained eight sources of household’s income which are salaries/wages business, remittances, pensions, grants, social grants, sales of farming products and services, and other incomes. These variables were coded 1 for households who responded yes to the different income sources and 0 otherwise. The Average Treatment Effect (ATE) and the Average Treatment Effect on the Treated (ATET) were estimated to estimate the impacts of those income sources. The ATE shows the difference between the expected outcomes for the treated and control groups while the ATET shows the difference between the average outcomes for the treated when they were treated, and their counterfactuals .

Results and Discussion

Child’s Demographic Dynamics

The results presented in Table 3 revealed that male children constituted 50.6%. Also, 54.3% of the households were male headed. The results further showed that in the combined data, 12.4% of the households’ heads were between 30 and 35 years old, while 8.6% were less than 15 years. In terms of population group, Black/African children had the highest proportion (87.8%). This was followed by 8.3% for Coloured children, 2.9% for White children and 1.0% for Indian/Asian children. The results also showed that majority of the children had their fathers being alive across all the year, although most of them were not residing with their children. Also, 93.3% of the children had living mothers while 70.3% of these mothers resided with their children. In terms of the areas of residence, the results showed that majority of the sampled children (54.5%) resided in urban areas, while 42.1% resided in traditional areas. The Table also revealed that Child Support Grant (CSG) was accessed by 80.7%. The results also revealed that 72.0% of the sampled children did not receive social grants. On the other hand, 28.0% of the sampled children were recipients of social grants. In terms of gardening activities 79.7% of the children were from households with backyard gardens.

Table 3.

