An Evaluation of the Impact of the Oportunidades Program on Child's Linear Growth: A Regression Discontinuity Design
Posted: 21 Jun 2007
Date Written: June 2007
The purpose of this paper is to examine the causal impact of a social program called Oportunidades on the nutritional status of preschool children that live in rural areas of Mexico. Oportunidades is a federal program whose aim is to increase the human capital of the poor through investments in health, education and nutrition. Program benefits include cash transfers that are disbursed conditional on the household engaging in a set of behaviors: (i) every family member accepts preventive health services; (ii) children 0-5 years, pregnant and lactating women obtain food supplements; and (iii) every month an adult attends community meetings where education on nutrition is given. Hence, there are economic and clinical reasons to expect that these benefits might have contributed to increasing child's linear growth (measured by Z-scores of height-for-age) among the beneficiary population. Previous work on the effectiveness of this program (Neufeld et al., 2005) compares poor children of treatment and control communities in 2003. To control for the fact that the control group was not generated through a random assignment, it applies the methodology of Propensity Score Matching (PSM) at a household level using retrospective information of 1997. This paper exploits the discontinuity in the rule that determined eligibility for the program based on poverty status. The rule was a discriminant score (DS), computed for each household before the program was implemented, which represents parsimoniously the differences between the poor and non-poor households. Using the ENCEL 2003, I compare children aged 2 to 6 living in rural households who score just below a preset DS and are thus eligible to participate in the program to those children scoring just above the cutoff. Children above and below the cutoff are shown to be nearly identical in terms of all other factors that might affect growth, suggesting that differences in their subsequent mean Z-scores might be causally attributed to participation in the Program. Preliminary results indicate that, after two years of exposure to the program, children aged 2-6 years who were eligible for the program are taller than the non-eligible. The effect size estimates are even bigger for children 24-30 months. These results are two times bigger than those found by Neufeld et al. (2005) but are closer to those found in other Latin American countries where similar programs have been implemented. The methodology of the Regression Discontinuity Design (RDD) used in this paper improves upon previous studies in the following ways. First, children above the cutoff (the control group in this study) is drawn from the same "market" as children below the cutoff so that local area effects can be controlled for. Second, avoiding the use of (constructed) control communities implies that: i) no (potentially noisy) retrospective information is used; and b) if more credible research is done using RDD, future evaluations of the program might cut costs by eliminating the design of new controls and the gathering of information in these communities.
Keywords: Regression Discontinuity Design, Mexico, Oportunidades, Nutrition
JEL Classification: I12
Suggested Citation: Suggested Citation