Multidimensional Poverty: CMPI Development, Spatial Analysis and Clustering

The multidimensional poverty index of poverty is the most widely used method for estimating the poverty rate. It uses preassigned weights for poverty indicator that plays a critical role in determining an individual’s poverty status. But the question immediately arises: how do we justify these weights of poverty indicators? In order to overcome this issue, we have formulated composite multidimensional poverty index and composite headcount ratio using principal component analysis. Further, the self-organizing maps clustering algorithm, the K-means clustering algorithm, and spatial analysis are used to determine the spatial and temporal disparity of different poverty levels. In this respect, an Indian state viz. Odisha has been considered as a case study. These methods cluster poverty data into different poverty levels by analysing the poverty indicators only. This study concluded that multidimensional poverty in the considered state is widespread and severe, requiring prompt government action. In comparison to other methods, this one is proven to be more robust, all-encompassing, and indicator-neutral.