Child’s selected demographic characteristics

Characteristics		2017		2018		2019		All
		Freq.	%	Freq.	%	Freq.	%	Freq.	%
Province	Western Cape	1990	7.7	1989	7.9	1450	7.0	5429	7.6
	Eastern Cape	3900	15.0	3689	14.6	3050	14.8	10,639	14.8
	Northern Cape	1211	4.7	1207	4.8	1016	4.9	3434	4.8
	Free State	1566	6.0	1510	6.0	1246	6.1	4322	6.0
	KwaZulu Natal	4944	19.1	4898	19.4	4141	20.1	13,983	19.5
	North-West	1615	6.2	1519	6.0	1368	6.6	4502	6.3
	Gauteng	4883	18.8	4674	18.5	4052	19.7	13,609	19.0
	Mpumalanga	2400	9.3	2346	9.3	1985	9.6	6731	9.4
	Limpopo	3406	13.1	3392	13.4	2264	11.0	9062	12.6
Child gender	Male	13,049	50.4	12,643	50.1	10,570	51.4	36,262	50.6
	Female	12,866	49.6	12,581	49.9	10,002	48.6	35,449	49.4
Household Head gender	Male	14,100	54.4	13,843	54.9	10,972	53.3	39,404	54.3
	Female	11,815	45.6	11,381	45.1	20,572	44.3	32,307	45.1
	Total	25,915	100.0	25,224	100.0	20,572	100.0	71,711	100.0
Household head Age group	< 15	2702	10.4	2517	10.0	981	4.8	6200	8.6
	15 < 20	2916	11.3	2737	10.9	2234	10.9	7887	11.0
	20 < 25	2978	11.5	2961	11.7	2472	12.0	8411	11.7
	25 < 30	3041	11.7	2924	11.6	2499	12.1	8464	11.8
	30 < 35	3173	12.2	3109	12.3	2608	12.7	8890	12.4
	35 < 40	3000	11.6	2992	11.9	2659	12.9	8651	12.1
	40 < 45	2878	11.1	2942	11.7	2523	12.3	8343	11.6
	45 < 50	2465	9.5	2545	10.1	2412	11.7	7422	10.3
	50 < 55	2762	10.7	2497	9.9	2184	10.6	7443	10.4
	Total	25,915	100	25,224	100	20,572	100	71,711	100
Population group	Black/African	22,702	87.6	22,212	88.1	18,084	87.9	62,998	87.8
	Coloured	2204	8.5	2085	8.3	1650	8.0	5939	8.3
	Indian/Asian	253	1.0	250	1.0	225	1.1	728	1.0
	White	756	2.9	677	2.7	613	3.0	2046	2.9
	Total	25,915	100	25,224	100	20,572	100	71,711	100
Father alive	No	2952	11.4	2748	10.9	2794	13.6	8494	11.8
	Yes	22,488	86.8	21,985	87.2	17,341	84.3	61,814	86.2
	Do not know	475	1.8	491	1.9	437	2.1	1403	2.0
	Total	25,915	100	25,224	100	20,572	100	71,711	100
Father part of the household	No	17,746	68.5	17,489	69.3	14,166	68.9	49,401	68.9
	Yes	8169	31.5	7735	30.7	6406	31.1	22,310	31.1
	Total	25,915	100.0	25,224	100.0	20,572	100.0	71,711	100.0
Mother alive	No	1578	6.1	1442	5.7	1527	7.4	4547	6.3
	Yes	24,251	93.6	23,684	93.9	18,986	92.3	66,921	93.3
	Do not know	86	0.3	98	0.4	59	0.3	243	0.3
Total		25,915	100.0	25,224	100.0	20,572	100.0	71,711	100.0
Mother part of the household	No	7543	29.1	7306	29.0	6441	31.3	21,290	29.7
	Yes	18,372	70.9	17,918	71.0	14,131	68.7	50,421	70.3
	Total	25,915	100.0	25,224	100	20,572	100.0	71,711	100.0
Household size group	< 2	855	3.3	823	3.3	724	3.5	2402	3.3
	2 < 4	7914	30.5	7843	31.1	6222	30.2	21,979	30.6
	4 < 6	8506	32.8	8226	32.6	6656	32.4	23,388	32.6
	6 < 9	4681	18.1	4228	16.8	3689	17.9	12,598	17.6
	9 < 11	2571	9.9	2721	10.8	1740	8.5	7032	9.8
	11 < 13	663	2.6	687	2.7	858	4.2	2208	3.1
	13 < 15	410	1.6	411	1.6	408	2.0	1229	1.7
	15 < 17	173	0.7	132	0.5	161	0.8	466	0.6
	17 < 19	90	0.3	58	0.2	67	0.3	215	0.3
	≥ 19	52	0.2	95	0.4	47	0.2	194	0.3
	Total	25,915	100.0	25,224	100.0	20,572	100.0	71,711	100.0
Geographical type	Urban	14,019	54.1	13,607	53.9	11,477	55.8	39,103	54.5
	Traditional	10,992	42.4	10,753	42.6	8454	41.1	30,199	42.1
	Farms	904	3.5	864	3.4	641	3.1	2409	3.4
	Total	25,915	100.0	25,224	100.0	20,572	100.0	71,711	100.0
Grants	No	4777	18.4	4383	17.4	4656	22.6	13,816	19.3
	Yes	21,138	81.6	20,841	82.6	15,916	77.4	57,895	80.7
	Total	25,915	100.0	25,224	100.0	20,572	100.0	71,711	100.0
Social grants	No	18,542	71.5	18,317	72.6	14,797	71.9	51,656	72.0
	Yes	7373	28.5	6907	27.4	5775	28.1	20,055	28.0
	Total	25,915	100.0	25,224	100.0	20,572	100.0	71,711	100.0
Backyard garden	No	20,467	79.0	20,096	79.7	16,564	80.5	57,127	79.7
	Yes	5448	21.0	5128	20.3	4008	19.5	14,584	20.3
	Total	25,915	100.0	25,224	100.0	20,572	100.0	71,711	100.0
School garden	No	25,890	99.9	25,198	99.9	20,538	99.8	71,626	99.9
	Yes	25	0.1	26	0.1	34	0.2	85	0.1
	Total	25,915	100.0	25,224	100.0	20,572	100.0	71,711	100.0
Communal garden	No	25,845	99.7	25,130	99.6	20,478	99.5	71,453	99.6
	Yes	70	0.3	94	0.4	94	0.5	258	0.4
	Total	25,915	100.0	25,224	100.0	20,572	100.0	71,711	100.0

Distribution of Income Sources

Table 4 presents the distribution of income sources among the children’s households, while Tables 5 and 6 present the income sources across population and geography groups. Table 4 shows that once-off grants and salaries were the most received income sources. These were followed by social grants and remittances. Specifically, Table 5 reveals that the Black/Africans least received salaries, while the Coloured received the least of business incomes. The highest proportions of the Whites received wage, business and social grant incomes. However, remittances were mostly received by the Black/Africans. Table 6 further reveals that urban households had the highest access to salaries business and social grant incomes, while traditional households had the highest access to pension, remittances and once-off grants. These results vividly reflect the deprivations that were suffered by many Black/African in South Africa since the apartheid era [18, 61]. The distributional pattern of incomes in South Africa has long favoured the Whites, who have benefited from several government grants and other welfare packages [43, 82].

