The Effect of Compulsory Preschool Education on Maternal Labour Supply

Abstract

The Brazilian government implemented the school reform that mandated preschool education compulsory in 2009. I exploit discontinuities in the preschool eligibility rules to examine the effects of preschool enrolment on mothers’ labour market outcomes. I found that preschool enrolment significantly increased the time spent working rather than performing household chores among mothers living without additional younger children and other relatives. Moreover, these mothers were more likely to take the jobs that guaranteed employee rights and benefits. However, I found no effects among the mothers who have an additional younger child and/or live with other relatives. Taken together, the findings reveal a potential benefit of lowering school entry age in a context where limited availability of early childhood education constrains mothers’ labour market activities.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. For example, Mexico and Brazil announced the policy to lower the school entry age to 4 years old in 2004 and 2008, respectively, and France recently lowered the school starting age to 3 years old in 2018.

2. Pesquisa Nacional por Amostra de Domicílios (PNAD) in 2015.

3. Among the studies introduced in the previous paragraph, Berlinski and Galiani (2007), Berlinski et al. (2011), Attanasio et al. (2017) are the only ones that evaluated the effect of children’s school enrolment on maternal labour outcomes in a developing country. Attanasio et al. (2017) are more closely related to this study as they examined the effect of access to free daycare on maternal employment in Rio de Janeiro, Brazil. While their study evaluated the lottery program for daycare access in one state of Brazil, the current study evaluated the school reform and included more than half of the Brazilian states.

4. According to 2013 school census data, approximately 85 per cent of preschools had equal to or less than 5 hours of instruction per day.

5. Tocantins is excluded because it allowed unrestricted access to preschool education regardless of the age at the cut-off date. Also, Rio de Janeiro, Sergipe, Paraiba, Santa Catarina and Paraná are not included in the current study because the information on preschool cut-off date was not available. Most of these states enforced Dec 31 cut-off date and often made enrolment decision based on students’ readiness. Lastly, Rondonia, Mato Grosso, and Mato Grosso do Sul enforced March 31 cut-off date until 2013, so I include only earlier observations for these states.

6. PNAD data are collected every September of the year. Therefore, 4-year-old students born after March 31 are not eligible for preschool entry even though they turned 4 by the time of survey.

7. The state government of Sao Paulo chose June 30, and the state government of Minas Gerais changed its cut-off date from March 31 to June 30 in 2013. I included these states and rescaled the running variable between students’ birth dates and the cut-off date to make them compatible with those of other states.

8. The impact of preschool eligibility on public/private school enrolment is examined in Figure A2(b).

9. Mothers who were public employees, self-employed, and employers were separately analysed in Figure A2(a), but the school eligibility rule did not make significant changes in these statuses. Self-employed status was significant with the use of second polynomial degrees, but the result was not robust to the use of different polynomial degrees. I did not analyse mothers who worked to produce goods for their own consumption, worked as construction workers to repair their own house, and had unpaid jobs because these mothers were not paid. Lastly, mothers who worked in the military were not analysed as they represent less than 0.1 per cent of the entire sample.

10. Discontinuities of the other occupations in a reference week were examined in the Appendix Figure A2(a).

11. The results from the entire sample and the subsample divided based on only one criterion (presence of an additional younger child or relatives) are reported in the Tables A1 and A2. The cut-off dates still make a significant effect on preschool enrolment regardless of how I define control groups, but the reduced form results were weaker or significant only for a few of outcome variables. This analysis indicates that both a younger child and relatives play important roles on the effect of preschool enrolment.

12. Despite these possibilities, the estimates in the first-stage based on survey year did not show much difference. Among the families where four-year-old children are the youngest and no other relatives present, the preschool eligibility increased school enrolment about 7 per cent point before 2014, and 8 per cent from 2014. Both coefficients were statistically significant.

13. I also used the data from the years previous to reform (2007 to 2009) and ran the placebo test. As expected, there is no significant effect of the cut-off dates on preschool enrolment and maternal labour market outcomes. The results are reported in Tables A3. Average preschool enrolment rate in 2007-2009 is reported higher compared to the recent years, 2011-2015, but this was because households mostly did not differentiate preschool and daycare enrolment. Overall school enrolment rate (daycare + preschool) was higher in recent years, but in previous years daycare enrolment rate was only reported as 5 per cent point, which was certainly underestimated.