Table 2 Methodological steps for estimating the annual costs of a maternity cash transfer for informally employed women in Brazil and Ghana

1

Compute the probability of a women having a baby in the previous year, given a set of women’s characteristics, needed to compute the value of α in Step 2

Fertility data sources:

• Brazil: National Household Sample Survey 2015 (PNAD) [44]

• Ghana: Ghana Living Standard Survey 2017 (GLSS7) [45]

• Identify women of reproductive age (16-49 years)

• Among this subset of women, generate combinations based on:

Age

Marital status

Educational level

Locality (based on country level definitions)

• For each of the combinations, calculate the percentage that had a live birth in the prior year (as a proportion of the total number of women of reproductive age)

Age groups

• Brazil& Ghana: 16-24; 25-29: 30-34; 35-39; 40-49.

Marital status

• Brazil& Ghana: single; married/living with a man; widow/divorced/separated.

Educational level

• Brazil: no education; kindergarten or incomplete primary; complete primary or incomplete middle; complete middle or incomplete high school; complete high school; higher or any technical career.

• Ghana: no education; primary or kindergarten; secondary/middle or incomplete high school; complete high school or higher incomplete or technical career; higher complete or more.

Locality

• Brazil & Ghana: rural; urban

Number of combinations:

• Brazil: 180

• Ghana: 150

2

Estimate the probability of a women working in the informal sector having a baby in the prior year (α), given a set of women’s characteristics

Fertility and employment data:

• Brazil: National Household Sample Survey 2015 (PNAD) [44]

• Ghana: Ghana Living Standard Survey 2017 (GLSS7) [45]

• Define informal employment

• Using the demographic groups generated in step 1, add employment information to estimate the probability of having a baby only among informally employed women

Informal employment

• Brazil: individuals without a formal contract, including domestic workers, employers and self-employed workers who do not contribute to social security, unpaid workers, as well as workers in production for own consumption and construction for own use. Variables to operationalize: occupation and social security contribution [46]

• Ghana: individuals who don’t have at least one social benefit (maternity leave, sick leave or holidays) and were without a written or verbal contract [47]. Variables to operationalize: holidays, paid leaves and contract

Employment in the formal and informal sector can vary by each country, national definitions should be prioritized [6]

3

Estimate the population of women of reproductive age weighted by the probability of having a baby in the previous year based on individual characteristics (α ∗ Pop_y)

This step seeks to generate a more realistic estimate the number of women employed in the informal sector who may claim maternity leave in a given year (i.e., target beneficiaries)

Census data:

• Brazil: 2010 Census (IBGE 2012)

• Ghana: 2010 Census (GSS 2010)

Population projections:

• Brazil: World Bank 2015 population projections for age group [48]

• Ghana: World Bank 2017 population projections for age group [48]

Employment Data:

• Brazil: National Household Sample Survey 2015 (PNAD) [44]

• Ghana: Ghana Living Standard Survey 2017 (GSS 2017) [45]

• Identify national estimates of women of reproductive age (16-49 years) currently working in the informal sector for Pop_y

• Multiply the population by each of the values of α’s generated in step 2

• Number of women 16-49 years currently working in the informal sector

While some surveys used in steps 1 and 2 may have expansion factors (e.g., Brazil), we strongly recommend not using them as they were generated for expanding other population subgroups. This may increase the error of any estimated parameter.

4

Estimate the weekly cost (UC_CT) of the maternity cash transfer (CT) using common welfare measures (i.e., minimum wages, poverty lines).

Minimum wage:

• For both countries were retrieved from the WageIndicator [34]. This information can be retrieved from national offices as well.

Poverty lines:

• For both countries were estimated based on the oncome poverty line from the World Bank [36]

• Determine the weekly maternity cash transfer through minimum wages and poverty lines

Cash transfers:

• Estimations can be performed through different operationalizations of the minimum wage and the poverty line, which may depend on contextual aspects.

• For purposes of the current analysis the maternity cash transfer was estimated at:

the minimum wage

the poverty line

twice the poverty line

The assumption for the two countries was that maternity cash transfer would be provided to all eligible women in one year, but incremental coverage could be modelled.

5

Determine the number of weeks to be covered, or incremental weekly coverage of the maternity cash transfer (IC_y) according to relevant thresholds

International and national organization documents establishing length of maternity leave coverage

• Determine the duration of the maternity leave cash transfer program.

For both Brazil and Ghana, the following durations were used for comparing estimates:

• 12 weeks: current duration of maternity leave in the formal sector in Ghana

• 14 weeks: duration recommended by the ILO

• 18 weeks: duration of maternity leave for formal workers currently discussed by key stakeholders in Ghana and approximate current duration of maternity leave in the formal sector in Brazil; extension recommended by the ILO R191.

• 26 weeks: durations to support EBF

Could include durations established policies for the formal sector

6

Determine the administrative cost of operating the maternity leave cash transfer program (AdmCost_y). Multiply the weekly cost of the maternity CT (UC_CT) by incremental coverage (IC_y) by the weighted population (α ∗ Pop_y) andadd the yearly administrative costs (AdmCost_y) to determine the total annual cost of the maternity cash transfer program (MatCT_y).

Administrative costs of programs similar in structure (i.e. one-time subsidy for a specific purpose) or from the same intervention in similar countries:

• Brazil: the estimated administrative costs for the Mexican maternity cash transfer [27]

• Ghana: average of the administrative costs of two programs (i.e., LESDEP and NHIS) [49]

Multiply the number of weeks to be covered (UC_CT) by (IC_y) by (α ∗ Pop_y). This will estimate the annual cost of the expansion in the maternity leave coverage.

Add the administrative costs to (α ∗ Pop_y) ∗ CT ∗ IC_y

Administrative costs:

• Brazil= 5.6%

• Ghana=5.8%

This step requires gathering the best locally available data to estimate the administrative costs. Sometimes it can be retrieved from national budgets if they are publicly available [27]. If unavailable, regional data may be used [32].