You are on page 1of 1

Oportunidades, a move from underweight to overweight?

Mabel A. Andalón L.
Introduction Cornell University, Department of Policy Analysis and Management Methods (IV-Probit) and Results
• 7% of the Mexican children are underweight and 20% are obese.
1st Stage: Using people around the threshold (N=1276), I estimate
Policies should not uniformly decrease food intake and/or Identification: RDD P( pi  1 / DSR )  g ( DS * State  i ) p : participat ion
increase physical activity.
• However, Oportunidades -a Federal program in Mexico- Oportunidades benefits poor households in highly Results: Wald chi2(7)= 282.81 P value = 0.0000
provides income, nutritional supplements, and health education marginalized communities since 1997
to all the poor. 2nd Stage: With pˆ  0.25 & pˆ  0.75 , I estimate
Measuring the effect of the program is possible because:
a) participation is independent of unobservables… P(Y  1)  f ( X i   pˆ i   i )
Objectives
… because it was determined by a regional discrimininant score (DS), Descriptive Statistics Marginal Effects
• Evaluate the overall impact of the Oportunidades program on Short Run Variables mean std. dev under over
which assessed whether the household was poor
underweight and overweight children (2 to 6 years of age)
b) comparison groups are very similar in the absence of the program… oportunidades 0.49a 0.140 -0.002 -0.292 **
in Mexico’s rural areas.

child
…because I only use individuals within a very small range around the cut-off gender (1=female) 48.16 0.50 0.013 -0.012
• Neufield et al (2005) do this using Propensity Score Matching Poor Non-poor
(PSM). Assumptions: age (months) 49.53 13.43 0.000 0.001
1. People to the left or to the right of the age 43.03 14.58 0.001 0.000
threshold are identical gender (1= female) 6.74 0.25 0.138 0.363 **

head of hh
Why is this important? Randomization around the cut-off indigenous (1=yes) 40.43 0.49 -0.008 -0.124 ***
• Unhealthy weight is associated with mortality. 2. No spillovers middle school 10.99 0.31 0.087 0.057
• Underweight children have difficulty acquiring and Local Average Treatment Effect (LATE) divorced 7.09 0.26
developing human capital. Still good as there are many unemployed 11.35 0.32 0.039 0.067
• Some overweight children suffer from discrimination at school, people around cut-off earnings (pesos) 186.03 526.48 -0.0002 *** 0.0000
depression, and low self-esteem. Wald 15.34 * 22.340 ***
• Overweight children are more likely to remain overweight in
adulthood and develop serious illnesses. Data Long Run Variables mean std. dev under over
• The program was implemented in phases: some marginalized communities started oportunidades 0.51a 0.14 0.080 -0.181

child
receiving the benefits in 1997 and others in 2000. Information on weight and height is gender (1=female) 47.02 0.50 -0.031 -0.057
Framework and Hypothesis Cawley (2004) only available in the “Household Evaluation Survey, ENCEL 2003”. age (months) 54.18 10.48 0.000 0.006
• The groups surveyed are poor and non-poor living in communities treated since 1997,
Max U [ Z, F , Y ; W (Z, F); H ( Z, F , Y , W ) ] (1) age 42.91 13.73 -0.002 -0.001
in communities treated since 2000, and in non-treated communities (control).
gender (1= female) 6.39 0.24 0.001 0.036
sleeping (S)

head of hh
F: Food indigenous (1=yes) 41.53 0.49 -0.008 0.010
leisure (L) Type of evaluation Treat 00 & Treat 97 ¥ Control
Y: All other goods middle school 12.14 0.33 -0.045 -0.044
Z : occupation (O) Poor Non-poor Poor Non-poor Poor Non-poor
Households

(time) transportation (T) W: Weight Exposure (years) 3 0 6 0 0 0 divorced 6.07 0.24 -0.003 -0.068
H: Health Short Run Vs. No Part.
home prod. (Ho) G: Genes
*ª * ª unemployed 11.18 0.32 0.105 -0.082
Budget (2) Time (3) Long Run Vs. No Part. *ª * ª earnings (pesos) 248.00 607.88 0.000 0.000
Constraints

Y + FPF =w * O + I S  L  O  T  H  24 Short Run Vs. Long Run *ª * *ª * Wald 5.18 9.53


a'
* This paper a Neufield et al (2005) & 2-6 years of age ¥ 3-6 years of age ***' sign. at 1% '**'sign. at 5% '*'sign. at 10% predicted probab. from 1st stage
Biologic Constraint (4)
Neufield et al (2005) match poor individuals based on characteristics of 1997. For the
W  c( F )  f (S , L, O, T , H , G)  S (G) control group these were gathered in 2003. When using RDD individuals are drawn
Conclusions
from the same local market and there is no need to use the control communities (i.e. to • There is no effect of the Oportunidades program in underweight either in
Food intake Physical activity Metabolic rate
use noisy retrospective information) the short or in the long run. In the short run, labor earnings slightly decrease
Benefits Recipient Impact on? Δ weight the probability of being underweight. Oportunidades decreases the
Underweight (%) Overweight (%)
The Z-core probability of being overweight among young children by 22% in the short
17 USD Mother (2) -/+ N Non-poor Poor Non-poor Poor
determines run. This effect disappears in the long run.
Preventive All (1) -/+ underweight/ Short Run 299 6.88 7.19 13.75 9.35 • In the long run none of the characteristics of the child or the head of the
health overweight Long Run 336 7.59 10.11 20.25 14.61 household play any role in determining weight for height.
Nutritional 0-5 children (1) or (4) + • The results are similar to those found by Neufield et al (2005). Thus, in
supplements & Mothers evaluating impacts on weight, RDD performs as good as the PSM method.
Acknowledgements
• Perhaps the money devoted to interviewing individuals in “control
Prenatal care Mothers (4) + I am grateful to my PAM 691 classmates and to John Cawley for helpful comments in communities” should have a better use if it is directed to the poor.
Overall effect …… + or -? different stages of this research.

You might also like