Table 4.

Distribution of children’s households’ income sources

Income Source	2017 (%)	2018 (%)	2019 (%)	Overall (%)
Salaries, wages, and commission	59.04	58.57	55.97	57.99
Income from a business	13.74	13.07	15.14	13.91
Remittances	20.47	20.77	16.33	19.39
Pensions	2.67	3.01	2.27	2.68
Sales of farming products and services	2.42	2.53	2.19	2.39
Social grants	28.45	27.38	28.07	27.97
Once-off grants	81.57	82.62	77.37	80.73
Other income sources	1.84	1.52	1.68	1.68

Table 5.

Distribution of income sources across households’ population groups

Salaries	2017 (%)	2018 (%)	2019 (%)	Combined (%)
Black/Africans	55.43	55.48	56.11	55.64
Coloured	83.39	79.94	55.80	74.52
Indian/Asian	83.92	76.68	48.32	70.11
White	88.66	88.12	57.94	79.37
Business
Black Africans	13.71	12.96	15.13	13.84
Coloured	7.06	7.60	14.86	9.41
Indian/Asian	27.06	23.72	17.23	22.79
White	29.20	29.03	15.55	25.10
Remittances
Black Africans	22.08	22.11	16.35	20.48
Coloured	9.96	12.27	14.30	11.98
Indian/Asian	5.10	7.91	17.23	9.92
White	7.56	7.48	18.00	10.63
Pension
Black Africans	2.57	2.90	2.33	2.62
Coloured	2.72	3.55	1.48	2.67
Indian/Asian	5.10	1.98	1.26	2.82
White	4.69	5.57	2.45	4.32
Once off grants
Black Africans	85.16	86.05	77.46	83.31
Coloured	75.37	75.80	76.57	75.86
Indian/Asian	29.80	29.25	77.73	44.91
White	9.65	11.29	76.76	30.10
Social grants
Black Africans	24.63	23.77	24.09	24.17
Coloured	38.93	37.29	39.33	38.47
Indian/Asian	81.96	84.98	79.83	82.31
White	94.13	93.84	94.44	94.13
Sale of farming products
Black Africans	2.61	2.73	2.21	2.54
Coloured	0.37	0.63	2.40	1.03
Indian/Asian	0.00	0.00	1.26	0.40
White	3.52	2.64	1.80	2.72
Other income sources
Black Africans	1.75	1.41	1.60	1.59
Coloured	2.49	2.19	1.97	2.24
Indian/Asian	0.78	0.79	3.36	1.61
White	2.87	3.37	2.13	2.82

Table 6.

Distribution of income sources across households’ geography types

Salary	2017 (%)	2018 (%)	2019 (%)	Combined (%)
Urban	73.23	71.70	55.56	67.48
Traditional	40.44	41.29	56.29	45.22
Farm	64.93	68.20	59.28	64.61
Business
Urban	15.01	14.06	15.15	14.73
Traditional	12.43	12.13	15.16	13.09
Farm	10.07	9.03	14.66	10.92
Remittances
Urban	13.91	14.63	16.32	14.88
Traditional	28.71	28.97	16.35	25.34
Farm	21.79	15.40	16.38	18.05
Pension
Urban	2.74	2.73	2.25	2.59
Traditional	2.63	3.50	2.30	2.85
Farm	2.10	1.27	2.18	1.83
Once off grant
Urban	72.00	73.30	77.29	74.01
Traditional	93.43	94.29	77.39	89.31
Farm	84.29	84.38	78.47	82.74
Social grants
Urban	38.37	36.98	37.21	37.58
Traditional	16.25	15.57	15.94	15.93
Farm	22.68	22.80	24.34	23.14
Sale of Agric products
Urban	0.56	0.67	2.10	1.05
Traditional	4.37	4.43	2.24	3.79
Farm	7.64	8.33	3.12	6.68
Other incomes
Urban	2.85	2.56	1.73	2.42
Traditional	0.65	0.31	1.66	0.81
Farm	0.66	0.35	1.09	0.66

