Evaluating the demand, supply and impact of early childhood development programmes in South Africa

ABSTRACT

Early childhood development (ECD) access and quality inequalities can manifest as school and labour market outcomes that are persistently unequal by race, gender and geography. Using the General Household Survey 2019 and Early Childhood Development Census 2009–2014 data, we find that household socio-economic status, proximity to ECD centres and the mother's educational attainment are positively associated with ECD enrolment. Having an economically inactive or unemployed adult in the household is negatively associated with enrolment rates. This paper also evaluates the programme gains (value added) of five ECD programmes in the large-scale cross-sectional the Early Learning Outcomes Measure study. We find gains that are surprisingly large, considering that these programmes served mainly poorer children, and because contact time per week for three of the programmes was only 2½ to 8 hours.

KEYWORDS:

• Early childhood development
• enrolment
• education quality
• learning outcomes

SUBJECT CLASSIFICATION CODES:

• I20
• I24
• I25

Acknowledgement

The authors wish to thank members of the ELOM team, in particular Andy Dawes, Linda Biersteker and Elizabeth Girdwood, as well as Colin Almeleh, Laura Brooks and Zaheera Mohamed of Ilifa Labantwana, participants in a Resep internal seminar, and Resep colleagues Gabrielle Wills and Jesal Kika-Mistry for inputs to the development of this paper.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 The term ECCE or ECD centre, according to the Audit of Early Childhood Development Centres National Report (DSD, 2014: xi), who commissioned the ECD Census data collection, can refer to:

‘ … a crèche, day care centre for young children, a pre-school, and/or after school care. Exclusions include Grade R classes attached to a primary school and home-based programmes with less than six children run by day mothers and child minders and/or play groups and for the purposes of the audit those with no children under the age of 6.’

2 Total monthly household income is derived by Statistics South Africa, which combines all earnings, income from grants and remittance income. The total monthly household income for all households in GHS is divided by the number of individuals within that household to produce a household income per capita measure. These per capita incomes are then ordered from lowest to highest, and then grouped into five quintiles, where quintile 1 represents the poorest 20 percent of all South African households, and quintile 5 represents the richest 20 percent of all households.

3 The centre or centroid of the enumeration area (EA) polygon is assumed to be a reasonable proxy for the child's residential location. For the regression analysis, we restrict our sample to children living in EA's that are 25km or smaller in area so that very large, sparsely populated EA's do not influence our results.

4 Registration refers to whether the centre is registered with South Africa's Department of Education.

5 The GHS 2019 asks if individuals are engaged in work for remuneration and work without remuneration for "any kind of business, big or small’. We assume that individuals doing work without remuneration are unavailable for childcare commitments within the household.

6 21% of children in our sample do not have a mother present in the same household. In these cases, the household head's educational attainment is substituted as a proxy for caregiver educational attainment.

7 OLS regression is used for the sake of interpretability. Logistic regression coefficients converted to marginal effects are remarkably similar to the OLS estimates.

8 Learning gains amongst children attending two centre development models and three playgroups aimed at children from low-income households (quintiles 1 to 3) were evaluated in the Western Cape, Mpumalanga, Free State and Kwazulu-Natal.

9 Implicitly, the constant refers to age 0 months, and the year dummies have to be added to this.

10 Confining the regressions to Quintile 1–3 centres only, the maturation effect is slightly lower, which would imply that the measured impacts of the five programmes are larger, closer to the upper end of the results reported below.

11 For Playgroup 1, that showed the largest gains and really consists of two playgroups in different provinces, the gains were significantly larger in a site in Mpumalanga where children attended three times a week for 4 h, than in a site in the Western Cape where only two sessions of the same length are offered per week.

Additional information

Funding

This work was supported by Ilifa Labantwana.