January 11, 2008, Emanuela Galasso & Nithin Umapathi
|A recent impact evaluation study of a community based nutrition program in Madagascar shows that malnutrition can be improved over the short- and long-term when mothers participate in community health programs that promote behavioral change in nutrition, feeding, and hygiene practices. The study highlights important complementarities between maternal education, knowledge, and community infrastructure to achieve improvements in children’s nutritional status.|
Nearly one-third of children in the developing world are either underweight or stunted. Child nutrition has direct, short-term impacts on health outcomes and important long-term cognitive functions, school performance, and productivity consequences later in life.1 Research has shown that economic growth alone (by reducing poverty over the long run) is not sufficient to achieve desired reductions in malnutrition rates.2
Understanding direct nutrition interventions and how they work best under different conditions has great policy relevance for improving immediate health outcomes and stemming the intergenerational transmission of poverty.3
Direct nutrition interventions are generally based on a heuristic model of the production of nutrition based on three pillars: nutrients (through food or supplementation), health and sanitation services (to protect from diseases), and the role of child care.
Evidence from programs that focus on child care and promoting behavioral change in child care practices is relatively scant
Policy research working paper 4424 (December 2007) provides one of the few rigorous ex post evaluations on the effectiveness of community based nutrition programs.4 The program in Madagascar (SEECALINE) targets children from birth to three years of age and pregnant/lactating women.
The preventive approach involves a monthly growth-monitoring and promotion activity facilitated by a local nutrition worker. Adoption of improved hygiene, child care, and nutrition practices is promoted through a community based approach of communication with locally adapted messages.
To identify the causal effect of the program, the study design addresses the non-random selection of communities into the program
The selection process takes into account the purposive targeting of the program to the most malnourished districts of the country, as well as the potential differences in unobserved characteristics (such as motivation) arising from the decision of the communities to participate in the program.
The evaluation design builds on two nationally representative surveys administered before and after the program over seven years, which are longitudinal at the community level and done in close collaboration with the Madagascar National Statistical Institute.
The longitudinal community level design allows the authors to difference out any time-invariant component of the selection bias and use propensity-score weighted regressions to balance the observable characteristics between sampled program and nonprogram villages.
The program targeted the most malnourished and vulnerable districts of the country
The figure to the right shows that communities participating in the program bridged the weight gap with respect to non-participating communities in terms of short-term nutritional outcomes (weight for age z-scores reduced by 0.15-0.22 standard deviations and the incidence of underweight children reduced by 5.2-7.5 percentage points).
The overall share of observed improvements in nutrition nationwide (reduced malnutrition rates) between 1997/98 and 2004 attributable to the program (using the estimates above) ranges between ¼ and ⅓.
The figure on height shows the protective effect of the program on longer-term nutritional outcomes (reversing the trend in stunting). This result is particularly important in light of the fact that program communities had a higher incidence of shocks and experienced higher food insecurity.
The observed program effects were obtained through significant improvements in feeding practices (exclusive breastfeeding, timing of weaning, and proactive and responsive feeding) and hygiene practices (appropriate disposal of garbage and toilet use, and improved methods of water purification).
More educated mothers and households living in better-off areas gain more from the program
The results reveal important socioeconomic gradients—characteristics of the household as well as conditions of the living environment—that affect the uptake of improvements in practices and nutritional outcomes.
Worse-off households are more likely to have gained from adopting beneficial child care practices, but have greater difficulty translating new practices into improved nutritional outcomes.
Nutritional gains are larger for households in relatively less poor and remote areas with relatively better infrastructure (access to electricity, secondary schools, and hospitals).
Knowledge is a necessary but not sufficient condition for improving nutritional outcomes for households with limited access to complementary resources
The results highlight important policy trade-offs between targeting areas with higher needs—where the initial stock of malnutrition is higher—and targeting areas with the highest expected impact.
Direct nutrition intervention bundled with complementary interventions in other sectors (health, infrastructure, and water and sanitation) is the best way to maximize effectiveness in the neediest areas.
1. Alderman Harold, John Hoddinott, and Bill Kinsey. 2006. “Long Term Consequences Of Early Childhood Malnutrition.” Oxford Economic Papers 58: 450–74.
2. Alderman Harold, Jere Behrman, and John Hoddinott. 2004. "Malnutrition and Hunger." In Global Crises, Global Solutions, ed. Bjorn Lomborg. Cambridge (UK): Cambridge University Press.
3. World Bank. 2006. Repositioning Nutrition as Central for Development. A Strategy for Large-Scale Action. Washington, DC: World Bank. Download book| Order book
4. Galasso, Emanuela, and Nithin Umapathi. 2007. “Improving Nutritional Status through Behavioral Change: Lessons from Madagascar.” Policy Research Working Paper 4424, World Bank, Washington, DC.
Development Impact Evaluation Initiative (website
Bank Conference “Making Smart Policy: Using Impact Evaluation for Policy Making,” January 15-16, 2008 (website
EMANUELA GALASSO is an economist in the Development Research Group. She serves as the evaluation coordinator for the Poverty Research Team. Her ongoing work focuses on the medium term and dynamic effects of anti-poverty programs using multi-year longitudinal surveys in Madagascar and Chile, with particular emphasis on their distributional impacts. Email: Egalasso@worldbank.org.
NITHIN UMAPATHI is a scholar at the Centre for Microdata Methods and Practice at University College of London. His research interests include development economics and the econometrics of program evaluation. Email: firstname.lastname@example.org.