Distribution of Child’s Fuzzy MPI Across Children’s Demographic and Household Characteristics

Distribution of Child’s Fuzzy MPI Across Children’s Provinces

The results presented in Fig. 1 show the distribution of average child fuzzy MPI across different provinces in South Africa. The results showed that the Eastern Cape province had the highest average child fuzzy MPI of 0.12. It was followed KwaZulu-Natal and Mpumalanga provinces with an average child MPI of 0.10. The results further revealed that the North West and Free State provinces had average child fuzzy MPI of 0.09, followed by Limpopo, Gauteng and the Northern Cape provinces with average child fuzzy MPI of 0.08. Lastly, the Western Cape province had the lowest average child MPI of 0.07.

Fig. 1.

The average child fuzzy Set MPI across provinces in 2017, 2018, 2019 and all years. Source: Own computation

Distribution of Child’s Fuzzy MPI Across Children’s Population Groups

The results in Fig. 2 show the distribution of fuzzy MPI across child’s population groups. The results revealed that Blacks/Africans children had the highest average fuzzy MPI with a total fuzzy MPI average of 0.10, followed by Coloured children with a total fuzzy MPI average of 0.07, 0.06 for white children and 0.05 for Indian/Asian children.

Fig. 2.

Average fuzzy set child MPI across population groups. Source: Own computation

Distribution of Child’s Fuzzy MPI Across Children’s Geography Type

The results presented in Fig. 3 shows the distribution of child’s fuzzy MPI across children’s geography type. It is revealed that children from the urban areas had the lowest total average fuzzy MPI of 0.08, followed by children from traditional areas with a total fuzzy MPI of 0.11 and children from farms with 0.12.

Fig. 3.

Average fuzzy set child MPI across geography type. Source: Own Computation

Evaluation of Model Fit and Multicollinearity Diagnostics

Table 7 shows the determinants of child’s fuzzy MPI using the Tobit model. The results showed a strong model fit, with a highly significant likelihood ratio chi square (Prob> Chi2 = 0.0000) indicating that the explanatory variables jointly influenced the child’s fuzzy MPI. The model was tested for multicollinearity using the Variance Inflation Factor (VIF) which revealed absence of serious multicollinearity with an average VIF of 3.76. This is according to Shrestha [84] who emphasized that a VIF value of 1 indicates that there is no correlation among independent variable, while VIF values between 1 and 5 suggest moderate correlation. Moreover, VIF values above 5, especially over 10 indicate high multicollinearity which can weaken the reliability of the regression coefficients.

Table 7.

Tobit model results on factors influencing child’s fuzzy set multidimensional poverty indicators

Fuzzy	Coefficient	t
Seeing	− 0.0064699	-4.79***
Hearing	− 0.0085841	-4.71***
Walking	0.0149214	7.54***
Remembering	0.0010963	0.71
Selfcare	− 0.0004096	-0.45
Communication	0.0061876	3.56***
Province (Western Cape as reference)
Eastern Cape	0.0202143	19.97***
Northern Cape	− 0.0003651	-0.31
Free State	0.0010275	0.88
KwaZulu-Natal	0.0008254	0.82
North-West	− 0.0038428	-3.25***
Gauteng	− 0.0002041	-0.21
Mpumalanga	0.0017435	1.57
Limpopo	− 0.0165127	-15.16***
Gender of child	− 0.0011164	-2.82***
Age	0.0000936	1.43
Population group (Black/African as reference)
Coloured	− 0.014303	-16.11***
Indian/Asian	− 0.0169717	-8.37***
White	− 0.012433	-9.60***
Son/daughter	0.008982	16.53***
Father alive (No as reference)
Yes	− 0.0030002	-4.47***
Do not Know	− 0.0006626	-0.43
Father part of the household	− 0.0050681	-9.32***
Mother still alive (No as reference)
Yes	− 0.000541	-0.59
Do not know	0.0044519	1.27
Mother part of household	− 0.0022775	-4.09***
Domestic or household workers services	0.0166665	18.89***
Household size	0.0002599	3.40***
Geography type
Traditional	0.0164765	30.34***
Farms	0.0333984	29.23***
Salaries/wages commission	− 0.0124142	-28.29***
Income from a business	− 0.0069858	-11.98***
Remittances	− 0.0062276	-11.65***
Pensions	− 0.0112907	-9.18***
Grants	0.0096668	15.80***
Social grants	− 0.0073088	-13.50***
Sales of farming products and service	− 0.0000583	-0.04
Other income sources	− 0.0102879	-6.65***
Backyard garden	0.0047083	8.95***
School garden	− 0.0013766	-0.24
Communal garden	0.0031533	0.95
Year (2017 as reference)
2018	− 0.0048383	-10.33***
2019	0.0137769	27.71***
_cons	0.0681964	35.71***
var(e.fuzzy)	0.0028002
Prob > chi2	0.0000
Number of obs	71 711
Log likelihood	109007.43

