MULTIDIMENSIONAL RURAL POVERTY ACROSS AGRO ECOLOGIES IN ETHIOPIA: EVIDENCE FROM THE THREE DISTRICTS

Abstract

Ethiopia is home to around 114 million people, whereas it holds 85 percent of the agricultural economy. Agro ecology differences matter to analyze poverty. However, there is a shortage of studies based on agro ecological studies across districts in Ethiopia. As far as the Ethiopian notion of poverty is concerned, identifying the poor is mainly limited to unidimensional poverty, that is, consumption/income dimension, following prescribed norms of either minimum calories or subsistence nutritional requirements. However, research interest demands a multidimensional poverty analysis across agro ecology, districts, and impact of microcredit services on multidimensional poverty, in addition to its determinants within the districts of Ethiopia. The present study undertakes the multidimensional notion of poverty starting from the unidimensional poverty to build a comprehensive scenario of poverty prevailing in Ethiopia. It has been done by choosing a cross-sectional survey data from 450 rural households using the three known agro ecologies of Ethiopia: Qolla, Woyinadega, and Dega, and among three districts. The unidimensional conception of poverty is measured in terms of consumption expenditure, using the household level survey in the three districts of these agro ecologies in 2019. The consumption poverty is aggregated considering the aggregate number of the poor, levels of poverty, and differences in poverty. The levels of poverty are assessed in terms of Headcount Ratio (HCR), Poverty Gap Ratio (PGR), and Square Poverty Gap Ratio (SPGR). These measures target the nearly poor, moderately poor, and severely poor population, respectively. It also examines the factors underlining poverty differences across districts and agro ecologies. The second conception, Multidimensional Poverty, comprises the construction of the Multidimensional Poverty Index (MPI) that includes three dimensions: education, health, and standard of living, which are represented by eleven indicators such as adult education, child school attendance, under nutrition, child mortality, access to light, safe drinking water sources, improved sanitation, safe cooking fuel, house roofing, access to a road and assets. The three dimensions are equally weighted (33.3% each), distributed evenly across indicators. Finally, any household whose total deprivation count is higher than or equal to poverty cut (k=33.3%) is considered as MD poor. After identification, the aggregation of MD poverty is i

ii

estimated by way of Headcount (H) and Intensity (A) components of MPI. The former defines the incidence (or proportion) of MD poor people, and the latter refers to the intensity of poverty, which is the average deprivation of the MD poor people. The MPI in the form of the index is computed as a product of H and A. The value of the Multidimensional Poverty Index denotes the population share that is MD poor accustomed by the intensity of the deprivation suffered. Next is the decomposition of MPI in the sense of contributions to overall poverty, first by dimensions and indicators and then by agro ecologies and districts. The estimation of MPI is based on the cross-sectional household survey taken from the three agro-ecologies of Ethiopia for 2019. Both Consumption poverty and MD poverty are estimated using the same survey data, and both can be compared meaningfully. The disaggregated and the sub-group analysis are also done with the same data to show whether there is a difference in poverty across agro-ecologies. The precision of the MPI estimates is tested on the grounds of mismatches in the identification of MD poor when equated with other notions of poverty (wealth and consumption), followed by correlation analysis and conditional probabilities associated with them. The robustness of MPI estimates is determined by a sensitivity analysis of the changes in deprivation weights and poverty cut-offs (k=33.3%). Multidimensional poverty measurements for one region or country may not fit for other regions or countries due to contextual differences in the indicators. It is in this case that this thesis, besides applying the existing dimension indicators, uses some other reasonable dimension indicators that show the reliable measurement of MDP in agro-ecologies of Ethiopia. Our study shows that the prevalence (H), average deprivation (A), and adjusted multidimensional rural poverty (Mo) are 84.2, 61.0, and 51.3 percent, respectively. The results indicate that multidimensional poverty is severe in the study area. The higher prevalence of MDP is due to inadequate access to necessary infrastructure and service privilege in general and poor access to electricity, solar energy, improved fuel sources, sanitation privilege, safe drinking water, and road in particular. The result shows that these indicators are severe in Qolla and Dega agro-ecologies than Woyinadega agro-ecology. Besides, if the rural household lives in Qolla agro-ecologies, on average, marginal effect, the probability of being multidimensional poor increases by 3.6 percent, whereas, in Woyinadega, it goes down by 11.3 percent, keeping all other factors constant.

