Childcare for Childhood and Business Development

The first intervention is offering a subsidy to cover the tuition and lunch fees for the target child to attend nursery school for a full year (three terms). The second intervention is providing a cash grant, equivalent to the average value of the subsidy received by individuals in the first intervention (average at the community level). The cash transfers will be paid in three installments, corresponding to the start of school terms so that the timing of the cash payments is identical to the payment for child care.

As discussed above, other studies have found rather modest impacts of providing grants to women, but our hypothesis is that this may be due to binding household constraints. Therefore, we will explore whether there are complementarities between child care and cash grants: the combination of a cash grant and childcare is the third treatment arm. In addition, there will be a control group that receives neither family nor business support. As a result, there are four different treatment arms:

- C – Control - Women who will not receive child care support, nor cash grant.
- T1 – Treatment 1 – Women who will receive child care only.
- T2 – Treatment 2 – Women who will receive cash grant only.
- T3 – Complementarities – Women who will receive both child care and cash grant.

2019-01-15

2019-12-31

We have 6 families of outcomes (please see below for further details): female respondent's labor supply, female respondent's labor earnings, female respondent's well-being, labor supply of household members, labor earnings of household members, productive assets owned by the household. In addition, we have 2 child-level outcomes: early childhood development and school enrollment.

Respondent-level outcomes

We focus on three categories of respondent outcomes: Labor supply, labor earnings and well-being.

Category 1: Labor supply of the respondent:

a. Total labor supply:
i. Extensive margin: an indicator for whether the respondent spent any time working in any income-generating activity during the last month.
ii. Intensive margin: hours spent by the respondent working in income-generating activities during the last month.
b. Self-employment:
i. Extensive margin: an indicator for whether the respondent spent any time working in any income-generating activity where the respondent was self-employed (i.e. main person in charge of the business) during the last month.
ii. Intensive margin: hours spent by the respondent working in any income-generating activity where the respondent was self-employed (i.e. main person in charge of the business) during the last month.
c. Wage-employment:
i. Extensive margin: an indicator for whether the respondent spent any time working in any income-generating activity where the respondent was working for an employer during the last month.
ii. Intensive margin: hours spent by the respondent working in any income-generating activity where the respondent was working for an employer during the last month.

Category 2: Labor earnings of the respondent:
a. Total labor earnings: total labor income the respondent generated during the last month.
b. Earnings from self-employment: total labor income the respondent generated from self-employment during the last month. We will use two measures for self-employment income:
i. Total revenues from business activities
ii. Total profits from business activities
c. Earnings from wage-employment: total labor income the respondent generated from wage-employment during the last month.

Category 3: Well-being of the respondent: Three indicators for well-being of the respondent:
a. Happiness: “How happy are you with your life?”, answered on a scale from 0 to 10. We will standardize the answers.
b. Satisfaction with life: “In your opinion, where are you on the ladder of life at the moment?”, answered on a scale from 1 to 10. We will standardize the answers.
c. Perceived stress: We use Cohen’s perceived stress scale. We will follow the Cohen’s scale guidelines and construct binary variables that indicate if the individual is in the low stress category (score from 0 to 13), moderate stress category (score from 14 to 26) or high stress category (score above 26).

Household-level outcomes

We focus on three categories of household outcomes: Total labor supply of all household members, total labor earnings of the household and assets owned by the household.
Category 4: Labor supply of all household members:
a. Total labor supply:
i. Extensive margin: an indicator for whether household members spent any time working in any income-generating activities during the last month.
ii. Intensive margin: total hours spent by household members working in income-generating activities during the last month.
b. Self-employment:
i. Extensive margin: an indicator for whether household members spent any time working in any income-generating activity where they were self-employed during the last month.
ii. Intensive margin: total hours spent by the household members working in any income-generating activity where they were self-employed during the last month.
c. Wage-employment:
i. Extensive margin: an indicator for whether household members spent any time working in any income-generating activity where they were working for an employer during the last month.
ii. Intensive margin: total hours spent by household members working in any income-generating activity where they were working for an employer during the last month.

Category 5: Labor earnings of the household:
a. Total labor earnings: total labor income generated by all household members during the last month.
b. Earnings from self-employment: total labor income generated by all household members household businesses during the last month.
c. Earnings from wage-employment: total labor income generated by all household members from wage-employment during the last month.