Source: Own computation. NB ***, implies statistically significant at 1% level of significance

The Tobit regression results presented in Table 7 shows the household head’s functional impairments that affected their fuzzy MPI. The eyesight impairment parameter showed statistical significance (p < 0.01) with negative sign. This implies that children from households which were affected by visual impairment had lower fuzzy MPI. The estimated parameter for being affected by hearing impairment also showed statistical significance (p < 0.01) with negative sign. This implies that children who were from hearing impaired households had lower fuzzy MPI. These results suggest that children who were from sight impaired households were less multidimensionally poor possibly due to stable support from social grants [33]; Bryning et al., [7]. However, Lindsay et al., [41] and Hanass-Hancock and McKenzie [27] found contrary results. The parameter for walking impairment also showed statistical significance with a positive sign (p < 0.01). This means that children who were from households’ heads with walking difficulties had higher fuzzy MPI. These findings can be associated with those of Buthelezi et al., [9]. The communication impairment parameter showed statistical significance (p < 0.01) and a positive sign. This implies that in accordance with the findings of Disability Data Initiative [16], children from households with communication problem had higher multidimensional poverty.

The Eastern Cape province showed statistical significance (p < 0.01) and a positive sign. This means that sampled children from the Eastern Cape had higher fuzzy MPI, when compared to children from the Western Cape province. These findings suggest that persistent developmental backlogs and high rural poverty in the Eastern Cape continue to affect child well-being adversely (Stats SA, 2022). In addition, these findings are in alignment with those of Chirowamhangu [12] and Mpongwana and Cishe [52] which show that the Eastern Cape remains one of the most deprived provinces in South Africa. Specifically, Mtize , [53] revealed that over 80% of young children in the Eastern Cape live below the upper-bound poverty line, and approximately 27% experience chronic malnutrition or stunting. Moreover, the Mtize [53] also revealed a widespread child food insecurity and systemic failures in service delivery across the province. The North-West province also showed statistical significance (p < 0.01) and a negative sign, implying that when compared to children from the Western Cape, children from the North-West had their fuzzy MPI reduced by 0.0038. The North West province’s encouraging signs of poverty reduction can be supported by the findings of Mutanga et al., [55] through improved child nutrition and feeding practices. With a mean Dietary Diversity Score of 4.18 and 86.0% of children aged 0–2 years breastfed within an hour of birth, the province reflects progress in addressing nutritional aspects of poverty. The results further revealed that Limpopo children had their fuzzy MPI significantly (p < 0.01) reduced by 0.0165. This result is contrary to those of Mathebula and Alaba [48] who revealed that poverty is persistent in Limpopo from the highest national rates at 88.2% due to its legacy as a former homeland area with limited infrastructure and economic opportunities. They further emphasised that the severe deprivations transcend income, indicating inadequate access to basic services like clean water, sanitation, education and healthcare [48].

The results in Table 7 show that male children had their fuzzy MPI significantly reduced by 0.0009 (p < 0.1) when compared to female children. The gender difference revealed can be associated with the findings by Akala [1], who found that in post-apartheid South Africa, female children, specifically in rural and semi-urban areas were facing higher deprivations in school attendance, nutritional intake, and personal care responsibilities due to persistent gender norms. Also, Raskind et al., [75] noted that female children, specifically in low-income households, are more likely to be kept at home for household chores or caregiving, especially during school disruptions, increasing their exposure to multidimensional poverty.

