The effects of motherhood on wages and labor force participation: evidence for Bolivia, Brazil, Ecuador and Peru Claudia Piras and Laura Ripani SDS / WID

Schedule n Introduction –Empirical issues about gender wage differentials –Motivation –Literature Review –Objective of this paper n Data –Data: why these four countries? –Sample restrictions n Labor Force Participation by Motherhood Status n The effect of Motherhood on Wages n Preliminary conclusions

Empirical issues about gender wage differentials n Empirical finding 1: Narrowing of the gender wage gap in the US and Latin America in the last decades. n Empirical finding 2: In general, women have less labor market experience than men and this fact played a big role in the explanation of gender wage differentials. n One of the reasons for having less labor market experience might be that women face more responsibilities in terms of childcare and nurturing: motherhood takes women away from the labor market or it makes them work part-time.

Motivation n From what we said before, the main motivation for this study is the importance of this question for its relevance for larger issues of gender inequality. n Most women are mothers, and a main aspect of intra-household gender division is assigning most child-rearing responsibilities to women. n Therefore, any “price” of being a mother will affect most women and contribute to gender inequality.

Motivation n On the other hand, if well-reared children are “public goods”, then the motherhood penalty is of theoretical and policy interest for two reasons. First, there exist an equity problem since mothers are not paid for their contribution to society. Second, the society must have a future problem of optimal level of care of future generations of responsible citizens. n Finally, if mothers are the only source of financial support in a family, we have an additional problem related to the lack of resources for their children’s development.

Literature Review n US: women with children earn lower wages than those who do not have any children, according to a number of studies (Hill 1979; Korenman and Neumark 1992, 1994; Waldfogel 1997, 1998; Lundberg and Rose 1999; England and Budig 1999). n There exists also a child penalty in the UK (Harkness and Waldfogel 1997), in Australia (Baxter 1992) and in Germany (Harkness and Waldfogel 1997). n Even after controlling for experience, some wage differential remains...

Child penalty: five mechanisms 1) Mothers spend more time at home caring for children, interrupting their experience and seniority, or at least interrupting full-time employment (human capital or labor supply explanation) 2) Mothers may choose jobs that trade off higher wages for some aspect of “mother-friendliness”. 3) Mothers exert less effort (per hour) on the job to conserve effort for household production, and this would affect wages through productivity. 4) Employers discriminate against mothers, treating them worse than other women. 5) Causal effects of having children may be spurious. There is heterogeneity in who selects into motherhood on unmeasured variables (i.e. career ambition) that also affect earnings.

Objective n The effect of being a mother on wages might differ substantially in Latin American countries compared to developed countries. n The objective of this paper is to calculate the effects of motherhood on wages and labor force participation for four Latin American countries. The countries chosen were Peru, Bolivia, Ecuador and Brazil.

Data: Why these four countries? n Household surveys for each country: Pesquisa Nacional para Amostra de Domicilios for Brazil Encuesta de Condiciones de Vida for Ecuador Encuesta Continua de Hogares for Bolivia Encuesta Nacional de Hogares for Peru All these surveys are nationwide. n Why these four countries? Only in these countries’ surveys it is clearly specified who the mothers of the kids living in the household are. There is also information about the fertility history of women 15 to 49 (or 50) years old as well as other socioeconomic characteristics.

Sample restrictions –Labor Force Participation analysis: women 14 to 45 years old. The geographical coverage is restricted to urban population only. –Wage regression analysis: we restrict the sample to those earning a salary at their job and those working more than 1 hour per day but not more than 16 hours per day. Self-employed workers are not included for this analysis given the difficulties in separating returns to labor and capital.

Labor Force Participation (LFP) Analysis n The approach of this section is based on a descriptive set of labor force participation rates and total hours worked, divided by certain women characteristics (marital status, age groups, etc). n Why do we have this approach and not a regression analysis? Skepticism regarding the causal interpretation of associations between fertility and labor supply arises in part from the fact that there are strong theoretical reasons to believe that fertility and labor supply are jointly determined (endogeneity problem).