iii

After estimation of multidimensional poverty across agro-ecologies, we identify determinants of MDP status of rural households in Ethiopia. The Binary Logit Model is used to study the determinants. The finding shows that household empowerment index, household head's education, agricultural extension services, irrigation, credit access, total livestock units, road access, and Woyinadega agro-ecology have odds ratios less than one. These variables are negatively related to the likelihood of being poor; implying that rise in these variables lessens the probability of being MPI-poor. On the contrary, variables dependency ratio, gender, and Qolla agro-ecology have odd-ratios higher than one, which implies that these variables have a positive statistically significant impact on the likelihood of being poor. Moreover, the marginal effects reveal that keeping other things constant, as the amount of each independent variable increases by a unit, on average, the likelihood of being multidimensional poor decreases by the coefficient percent, that is, education of the household's head decreases MDP by 0.9 percent, agriculture extension services by 2.9 percent, access to credit by 5.7 percent, livestock units by 0.2 percent, access to irrigation by 1.4 percent, access to the road by 3.1 percent and living in Woyina-Dega agro-ecology by 11.3 percent. In addition, keeping all other things constant, as the dependency ratio increases by a unit, the likelihood of being multidimensional poor rises by 0.4 percent; living in Qolla agro-ecology increases it by 1.9 percent; and being female-headed household raises it by 3.3 percent. This study also analyses the impact of the potential household empowerment index (HEMPI) on the MDP status of households. The HEMPI is constructed through Principal Component Analysis, and the Binary Logit Model is applied to assess the effect of the HEMPI along with control variables on the MDP. Results confirm that HEMPI has a significant impact on reducing MDP. The marginal effects reveal that citrus paribus, as the HEMPI increases by a unit percent, the likelihood of being MPI poor decreases by 1.17 percent. Finally, the study examines the impact of credit services on the households’ MDP status. The Propensity Score Matching Model is used to investigate the impact of credit services on the MDP status. The findings show that access to credit services reduces the MDP index by 0.2538, living standard deprivation by 0.1490, education deprivation by 0.1023, and health deprivation by 0.0897, on average, for the microcredit members.

iv

Our study's significant contribution is that variations in unidimensional poverty are independent of changes in multidimensional welfare, showing its severity than unidimensional one. MDP index varies across agro-ecologies with all the three dimensions, showing more severity at Qolla agro-ecology. Microcredit service and household empowerment are significant factors in reducing MDP index. The study recommends that Government of Ethiopia and concerned organizations should expend their efforts on improving living standards. They need to familiarize the households with alternative sources of light and fuel energy; and mobilize more resources to construct safe drinking water points, sanitation, and road infrastructure with a priority for Qolla and Dega agro-ecologies. The government should give special attention to alleviate MDP through raising livestock production, training and capacity of farmers, work against decreasing dependency of young age, increasing road accessibility, increasing the availability of agronomic extension employees, expanding the irrigation facilities, and giving an individual access of rural households to credit in general, and giving special attention for gender lines with particular focus in Qolla agro-ecology. The study suggests that the country administration and concerned groups ought to give special attention to alleviate the MDP status of the community by increasing the potential human empowerment (economically, socially, and politically) for rural households in Ethiopia. It is policy advice that the Ethiopian régime should give extraordinary responsiveness to the organizations and promote financial cooperatives. Cooperative extension led labor-force had better advice for non-membership to join the micro-credit institutions, and microcredit institutions need to expand their commercial services for improving the living, education, and health standards in the rural areas.