In terms of population group, the parameter for Coloured children - people of mixed racial ancestry - showed statistical significance and negative signs. This implies that when compared to Black/African children which are the indigenous African groups, including Zulu, Xhosa, Sotho, Tswana, Venda, Ndebele, Tsonga, Swazi, and others, Coloured children had their fuzzy MPI reduced by 0.0143. The parameter for Indian/Asian (descendants of Indian or other Asian immigrants) also showed statistical significance (p < 0.01) and negative signs. This implies that Indian/Asian children had their fuzzy MPI reduced by 0.0170. The results also revealed that White children (people of European descent) had their fuzzy MPI significantly (p < 0.01) reduced by 0.0124. These results are in alignment with those of Omotoso et al., [65], who found that Black African children remain the most vulnerable to multidimensional poverty in South Africa. This is largely due to geographic concentration in under-resourced provinces, lower parental education levels, and weaker access to public services. Similarly, Hale et al., [23] emphasized that while there has been an overall decline on poverty, race remains a powerful predictor of poverty exposure and intergenerational disadvantage. Furthermore, Nkhoma , [60], used the data from the National Income Dynamics Study (NIDS) and showed that White and Indian/Asian households consistently score better on indicators related to child education, nutrition, and access to utilities, which contributes to lower poverty among children in these groups.

The results in Table 7 revealed statistical significance (p < 0.01) for biological children in the household and negative sign. This implies that children who were biological daughters/sons in a household had their fuzzy MPI reduced by 0.0090. The results also revealed that children who had fathers that were alive had their fuzzy MPI significantly reduced by 0.0030 compared to those who did not have alive fathers. The parameter for father as part of the household also showed statistical significance at 1% level with negative sign. This implies that children who were residing in the same household as their fathers had their fuzzy MPI significantly reduced by 0.0051. The results suggest that co-residence of fathers contributes positively to child welfare, possibly through financial stability, emotional support, and enhanced supervision (Daniel & Madugu [15, 83]). Moreover, Hall and Mokomane , [26] noted that the absence of fathers is common in South Africa due to labour migration, non-marital births, and family breakdown and has also been linked to higher rates of child poverty, lower educational attainment, and food insecurity. Additionally, Chung , [13] also revealed that father engagement in the same household is positively associated with improved child developmental outcomes and household economic wellbeing, echoing the present findings. These results underscore the critical role of household structure in shaping children’s experiences of poverty and support concerns for strengthening social protection systems and family support programs that encourage father involvement and alleviate gender disparities.

Residing with mother has emerged as a critical household dynamic influencing child’s experience of multidimensional poverty. The parameter for mother being alive did not show any statistical significance. However, the parameter for mother being part of the household shows statistical significance (p < 0.01) and negative sign. This result implies that children who were residing with their mothers had their fuzzy MPI reduced by 0.0023. This result aligns with a report by Spenner [85] who emphasized that children residing with their mothers are less likely to experience multiple deprivations due to the secure, emotional attachment of maternal presence. It was further mentioned that mothers also promote cognitive and social development, equipping children with the resilience and skills needed to thrive despite poverty Spenner [85]. The parameter for domestic worker also showed statistical significance (p < 0.01) with positive a sign. This implies that children who had domestic services in their households had their fuzzy MPI increased by 0.0167. This might be due to diverting household resources away from children’s basic needs and reducing parental involvement in their development. The parameter for household size showed statistical significance (p < 0.01) with a positive sign. This implies that an additional member to the household had their fuzzy MPI increased by 0.0003. This result is in line with those of Drammeh et al., [17] and Fusco and Islam [22] who reported that additional children raise the risk of being poor as it increases dependency without a corresponding rise in household income or their resources, straining access to basics like food, education and proper healthcare.

The parameter for traditional geography type showed statistical significance (p < 0.01) and positive sign. This result implies that when holding other variables constant, children who resided in traditional areas, which are semi-urban, had their fuzzy MPI increased by 0.0165 when compared to children from urban areas. The results further revealed that when compared to children from urban areas, children from farms, which are rural areas outside urban centers where the primary land use is agricultural, had their fuzzy MPI significantly (p < 0.01) increased by 0.0334. this result is expected an in line with those of Mathebula [48]. Children from urban households are expected to have access to functioning housing with access to basic assets [46, 95].

