Using food consumption data to assess nutrition and health impacts of interventions: An example from Malawi Matin Qaim International Food Economics Georg-August-University of Göttingen LCIRAH Workshop on "Measuring Effects of Agri-Health Interventions", 12-13 May 2011, London
Billion people Global prevalence of malnutrition 7 6 5 4 3 2 1 0 Sufficient Deficient Calories Iron Zinc Iodine Vitamin A Sources: FAO (2010), WHO (2010). 2
What type of interventions are likely to have nutritional effects? Interventions directly targeted at food consumption Food price subsidies / food transfers Promotion of kitchen gardens Interventions where changes in food consumption occur as an indirect effect Any policy that changes agricultural productivity (e.g., adoption of innovations) Any policy that changes prices or household incomes (e.g., infrastructure, employment) 3
Evaluation approach Ex post evaluation of interventions Compile consumption data from treatment/control groups. Use food composition tables to convert food consumption to calorie and nutrient consumption. Compare actually consumed with required/recommended amounts for particular nutrients to establish deficiencies. Estimate net treatment effect of intervention on nutrient consumption (or deficiency prevalence) through statistical approach, controlling for confounding factors. Ex ante evaluation of interventions Use representative household level data and estimate elasticities of calorie and nutrient consumption. Simulate consumption effects of the policy/intervention. 4
Types of food consumption data Household surveys sometimes include a module on household level food expenditure/consumption, which is captured for a certain recall period (e.g., 7, 14, or 30 days). The shorter the recall period, the better for nutritional analysis (data accuracy). The more disaggregated this consumption recall, the better for nutritional analysis (100+ different food items). Many living standard monitoring surveys (LSMS) have a food consumption module (advantage of nationally representative data). Use of food consumption data from household surveys has drawbacks (intra-household distribution, seasonality, issues of bioavailability etc.), but data are rarely perfect. PAS Study Week 2009 5
Example from Malawi Question: What are the impacts of food price and income related policies on undernutrition and micronutrient deficiency? We need to estimate price and income elasticities of calorie and micronutrient (MN) consumption. Building on available literature on calories (Behrman/ Deolalikar 1987, Bouis et al. 1992, Subramanian/ Deaton 1996) we developed a method to estimate MN elasticities. Estimation of three-stage QUAIDS for highly disaggregated food groups and calculation of elasticities (with correction for censoring). Details described in Ecker/Qaim (World Dev. 2011) 6
Data Malawi Poverty rate: 52% Net buyers of staple foods: 78% Average food budget share: 73% Second Integrated Household Budget Survey (IHS), 2004/05 11,280 households (representative at national and district level) Food consumption module: 7-day recall (>100 food categories) Food composition tables from FAO and WHO 7
Prevalence of malnutrition in Malawi Vitamin C Vitamin B12 Folate Vitamin B6 Niacin Riboflavin Thiamin Vitamin A Zinc Iron Protein Calories 0 10 20 30 40 50 60 70 80 90 Percent of households 8
Relative contribution to nutrient supply 100% Other Animal products Vegetables & fruits Pulses Staple foods 80% 60% 40% 20% 0% Calories Protein Iron Zinc Vit A Folate 9
Nutrient consumption elasticities Income elasticity Price elasticity Maize Beans Leafy veg. Fish Calories 0.92-0.17-0.01 0.07 0.08 Protein 0.95-0.02-0.08 0.07-0.02 Iron 0.87-0.19-0.09 0.06 0.08 Zinc 0.91-0.16-0.06 0.07 0.07 Vit. A 0.36 0.32 0.02-0.43 0.07 Folate 0.79 0.15-0.25-0.04 0.10 For reasons of space, not all estimated elasticities are shown. 10
Percentage change Policy simulation Change in the prevalence of deficiency 15 Calories Protein Iron Zinc Vitamin A Folate 10 5 0-5 -10-15 -20-25 -30 50% maize price subsidy Income transfer of equal value 11
Possible other uses and extensions From a food security perspective, we need to better understand the nutrition implications of policies and shocks (ad-hoc statements are not sufficiently reliable) This requires suitable data and methodologies Proposed framework can be used for: Ex ante simulation (e.g. introduction of new technology in vegetables, input subsidies, price spikes) Ex post evaluation (e.g., supermarket revolution, oil palm boom, technology adoption) PAS Study Week 2009 12
Assessing health effects Malnutrition (nutrient deficiencies) entails adverse health outcomes, causing a health burden for individuals & society. The DALYs approach (disability-adjusted life years) can measure health burden by combining mortality and morbidity within a single index (Murray/Lopez 1996, Stein/Qaim 2007): DALYs Lost = Years lost to mortality + Years with disability x Disability weight DALYs Lost Health benefit of intervention Without intervention With intervention EPSO Conference 2008 13