Comparison of Multidimensional Measurements of Poverty Analysis: Cross Sectional Data Evidence from Tigray

Abstract

This paper examines the extent of poverty using the data from Tigray rural and urban baseline socio economic survey of 2200 households collected in 2011 by comparing and contrasting Alkire and Foster dual cut off approach, with Cluster analysis multidimensional poverty approaches. In addition for comparison purposes, a unidimensional poverty using per capita income approach was used.  The study measure multidimensional deprivation in ten dimensions: education, health condition, housing quality, electrification, and access to safe drinking water, sanitation, energy for cooking, per capita income, house congestion and child health. The results indicate that the multidimensional deprivation far exceeds the unidimensional poverty. It has been estimated that about 69 percent, 56.45 percent and 41.6 percent of the households are poor in Alkire and Foster counting approach, in cluster analysis approach and unidimensionally poor respectively.  In addition, the results also show that the decomposition of multidimensional and unidimensional poverty by location, indicates that in both methods of analysis, poverty is more prevalent in rural than urban areas. In comparing  Dual cutoff   and cluster analysis we find that at k=4, as cut off point 69 percent and 54.6 percent of the households are multidimensional poor in Alkire and Foster counting approach and in cluster analysis approach respectively. This shows that, Alkire and Foster dual cut off approach is the best estimation to measure the magnitude of multidimensional poverty and the level of deprivation in many dimensions. Using the intersection method at k=1, 99.6 percent of the total households are deprived in one or more dimensions. Among dimensions, above 88 percent of household head deprivation was due to lack of source of energy for cooking, i.e. the highest contributor to overall multidimensional poverty. Finally, the comparison results of the dual cutoff and counting approach with the cluster analysis of multidimensional poverty approach shows that the former one is the best suitable approach in estimation of multidimensional poverty analysis using different methods of poverty estimation.

Keywords: Multidimensional, Unidimensional, Dual cut off, Cluster, Tigray