Regarding the sources of income, the parameter for salaries/wages showed statistical significance (p < 0.01) with negative signs. This implies that the null hypothesis should be rejected because children from households that had incomes/wages as a source of income had their fuzzy MPI reduced by 0.0124. This result is expected and it is in line with those of Fransman and Yu [21] and Fonta et al., [19] because depending on the income levels, salary earners may face lesser barriers in renting or buying good houses, procurement of assets and ability to ensure adequate nutrition for children [87].

The parameter for income from a business showed statistical significance (p < 0.01) and a negative sign. This implies that the null hypothesis should be rejected and children from households that had income from a business had their fuzzy MPI significantly reduced by 0.0070. Businesses are significant sources of income in many South African households [74]. Therefore, engagement in business can promote access to requisite incomes to procure better housing and other essential household assets [45]. The remittances parameter also showed statistical significance (p < 0.01) with a negative sign. This result implies the null hypothesis should be rejected and that children from households that were receiving remittances had their fuzzy MPI significantly reduced by 0.0062. The results also showed that children from households that received pensions had their fuzzy MPI statistically (p < 0.01) reduced by 0.0113. The results are expected and in line with those of Hu et al., [32] and Koomson et al., [37] because remittances can improve their access to healthcare, nutrition, education and other basic needs essential for child’s wellbeing.

The parameter for once off grants also revealed statistical significance (p < 0.01). This implies that children from households that received once off grants had their fuzzy MPI significantly increased by 0.0097. This result aligns with those of Zaneva et al., [96] who reported that once off grants may not lift people out of poverty because of their insufficiency. It is also revealed that children from households that received monthly social grants had their fuzzy MPI significantly (p < 0.01) reduced by 0.0073. This result is expected an in line with that of Osei and Turkson [68] because monthly social grant is a targeted, regular payment that can assist in covering basic needs. The parameter for sales of farming product/services did not show any statistical significance. Regarding other income sources, the results showed that children from households that were receiving income from other streams had their fuzzy MPI reduced by 0.0103. This implies that other income sources such as rental and sale of properties are essential for child poverty reduction [29, 30].

Even though the parameters for school and communal garden did not show any statistical significance, the parameter for backyard garden showed statistical significance. This implies that at 1% level of significance, children who had backyard gardens in their homes had their fuzzy MPI significantly increased by 0.0047 (p < 0.01). This result is expected and in line with those of Song & Imai (2019) because having other income streams provide greater financial stability and a better access to basic needs. Moreover, the parameter for 2018 showed statistical significance and a negative sign. This implies that when holding all other variables constant, in 2018 children had their fuzzy MPI reduced when compared to 2017. In 2019, however, it is revealed that when compared to 2017, children had their fuzzy MPI significantly (p < 0.01) increased by 0.0138.

Impacts of Income Sources on Fuzzy MPI

The results presented in Table 8 show the parameters for POM, ATE and ATET on the treated group across sampled children’s households’ different income sources. The POM for the treated group showed statistical significance at 1% level of significance with 0.0763, 0.1012, 0.093, 0.9332, 0.0923, 0.0954 and 0.092 for sampled children’s households that reported to have been receiving grants, salaries, income from business, remittances, pension, social grants and sale of farming products and services, respectively. These results can be compared with those of the control group which also showed statistical significance at 1% with 0.0773, 0.0995, 0.0915233, 0.1019, 0.0914, 0.0872 and 0.0998 of the sampled children’s households that reported to have received grants, salaries, income from business, remittances, pension, social grants and sale of farming products and services, respectively.

Table 8.

Treatment effects parameters for child’s fuzzy MPI

Indicators/Treatment	Grants	Salaries	Income from business	Remittances	Pension	Social Grants	Sales of farming product and service
Potential Outcome Mean (treated)	0.0763029	0.1011647	0.0930381	0.0933221	0.0923357	0.0954227	0.092037
Z statistics	86.31	257.91	401.19	379.25	423.70	344.78	422.46
Potential Outcome Mean (control)	0.077396	0.0994908	0.0915233	0.1018939	0.0913813	0.0872016	0.0998041
Z statistics	75.23	200.03	261.90	278.61	150.82	153.38	171.67
Average Treatment Effect (ATE)	0.0181606	− 0.0140565	− 0.0064875	− 0.0044692	− 0.0093761	− 0.0090205	0.0086012
Z statistics	19.93	-29.65	-10.53	-7.40	-6.54	-15.33	4.83
Average Treatment Effect on the Treated (ATET)	0.194826	− 0.0157485	− 0.0072355	− 0.006654	− 0.0113488	− 0.0121242	− 0.0002076
Z statistics	18.85	-28.26	-12.88	-12.52	-9.78	-17.77	0.853

