! ! additional 1,500 urban households; the panel sample includes rural and small town households only. Dynamics of wasting and The ESS uses a stratified, two-stage sampling scheme. underweight in Ethiopian Regions of Ethiopia served as the strata and children enumeration areas (EAs) were randomly selected in proportion to population size. A total of 290 and 43 Underweight prevalence among children 6-59 months EAs were selected from rural and small town areas, in rural and small town areas of Ethiopia declined from respectively, and twelve households were then chosen 26.9 percent in 2012 to 24.9 percent in 2014, while from each EA. Tracking between waves was done at wasting prevalence stalled at 11 percent. Using panel the household level-- with a low attrition rate of 4.9%- data from the Ethiopia Socioeconomic Survey (ESS), - leading to a final panel sample of 3,776 households. we perform cross-sectional and fixed effects analyses to uncover the factors that drive children in and out of Methods!! each state and how they compare to static correlates at Our analysis was restricted to children who were 6-59 one point in time.! months during both waves of the ESS, with the main outcomes of interest as weight-for-height and weight- Background!! for-age z-scores, and binary indicators for whether the Undernutrition in all forms is responsible for 51% of child is wasted or underweight. Independent variables childhood deaths in Ethiopia. Wasting, defined by a are drawn from individual, household, and community weight-for height z-score below -2, is a measure of level characteristics found to be associated with each acute and severe malnutrition, while underweight, outcome in the literature. defined by a weight-for-age z-score below -2, is a Cross-sectional analysis using Ordinary Least Squares broader measure of malnutrition in children 6-59 (OLS) regression within each round of the data helps months. Both forms are associated with negative us understand the correlates of wasting and health, development, and long-term outcomes. While underweight within a static framework and compare many studies have looked at correlates of stunting and our findings to existing literature. Next, we exploit the underweight in Ethiopia using cross-sectional analysis, panel setup of the ESS to estimate fixed effect models few have used panel data to identify drivers of changes, for changes in each outcome, focusing on the sub-set particularly with respect to wasting. of individuals that were 6-41 months at baseline and 24-59 months at follow-up (n=1,048). The fixed Data-- effects model improves upon the cross-sectional We analyze data from two waves of the ESS, a analysis by allowing us to control for all time invariant collaborative project between the Central Statistics characteristics that influence our explanatory variables Agency of Ethiopia (CSA) and the World Bank’s in addition to wasting and underweight. We also Living Standards Measurement Study- Integrated examine how dynamics in and out of each state vary in Surveys of Agriculture (LSMS-ISA) project that a further specification of the fixed effects model that collects multi-topic panel data at the household level. controls for baseline wasting and underweight status. The ESS began in 2011 as the Ethiopia Rural Socioeconomic Survey (ESS1), during which 3,969 Results- rural and small town households were surveyed. In Table 1 displays the movement in and out of wasting 2013, a second wave of the survey (ESS2) was between 2012 and 2014, with 13% of children administered, revisiting the ESS1 households and an experiencing an improvement and only 2% remaining perpetually wasted. Similarly, more children recovered from being underweight (16%) than became Discussion- &- Policy- Implications-! underweight (11%), but the fact that 128 children Overall, our results suggest childhood nutrition is (12%) remained underweight from baseline to follow improving in rural Ethiopia, though further research is up emphasizes the need to better understand what warranted on a number of topics. Foremost, responses drives changes out of undernourishment. to illness in young children require investigation to determine what actions are taken at households and Table 1. Dynamics of Wasting health centers, and how these responses relate to the Always Became Became Never negative nutritional outcomes we observed. Similarly, Total more research is needed on the differences between wasted wasted un-wasted wasted male and female-headed households with regard to 1.92% 7.07% 13.02% 77.99% 100% children’s nutrition, as outcomes were significantly improved in the latter. The combination of our n=18 n=66 n=121 n=725 n=929 findings and those obtained from the proposed research could then inform nutrition-related guidelines Our cross-sectional findings generally agree with the for health centers, community health workers, and existing literature, particularly other studies using data households. Community-level main road access is also from Ethiopia. Male children, those with illiterate worth further investigation given our findings that it mothers, male household heads, and older household was associated with improved nutritional outcomes. It heads, and those experiencing illness in the last two is likely that increased accessibility also improves months were significantly more likely to have negative economic outcomes in rural communities and would nutrition outcomes (lower z-scores or higher therefore further Ethiopia’s progress on related likelihood of wasting/underweight). Futhermore, national goals. having a solid roof, improved toilet, radio, cellphone, Finally, our findings support differential policy female cow, and laying hen were repeatedly approaches for aiming to prevent vs. treat acute significantly associated with positive outcomes. We undernutrition. Children undernourished at baseline also observe some departures from the existing were more sensitive to household level changes such literature that are worth noting; we find better weight- as asset ownership and food aid receipt than those who for-age z-scores and underweight outcomes for female were adequately nourished. Therefore, while new children and those in female-headed households, efforts to improve living standards may simultaneously unlike regional studies in Ethiopia. lift undernourished children to healthier states, After controlling for individual fixed effects, illness in emphasis should also be placed on preventing the last two months remained significantly associated undernutrition altogether, thereby imbuing children with changes in both z-scores and underweight status, with greater resiliency to external changes. increasing negative outcomes for each. Additionally, community level main road access, which was not significant in our cross-sectional models, was Generous funding assistance for this research came from associated with positive changes in z-score. When also the UK Department for International Development controlling for baseline status, we observed that factors Ethiopia. driving changes to or from undernourished states vary, The findings outlined in this brief are drawn from: and children wasted at baseline were generally more Cintron, C. (forthcoming) “Dynamics of wasting and responsive to household level changes than non- underweight in Ethiopian children” wasted children. For example, children wasted at baseline saw improvements in weight-for-height z- To access the ESS data: score when they gained a solid roof, improved toilet, hhtp://go.wordlbank.org/ZK2ZDZYDD0! or food assistance. Non-wasted children were not statistically significantly affected by such changes. !!!!!!!!!!!! ! !