Category 6: Productive assets owned by the household
a. Total value of productive assets owned by the household

Child-level outcomes

To measure child development, we use the International Development and Early Learning Assessment (IDELA) tool developed by Save the Children. IDELA measures development across four domains: motor skills, early literacy, early numeracy and socio-emotional skills. The tool is adapted to the age range of the children that we target and has been translated for and used in Uganda in the past. The complete tool can be obtained from Save the Children: https://idela-network.org/.
We construct an aggregate standardized score. For each of the four domains of the IDELA tool, we will subtract the control group mean and divide by the control group standard deviation. We then take the unweighted average of all the z-scores, and again standardize to the control group.
We will focus on two child-level outcomes:
• Child-outcome 1: The standardized IDELA test score for the target child.
• Child-outcome 2: School attendance: a binary variable equal to one if the target child is enrolled in school.

The sample:

First, we selected three districts in Western Uganda (Kasese, Kyenjojo and Kabarole), three districts in central Uganda (Mukono, Masaka and Mityana) and three districts in Eastern Uganda (Mbale, Iganga and Jinja).

In these districts we identified 454 communities containing at least one nursery school. To identify eligible households, we conducted a census in September 2018 in each of the 454 communities.

To be eligible, a household has to satisfy three criteria: (i) the female caregiver should be present within the household (e.g. mother or grandmother), (ii) the household should have one (and only one) child in the age range 3-5 and (iii) the children shouldn’t already be attending full time childcare (but we allowed for children attending part-time child-care).

We also wanted to have a sufficiently large group of households both with and without younger siblings (below 3 years old). To that end we restricted the study samples to villages that have at least three households that satisfy the three criteria and do not have younger siblings, and at least one household that satisfies the three criteria and has younger siblings.

Finally, from the list of eligible households, we randomly selected 1496 in 400 communities to participate in the baseline survey.

Data collection:

The baseline was conducted in November and December 2018 and consisted of two surveys: a “household” survey and a “childhood development” survey. The two surveys were conducted by separate teams. The respondent for the household survey was the main female caregiver of the target child.

The household survey collected four main categories of information: (i) The socio-economic status of the household, including detailed information on business ownership (where applicable), assets, financial loans and savings; (ii) labor supply of the respondent and other household members; (iii) time use of the mother, father, and/or other main caregivers of the target child; and (iv) data on attitudes towards risk, happiness, stress, social attitudes and financial literacy.

The childhood development survey collected information on the target child using the International Development and Early Learning Assessment (IDELA) instrument developed by Save the Children. The tool consists of a battery of questions and tests that aim to measure the level of competency or mastery that children possess across four domains - motor skills, early literacy, early numeracy and socio-emotional skills.

We will conduct three follow-up surveys:

July-August 2019: Short-run follow-up survey – We will collect information on a limited set of outcomes from the household survey, but will not collect data on childhood development outcomes. From the households, we will collect the following information: questions related to the socio-economic status (excluding information related to loans and savings), time use of the mother, father, and/or other main caregivers of the target child and information on happiness and stress.

November-December 2019: Midline survey – We will collect data on the same set of indicators as at baseline, including the childhood development survey.

November-December 2020: Midline survey – We will collect data on the same set of indicators as at baseline and midline, including the childhood development survey.

Randomization:

Following the baseline survey, we randomized the 1496 households that were surveyed into the four different groups: C, T1, T2 and T3.
The randomization was conducted at the individual level. The following variables were used as randomization strata: district indicators, an indicator for whether the target child has younger siblings, whether the target child was already attending daycare at baseline, respondent’s occupation at baseline, and whether the mother of the target child resided in the same household at baseline.

As a result, out of the 1496 households that were part of the baseline survey, 363 were randomly allocated to T1, 364 to T2, 357 to T3 and 412 to C. These are not symmetric, because the number of observations differed across strata and it was not always divisible by four.

Randomization was done in office by a computer.

Household.

No

1496 households.

1496.

363 households in T1, 364 households in T2, 357 households in T3 and 412 households in the control group.

At the conventional level of significance of 0.05 and a power of 0.8, assuming the correlation between the baseline and the follow-up measure of the outcome of interest is 0.33, with 350 individuals per treatment arm, our sample size would allow for a minimum detectable effect of 0.2 standard deviations. If we pool the 3 follow-up surveys and estimate a pooled ANCOVA specification, then under the same assumptions as before, we can detect a minimum effect of 0.14 standard deviations.