Moreover, the values for ATEs are positive for grants, negative for all other sources of income and statistically significant at 1%. These imply that the MPI for children from households who received grants, salaries, income from business, remittances, pension, social grants and sale of farming products and services is significantly higher by 0.0182 for households that indicated to have received grants and lower by 0.0141, 0.0065, 0.0047, 0.0094, 0.0090 and 0.0086 for children from households that reported to have received salaries, income from business, remittances, pension, social grants and sale of farming products and services, respectively when compared to the control group. The ATET results showed statistical significance at 1% and revealed that children from households that reported to have received grants had their MPI being significantly higher by 0.0195 compared to if they never had such source of income. In addition, children from households that reported to have received salaries, income from business, remittances, pension, social grants and sale of farming products and services had their average probability of being multidimensionally deprived significantly lower by 0.0195, 0.0157, 0.0072, 0.0067, 0.0113, 0.0121 and 0.0002, respectively compared to if they never had such sources of income.

These results are expected and in line with those of Koswana [38] because while grants provide a crucial safety net, they are solely used for food and cannot cover other needs like cost of transport, healthcare, clothes, among others, making them insufficient to bring children and their households out of poverty. Moreover, these grants are exceeded by inflation, especially the rising prices of food [38]. On the other hand, income sources like salaries, income from business, remittances, pensions, social grants and sales of farming products/services are more stable and belong to higher financial resources which can allow households to cover most of their basic needs [2]. Moreover, Askarov and Doucouliagos , [2] emphasized that remittances can be used for consumption and investment, specifically on education which can address the dimension of education. The positive impact on social grants might be in alignment with targeted social protection programmes which significantly reduce child poverty when they are supplementing other income sources. However, the social grants results are contrary to those of Belete [6] who revealed that households that were receiving social grants were poorer compared to those who were non-recipients. The other sources of income results are in line with those of Koomson et al., [37] because household diversified income streams reduce poverty among children.

Conclusion

This study highlighted the impact of households’ income sources on child’s fuzzy set multidimensional poverty indicators in South Africa. Thus, some light was shed on the complex and unequal landscape of child’s fuzzy set multidimensional poverty, highlighting how household income sources significantly shape the poverty status outcomes among children in South Africa. The analysis revealed that disparities persist across racial, geographical and provincial lines, with children in traditional and farming communities, specifically those from Black/African households, experiencing higher levels of depravations when compared to other three races in the country. Moreover, the elevated child multidimensional poverty levels in grant-receiving households further underscores that social assistance alone is insufficient in breaking intergenerational poverty cycles. In contrast, it was revealed that households who were benefiting from stable income sources such as salaries, remittances, and business earnings were able to provide better outcomes for their children, reducing their fuzzy set multidimensional poverty. This also underscores the central role of economic empowerment in reducing poverty.

The targeted poverty reduction efforts must therefore not only focus on geographic and racial equity but also prioritize the diversification and strengthening of household livelihoods by promoting inclusivity such as promotion of access to sustainable employment opportunities, offering more support to small-scale entrepreneurship, enhancing business skills training programmes and expending access to financial services and credit to enable informal sector growth. These strategies require to be tailored to the specific needs of households in both the traditional and farming communities. Moreover, they must be inclusive of marginalised racial groups that are faced with structural barriers to economic participation. This can assist households in reducing dependency on grants and break the cycle of intergenerational poverty. In addition, the results revealed that the presence of biological parents are protective factors against child’s multidimensional poverty. These findings highlight the urgent need for integrated policies that combine economic, social and family interventions to address the roots of child poverty in its many dimensions. Addressing these determinants through informed, data-driven strategies can be of importance for advancing the wellbeing of South African children and achieving sustainable progress on the Sustainable Development Goals (SDGs).

Author Contributions

TC conceptualized and wrote the paper.

Funding

None.

Data Availability

No datasets were generated or analysed during the current study.

Declarations

Competing interests

The authors declare no competing interests.