!!! ! Once poor always poor? Exploring selected from rural and small town areas, respectively, and twelve households were chosen from each EA. poverty dynamics in Ethiopia Tracking between waves was done at the household Using two waves of panel data from the Ethiopia level-- with a low attrition rate of 4.9%-- leading to a Socioeconomic Survey (ESS), we explore the dynamics panel sample of 3,776 households. We further exclude of wellbeing in rural and small town Ethiopia by households based on the criteria outlined in Table 1, assessing changes in poverty status based on for a final balanced sample of 3,480 households. consumption expenditures in 2012 and 2014. We Table 1. Sample size and exclusion criteria discover that although the prevalence of poverty in Wave 1 Wave 2 rural and small town Ethiopia remains relatively Excluded if: Total Excluded Total Excluded unchanged (approximately 30% in both 2012 and 1. Lost to attrition 3969 193 3776 - 2014), the proportion of the population facing poverty 2. Missing info for 3776 65 3776 128 in either year is much higher, at 47%. consumption aggregates 3. Zero total 3711 46 3648 22 Background!! consumption 4. Unmatched in two Many studies assessing poverty reduction use repeated 3665 139 3626 100 waves cross-sectional data to track trends over time at the 5. |∆| in consumption national level. However, panel data allows us to track 3526 46 3526 46 >25k Birr/year individuals over time, leading to a better understanding Sample size 3480 489 3480 296 of poverty dynamics. For example, are the same individuals poor at each point in time or is there Methods!! movement in wellbeing status? What characteristics To establish poverty lines in each wave, we use annual distinguish those facing chronic poverty from those consumption per adult equivalent. We inflate this wave facing transitory poverty? Answers to these questions 1 consumption to wave 2 levels by a factor of 1.21 as can help policymakers develop more effective, targeted reported by the CSA. We then set the poverty line to policies and interventions. a value that corresponds to the 30th percentile of total consumption in wave 1. For wave 2, we hold the value Data-- of the poverty line (3218 Birr/year per adult equivalent We analyze panel data from two waves of the ESS, a in 2014 terms), fixed. We then identify households that collaboration between the Central Statistics Agency of descended into poverty (backward movers), moved out Ethiopia (CSA) and the World Bank’s Living Standards of poverty (forward movers), were poor in both waves Measurement Study- Integrated Surveys of Agriculture (chronic poor), and non-poor in both waves (always (LSMS-ISA) project that collects multi-topic panel data non-poor). We also break down the consumption at the household level. The ESS began in 2011 (ESS1), aggregate into food and nonfood levels and shares to with 3,969 rural and small town households. In 2013, a assess the extent to which the nature of consumption second wave (ESS2) was administered, revisiting the changed over time according to poverty dynamic ESS1 households and an additional 1,500 urban status. households; the panel sample includes rural and small town households only. Results-- The ESS uses a stratified, two-stage sampling scheme. In the aggregate, we find that total and food Enumeration areas (EAs) were randomly selected in expenditures decreased between 2012 and 2014, while proportion to population size; 290 and 43 EAs were nonfood expenditures increased (see Table 2). Additionally, the composition of expenditure appears non-food’. As Bennett’s law of food demand predicts1, to have shifted slightly; on average, households shifted forward movers spend smaller shares on starchy their relative share of consumption to nonfood items. staples, but larger shares on nutritious foods like ASF and vegetables and fruits; conversely, backward Table 2. Trends in consumption expenditures movers increase the proportion spent on staples and Full Sample decrease the relative share spent on more nutritious Wave 1 Wave 2 foods. These results show that movement in and out Diff Consumption (2011/12) (2013/14) of poverty is also accompanied by shifts in wellbeing Total 5261 4889 -372** as measured through quality of food consumption. (168.25) (144.01) Food 4358 3874 -484*** (147.54) (125.53) Figure 1. Change in shares of food and nonfood Nonfood 903 1014 111 ** expenditures, by poverty dynamic category (45.85) (42.43) 8.0 Food and nonfood shares Food 0.82 0.79 -0.04*** 4.0 (0.01 (0.01) *** Nonfood 0.18 0.21 0.04 (0.01 (0.01) 0.0 % Observations 3480 3480 -4.0 Despite a statistically significant drop in mean consumption, we find that the prevalence of poverty -8.0 remained approximately unchanged at 30% and 32% Chronic poor Forward movers in waves 1 and 2, respectively. However, the minimal Backward movers Always Nonpoor change in poverty in the aggregate masks substantial movement in and out of poverty at the individual level, as shown in Table 3. Only 15% of individuals Discussion-&-Policy-Implications- were poor in both waves; in contrast, 15% escaped Our analysis emphasizes the point that household-level poverty and 17% fell into poverty between waves. In panel data provide a clearer profile of poverty (relative fact, nearly 47% of individuals were poor at either to cross-sectional data) because they can distinguish point in time. This result emphasizes that poverty the transient from the chronic poor. This distinction estimates based on cross-sectional data may between transitory and chronic poverty has important underestimate the extent of poverty as they do not implications for policy design; food assistance or other capture the proportion of individuals moving in and safety nets can effectively aid the transient poor, while out of poverty over time. policies for the chronic poor most likely require more significant long-run investments in human capital and Table 3. Poverty dynamics physical infrastructure. Wave 2 (2013/14) Wave 1 (2011/12) Non-poor Poor Total Generous funding assistance for this research came from the Non-poor 53.2 16.5 69.7 UK Department for International Development Ethiopia. Poor 14.9 15.4 30.3 The findings outlined in this brief are drawn from: Kafle, K., Total 68.1 31.9 100 McGee, K., Ambel, A., & Seff, I. (forthcoming) “Once poor always poor? Exploring consumption- and asset-based poverty Examining changes in expenditure shares, we find that dynamics in Ethiopia” chronically poor households decrease expenditure To access the ESS data: shares on all food and non-food items excluding ‘other hhtp://go.wordlbank.org/ZK2ZDZYDD0! !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! 1The Bennett’s law of food demand states ‘As income rises the proportion of starchy staples in the diet falls’ !!!!!!!!!!!! ! !