![Sample our Economics, Finance,Business & Industry journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5931/TOC-Subject-Sample-2021-Economics-Finance-Business-Industry.jpg"})
ABSTRACT
Early childhood interventions are high on the social policy agenda. Yet their impact on disciplining practices is scarce. This paper examines how early childhood interventions affect disciplining methods in Colombia, where poor households are eligible for different social programmes based on a proxy means index. Using a regression discontinuity design, I found that benefiting on a larger extent of childcare, nutritional programmes and health checks, decreases parents’ use of physical disciplining. Using a different data set and through a propensity score matching, I show that mothers of children exposed longer to a child care service use more often non-physical methods.
Acknowledgments
I’m grateful to Karen Macours for her thorough guidance in this work. This work received the support of the French National Research Agency, through the program Investissements d’Avenir, ANR-10–LABX_93-01 to visit some Hogares Comunitarios in Colombia in December 2013. I’m grateful to Maria Consuelo Gaitan Clavijo at Instituto Colombiano de Bienestar Familiar (ICBF) for her help contacting the regional offices of ICBF to perform the visits, as well as the people who helped in the regional offices of Cundinamarca and Quindío. I also want to thank Francisco Espinosa at Departamento Nacional de Planeación (DNP) and Raquel Bernal at Universidad de Los Andes who facilitated me the data to make this work possible. I also thank Carlos Medina at Banco de la República in Medellín, Lina Sanchez and Maritza Lopez for their help. I also thank my thesis committee for all useful comments received, I thank Sylvie Lambert, Raquel Bernal and Orazio Attanasio. Any errors are mine.
Disclosure statement
No potential conflict of interest was reported by the author.
Notes
1. Proyecto Integral de Desarrollo Infantil.
2. Home Observation for Measurement of the Environment scale.
3. Those who belonged to Sisben levels 1 and 2.
4. A Conditional Cash Transfer programme implemented since 1992 targeted to population in Sisben level 1.
5. Every check last approximately 1 h.
6. The protocol that should be followed in the G&D check is available upon request.
7. The target population of G&D checks is a quite vulnerable population. From qualitative interviews with people in charge of the G&D checks, it seems that mothers see a lot of advantages in taking their children to the checks, even if they may receive a sanction in case the child has been physically punished. Also, it may be that in few cases, mothers do not dare to report that their partners punish their children physically, and prefer for doctors to find out.
8. The ICBF encourages only hosting children older than 2 y.o. given that younger children require special care, when a madre comunitaria receives very young chuldren, she should not receive more than 1. HCB usually host children from 8am to 3pm.
9. All Madres Comunitarias are required to attend a 40-hours pre-service training on child development, health, nutrition, safety, and community participation (Bernal and Fernandez Citation2013). Studies have shown that there is great variation on the quality provided by the HCBs (Bernal et al. Citation2009), (DNP Citation2009).
10. Madres Comunitarias encourage both parents to participate in these meetings, though is mostly mothers the ones who attend. Yet, through the qualitative work, fathers reported that their partners (ie. mothers of the children) used to share the information received during the meeting.
11. Families registered in the Red Unidos correspond to families in the lowest quintile of families belonging to Sisben level 1.
12. From 2011–2012 programmes began to be targeted with the newest methodology of the Sisben, Methodology III. Now, thresholds for programme eligibility change from programme to programme (ICBF Citation2013).
13. Based on assessment of the outside characteristics of neighbourhoods and dwellings.
14. Methodology I from 1995–2002, methodology II from 2003–2010 and methodology III from 2011 on.
15. The official formula and weights are not publicly available.
16. As nearly 80% of the population falls in the urban classification, there are not enough observations to estimate an impact for rural ones.
17. All questions are answered by the mother.
18. As the DHS05 is not used for targeting purposes, we expect the information reported in the DHS to be more accurate as people don’t have incentives to under report to be classified as eligible for a program.
19. Household composition by age group (0–6 y.o., 7–14 y.o., 15–24 y.o., 25–64 y.o. and 65 y.o. and above), woman’s and partner’s education and age; and the household wealth index.
20. I also run the estimations on the rank [19–25], which provided similar results. Similarly, estimations with a linear polynomial were also done. Results are available upon request.
21. shows the optimal number of G&D checks children must attend according to their age, on average, children up to 4 y.o should attend 3 checks during a year.
22. I focus on children 0–4 y.o because for them I have information on both programmes.
23. Also, mothers may not distinguish between the different childcare options, such as HCB and a public kinder garden. Hence, the answers may not be completely accurate.
24. Given that no particular difference arises by whether or not the father is present at home, all the subsequent analysis will focus on all type of households.
25. In 46% households have at least 1 member 0–4 y.o.
26. Around 40% of women work outside the household. There is no discontinuity on the likelihood that mothers work outside the household at the threshold of interest.
27. .
28. Basically, belonging to Sisben levels 1 and 2.
29. Data from the HCB only provides information on whether the mother works.
30. The possible answers are never (1), sometimes (2), almost every time (3) and always (4).
31. To estimate the factor, I use a principal-component factor method with quartimin rotation as we search for dedicates measures for each factor. The aim of the quartimin rotation is to re-weight the factor loadings obtained from the explanatory factor analysis in a way that leads to a structure of factor loadings such that measures only heavily load on one factor. If a measure does not load heavily on a factor or if it’s not clearly related to only one factor, we exclude it from our measurement system, following (Attanasio et al. Citation2015).
32. For each of the behaviours, mothers were asked to classify children’s behaviour into always, sometimes and never.
33. p(x; γ) = exp(γ0 + xjγ1)/(1 + exp(γ0 + xjγ1)).
34. The graphs and balance tests for the other sub-samples of interest are available upon request.
35. Caliper=0.2.
36. The common support is defined by dropping treatment observations whose pscore is higher than the maximum or less than the minimum pscore of the controls.
37. Following (Imbens Citation2004).
38. (Xi) includes all the covariates used to estimate the propensity score, one these covariates is the age of the child and its interactions with other variables such as parents’ age and education.
39. This data set only contains information on whether the mother works, not on whether the mothers works outside or inside the HH. 51% of mothers report to work and 49% do not.
40. The survey includes a question on for attendance to the meetings, but only asks for attendance over the last 6 months. Hence, mothers of children who have been exposed less time to the programme, have had less chances to attend the meetings.
Additional information
Funding
Notes on contributors
Diana Lopez-Avila
Diana Lopez-Avila is an assistant professor at the Department of Economics of Universidad Javeriana. She has previously worked at the National Planning Department of Colombia and the International Initiative for Impact Evaluation. She holds a PhD in Economics from Paris School of Economics.