LFP by age group, marital and motherhood status

LFP by age group for Brazil and motherhood status (no information about marital status)

LFP by age group and number of children

LFP by Occupational Category and Motherhood Status

Self-employment by Motherhood Status

Hours worked by occupational category and motherhood

LFP by Public Sector and Motherhood Status

The effect of motherhood on wages –First specification : lnW i =  1 +  1 (mother children less than 7) i +  2 (mother children between 7 and 12) i +  3 (mother children between 13 and 18) i +  1 X i +  i –Second specification: lnW i =  1 +  1 (one child) i +  1 (two or more children) i +  1 X i +  i –Third specification: lnW i =  2 +  1 (number of children 0 to 6) i +  2 (number of children 7 to 12) i +  3 (number of children 13 to 18) i +  2 X i +  i

Model n The dependent variable is the natural logarithm of the hourly wage in the respondent’s current job. n For this presentation, we are going to show the results for a model that controls for age, age squared, level of education (in years), level of education squared, tenure (in months), tenure squared, marital status, head of household status, regional location, ethnicity, part-time status, public sector status, industry dummy variables and occupational dummy variables.

Model: caveat n Lack of information about actual work experience. The only variable available is tenure in the current job. n Anderson, Binder and Krause (2003) show that the gap between potential experience (age – schooling – 6) and actual work experience is three years for mothers compared with only 1-1/2 months for non- mothers, using a panel data set for the US. n Problem for this paper: the lack of information on actual work experience can create a bias in our motherhood coefficients. Experiment for Brazil.

Descriptive Statistics n Mothers are older than non-mothers, with differences of 7 to 8 years. n The number of children aged less than 18 years old is 1.8 to 2.2 children. n The average hourly wage is always higher for mothers. n Regarding levels of education, we did not find statistically significant differences among mothers and non-mothers with the exception of Brazil, where mothers tend to be less educated than mothers in average. n Mothers are more likely to be married, as we might have expected.

Descriptive Statistics n Potential experience and tenure are always smaller for non-mothers, basically because of the age difference among mothers and non-mothers. n Mothers are more likely to work in the public sector. n Mothers are more likely to be heads of household for Bolivia and Ecuador with around 17% of mothers being head of household, but not for Brazil, where only 2% are heads of household.

Regression Results

By age groups n Age groups: 14-25, 26-35, n Bolivia: youngest group of mothers has a penalty for having children less than 6 years old in terms of wages (38.6% smaller wages than non-mothers) n Ecuador: positive effect can be seen now for the oldest group of mothers if they have children 7 to 12 years old (the hourly wage differential is 28.6% more than non-mothers). n Peru: penalties appear in the middle-aged group (18.5% smaller wages, almost double of the penalty found for the pooled sample). n Brazil: significant effects disappear for the oldest group of mothers

By educational levels n Educational levels: less than High School, High School or more. n Bolivia: premium for having children 13 to 18 years old only for educated mothers. This premium is equal to 26.4% more for mothers than non-mothers n Ecuador: for the less educated group of mothers, there is a penalty of 15.1% for having children less than 6 years old as well as a premium of 17.9% for having children 7 to 12 years old n Peru: having children less than 6 years old has a negative impact on wages (hourly wage differential of 11% less for mothers) if you have High School or more n Brazil: similar results for both groups

By public or private sector n Bolivia: for those working in the public sector there is a 15.4% premium for those who have children 13-18, whereas there is a penalty of 11.5% for having children 7-12 years old n Ecuador: no significant results n Peru: negative impact on wages for having two or more children for private sector workers. Having more children less than 6 years old also shows a negative effect on wages for those who work in the private sector n Brazil: similar results for both groups

Conclusions n Studies in developed countries regularly observe a wage penalty for working mothers. n In this paper we explore the effects of motherhood on wages and labor force participation for four Latin American countries. n LFP: The results of this paper show that mothers with children less than 6 years old participate less than those with no children, except for single mothers. Mothers are over-represented among the self-employed and work fewer hours than non- mothers if employed in the formal sector.

Conclusions n WAGES: Conversely from the evidence found in the US, UK, Australia and Germany, our results for Latin America do not show a clear impact of being a mother on wages. While for Peru there exists a penalty for mothers of children less than 6 years old, for Bolivia and Brazil we find premiums for being a mother. For Ecuador there are no significant effects. n These very heterogeneous effects are further investigated looking at samples divided by public and private sector, by educational level and by age groups. We find that wage penalties and premiums are not borne equally among all mothers.

Sample restrictions