Table 2. Steps for estimating the annual costs of extending maternity leave for women in formal employment in Brazil, Ghana and Mexico.
| Step | Aim | Data used | Process | Variables input | Notes |
|---|---|---|---|---|---|
| Step 1 | Compute the probability of women having a baby in the previous year, given a set of women’s characteristics, needed to compute the value of α in Equation 2 in the methods section | Fertility data Brazil: National Household Sample Survey 201533 Ghana: Ghana Living Standard Survey 201734 Mexico: National Survey of Demographic Dynamics 201435 |
Identify women of reproductive age. Among this subset of women, generate combinations according the available sociodemographic variables. For each of the combinations, calculate the percentage of women who had a baby in the previous year (as a proportion of the total number of women of reproductive age) |
Reproductive age Brazil & Ghana: 16–49 years; Mexico: 18–49 years. Marital status Brazil & Ghana: single; married or living with partner; widow or divorced or separated; Mexico: single; married; divorced. 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 or middle or incomplete high school; complete high school or higher incomplete or technical career; higher complete or more. Mexico: incomplete primary or less; primary or some secondary; secondary or some high school; high school completed; technical training or incomplete professional education; university degree. Locality Brazil & Ghana: rural; urban. Mexico: rural; semi-urban; urban. |
Number of combinations Brazil: 180 Ghana: 150 Mexico: 270 |
| Step 2 | Estimate the probability of women working in the formal sector having a baby in the previous year (variable α), given a set of women’s characteristics | Fertility and employment data Brazil: National Household Sample Survey, 201533 Ghana: Ghana Living Standard Survey, 201734 Mexico: National Survey of Demographic Dynamics, 201435 and the National Survey of Occupation and Employment, 2013–201436 |
Define formal employment. Considering the combinations generated in Step 1, add employment information to estimate the probability of having a baby only among formally employed women. This may be done by tabulating data from a single survey (such as in Brazil and Ghana) or through merging different data sets (as in Mexico) |
Formal employment Brazil: women with a formal contract, including domestic workers, military and civil servants, as well as employers and self-employed persons who contribute to social security (variables to operationalize: occupation and social security contribution). Ghana: women who have at least one social benefit (maternity leave, sick leave or holidays) and a written or verbal contract (variables to operationalize: holidays, paid leave and contract). Mexico: women who have access to social security and have the right to a paid maternity leave (variable to operationalize: social security) |
NA |
| Step 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
× Popy). This step seeks to generate a more realistic estimate of the women employed in the formal sector who may claim maternity leave in a given year |
Census data or demographic projections. Brazil: World Bank 2015 population projections for age group37 Ghana: World Bank 2017 population projections for age group37 Mexico: Inter-census Mexican Survey, 201538 |
Identify national estimates of women in reproductive ages Popy Multiply the population by each of the values of α’s generated in Step 2 |
No additional variables | 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 |
| Step 4 | Estimate the mean or median weekly wages of women working in the formal sector, given a set of women’s characteristics (W). Multiply the wage by the weighted population of women of reproductive age |
Employment or wage data. Brazil: National Household Sample Survey 201533 Ghana: Ghana Labour Force Survey 201539 Mexico: National Survey of Occupation and Employment 2013–201436 |
For each group of women (combinations) identify the mean or median weekly wage. To decide whether to use the mean or the median, plot a density function graph of weekly wages to see if its distribution is symmetrical (see Fig. 1 for example). If the distribution is not symmetrical and the mean is not centred, use the median. Determine the percentage of the salary that would be covered by the maternity leave benefit and multiply it by the weekly wage. Multiply the covered wage by the weighted population computed in Step 3. To estimate the mean and median weekly cost per woman, W × (α × Popy) can be divided by the estimated number of women expected to receive maternity leave |
Weekly wages Brazil: full-time weekly wages (at least 44 hours of work per week). Ghana: full-time weekly wages (at least 40 hours of work per week). Mexico: full-time weekly wages (at least 40 hours of work per week) |
The assumption for the three countries was that maternity leave benefits would cover 100% of the salaries |
| Step 5 | Determine the incremental weekly coverage of the maternity leave IC according to relevant thresholds. Estimate the annual cost of expanding maternity leave |
Laws, international and national organization documents establishing length of maternity leave coverage | Multiply the number of weeks to be covered by W × (α × Popy) to estimate the annual cost of the expansion in the maternity leave coverage | NA | NA |
NA: not applicable.